quick_xml/de/mod.rs
1//! Serde `Deserializer` module.
2//!
3//! Due to the complexity of the XML standard and the fact that Serde was developed
4//! with JSON in mind, not all Serde concepts apply smoothly to XML. This leads to
5//! that fact that some XML concepts are inexpressible in terms of Serde derives
6//! and may require manual deserialization.
7//!
8//! The most notable restriction is the ability to distinguish between _elements_
9//! and _attributes_, as no other format used by serde has such a conception.
10//!
11//! Due to that the mapping is performed in a best effort manner.
12//!
13//!
14//!
15//! Table of Contents
16//! =================
17//! - [Mapping XML to Rust types](#mapping-xml-to-rust-types)
18//! - [Basics](#basics)
19//! - [Optional attributes and elements](#optional-attributes-and-elements)
20//! - [Choices (`xs:choice` XML Schema type)](#choices-xschoice-xml-schema-type)
21//! - [Sequences (`xs:all` and `xs:sequence` XML Schema types)](#sequences-xsall-and-xssequence-xml-schema-types)
22//! - [Mapping of `xsi:nil`](#mapping-of-xsinil)
23//! - [Generate Rust types from XML](#generate-rust-types-from-xml)
24//! - [Composition Rules](#composition-rules)
25//! - [Enum Representations](#enum-representations)
26//! - [Normal enum variant](#normal-enum-variant)
27//! - [`$text` enum variant](#text-enum-variant)
28//! - [`$text` and `$value` special names](#text-and-value-special-names)
29//! - [`$text`](#text)
30//! - [`$value`](#value)
31//! - [Primitives and sequences of primitives](#primitives-and-sequences-of-primitives)
32//! - [Structs and sequences of structs](#structs-and-sequences-of-structs)
33//! - [Enums and sequences of enums](#enums-and-sequences-of-enums)
34//! - [Frequently Used Patterns](#frequently-used-patterns)
35//! - [`<element>` lists](#element-lists)
36//! - [Overlapped (Out-of-Order) Elements](#overlapped-out-of-order-elements)
37//! - [Internally Tagged Enums](#internally-tagged-enums)
38//!
39//!
40//!
41//! Mapping XML to Rust types
42//! =========================
43//!
44//! Type names are never considered when deserializing, so you can name your
45//! types as you wish. Other general rules:
46//! - `struct` field name could be represented in XML only as an attribute name
47//! or an element name;
48//! - `enum` variant name could be represented in XML only as an attribute name
49//! or an element name;
50//! - the unit struct, unit type `()` and unit enum variant can be deserialized
51//! from any valid XML content:
52//! - attribute and element names;
53//! - attribute and element values;
54//! - text or CDATA content (including mixed text and CDATA content).
55//!
56//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
57//!
58//! NOTE: All tests are marked with an `ignore` option, even though they do
59//! compile. This is because rustdoc marks such blocks with an information
60//! icon unlike `no_run` blocks.
61//!
62//! </div>
63//!
64//! <table>
65//! <thead>
66//! <tr><th colspan="2">
67//!
68//! ## Basics
69//!
70//! </th></tr>
71//! <tr><th>To parse all these XML's...</th><th>...use these Rust type(s)</th></tr>
72//! </thead>
73//! <tbody style="vertical-align:top;">
74//! <tr>
75//! <td>
76//! Content of attributes and text / CDATA content of elements (including mixed
77//! text and CDATA content):
78//!
79//! ```xml
80//! <... ...="content" />
81//! ```
82//! ```xml
83//! <...>content</...>
84//! ```
85//! ```xml
86//! <...><![CDATA[content]]></...>
87//! ```
88//! ```xml
89//! <...>text<![CDATA[cdata]]>text</...>
90//! ```
91//! Mixed text / CDATA content represents one logical string, `"textcdatatext"` in that case.
92//! </td>
93//! <td>
94//!
95//! You can use any type that can be deserialized from an `&str`, for example:
96//! - [`String`] and [`&str`]
97//! - [`Cow<str>`]
98//! - [`u32`], [`f32`] and other numeric types
99//! - `enum`s, like
100//! ```
101//! # use pretty_assertions::assert_eq;
102//! # use serde::Deserialize;
103//! # #[derive(Debug, PartialEq)]
104//! #[derive(Deserialize)]
105//! enum Language {
106//! Rust,
107//! Cpp,
108//! #[serde(other)]
109//! Other,
110//! }
111//! # #[derive(Debug, PartialEq, Deserialize)]
112//! # struct X { #[serde(rename = "$text")] x: Language }
113//! # assert_eq!(X { x: Language::Rust }, quick_xml::de::from_str("<x>Rust</x>").unwrap());
114//! # assert_eq!(X { x: Language::Cpp }, quick_xml::de::from_str("<x>C<![CDATA[p]]>p</x>").unwrap());
115//! # assert_eq!(X { x: Language::Other }, quick_xml::de::from_str("<x><![CDATA[other]]></x>").unwrap());
116//! ```
117//!
118//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
119//!
120//! NOTE: deserialization to non-owned types (i.e. borrow from the input),
121//! such as `&str`, is possible only if you parse document in the UTF-8
122//! encoding and content does not contain entity references such as `&`,
123//! or character references such as `
`, as well as text content represented
124//! by one piece of [text] or [CDATA] element.
125//! </div>
126//! <!-- TODO: document an error type returned -->
127//!
128//! [text]: Event::Text
129//! [CDATA]: Event::CData
130//! </td>
131//! </tr>
132//! <!-- 2 ===================================================================================== -->
133//! <tr>
134//! <td>
135//!
136//! Content of attributes and text / CDATA content of elements (including mixed
137//! text and CDATA content), which represents a space-delimited lists, as
138//! specified in the XML Schema specification for [`xs:list`] `simpleType`:
139//!
140//! ```xml
141//! <... ...="element1 element2 ..." />
142//! ```
143//! ```xml
144//! <...>
145//! element1
146//! element2
147//! ...
148//! </...>
149//! ```
150//! ```xml
151//! <...><![CDATA[
152//! element1
153//! element2
154//! ...
155//! ]]></...>
156//! ```
157//!
158//! [`xs:list`]: https://www.w3.org/TR/xmlschema11-2/#list-datatypes
159//! </td>
160//! <td>
161//!
162//! Use any type that deserialized using [`deserialize_seq()`] call, for example:
163//!
164//! ```
165//! type List = Vec<u32>;
166//! ```
167//!
168//! See the next row to learn where in your struct definition you should
169//! use that type.
170//!
171//! According to the XML Schema specification, delimiters for elements is one
172//! or more space (`' '`, `'\r'`, `'\n'`, and `'\t'`) character(s).
173//!
174//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
175//!
176//! NOTE: according to the XML Schema restrictions, you cannot escape those
177//! white-space characters, so list elements will _never_ contain them.
178//! In practice you will usually use `xs:list`s for lists of numbers or enumerated
179//! values which looks like identifiers in many languages, for example, `item`,
180//! `some_item` or `some-item`, so that shouldn't be a problem.
181//!
182//! NOTE: according to the XML Schema specification, list elements can be
183//! delimited only by spaces. Other delimiters (for example, commas) are not
184//! allowed.
185//!
186//! </div>
187//!
188//! [`deserialize_seq()`]: de::Deserializer::deserialize_seq
189//! </td>
190//! </tr>
191//! <!-- 3 ===================================================================================== -->
192//! <tr>
193//! <td>
194//! A typical XML with attributes. The root tag name does not matter:
195//!
196//! ```xml
197//! <any-tag one="..." two="..."/>
198//! ```
199//! </td>
200//! <td>
201//!
202//! A structure where each XML attribute is mapped to a field with a name
203//! starting with `@`. Because Rust identifiers do not permit the `@` character,
204//! you should use the `#[serde(rename = "@...")]` attribute to rename it.
205//! The name of the struct itself does not matter:
206//!
207//! ```
208//! # use serde::Deserialize;
209//! # type T = ();
210//! # type U = ();
211//! // Get both attributes
212//! # #[derive(Debug, PartialEq)]
213//! #[derive(Deserialize)]
214//! struct AnyName {
215//! #[serde(rename = "@one")]
216//! one: T,
217//!
218//! #[serde(rename = "@two")]
219//! two: U,
220//! }
221//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..." two="..."/>"#).unwrap();
222//! ```
223//! ```
224//! # use serde::Deserialize;
225//! # type T = ();
226//! // Get only the one attribute, ignore the other
227//! # #[derive(Debug, PartialEq)]
228//! #[derive(Deserialize)]
229//! struct AnyName {
230//! #[serde(rename = "@one")]
231//! one: T,
232//! }
233//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..." two="..."/>"#).unwrap();
234//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..."/>"#).unwrap();
235//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..."><one>...</one></any-tag>"#).unwrap();
236//! ```
237//! ```
238//! # use serde::Deserialize;
239//! // Ignore all attributes
240//! // You can also use the `()` type (unit type)
241//! # #[derive(Debug, PartialEq)]
242//! #[derive(Deserialize)]
243//! struct AnyName;
244//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..." two="..."/>"#).unwrap();
245//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..."><one>...</one></any-tag>"#).unwrap();
246//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag><one>...</one><two>...</two></any-tag>"#).unwrap();
247//! ```
248//!
249//! All these structs can be used to deserialize from an XML on the
250//! left side depending on amount of information that you want to get.
251//! Of course, you can combine them with elements extractor structs (see below).
252//!
253//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
254//!
255//! NOTE: XML allows you to have an attribute and an element with the same name
256//! inside the one element. quick-xml deals with that by prepending a `@` prefix
257//! to the name of attributes.
258//! </div>
259//! </td>
260//! </tr>
261//! <!-- 4 ===================================================================================== -->
262//! <tr>
263//! <td>
264//! A typical XML with child elements. The root tag name does not matter:
265//!
266//! ```xml
267//! <any-tag>
268//! <one>...</one>
269//! <two>...</two>
270//! </any-tag>
271//! ```
272//! </td>
273//! <td>
274//! A structure where each XML child element is mapped to the field.
275//! Each element name becomes a name of field. The name of the struct itself
276//! does not matter:
277//!
278//! ```
279//! # use serde::Deserialize;
280//! # type T = ();
281//! # type U = ();
282//! // Get both elements
283//! # #[derive(Debug, PartialEq)]
284//! #[derive(Deserialize)]
285//! struct AnyName {
286//! one: T,
287//! two: U,
288//! }
289//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag><one>...</one><two>...</two></any-tag>"#).unwrap();
290//! #
291//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..." two="..."/>"#).unwrap_err();
292//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..."><two>...</two></any-tag>"#).unwrap_err();
293//! ```
294//! ```
295//! # use serde::Deserialize;
296//! # type T = ();
297//! // Get only the one element, ignore the other
298//! # #[derive(Debug, PartialEq)]
299//! #[derive(Deserialize)]
300//! struct AnyName {
301//! one: T,
302//! }
303//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag><one>...</one><two>...</two></any-tag>"#).unwrap();
304//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..."><one>...</one></any-tag>"#).unwrap();
305//! ```
306//! ```
307//! # use serde::Deserialize;
308//! // Ignore all elements
309//! // You can also use the `()` type (unit type)
310//! # #[derive(Debug, PartialEq)]
311//! #[derive(Deserialize)]
312//! struct AnyName;
313//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..." two="..."/>"#).unwrap();
314//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag><one>...</one><two>...</two></any-tag>"#).unwrap();
315//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..."><two>...</two></any-tag>"#).unwrap();
316//! # quick_xml::de::from_str::<AnyName>(r#"<any-tag one="..."><one>...</one></any-tag>"#).unwrap();
317//! ```
318//!
319//! All these structs can be used to deserialize from an XML on the
320//! left side depending on amount of information that you want to get.
321//! Of course, you can combine them with attributes extractor structs (see above).
322//!
323//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
324//!
325//! NOTE: XML allows you to have an attribute and an element with the same name
326//! inside the one element. quick-xml deals with that by prepending a `@` prefix
327//! to the name of attributes.
328//! </div>
329//! </td>
330//! </tr>
331//! <!-- 5 ===================================================================================== -->
332//! <tr>
333//! <td>
334//! An XML with an attribute and a child element named equally:
335//!
336//! ```xml
337//! <any-tag field="...">
338//! <field>...</field>
339//! </any-tag>
340//! ```
341//! </td>
342//! <td>
343//!
344//! You MUST specify `#[serde(rename = "@field")]` on a field that will be used
345//! for an attribute:
346//!
347//! ```
348//! # use pretty_assertions::assert_eq;
349//! # use serde::Deserialize;
350//! # type T = ();
351//! # type U = ();
352//! # #[derive(Debug, PartialEq)]
353//! #[derive(Deserialize)]
354//! struct AnyName {
355//! #[serde(rename = "@field")]
356//! attribute: T,
357//! field: U,
358//! }
359//! # assert_eq!(
360//! # AnyName { attribute: (), field: () },
361//! # quick_xml::de::from_str(r#"
362//! # <any-tag field="...">
363//! # <field>...</field>
364//! # </any-tag>
365//! # "#).unwrap(),
366//! # );
367//! ```
368//! </td>
369//! </tr>
370//! <!-- ======================================================================================= -->
371//! <tr><th colspan="2">
372//!
373//! ## Optional attributes and elements
374//!
375//! </th></tr>
376//! <tr><th>To parse all these XML's...</th><th>...use these Rust type(s)</th></tr>
377//! <!-- 6 ===================================================================================== -->
378//! <tr>
379//! <td>
380//! An optional XML attribute that you want to capture.
381//! The root tag name does not matter:
382//!
383//! ```xml
384//! <any-tag optional="..."/>
385//! ```
386//! ```xml
387//! <any-tag/>
388//! ```
389//! </td>
390//! <td>
391//!
392//! A structure with an optional field, renamed according to the requirements
393//! for attributes:
394//!
395//! ```
396//! # use pretty_assertions::assert_eq;
397//! # use serde::Deserialize;
398//! # type T = ();
399//! # #[derive(Debug, PartialEq)]
400//! #[derive(Deserialize)]
401//! struct AnyName {
402//! #[serde(rename = "@optional")]
403//! optional: Option<T>,
404//! }
405//! # assert_eq!(AnyName { optional: Some(()) }, quick_xml::de::from_str(r#"<any-tag optional="..."/>"#).unwrap());
406//! # assert_eq!(AnyName { optional: None }, quick_xml::de::from_str(r#"<any-tag/>"#).unwrap());
407//! ```
408//! When the XML attribute is present, type `T` will be deserialized from
409//! an attribute value (which is a string). Note, that if `T = String` or other
410//! string type, the empty attribute is mapped to a `Some("")`, whereas `None`
411//! represents the missed attribute:
412//! ```xml
413//! <any-tag optional="..."/><!-- Some("...") -->
414//! <any-tag optional=""/> <!-- Some("") -->
415//! <any-tag/> <!-- None -->
416//! ```
417//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
418//!
419//! NOTE: The behaviour is not symmetric by default. `None` will be serialized as
420//! `optional=""`. This behaviour is consistent across serde crates. You should add
421//! `#[serde(skip_serializing_if = "Option::is_none")]` attribute to the field to
422//! skip `None`s.
423//! </div>
424//! </td>
425//! </tr>
426//! <!-- 7 ===================================================================================== -->
427//! <tr>
428//! <td>
429//! An optional XML elements that you want to capture.
430//! The root tag name does not matter:
431//!
432//! ```xml
433//! <any-tag/>
434//! <optional>...</optional>
435//! </any-tag>
436//! ```
437//! ```xml
438//! <any-tag/>
439//! <optional/>
440//! </any-tag>
441//! ```
442//! ```xml
443//! <any-tag/>
444//! ```
445//! </td>
446//! <td>
447//!
448//! A structure with an optional field:
449//!
450//! ```
451//! # use pretty_assertions::assert_eq;
452//! # use serde::Deserialize;
453//! # type T = ();
454//! # #[derive(Debug, PartialEq)]
455//! #[derive(Deserialize)]
456//! struct AnyName {
457//! optional: Option<T>,
458//! }
459//! # assert_eq!(AnyName { optional: Some(()) }, quick_xml::de::from_str(r#"<any-tag><optional>...</optional></any-tag>"#).unwrap());
460//! # assert_eq!(AnyName { optional: None }, quick_xml::de::from_str(r#"<any-tag/>"#).unwrap());
461//! ```
462//! When the XML element is present, type `T` will be deserialized from an
463//! element (which is a string or a multi-mapping -- i.e. mapping which can have
464//! duplicated keys).
465//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
466//!
467//! NOTE: The behaviour is not symmetric by default. `None` will be serialized as
468//! `<optional/>`. This behaviour is consistent across serde crates. You should add
469//! `#[serde(skip_serializing_if = "Option::is_none")]` attribute to the field to
470//! skip `None`s.
471//!
472//! NOTE: Deserializer will automatically handle a [`xsi:nil`] attribute and set field to `None`.
473//! For more info see [Mapping of `xsi:nil`](#mapping-of-xsinil).
474//! </div>
475//! </td>
476//! </tr>
477//! <!-- ======================================================================================= -->
478//! <tr><th colspan="2">
479//!
480//! ## Choices (`xs:choice` XML Schema type)
481//!
482//! </th></tr>
483//! <tr><th>To parse all these XML's...</th><th>...use these Rust type(s)</th></tr>
484//! <!-- 8 ===================================================================================== -->
485//! <tr>
486//! <td>
487//! An XML with different root tag names, as well as text / CDATA content:
488//!
489//! ```xml
490//! <one field1="...">...</one>
491//! ```
492//! ```xml
493//! <two>
494//! <field2>...</field2>
495//! </two>
496//! ```
497//! ```xml
498//! Text <![CDATA[or (mixed)
499//! CDATA]]> content
500//! ```
501//! </td>
502//! <td>
503//!
504//! An enum where each variant has the name of a possible root tag. The name of
505//! the enum itself does not matter.
506//!
507//! If you need to get the textual content, mark a variant with `#[serde(rename = "$text")]`.
508//!
509//! All these structs can be used to deserialize from any XML on the
510//! left side depending on amount of information that you want to get:
511//!
512//! ```
513//! # use pretty_assertions::assert_eq;
514//! # use serde::Deserialize;
515//! # type T = ();
516//! # type U = ();
517//! # #[derive(Debug, PartialEq)]
518//! #[derive(Deserialize)]
519//! #[serde(rename_all = "snake_case")]
520//! enum AnyName {
521//! One { #[serde(rename = "@field1")] field1: T },
522//! Two { field2: U },
523//!
524//! /// Use unit variant, if you do not care of a content.
525//! /// You can use tuple variant if you want to parse
526//! /// textual content as an xs:list.
527//! /// Struct variants are will pass a string to the
528//! /// struct enum variant visitor, which typically
529//! /// returns Err(Custom)
530//! #[serde(rename = "$text")]
531//! Text(String),
532//! }
533//! # assert_eq!(AnyName::One { field1: () }, quick_xml::de::from_str(r#"<one field1="...">...</one>"#).unwrap());
534//! # assert_eq!(AnyName::Two { field2: () }, quick_xml::de::from_str(r#"<two><field2>...</field2></two>"#).unwrap());
535//! # assert_eq!(AnyName::Text("text cdata ".into()), quick_xml::de::from_str(r#"text <![CDATA[ cdata ]]>"#).unwrap());
536//! ```
537//! ```
538//! # use pretty_assertions::assert_eq;
539//! # use serde::Deserialize;
540//! # type T = ();
541//! # #[derive(Debug, PartialEq)]
542//! #[derive(Deserialize)]
543//! struct Two {
544//! field2: T,
545//! }
546//! # #[derive(Debug, PartialEq)]
547//! #[derive(Deserialize)]
548//! #[serde(rename_all = "snake_case")]
549//! enum AnyName {
550//! // `field1` content discarded
551//! One,
552//! Two(Two),
553//! #[serde(rename = "$text")]
554//! Text,
555//! }
556//! # assert_eq!(AnyName::One, quick_xml::de::from_str(r#"<one field1="...">...</one>"#).unwrap());
557//! # assert_eq!(AnyName::Two(Two { field2: () }), quick_xml::de::from_str(r#"<two><field2>...</field2></two>"#).unwrap());
558//! # assert_eq!(AnyName::Text, quick_xml::de::from_str(r#"text <![CDATA[ cdata ]]>"#).unwrap());
559//! ```
560//! ```
561//! # use pretty_assertions::assert_eq;
562//! # use serde::Deserialize;
563//! # #[derive(Debug, PartialEq)]
564//! #[derive(Deserialize)]
565//! #[serde(rename_all = "snake_case")]
566//! enum AnyName {
567//! One,
568//! // the <two> and textual content will be mapped to this
569//! #[serde(other)]
570//! Other,
571//! }
572//! # assert_eq!(AnyName::One, quick_xml::de::from_str(r#"<one field1="...">...</one>"#).unwrap());
573//! # assert_eq!(AnyName::Other, quick_xml::de::from_str(r#"<two><field2>...</field2></two>"#).unwrap());
574//! # assert_eq!(AnyName::Other, quick_xml::de::from_str(r#"text <![CDATA[ cdata ]]>"#).unwrap());
575//! ```
576//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
577//!
578//! NOTE: You should have variants for all possible tag names in your enum
579//! or have an `#[serde(other)]` variant.
580//! <!-- TODO: document an error type if that requirement is violated -->
581//! </div>
582//! </td>
583//! </tr>
584//! <!-- 9 ===================================================================================== -->
585//! <tr>
586//! <td>
587//!
588//! `<xs:choice>` embedded in the other element, and at the same time you want
589//! to get access to other attributes that can appear in the same container
590//! (`<any-tag>`). Also this case can be described, as if you want to choose
591//! Rust enum variant based on a tag name:
592//!
593//! ```xml
594//! <any-tag field="...">
595//! <one>...</one>
596//! </any-tag>
597//! ```
598//! ```xml
599//! <any-tag field="...">
600//! <two>...</two>
601//! </any-tag>
602//! ```
603//! ```xml
604//! <any-tag field="...">
605//! Text <![CDATA[or (mixed)
606//! CDATA]]> content
607//! </any-tag>
608//! ```
609//! </td>
610//! <td>
611//!
612//! A structure with a field which type is an `enum`.
613//!
614//! If you need to get a textual content, mark a variant with `#[serde(rename = "$text")]`.
615//!
616//! Names of the enum, struct, and struct field with `Choice` type does not matter:
617//!
618//! ```
619//! # use pretty_assertions::assert_eq;
620//! # use serde::Deserialize;
621//! # type T = ();
622//! # #[derive(Debug, PartialEq)]
623//! #[derive(Deserialize)]
624//! #[serde(rename_all = "snake_case")]
625//! enum Choice {
626//! One,
627//! Two,
628//!
629//! /// Use unit variant, if you do not care of a content.
630//! /// You can use tuple variant if you want to parse
631//! /// textual content as an xs:list.
632//! /// Struct variants are will pass a string to the
633//! /// struct enum variant visitor, which typically
634//! /// returns Err(Custom)
635//! #[serde(rename = "$text")]
636//! Text(String),
637//! }
638//! # #[derive(Debug, PartialEq)]
639//! #[derive(Deserialize)]
640//! struct AnyName {
641//! #[serde(rename = "@field")]
642//! field: T,
643//!
644//! #[serde(rename = "$value")]
645//! any_name: Choice,
646//! }
647//! # assert_eq!(
648//! # AnyName { field: (), any_name: Choice::One },
649//! # quick_xml::de::from_str(r#"<any-tag field="..."><one>...</one></any-tag>"#).unwrap(),
650//! # );
651//! # assert_eq!(
652//! # AnyName { field: (), any_name: Choice::Two },
653//! # quick_xml::de::from_str(r#"<any-tag field="..."><two>...</two></any-tag>"#).unwrap(),
654//! # );
655//! # assert_eq!(
656//! # AnyName { field: (), any_name: Choice::Text("text cdata ".into()) },
657//! # quick_xml::de::from_str(r#"<any-tag field="...">text <![CDATA[ cdata ]]></any-tag>"#).unwrap(),
658//! # );
659//! ```
660//! </td>
661//! </tr>
662//! <!-- 10 ==================================================================================== -->
663//! <tr>
664//! <td>
665//!
666//! `<xs:choice>` embedded in the other element, and at the same time you want
667//! to get access to other elements that can appear in the same container
668//! (`<any-tag>`). Also this case can be described, as if you want to choose
669//! Rust enum variant based on a tag name:
670//!
671//! ```xml
672//! <any-tag>
673//! <field>...</field>
674//! <one>...</one>
675//! </any-tag>
676//! ```
677//! ```xml
678//! <any-tag>
679//! <two>...</two>
680//! <field>...</field>
681//! </any-tag>
682//! ```
683//! </td>
684//! <td>
685//!
686//! A structure with a field which type is an `enum`.
687//!
688//! Names of the enum, struct, and struct field with `Choice` type does not matter:
689//!
690//! ```
691//! # use pretty_assertions::assert_eq;
692//! # use serde::Deserialize;
693//! # type T = ();
694//! # #[derive(Debug, PartialEq)]
695//! #[derive(Deserialize)]
696//! #[serde(rename_all = "snake_case")]
697//! enum Choice {
698//! One,
699//! Two,
700//! }
701//! # #[derive(Debug, PartialEq)]
702//! #[derive(Deserialize)]
703//! struct AnyName {
704//! field: T,
705//!
706//! #[serde(rename = "$value")]
707//! any_name: Choice,
708//! }
709//! # assert_eq!(
710//! # AnyName { field: (), any_name: Choice::One },
711//! # quick_xml::de::from_str(r#"<any-tag><field>...</field><one>...</one></any-tag>"#).unwrap(),
712//! # );
713//! # assert_eq!(
714//! # AnyName { field: (), any_name: Choice::Two },
715//! # quick_xml::de::from_str(r#"<any-tag><two>...</two><field>...</field></any-tag>"#).unwrap(),
716//! # );
717//! ```
718//!
719//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
720//!
721//! NOTE: if your `Choice` enum would contain an `#[serde(other)]`
722//! variant, element `<field>` will be mapped to the `field` and not to the enum
723//! variant.
724//! </div>
725//!
726//! </td>
727//! </tr>
728//! <!-- 11 ==================================================================================== -->
729//! <tr>
730//! <td>
731//!
732//! `<xs:choice>` encapsulated in other element with a fixed name:
733//!
734//! ```xml
735//! <any-tag field="...">
736//! <choice>
737//! <one>...</one>
738//! </choice>
739//! </any-tag>
740//! ```
741//! ```xml
742//! <any-tag field="...">
743//! <choice>
744//! <two>...</two>
745//! </choice>
746//! </any-tag>
747//! ```
748//! </td>
749//! <td>
750//!
751//! A structure with a field of an intermediate type with one field of `enum` type.
752//! Actually, this example is not necessary, because you can construct it by yourself
753//! using the composition rules that were described above. However the XML construction
754//! described here is very common, so it is shown explicitly.
755//!
756//! Names of the enum and struct does not matter:
757//!
758//! ```
759//! # use pretty_assertions::assert_eq;
760//! # use serde::Deserialize;
761//! # type T = ();
762//! # #[derive(Debug, PartialEq)]
763//! #[derive(Deserialize)]
764//! #[serde(rename_all = "snake_case")]
765//! enum Choice {
766//! One,
767//! Two,
768//! }
769//! # #[derive(Debug, PartialEq)]
770//! #[derive(Deserialize)]
771//! struct Holder {
772//! #[serde(rename = "$value")]
773//! any_name: Choice,
774//! }
775//! # #[derive(Debug, PartialEq)]
776//! #[derive(Deserialize)]
777//! struct AnyName {
778//! #[serde(rename = "@field")]
779//! field: T,
780//!
781//! choice: Holder,
782//! }
783//! # assert_eq!(
784//! # AnyName { field: (), choice: Holder { any_name: Choice::One } },
785//! # quick_xml::de::from_str(r#"<any-tag field="..."><choice><one>...</one></choice></any-tag>"#).unwrap(),
786//! # );
787//! # assert_eq!(
788//! # AnyName { field: (), choice: Holder { any_name: Choice::Two } },
789//! # quick_xml::de::from_str(r#"<any-tag field="..."><choice><two>...</two></choice></any-tag>"#).unwrap(),
790//! # );
791//! ```
792//! </td>
793//! </tr>
794//! <!-- 12 ==================================================================================== -->
795//! <tr>
796//! <td>
797//!
798//! `<xs:choice>` encapsulated in other element with a fixed name:
799//!
800//! ```xml
801//! <any-tag>
802//! <field>...</field>
803//! <choice>
804//! <one>...</one>
805//! </choice>
806//! </any-tag>
807//! ```
808//! ```xml
809//! <any-tag>
810//! <choice>
811//! <two>...</two>
812//! </choice>
813//! <field>...</field>
814//! </any-tag>
815//! ```
816//! </td>
817//! <td>
818//!
819//! A structure with a field of an intermediate type with one field of `enum` type.
820//! Actually, this example is not necessary, because you can construct it by yourself
821//! using the composition rules that were described above. However the XML construction
822//! described here is very common, so it is shown explicitly.
823//!
824//! Names of the enum and struct does not matter:
825//!
826//! ```
827//! # use pretty_assertions::assert_eq;
828//! # use serde::Deserialize;
829//! # type T = ();
830//! # #[derive(Debug, PartialEq)]
831//! #[derive(Deserialize)]
832//! #[serde(rename_all = "snake_case")]
833//! enum Choice {
834//! One,
835//! Two,
836//! }
837//! # #[derive(Debug, PartialEq)]
838//! #[derive(Deserialize)]
839//! struct Holder {
840//! #[serde(rename = "$value")]
841//! any_name: Choice,
842//! }
843//! # #[derive(Debug, PartialEq)]
844//! #[derive(Deserialize)]
845//! struct AnyName {
846//! field: T,
847//!
848//! choice: Holder,
849//! }
850//! # assert_eq!(
851//! # AnyName { field: (), choice: Holder { any_name: Choice::One } },
852//! # quick_xml::de::from_str(r#"<any-tag><field>...</field><choice><one>...</one></choice></any-tag>"#).unwrap(),
853//! # );
854//! # assert_eq!(
855//! # AnyName { field: (), choice: Holder { any_name: Choice::Two } },
856//! # quick_xml::de::from_str(r#"<any-tag><choice><two>...</two></choice><field>...</field></any-tag>"#).unwrap(),
857//! # );
858//! ```
859//! </td>
860//! </tr>
861//! <!-- ======================================================================================== -->
862//! <tr><th colspan="2">
863//!
864//! ## Sequences (`xs:all` and `xs:sequence` XML Schema types)
865//!
866//! </th></tr>
867//! <tr><th>To parse all these XML's...</th><th>...use these Rust type(s)</th></tr>
868//! <!-- 13 ==================================================================================== -->
869//! <tr>
870//! <td>
871//! A sequence inside of a tag without a dedicated name:
872//!
873//! ```xml
874//! <any-tag/>
875//! ```
876//! ```xml
877//! <any-tag>
878//! <item/>
879//! </any-tag>
880//! ```
881//! ```xml
882//! <any-tag>
883//! <item/>
884//! <item/>
885//! <item/>
886//! </any-tag>
887//! ```
888//! </td>
889//! <td>
890//!
891//! A structure with a field which is a sequence type, for example, [`Vec`].
892//! Because XML syntax does not distinguish between empty sequences and missed
893//! elements, we should indicate that on the Rust side, because serde will require
894//! that field `item` exists. You can do that in two possible ways:
895//!
896//! Use the `#[serde(default)]` attribute for a [field] or the entire [struct]:
897//! ```
898//! # use pretty_assertions::assert_eq;
899//! # use serde::Deserialize;
900//! # type Item = ();
901//! # #[derive(Debug, PartialEq)]
902//! #[derive(Deserialize)]
903//! struct AnyName {
904//! #[serde(default)]
905//! item: Vec<Item>,
906//! }
907//! # assert_eq!(
908//! # AnyName { item: vec![] },
909//! # quick_xml::de::from_str(r#"<any-tag/>"#).unwrap(),
910//! # );
911//! # assert_eq!(
912//! # AnyName { item: vec![()] },
913//! # quick_xml::de::from_str(r#"<any-tag><item/></any-tag>"#).unwrap(),
914//! # );
915//! # assert_eq!(
916//! # AnyName { item: vec![(), (), ()] },
917//! # quick_xml::de::from_str(r#"<any-tag><item/><item/><item/></any-tag>"#).unwrap(),
918//! # );
919//! ```
920//!
921//! Use the [`Option`]. In that case inner array will always contains at least one
922//! element after deserialization:
923//! ```ignore
924//! # use pretty_assertions::assert_eq;
925//! # use serde::Deserialize;
926//! # type Item = ();
927//! # #[derive(Debug, PartialEq)]
928//! #[derive(Deserialize)]
929//! struct AnyName {
930//! item: Option<Vec<Item>>,
931//! }
932//! # assert_eq!(
933//! # AnyName { item: None },
934//! # quick_xml::de::from_str(r#"<any-tag/>"#).unwrap(),
935//! # );
936//! # assert_eq!(
937//! # AnyName { item: Some(vec![()]) },
938//! # quick_xml::de::from_str(r#"<any-tag><item/></any-tag>"#).unwrap(),
939//! # );
940//! # assert_eq!(
941//! # AnyName { item: Some(vec![(), (), ()]) },
942//! # quick_xml::de::from_str(r#"<any-tag><item/><item/><item/></any-tag>"#).unwrap(),
943//! # );
944//! ```
945//!
946//! See also [Frequently Used Patterns](#element-lists).
947//!
948//! [field]: https://serde.rs/field-attrs.html#default
949//! [struct]: https://serde.rs/container-attrs.html#default
950//! </td>
951//! </tr>
952//! <!-- 14 ==================================================================================== -->
953//! <tr>
954//! <td>
955//! A sequence with a strict order, probably with mixed content
956//! (text / CDATA and tags):
957//!
958//! ```xml
959//! <one>...</one>
960//! text
961//! <![CDATA[cdata]]>
962//! <two>...</two>
963//! <one>...</one>
964//! ```
965//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
966//!
967//! NOTE: this is just an example for showing mapping. XML does not allow
968//! multiple root tags -- you should wrap the sequence into a tag.
969//! </div>
970//! </td>
971//! <td>
972//!
973//! All elements mapped to the heterogeneous sequential type: tuple or named tuple.
974//! Each element of the tuple should be able to be deserialized from the nested
975//! element content (`...`), except the enum types which would be deserialized
976//! from the full element (`<one>...</one>`), so they could use the element name
977//! to choose the right variant:
978//!
979//! ```
980//! # use pretty_assertions::assert_eq;
981//! # use serde::Deserialize;
982//! # type One = ();
983//! # type Two = ();
984//! # /*
985//! type One = ...;
986//! type Two = ...;
987//! # */
988//! # #[derive(Debug, PartialEq)]
989//! #[derive(Deserialize)]
990//! struct AnyName(One, String, Two, One);
991//! # assert_eq!(
992//! # AnyName((), "text cdata".into(), (), ()),
993//! # quick_xml::de::from_str(r#"<one>...</one>text <![CDATA[cdata]]><two>...</two><one>...</one>"#).unwrap(),
994//! # );
995//! ```
996//! ```
997//! # use pretty_assertions::assert_eq;
998//! # use serde::Deserialize;
999//! # #[derive(Debug, PartialEq)]
1000//! #[derive(Deserialize)]
1001//! #[serde(rename_all = "snake_case")]
1002//! enum Choice {
1003//! One,
1004//! }
1005//! # type Two = ();
1006//! # /*
1007//! type Two = ...;
1008//! # */
1009//! type AnyName = (Choice, String, Two, Choice);
1010//! # assert_eq!(
1011//! # (Choice::One, "text cdata".to_string(), (), Choice::One),
1012//! # quick_xml::de::from_str(r#"<one>...</one>text <![CDATA[cdata]]><two>...</two><one>...</one>"#).unwrap(),
1013//! # );
1014//! ```
1015//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
1016//!
1017//! NOTE: consequent text and CDATA nodes are merged into the one text node,
1018//! so you cannot have two adjacent string types in your sequence.
1019//!
1020//! NOTE: In the case that the list might contain tags that are overlapped with
1021//! tags that do not correspond to the list you should add the feature [`overlapped-lists`].
1022//! </div>
1023//! </td>
1024//! </tr>
1025//! <!-- 15 ==================================================================================== -->
1026//! <tr>
1027//! <td>
1028//! A sequence with a non-strict order, probably with a mixed content
1029//! (text / CDATA and tags).
1030//!
1031//! ```xml
1032//! <one>...</one>
1033//! text
1034//! <![CDATA[cdata]]>
1035//! <two>...</two>
1036//! <one>...</one>
1037//! ```
1038//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
1039//!
1040//! NOTE: this is just an example for showing mapping. XML does not allow
1041//! multiple root tags -- you should wrap the sequence into a tag.
1042//! </div>
1043//! </td>
1044//! <td>
1045//! A homogeneous sequence of elements with a fixed or dynamic size:
1046//!
1047//! ```
1048//! # use pretty_assertions::assert_eq;
1049//! # use serde::Deserialize;
1050//! # #[derive(Debug, PartialEq)]
1051//! #[derive(Deserialize)]
1052//! #[serde(rename_all = "snake_case")]
1053//! enum Choice {
1054//! One,
1055//! Two,
1056//! #[serde(other)]
1057//! Other,
1058//! }
1059//! type AnyName = [Choice; 4];
1060//! # assert_eq!(
1061//! # [Choice::One, Choice::Other, Choice::Two, Choice::One],
1062//! # quick_xml::de::from_str::<AnyName>(r#"<one>...</one>text <![CDATA[cdata]]><two>...</two><one>...</one>"#).unwrap(),
1063//! # );
1064//! ```
1065//! ```
1066//! # use pretty_assertions::assert_eq;
1067//! # use serde::Deserialize;
1068//! # #[derive(Debug, PartialEq)]
1069//! #[derive(Deserialize)]
1070//! #[serde(rename_all = "snake_case")]
1071//! enum Choice {
1072//! One,
1073//! Two,
1074//! #[serde(rename = "$text")]
1075//! Other(String),
1076//! }
1077//! type AnyName = Vec<Choice>;
1078//! # assert_eq!(
1079//! # vec![
1080//! # Choice::One,
1081//! # Choice::Other("text cdata".into()),
1082//! # Choice::Two,
1083//! # Choice::One,
1084//! # ],
1085//! # quick_xml::de::from_str::<AnyName>(r#"<one>...</one>text <![CDATA[cdata]]><two>...</two><one>...</one>"#).unwrap(),
1086//! # );
1087//! ```
1088//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
1089//!
1090//! NOTE: consequent text and CDATA nodes are merged into the one text node,
1091//! so you cannot have two adjacent string types in your sequence.
1092//! </div>
1093//! </td>
1094//! </tr>
1095//! <!-- 16 ==================================================================================== -->
1096//! <tr>
1097//! <td>
1098//! A sequence with a strict order, probably with a mixed content,
1099//! (text and tags) inside of the other element:
1100//!
1101//! ```xml
1102//! <any-tag attribute="...">
1103//! <one>...</one>
1104//! text
1105//! <![CDATA[cdata]]>
1106//! <two>...</two>
1107//! <one>...</one>
1108//! </any-tag>
1109//! ```
1110//! </td>
1111//! <td>
1112//!
1113//! A structure where all child elements mapped to the one field which have
1114//! a heterogeneous sequential type: tuple or named tuple. Each element of the
1115//! tuple should be able to be deserialized from the full element (`<one>...</one>`).
1116//!
1117//! You MUST specify `#[serde(rename = "$value")]` on that field:
1118//!
1119//! ```
1120//! # use pretty_assertions::assert_eq;
1121//! # use serde::Deserialize;
1122//! # type One = ();
1123//! # type Two = ();
1124//! # /*
1125//! type One = ...;
1126//! type Two = ...;
1127//! # */
1128//!
1129//! # #[derive(Debug, PartialEq)]
1130//! #[derive(Deserialize)]
1131//! struct AnyName {
1132//! #[serde(rename = "@attribute")]
1133//! # attribute: (),
1134//! # /*
1135//! attribute: ...,
1136//! # */
1137//! // Does not (yet?) supported by the serde
1138//! // https://github.com/serde-rs/serde/issues/1905
1139//! // #[serde(flatten)]
1140//! #[serde(rename = "$value")]
1141//! any_name: (One, String, Two, One),
1142//! }
1143//! # assert_eq!(
1144//! # AnyName { attribute: (), any_name: ((), "text cdata".into(), (), ()) },
1145//! # quick_xml::de::from_str("\
1146//! # <any-tag attribute='...'>\
1147//! # <one>...</one>\
1148//! # text \
1149//! # <![CDATA[cdata]]>\
1150//! # <two>...</two>\
1151//! # <one>...</one>\
1152//! # </any-tag>"
1153//! # ).unwrap(),
1154//! # );
1155//! ```
1156//! ```
1157//! # use pretty_assertions::assert_eq;
1158//! # use serde::Deserialize;
1159//! # type One = ();
1160//! # type Two = ();
1161//! # /*
1162//! type One = ...;
1163//! type Two = ...;
1164//! # */
1165//!
1166//! # #[derive(Debug, PartialEq)]
1167//! #[derive(Deserialize)]
1168//! struct NamedTuple(One, String, Two, One);
1169//!
1170//! # #[derive(Debug, PartialEq)]
1171//! #[derive(Deserialize)]
1172//! struct AnyName {
1173//! #[serde(rename = "@attribute")]
1174//! # attribute: (),
1175//! # /*
1176//! attribute: ...,
1177//! # */
1178//! // Does not (yet?) supported by the serde
1179//! // https://github.com/serde-rs/serde/issues/1905
1180//! // #[serde(flatten)]
1181//! #[serde(rename = "$value")]
1182//! any_name: NamedTuple,
1183//! }
1184//! # assert_eq!(
1185//! # AnyName { attribute: (), any_name: NamedTuple((), "text cdata".into(), (), ()) },
1186//! # quick_xml::de::from_str("\
1187//! # <any-tag attribute='...'>\
1188//! # <one>...</one>\
1189//! # text \
1190//! # <![CDATA[cdata]]>\
1191//! # <two>...</two>\
1192//! # <one>...</one>\
1193//! # </any-tag>"
1194//! # ).unwrap(),
1195//! # );
1196//! ```
1197//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
1198//!
1199//! NOTE: consequent text and CDATA nodes are merged into the one text node,
1200//! so you cannot have two adjacent string types in your sequence.
1201//! </div>
1202//! </td>
1203//! </tr>
1204//! <!-- 17 ==================================================================================== -->
1205//! <tr>
1206//! <td>
1207//! A sequence with a non-strict order, probably with a mixed content
1208//! (text / CDATA and tags) inside of the other element:
1209//!
1210//! ```xml
1211//! <any-tag>
1212//! <one>...</one>
1213//! text
1214//! <![CDATA[cdata]]>
1215//! <two>...</two>
1216//! <one>...</one>
1217//! </any-tag>
1218//! ```
1219//! </td>
1220//! <td>
1221//!
1222//! A structure where all child elements mapped to the one field which have
1223//! a homogeneous sequential type: array-like container. A container type `T`
1224//! should be able to be deserialized from the nested element content (`...`),
1225//! except if it is an enum type which would be deserialized from the full
1226//! element (`<one>...</one>`).
1227//!
1228//! You MUST specify `#[serde(rename = "$value")]` on that field:
1229//!
1230//! ```
1231//! # use pretty_assertions::assert_eq;
1232//! # use serde::Deserialize;
1233//! # #[derive(Debug, PartialEq)]
1234//! #[derive(Deserialize)]
1235//! #[serde(rename_all = "snake_case")]
1236//! enum Choice {
1237//! One,
1238//! Two,
1239//! #[serde(rename = "$text")]
1240//! Other(String),
1241//! }
1242//! # #[derive(Debug, PartialEq)]
1243//! #[derive(Deserialize)]
1244//! struct AnyName {
1245//! #[serde(rename = "@attribute")]
1246//! # attribute: (),
1247//! # /*
1248//! attribute: ...,
1249//! # */
1250//! // Does not (yet?) supported by the serde
1251//! // https://github.com/serde-rs/serde/issues/1905
1252//! // #[serde(flatten)]
1253//! #[serde(rename = "$value")]
1254//! any_name: [Choice; 4],
1255//! }
1256//! # assert_eq!(
1257//! # AnyName { attribute: (), any_name: [
1258//! # Choice::One,
1259//! # Choice::Other("text cdata".into()),
1260//! # Choice::Two,
1261//! # Choice::One,
1262//! # ] },
1263//! # quick_xml::de::from_str("\
1264//! # <any-tag attribute='...'>\
1265//! # <one>...</one>\
1266//! # text \
1267//! # <![CDATA[cdata]]>\
1268//! # <two>...</two>\
1269//! # <one>...</one>\
1270//! # </any-tag>"
1271//! # ).unwrap(),
1272//! # );
1273//! ```
1274//! ```
1275//! # use pretty_assertions::assert_eq;
1276//! # use serde::Deserialize;
1277//! # #[derive(Debug, PartialEq)]
1278//! #[derive(Deserialize)]
1279//! #[serde(rename_all = "snake_case")]
1280//! enum Choice {
1281//! One,
1282//! Two,
1283//! #[serde(rename = "$text")]
1284//! Other(String),
1285//! }
1286//! # #[derive(Debug, PartialEq)]
1287//! #[derive(Deserialize)]
1288//! struct AnyName {
1289//! #[serde(rename = "@attribute")]
1290//! # attribute: (),
1291//! # /*
1292//! attribute: ...,
1293//! # */
1294//! // Does not (yet?) supported by the serde
1295//! // https://github.com/serde-rs/serde/issues/1905
1296//! // #[serde(flatten)]
1297//! #[serde(rename = "$value")]
1298//! any_name: Vec<Choice>,
1299//! }
1300//! # assert_eq!(
1301//! # AnyName { attribute: (), any_name: vec![
1302//! # Choice::One,
1303//! # Choice::Other("text cdata".into()),
1304//! # Choice::Two,
1305//! # Choice::One,
1306//! # ] },
1307//! # quick_xml::de::from_str("\
1308//! # <any-tag attribute='...'>\
1309//! # <one>...</one>\
1310//! # text \
1311//! # <![CDATA[cdata]]>\
1312//! # <two>...</two>\
1313//! # <one>...</one>\
1314//! # </any-tag>"
1315//! # ).unwrap(),
1316//! # );
1317//! ```
1318//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
1319//!
1320//! NOTE: consequent text and CDATA nodes are merged into the one text node,
1321//! so you cannot have two adjacent string types in your sequence.
1322//! </div>
1323//! </td>
1324//! </tr>
1325//! </tbody>
1326//! </table>
1327//!
1328//!
1329//! Mapping of `xsi:nil`
1330//! ====================
1331//!
1332//! quick-xml supports handling of [`xsi:nil`] special attribute. When field of optional
1333//! type is mapped to the XML element which have `xsi:nil="true"` set, or if that attribute
1334//! is placed on parent XML element, the deserializer will call [`Visitor::visit_none`]
1335//! and skip XML element corresponding to a field.
1336//!
1337//! Examples:
1338//!
1339//! ```
1340//! # use pretty_assertions::assert_eq;
1341//! # use serde::Deserialize;
1342//! #[derive(Deserialize, Debug, PartialEq)]
1343//! struct TypeWithOptionalField {
1344//! element: Option<String>,
1345//! }
1346//!
1347//! assert_eq!(
1348//! TypeWithOptionalField {
1349//! element: None,
1350//! },
1351//! quick_xml::de::from_str("
1352//! <any-tag xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance'>
1353//! <element xsi:nil='true'>Content is skiped because of xsi:nil='true'</element>
1354//! </any-tag>
1355//! ").unwrap(),
1356//! );
1357//! ```
1358//!
1359//! You can capture attributes from the optional type, because ` xsi:nil="true"` elements can have
1360//! attributes:
1361//! ```
1362//! # use pretty_assertions::assert_eq;
1363//! # use serde::Deserialize;
1364//! #[derive(Deserialize, Debug, PartialEq)]
1365//! struct TypeWithOptionalField {
1366//! #[serde(rename = "@attribute")]
1367//! attribute: usize,
1368//!
1369//! element: Option<String>,
1370//! non_optional: String,
1371//! }
1372//!
1373//! assert_eq!(
1374//! TypeWithOptionalField {
1375//! attribute: 42,
1376//! element: None,
1377//! non_optional: "Note, that non-optional fields will be deserialized as usual".to_string(),
1378//! },
1379//! quick_xml::de::from_str("
1380//! <any-tag attribute='42' xsi:nil='true' xmlns:xsi='http://www.w3.org/2001/XMLSchema-instance'>
1381//! <element>Content is skiped because of xsi:nil='true'</element>
1382//! <non_optional>Note, that non-optional fields will be deserialized as usual</non_optional>
1383//! </any-tag>
1384//! ").unwrap(),
1385//! );
1386//! ```
1387//!
1388//! Generate Rust types from XML
1389//! ============================
1390//!
1391//! To speed up the creation of Rust types that represent a given XML file you can
1392//! use the [xml_schema_generator](https://github.com/Thomblin/xml_schema_generator).
1393//! It provides a standalone binary and a Rust library that parses one or more XML files
1394//! and generates a collection of structs that are compatible with quick_xml::de.
1395//!
1396//!
1397//!
1398//! Composition Rules
1399//! =================
1400//!
1401//! The XML format is very different from other formats supported by `serde`.
1402//! One such difference it is how data in the serialized form is related to
1403//! the Rust type. Usually each byte in the data can be associated only with
1404//! one field in the data structure. However, XML is an exception.
1405//!
1406//! For example, took this XML:
1407//!
1408//! ```xml
1409//! <any>
1410//! <key attr="value"/>
1411//! </any>
1412//! ```
1413//!
1414//! and try to deserialize it to the struct `AnyName`:
1415//!
1416//! ```no_run
1417//! # use serde::Deserialize;
1418//! #[derive(Deserialize)]
1419//! struct AnyName { // AnyName calls `deserialize_struct` on `<any><key attr="value"/></any>`
1420//! // Used data: ^^^^^^^^^^^^^^^^^^^
1421//! key: Inner, // Inner calls `deserialize_struct` on `<key attr="value"/>`
1422//! // Used data: ^^^^^^^^^^^^
1423//! }
1424//! #[derive(Deserialize)]
1425//! struct Inner {
1426//! #[serde(rename = "@attr")]
1427//! attr: String, // String calls `deserialize_string` on `value`
1428//! // Used data: ^^^^^
1429//! }
1430//! ```
1431//!
1432//! Comments shows what methods of a [`Deserializer`] called by each struct
1433//! `deserialize` method and which input their seen. **Used data** shows, what
1434//! content is actually used for deserializing. As you see, name of the inner
1435//! `<key>` tag used both as a map key / outer struct field name and as part
1436//! of the inner struct (although _value_ of the tag, i.e. `key` is not used
1437//! by it).
1438//!
1439//!
1440//!
1441//! Enum Representations
1442//! ====================
1443//!
1444//! `quick-xml` represents enums differently in normal fields, `$text` fields and
1445//! `$value` fields. A normal representation is compatible with serde's adjacent
1446//! and internal tags feature -- tag for adjacently and internally tagged enums
1447//! are serialized using [`Serializer::serialize_unit_variant`] and deserialized
1448//! using [`Deserializer::deserialize_enum`].
1449//!
1450//! Use those simple rules to remember, how enum would be represented in XML:
1451//! - In `$value` field the representation is always the same as top-level representation;
1452//! - In `$text` field the representation is always the same as in normal field,
1453//! but surrounding tags with field name are removed;
1454//! - In normal field the representation is always contains a tag with a field name.
1455//!
1456//! Normal enum variant
1457//! -------------------
1458//!
1459//! To model an `xs:choice` XML construct use `$value` field.
1460//! To model a top-level `xs:choice` just use the enum type.
1461//!
1462//! |Kind |Top-level and in `$value` field |In normal field |In `$text` field |
1463//! |-------|-----------------------------------------|---------------------|---------------------|
1464//! |Unit |`<Unit/>` |`<field>Unit</field>`|`Unit` |
1465//! |Newtype|`<Newtype>42</Newtype>` |Err(Custom) [^0] |Err(Custom) [^0] |
1466//! |Tuple |`<Tuple>42</Tuple><Tuple>answer</Tuple>` |Err(Custom) [^0] |Err(Custom) [^0] |
1467//! |Struct |`<Struct><q>42</q><a>answer</a></Struct>`|Err(Custom) [^0] |Err(Custom) [^0] |
1468//!
1469//! `$text` enum variant
1470//! --------------------
1471//!
1472//! |Kind |Top-level and in `$value` field |In normal field |In `$text` field |
1473//! |-------|-----------------------------------------|---------------------|---------------------|
1474//! |Unit |_(empty)_ |`<field/>` |_(empty)_ |
1475//! |Newtype|`42` |Err(Custom) [^0] [^1]|Err(Custom) [^0] [^2]|
1476//! |Tuple |`42 answer` |Err(Custom) [^0] [^3]|Err(Custom) [^0] [^4]|
1477//! |Struct |Err(Custom) [^0] |Err(Custom) [^0] |Err(Custom) [^0] |
1478//!
1479//! [^0]: Error is returned by the deserialized type. In case of derived implementation a `Custom`
1480//! error will be returned, but custom deserialize implementation can successfully deserialize
1481//! value from a string which will be passed to it.
1482//!
1483//! [^1]: If this serialize as `<field>42</field>` then it will be ambiguity during deserialization,
1484//! because it clash with `Unit` representation in normal field.
1485//!
1486//! [^2]: If this serialize as `42` then it will be ambiguity during deserialization,
1487//! because it clash with `Unit` representation in `$text` field.
1488//!
1489//! [^3]: If this serialize as `<field>42 answer</field>` then it will be ambiguity during deserialization,
1490//! because it clash with `Unit` representation in normal field.
1491//!
1492//! [^4]: If this serialize as `42 answer` then it will be ambiguity during deserialization,
1493//! because it clash with `Unit` representation in `$text` field.
1494//!
1495//!
1496//!
1497//! `$text` and `$value` special names
1498//! ==================================
1499//!
1500//! quick-xml supports two special names for fields -- `$text` and `$value`.
1501//! Although they may seem the same, there is a distinction. Two different
1502//! names is required mostly for serialization, because quick-xml should know
1503//! how you want to serialize certain constructs, which could be represented
1504//! through XML in multiple different ways.
1505//!
1506//! The only difference is in how complex types and sequences are serialized.
1507//! If you doubt which one you should select, begin with [`$value`](#value).
1508//!
1509//! If you have both `$text` and `$value` in you struct, then text events will be
1510//! mapped to the `$text` field:
1511//!
1512//! ```
1513//! # use serde::Deserialize;
1514//! # use quick_xml::de::from_str;
1515//! #[derive(Deserialize, PartialEq, Debug)]
1516//! struct TextAndValue {
1517//! #[serde(rename = "$text")]
1518//! text: Option<String>,
1519//!
1520//! #[serde(rename = "$value")]
1521//! value: Option<String>,
1522//! }
1523//!
1524//! let object: TextAndValue = from_str("<AnyName>text <![CDATA[and CDATA]]></AnyName>").unwrap();
1525//! assert_eq!(object, TextAndValue {
1526//! text: Some("text and CDATA".to_string()),
1527//! value: None,
1528//! });
1529//! ```
1530//!
1531//! ## `$text`
1532//! `$text` is used when you want to write your XML as a text or a CDATA content.
1533//! More formally, field with that name represents simple type definition with
1534//! `{variety} = atomic` or `{variety} = union` whose basic members are all atomic,
1535//! as described in the [specification].
1536//!
1537//! As a result, not all types of such fields can be serialized. Only serialization
1538//! of following types are supported:
1539//! - all primitive types (strings, numbers, booleans)
1540//! - unit variants of enumerations (serializes to a name of a variant)
1541//! - newtypes (delegates serialization to inner type)
1542//! - [`Option`] of above (`None` serializes to nothing)
1543//! - sequences (including tuples and tuple variants of enumerations) of above,
1544//! excluding `None` and empty string elements (because it will not be possible
1545//! to deserialize them back). The elements are separated by space(s)
1546//! - unit type `()` and unit structs (serializes to nothing)
1547//!
1548//! Complex types, such as structs and maps, are not supported in this field.
1549//! If you want them, you should use `$value`.
1550//!
1551//! Sequences serialized to a space-delimited string, that is why only certain
1552//! types are allowed in this mode:
1553//!
1554//! ```
1555//! # use serde::{Deserialize, Serialize};
1556//! # use quick_xml::de::from_str;
1557//! # use quick_xml::se::to_string;
1558//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1559//! struct AnyName {
1560//! #[serde(rename = "$text")]
1561//! field: Vec<usize>,
1562//! }
1563//!
1564//! let obj = AnyName { field: vec![1, 2, 3] };
1565//! let xml = to_string(&obj).unwrap();
1566//! assert_eq!(xml, "<AnyName>1 2 3</AnyName>");
1567//!
1568//! let object: AnyName = from_str(&xml).unwrap();
1569//! assert_eq!(object, obj);
1570//! ```
1571//!
1572//! ## `$value`
1573//! <div style="background:rgba(120,145,255,0.45);padding:0.75em;">
1574//!
1575//! NOTE: a name `#content` would better explain the purpose of that field,
1576//! but `$value` is used for compatibility with other XML serde crates, which
1577//! uses that name. This will allow you to switch XML crates more smoothly if required.
1578//! </div>
1579//!
1580//! The representation of primitive types in `$value` does not differ from their
1581//! representation in `$text` fields. The difference is how sequences are serialized
1582//! and deserialized. `$value` serializes each sequence item as a separate XML element.
1583//! How the name of the XML element is chosen depends on the field's type. For
1584//! `enum`s, the variant name is used. For `struct`s, the name of the `struct`
1585//! is used.
1586//!
1587//! During deserialization, if the `$value` field is an enum, then the variant's
1588//! name is matched against. That's **not** the case with structs, however, since
1589//! `serde` does not expose type names of nested fields. This does mean that **any**
1590//! type could be deserialized into a `$value` struct-type field, so long as the
1591//! struct's fields have compatible types (or are captured as text by `String`
1592//! or similar-behaving types). This can be handy when using generic types in fields
1593//! where one knows in advance what to expect. If you do not know what to expect,
1594//! however, prefer an enum with all possible variants.
1595//!
1596//! Unit structs and unit type `()` serialize to nothing and can be deserialized
1597//! from any content.
1598//!
1599//! Serialization and deserialization of `$value` field performed as usual, except
1600//! that name for an XML element will be given by the serialized type, instead of
1601//! field. The latter allow to serialize enumerated types, where variant is encoded
1602//! as a tag name, and, so, represent an XSD `xs:choice` schema by the Rust `enum`.
1603//!
1604//! In the example below, field will be serialized as `<field/>`, because elements
1605//! get their names from the field name. It cannot be deserialized, because `Enum`
1606//! expects elements `<A/>`, `<B/>` or `<C/>`, but `AnyName` looked only for `<field/>`:
1607//!
1608//! ```
1609//! # use serde::{Deserialize, Serialize};
1610//! # use pretty_assertions::assert_eq;
1611//! # #[derive(PartialEq, Debug)]
1612//! #[derive(Deserialize, Serialize)]
1613//! enum Enum { A, B, C }
1614//!
1615//! # #[derive(PartialEq, Debug)]
1616//! #[derive(Deserialize, Serialize)]
1617//! struct AnyName {
1618//! // <field>A</field>, <field>B</field>, or <field>C</field>
1619//! field: Enum,
1620//! }
1621//! # assert_eq!(
1622//! # quick_xml::se::to_string(&AnyName { field: Enum::A }).unwrap(),
1623//! # "<AnyName><field>A</field></AnyName>",
1624//! # );
1625//! # assert_eq!(
1626//! # AnyName { field: Enum::B },
1627//! # quick_xml::de::from_str("<root><field>B</field></root>").unwrap(),
1628//! # );
1629//! ```
1630//!
1631//! If you rename field to `$value`, then `field` would be serialized as `<A/>`,
1632//! `<B/>` or `<C/>`, depending on the its content. It is also possible to
1633//! deserialize it from the same elements:
1634//!
1635//! ```
1636//! # use serde::{Deserialize, Serialize};
1637//! # use pretty_assertions::assert_eq;
1638//! # #[derive(Deserialize, Serialize, PartialEq, Debug)]
1639//! # enum Enum { A, B, C }
1640//! #
1641//! # #[derive(PartialEq, Debug)]
1642//! #[derive(Deserialize, Serialize)]
1643//! struct AnyName {
1644//! // <A/>, <B/> or <C/>
1645//! #[serde(rename = "$value")]
1646//! field: Enum,
1647//! }
1648//! # assert_eq!(
1649//! # quick_xml::se::to_string(&AnyName { field: Enum::A }).unwrap(),
1650//! # "<AnyName><A/></AnyName>",
1651//! # );
1652//! # assert_eq!(
1653//! # AnyName { field: Enum::B },
1654//! # quick_xml::de::from_str("<root><B/></root>").unwrap(),
1655//! # );
1656//! ```
1657//!
1658//! The next example demonstrates how generic types can be used in conjunction
1659//! with `$value`-named fields to allow the reuse of wrapping structs. A common
1660//! example use case for this feature is SOAP messages, which can be commmonly
1661//! found wrapped around `<soapenv:Envelope> ... </soapenv:Envelope>`.
1662//!
1663//! ```rust
1664//! # use pretty_assertions::assert_eq;
1665//! # use quick_xml::de::from_str;
1666//! # use quick_xml::se::to_string;
1667//! # use serde::{Deserialize, Serialize};
1668//! #
1669//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1670//! struct Envelope<T> {
1671//! body: Body<T>,
1672//! }
1673//!
1674//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1675//! struct Body<T> {
1676//! #[serde(rename = "$value")]
1677//! inner: T,
1678//! }
1679//!
1680//! #[derive(Serialize, PartialEq, Debug)]
1681//! struct Example {
1682//! a: i32,
1683//! }
1684//!
1685//! assert_eq!(
1686//! to_string(&Envelope { body: Body { inner: Example { a: 42 } } }).unwrap(),
1687//! // Notice how `inner` is not present in the XML
1688//! "<Envelope><body><Example><a>42</a></Example></body></Envelope>",
1689//! );
1690//!
1691//! #[derive(Deserialize, PartialEq, Debug)]
1692//! struct AnotherExample {
1693//! a: i32,
1694//! }
1695//!
1696//! assert_eq!(
1697//! // Notice that tag the name does nothing for struct in `$value` field
1698//! Envelope { body: Body { inner: AnotherExample { a: 42 } } },
1699//! from_str("<Envelope><body><Example><a>42</a></Example></body></Envelope>").unwrap(),
1700//! );
1701//! ```
1702//!
1703//! ### Primitives and sequences of primitives
1704//!
1705//! Sequences serialized to a list of elements. Note, that types that does not
1706//! produce their own tag (i. e. primitives) will produce [`SeError::Unsupported`]
1707//! if they contains more that one element, because such sequence cannot be
1708//! deserialized to the same value:
1709//!
1710//! ```
1711//! # use serde::{Deserialize, Serialize};
1712//! # use pretty_assertions::assert_eq;
1713//! # use quick_xml::de::from_str;
1714//! # use quick_xml::se::to_string;
1715//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1716//! struct AnyName {
1717//! #[serde(rename = "$value")]
1718//! field: Vec<usize>,
1719//! }
1720//!
1721//! let obj = AnyName { field: vec![1, 2, 3] };
1722//! // If this object were serialized, it would be represented as "<AnyName>123</AnyName>"
1723//! to_string(&obj).unwrap_err();
1724//!
1725//! let object: AnyName = from_str("<AnyName>123</AnyName>").unwrap();
1726//! assert_eq!(object, AnyName { field: vec![123] });
1727//!
1728//! // `1 2 3` is mapped to a single `usize` element
1729//! // It is impossible to deserialize list of primitives to such field
1730//! from_str::<AnyName>("<AnyName>1 2 3</AnyName>").unwrap_err();
1731//! ```
1732//!
1733//! A particular case of that example is a string `$value` field, which probably
1734//! would be a most used example of that attribute:
1735//!
1736//! ```
1737//! # use serde::{Deserialize, Serialize};
1738//! # use pretty_assertions::assert_eq;
1739//! # use quick_xml::de::from_str;
1740//! # use quick_xml::se::to_string;
1741//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1742//! struct AnyName {
1743//! #[serde(rename = "$value")]
1744//! field: String,
1745//! }
1746//!
1747//! let obj = AnyName { field: "content".to_string() };
1748//! let xml = to_string(&obj).unwrap();
1749//! assert_eq!(xml, "<AnyName>content</AnyName>");
1750//! ```
1751//!
1752//! ### Structs and sequences of structs
1753//!
1754//! Note, that structures do not have a serializable name as well (name of the
1755//! type is never used), so it is impossible to serialize non-unit struct or
1756//! sequence of non-unit structs in `$value` field. (sequences of) unit structs
1757//! are serialized as empty string, because units itself serializing
1758//! to nothing:
1759//!
1760//! ```
1761//! # use serde::{Deserialize, Serialize};
1762//! # use pretty_assertions::assert_eq;
1763//! # use quick_xml::de::from_str;
1764//! # use quick_xml::se::to_string;
1765//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1766//! struct Unit;
1767//!
1768//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1769//! struct AnyName {
1770//! // #[serde(default)] is required to deserialization of empty lists
1771//! // This is a general note, not related to $value
1772//! #[serde(rename = "$value", default)]
1773//! field: Vec<Unit>,
1774//! }
1775//!
1776//! let obj = AnyName { field: vec![Unit, Unit, Unit] };
1777//! let xml = to_string(&obj).unwrap();
1778//! assert_eq!(xml, "<AnyName/>");
1779//!
1780//! let object: AnyName = from_str("<AnyName/>").unwrap();
1781//! assert_eq!(object, AnyName { field: vec![] });
1782//!
1783//! let object: AnyName = from_str("<AnyName></AnyName>").unwrap();
1784//! assert_eq!(object, AnyName { field: vec![] });
1785//!
1786//! let object: AnyName = from_str("<AnyName><A/><B/><C/></AnyName>").unwrap();
1787//! assert_eq!(object, AnyName { field: vec![Unit, Unit, Unit] });
1788//! ```
1789//!
1790//! ### Enums and sequences of enums
1791//!
1792//! Enumerations uses the variant name as an element name:
1793//!
1794//! ```
1795//! # use serde::{Deserialize, Serialize};
1796//! # use pretty_assertions::assert_eq;
1797//! # use quick_xml::de::from_str;
1798//! # use quick_xml::se::to_string;
1799//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1800//! struct AnyName {
1801//! #[serde(rename = "$value")]
1802//! field: Vec<Enum>,
1803//! }
1804//!
1805//! #[derive(Deserialize, Serialize, PartialEq, Debug)]
1806//! enum Enum { A, B, C }
1807//!
1808//! let obj = AnyName { field: vec![Enum::A, Enum::B, Enum::C] };
1809//! let xml = to_string(&obj).unwrap();
1810//! assert_eq!(
1811//! xml,
1812//! "<AnyName>\
1813//! <A/>\
1814//! <B/>\
1815//! <C/>\
1816//! </AnyName>"
1817//! );
1818//!
1819//! let object: AnyName = from_str(&xml).unwrap();
1820//! assert_eq!(object, obj);
1821//! ```
1822//!
1823//!
1824//!
1825//! Frequently Used Patterns
1826//! ========================
1827//!
1828//! Some XML constructs used so frequent, that it is worth to document the recommended
1829//! way to represent them in the Rust. The sections below describes them.
1830//!
1831//! `<element>` lists
1832//! -----------------
1833//! Many XML formats wrap lists of elements in the additional container,
1834//! although this is not required by the XML rules:
1835//!
1836//! ```xml
1837//! <root>
1838//! <field1/>
1839//! <field2/>
1840//! <list><!-- Container -->
1841//! <element/>
1842//! <element/>
1843//! <element/>
1844//! </list>
1845//! <field3/>
1846//! </root>
1847//! ```
1848//! In this case, there is a great desire to describe this XML in this way:
1849//! ```
1850//! /// Represents <element/>
1851//! type Element = ();
1852//!
1853//! /// Represents <root>...</root>
1854//! struct AnyName {
1855//! // Incorrect
1856//! list: Vec<Element>,
1857//! }
1858//! ```
1859//! This will not work, because potentially `<list>` element can have attributes
1860//! and other elements inside. You should define the struct for the `<list>`
1861//! explicitly, as you do that in the XSD for that XML:
1862//! ```
1863//! /// Represents <element/>
1864//! type Element = ();
1865//!
1866//! /// Represents <root>...</root>
1867//! struct AnyName {
1868//! // Correct
1869//! list: List,
1870//! }
1871//! /// Represents <list>...</list>
1872//! struct List {
1873//! element: Vec<Element>,
1874//! }
1875//! ```
1876//!
1877//! If you want to simplify your API, you could write a simple function for unwrapping
1878//! inner list and apply it via [`deserialize_with`]:
1879//!
1880//! ```
1881//! # use pretty_assertions::assert_eq;
1882//! use quick_xml::de::from_str;
1883//! use serde::{Deserialize, Deserializer};
1884//!
1885//! /// Represents <element/>
1886//! type Element = ();
1887//!
1888//! /// Represents <root>...</root>
1889//! #[derive(Deserialize, Debug, PartialEq)]
1890//! struct AnyName {
1891//! #[serde(deserialize_with = "unwrap_list")]
1892//! list: Vec<Element>,
1893//! }
1894//!
1895//! fn unwrap_list<'de, D>(deserializer: D) -> Result<Vec<Element>, D::Error>
1896//! where
1897//! D: Deserializer<'de>,
1898//! {
1899//! /// Represents <list>...</list>
1900//! #[derive(Deserialize)]
1901//! struct List {
1902//! // default allows empty list
1903//! #[serde(default)]
1904//! element: Vec<Element>,
1905//! }
1906//! Ok(List::deserialize(deserializer)?.element)
1907//! }
1908//!
1909//! assert_eq!(
1910//! AnyName { list: vec![(), (), ()] },
1911//! from_str("
1912//! <root>
1913//! <list>
1914//! <element/>
1915//! <element/>
1916//! <element/>
1917//! </list>
1918//! </root>
1919//! ").unwrap(),
1920//! );
1921//! ```
1922//!
1923//! Instead of writing such functions manually, you also could try <https://lib.rs/crates/serde-query>.
1924//!
1925//! Overlapped (Out-of-Order) Elements
1926//! ----------------------------------
1927//! In the case that the list might contain tags that are overlapped with
1928//! tags that do not correspond to the list (this is a usual case in XML
1929//! documents) like this:
1930//! ```xml
1931//! <any-name>
1932//! <item/>
1933//! <another-item/>
1934//! <item/>
1935//! <item/>
1936//! </any-name>
1937//! ```
1938//! you should enable the [`overlapped-lists`] feature to make it possible
1939//! to deserialize this to:
1940//! ```no_run
1941//! # use serde::Deserialize;
1942//! #[derive(Deserialize)]
1943//! #[serde(rename_all = "kebab-case")]
1944//! struct AnyName {
1945//! item: Vec<()>,
1946//! another_item: (),
1947//! }
1948//! ```
1949//!
1950//!
1951//! Internally Tagged Enums
1952//! -----------------------
1953//! [Tagged enums] are currently not supported because of an issue in the Serde
1954//! design (see [serde#1183] and [quick-xml#586]) and missing optimizations in
1955//! Serde which could be useful for XML parsing ([serde#1495]). This can be worked
1956//! around by manually implementing deserialize with `#[serde(deserialize_with = "func")]`
1957//! or implementing [`Deserialize`], but this can get very tedious very fast for
1958//! files with large amounts of tagged enums. To help with this issue quick-xml
1959//! provides a macro [`impl_deserialize_for_internally_tagged_enum!`]. See the
1960//! macro documentation for details.
1961//!
1962//!
1963//! [`overlapped-lists`]: ../index.html#overlapped-lists
1964//! [specification]: https://www.w3.org/TR/xmlschema11-1/#Simple_Type_Definition
1965//! [`deserialize_with`]: https://serde.rs/field-attrs.html#deserialize_with
1966//! [`xsi:nil`]: https://www.w3.org/TR/xmlschema-1/#xsi_nil
1967//! [`Serializer::serialize_unit_variant`]: serde::Serializer::serialize_unit_variant
1968//! [`Deserializer::deserialize_enum`]: serde::Deserializer::deserialize_enum
1969//! [`SeError::Unsupported`]: crate::errors::serialize::SeError::Unsupported
1970//! [Tagged enums]: https://serde.rs/enum-representations.html#internally-tagged
1971//! [serde#1183]: https://github.com/serde-rs/serde/issues/1183
1972//! [serde#1495]: https://github.com/serde-rs/serde/issues/1495
1973//! [quick-xml#586]: https://github.com/tafia/quick-xml/issues/586
1974//! [`impl_deserialize_for_internally_tagged_enum!`]: crate::impl_deserialize_for_internally_tagged_enum
1975
1976// Macros should be defined before the modules that using them
1977// Also, macros should be imported before using them
1978use serde::serde_if_integer128;
1979
1980macro_rules! forward_to_simple_type {
1981 ($deserialize:ident, $($mut:tt)?) => {
1982 #[inline]
1983 fn $deserialize<V>($($mut)? self, visitor: V) -> Result<V::Value, DeError>
1984 where
1985 V: Visitor<'de>,
1986 {
1987 SimpleTypeDeserializer::from_text(self.read_string()?).$deserialize(visitor)
1988 }
1989 };
1990}
1991
1992/// Implement deserialization methods for scalar types, such as numbers, strings,
1993/// byte arrays, booleans and identifiers.
1994macro_rules! deserialize_primitives {
1995 ($($mut:tt)?) => {
1996 forward_to_simple_type!(deserialize_i8, $($mut)?);
1997 forward_to_simple_type!(deserialize_i16, $($mut)?);
1998 forward_to_simple_type!(deserialize_i32, $($mut)?);
1999 forward_to_simple_type!(deserialize_i64, $($mut)?);
2000
2001 forward_to_simple_type!(deserialize_u8, $($mut)?);
2002 forward_to_simple_type!(deserialize_u16, $($mut)?);
2003 forward_to_simple_type!(deserialize_u32, $($mut)?);
2004 forward_to_simple_type!(deserialize_u64, $($mut)?);
2005
2006 serde_if_integer128! {
2007 forward_to_simple_type!(deserialize_i128, $($mut)?);
2008 forward_to_simple_type!(deserialize_u128, $($mut)?);
2009 }
2010
2011 forward_to_simple_type!(deserialize_f32, $($mut)?);
2012 forward_to_simple_type!(deserialize_f64, $($mut)?);
2013
2014 forward_to_simple_type!(deserialize_bool, $($mut)?);
2015 forward_to_simple_type!(deserialize_char, $($mut)?);
2016
2017 forward_to_simple_type!(deserialize_str, $($mut)?);
2018 forward_to_simple_type!(deserialize_string, $($mut)?);
2019
2020 /// Forwards deserialization to the [`deserialize_any`](#method.deserialize_any).
2021 #[inline]
2022 fn deserialize_bytes<V>(self, visitor: V) -> Result<V::Value, DeError>
2023 where
2024 V: Visitor<'de>,
2025 {
2026 self.deserialize_any(visitor)
2027 }
2028
2029 /// Forwards deserialization to the [`deserialize_bytes`](#method.deserialize_bytes).
2030 #[inline]
2031 fn deserialize_byte_buf<V>(self, visitor: V) -> Result<V::Value, DeError>
2032 where
2033 V: Visitor<'de>,
2034 {
2035 self.deserialize_bytes(visitor)
2036 }
2037
2038 /// Representation of the named units the same as [unnamed units](#method.deserialize_unit).
2039 #[inline]
2040 fn deserialize_unit_struct<V>(
2041 self,
2042 _name: &'static str,
2043 visitor: V,
2044 ) -> Result<V::Value, DeError>
2045 where
2046 V: Visitor<'de>,
2047 {
2048 self.deserialize_unit(visitor)
2049 }
2050
2051 /// Representation of tuples the same as [sequences](#method.deserialize_seq).
2052 #[inline]
2053 fn deserialize_tuple<V>(self, _len: usize, visitor: V) -> Result<V::Value, DeError>
2054 where
2055 V: Visitor<'de>,
2056 {
2057 self.deserialize_seq(visitor)
2058 }
2059
2060 /// Representation of named tuples the same as [unnamed tuples](#method.deserialize_tuple).
2061 #[inline]
2062 fn deserialize_tuple_struct<V>(
2063 self,
2064 _name: &'static str,
2065 len: usize,
2066 visitor: V,
2067 ) -> Result<V::Value, DeError>
2068 where
2069 V: Visitor<'de>,
2070 {
2071 self.deserialize_tuple(len, visitor)
2072 }
2073
2074 /// Forwards deserialization to the [`deserialize_struct`](#method.deserialize_struct)
2075 /// with empty name and fields.
2076 #[inline]
2077 fn deserialize_map<V>(self, visitor: V) -> Result<V::Value, DeError>
2078 where
2079 V: Visitor<'de>,
2080 {
2081 self.deserialize_struct("", &[], visitor)
2082 }
2083
2084 /// Identifiers represented as [strings](#method.deserialize_str).
2085 #[inline]
2086 fn deserialize_identifier<V>(self, visitor: V) -> Result<V::Value, DeError>
2087 where
2088 V: Visitor<'de>,
2089 {
2090 self.deserialize_str(visitor)
2091 }
2092
2093 /// Forwards deserialization to the [`deserialize_unit`](#method.deserialize_unit).
2094 #[inline]
2095 fn deserialize_ignored_any<V>(self, visitor: V) -> Result<V::Value, DeError>
2096 where
2097 V: Visitor<'de>,
2098 {
2099 self.deserialize_unit(visitor)
2100 }
2101 };
2102}
2103
2104mod attributes;
2105mod key;
2106mod map;
2107mod resolver;
2108mod simple_type;
2109mod text;
2110mod var;
2111
2112pub use self::attributes::AttributesDeserializer;
2113pub use self::resolver::{EntityResolver, PredefinedEntityResolver};
2114pub use self::simple_type::SimpleTypeDeserializer;
2115pub use crate::errors::serialize::DeError;
2116
2117use crate::{
2118 de::map::ElementMapAccess,
2119 encoding::Decoder,
2120 errors::Error,
2121 escape::{parse_number, EscapeError},
2122 events::{BytesCData, BytesEnd, BytesRef, BytesStart, BytesText, Event},
2123 name::QName,
2124 reader::NsReader,
2125};
2126use serde::de::{
2127 self, Deserialize, DeserializeOwned, DeserializeSeed, IntoDeserializer, SeqAccess, Visitor,
2128};
2129use std::borrow::Cow;
2130#[cfg(feature = "overlapped-lists")]
2131use std::collections::VecDeque;
2132use std::io::BufRead;
2133use std::mem::replace;
2134#[cfg(feature = "overlapped-lists")]
2135use std::num::NonZeroUsize;
2136use std::ops::{Deref, Range};
2137
2138/// Data represented by a text node or a CDATA node. XML markup is not expected
2139pub(crate) const TEXT_KEY: &str = "$text";
2140/// Data represented by any XML markup inside
2141pub(crate) const VALUE_KEY: &str = "$value";
2142
2143/// A function to check whether the character is a whitespace (blank, new line, carriage return or tab).
2144#[inline]
2145const fn is_non_whitespace(ch: char) -> bool {
2146 !matches!(ch, ' ' | '\r' | '\n' | '\t')
2147}
2148
2149/// Decoded and concatenated content of consequent [`Text`] and [`CData`]
2150/// events. _Consequent_ means that events should follow each other or be
2151/// delimited only by (any count of) [`Comment`] or [`PI`] events.
2152///
2153/// Internally text is stored in `Cow<str>`. Cloning of text is cheap while it
2154/// is borrowed and makes copies of data when it is owned.
2155///
2156/// [`Text`]: Event::Text
2157/// [`CData`]: Event::CData
2158/// [`Comment`]: Event::Comment
2159/// [`PI`]: Event::PI
2160#[derive(Clone, Debug, PartialEq, Eq)]
2161pub struct Text<'a> {
2162 /// Untrimmed text after concatenating content of all
2163 /// [`Text`] and [`CData`] events
2164 ///
2165 /// [`Text`]: Event::Text
2166 /// [`CData`]: Event::CData
2167 text: Cow<'a, str>,
2168 /// A range into `text` which contains data after trimming
2169 content: Range<usize>,
2170}
2171
2172impl<'a> Text<'a> {
2173 fn new(text: Cow<'a, str>) -> Self {
2174 let start = text.find(is_non_whitespace).unwrap_or(0);
2175 let end = text.rfind(is_non_whitespace).map_or(0, |i| i + 1);
2176
2177 let content = if start >= end { 0..0 } else { start..end };
2178
2179 Self { text, content }
2180 }
2181
2182 /// Returns text without leading and trailing whitespaces as [defined] by XML specification.
2183 ///
2184 /// If you want to only check if text contains only whitespaces, use [`is_blank`](Self::is_blank),
2185 /// which will not allocate.
2186 ///
2187 /// # Example
2188 ///
2189 /// ```
2190 /// # use quick_xml::de::Text;
2191 /// # use pretty_assertions::assert_eq;
2192 /// #
2193 /// let text = Text::from("");
2194 /// assert_eq!(text.trimmed(), "");
2195 ///
2196 /// let text = Text::from(" \r\n\t ");
2197 /// assert_eq!(text.trimmed(), "");
2198 ///
2199 /// let text = Text::from(" some useful text ");
2200 /// assert_eq!(text.trimmed(), "some useful text");
2201 /// ```
2202 ///
2203 /// [defined]: https://www.w3.org/TR/xml11/#NT-S
2204 pub fn trimmed(&self) -> Cow<'a, str> {
2205 match self.text {
2206 Cow::Borrowed(text) => Cow::Borrowed(&text[self.content.clone()]),
2207 Cow::Owned(ref text) => Cow::Owned(text[self.content.clone()].to_string()),
2208 }
2209 }
2210
2211 /// Returns `true` if text is empty or contains only whitespaces as [defined] by XML specification.
2212 ///
2213 /// # Example
2214 ///
2215 /// ```
2216 /// # use quick_xml::de::Text;
2217 /// # use pretty_assertions::assert_eq;
2218 /// #
2219 /// let text = Text::from("");
2220 /// assert_eq!(text.is_blank(), true);
2221 ///
2222 /// let text = Text::from(" \r\n\t ");
2223 /// assert_eq!(text.is_blank(), true);
2224 ///
2225 /// let text = Text::from(" some useful text ");
2226 /// assert_eq!(text.is_blank(), false);
2227 /// ```
2228 ///
2229 /// [defined]: https://www.w3.org/TR/xml11/#NT-S
2230 pub fn is_blank(&self) -> bool {
2231 self.content.is_empty()
2232 }
2233}
2234
2235impl<'a> Deref for Text<'a> {
2236 type Target = str;
2237
2238 #[inline]
2239 fn deref(&self) -> &Self::Target {
2240 self.text.deref()
2241 }
2242}
2243
2244impl<'a> From<&'a str> for Text<'a> {
2245 #[inline]
2246 fn from(text: &'a str) -> Self {
2247 Self::new(Cow::Borrowed(text))
2248 }
2249}
2250
2251impl<'a> From<String> for Text<'a> {
2252 #[inline]
2253 fn from(text: String) -> Self {
2254 Self::new(Cow::Owned(text))
2255 }
2256}
2257
2258impl<'a> From<Cow<'a, str>> for Text<'a> {
2259 #[inline]
2260 fn from(text: Cow<'a, str>) -> Self {
2261 Self::new(text)
2262 }
2263}
2264
2265////////////////////////////////////////////////////////////////////////////////////////////////////
2266
2267/// Simplified event which contains only these variants that used by deserializer
2268#[derive(Clone, Debug, PartialEq, Eq)]
2269pub enum DeEvent<'a> {
2270 /// Start tag (with attributes) `<tag attr="value">`.
2271 Start(BytesStart<'a>),
2272 /// End tag `</tag>`.
2273 End(BytesEnd<'a>),
2274 /// Decoded and concatenated content of consequent [`Text`] and [`CData`]
2275 /// events. _Consequent_ means that events should follow each other or be
2276 /// delimited only by (any count of) [`Comment`] or [`PI`] events.
2277 ///
2278 /// [`Text`]: Event::Text
2279 /// [`CData`]: Event::CData
2280 /// [`Comment`]: Event::Comment
2281 /// [`PI`]: Event::PI
2282 Text(Text<'a>),
2283 /// End of XML document.
2284 Eof,
2285}
2286
2287////////////////////////////////////////////////////////////////////////////////////////////////////
2288
2289/// Simplified event which contains only these variants that used by deserializer,
2290/// but [`Text`] events not yet fully processed.
2291///
2292/// [`Text`] events should be trimmed if they does not surrounded by the other
2293/// [`Text`] or [`CData`] events. This event contains intermediate state of [`Text`]
2294/// event, where they are trimmed from the start, but not from the end. To trim
2295/// end spaces we should lookahead by one deserializer event (i. e. skip all
2296/// comments and processing instructions).
2297///
2298/// [`Text`]: Event::Text
2299/// [`CData`]: Event::CData
2300#[derive(Clone, Debug, PartialEq, Eq)]
2301pub enum PayloadEvent<'a> {
2302 /// Start tag (with attributes) `<tag attr="value">`.
2303 Start(BytesStart<'a>),
2304 /// End tag `</tag>`.
2305 End(BytesEnd<'a>),
2306 /// Escaped character data between tags.
2307 Text(BytesText<'a>),
2308 /// Unescaped character data stored in `<![CDATA[...]]>`.
2309 CData(BytesCData<'a>),
2310 /// Document type definition data (DTD) stored in `<!DOCTYPE ...>`.
2311 DocType(BytesText<'a>),
2312 /// Reference `&ref;` in the textual data.
2313 GeneralRef(BytesRef<'a>),
2314 /// End of XML document.
2315 Eof,
2316}
2317
2318impl<'a> PayloadEvent<'a> {
2319 /// Ensures that all data is owned to extend the object's lifetime if necessary.
2320 #[inline]
2321 fn into_owned(self) -> PayloadEvent<'static> {
2322 match self {
2323 PayloadEvent::Start(e) => PayloadEvent::Start(e.into_owned()),
2324 PayloadEvent::End(e) => PayloadEvent::End(e.into_owned()),
2325 PayloadEvent::Text(e) => PayloadEvent::Text(e.into_owned()),
2326 PayloadEvent::CData(e) => PayloadEvent::CData(e.into_owned()),
2327 PayloadEvent::DocType(e) => PayloadEvent::DocType(e.into_owned()),
2328 PayloadEvent::GeneralRef(e) => PayloadEvent::GeneralRef(e.into_owned()),
2329 PayloadEvent::Eof => PayloadEvent::Eof,
2330 }
2331 }
2332}
2333
2334/// An intermediate reader that consumes [`PayloadEvent`]s and produces final [`DeEvent`]s.
2335/// [`PayloadEvent::Text`] events, that followed by any event except
2336/// [`PayloadEvent::Text`] or [`PayloadEvent::CData`], are trimmed from the end.
2337struct XmlReader<'i, R: XmlRead<'i>, E: EntityResolver = PredefinedEntityResolver> {
2338 /// A source of low-level XML events
2339 reader: R,
2340 /// Intermediate event, that could be returned by the next call to `next()`.
2341 /// If that is the `Text` event then leading spaces already trimmed, but
2342 /// trailing spaces is not. Before the event will be returned, trimming of
2343 /// the spaces could be necessary
2344 lookahead: Result<PayloadEvent<'i>, DeError>,
2345
2346 /// Used to resolve unknown entities that would otherwise cause the parser
2347 /// to return an [`EscapeError::UnrecognizedEntity`] error.
2348 ///
2349 /// [`EscapeError::UnrecognizedEntity`]: crate::escape::EscapeError::UnrecognizedEntity
2350 entity_resolver: E,
2351}
2352
2353impl<'i, R: XmlRead<'i>, E: EntityResolver> XmlReader<'i, R, E> {
2354 fn new(mut reader: R, entity_resolver: E) -> Self {
2355 // Lookahead by one event immediately, so we do not need to check in the
2356 // loop if we need lookahead or not
2357 let lookahead = reader.next();
2358
2359 Self {
2360 reader,
2361 lookahead,
2362 entity_resolver,
2363 }
2364 }
2365
2366 /// Returns `true` if all events was consumed
2367 const fn is_empty(&self) -> bool {
2368 matches!(self.lookahead, Ok(PayloadEvent::Eof))
2369 }
2370
2371 /// Read next event and put it in lookahead, return the current lookahead
2372 #[inline(always)]
2373 fn next_impl(&mut self) -> Result<PayloadEvent<'i>, DeError> {
2374 replace(&mut self.lookahead, self.reader.next())
2375 }
2376
2377 /// Returns `true` when next event is not a text event in any form.
2378 #[inline(always)]
2379 const fn current_event_is_last_text(&self) -> bool {
2380 // If next event is a text or CDATA, we should not trim trailing spaces
2381 !matches!(
2382 self.lookahead,
2383 Ok(PayloadEvent::Text(_)) | Ok(PayloadEvent::CData(_) | PayloadEvent::GeneralRef(_))
2384 )
2385 }
2386
2387 /// Read all consequent [`Text`] and [`CData`] events until non-text event
2388 /// occurs. Content of all events would be appended to `result` and returned
2389 /// as [`DeEvent::Text`].
2390 ///
2391 /// [`Text`]: PayloadEvent::Text
2392 /// [`CData`]: PayloadEvent::CData
2393 fn drain_text(&mut self, mut result: Cow<'i, str>) -> Result<DeEvent<'i>, DeError> {
2394 loop {
2395 if self.current_event_is_last_text() {
2396 break;
2397 }
2398
2399 match self.next_impl()? {
2400 PayloadEvent::Text(e) => result.to_mut().push_str(&e.xml_content()?),
2401 PayloadEvent::CData(e) => result.to_mut().push_str(&e.xml_content()?),
2402 PayloadEvent::GeneralRef(e) => self.resolve_reference(result.to_mut(), e)?,
2403
2404 // SAFETY: current_event_is_last_text checks that event is Text, CData or GeneralRef
2405 _ => unreachable!("Only `Text`, `CData` or `GeneralRef` events can come here"),
2406 }
2407 }
2408 Ok(DeEvent::Text(Text::new(result)))
2409 }
2410
2411 /// Return an input-borrowing event.
2412 fn next(&mut self) -> Result<DeEvent<'i>, DeError> {
2413 loop {
2414 return match self.next_impl()? {
2415 PayloadEvent::Start(e) => Ok(DeEvent::Start(e)),
2416 PayloadEvent::End(e) => Ok(DeEvent::End(e)),
2417 PayloadEvent::Text(e) => self.drain_text(e.xml_content()?),
2418 PayloadEvent::CData(e) => self.drain_text(e.xml_content()?),
2419 PayloadEvent::DocType(e) => {
2420 self.entity_resolver
2421 .capture(e)
2422 .map_err(|err| DeError::Custom(format!("cannot parse DTD: {}", err)))?;
2423 continue;
2424 }
2425 PayloadEvent::GeneralRef(e) => {
2426 let mut text = String::new();
2427 self.resolve_reference(&mut text, e)?;
2428 self.drain_text(text.into())
2429 }
2430 PayloadEvent::Eof => Ok(DeEvent::Eof),
2431 };
2432 }
2433 }
2434
2435 fn resolve_reference(&mut self, result: &mut String, event: BytesRef) -> Result<(), DeError> {
2436 let len = event.len();
2437 let reference = self.decoder().decode(&event)?;
2438
2439 if let Some(num) = reference.strip_prefix('#') {
2440 let codepoint = parse_number(num).map_err(EscapeError::InvalidCharRef)?;
2441 result.push_str(codepoint.encode_utf8(&mut [0u8; 4]));
2442 return Ok(());
2443 }
2444 if let Some(value) = self.entity_resolver.resolve(reference.as_ref()) {
2445 result.push_str(value);
2446 return Ok(());
2447 }
2448 Err(EscapeError::UnrecognizedEntity(0..len, reference.to_string()).into())
2449 }
2450
2451 #[inline]
2452 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
2453 match self.lookahead {
2454 // We pre-read event with the same name that is required to be skipped.
2455 // First call of `read_to_end` will end out pre-read event, the second
2456 // will consume other events
2457 Ok(PayloadEvent::Start(ref e)) if e.name() == name => {
2458 let result1 = self.reader.read_to_end(name);
2459 let result2 = self.reader.read_to_end(name);
2460
2461 // In case of error `next_impl` returns `Eof`
2462 let _ = self.next_impl();
2463 result1?;
2464 result2?;
2465 }
2466 // We pre-read event with the same name that is required to be skipped.
2467 // Because this is end event, we already consume the whole tree, so
2468 // nothing to do, just update lookahead
2469 Ok(PayloadEvent::End(ref e)) if e.name() == name => {
2470 let _ = self.next_impl();
2471 }
2472 Ok(_) => {
2473 let result = self.reader.read_to_end(name);
2474
2475 // In case of error `next_impl` returns `Eof`
2476 let _ = self.next_impl();
2477 result?;
2478 }
2479 // Read next lookahead event, unpack error from the current lookahead
2480 Err(_) => {
2481 self.next_impl()?;
2482 }
2483 }
2484 Ok(())
2485 }
2486
2487 #[inline]
2488 fn decoder(&self) -> Decoder {
2489 self.reader.decoder()
2490 }
2491}
2492
2493////////////////////////////////////////////////////////////////////////////////////////////////////
2494
2495/// Deserialize an instance of type `T` from a string of XML text.
2496pub fn from_str<'de, T>(s: &'de str) -> Result<T, DeError>
2497where
2498 T: Deserialize<'de>,
2499{
2500 let mut de = Deserializer::from_str(s);
2501 T::deserialize(&mut de)
2502}
2503
2504/// Deserialize from a reader. This method will do internal copies of data
2505/// read from `reader`. If you want have a `&str` input and want to borrow
2506/// as much as possible, use [`from_str`].
2507pub fn from_reader<R, T>(reader: R) -> Result<T, DeError>
2508where
2509 R: BufRead,
2510 T: DeserializeOwned,
2511{
2512 let mut de = Deserializer::from_reader(reader);
2513 T::deserialize(&mut de)
2514}
2515
2516////////////////////////////////////////////////////////////////////////////////////////////////////
2517
2518/// A structure that deserializes XML into Rust values.
2519pub struct Deserializer<'de, R, E: EntityResolver = PredefinedEntityResolver>
2520where
2521 R: XmlRead<'de>,
2522{
2523 /// An XML reader that streams events into this deserializer
2524 reader: XmlReader<'de, R, E>,
2525
2526 /// When deserializing sequences sometimes we have to skip unwanted events.
2527 /// That events should be stored and then replayed. This is a replay buffer,
2528 /// that streams events while not empty. When it exhausted, events will
2529 /// requested from [`Self::reader`].
2530 #[cfg(feature = "overlapped-lists")]
2531 read: VecDeque<DeEvent<'de>>,
2532 /// When deserializing sequences sometimes we have to skip events, because XML
2533 /// is tolerant to elements order and even if in the XSD order is strictly
2534 /// specified (using `xs:sequence`) most of XML parsers allows order violations.
2535 /// That means, that elements, forming a sequence, could be overlapped with
2536 /// other elements, do not related to that sequence.
2537 ///
2538 /// In order to support this, deserializer will scan events and skip unwanted
2539 /// events, store them here. After call [`Self::start_replay()`] all events
2540 /// moved from this to [`Self::read`].
2541 #[cfg(feature = "overlapped-lists")]
2542 write: VecDeque<DeEvent<'de>>,
2543 /// Maximum number of events that can be skipped when processing sequences
2544 /// that occur out-of-order. This field is used to prevent potential
2545 /// denial-of-service (DoS) attacks which could cause infinite memory
2546 /// consumption when parsing a very large amount of XML into a sequence field.
2547 #[cfg(feature = "overlapped-lists")]
2548 limit: Option<NonZeroUsize>,
2549
2550 #[cfg(not(feature = "overlapped-lists"))]
2551 peek: Option<DeEvent<'de>>,
2552
2553 /// Buffer to store attribute name as a field name exposed to serde consumers
2554 key_buf: String,
2555}
2556
2557impl<'de, R, E> Deserializer<'de, R, E>
2558where
2559 R: XmlRead<'de>,
2560 E: EntityResolver,
2561{
2562 /// Create an XML deserializer from one of the possible quick_xml input sources.
2563 ///
2564 /// Typically it is more convenient to use one of these methods instead:
2565 ///
2566 /// - [`Deserializer::from_str`]
2567 /// - [`Deserializer::from_reader`]
2568 fn new(reader: R, entity_resolver: E) -> Self {
2569 Self {
2570 reader: XmlReader::new(reader, entity_resolver),
2571
2572 #[cfg(feature = "overlapped-lists")]
2573 read: VecDeque::new(),
2574 #[cfg(feature = "overlapped-lists")]
2575 write: VecDeque::new(),
2576 #[cfg(feature = "overlapped-lists")]
2577 limit: None,
2578
2579 #[cfg(not(feature = "overlapped-lists"))]
2580 peek: None,
2581
2582 key_buf: String::new(),
2583 }
2584 }
2585
2586 /// Returns `true` if all events was consumed.
2587 pub fn is_empty(&self) -> bool {
2588 #[cfg(feature = "overlapped-lists")]
2589 let event = self.read.front();
2590
2591 #[cfg(not(feature = "overlapped-lists"))]
2592 let event = self.peek.as_ref();
2593
2594 match event {
2595 None | Some(DeEvent::Eof) => self.reader.is_empty(),
2596 _ => false,
2597 }
2598 }
2599
2600 /// Returns the underlying XML reader.
2601 ///
2602 /// ```
2603 /// # use pretty_assertions::assert_eq;
2604 /// use serde::Deserialize;
2605 /// use quick_xml::de::Deserializer;
2606 /// use quick_xml::NsReader;
2607 ///
2608 /// #[derive(Deserialize)]
2609 /// struct SomeStruct {
2610 /// field1: String,
2611 /// field2: String,
2612 /// }
2613 ///
2614 /// // Try to deserialize from broken XML
2615 /// let mut de = Deserializer::from_str(
2616 /// "<SomeStruct><field1><field2></SomeStruct>"
2617 /// // 0 ^= 28 ^= 41
2618 /// );
2619 ///
2620 /// let err = SomeStruct::deserialize(&mut de);
2621 /// assert!(err.is_err());
2622 ///
2623 /// let reader: &NsReader<_> = de.get_ref().get_ref();
2624 ///
2625 /// assert_eq!(reader.error_position(), 28);
2626 /// assert_eq!(reader.buffer_position(), 41);
2627 /// ```
2628 pub const fn get_ref(&self) -> &R {
2629 &self.reader.reader
2630 }
2631
2632 /// Set the maximum number of events that could be skipped during deserialization
2633 /// of sequences.
2634 ///
2635 /// If `<element>` contains more than specified nested elements, `$text` or
2636 /// CDATA nodes, then [`DeError::TooManyEvents`] will be returned during
2637 /// deserialization of sequence field (any type that uses [`deserialize_seq`]
2638 /// for the deserialization, for example, `Vec<T>`).
2639 ///
2640 /// This method can be used to prevent a [DoS] attack and infinite memory
2641 /// consumption when parsing a very large XML to a sequence field.
2642 ///
2643 /// It is strongly recommended to set limit to some value when you parse data
2644 /// from untrusted sources. You should choose a value that your typical XMLs
2645 /// can have _between_ different elements that corresponds to the same sequence.
2646 ///
2647 /// # Examples
2648 ///
2649 /// Let's imagine, that we deserialize such structure:
2650 /// ```
2651 /// struct List {
2652 /// item: Vec<()>,
2653 /// }
2654 /// ```
2655 ///
2656 /// The XML that we try to parse look like this:
2657 /// ```xml
2658 /// <any-name>
2659 /// <item/>
2660 /// <!-- Bufferization starts at this point -->
2661 /// <another-item>
2662 /// <some-element>with text</some-element>
2663 /// <yet-another-element/>
2664 /// </another-item>
2665 /// <!-- Buffer will be emptied at this point; 7 events were buffered -->
2666 /// <item/>
2667 /// <!-- There is nothing to buffer, because elements follows each other -->
2668 /// <item/>
2669 /// </any-name>
2670 /// ```
2671 ///
2672 /// There, when we deserialize the `item` field, we need to buffer 7 events,
2673 /// before we can deserialize the second `<item/>`:
2674 ///
2675 /// - `<another-item>`
2676 /// - `<some-element>`
2677 /// - `$text(with text)`
2678 /// - `</some-element>`
2679 /// - `<yet-another-element/>` (virtual start event)
2680 /// - `<yet-another-element/>` (virtual end event)
2681 /// - `</another-item>`
2682 ///
2683 /// Note, that `<yet-another-element/>` internally represented as 2 events:
2684 /// one for the start tag and one for the end tag. In the future this can be
2685 /// eliminated, but for now we use [auto-expanding feature] of a reader,
2686 /// because this simplifies deserializer code.
2687 ///
2688 /// [`deserialize_seq`]: serde::Deserializer::deserialize_seq
2689 /// [DoS]: https://en.wikipedia.org/wiki/Denial-of-service_attack
2690 /// [auto-expanding feature]: crate::reader::Config::expand_empty_elements
2691 #[cfg(feature = "overlapped-lists")]
2692 pub fn event_buffer_size(&mut self, limit: Option<NonZeroUsize>) -> &mut Self {
2693 self.limit = limit;
2694 self
2695 }
2696
2697 #[cfg(feature = "overlapped-lists")]
2698 fn peek(&mut self) -> Result<&DeEvent<'de>, DeError> {
2699 if self.read.is_empty() {
2700 self.read.push_front(self.reader.next()?);
2701 }
2702 if let Some(event) = self.read.front() {
2703 return Ok(event);
2704 }
2705 // SAFETY: `self.read` was filled in the code above.
2706 // NOTE: Can be replaced with `unsafe { std::hint::unreachable_unchecked() }`
2707 // if unsafe code will be allowed
2708 unreachable!()
2709 }
2710 #[cfg(not(feature = "overlapped-lists"))]
2711 fn peek(&mut self) -> Result<&DeEvent<'de>, DeError> {
2712 match &mut self.peek {
2713 Some(event) => Ok(event),
2714 empty_peek @ None => Ok(empty_peek.insert(self.reader.next()?)),
2715 }
2716 }
2717
2718 #[inline]
2719 fn last_peeked(&self) -> &DeEvent<'de> {
2720 #[cfg(feature = "overlapped-lists")]
2721 {
2722 self.read
2723 .front()
2724 .expect("`Deserializer::peek()` should be called")
2725 }
2726 #[cfg(not(feature = "overlapped-lists"))]
2727 {
2728 self.peek
2729 .as_ref()
2730 .expect("`Deserializer::peek()` should be called")
2731 }
2732 }
2733
2734 fn next(&mut self) -> Result<DeEvent<'de>, DeError> {
2735 // Replay skipped or peeked events
2736 #[cfg(feature = "overlapped-lists")]
2737 if let Some(event) = self.read.pop_front() {
2738 return Ok(event);
2739 }
2740 #[cfg(not(feature = "overlapped-lists"))]
2741 if let Some(e) = self.peek.take() {
2742 return Ok(e);
2743 }
2744 self.reader.next()
2745 }
2746
2747 fn skip_whitespaces(&mut self) -> Result<(), DeError> {
2748 loop {
2749 match self.peek()? {
2750 DeEvent::Text(e) if e.is_blank() => {
2751 self.next()?;
2752 }
2753 _ => break,
2754 }
2755 }
2756 Ok(())
2757 }
2758
2759 /// Returns the mark after which all events, skipped by [`Self::skip()`] call,
2760 /// should be replayed after calling [`Self::start_replay()`].
2761 #[cfg(feature = "overlapped-lists")]
2762 #[inline]
2763 #[must_use = "returned checkpoint should be used in `start_replay`"]
2764 fn skip_checkpoint(&self) -> usize {
2765 self.write.len()
2766 }
2767
2768 /// Extracts XML tree of events from and stores them in the skipped events
2769 /// buffer from which they can be retrieved later. You MUST call
2770 /// [`Self::start_replay()`] after calling this to give access to the skipped
2771 /// events and release internal buffers.
2772 #[cfg(feature = "overlapped-lists")]
2773 fn skip(&mut self) -> Result<(), DeError> {
2774 let event = self.next()?;
2775 self.skip_event(event)?;
2776 // Skip all subtree, if we skip a start event
2777 if let Some(DeEvent::Start(e)) = self.write.back() {
2778 let end = e.name().as_ref().to_owned();
2779 let mut depth = 0;
2780 loop {
2781 let event = self.next()?;
2782 match event {
2783 DeEvent::Start(ref e) if e.name().as_ref() == end => {
2784 self.skip_event(event)?;
2785 depth += 1;
2786 }
2787 DeEvent::End(ref e) if e.name().as_ref() == end => {
2788 self.skip_event(event)?;
2789 if depth == 0 {
2790 break;
2791 }
2792 depth -= 1;
2793 }
2794 DeEvent::Eof => {
2795 self.skip_event(event)?;
2796 break;
2797 }
2798 _ => self.skip_event(event)?,
2799 }
2800 }
2801 }
2802 Ok(())
2803 }
2804
2805 #[cfg(feature = "overlapped-lists")]
2806 #[inline]
2807 fn skip_event(&mut self, event: DeEvent<'de>) -> Result<(), DeError> {
2808 if let Some(max) = self.limit {
2809 if self.write.len() >= max.get() {
2810 return Err(DeError::TooManyEvents(max));
2811 }
2812 }
2813 self.write.push_back(event);
2814 Ok(())
2815 }
2816
2817 /// Moves buffered events, skipped after given `checkpoint` from [`Self::write`]
2818 /// skip buffer to [`Self::read`] buffer.
2819 ///
2820 /// After calling this method, [`Self::peek()`] and [`Self::next()`] starts
2821 /// return events that was skipped previously by calling [`Self::skip()`],
2822 /// and only when all that events will be consumed, the deserializer starts
2823 /// to drain events from underlying reader.
2824 ///
2825 /// This method MUST be called if any number of [`Self::skip()`] was called
2826 /// after [`Self::new()`] or `start_replay()` or you'll lost events.
2827 #[cfg(feature = "overlapped-lists")]
2828 fn start_replay(&mut self, checkpoint: usize) {
2829 if checkpoint == 0 {
2830 self.write.append(&mut self.read);
2831 std::mem::swap(&mut self.read, &mut self.write);
2832 } else {
2833 let mut read = self.write.split_off(checkpoint);
2834 read.append(&mut self.read);
2835 self.read = read;
2836 }
2837 }
2838
2839 #[inline]
2840 fn read_string(&mut self) -> Result<Cow<'de, str>, DeError> {
2841 self.read_string_impl(true)
2842 }
2843
2844 /// Consumes consequent [`Text`] and [`CData`] (both a referred below as a _text_)
2845 /// events, merge them into one string. If there are no such events, returns
2846 /// an empty string.
2847 ///
2848 /// If `allow_start` is `false`, then only text events are consumed, for other
2849 /// events an error is returned (see table below).
2850 ///
2851 /// If `allow_start` is `true`, then two or three events are expected:
2852 /// - [`DeEvent::Start`];
2853 /// - _(optional)_ [`DeEvent::Text`] which content is returned;
2854 /// - [`DeEvent::End`]. If text event was missed, an empty string is returned.
2855 ///
2856 /// Corresponding events are consumed.
2857 ///
2858 /// # Handling events
2859 ///
2860 /// The table below shows how events is handled by this method:
2861 ///
2862 /// |Event |XML |Handling
2863 /// |------------------|---------------------------|----------------------------------------
2864 /// |[`DeEvent::Start`]|`<tag>...</tag>` |if `allow_start == true`, result determined by the second table, otherwise emits [`UnexpectedStart("tag")`](DeError::UnexpectedStart)
2865 /// |[`DeEvent::End`] |`</any-tag>` |This is impossible situation, the method will panic if it happens
2866 /// |[`DeEvent::Text`] |`text content` or `<![CDATA[cdata content]]>` (probably mixed)|Returns event content unchanged
2867 /// |[`DeEvent::Eof`] | |Emits [`UnexpectedEof`](DeError::UnexpectedEof)
2868 ///
2869 /// Second event, consumed if [`DeEvent::Start`] was received and `allow_start == true`:
2870 ///
2871 /// |Event |XML |Handling
2872 /// |------------------|---------------------------|----------------------------------------------------------------------------------
2873 /// |[`DeEvent::Start`]|`<any-tag>...</any-tag>` |Emits [`UnexpectedStart("any-tag")`](DeError::UnexpectedStart)
2874 /// |[`DeEvent::End`] |`</tag>` |Returns an empty slice. The reader guarantee that tag will match the open one
2875 /// |[`DeEvent::Text`] |`text content` or `<![CDATA[cdata content]]>` (probably mixed)|Returns event content unchanged, expects the `</tag>` after that
2876 /// |[`DeEvent::Eof`] | |Emits [`InvalidXml(IllFormed(MissingEndTag))`](DeError::InvalidXml)
2877 ///
2878 /// [`Text`]: Event::Text
2879 /// [`CData`]: Event::CData
2880 fn read_string_impl(&mut self, allow_start: bool) -> Result<Cow<'de, str>, DeError> {
2881 match self.next()? {
2882 // Reached by doc tests only: this file, lines 979 and 996
2883 DeEvent::Text(e) => Ok(e.text),
2884 // allow one nested level
2885 // Reached by trivial::{...}::{field, field_nested, field_tag_after, field_tag_before, nested, tag_after, tag_before, wrapped}
2886 DeEvent::Start(e) if allow_start => self.read_text(e.name()),
2887 // TODO: not reached by any tests
2888 DeEvent::Start(e) => Err(DeError::UnexpectedStart(e.name().as_ref().to_owned())),
2889 // SAFETY: The reader is guaranteed that we don't have unmatched tags
2890 // If we here, then our deserializer has a bug
2891 DeEvent::End(e) => unreachable!("{:?}", e),
2892 // Reached by trivial::{empty_doc, only_comment}
2893 DeEvent::Eof => Err(DeError::UnexpectedEof),
2894 }
2895 }
2896 /// Consumes one [`DeEvent::Text`] event and ensures that it is followed by the
2897 /// [`DeEvent::End`] event.
2898 ///
2899 /// # Parameters
2900 /// - `name`: name of a tag opened before reading text. The corresponding end tag
2901 /// should present in input just after the text
2902 fn read_text(&mut self, name: QName) -> Result<Cow<'de, str>, DeError> {
2903 match self.next()? {
2904 DeEvent::Text(e) => match self.next()? {
2905 // The matching tag name is guaranteed by the reader
2906 // Reached by trivial::{...}::{field, wrapped}
2907 DeEvent::End(_) => Ok(e.text),
2908 // SAFETY: Cannot be two consequent Text events, they would be merged into one
2909 DeEvent::Text(_) => unreachable!(),
2910 // Reached by trivial::{...}::{field_tag_after, tag_after}
2911 DeEvent::Start(e) => Err(DeError::UnexpectedStart(e.name().as_ref().to_owned())),
2912 // Reached by struct_::non_closed::elements_child
2913 DeEvent::Eof => Err(Error::missed_end(name, self.reader.decoder()).into()),
2914 },
2915 // We can get End event in case of `<tag></tag>` or `<tag/>` input
2916 // Return empty text in that case
2917 // The matching tag name is guaranteed by the reader
2918 // Reached by {...}::xs_list::empty
2919 DeEvent::End(_) => Ok("".into()),
2920 // Reached by trivial::{...}::{field_nested, field_tag_before, nested, tag_before}
2921 DeEvent::Start(s) => Err(DeError::UnexpectedStart(s.name().as_ref().to_owned())),
2922 // Reached by struct_::non_closed::elements_child
2923 DeEvent::Eof => Err(Error::missed_end(name, self.reader.decoder()).into()),
2924 }
2925 }
2926
2927 /// Drops all events until event with [name](BytesEnd::name()) `name` won't be
2928 /// dropped. This method should be called after [`Self::next()`]
2929 #[cfg(feature = "overlapped-lists")]
2930 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
2931 let mut depth = 0;
2932 loop {
2933 match self.read.pop_front() {
2934 Some(DeEvent::Start(e)) if e.name() == name => {
2935 depth += 1;
2936 }
2937 Some(DeEvent::End(e)) if e.name() == name => {
2938 if depth == 0 {
2939 break;
2940 }
2941 depth -= 1;
2942 }
2943
2944 // Drop all other skipped events
2945 Some(_) => continue,
2946
2947 // If we do not have skipped events, use effective reading that will
2948 // not allocate memory for events
2949 None => {
2950 // We should close all opened tags, because we could buffer
2951 // Start events, but not the corresponding End events. So we
2952 // keep reading events until we exit all nested tags.
2953 // `read_to_end()` will return an error if an Eof was encountered
2954 // preliminary (in case of malformed XML).
2955 //
2956 // <tag><tag></tag></tag>
2957 // ^^^^^^^^^^ - buffered in `self.read`, when `self.read_to_end()` is called, depth = 2
2958 // ^^^^^^ - read by the first call of `self.reader.read_to_end()`
2959 // ^^^^^^ - read by the second call of `self.reader.read_to_end()`
2960 loop {
2961 self.reader.read_to_end(name)?;
2962 if depth == 0 {
2963 break;
2964 }
2965 depth -= 1;
2966 }
2967 break;
2968 }
2969 }
2970 }
2971 Ok(())
2972 }
2973 #[cfg(not(feature = "overlapped-lists"))]
2974 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
2975 // First one might be in self.peek
2976 match self.next()? {
2977 DeEvent::Start(e) => self.reader.read_to_end(e.name())?,
2978 DeEvent::End(e) if e.name() == name => return Ok(()),
2979 _ => (),
2980 }
2981 self.reader.read_to_end(name)
2982 }
2983
2984 fn skip_next_tree(&mut self) -> Result<(), DeError> {
2985 let DeEvent::Start(start) = self.next()? else {
2986 unreachable!("Only call this if the next event is a start event")
2987 };
2988 let name = start.name();
2989 self.read_to_end(name)
2990 }
2991
2992 /// Method for testing Deserializer implementation. Checks that all events was consumed during
2993 /// deserialization. Panics if the next event will not be [`DeEvent::Eof`].
2994 #[doc(hidden)]
2995 #[track_caller]
2996 pub fn check_eof_reached(&mut self) {
2997 // Deserializer may not consume trailing spaces, that is normal
2998 self.skip_whitespaces().expect("cannot skip whitespaces");
2999 let event = self.peek().expect("cannot peek event");
3000 assert_eq!(
3001 *event,
3002 DeEvent::Eof,
3003 "the whole XML document should be consumed, expected `Eof`",
3004 );
3005 }
3006}
3007
3008impl<'de> Deserializer<'de, SliceReader<'de>> {
3009 /// Create a new deserializer that will borrow data from the specified string.
3010 ///
3011 /// Deserializer created with this method will not resolve custom entities.
3012 #[allow(clippy::should_implement_trait)]
3013 pub fn from_str(source: &'de str) -> Self {
3014 Self::from_str_with_resolver(source, PredefinedEntityResolver)
3015 }
3016
3017 /// Create a new deserializer that will borrow data from the specified preconfigured
3018 /// reader.
3019 ///
3020 /// Deserializer created with this method will not resolve custom entities.
3021 ///
3022 /// Note, that config option [`Config::expand_empty_elements`] will be set to `true`.
3023 ///
3024 /// # Example
3025 ///
3026 /// ```
3027 /// # use pretty_assertions::assert_eq;
3028 /// # use quick_xml::de::Deserializer;
3029 /// # use quick_xml::NsReader;
3030 /// # use serde::Deserialize;
3031 /// #
3032 /// #[derive(Deserialize, PartialEq, Debug)]
3033 /// struct Object<'a> {
3034 /// tag: &'a str,
3035 /// }
3036 ///
3037 /// let mut reader = NsReader::from_str("<xml><tag> test </tag></xml>");
3038 ///
3039 /// let mut de = Deserializer::borrowing(reader.clone());
3040 /// let obj = Object::deserialize(&mut de).unwrap();
3041 /// assert_eq!(obj, Object { tag: " test " });
3042 ///
3043 /// reader.config_mut().trim_text(true);
3044 ///
3045 /// let mut de = Deserializer::borrowing(reader);
3046 /// let obj = Object::deserialize(&mut de).unwrap();
3047 /// assert_eq!(obj, Object { tag: "test" });
3048 /// ```
3049 ///
3050 /// [`Config::expand_empty_elements`]: crate::reader::Config::expand_empty_elements
3051 #[inline]
3052 pub fn borrowing(reader: NsReader<&'de [u8]>) -> Self {
3053 Self::borrowing_with_resolver(reader, PredefinedEntityResolver)
3054 }
3055}
3056
3057impl<'de, E> Deserializer<'de, SliceReader<'de>, E>
3058where
3059 E: EntityResolver,
3060{
3061 /// Create a new deserializer that will borrow data from the specified string
3062 /// and use the specified entity resolver.
3063 pub fn from_str_with_resolver(source: &'de str, entity_resolver: E) -> Self {
3064 Self::borrowing_with_resolver(NsReader::from_str(source), entity_resolver)
3065 }
3066
3067 /// Create a new deserializer that will borrow data from the specified preconfigured
3068 /// reader and use the specified entity resolver.
3069 ///
3070 /// Note, that config option [`Config::expand_empty_elements`] will be set to `true`.
3071 ///
3072 /// [`Config::expand_empty_elements`]: crate::reader::Config::expand_empty_elements
3073 pub fn borrowing_with_resolver(mut reader: NsReader<&'de [u8]>, entity_resolver: E) -> Self {
3074 let config = reader.config_mut();
3075 config.expand_empty_elements = true;
3076
3077 Self::new(SliceReader { reader }, entity_resolver)
3078 }
3079}
3080
3081impl<'de, R> Deserializer<'de, IoReader<R>>
3082where
3083 R: BufRead,
3084{
3085 /// Create a new deserializer that will copy data from the specified reader
3086 /// into internal buffer.
3087 ///
3088 /// If you already have a string use [`Self::from_str`] instead, because it
3089 /// will borrow instead of copy. If you have `&[u8]` which is known to represent
3090 /// UTF-8, you can decode it first before using [`from_str`].
3091 ///
3092 /// Deserializer created with this method will not resolve custom entities.
3093 pub fn from_reader(reader: R) -> Self {
3094 Self::with_resolver(reader, PredefinedEntityResolver)
3095 }
3096
3097 /// Create a new deserializer that will copy data from the specified preconfigured
3098 /// reader into internal buffer.
3099 ///
3100 /// Deserializer created with this method will not resolve custom entities.
3101 ///
3102 /// Note, that config option [`Config::expand_empty_elements`] will be set to `true`.
3103 ///
3104 /// # Example
3105 ///
3106 /// ```
3107 /// # use pretty_assertions::assert_eq;
3108 /// # use quick_xml::de::Deserializer;
3109 /// # use quick_xml::NsReader;
3110 /// # use serde::Deserialize;
3111 /// #
3112 /// #[derive(Deserialize, PartialEq, Debug)]
3113 /// struct Object {
3114 /// tag: String,
3115 /// }
3116 ///
3117 /// let mut reader = NsReader::from_str("<xml><tag> test </tag></xml>");
3118 ///
3119 /// let mut de = Deserializer::buffering(reader.clone());
3120 /// let obj = Object::deserialize(&mut de).unwrap();
3121 /// assert_eq!(obj, Object { tag: " test ".to_string() });
3122 ///
3123 /// reader.config_mut().trim_text(true);
3124 ///
3125 /// let mut de = Deserializer::buffering(reader);
3126 /// let obj = Object::deserialize(&mut de).unwrap();
3127 /// assert_eq!(obj, Object { tag: "test".to_string() });
3128 /// ```
3129 ///
3130 /// [`Config::expand_empty_elements`]: crate::reader::Config::expand_empty_elements
3131 #[inline]
3132 pub fn buffering(reader: NsReader<R>) -> Self {
3133 Self::buffering_with_resolver(reader, PredefinedEntityResolver)
3134 }
3135}
3136
3137impl<'de, R, E> Deserializer<'de, IoReader<R>, E>
3138where
3139 R: BufRead,
3140 E: EntityResolver,
3141{
3142 /// Create a new deserializer that will copy data from the specified reader
3143 /// into internal buffer and use the specified entity resolver.
3144 ///
3145 /// If you already have a string use [`Self::from_str`] instead, because it
3146 /// will borrow instead of copy. If you have `&[u8]` which is known to represent
3147 /// UTF-8, you can decode it first before using [`from_str`].
3148 pub fn with_resolver(reader: R, entity_resolver: E) -> Self {
3149 let mut reader = NsReader::from_reader(reader);
3150 let config = reader.config_mut();
3151 config.expand_empty_elements = true;
3152
3153 Self::new(
3154 IoReader {
3155 reader,
3156 buf: Vec::new(),
3157 },
3158 entity_resolver,
3159 )
3160 }
3161
3162 /// Create new deserializer that will copy data from the specified preconfigured reader
3163 /// into internal buffer and use the specified entity resolver.
3164 ///
3165 /// Note, that config option [`Config::expand_empty_elements`] will be set to `true`.
3166 ///
3167 /// [`Config::expand_empty_elements`]: crate::reader::Config::expand_empty_elements
3168 pub fn buffering_with_resolver(mut reader: NsReader<R>, entity_resolver: E) -> Self {
3169 let config = reader.config_mut();
3170 config.expand_empty_elements = true;
3171
3172 Self::new(
3173 IoReader {
3174 reader,
3175 buf: Vec::new(),
3176 },
3177 entity_resolver,
3178 )
3179 }
3180}
3181
3182impl<'de, R, E> de::Deserializer<'de> for &mut Deserializer<'de, R, E>
3183where
3184 R: XmlRead<'de>,
3185 E: EntityResolver,
3186{
3187 type Error = DeError;
3188
3189 deserialize_primitives!();
3190
3191 fn deserialize_struct<V>(
3192 self,
3193 _name: &'static str,
3194 fields: &'static [&'static str],
3195 visitor: V,
3196 ) -> Result<V::Value, DeError>
3197 where
3198 V: Visitor<'de>,
3199 {
3200 // When document is pretty-printed there could be whitespaces before the root element
3201 self.skip_whitespaces()?;
3202 match self.next()? {
3203 DeEvent::Start(e) => visitor.visit_map(ElementMapAccess::new(self, e, fields)),
3204 // SAFETY: The reader is guaranteed that we don't have unmatched tags
3205 // If we here, then our deserializer has a bug
3206 DeEvent::End(e) => unreachable!("{:?}", e),
3207 // Deserializer methods are only hints, if deserializer could not satisfy
3208 // request, it should return the data that it has. It is responsibility
3209 // of a Visitor to return an error if it does not understand the data
3210 DeEvent::Text(e) => match e.text {
3211 Cow::Borrowed(s) => visitor.visit_borrowed_str(s),
3212 Cow::Owned(s) => visitor.visit_string(s),
3213 },
3214 DeEvent::Eof => Err(DeError::UnexpectedEof),
3215 }
3216 }
3217
3218 /// Unit represented in XML as a `xs:element` or text/CDATA content.
3219 /// Any content inside `xs:element` is ignored and skipped.
3220 ///
3221 /// Produces unit struct from any of following inputs:
3222 /// - any `<tag ...>...</tag>`
3223 /// - any `<tag .../>`
3224 /// - any consequent text / CDATA content (can consist of several parts
3225 /// delimited by comments and processing instructions)
3226 ///
3227 /// # Events handling
3228 ///
3229 /// |Event |XML |Handling
3230 /// |------------------|---------------------------|-------------------------------------------
3231 /// |[`DeEvent::Start`]|`<tag>...</tag>` |Calls `visitor.visit_unit()`, consumes all events up to and including corresponding `End` event
3232 /// |[`DeEvent::End`] |`</tag>` |This is impossible situation, the method will panic if it happens
3233 /// |[`DeEvent::Text`] |`text content` or `<![CDATA[cdata content]]>` (probably mixed)|Calls `visitor.visit_unit()`. The content is ignored
3234 /// |[`DeEvent::Eof`] | |Emits [`UnexpectedEof`](DeError::UnexpectedEof)
3235 fn deserialize_unit<V>(self, visitor: V) -> Result<V::Value, DeError>
3236 where
3237 V: Visitor<'de>,
3238 {
3239 match self.next()? {
3240 DeEvent::Start(s) => {
3241 self.read_to_end(s.name())?;
3242 visitor.visit_unit()
3243 }
3244 DeEvent::Text(_) => visitor.visit_unit(),
3245 // SAFETY: The reader is guaranteed that we don't have unmatched tags
3246 // If we here, then our deserializer has a bug
3247 DeEvent::End(e) => unreachable!("{:?}", e),
3248 DeEvent::Eof => Err(DeError::UnexpectedEof),
3249 }
3250 }
3251
3252 /// Forwards deserialization of the inner type. Always calls [`Visitor::visit_newtype_struct`]
3253 /// with the same deserializer.
3254 fn deserialize_newtype_struct<V>(
3255 self,
3256 _name: &'static str,
3257 visitor: V,
3258 ) -> Result<V::Value, DeError>
3259 where
3260 V: Visitor<'de>,
3261 {
3262 visitor.visit_newtype_struct(self)
3263 }
3264
3265 fn deserialize_enum<V>(
3266 self,
3267 _name: &'static str,
3268 _variants: &'static [&'static str],
3269 visitor: V,
3270 ) -> Result<V::Value, DeError>
3271 where
3272 V: Visitor<'de>,
3273 {
3274 // When document is pretty-printed there could be whitespaces before the root element
3275 // which represents the enum variant
3276 // Checked by `top_level::list_of_enum` test in serde-de-seq
3277 self.skip_whitespaces()?;
3278 visitor.visit_enum(var::EnumAccess::new(self))
3279 }
3280
3281 fn deserialize_seq<V>(self, visitor: V) -> Result<V::Value, DeError>
3282 where
3283 V: Visitor<'de>,
3284 {
3285 visitor.visit_seq(self)
3286 }
3287
3288 fn deserialize_option<V>(self, visitor: V) -> Result<V::Value, DeError>
3289 where
3290 V: Visitor<'de>,
3291 {
3292 // We cannot use result of `peek()` directly because of borrow checker
3293 let _ = self.peek()?;
3294 match self.last_peeked() {
3295 DeEvent::Text(t) if t.is_empty() => visitor.visit_none(),
3296 DeEvent::Eof => visitor.visit_none(),
3297 // if the `xsi:nil` attribute is set to true we got a none value
3298 DeEvent::Start(start) if self.reader.reader.has_nil_attr(start) => {
3299 self.skip_next_tree()?;
3300 visitor.visit_none()
3301 }
3302 _ => visitor.visit_some(self),
3303 }
3304 }
3305
3306 fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, DeError>
3307 where
3308 V: Visitor<'de>,
3309 {
3310 match self.peek()? {
3311 DeEvent::Text(_) => self.deserialize_str(visitor),
3312 _ => self.deserialize_map(visitor),
3313 }
3314 }
3315}
3316
3317/// An accessor to sequence elements forming a value for top-level sequence of XML
3318/// elements.
3319///
3320/// Technically, multiple top-level elements violates XML rule of only one top-level
3321/// element, but we consider this as several concatenated XML documents.
3322impl<'de, R, E> SeqAccess<'de> for &mut Deserializer<'de, R, E>
3323where
3324 R: XmlRead<'de>,
3325 E: EntityResolver,
3326{
3327 type Error = DeError;
3328
3329 fn next_element_seed<T>(&mut self, seed: T) -> Result<Option<T::Value>, Self::Error>
3330 where
3331 T: DeserializeSeed<'de>,
3332 {
3333 // When document is pretty-printed there could be whitespaces before, between
3334 // and after root elements. We cannot defer decision if we need to skip spaces
3335 // or not: if we have a sequence of type that does not accept blank text, it
3336 // will need to return something and it can return only error. For example,
3337 // it can be enum without `$text` variant
3338 // Checked by `top_level::list_of_enum` test in serde-de-seq
3339 self.skip_whitespaces()?;
3340 match self.peek()? {
3341 DeEvent::Eof => Ok(None),
3342
3343 // Start(tag), End(tag), Text
3344 _ => seed.deserialize(&mut **self).map(Some),
3345 }
3346 }
3347}
3348
3349impl<'de, R, E> IntoDeserializer<'de, DeError> for &mut Deserializer<'de, R, E>
3350where
3351 R: XmlRead<'de>,
3352 E: EntityResolver,
3353{
3354 type Deserializer = Self;
3355
3356 #[inline]
3357 fn into_deserializer(self) -> Self {
3358 self
3359 }
3360}
3361
3362////////////////////////////////////////////////////////////////////////////////////////////////////
3363
3364/// Converts raw reader's event into a payload event.
3365/// Returns `None`, if event should be skipped.
3366#[inline(always)]
3367fn skip_uninterested<'a>(event: Event<'a>) -> Option<PayloadEvent<'a>> {
3368 let event = match event {
3369 Event::DocType(e) => PayloadEvent::DocType(e),
3370 Event::Start(e) => PayloadEvent::Start(e),
3371 Event::End(e) => PayloadEvent::End(e),
3372 Event::Eof => PayloadEvent::Eof,
3373
3374 // Do not trim next text event after Text, CDATA or reference event
3375 Event::CData(e) => PayloadEvent::CData(e),
3376 Event::Text(e) => PayloadEvent::Text(e),
3377 Event::GeneralRef(e) => PayloadEvent::GeneralRef(e),
3378
3379 _ => return None,
3380 };
3381 Some(event)
3382}
3383
3384////////////////////////////////////////////////////////////////////////////////////////////////////
3385
3386/// Trait used by the deserializer for iterating over input. This is manually
3387/// "specialized" for iterating over `&[u8]`.
3388///
3389/// You do not need to implement this trait, it is needed to abstract from
3390/// [borrowing](SliceReader) and [copying](IoReader) data sources and reuse code in
3391/// deserializer
3392pub trait XmlRead<'i> {
3393 /// Return an input-borrowing event.
3394 fn next(&mut self) -> Result<PayloadEvent<'i>, DeError>;
3395
3396 /// Skips until end element is found. Unlike `next()` it will not allocate
3397 /// when it cannot satisfy the lifetime.
3398 fn read_to_end(&mut self, name: QName) -> Result<(), DeError>;
3399
3400 /// A copy of the reader's decoder used to decode strings.
3401 fn decoder(&self) -> Decoder;
3402
3403 /// Checks if the `start` tag has a [`xsi:nil`] attribute. This method ignores
3404 /// any errors in attributes.
3405 ///
3406 /// [`xsi:nil`]: https://www.w3.org/TR/xmlschema-1/#xsi_nil
3407 fn has_nil_attr(&self, start: &BytesStart) -> bool;
3408}
3409
3410/// XML input source that reads from a std::io input stream.
3411///
3412/// You cannot create it, it is created automatically when you call
3413/// [`Deserializer::from_reader`]
3414pub struct IoReader<R: BufRead> {
3415 reader: NsReader<R>,
3416 buf: Vec<u8>,
3417}
3418
3419impl<R: BufRead> IoReader<R> {
3420 /// Returns the underlying XML reader.
3421 ///
3422 /// ```
3423 /// # use pretty_assertions::assert_eq;
3424 /// use serde::Deserialize;
3425 /// use std::io::Cursor;
3426 /// use quick_xml::de::Deserializer;
3427 /// use quick_xml::NsReader;
3428 ///
3429 /// #[derive(Deserialize)]
3430 /// struct SomeStruct {
3431 /// field1: String,
3432 /// field2: String,
3433 /// }
3434 ///
3435 /// // Try to deserialize from broken XML
3436 /// let mut de = Deserializer::from_reader(Cursor::new(
3437 /// "<SomeStruct><field1><field2></SomeStruct>"
3438 /// // 0 ^= 28 ^= 41
3439 /// ));
3440 ///
3441 /// let err = SomeStruct::deserialize(&mut de);
3442 /// assert!(err.is_err());
3443 ///
3444 /// let reader: &NsReader<Cursor<&str>> = de.get_ref().get_ref();
3445 ///
3446 /// assert_eq!(reader.error_position(), 28);
3447 /// assert_eq!(reader.buffer_position(), 41);
3448 /// ```
3449 pub const fn get_ref(&self) -> &NsReader<R> {
3450 &self.reader
3451 }
3452}
3453
3454impl<'i, R: BufRead> XmlRead<'i> for IoReader<R> {
3455 fn next(&mut self) -> Result<PayloadEvent<'static>, DeError> {
3456 loop {
3457 self.buf.clear();
3458
3459 let event = self.reader.read_event_into(&mut self.buf)?;
3460 if let Some(event) = skip_uninterested(event) {
3461 return Ok(event.into_owned());
3462 }
3463 }
3464 }
3465
3466 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
3467 match self.reader.read_to_end_into(name, &mut self.buf) {
3468 Err(e) => Err(e.into()),
3469 Ok(_) => Ok(()),
3470 }
3471 }
3472
3473 fn decoder(&self) -> Decoder {
3474 self.reader.decoder()
3475 }
3476
3477 fn has_nil_attr(&self, start: &BytesStart) -> bool {
3478 start.attributes().has_nil(&self.reader)
3479 }
3480}
3481
3482/// XML input source that reads from a slice of bytes and can borrow from it.
3483///
3484/// You cannot create it, it is created automatically when you call
3485/// [`Deserializer::from_str`].
3486pub struct SliceReader<'de> {
3487 reader: NsReader<&'de [u8]>,
3488}
3489
3490impl<'de> SliceReader<'de> {
3491 /// Returns the underlying XML reader.
3492 ///
3493 /// ```
3494 /// # use pretty_assertions::assert_eq;
3495 /// use serde::Deserialize;
3496 /// use quick_xml::de::Deserializer;
3497 /// use quick_xml::NsReader;
3498 ///
3499 /// #[derive(Deserialize)]
3500 /// struct SomeStruct {
3501 /// field1: String,
3502 /// field2: String,
3503 /// }
3504 ///
3505 /// // Try to deserialize from broken XML
3506 /// let mut de = Deserializer::from_str(
3507 /// "<SomeStruct><field1><field2></SomeStruct>"
3508 /// // 0 ^= 28 ^= 41
3509 /// );
3510 ///
3511 /// let err = SomeStruct::deserialize(&mut de);
3512 /// assert!(err.is_err());
3513 ///
3514 /// let reader: &NsReader<&[u8]> = de.get_ref().get_ref();
3515 ///
3516 /// assert_eq!(reader.error_position(), 28);
3517 /// assert_eq!(reader.buffer_position(), 41);
3518 /// ```
3519 pub const fn get_ref(&self) -> &NsReader<&'de [u8]> {
3520 &self.reader
3521 }
3522}
3523
3524impl<'de> XmlRead<'de> for SliceReader<'de> {
3525 fn next(&mut self) -> Result<PayloadEvent<'de>, DeError> {
3526 loop {
3527 let event = self.reader.read_event()?;
3528 if let Some(event) = skip_uninterested(event) {
3529 return Ok(event);
3530 }
3531 }
3532 }
3533
3534 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
3535 match self.reader.read_to_end(name) {
3536 Err(e) => Err(e.into()),
3537 Ok(_) => Ok(()),
3538 }
3539 }
3540
3541 fn decoder(&self) -> Decoder {
3542 self.reader.decoder()
3543 }
3544
3545 fn has_nil_attr(&self, start: &BytesStart) -> bool {
3546 start.attributes().has_nil(&self.reader)
3547 }
3548}
3549
3550#[cfg(test)]
3551mod tests {
3552 use super::*;
3553 use crate::errors::IllFormedError;
3554 use pretty_assertions::assert_eq;
3555
3556 fn make_de<'de>(source: &'de str) -> Deserializer<'de, SliceReader<'de>> {
3557 dbg!(source);
3558 Deserializer::from_str(source)
3559 }
3560
3561 #[cfg(feature = "overlapped-lists")]
3562 mod skip {
3563 use super::*;
3564 use crate::de::DeEvent::*;
3565 use crate::events::BytesEnd;
3566 use pretty_assertions::assert_eq;
3567
3568 /// Checks that `peek()` and `read()` behaves correctly after `skip()`
3569 #[test]
3570 fn read_and_peek() {
3571 let mut de = make_de(
3572 "\
3573 <root>\
3574 <inner>\
3575 text\
3576 <inner/>\
3577 </inner>\
3578 <next/>\
3579 <target/>\
3580 </root>\
3581 ",
3582 );
3583
3584 // Initial conditions - both are empty
3585 assert_eq!(de.read, vec![]);
3586 assert_eq!(de.write, vec![]);
3587
3588 assert_eq!(de.next().unwrap(), Start(BytesStart::new("root")));
3589 assert_eq!(de.peek().unwrap(), &Start(BytesStart::new("inner")));
3590
3591 // Mark that start_replay() should begin replay from this point
3592 let checkpoint = de.skip_checkpoint();
3593 assert_eq!(checkpoint, 0);
3594
3595 // Should skip first <inner> tree
3596 de.skip().unwrap();
3597 assert_eq!(de.read, vec![]);
3598 assert_eq!(
3599 de.write,
3600 vec![
3601 Start(BytesStart::new("inner")),
3602 Text("text".into()),
3603 Start(BytesStart::new("inner")),
3604 End(BytesEnd::new("inner")),
3605 End(BytesEnd::new("inner")),
3606 ]
3607 );
3608
3609 // Consume <next/>. Now unconsumed XML looks like:
3610 //
3611 // <inner>
3612 // text
3613 // <inner/>
3614 // </inner>
3615 // <target/>
3616 // </root>
3617 assert_eq!(de.next().unwrap(), Start(BytesStart::new("next")));
3618 assert_eq!(de.next().unwrap(), End(BytesEnd::new("next")));
3619
3620 // We finish writing. Next call to `next()` should start replay that messages:
3621 //
3622 // <inner>
3623 // text
3624 // <inner/>
3625 // </inner>
3626 //
3627 // and after that stream that messages:
3628 //
3629 // <target/>
3630 // </root>
3631 de.start_replay(checkpoint);
3632 assert_eq!(
3633 de.read,
3634 vec![
3635 Start(BytesStart::new("inner")),
3636 Text("text".into()),
3637 Start(BytesStart::new("inner")),
3638 End(BytesEnd::new("inner")),
3639 End(BytesEnd::new("inner")),
3640 ]
3641 );
3642 assert_eq!(de.write, vec![]);
3643 assert_eq!(de.next().unwrap(), Start(BytesStart::new("inner")));
3644
3645 // Mark that start_replay() should begin replay from this point
3646 let checkpoint = de.skip_checkpoint();
3647 assert_eq!(checkpoint, 0);
3648
3649 // Skip `$text` node and consume <inner/> after it
3650 de.skip().unwrap();
3651 assert_eq!(
3652 de.read,
3653 vec![
3654 Start(BytesStart::new("inner")),
3655 End(BytesEnd::new("inner")),
3656 End(BytesEnd::new("inner")),
3657 ]
3658 );
3659 assert_eq!(
3660 de.write,
3661 vec![
3662 // This comment here to keep the same formatting of both arrays
3663 // otherwise rustfmt suggest one-line it
3664 Text("text".into()),
3665 ]
3666 );
3667
3668 assert_eq!(de.next().unwrap(), Start(BytesStart::new("inner")));
3669 assert_eq!(de.next().unwrap(), End(BytesEnd::new("inner")));
3670
3671 // We finish writing. Next call to `next()` should start replay messages:
3672 //
3673 // text
3674 // </inner>
3675 //
3676 // and after that stream that messages:
3677 //
3678 // <target/>
3679 // </root>
3680 de.start_replay(checkpoint);
3681 assert_eq!(
3682 de.read,
3683 vec![
3684 // This comment here to keep the same formatting as others
3685 // otherwise rustfmt suggest one-line it
3686 Text("text".into()),
3687 End(BytesEnd::new("inner")),
3688 ]
3689 );
3690 assert_eq!(de.write, vec![]);
3691 assert_eq!(de.next().unwrap(), Text("text".into()));
3692 assert_eq!(de.next().unwrap(), End(BytesEnd::new("inner")));
3693 assert_eq!(de.next().unwrap(), Start(BytesStart::new("target")));
3694 assert_eq!(de.next().unwrap(), End(BytesEnd::new("target")));
3695 assert_eq!(de.next().unwrap(), End(BytesEnd::new("root")));
3696 assert_eq!(de.next().unwrap(), Eof);
3697 }
3698
3699 /// Checks that `read_to_end()` behaves correctly after `skip()`
3700 #[test]
3701 fn read_to_end() {
3702 let mut de = make_de(
3703 "\
3704 <root>\
3705 <skip>\
3706 text\
3707 <skip/>\
3708 </skip>\
3709 <target>\
3710 <target/>\
3711 </target>\
3712 </root>\
3713 ",
3714 );
3715
3716 // Initial conditions - both are empty
3717 assert_eq!(de.read, vec![]);
3718 assert_eq!(de.write, vec![]);
3719
3720 assert_eq!(de.next().unwrap(), Start(BytesStart::new("root")));
3721
3722 // Mark that start_replay() should begin replay from this point
3723 let checkpoint = de.skip_checkpoint();
3724 assert_eq!(checkpoint, 0);
3725
3726 // Skip the <skip> tree
3727 de.skip().unwrap();
3728 assert_eq!(de.read, vec![]);
3729 assert_eq!(
3730 de.write,
3731 vec![
3732 Start(BytesStart::new("skip")),
3733 Text("text".into()),
3734 Start(BytesStart::new("skip")),
3735 End(BytesEnd::new("skip")),
3736 End(BytesEnd::new("skip")),
3737 ]
3738 );
3739
3740 // Drop all events that represents <target> tree. Now unconsumed XML looks like:
3741 //
3742 // <skip>
3743 // text
3744 // <skip/>
3745 // </skip>
3746 // </root>
3747 assert_eq!(de.next().unwrap(), Start(BytesStart::new("target")));
3748 de.read_to_end(QName(b"target")).unwrap();
3749 assert_eq!(de.read, vec![]);
3750 assert_eq!(
3751 de.write,
3752 vec![
3753 Start(BytesStart::new("skip")),
3754 Text("text".into()),
3755 Start(BytesStart::new("skip")),
3756 End(BytesEnd::new("skip")),
3757 End(BytesEnd::new("skip")),
3758 ]
3759 );
3760
3761 // We finish writing. Next call to `next()` should start replay that messages:
3762 //
3763 // <skip>
3764 // text
3765 // <skip/>
3766 // </skip>
3767 //
3768 // and after that stream that messages:
3769 //
3770 // </root>
3771 de.start_replay(checkpoint);
3772 assert_eq!(
3773 de.read,
3774 vec![
3775 Start(BytesStart::new("skip")),
3776 Text("text".into()),
3777 Start(BytesStart::new("skip")),
3778 End(BytesEnd::new("skip")),
3779 End(BytesEnd::new("skip")),
3780 ]
3781 );
3782 assert_eq!(de.write, vec![]);
3783
3784 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skip")));
3785 de.read_to_end(QName(b"skip")).unwrap();
3786
3787 assert_eq!(de.next().unwrap(), End(BytesEnd::new("root")));
3788 assert_eq!(de.next().unwrap(), Eof);
3789 }
3790
3791 /// Checks that replay replayes only part of events
3792 /// Test for https://github.com/tafia/quick-xml/issues/435
3793 #[test]
3794 fn partial_replay() {
3795 let mut de = make_de(
3796 "\
3797 <root>\
3798 <skipped-1/>\
3799 <skipped-2/>\
3800 <inner>\
3801 <skipped-3/>\
3802 <skipped-4/>\
3803 <target-2/>\
3804 </inner>\
3805 <target-1/>\
3806 </root>\
3807 ",
3808 );
3809
3810 // Initial conditions - both are empty
3811 assert_eq!(de.read, vec![]);
3812 assert_eq!(de.write, vec![]);
3813
3814 assert_eq!(de.next().unwrap(), Start(BytesStart::new("root")));
3815
3816 // start_replay() should start replay from this point
3817 let checkpoint1 = de.skip_checkpoint();
3818 assert_eq!(checkpoint1, 0);
3819
3820 // Should skip first and second <skipped-N/> elements
3821 de.skip().unwrap(); // skipped-1
3822 de.skip().unwrap(); // skipped-2
3823 assert_eq!(de.read, vec![]);
3824 assert_eq!(
3825 de.write,
3826 vec![
3827 Start(BytesStart::new("skipped-1")),
3828 End(BytesEnd::new("skipped-1")),
3829 Start(BytesStart::new("skipped-2")),
3830 End(BytesEnd::new("skipped-2")),
3831 ]
3832 );
3833
3834 ////////////////////////////////////////////////////////////////////////////////////////
3835
3836 assert_eq!(de.next().unwrap(), Start(BytesStart::new("inner")));
3837 assert_eq!(de.peek().unwrap(), &Start(BytesStart::new("skipped-3")));
3838 assert_eq!(
3839 de.read,
3840 vec![
3841 // This comment here to keep the same formatting of both arrays
3842 // otherwise rustfmt suggest one-line it
3843 Start(BytesStart::new("skipped-3")),
3844 ]
3845 );
3846 assert_eq!(
3847 de.write,
3848 vec![
3849 Start(BytesStart::new("skipped-1")),
3850 End(BytesEnd::new("skipped-1")),
3851 Start(BytesStart::new("skipped-2")),
3852 End(BytesEnd::new("skipped-2")),
3853 ]
3854 );
3855
3856 // start_replay() should start replay from this point
3857 let checkpoint2 = de.skip_checkpoint();
3858 assert_eq!(checkpoint2, 4);
3859
3860 // Should skip third and forth <skipped-N/> elements
3861 de.skip().unwrap(); // skipped-3
3862 de.skip().unwrap(); // skipped-4
3863 assert_eq!(de.read, vec![]);
3864 assert_eq!(
3865 de.write,
3866 vec![
3867 // checkpoint 1
3868 Start(BytesStart::new("skipped-1")),
3869 End(BytesEnd::new("skipped-1")),
3870 Start(BytesStart::new("skipped-2")),
3871 End(BytesEnd::new("skipped-2")),
3872 // checkpoint 2
3873 Start(BytesStart::new("skipped-3")),
3874 End(BytesEnd::new("skipped-3")),
3875 Start(BytesStart::new("skipped-4")),
3876 End(BytesEnd::new("skipped-4")),
3877 ]
3878 );
3879 assert_eq!(de.next().unwrap(), Start(BytesStart::new("target-2")));
3880 assert_eq!(de.next().unwrap(), End(BytesEnd::new("target-2")));
3881 assert_eq!(de.peek().unwrap(), &End(BytesEnd::new("inner")));
3882 assert_eq!(
3883 de.read,
3884 vec![
3885 // This comment here to keep the same formatting of both arrays
3886 // otherwise rustfmt suggest one-line it
3887 End(BytesEnd::new("inner")),
3888 ]
3889 );
3890 assert_eq!(
3891 de.write,
3892 vec![
3893 // checkpoint 1
3894 Start(BytesStart::new("skipped-1")),
3895 End(BytesEnd::new("skipped-1")),
3896 Start(BytesStart::new("skipped-2")),
3897 End(BytesEnd::new("skipped-2")),
3898 // checkpoint 2
3899 Start(BytesStart::new("skipped-3")),
3900 End(BytesEnd::new("skipped-3")),
3901 Start(BytesStart::new("skipped-4")),
3902 End(BytesEnd::new("skipped-4")),
3903 ]
3904 );
3905
3906 // Start replay events from checkpoint 2
3907 de.start_replay(checkpoint2);
3908 assert_eq!(
3909 de.read,
3910 vec![
3911 Start(BytesStart::new("skipped-3")),
3912 End(BytesEnd::new("skipped-3")),
3913 Start(BytesStart::new("skipped-4")),
3914 End(BytesEnd::new("skipped-4")),
3915 End(BytesEnd::new("inner")),
3916 ]
3917 );
3918 assert_eq!(
3919 de.write,
3920 vec![
3921 Start(BytesStart::new("skipped-1")),
3922 End(BytesEnd::new("skipped-1")),
3923 Start(BytesStart::new("skipped-2")),
3924 End(BytesEnd::new("skipped-2")),
3925 ]
3926 );
3927
3928 // Replayed events
3929 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skipped-3")));
3930 assert_eq!(de.next().unwrap(), End(BytesEnd::new("skipped-3")));
3931 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skipped-4")));
3932 assert_eq!(de.next().unwrap(), End(BytesEnd::new("skipped-4")));
3933
3934 assert_eq!(de.next().unwrap(), End(BytesEnd::new("inner")));
3935 assert_eq!(de.read, vec![]);
3936 assert_eq!(
3937 de.write,
3938 vec![
3939 Start(BytesStart::new("skipped-1")),
3940 End(BytesEnd::new("skipped-1")),
3941 Start(BytesStart::new("skipped-2")),
3942 End(BytesEnd::new("skipped-2")),
3943 ]
3944 );
3945
3946 ////////////////////////////////////////////////////////////////////////////////////////
3947
3948 // New events
3949 assert_eq!(de.next().unwrap(), Start(BytesStart::new("target-1")));
3950 assert_eq!(de.next().unwrap(), End(BytesEnd::new("target-1")));
3951
3952 assert_eq!(de.read, vec![]);
3953 assert_eq!(
3954 de.write,
3955 vec![
3956 Start(BytesStart::new("skipped-1")),
3957 End(BytesEnd::new("skipped-1")),
3958 Start(BytesStart::new("skipped-2")),
3959 End(BytesEnd::new("skipped-2")),
3960 ]
3961 );
3962
3963 // Start replay events from checkpoint 1
3964 de.start_replay(checkpoint1);
3965 assert_eq!(
3966 de.read,
3967 vec![
3968 Start(BytesStart::new("skipped-1")),
3969 End(BytesEnd::new("skipped-1")),
3970 Start(BytesStart::new("skipped-2")),
3971 End(BytesEnd::new("skipped-2")),
3972 ]
3973 );
3974 assert_eq!(de.write, vec![]);
3975
3976 // Replayed events
3977 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skipped-1")));
3978 assert_eq!(de.next().unwrap(), End(BytesEnd::new("skipped-1")));
3979 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skipped-2")));
3980 assert_eq!(de.next().unwrap(), End(BytesEnd::new("skipped-2")));
3981
3982 assert_eq!(de.read, vec![]);
3983 assert_eq!(de.write, vec![]);
3984
3985 // New events
3986 assert_eq!(de.next().unwrap(), End(BytesEnd::new("root")));
3987 assert_eq!(de.next().unwrap(), Eof);
3988 }
3989
3990 /// Checks that limiting buffer size works correctly
3991 #[test]
3992 fn limit() {
3993 use serde::Deserialize;
3994
3995 #[derive(Debug, Deserialize)]
3996 #[allow(unused)]
3997 struct List {
3998 item: Vec<()>,
3999 }
4000
4001 let mut de = make_de(
4002 "\
4003 <any-name>\
4004 <item/>\
4005 <another-item>\
4006 <some-element>with text</some-element>\
4007 <yet-another-element/>\
4008 </another-item>\
4009 <item/>\
4010 <item/>\
4011 </any-name>\
4012 ",
4013 );
4014 de.event_buffer_size(NonZeroUsize::new(3));
4015
4016 match List::deserialize(&mut de) {
4017 Err(DeError::TooManyEvents(count)) => assert_eq!(count.get(), 3),
4018 e => panic!("Expected `Err(TooManyEvents(3))`, but got `{:?}`", e),
4019 }
4020 }
4021
4022 /// Without handling Eof in `skip` this test failed with memory allocation
4023 #[test]
4024 fn invalid_xml() {
4025 use crate::de::DeEvent::*;
4026
4027 let mut de = make_de("<root>");
4028
4029 // Cache all events
4030 let checkpoint = de.skip_checkpoint();
4031 de.skip().unwrap();
4032 de.start_replay(checkpoint);
4033 assert_eq!(de.read, vec![Start(BytesStart::new("root")), Eof]);
4034 }
4035 }
4036
4037 mod read_to_end {
4038 use super::*;
4039 use crate::de::DeEvent::*;
4040 use pretty_assertions::assert_eq;
4041
4042 #[test]
4043 fn complex() {
4044 let mut de = make_de(
4045 r#"
4046 <root>
4047 <tag a="1"><tag>text</tag>content</tag>
4048 <tag a="2"><![CDATA[cdata content]]></tag>
4049 <self-closed/>
4050 </root>
4051 "#,
4052 );
4053
4054 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4055 assert_eq!(de.next().unwrap(), Start(BytesStart::new("root")));
4056
4057 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4058 assert_eq!(
4059 de.next().unwrap(),
4060 Start(BytesStart::from_content(r#"tag a="1""#, 3))
4061 );
4062 assert_eq!(de.read_to_end(QName(b"tag")).unwrap(), ());
4063
4064 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4065 assert_eq!(
4066 de.next().unwrap(),
4067 Start(BytesStart::from_content(r#"tag a="2""#, 3))
4068 );
4069 assert_eq!(de.next().unwrap(), Text("cdata content".into()));
4070 assert_eq!(de.next().unwrap(), End(BytesEnd::new("tag")));
4071
4072 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4073 assert_eq!(de.next().unwrap(), Start(BytesStart::new("self-closed")));
4074 assert_eq!(de.read_to_end(QName(b"self-closed")).unwrap(), ());
4075
4076 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4077 assert_eq!(de.next().unwrap(), End(BytesEnd::new("root")));
4078 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4079 assert_eq!(de.next().unwrap(), Eof);
4080 }
4081
4082 #[test]
4083 fn invalid_xml1() {
4084 let mut de = make_de("<tag><tag></tag>");
4085
4086 assert_eq!(de.next().unwrap(), Start(BytesStart::new("tag")));
4087 assert_eq!(de.peek().unwrap(), &Start(BytesStart::new("tag")));
4088
4089 match de.read_to_end(QName(b"tag")) {
4090 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4091 assert_eq!(cause, IllFormedError::MissingEndTag("tag".into()))
4092 }
4093 x => panic!(
4094 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4095 x
4096 ),
4097 }
4098 assert_eq!(de.next().unwrap(), Eof);
4099 }
4100
4101 #[test]
4102 fn invalid_xml2() {
4103 let mut de = make_de("<tag><![CDATA[]]><tag></tag>");
4104
4105 assert_eq!(de.next().unwrap(), Start(BytesStart::new("tag")));
4106 assert_eq!(de.peek().unwrap(), &Text("".into()));
4107
4108 match de.read_to_end(QName(b"tag")) {
4109 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4110 assert_eq!(cause, IllFormedError::MissingEndTag("tag".into()))
4111 }
4112 x => panic!(
4113 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4114 x
4115 ),
4116 }
4117 assert_eq!(de.next().unwrap(), Eof);
4118 }
4119 }
4120
4121 #[test]
4122 fn borrowing_reader_parity() {
4123 let s = r#"
4124 <item name="hello" source="world.rs">Some text</item>
4125 <item2/>
4126 <item3 value="world" />
4127 "#;
4128
4129 let mut reader1 = IoReader {
4130 reader: NsReader::from_reader(s.as_bytes()),
4131 buf: Vec::new(),
4132 };
4133 let mut reader2 = SliceReader {
4134 reader: NsReader::from_str(s),
4135 };
4136
4137 loop {
4138 let event1 = reader1.next().unwrap();
4139 let event2 = reader2.next().unwrap();
4140
4141 if let (PayloadEvent::Eof, PayloadEvent::Eof) = (&event1, &event2) {
4142 break;
4143 }
4144
4145 assert_eq!(event1, event2);
4146 }
4147 }
4148
4149 #[test]
4150 fn borrowing_reader_events() {
4151 let s = r#"
4152 <item name="hello" source="world.rs">Some text</item>
4153 <item2></item2>
4154 <item3/>
4155 <item4 value="world" />
4156 "#;
4157
4158 let mut reader = SliceReader {
4159 reader: NsReader::from_str(s),
4160 };
4161
4162 let config = reader.reader.config_mut();
4163 config.expand_empty_elements = true;
4164
4165 let mut events = Vec::new();
4166
4167 loop {
4168 let event = reader.next().unwrap();
4169 if let PayloadEvent::Eof = event {
4170 break;
4171 }
4172 events.push(event);
4173 }
4174
4175 use crate::de::PayloadEvent::*;
4176
4177 assert_eq!(
4178 events,
4179 vec![
4180 Text(BytesText::from_escaped("\n ")),
4181 Start(BytesStart::from_content(
4182 r#"item name="hello" source="world.rs""#,
4183 4
4184 )),
4185 Text(BytesText::from_escaped("Some text")),
4186 End(BytesEnd::new("item")),
4187 Text(BytesText::from_escaped("\n ")),
4188 Start(BytesStart::from_content("item2", 5)),
4189 End(BytesEnd::new("item2")),
4190 Text(BytesText::from_escaped("\n ")),
4191 Start(BytesStart::from_content("item3", 5)),
4192 End(BytesEnd::new("item3")),
4193 Text(BytesText::from_escaped("\n ")),
4194 Start(BytesStart::from_content(r#"item4 value="world" "#, 5)),
4195 End(BytesEnd::new("item4")),
4196 Text(BytesText::from_escaped("\n ")),
4197 ]
4198 )
4199 }
4200
4201 /// Ensures, that [`Deserializer::read_string()`] never can get an `End` event,
4202 /// because parser reports error early
4203 #[test]
4204 fn read_string() {
4205 match from_str::<String>(r#"</root>"#) {
4206 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4207 assert_eq!(cause, IllFormedError::UnmatchedEndTag("root".into()));
4208 }
4209 x => panic!(
4210 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4211 x
4212 ),
4213 }
4214
4215 let s: String = from_str(r#"<root></root>"#).unwrap();
4216 assert_eq!(s, "");
4217
4218 match from_str::<String>(r#"<root></other>"#) {
4219 Err(DeError::InvalidXml(Error::IllFormed(cause))) => assert_eq!(
4220 cause,
4221 IllFormedError::MismatchedEndTag {
4222 expected: "root".into(),
4223 found: "other".into(),
4224 }
4225 ),
4226 x => panic!("Expected `Err(InvalidXml(IllFormed(_))`, but got `{:?}`", x),
4227 }
4228 }
4229
4230 /// Tests for https://github.com/tafia/quick-xml/issues/474.
4231 ///
4232 /// That tests ensures that comments and processed instructions is ignored
4233 /// and can split one logical string in pieces.
4234 mod merge_text {
4235 use super::*;
4236 use pretty_assertions::assert_eq;
4237
4238 #[test]
4239 fn text() {
4240 let mut de = make_de("text");
4241 assert_eq!(de.next().unwrap(), DeEvent::Text("text".into()));
4242 }
4243
4244 #[test]
4245 fn cdata() {
4246 let mut de = make_de("<![CDATA[cdata]]>");
4247 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata".into()));
4248 }
4249
4250 #[test]
4251 fn text_and_cdata() {
4252 let mut de = make_de("text and <![CDATA[cdata]]>");
4253 assert_eq!(de.next().unwrap(), DeEvent::Text("text and cdata".into()));
4254 }
4255
4256 #[test]
4257 fn text_and_empty_cdata() {
4258 let mut de = make_de("text and <![CDATA[]]>");
4259 assert_eq!(de.next().unwrap(), DeEvent::Text("text and ".into()));
4260 }
4261
4262 #[test]
4263 fn cdata_and_text() {
4264 let mut de = make_de("<![CDATA[cdata]]> and text");
4265 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata and text".into()));
4266 }
4267
4268 #[test]
4269 fn empty_cdata_and_text() {
4270 let mut de = make_de("<![CDATA[]]> and text");
4271 assert_eq!(de.next().unwrap(), DeEvent::Text(" and text".into()));
4272 }
4273
4274 #[test]
4275 fn cdata_and_cdata() {
4276 let mut de = make_de(
4277 "\
4278 <![CDATA[cdata]]]]>\
4279 <![CDATA[>cdata]]>\
4280 ",
4281 );
4282 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4283 }
4284
4285 mod comment_between {
4286 use super::*;
4287 use pretty_assertions::assert_eq;
4288
4289 #[test]
4290 fn text() {
4291 let mut de = make_de(
4292 "\
4293 text \
4294 <!--comment 1--><!--comment 2--> \
4295 text\
4296 ",
4297 );
4298 assert_eq!(de.next().unwrap(), DeEvent::Text("text text".into()));
4299 }
4300
4301 #[test]
4302 fn cdata() {
4303 let mut de = make_de(
4304 "\
4305 <![CDATA[cdata]]]]>\
4306 <!--comment 1--><!--comment 2-->\
4307 <![CDATA[>cdata]]>\
4308 ",
4309 );
4310 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4311 }
4312
4313 #[test]
4314 fn text_and_cdata() {
4315 let mut de = make_de(
4316 "\
4317 text \
4318 <!--comment 1--><!--comment 2-->\
4319 <![CDATA[ cdata]]>\
4320 ",
4321 );
4322 assert_eq!(de.next().unwrap(), DeEvent::Text("text cdata".into()));
4323 }
4324
4325 #[test]
4326 fn text_and_empty_cdata() {
4327 let mut de = make_de(
4328 "\
4329 text \
4330 <!--comment 1--><!--comment 2-->\
4331 <![CDATA[]]>\
4332 ",
4333 );
4334 assert_eq!(de.next().unwrap(), DeEvent::Text("text ".into()));
4335 }
4336
4337 #[test]
4338 fn cdata_and_text() {
4339 let mut de = make_de(
4340 "\
4341 <![CDATA[cdata ]]>\
4342 <!--comment 1--><!--comment 2--> \
4343 text \
4344 ",
4345 );
4346 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata text ".into()));
4347 }
4348
4349 #[test]
4350 fn empty_cdata_and_text() {
4351 let mut de = make_de(
4352 "\
4353 <![CDATA[]]>\
4354 <!--comment 1--><!--comment 2--> \
4355 text \
4356 ",
4357 );
4358 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4359 }
4360
4361 #[test]
4362 fn cdata_and_cdata() {
4363 let mut de = make_de(
4364 "\
4365 <![CDATA[cdata]]]>\
4366 <!--comment 1--><!--comment 2-->\
4367 <![CDATA[]>cdata]]>\
4368 ",
4369 );
4370 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4371 }
4372 }
4373
4374 mod pi_between {
4375 use super::*;
4376 use pretty_assertions::assert_eq;
4377
4378 #[test]
4379 fn text() {
4380 let mut de = make_de(
4381 "\
4382 text \
4383 <?pi 1?><?pi 2?> \
4384 text\
4385 ",
4386 );
4387 assert_eq!(de.next().unwrap(), DeEvent::Text("text text".into()));
4388 }
4389
4390 #[test]
4391 fn cdata() {
4392 let mut de = make_de(
4393 "\
4394 <![CDATA[cdata]]]]>\
4395 <?pi 1?><?pi 2?>\
4396 <![CDATA[>cdata]]>\
4397 ",
4398 );
4399 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4400 }
4401
4402 #[test]
4403 fn text_and_cdata() {
4404 let mut de = make_de(
4405 "\
4406 text \
4407 <?pi 1?><?pi 2?>\
4408 <![CDATA[ cdata]]>\
4409 ",
4410 );
4411 assert_eq!(de.next().unwrap(), DeEvent::Text("text cdata".into()));
4412 }
4413
4414 #[test]
4415 fn text_and_empty_cdata() {
4416 let mut de = make_de(
4417 "\
4418 text \
4419 <?pi 1?><?pi 2?>\
4420 <![CDATA[]]>\
4421 ",
4422 );
4423 assert_eq!(de.next().unwrap(), DeEvent::Text("text ".into()));
4424 }
4425
4426 #[test]
4427 fn cdata_and_text() {
4428 let mut de = make_de(
4429 "\
4430 <![CDATA[cdata ]]>\
4431 <?pi 1?><?pi 2?> \
4432 text \
4433 ",
4434 );
4435 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata text ".into()));
4436 }
4437
4438 #[test]
4439 fn empty_cdata_and_text() {
4440 let mut de = make_de(
4441 "\
4442 <![CDATA[]]>\
4443 <?pi 1?><?pi 2?> \
4444 text \
4445 ",
4446 );
4447 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4448 }
4449
4450 #[test]
4451 fn cdata_and_cdata() {
4452 let mut de = make_de(
4453 "\
4454 <![CDATA[cdata]]]>\
4455 <?pi 1?><?pi 2?>\
4456 <![CDATA[]>cdata]]>\
4457 ",
4458 );
4459 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4460 }
4461 }
4462 }
4463
4464 /// Tests for https://github.com/tafia/quick-xml/issues/474.
4465 ///
4466 /// This tests ensures that any combination of payload data is processed
4467 /// as expected.
4468 mod triples {
4469 use super::*;
4470 use pretty_assertions::assert_eq;
4471
4472 mod start {
4473 use super::*;
4474
4475 /// <tag1><tag2>...
4476 // The same name is intentional
4477 #[allow(clippy::module_inception)]
4478 mod start {
4479 use super::*;
4480 use pretty_assertions::assert_eq;
4481
4482 #[test]
4483 fn start() {
4484 let mut de = make_de("<tag1><tag2><tag3>");
4485 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4486 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4487 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag3")));
4488 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4489 }
4490
4491 /// Not matching end tag will result to error
4492 #[test]
4493 fn end() {
4494 let mut de = make_de("<tag1><tag2></tag2>");
4495 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4496 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4497 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag2")));
4498 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4499 }
4500
4501 #[test]
4502 fn text() {
4503 let mut de = make_de("<tag1><tag2> text ");
4504 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4505 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4506 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4507 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4508 }
4509
4510 #[test]
4511 fn cdata() {
4512 let mut de = make_de("<tag1><tag2><![CDATA[ cdata ]]>");
4513 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4514 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4515 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4516 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4517 }
4518
4519 #[test]
4520 fn eof() {
4521 let mut de = make_de("<tag1><tag2>");
4522 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4523 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4524 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4525 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4526 }
4527 }
4528
4529 /// <tag></tag>...
4530 mod end {
4531 use super::*;
4532 use pretty_assertions::assert_eq;
4533
4534 #[test]
4535 fn start() {
4536 let mut de = make_de("<tag></tag><tag2>");
4537 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4538 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4539 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4540 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4541 }
4542
4543 #[test]
4544 fn end() {
4545 let mut de = make_de("<tag></tag></tag2>");
4546 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4547 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4548 match de.next() {
4549 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4550 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag2".into()));
4551 }
4552 x => panic!(
4553 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4554 x
4555 ),
4556 }
4557 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4558 }
4559
4560 #[test]
4561 fn text() {
4562 let mut de = make_de("<tag></tag> text ");
4563 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4564 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4565 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4566 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4567 }
4568
4569 #[test]
4570 fn cdata() {
4571 let mut de = make_de("<tag></tag><![CDATA[ cdata ]]>");
4572 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4573 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4574 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4575 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4576 }
4577
4578 #[test]
4579 fn eof() {
4580 let mut de = make_de("<tag></tag>");
4581 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4582 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4583 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4584 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4585 }
4586 }
4587
4588 /// <tag> text ...
4589 mod text {
4590 use super::*;
4591 use pretty_assertions::assert_eq;
4592
4593 #[test]
4594 fn start() {
4595 let mut de = make_de("<tag> text <tag2>");
4596 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4597 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4598 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4599 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4600 }
4601
4602 #[test]
4603 fn end() {
4604 let mut de = make_de("<tag> text </tag>");
4605 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4606 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4607 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4608 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4609 }
4610
4611 // start::text::text has no difference from start::text
4612
4613 #[test]
4614 fn cdata() {
4615 let mut de = make_de("<tag> text <![CDATA[ cdata ]]>");
4616 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4617 assert_eq!(de.next().unwrap(), DeEvent::Text(" text cdata ".into()));
4618 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4619 }
4620
4621 #[test]
4622 fn eof() {
4623 let mut de = make_de("<tag> text ");
4624 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4625 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4626 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4627 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4628 }
4629 }
4630
4631 /// <tag><![CDATA[ cdata ]]>...
4632 mod cdata {
4633 use super::*;
4634 use pretty_assertions::assert_eq;
4635
4636 #[test]
4637 fn start() {
4638 let mut de = make_de("<tag><![CDATA[ cdata ]]><tag2>");
4639 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4640 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4641 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4642 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4643 }
4644
4645 #[test]
4646 fn end() {
4647 let mut de = make_de("<tag><![CDATA[ cdata ]]></tag>");
4648 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4649 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4650 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4651 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4652 }
4653
4654 #[test]
4655 fn text() {
4656 let mut de = make_de("<tag><![CDATA[ cdata ]]> text ");
4657 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4658 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata text ".into()));
4659 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4660 }
4661
4662 #[test]
4663 fn cdata() {
4664 let mut de = make_de("<tag><![CDATA[ cdata ]]><![CDATA[ cdata2 ]]>");
4665 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4666 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata cdata2 ".into()));
4667 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4668 }
4669
4670 #[test]
4671 fn eof() {
4672 let mut de = make_de("<tag><![CDATA[ cdata ]]>");
4673 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4674 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4675 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4676 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4677 }
4678 }
4679 }
4680
4681 /// Start from End event will always generate an error
4682 #[test]
4683 fn end() {
4684 let mut de = make_de("</tag>");
4685 match de.next() {
4686 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4687 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4688 }
4689 x => panic!(
4690 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4691 x
4692 ),
4693 }
4694 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4695 }
4696
4697 mod text {
4698 use super::*;
4699 use pretty_assertions::assert_eq;
4700
4701 mod start {
4702 use super::*;
4703 use pretty_assertions::assert_eq;
4704
4705 #[test]
4706 fn start() {
4707 let mut de = make_de(" text <tag1><tag2>");
4708 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4709 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4710 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4711 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4712 }
4713
4714 /// Not matching end tag will result in error
4715 #[test]
4716 fn end() {
4717 let mut de = make_de(" text <tag></tag>");
4718 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4719 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4720 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4721 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4722 }
4723
4724 #[test]
4725 fn text() {
4726 let mut de = make_de(" text <tag> text2 ");
4727 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4728 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4729 assert_eq!(de.next().unwrap(), DeEvent::Text(" text2 ".into()));
4730 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4731 }
4732
4733 #[test]
4734 fn cdata() {
4735 let mut de = make_de(" text <tag><![CDATA[ cdata ]]>");
4736 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4737 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4738 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4739 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4740 }
4741
4742 #[test]
4743 fn eof() {
4744 let mut de = make_de(" text <tag>");
4745 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4746 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4747 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4748 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4749 }
4750 }
4751
4752 /// End event without corresponding start event will always generate an error
4753 #[test]
4754 fn end() {
4755 let mut de = make_de(" text </tag>");
4756 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4757 match de.next() {
4758 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4759 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4760 }
4761 x => panic!(
4762 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4763 x
4764 ),
4765 }
4766 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4767 }
4768
4769 // text::text::something is equivalent to text::something
4770
4771 mod cdata {
4772 use super::*;
4773 use pretty_assertions::assert_eq;
4774
4775 #[test]
4776 fn start() {
4777 let mut de = make_de(" text <![CDATA[ cdata ]]><tag>");
4778 assert_eq!(de.next().unwrap(), DeEvent::Text(" text cdata ".into()));
4779 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4780 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4781 }
4782
4783 #[test]
4784 fn end() {
4785 let mut de = make_de(" text <![CDATA[ cdata ]]></tag>");
4786 assert_eq!(de.next().unwrap(), DeEvent::Text(" text cdata ".into()));
4787 match de.next() {
4788 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4789 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4790 }
4791 x => panic!(
4792 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4793 x
4794 ),
4795 }
4796 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4797 }
4798
4799 #[test]
4800 fn text() {
4801 let mut de = make_de(" text <![CDATA[ cdata ]]> text2 ");
4802 assert_eq!(
4803 de.next().unwrap(),
4804 DeEvent::Text(" text cdata text2 ".into())
4805 );
4806 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4807 }
4808
4809 #[test]
4810 fn cdata() {
4811 let mut de = make_de(" text <![CDATA[ cdata ]]><![CDATA[ cdata2 ]]>");
4812 assert_eq!(
4813 de.next().unwrap(),
4814 DeEvent::Text(" text cdata cdata2 ".into())
4815 );
4816 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4817 }
4818
4819 #[test]
4820 fn eof() {
4821 let mut de = make_de(" text <![CDATA[ cdata ]]>");
4822 assert_eq!(de.next().unwrap(), DeEvent::Text(" text cdata ".into()));
4823 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4824 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4825 }
4826 }
4827 }
4828
4829 mod cdata {
4830 use super::*;
4831 use pretty_assertions::assert_eq;
4832
4833 mod start {
4834 use super::*;
4835 use pretty_assertions::assert_eq;
4836
4837 #[test]
4838 fn start() {
4839 let mut de = make_de("<![CDATA[ cdata ]]><tag1><tag2>");
4840 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4841 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4842 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4843 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4844 }
4845
4846 /// Not matching end tag will result in error
4847 #[test]
4848 fn end() {
4849 let mut de = make_de("<![CDATA[ cdata ]]><tag></tag>");
4850 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4851 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4852 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4853 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4854 }
4855
4856 #[test]
4857 fn text() {
4858 let mut de = make_de("<![CDATA[ cdata ]]><tag> text ");
4859 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4860 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4861 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4862 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4863 }
4864
4865 #[test]
4866 fn cdata() {
4867 let mut de = make_de("<![CDATA[ cdata ]]><tag><![CDATA[ cdata2 ]]>");
4868 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4869 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4870 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata2 ".into()));
4871 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4872 }
4873
4874 #[test]
4875 fn eof() {
4876 let mut de = make_de("<![CDATA[ cdata ]]><tag>");
4877 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4878 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4879 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4880 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4881 }
4882 }
4883
4884 /// End event without corresponding start event will always generate an error
4885 #[test]
4886 fn end() {
4887 let mut de = make_de("<![CDATA[ cdata ]]></tag>");
4888 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4889 match de.next() {
4890 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4891 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4892 }
4893 x => panic!(
4894 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4895 x
4896 ),
4897 }
4898 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4899 }
4900
4901 mod text {
4902 use super::*;
4903 use pretty_assertions::assert_eq;
4904
4905 #[test]
4906 fn start() {
4907 let mut de = make_de("<![CDATA[ cdata ]]> text <tag>");
4908 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata text ".into()));
4909 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4910 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4911 }
4912
4913 #[test]
4914 fn end() {
4915 let mut de = make_de("<![CDATA[ cdata ]]> text </tag>");
4916 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata text ".into()));
4917 match de.next() {
4918 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4919 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4920 }
4921 x => panic!(
4922 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4923 x
4924 ),
4925 }
4926 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4927 }
4928
4929 // cdata::text::text is equivalent to cdata::text
4930
4931 #[test]
4932 fn cdata() {
4933 let mut de = make_de("<![CDATA[ cdata ]]> text <![CDATA[ cdata2 ]]>");
4934 assert_eq!(
4935 de.next().unwrap(),
4936 DeEvent::Text(" cdata text cdata2 ".into())
4937 );
4938 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4939 }
4940
4941 #[test]
4942 fn eof() {
4943 let mut de = make_de("<![CDATA[ cdata ]]> text ");
4944 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata text ".into()));
4945 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4946 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4947 }
4948 }
4949
4950 // The same name is intentional
4951 #[allow(clippy::module_inception)]
4952 mod cdata {
4953 use super::*;
4954 use pretty_assertions::assert_eq;
4955
4956 #[test]
4957 fn start() {
4958 let mut de = make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]><tag>");
4959 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata cdata2 ".into()));
4960 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4961 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4962 }
4963
4964 #[test]
4965 fn end() {
4966 let mut de = make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]></tag>");
4967 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata cdata2 ".into()));
4968 match de.next() {
4969 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4970 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4971 }
4972 x => panic!(
4973 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4974 x
4975 ),
4976 }
4977 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4978 }
4979
4980 #[test]
4981 fn text() {
4982 let mut de = make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]> text ");
4983 assert_eq!(
4984 de.next().unwrap(),
4985 DeEvent::Text(" cdata cdata2 text ".into())
4986 );
4987 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4988 }
4989
4990 #[test]
4991 fn cdata() {
4992 let mut de =
4993 make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]><![CDATA[ cdata3 ]]>");
4994 assert_eq!(
4995 de.next().unwrap(),
4996 DeEvent::Text(" cdata cdata2 cdata3 ".into())
4997 );
4998 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4999 }
5000
5001 #[test]
5002 fn eof() {
5003 let mut de = make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]>");
5004 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata cdata2 ".into()));
5005 assert_eq!(de.next().unwrap(), DeEvent::Eof);
5006 assert_eq!(de.next().unwrap(), DeEvent::Eof);
5007 }
5008 }
5009 }
5010 }
5011}