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
1976macro_rules! forward_to_simple_type {
1977 ($deserialize:ident, $($mut:tt)?) => {
1978 #[inline]
1979 fn $deserialize<V>($($mut)? self, visitor: V) -> Result<V::Value, DeError>
1980 where
1981 V: Visitor<'de>,
1982 {
1983 SimpleTypeDeserializer::from_text(self.read_string()?).$deserialize(visitor)
1984 }
1985 };
1986}
1987
1988/// Implement deserialization methods for scalar types, such as numbers, strings,
1989/// byte arrays, booleans and identifiers.
1990macro_rules! deserialize_primitives {
1991 ($($mut:tt)?) => {
1992 forward_to_simple_type!(deserialize_i8, $($mut)?);
1993 forward_to_simple_type!(deserialize_i16, $($mut)?);
1994 forward_to_simple_type!(deserialize_i32, $($mut)?);
1995 forward_to_simple_type!(deserialize_i64, $($mut)?);
1996
1997 forward_to_simple_type!(deserialize_u8, $($mut)?);
1998 forward_to_simple_type!(deserialize_u16, $($mut)?);
1999 forward_to_simple_type!(deserialize_u32, $($mut)?);
2000 forward_to_simple_type!(deserialize_u64, $($mut)?);
2001
2002 forward_to_simple_type!(deserialize_i128, $($mut)?);
2003 forward_to_simple_type!(deserialize_u128, $($mut)?);
2004
2005 forward_to_simple_type!(deserialize_f32, $($mut)?);
2006 forward_to_simple_type!(deserialize_f64, $($mut)?);
2007
2008 forward_to_simple_type!(deserialize_bool, $($mut)?);
2009 forward_to_simple_type!(deserialize_char, $($mut)?);
2010
2011 forward_to_simple_type!(deserialize_str, $($mut)?);
2012 forward_to_simple_type!(deserialize_string, $($mut)?);
2013
2014 /// Forwards deserialization to the [`deserialize_any`](#method.deserialize_any).
2015 #[inline]
2016 fn deserialize_bytes<V>(self, visitor: V) -> Result<V::Value, DeError>
2017 where
2018 V: Visitor<'de>,
2019 {
2020 self.deserialize_any(visitor)
2021 }
2022
2023 /// Forwards deserialization to the [`deserialize_bytes`](#method.deserialize_bytes).
2024 #[inline]
2025 fn deserialize_byte_buf<V>(self, visitor: V) -> Result<V::Value, DeError>
2026 where
2027 V: Visitor<'de>,
2028 {
2029 self.deserialize_bytes(visitor)
2030 }
2031
2032 /// Representation of the named units the same as [unnamed units](#method.deserialize_unit).
2033 #[inline]
2034 fn deserialize_unit_struct<V>(
2035 self,
2036 _name: &'static str,
2037 visitor: V,
2038 ) -> Result<V::Value, DeError>
2039 where
2040 V: Visitor<'de>,
2041 {
2042 self.deserialize_unit(visitor)
2043 }
2044
2045 /// Representation of tuples the same as [sequences](#method.deserialize_seq).
2046 #[inline]
2047 fn deserialize_tuple<V>(self, _len: usize, visitor: V) -> Result<V::Value, DeError>
2048 where
2049 V: Visitor<'de>,
2050 {
2051 self.deserialize_seq(visitor)
2052 }
2053
2054 /// Representation of named tuples the same as [unnamed tuples](#method.deserialize_tuple).
2055 #[inline]
2056 fn deserialize_tuple_struct<V>(
2057 self,
2058 _name: &'static str,
2059 len: usize,
2060 visitor: V,
2061 ) -> Result<V::Value, DeError>
2062 where
2063 V: Visitor<'de>,
2064 {
2065 self.deserialize_tuple(len, visitor)
2066 }
2067
2068 /// Forwards deserialization to the [`deserialize_struct`](#method.deserialize_struct)
2069 /// with empty name and fields.
2070 #[inline]
2071 fn deserialize_map<V>(self, visitor: V) -> Result<V::Value, DeError>
2072 where
2073 V: Visitor<'de>,
2074 {
2075 self.deserialize_struct("", &[], visitor)
2076 }
2077
2078 /// Identifiers represented as [strings](#method.deserialize_str).
2079 #[inline]
2080 fn deserialize_identifier<V>(self, visitor: V) -> Result<V::Value, DeError>
2081 where
2082 V: Visitor<'de>,
2083 {
2084 self.deserialize_str(visitor)
2085 }
2086
2087 /// Forwards deserialization to the [`deserialize_unit`](#method.deserialize_unit).
2088 #[inline]
2089 fn deserialize_ignored_any<V>(self, visitor: V) -> Result<V::Value, DeError>
2090 where
2091 V: Visitor<'de>,
2092 {
2093 self.deserialize_unit(visitor)
2094 }
2095 };
2096}
2097
2098mod attributes;
2099mod key;
2100mod map;
2101mod resolver;
2102mod simple_type;
2103mod text;
2104mod var;
2105
2106pub use self::attributes::AttributesDeserializer;
2107pub use self::resolver::{EntityResolver, PredefinedEntityResolver};
2108pub use self::simple_type::SimpleTypeDeserializer;
2109pub use crate::errors::serialize::DeError;
2110
2111use crate::{
2112 de::map::ElementMapAccess,
2113 encoding::Decoder,
2114 errors::Error,
2115 escape::{parse_number, EscapeError},
2116 events::{BytesCData, BytesEnd, BytesRef, BytesStart, BytesText, Event},
2117 name::QName,
2118 reader::NsReader,
2119};
2120use serde::de::{
2121 self, Deserialize, DeserializeOwned, DeserializeSeed, IntoDeserializer, SeqAccess, Visitor,
2122};
2123use std::borrow::Cow;
2124#[cfg(feature = "overlapped-lists")]
2125use std::collections::VecDeque;
2126use std::io::BufRead;
2127use std::mem::replace;
2128#[cfg(feature = "overlapped-lists")]
2129use std::num::NonZeroUsize;
2130use std::ops::{Deref, Range};
2131
2132/// Data represented by a text node or a CDATA node. XML markup is not expected
2133pub(crate) const TEXT_KEY: &str = "$text";
2134/// Data represented by any XML markup inside
2135pub(crate) const VALUE_KEY: &str = "$value";
2136
2137/// A function to check whether the character is a whitespace (blank, new line, carriage return or tab).
2138#[inline]
2139const fn is_non_whitespace(ch: char) -> bool {
2140 !matches!(ch, ' ' | '\r' | '\n' | '\t')
2141}
2142
2143/// Decoded and concatenated content of consequent [`Text`] and [`CData`]
2144/// events. _Consequent_ means that events should follow each other or be
2145/// delimited only by (any count of) [`Comment`] or [`PI`] events.
2146///
2147/// Internally text is stored in `Cow<str>`. Cloning of text is cheap while it
2148/// is borrowed and makes copies of data when it is owned.
2149///
2150/// [`Text`]: Event::Text
2151/// [`CData`]: Event::CData
2152/// [`Comment`]: Event::Comment
2153/// [`PI`]: Event::PI
2154#[derive(Clone, Debug, PartialEq, Eq)]
2155pub struct Text<'a> {
2156 /// Untrimmed text after concatenating content of all
2157 /// [`Text`] and [`CData`] events
2158 ///
2159 /// [`Text`]: Event::Text
2160 /// [`CData`]: Event::CData
2161 text: Cow<'a, str>,
2162 /// A range into `text` which contains data after trimming
2163 content: Range<usize>,
2164}
2165
2166impl<'a> Text<'a> {
2167 fn new(text: Cow<'a, str>) -> Self {
2168 let start = text.find(is_non_whitespace).unwrap_or(0);
2169 let end = text.rfind(is_non_whitespace).map_or(0, |i| i + 1);
2170
2171 let content = if start >= end { 0..0 } else { start..end };
2172
2173 Self { text, content }
2174 }
2175
2176 /// Returns text without leading and trailing whitespaces as [defined] by XML specification.
2177 ///
2178 /// If you want to only check if text contains only whitespaces, use [`is_blank`](Self::is_blank),
2179 /// which will not allocate.
2180 ///
2181 /// # Example
2182 ///
2183 /// ```
2184 /// # use quick_xml::de::Text;
2185 /// # use pretty_assertions::assert_eq;
2186 /// #
2187 /// let text = Text::from("");
2188 /// assert_eq!(text.trimmed(), "");
2189 ///
2190 /// let text = Text::from(" \r\n\t ");
2191 /// assert_eq!(text.trimmed(), "");
2192 ///
2193 /// let text = Text::from(" some useful text ");
2194 /// assert_eq!(text.trimmed(), "some useful text");
2195 /// ```
2196 ///
2197 /// [defined]: https://www.w3.org/TR/xml11/#NT-S
2198 pub fn trimmed(&self) -> Cow<'a, str> {
2199 match self.text {
2200 Cow::Borrowed(text) => Cow::Borrowed(&text[self.content.clone()]),
2201 Cow::Owned(ref text) => Cow::Owned(text[self.content.clone()].to_string()),
2202 }
2203 }
2204
2205 /// Returns `true` if text is empty or contains only whitespaces as [defined] by XML specification.
2206 ///
2207 /// # Example
2208 ///
2209 /// ```
2210 /// # use quick_xml::de::Text;
2211 /// # use pretty_assertions::assert_eq;
2212 /// #
2213 /// let text = Text::from("");
2214 /// assert_eq!(text.is_blank(), true);
2215 ///
2216 /// let text = Text::from(" \r\n\t ");
2217 /// assert_eq!(text.is_blank(), true);
2218 ///
2219 /// let text = Text::from(" some useful text ");
2220 /// assert_eq!(text.is_blank(), false);
2221 /// ```
2222 ///
2223 /// [defined]: https://www.w3.org/TR/xml11/#NT-S
2224 pub fn is_blank(&self) -> bool {
2225 self.content.is_empty()
2226 }
2227}
2228
2229impl<'a> Deref for Text<'a> {
2230 type Target = str;
2231
2232 #[inline]
2233 fn deref(&self) -> &Self::Target {
2234 self.text.deref()
2235 }
2236}
2237
2238impl<'a> From<&'a str> for Text<'a> {
2239 #[inline]
2240 fn from(text: &'a str) -> Self {
2241 Self::new(Cow::Borrowed(text))
2242 }
2243}
2244
2245impl<'a> From<String> for Text<'a> {
2246 #[inline]
2247 fn from(text: String) -> Self {
2248 Self::new(Cow::Owned(text))
2249 }
2250}
2251
2252impl<'a> From<Cow<'a, str>> for Text<'a> {
2253 #[inline]
2254 fn from(text: Cow<'a, str>) -> Self {
2255 Self::new(text)
2256 }
2257}
2258
2259////////////////////////////////////////////////////////////////////////////////////////////////////
2260
2261/// Simplified event which contains only these variants that used by deserializer
2262#[derive(Clone, Debug, PartialEq, Eq)]
2263pub enum DeEvent<'a> {
2264 /// Start tag (with attributes) `<tag attr="value">`.
2265 Start(BytesStart<'a>),
2266 /// End tag `</tag>`.
2267 End(BytesEnd<'a>),
2268 /// Decoded and concatenated content of consequent [`Text`] and [`CData`]
2269 /// events. _Consequent_ means that events should follow each other or be
2270 /// delimited only by (any count of) [`Comment`] or [`PI`] events.
2271 ///
2272 /// [`Text`]: Event::Text
2273 /// [`CData`]: Event::CData
2274 /// [`Comment`]: Event::Comment
2275 /// [`PI`]: Event::PI
2276 Text(Text<'a>),
2277 /// End of XML document.
2278 Eof,
2279}
2280
2281////////////////////////////////////////////////////////////////////////////////////////////////////
2282
2283/// Simplified event which contains only these variants that used by deserializer,
2284/// but [`Text`] events not yet fully processed.
2285///
2286/// [`Text`] events should be trimmed if they does not surrounded by the other
2287/// [`Text`] or [`CData`] events. This event contains intermediate state of [`Text`]
2288/// event, where they are trimmed from the start, but not from the end. To trim
2289/// end spaces we should lookahead by one deserializer event (i. e. skip all
2290/// comments and processing instructions).
2291///
2292/// [`Text`]: Event::Text
2293/// [`CData`]: Event::CData
2294#[derive(Clone, Debug, PartialEq, Eq)]
2295pub enum PayloadEvent<'a> {
2296 /// Start tag (with attributes) `<tag attr="value">`.
2297 Start(BytesStart<'a>),
2298 /// End tag `</tag>`.
2299 End(BytesEnd<'a>),
2300 /// Escaped character data between tags.
2301 Text(BytesText<'a>),
2302 /// Unescaped character data stored in `<![CDATA[...]]>`.
2303 CData(BytesCData<'a>),
2304 /// Document type definition data (DTD) stored in `<!DOCTYPE ...>`.
2305 DocType(BytesText<'a>),
2306 /// Reference `&ref;` in the textual data.
2307 GeneralRef(BytesRef<'a>),
2308 /// End of XML document.
2309 Eof,
2310}
2311
2312impl<'a> PayloadEvent<'a> {
2313 /// Ensures that all data is owned to extend the object's lifetime if necessary.
2314 #[inline]
2315 fn into_owned(self) -> PayloadEvent<'static> {
2316 match self {
2317 PayloadEvent::Start(e) => PayloadEvent::Start(e.into_owned()),
2318 PayloadEvent::End(e) => PayloadEvent::End(e.into_owned()),
2319 PayloadEvent::Text(e) => PayloadEvent::Text(e.into_owned()),
2320 PayloadEvent::CData(e) => PayloadEvent::CData(e.into_owned()),
2321 PayloadEvent::DocType(e) => PayloadEvent::DocType(e.into_owned()),
2322 PayloadEvent::GeneralRef(e) => PayloadEvent::GeneralRef(e.into_owned()),
2323 PayloadEvent::Eof => PayloadEvent::Eof,
2324 }
2325 }
2326}
2327
2328/// An intermediate reader that consumes [`PayloadEvent`]s and produces final [`DeEvent`]s.
2329/// [`PayloadEvent::Text`] events, that followed by any event except
2330/// [`PayloadEvent::Text`] or [`PayloadEvent::CData`], are trimmed from the end.
2331struct XmlReader<'i, R: XmlRead<'i>, E: EntityResolver = PredefinedEntityResolver> {
2332 /// A source of low-level XML events
2333 reader: R,
2334 /// Intermediate event, that could be returned by the next call to `next()`.
2335 /// If that is the `Text` event then leading spaces already trimmed, but
2336 /// trailing spaces is not. Before the event will be returned, trimming of
2337 /// the spaces could be necessary
2338 lookahead: Result<PayloadEvent<'i>, DeError>,
2339
2340 /// Used to resolve unknown entities that would otherwise cause the parser
2341 /// to return an [`EscapeError::UnrecognizedEntity`] error.
2342 ///
2343 /// [`EscapeError::UnrecognizedEntity`]: crate::escape::EscapeError::UnrecognizedEntity
2344 entity_resolver: E,
2345}
2346
2347impl<'i, R: XmlRead<'i>, E: EntityResolver> XmlReader<'i, R, E> {
2348 fn new(mut reader: R, entity_resolver: E) -> Self {
2349 // Lookahead by one event immediately, so we do not need to check in the
2350 // loop if we need lookahead or not
2351 let lookahead = reader.next();
2352
2353 Self {
2354 reader,
2355 lookahead,
2356 entity_resolver,
2357 }
2358 }
2359
2360 /// Returns `true` if all events was consumed
2361 const fn is_empty(&self) -> bool {
2362 matches!(self.lookahead, Ok(PayloadEvent::Eof))
2363 }
2364
2365 /// Read next event and put it in lookahead, return the current lookahead
2366 #[inline(always)]
2367 fn next_impl(&mut self) -> Result<PayloadEvent<'i>, DeError> {
2368 replace(&mut self.lookahead, self.reader.next())
2369 }
2370
2371 /// Returns `true` when next event is not a text event in any form.
2372 #[inline(always)]
2373 const fn current_event_is_last_text(&self) -> bool {
2374 // If next event is a text or CDATA, we should not trim trailing spaces
2375 !matches!(
2376 self.lookahead,
2377 Ok(PayloadEvent::Text(_)) | Ok(PayloadEvent::CData(_) | PayloadEvent::GeneralRef(_))
2378 )
2379 }
2380
2381 /// Read all consequent [`Text`] and [`CData`] events until non-text event
2382 /// occurs. Content of all events would be appended to `result` and returned
2383 /// as [`DeEvent::Text`].
2384 ///
2385 /// [`Text`]: PayloadEvent::Text
2386 /// [`CData`]: PayloadEvent::CData
2387 fn drain_text(&mut self, mut result: Cow<'i, str>) -> Result<DeEvent<'i>, DeError> {
2388 loop {
2389 if self.current_event_is_last_text() {
2390 break;
2391 }
2392
2393 match self.next_impl()? {
2394 PayloadEvent::Text(e) => result.to_mut().push_str(&e.xml_content()?),
2395 PayloadEvent::CData(e) => result.to_mut().push_str(&e.xml_content()?),
2396 PayloadEvent::GeneralRef(e) => self.resolve_reference(result.to_mut(), e)?,
2397
2398 // SAFETY: current_event_is_last_text checks that event is Text, CData or GeneralRef
2399 _ => unreachable!("Only `Text`, `CData` or `GeneralRef` events can come here"),
2400 }
2401 }
2402 Ok(DeEvent::Text(Text::new(result)))
2403 }
2404
2405 /// Return an input-borrowing event.
2406 fn next(&mut self) -> Result<DeEvent<'i>, DeError> {
2407 loop {
2408 return match self.next_impl()? {
2409 PayloadEvent::Start(e) => Ok(DeEvent::Start(e)),
2410 PayloadEvent::End(e) => Ok(DeEvent::End(e)),
2411 PayloadEvent::Text(e) => self.drain_text(e.xml_content()?),
2412 PayloadEvent::CData(e) => self.drain_text(e.xml_content()?),
2413 PayloadEvent::DocType(e) => {
2414 self.entity_resolver
2415 .capture(e)
2416 .map_err(|err| DeError::Custom(format!("cannot parse DTD: {}", err)))?;
2417 continue;
2418 }
2419 PayloadEvent::GeneralRef(e) => {
2420 let mut text = String::new();
2421 self.resolve_reference(&mut text, e)?;
2422 self.drain_text(text.into())
2423 }
2424 PayloadEvent::Eof => Ok(DeEvent::Eof),
2425 };
2426 }
2427 }
2428
2429 fn resolve_reference(&mut self, result: &mut String, event: BytesRef) -> Result<(), DeError> {
2430 let len = event.len();
2431 let reference = self.decoder().decode(&event)?;
2432
2433 if let Some(num) = reference.strip_prefix('#') {
2434 let codepoint = parse_number(num).map_err(EscapeError::InvalidCharRef)?;
2435 result.push_str(codepoint.encode_utf8(&mut [0u8; 4]));
2436 return Ok(());
2437 }
2438 if let Some(value) = self.entity_resolver.resolve(reference.as_ref()) {
2439 result.push_str(value);
2440 return Ok(());
2441 }
2442 Err(EscapeError::UnrecognizedEntity(0..len, reference.to_string()).into())
2443 }
2444
2445 #[inline]
2446 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
2447 match self.lookahead {
2448 // We pre-read event with the same name that is required to be skipped.
2449 // First call of `read_to_end` will end out pre-read event, the second
2450 // will consume other events
2451 Ok(PayloadEvent::Start(ref e)) if e.name() == name => {
2452 let result1 = self.reader.read_to_end(name);
2453 let result2 = self.reader.read_to_end(name);
2454
2455 // In case of error `next_impl` returns `Eof`
2456 let _ = self.next_impl();
2457 result1?;
2458 result2?;
2459 }
2460 // We pre-read event with the same name that is required to be skipped.
2461 // Because this is end event, we already consume the whole tree, so
2462 // nothing to do, just update lookahead
2463 Ok(PayloadEvent::End(ref e)) if e.name() == name => {
2464 let _ = self.next_impl();
2465 }
2466 Ok(_) => {
2467 let result = self.reader.read_to_end(name);
2468
2469 // In case of error `next_impl` returns `Eof`
2470 let _ = self.next_impl();
2471 result?;
2472 }
2473 // Read next lookahead event, unpack error from the current lookahead
2474 Err(_) => {
2475 self.next_impl()?;
2476 }
2477 }
2478 Ok(())
2479 }
2480
2481 #[inline]
2482 fn decoder(&self) -> Decoder {
2483 self.reader.decoder()
2484 }
2485}
2486
2487////////////////////////////////////////////////////////////////////////////////////////////////////
2488
2489/// Deserialize an instance of type `T` from a string of XML text.
2490pub fn from_str<'de, T>(s: &'de str) -> Result<T, DeError>
2491where
2492 T: Deserialize<'de>,
2493{
2494 let mut de = Deserializer::from_str(s);
2495 T::deserialize(&mut de)
2496}
2497
2498/// Deserialize from a reader. This method will do internal copies of data
2499/// read from `reader`. If you want have a `&str` input and want to borrow
2500/// as much as possible, use [`from_str`].
2501pub fn from_reader<R, T>(reader: R) -> Result<T, DeError>
2502where
2503 R: BufRead,
2504 T: DeserializeOwned,
2505{
2506 let mut de = Deserializer::from_reader(reader);
2507 T::deserialize(&mut de)
2508}
2509
2510////////////////////////////////////////////////////////////////////////////////////////////////////
2511
2512/// A structure that deserializes XML into Rust values.
2513pub struct Deserializer<'de, R, E: EntityResolver = PredefinedEntityResolver>
2514where
2515 R: XmlRead<'de>,
2516{
2517 /// An XML reader that streams events into this deserializer
2518 reader: XmlReader<'de, R, E>,
2519
2520 /// When deserializing sequences sometimes we have to skip unwanted events.
2521 /// That events should be stored and then replayed. This is a replay buffer,
2522 /// that streams events while not empty. When it exhausted, events will
2523 /// requested from [`Self::reader`].
2524 #[cfg(feature = "overlapped-lists")]
2525 read: VecDeque<DeEvent<'de>>,
2526 /// When deserializing sequences sometimes we have to skip events, because XML
2527 /// is tolerant to elements order and even if in the XSD order is strictly
2528 /// specified (using `xs:sequence`) most of XML parsers allows order violations.
2529 /// That means, that elements, forming a sequence, could be overlapped with
2530 /// other elements, do not related to that sequence.
2531 ///
2532 /// In order to support this, deserializer will scan events and skip unwanted
2533 /// events, store them here. After call [`Self::start_replay()`] all events
2534 /// moved from this to [`Self::read`].
2535 #[cfg(feature = "overlapped-lists")]
2536 write: VecDeque<DeEvent<'de>>,
2537 /// Maximum number of events that can be skipped when processing sequences
2538 /// that occur out-of-order. This field is used to prevent potential
2539 /// denial-of-service (DoS) attacks which could cause infinite memory
2540 /// consumption when parsing a very large amount of XML into a sequence field.
2541 #[cfg(feature = "overlapped-lists")]
2542 limit: Option<NonZeroUsize>,
2543
2544 #[cfg(not(feature = "overlapped-lists"))]
2545 peek: Option<DeEvent<'de>>,
2546
2547 /// Buffer to store attribute name as a field name exposed to serde consumers
2548 key_buf: String,
2549}
2550
2551impl<'de, R, E> Deserializer<'de, R, E>
2552where
2553 R: XmlRead<'de>,
2554 E: EntityResolver,
2555{
2556 /// Create an XML deserializer from one of the possible quick_xml input sources.
2557 ///
2558 /// Typically it is more convenient to use one of these methods instead:
2559 ///
2560 /// - [`Deserializer::from_str`]
2561 /// - [`Deserializer::from_reader`]
2562 fn new(reader: R, entity_resolver: E) -> Self {
2563 Self {
2564 reader: XmlReader::new(reader, entity_resolver),
2565
2566 #[cfg(feature = "overlapped-lists")]
2567 read: VecDeque::new(),
2568 #[cfg(feature = "overlapped-lists")]
2569 write: VecDeque::new(),
2570 #[cfg(feature = "overlapped-lists")]
2571 limit: None,
2572
2573 #[cfg(not(feature = "overlapped-lists"))]
2574 peek: None,
2575
2576 key_buf: String::new(),
2577 }
2578 }
2579
2580 /// Returns `true` if all events was consumed.
2581 pub fn is_empty(&self) -> bool {
2582 #[cfg(feature = "overlapped-lists")]
2583 let event = self.read.front();
2584
2585 #[cfg(not(feature = "overlapped-lists"))]
2586 let event = self.peek.as_ref();
2587
2588 match event {
2589 None | Some(DeEvent::Eof) => self.reader.is_empty(),
2590 _ => false,
2591 }
2592 }
2593
2594 /// Returns the underlying XML reader.
2595 ///
2596 /// ```
2597 /// # use pretty_assertions::assert_eq;
2598 /// use serde::Deserialize;
2599 /// use quick_xml::de::Deserializer;
2600 /// use quick_xml::NsReader;
2601 ///
2602 /// #[derive(Deserialize)]
2603 /// struct SomeStruct {
2604 /// field1: String,
2605 /// field2: String,
2606 /// }
2607 ///
2608 /// // Try to deserialize from broken XML
2609 /// let mut de = Deserializer::from_str(
2610 /// "<SomeStruct><field1><field2></SomeStruct>"
2611 /// // 0 ^= 28 ^= 41
2612 /// );
2613 ///
2614 /// let err = SomeStruct::deserialize(&mut de);
2615 /// assert!(err.is_err());
2616 ///
2617 /// let reader: &NsReader<_> = de.get_ref().get_ref();
2618 ///
2619 /// assert_eq!(reader.error_position(), 28);
2620 /// assert_eq!(reader.buffer_position(), 41);
2621 /// ```
2622 pub const fn get_ref(&self) -> &R {
2623 &self.reader.reader
2624 }
2625
2626 /// Set the maximum number of events that could be skipped during deserialization
2627 /// of sequences.
2628 ///
2629 /// If `<element>` contains more than specified nested elements, `$text` or
2630 /// CDATA nodes, then [`DeError::TooManyEvents`] will be returned during
2631 /// deserialization of sequence field (any type that uses [`deserialize_seq`]
2632 /// for the deserialization, for example, `Vec<T>`).
2633 ///
2634 /// This method can be used to prevent a [DoS] attack and infinite memory
2635 /// consumption when parsing a very large XML to a sequence field.
2636 ///
2637 /// It is strongly recommended to set limit to some value when you parse data
2638 /// from untrusted sources. You should choose a value that your typical XMLs
2639 /// can have _between_ different elements that corresponds to the same sequence.
2640 ///
2641 /// # Examples
2642 ///
2643 /// Let's imagine, that we deserialize such structure:
2644 /// ```
2645 /// struct List {
2646 /// item: Vec<()>,
2647 /// }
2648 /// ```
2649 ///
2650 /// The XML that we try to parse look like this:
2651 /// ```xml
2652 /// <any-name>
2653 /// <item/>
2654 /// <!-- Bufferization starts at this point -->
2655 /// <another-item>
2656 /// <some-element>with text</some-element>
2657 /// <yet-another-element/>
2658 /// </another-item>
2659 /// <!-- Buffer will be emptied at this point; 7 events were buffered -->
2660 /// <item/>
2661 /// <!-- There is nothing to buffer, because elements follows each other -->
2662 /// <item/>
2663 /// </any-name>
2664 /// ```
2665 ///
2666 /// There, when we deserialize the `item` field, we need to buffer 7 events,
2667 /// before we can deserialize the second `<item/>`:
2668 ///
2669 /// - `<another-item>`
2670 /// - `<some-element>`
2671 /// - `$text(with text)`
2672 /// - `</some-element>`
2673 /// - `<yet-another-element/>` (virtual start event)
2674 /// - `<yet-another-element/>` (virtual end event)
2675 /// - `</another-item>`
2676 ///
2677 /// Note, that `<yet-another-element/>` internally represented as 2 events:
2678 /// one for the start tag and one for the end tag. In the future this can be
2679 /// eliminated, but for now we use [auto-expanding feature] of a reader,
2680 /// because this simplifies deserializer code.
2681 ///
2682 /// [`deserialize_seq`]: serde::Deserializer::deserialize_seq
2683 /// [DoS]: https://en.wikipedia.org/wiki/Denial-of-service_attack
2684 /// [auto-expanding feature]: crate::reader::Config::expand_empty_elements
2685 #[cfg(feature = "overlapped-lists")]
2686 pub fn event_buffer_size(&mut self, limit: Option<NonZeroUsize>) -> &mut Self {
2687 self.limit = limit;
2688 self
2689 }
2690
2691 #[cfg(feature = "overlapped-lists")]
2692 fn peek(&mut self) -> Result<&DeEvent<'de>, DeError> {
2693 if self.read.is_empty() {
2694 self.read.push_front(self.reader.next()?);
2695 }
2696 if let Some(event) = self.read.front() {
2697 return Ok(event);
2698 }
2699 // SAFETY: `self.read` was filled in the code above.
2700 // NOTE: Can be replaced with `unsafe { std::hint::unreachable_unchecked() }`
2701 // if unsafe code will be allowed
2702 unreachable!()
2703 }
2704 #[cfg(not(feature = "overlapped-lists"))]
2705 fn peek(&mut self) -> Result<&DeEvent<'de>, DeError> {
2706 match &mut self.peek {
2707 Some(event) => Ok(event),
2708 empty_peek @ None => Ok(empty_peek.insert(self.reader.next()?)),
2709 }
2710 }
2711
2712 #[inline]
2713 fn last_peeked(&self) -> &DeEvent<'de> {
2714 #[cfg(feature = "overlapped-lists")]
2715 {
2716 self.read
2717 .front()
2718 .expect("`Deserializer::peek()` should be called")
2719 }
2720 #[cfg(not(feature = "overlapped-lists"))]
2721 {
2722 self.peek
2723 .as_ref()
2724 .expect("`Deserializer::peek()` should be called")
2725 }
2726 }
2727
2728 fn next(&mut self) -> Result<DeEvent<'de>, DeError> {
2729 // Replay skipped or peeked events
2730 #[cfg(feature = "overlapped-lists")]
2731 if let Some(event) = self.read.pop_front() {
2732 return Ok(event);
2733 }
2734 #[cfg(not(feature = "overlapped-lists"))]
2735 if let Some(e) = self.peek.take() {
2736 return Ok(e);
2737 }
2738 self.reader.next()
2739 }
2740
2741 fn skip_whitespaces(&mut self) -> Result<(), DeError> {
2742 loop {
2743 match self.peek()? {
2744 DeEvent::Text(e) if e.is_blank() => {
2745 self.next()?;
2746 }
2747 _ => break,
2748 }
2749 }
2750 Ok(())
2751 }
2752
2753 /// Returns the mark after which all events, skipped by [`Self::skip()`] call,
2754 /// should be replayed after calling [`Self::start_replay()`].
2755 #[cfg(feature = "overlapped-lists")]
2756 #[inline]
2757 #[must_use = "returned checkpoint should be used in `start_replay`"]
2758 fn skip_checkpoint(&self) -> usize {
2759 self.write.len()
2760 }
2761
2762 /// Extracts XML tree of events from and stores them in the skipped events
2763 /// buffer from which they can be retrieved later. You MUST call
2764 /// [`Self::start_replay()`] after calling this to give access to the skipped
2765 /// events and release internal buffers.
2766 #[cfg(feature = "overlapped-lists")]
2767 fn skip(&mut self) -> Result<(), DeError> {
2768 let event = self.next()?;
2769 self.skip_event(event)?;
2770 // Skip all subtree, if we skip a start event
2771 if let Some(DeEvent::Start(e)) = self.write.back() {
2772 let end = e.name().as_ref().to_owned();
2773 let mut depth = 0;
2774 loop {
2775 let event = self.next()?;
2776 match event {
2777 DeEvent::Start(ref e) if e.name().as_ref() == end => {
2778 self.skip_event(event)?;
2779 depth += 1;
2780 }
2781 DeEvent::End(ref e) if e.name().as_ref() == end => {
2782 self.skip_event(event)?;
2783 if depth == 0 {
2784 break;
2785 }
2786 depth -= 1;
2787 }
2788 DeEvent::Eof => {
2789 self.skip_event(event)?;
2790 break;
2791 }
2792 _ => self.skip_event(event)?,
2793 }
2794 }
2795 }
2796 Ok(())
2797 }
2798
2799 #[cfg(feature = "overlapped-lists")]
2800 #[inline]
2801 fn skip_event(&mut self, event: DeEvent<'de>) -> Result<(), DeError> {
2802 if let Some(max) = self.limit {
2803 if self.write.len() >= max.get() {
2804 return Err(DeError::TooManyEvents(max));
2805 }
2806 }
2807 self.write.push_back(event);
2808 Ok(())
2809 }
2810
2811 /// Moves buffered events, skipped after given `checkpoint` from [`Self::write`]
2812 /// skip buffer to [`Self::read`] buffer.
2813 ///
2814 /// After calling this method, [`Self::peek()`] and [`Self::next()`] starts
2815 /// return events that was skipped previously by calling [`Self::skip()`],
2816 /// and only when all that events will be consumed, the deserializer starts
2817 /// to drain events from underlying reader.
2818 ///
2819 /// This method MUST be called if any number of [`Self::skip()`] was called
2820 /// after [`Self::new()`] or `start_replay()` or you'll lost events.
2821 #[cfg(feature = "overlapped-lists")]
2822 fn start_replay(&mut self, checkpoint: usize) {
2823 if checkpoint == 0 {
2824 self.write.append(&mut self.read);
2825 std::mem::swap(&mut self.read, &mut self.write);
2826 } else {
2827 let mut read = self.write.split_off(checkpoint);
2828 read.append(&mut self.read);
2829 self.read = read;
2830 }
2831 }
2832
2833 #[inline]
2834 fn read_string(&mut self) -> Result<Cow<'de, str>, DeError> {
2835 self.read_string_impl(true)
2836 }
2837
2838 /// Consumes consequent [`Text`] and [`CData`] (both a referred below as a _text_)
2839 /// events, merge them into one string. If there are no such events, returns
2840 /// an empty string.
2841 ///
2842 /// If `allow_start` is `false`, then only text events are consumed, for other
2843 /// events an error is returned (see table below).
2844 ///
2845 /// If `allow_start` is `true`, then two or three events are expected:
2846 /// - [`DeEvent::Start`];
2847 /// - _(optional)_ [`DeEvent::Text`] which content is returned;
2848 /// - [`DeEvent::End`]. If text event was missed, an empty string is returned.
2849 ///
2850 /// Corresponding events are consumed.
2851 ///
2852 /// # Handling events
2853 ///
2854 /// The table below shows how events is handled by this method:
2855 ///
2856 /// |Event |XML |Handling
2857 /// |------------------|---------------------------|----------------------------------------
2858 /// |[`DeEvent::Start`]|`<tag>...</tag>` |if `allow_start == true`, result determined by the second table, otherwise emits [`UnexpectedStart("tag")`](DeError::UnexpectedStart)
2859 /// |[`DeEvent::End`] |`</any-tag>` |This is impossible situation, the method will panic if it happens
2860 /// |[`DeEvent::Text`] |`text content` or `<![CDATA[cdata content]]>` (probably mixed)|Returns event content unchanged
2861 /// |[`DeEvent::Eof`] | |Emits [`UnexpectedEof`](DeError::UnexpectedEof)
2862 ///
2863 /// Second event, consumed if [`DeEvent::Start`] was received and `allow_start == true`:
2864 ///
2865 /// |Event |XML |Handling
2866 /// |------------------|---------------------------|----------------------------------------------------------------------------------
2867 /// |[`DeEvent::Start`]|`<any-tag>...</any-tag>` |Emits [`UnexpectedStart("any-tag")`](DeError::UnexpectedStart)
2868 /// |[`DeEvent::End`] |`</tag>` |Returns an empty slice. The reader guarantee that tag will match the open one
2869 /// |[`DeEvent::Text`] |`text content` or `<![CDATA[cdata content]]>` (probably mixed)|Returns event content unchanged, expects the `</tag>` after that
2870 /// |[`DeEvent::Eof`] | |Emits [`InvalidXml(IllFormed(MissingEndTag))`](DeError::InvalidXml)
2871 ///
2872 /// [`Text`]: Event::Text
2873 /// [`CData`]: Event::CData
2874 fn read_string_impl(&mut self, allow_start: bool) -> Result<Cow<'de, str>, DeError> {
2875 match self.next()? {
2876 // Reached by doc tests only: this file, lines 979 and 996
2877 DeEvent::Text(e) => Ok(e.text),
2878 // allow one nested level
2879 // Reached by trivial::{...}::{field, field_nested, field_tag_after, field_tag_before, nested, tag_after, tag_before, wrapped}
2880 DeEvent::Start(e) if allow_start => self.read_text(e.name()),
2881 // TODO: not reached by any tests
2882 DeEvent::Start(e) => Err(DeError::UnexpectedStart(e.name().as_ref().to_owned())),
2883 // SAFETY: The reader is guaranteed that we don't have unmatched tags
2884 // If we here, then our deserializer has a bug
2885 DeEvent::End(e) => unreachable!("{:?}", e),
2886 // Reached by trivial::{empty_doc, only_comment}
2887 DeEvent::Eof => Err(DeError::UnexpectedEof),
2888 }
2889 }
2890 /// Consumes one [`DeEvent::Text`] event and ensures that it is followed by the
2891 /// [`DeEvent::End`] event.
2892 ///
2893 /// # Parameters
2894 /// - `name`: name of a tag opened before reading text. The corresponding end tag
2895 /// should present in input just after the text
2896 fn read_text(&mut self, name: QName) -> Result<Cow<'de, str>, DeError> {
2897 match self.next()? {
2898 DeEvent::Text(e) => match self.next()? {
2899 // The matching tag name is guaranteed by the reader
2900 // Reached by trivial::{...}::{field, wrapped}
2901 DeEvent::End(_) => Ok(e.text),
2902 // SAFETY: Cannot be two consequent Text events, they would be merged into one
2903 DeEvent::Text(_) => unreachable!(),
2904 // Reached by trivial::{...}::{field_tag_after, tag_after}
2905 DeEvent::Start(e) => Err(DeError::UnexpectedStart(e.name().as_ref().to_owned())),
2906 // Reached by struct_::non_closed::elements_child
2907 DeEvent::Eof => Err(Error::missed_end(name, self.reader.decoder()).into()),
2908 },
2909 // We can get End event in case of `<tag></tag>` or `<tag/>` input
2910 // Return empty text in that case
2911 // The matching tag name is guaranteed by the reader
2912 // Reached by {...}::xs_list::empty
2913 DeEvent::End(_) => Ok("".into()),
2914 // Reached by trivial::{...}::{field_nested, field_tag_before, nested, tag_before}
2915 DeEvent::Start(s) => Err(DeError::UnexpectedStart(s.name().as_ref().to_owned())),
2916 // Reached by struct_::non_closed::elements_child
2917 DeEvent::Eof => Err(Error::missed_end(name, self.reader.decoder()).into()),
2918 }
2919 }
2920
2921 /// Drops all events until event with [name](BytesEnd::name()) `name` won't be
2922 /// dropped. This method should be called after [`Self::next()`]
2923 #[cfg(feature = "overlapped-lists")]
2924 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
2925 let mut depth = 0;
2926 loop {
2927 match self.read.pop_front() {
2928 Some(DeEvent::Start(e)) if e.name() == name => {
2929 depth += 1;
2930 }
2931 Some(DeEvent::End(e)) if e.name() == name => {
2932 if depth == 0 {
2933 break;
2934 }
2935 depth -= 1;
2936 }
2937
2938 // Drop all other skipped events
2939 Some(_) => continue,
2940
2941 // If we do not have skipped events, use effective reading that will
2942 // not allocate memory for events
2943 None => {
2944 // We should close all opened tags, because we could buffer
2945 // Start events, but not the corresponding End events. So we
2946 // keep reading events until we exit all nested tags.
2947 // `read_to_end()` will return an error if an Eof was encountered
2948 // preliminary (in case of malformed XML).
2949 //
2950 // <tag><tag></tag></tag>
2951 // ^^^^^^^^^^ - buffered in `self.read`, when `self.read_to_end()` is called, depth = 2
2952 // ^^^^^^ - read by the first call of `self.reader.read_to_end()`
2953 // ^^^^^^ - read by the second call of `self.reader.read_to_end()`
2954 loop {
2955 self.reader.read_to_end(name)?;
2956 if depth == 0 {
2957 break;
2958 }
2959 depth -= 1;
2960 }
2961 break;
2962 }
2963 }
2964 }
2965 Ok(())
2966 }
2967 #[cfg(not(feature = "overlapped-lists"))]
2968 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
2969 // First one might be in self.peek
2970 match self.next()? {
2971 DeEvent::Start(e) => self.reader.read_to_end(e.name())?,
2972 DeEvent::End(e) if e.name() == name => return Ok(()),
2973 _ => (),
2974 }
2975 self.reader.read_to_end(name)
2976 }
2977
2978 fn skip_next_tree(&mut self) -> Result<(), DeError> {
2979 let DeEvent::Start(start) = self.next()? else {
2980 unreachable!("Only call this if the next event is a start event")
2981 };
2982 let name = start.name();
2983 self.read_to_end(name)
2984 }
2985
2986 /// Method for testing Deserializer implementation. Checks that all events was consumed during
2987 /// deserialization. Panics if the next event will not be [`DeEvent::Eof`].
2988 #[doc(hidden)]
2989 #[track_caller]
2990 pub fn check_eof_reached(&mut self) {
2991 // Deserializer may not consume trailing spaces, that is normal
2992 self.skip_whitespaces().expect("cannot skip whitespaces");
2993 let event = self.peek().expect("cannot peek event");
2994 assert_eq!(
2995 *event,
2996 DeEvent::Eof,
2997 "the whole XML document should be consumed, expected `Eof`",
2998 );
2999 }
3000}
3001
3002impl<'de> Deserializer<'de, SliceReader<'de>> {
3003 /// Create a new deserializer that will borrow data from the specified string.
3004 ///
3005 /// Deserializer created with this method will not resolve custom entities.
3006 #[allow(clippy::should_implement_trait)]
3007 pub fn from_str(source: &'de str) -> Self {
3008 Self::from_str_with_resolver(source, PredefinedEntityResolver)
3009 }
3010
3011 /// Create a new deserializer that will borrow data from the specified preconfigured
3012 /// reader.
3013 ///
3014 /// Deserializer created with this method will not resolve custom entities.
3015 ///
3016 /// Note, that config option [`Config::expand_empty_elements`] will be set to `true`.
3017 ///
3018 /// # Example
3019 ///
3020 /// ```
3021 /// # use pretty_assertions::assert_eq;
3022 /// # use quick_xml::de::Deserializer;
3023 /// # use quick_xml::NsReader;
3024 /// # use serde::Deserialize;
3025 /// #
3026 /// #[derive(Deserialize, PartialEq, Debug)]
3027 /// struct Object<'a> {
3028 /// tag: &'a str,
3029 /// }
3030 ///
3031 /// let mut reader = NsReader::from_str("<xml><tag> test </tag></xml>");
3032 ///
3033 /// let mut de = Deserializer::borrowing(reader.clone());
3034 /// let obj = Object::deserialize(&mut de).unwrap();
3035 /// assert_eq!(obj, Object { tag: " test " });
3036 ///
3037 /// reader.config_mut().trim_text(true);
3038 ///
3039 /// let mut de = Deserializer::borrowing(reader);
3040 /// let obj = Object::deserialize(&mut de).unwrap();
3041 /// assert_eq!(obj, Object { tag: "test" });
3042 /// ```
3043 ///
3044 /// [`Config::expand_empty_elements`]: crate::reader::Config::expand_empty_elements
3045 #[inline]
3046 pub fn borrowing(reader: NsReader<&'de [u8]>) -> Self {
3047 Self::borrowing_with_resolver(reader, PredefinedEntityResolver)
3048 }
3049}
3050
3051impl<'de, E> Deserializer<'de, SliceReader<'de>, E>
3052where
3053 E: EntityResolver,
3054{
3055 /// Create a new deserializer that will borrow data from the specified string
3056 /// and use the specified entity resolver.
3057 pub fn from_str_with_resolver(source: &'de str, entity_resolver: E) -> Self {
3058 Self::borrowing_with_resolver(NsReader::from_str(source), entity_resolver)
3059 }
3060
3061 /// Create a new deserializer that will borrow data from the specified preconfigured
3062 /// reader and use the specified entity resolver.
3063 ///
3064 /// Note, that config option [`Config::expand_empty_elements`] will be set to `true`.
3065 ///
3066 /// [`Config::expand_empty_elements`]: crate::reader::Config::expand_empty_elements
3067 pub fn borrowing_with_resolver(mut reader: NsReader<&'de [u8]>, entity_resolver: E) -> Self {
3068 let config = reader.config_mut();
3069 config.expand_empty_elements = true;
3070
3071 Self::new(SliceReader { reader }, entity_resolver)
3072 }
3073}
3074
3075impl<'de, R> Deserializer<'de, IoReader<R>>
3076where
3077 R: BufRead,
3078{
3079 /// Create a new deserializer that will copy data from the specified reader
3080 /// into internal buffer.
3081 ///
3082 /// If you already have a string use [`Self::from_str`] instead, because it
3083 /// will borrow instead of copy. If you have `&[u8]` which is known to represent
3084 /// UTF-8, you can decode it first before using [`from_str`].
3085 ///
3086 /// Deserializer created with this method will not resolve custom entities.
3087 pub fn from_reader(reader: R) -> Self {
3088 Self::with_resolver(reader, PredefinedEntityResolver)
3089 }
3090
3091 /// Create a new deserializer that will copy data from the specified preconfigured
3092 /// reader into internal buffer.
3093 ///
3094 /// Deserializer created with this method will not resolve custom entities.
3095 ///
3096 /// Note, that config option [`Config::expand_empty_elements`] will be set to `true`.
3097 ///
3098 /// # Example
3099 ///
3100 /// ```
3101 /// # use pretty_assertions::assert_eq;
3102 /// # use quick_xml::de::Deserializer;
3103 /// # use quick_xml::NsReader;
3104 /// # use serde::Deserialize;
3105 /// #
3106 /// #[derive(Deserialize, PartialEq, Debug)]
3107 /// struct Object {
3108 /// tag: String,
3109 /// }
3110 ///
3111 /// let mut reader = NsReader::from_str("<xml><tag> test </tag></xml>");
3112 ///
3113 /// let mut de = Deserializer::buffering(reader.clone());
3114 /// let obj = Object::deserialize(&mut de).unwrap();
3115 /// assert_eq!(obj, Object { tag: " test ".to_string() });
3116 ///
3117 /// reader.config_mut().trim_text(true);
3118 ///
3119 /// let mut de = Deserializer::buffering(reader);
3120 /// let obj = Object::deserialize(&mut de).unwrap();
3121 /// assert_eq!(obj, Object { tag: "test".to_string() });
3122 /// ```
3123 ///
3124 /// [`Config::expand_empty_elements`]: crate::reader::Config::expand_empty_elements
3125 #[inline]
3126 pub fn buffering(reader: NsReader<R>) -> Self {
3127 Self::buffering_with_resolver(reader, PredefinedEntityResolver)
3128 }
3129}
3130
3131impl<'de, R, E> Deserializer<'de, IoReader<R>, E>
3132where
3133 R: BufRead,
3134 E: EntityResolver,
3135{
3136 /// Create a new deserializer that will copy data from the specified reader
3137 /// into internal buffer and use the specified entity resolver.
3138 ///
3139 /// If you already have a string use [`Self::from_str`] instead, because it
3140 /// will borrow instead of copy. If you have `&[u8]` which is known to represent
3141 /// UTF-8, you can decode it first before using [`from_str`].
3142 pub fn with_resolver(reader: R, entity_resolver: E) -> Self {
3143 let mut reader = NsReader::from_reader(reader);
3144 let config = reader.config_mut();
3145 config.expand_empty_elements = true;
3146
3147 Self::new(
3148 IoReader {
3149 reader,
3150 buf: Vec::new(),
3151 },
3152 entity_resolver,
3153 )
3154 }
3155
3156 /// Create new deserializer that will copy data from the specified preconfigured reader
3157 /// into internal buffer and use the specified entity resolver.
3158 ///
3159 /// Note, that config option [`Config::expand_empty_elements`] will be set to `true`.
3160 ///
3161 /// [`Config::expand_empty_elements`]: crate::reader::Config::expand_empty_elements
3162 pub fn buffering_with_resolver(mut reader: NsReader<R>, entity_resolver: E) -> Self {
3163 let config = reader.config_mut();
3164 config.expand_empty_elements = true;
3165
3166 Self::new(
3167 IoReader {
3168 reader,
3169 buf: Vec::new(),
3170 },
3171 entity_resolver,
3172 )
3173 }
3174}
3175
3176impl<'de, R, E> de::Deserializer<'de> for &mut Deserializer<'de, R, E>
3177where
3178 R: XmlRead<'de>,
3179 E: EntityResolver,
3180{
3181 type Error = DeError;
3182
3183 deserialize_primitives!();
3184
3185 fn deserialize_struct<V>(
3186 self,
3187 _name: &'static str,
3188 fields: &'static [&'static str],
3189 visitor: V,
3190 ) -> Result<V::Value, DeError>
3191 where
3192 V: Visitor<'de>,
3193 {
3194 // When document is pretty-printed there could be whitespaces before the root element
3195 self.skip_whitespaces()?;
3196 match self.next()? {
3197 DeEvent::Start(e) => visitor.visit_map(ElementMapAccess::new(self, e, fields)),
3198 // SAFETY: The reader is guaranteed that we don't have unmatched tags
3199 // If we here, then our deserializer has a bug
3200 DeEvent::End(e) => unreachable!("{:?}", e),
3201 // Deserializer methods are only hints, if deserializer could not satisfy
3202 // request, it should return the data that it has. It is responsibility
3203 // of a Visitor to return an error if it does not understand the data
3204 DeEvent::Text(e) => match e.text {
3205 Cow::Borrowed(s) => visitor.visit_borrowed_str(s),
3206 Cow::Owned(s) => visitor.visit_string(s),
3207 },
3208 DeEvent::Eof => Err(DeError::UnexpectedEof),
3209 }
3210 }
3211
3212 /// Unit represented in XML as a `xs:element` or text/CDATA content.
3213 /// Any content inside `xs:element` is ignored and skipped.
3214 ///
3215 /// Produces unit struct from any of following inputs:
3216 /// - any `<tag ...>...</tag>`
3217 /// - any `<tag .../>`
3218 /// - any consequent text / CDATA content (can consist of several parts
3219 /// delimited by comments and processing instructions)
3220 ///
3221 /// # Events handling
3222 ///
3223 /// |Event |XML |Handling
3224 /// |------------------|---------------------------|-------------------------------------------
3225 /// |[`DeEvent::Start`]|`<tag>...</tag>` |Calls `visitor.visit_unit()`, consumes all events up to and including corresponding `End` event
3226 /// |[`DeEvent::End`] |`</tag>` |This is impossible situation, the method will panic if it happens
3227 /// |[`DeEvent::Text`] |`text content` or `<![CDATA[cdata content]]>` (probably mixed)|Calls `visitor.visit_unit()`. The content is ignored
3228 /// |[`DeEvent::Eof`] | |Emits [`UnexpectedEof`](DeError::UnexpectedEof)
3229 fn deserialize_unit<V>(self, visitor: V) -> Result<V::Value, DeError>
3230 where
3231 V: Visitor<'de>,
3232 {
3233 match self.next()? {
3234 DeEvent::Start(s) => {
3235 self.read_to_end(s.name())?;
3236 visitor.visit_unit()
3237 }
3238 DeEvent::Text(_) => visitor.visit_unit(),
3239 // SAFETY: The reader is guaranteed that we don't have unmatched tags
3240 // If we here, then our deserializer has a bug
3241 DeEvent::End(e) => unreachable!("{:?}", e),
3242 DeEvent::Eof => Err(DeError::UnexpectedEof),
3243 }
3244 }
3245
3246 /// Forwards deserialization of the inner type. Always calls [`Visitor::visit_newtype_struct`]
3247 /// with the same deserializer.
3248 fn deserialize_newtype_struct<V>(
3249 self,
3250 _name: &'static str,
3251 visitor: V,
3252 ) -> Result<V::Value, DeError>
3253 where
3254 V: Visitor<'de>,
3255 {
3256 visitor.visit_newtype_struct(self)
3257 }
3258
3259 fn deserialize_enum<V>(
3260 self,
3261 _name: &'static str,
3262 _variants: &'static [&'static str],
3263 visitor: V,
3264 ) -> Result<V::Value, DeError>
3265 where
3266 V: Visitor<'de>,
3267 {
3268 // When document is pretty-printed there could be whitespaces before the root element
3269 // which represents the enum variant
3270 // Checked by `top_level::list_of_enum` test in serde-de-seq
3271 self.skip_whitespaces()?;
3272 visitor.visit_enum(var::EnumAccess::new(self))
3273 }
3274
3275 fn deserialize_seq<V>(self, visitor: V) -> Result<V::Value, DeError>
3276 where
3277 V: Visitor<'de>,
3278 {
3279 visitor.visit_seq(self)
3280 }
3281
3282 fn deserialize_option<V>(self, visitor: V) -> Result<V::Value, DeError>
3283 where
3284 V: Visitor<'de>,
3285 {
3286 // We cannot use result of `peek()` directly because of borrow checker
3287 let _ = self.peek()?;
3288 match self.last_peeked() {
3289 DeEvent::Text(t) if t.is_empty() => visitor.visit_none(),
3290 DeEvent::Eof => visitor.visit_none(),
3291 // if the `xsi:nil` attribute is set to true we got a none value
3292 DeEvent::Start(start) if self.reader.reader.has_nil_attr(start) => {
3293 self.skip_next_tree()?;
3294 visitor.visit_none()
3295 }
3296 _ => visitor.visit_some(self),
3297 }
3298 }
3299
3300 fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, DeError>
3301 where
3302 V: Visitor<'de>,
3303 {
3304 match self.peek()? {
3305 DeEvent::Text(_) => self.deserialize_str(visitor),
3306 _ => self.deserialize_map(visitor),
3307 }
3308 }
3309}
3310
3311/// An accessor to sequence elements forming a value for top-level sequence of XML
3312/// elements.
3313///
3314/// Technically, multiple top-level elements violates XML rule of only one top-level
3315/// element, but we consider this as several concatenated XML documents.
3316impl<'de, R, E> SeqAccess<'de> for &mut Deserializer<'de, R, E>
3317where
3318 R: XmlRead<'de>,
3319 E: EntityResolver,
3320{
3321 type Error = DeError;
3322
3323 fn next_element_seed<T>(&mut self, seed: T) -> Result<Option<T::Value>, Self::Error>
3324 where
3325 T: DeserializeSeed<'de>,
3326 {
3327 // When document is pretty-printed there could be whitespaces before, between
3328 // and after root elements. We cannot defer decision if we need to skip spaces
3329 // or not: if we have a sequence of type that does not accept blank text, it
3330 // will need to return something and it can return only error. For example,
3331 // it can be enum without `$text` variant
3332 // Checked by `top_level::list_of_enum` test in serde-de-seq
3333 self.skip_whitespaces()?;
3334 match self.peek()? {
3335 DeEvent::Eof => Ok(None),
3336
3337 // Start(tag), End(tag), Text
3338 _ => seed.deserialize(&mut **self).map(Some),
3339 }
3340 }
3341}
3342
3343impl<'de, R, E> IntoDeserializer<'de, DeError> for &mut Deserializer<'de, R, E>
3344where
3345 R: XmlRead<'de>,
3346 E: EntityResolver,
3347{
3348 type Deserializer = Self;
3349
3350 #[inline]
3351 fn into_deserializer(self) -> Self {
3352 self
3353 }
3354}
3355
3356////////////////////////////////////////////////////////////////////////////////////////////////////
3357
3358/// Converts raw reader's event into a payload event.
3359/// Returns `None`, if event should be skipped.
3360#[inline(always)]
3361fn skip_uninterested<'a>(event: Event<'a>) -> Option<PayloadEvent<'a>> {
3362 let event = match event {
3363 Event::DocType(e) => PayloadEvent::DocType(e),
3364 Event::Start(e) => PayloadEvent::Start(e),
3365 Event::End(e) => PayloadEvent::End(e),
3366 Event::Eof => PayloadEvent::Eof,
3367
3368 // Do not trim next text event after Text, CDATA or reference event
3369 Event::CData(e) => PayloadEvent::CData(e),
3370 Event::Text(e) => PayloadEvent::Text(e),
3371 Event::GeneralRef(e) => PayloadEvent::GeneralRef(e),
3372
3373 _ => return None,
3374 };
3375 Some(event)
3376}
3377
3378////////////////////////////////////////////////////////////////////////////////////////////////////
3379
3380/// Trait used by the deserializer for iterating over input. This is manually
3381/// "specialized" for iterating over `&[u8]`.
3382///
3383/// You do not need to implement this trait, it is needed to abstract from
3384/// [borrowing](SliceReader) and [copying](IoReader) data sources and reuse code in
3385/// deserializer
3386pub trait XmlRead<'i> {
3387 /// Return an input-borrowing event.
3388 fn next(&mut self) -> Result<PayloadEvent<'i>, DeError>;
3389
3390 /// Skips until end element is found. Unlike `next()` it will not allocate
3391 /// when it cannot satisfy the lifetime.
3392 fn read_to_end(&mut self, name: QName) -> Result<(), DeError>;
3393
3394 /// A copy of the reader's decoder used to decode strings.
3395 fn decoder(&self) -> Decoder;
3396
3397 /// Checks if the `start` tag has a [`xsi:nil`] attribute. This method ignores
3398 /// any errors in attributes.
3399 ///
3400 /// [`xsi:nil`]: https://www.w3.org/TR/xmlschema-1/#xsi_nil
3401 fn has_nil_attr(&self, start: &BytesStart) -> bool;
3402}
3403
3404/// XML input source that reads from a std::io input stream.
3405///
3406/// You cannot create it, it is created automatically when you call
3407/// [`Deserializer::from_reader`]
3408pub struct IoReader<R: BufRead> {
3409 reader: NsReader<R>,
3410 buf: Vec<u8>,
3411}
3412
3413impl<R: BufRead> IoReader<R> {
3414 /// Returns the underlying XML reader.
3415 ///
3416 /// ```
3417 /// # use pretty_assertions::assert_eq;
3418 /// use serde::Deserialize;
3419 /// use std::io::Cursor;
3420 /// use quick_xml::de::Deserializer;
3421 /// use quick_xml::NsReader;
3422 ///
3423 /// #[derive(Deserialize)]
3424 /// struct SomeStruct {
3425 /// field1: String,
3426 /// field2: String,
3427 /// }
3428 ///
3429 /// // Try to deserialize from broken XML
3430 /// let mut de = Deserializer::from_reader(Cursor::new(
3431 /// "<SomeStruct><field1><field2></SomeStruct>"
3432 /// // 0 ^= 28 ^= 41
3433 /// ));
3434 ///
3435 /// let err = SomeStruct::deserialize(&mut de);
3436 /// assert!(err.is_err());
3437 ///
3438 /// let reader: &NsReader<Cursor<&str>> = de.get_ref().get_ref();
3439 ///
3440 /// assert_eq!(reader.error_position(), 28);
3441 /// assert_eq!(reader.buffer_position(), 41);
3442 /// ```
3443 pub const fn get_ref(&self) -> &NsReader<R> {
3444 &self.reader
3445 }
3446}
3447
3448impl<'i, R: BufRead> XmlRead<'i> for IoReader<R> {
3449 fn next(&mut self) -> Result<PayloadEvent<'static>, DeError> {
3450 loop {
3451 self.buf.clear();
3452
3453 let event = self.reader.read_event_into(&mut self.buf)?;
3454 if let Some(event) = skip_uninterested(event) {
3455 return Ok(event.into_owned());
3456 }
3457 }
3458 }
3459
3460 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
3461 match self.reader.read_to_end_into(name, &mut self.buf) {
3462 Err(e) => Err(e.into()),
3463 Ok(_) => Ok(()),
3464 }
3465 }
3466
3467 fn decoder(&self) -> Decoder {
3468 self.reader.decoder()
3469 }
3470
3471 fn has_nil_attr(&self, start: &BytesStart) -> bool {
3472 start.attributes().has_nil(self.reader.resolver())
3473 }
3474}
3475
3476/// XML input source that reads from a slice of bytes and can borrow from it.
3477///
3478/// You cannot create it, it is created automatically when you call
3479/// [`Deserializer::from_str`].
3480pub struct SliceReader<'de> {
3481 reader: NsReader<&'de [u8]>,
3482}
3483
3484impl<'de> SliceReader<'de> {
3485 /// Returns the underlying XML reader.
3486 ///
3487 /// ```
3488 /// # use pretty_assertions::assert_eq;
3489 /// use serde::Deserialize;
3490 /// use quick_xml::de::Deserializer;
3491 /// use quick_xml::NsReader;
3492 ///
3493 /// #[derive(Deserialize)]
3494 /// struct SomeStruct {
3495 /// field1: String,
3496 /// field2: String,
3497 /// }
3498 ///
3499 /// // Try to deserialize from broken XML
3500 /// let mut de = Deserializer::from_str(
3501 /// "<SomeStruct><field1><field2></SomeStruct>"
3502 /// // 0 ^= 28 ^= 41
3503 /// );
3504 ///
3505 /// let err = SomeStruct::deserialize(&mut de);
3506 /// assert!(err.is_err());
3507 ///
3508 /// let reader: &NsReader<&[u8]> = de.get_ref().get_ref();
3509 ///
3510 /// assert_eq!(reader.error_position(), 28);
3511 /// assert_eq!(reader.buffer_position(), 41);
3512 /// ```
3513 pub const fn get_ref(&self) -> &NsReader<&'de [u8]> {
3514 &self.reader
3515 }
3516}
3517
3518impl<'de> XmlRead<'de> for SliceReader<'de> {
3519 fn next(&mut self) -> Result<PayloadEvent<'de>, DeError> {
3520 loop {
3521 let event = self.reader.read_event()?;
3522 if let Some(event) = skip_uninterested(event) {
3523 return Ok(event);
3524 }
3525 }
3526 }
3527
3528 fn read_to_end(&mut self, name: QName) -> Result<(), DeError> {
3529 match self.reader.read_to_end(name) {
3530 Err(e) => Err(e.into()),
3531 Ok(_) => Ok(()),
3532 }
3533 }
3534
3535 fn decoder(&self) -> Decoder {
3536 self.reader.decoder()
3537 }
3538
3539 fn has_nil_attr(&self, start: &BytesStart) -> bool {
3540 start.attributes().has_nil(self.reader.resolver())
3541 }
3542}
3543
3544#[cfg(test)]
3545mod tests {
3546 use super::*;
3547 use crate::errors::IllFormedError;
3548 use pretty_assertions::assert_eq;
3549
3550 fn make_de<'de>(source: &'de str) -> Deserializer<'de, SliceReader<'de>> {
3551 dbg!(source);
3552 Deserializer::from_str(source)
3553 }
3554
3555 #[cfg(feature = "overlapped-lists")]
3556 mod skip {
3557 use super::*;
3558 use crate::de::DeEvent::*;
3559 use crate::events::BytesEnd;
3560 use pretty_assertions::assert_eq;
3561
3562 /// Checks that `peek()` and `read()` behaves correctly after `skip()`
3563 #[test]
3564 fn read_and_peek() {
3565 let mut de = make_de(
3566 "\
3567 <root>\
3568 <inner>\
3569 text\
3570 <inner/>\
3571 </inner>\
3572 <next/>\
3573 <target/>\
3574 </root>\
3575 ",
3576 );
3577
3578 // Initial conditions - both are empty
3579 assert_eq!(de.read, vec![]);
3580 assert_eq!(de.write, vec![]);
3581
3582 assert_eq!(de.next().unwrap(), Start(BytesStart::new("root")));
3583 assert_eq!(de.peek().unwrap(), &Start(BytesStart::new("inner")));
3584
3585 // Mark that start_replay() should begin replay from this point
3586 let checkpoint = de.skip_checkpoint();
3587 assert_eq!(checkpoint, 0);
3588
3589 // Should skip first <inner> tree
3590 de.skip().unwrap();
3591 assert_eq!(de.read, vec![]);
3592 assert_eq!(
3593 de.write,
3594 vec![
3595 Start(BytesStart::new("inner")),
3596 Text("text".into()),
3597 Start(BytesStart::new("inner")),
3598 End(BytesEnd::new("inner")),
3599 End(BytesEnd::new("inner")),
3600 ]
3601 );
3602
3603 // Consume <next/>. Now unconsumed XML looks like:
3604 //
3605 // <inner>
3606 // text
3607 // <inner/>
3608 // </inner>
3609 // <target/>
3610 // </root>
3611 assert_eq!(de.next().unwrap(), Start(BytesStart::new("next")));
3612 assert_eq!(de.next().unwrap(), End(BytesEnd::new("next")));
3613
3614 // We finish writing. Next call to `next()` should start replay that messages:
3615 //
3616 // <inner>
3617 // text
3618 // <inner/>
3619 // </inner>
3620 //
3621 // and after that stream that messages:
3622 //
3623 // <target/>
3624 // </root>
3625 de.start_replay(checkpoint);
3626 assert_eq!(
3627 de.read,
3628 vec![
3629 Start(BytesStart::new("inner")),
3630 Text("text".into()),
3631 Start(BytesStart::new("inner")),
3632 End(BytesEnd::new("inner")),
3633 End(BytesEnd::new("inner")),
3634 ]
3635 );
3636 assert_eq!(de.write, vec![]);
3637 assert_eq!(de.next().unwrap(), Start(BytesStart::new("inner")));
3638
3639 // Mark that start_replay() should begin replay from this point
3640 let checkpoint = de.skip_checkpoint();
3641 assert_eq!(checkpoint, 0);
3642
3643 // Skip `$text` node and consume <inner/> after it
3644 de.skip().unwrap();
3645 assert_eq!(
3646 de.read,
3647 vec![
3648 Start(BytesStart::new("inner")),
3649 End(BytesEnd::new("inner")),
3650 End(BytesEnd::new("inner")),
3651 ]
3652 );
3653 assert_eq!(
3654 de.write,
3655 vec![
3656 // This comment here to keep the same formatting of both arrays
3657 // otherwise rustfmt suggest one-line it
3658 Text("text".into()),
3659 ]
3660 );
3661
3662 assert_eq!(de.next().unwrap(), Start(BytesStart::new("inner")));
3663 assert_eq!(de.next().unwrap(), End(BytesEnd::new("inner")));
3664
3665 // We finish writing. Next call to `next()` should start replay messages:
3666 //
3667 // text
3668 // </inner>
3669 //
3670 // and after that stream that messages:
3671 //
3672 // <target/>
3673 // </root>
3674 de.start_replay(checkpoint);
3675 assert_eq!(
3676 de.read,
3677 vec![
3678 // This comment here to keep the same formatting as others
3679 // otherwise rustfmt suggest one-line it
3680 Text("text".into()),
3681 End(BytesEnd::new("inner")),
3682 ]
3683 );
3684 assert_eq!(de.write, vec![]);
3685 assert_eq!(de.next().unwrap(), Text("text".into()));
3686 assert_eq!(de.next().unwrap(), End(BytesEnd::new("inner")));
3687 assert_eq!(de.next().unwrap(), Start(BytesStart::new("target")));
3688 assert_eq!(de.next().unwrap(), End(BytesEnd::new("target")));
3689 assert_eq!(de.next().unwrap(), End(BytesEnd::new("root")));
3690 assert_eq!(de.next().unwrap(), Eof);
3691 }
3692
3693 /// Checks that `read_to_end()` behaves correctly after `skip()`
3694 #[test]
3695 fn read_to_end() {
3696 let mut de = make_de(
3697 "\
3698 <root>\
3699 <skip>\
3700 text\
3701 <skip/>\
3702 </skip>\
3703 <target>\
3704 <target/>\
3705 </target>\
3706 </root>\
3707 ",
3708 );
3709
3710 // Initial conditions - both are empty
3711 assert_eq!(de.read, vec![]);
3712 assert_eq!(de.write, vec![]);
3713
3714 assert_eq!(de.next().unwrap(), Start(BytesStart::new("root")));
3715
3716 // Mark that start_replay() should begin replay from this point
3717 let checkpoint = de.skip_checkpoint();
3718 assert_eq!(checkpoint, 0);
3719
3720 // Skip the <skip> tree
3721 de.skip().unwrap();
3722 assert_eq!(de.read, vec![]);
3723 assert_eq!(
3724 de.write,
3725 vec![
3726 Start(BytesStart::new("skip")),
3727 Text("text".into()),
3728 Start(BytesStart::new("skip")),
3729 End(BytesEnd::new("skip")),
3730 End(BytesEnd::new("skip")),
3731 ]
3732 );
3733
3734 // Drop all events that represents <target> tree. Now unconsumed XML looks like:
3735 //
3736 // <skip>
3737 // text
3738 // <skip/>
3739 // </skip>
3740 // </root>
3741 assert_eq!(de.next().unwrap(), Start(BytesStart::new("target")));
3742 de.read_to_end(QName(b"target")).unwrap();
3743 assert_eq!(de.read, vec![]);
3744 assert_eq!(
3745 de.write,
3746 vec![
3747 Start(BytesStart::new("skip")),
3748 Text("text".into()),
3749 Start(BytesStart::new("skip")),
3750 End(BytesEnd::new("skip")),
3751 End(BytesEnd::new("skip")),
3752 ]
3753 );
3754
3755 // We finish writing. Next call to `next()` should start replay that messages:
3756 //
3757 // <skip>
3758 // text
3759 // <skip/>
3760 // </skip>
3761 //
3762 // and after that stream that messages:
3763 //
3764 // </root>
3765 de.start_replay(checkpoint);
3766 assert_eq!(
3767 de.read,
3768 vec![
3769 Start(BytesStart::new("skip")),
3770 Text("text".into()),
3771 Start(BytesStart::new("skip")),
3772 End(BytesEnd::new("skip")),
3773 End(BytesEnd::new("skip")),
3774 ]
3775 );
3776 assert_eq!(de.write, vec![]);
3777
3778 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skip")));
3779 de.read_to_end(QName(b"skip")).unwrap();
3780
3781 assert_eq!(de.next().unwrap(), End(BytesEnd::new("root")));
3782 assert_eq!(de.next().unwrap(), Eof);
3783 }
3784
3785 /// Checks that replay replayes only part of events
3786 /// Test for https://github.com/tafia/quick-xml/issues/435
3787 #[test]
3788 fn partial_replay() {
3789 let mut de = make_de(
3790 "\
3791 <root>\
3792 <skipped-1/>\
3793 <skipped-2/>\
3794 <inner>\
3795 <skipped-3/>\
3796 <skipped-4/>\
3797 <target-2/>\
3798 </inner>\
3799 <target-1/>\
3800 </root>\
3801 ",
3802 );
3803
3804 // Initial conditions - both are empty
3805 assert_eq!(de.read, vec![]);
3806 assert_eq!(de.write, vec![]);
3807
3808 assert_eq!(de.next().unwrap(), Start(BytesStart::new("root")));
3809
3810 // start_replay() should start replay from this point
3811 let checkpoint1 = de.skip_checkpoint();
3812 assert_eq!(checkpoint1, 0);
3813
3814 // Should skip first and second <skipped-N/> elements
3815 de.skip().unwrap(); // skipped-1
3816 de.skip().unwrap(); // skipped-2
3817 assert_eq!(de.read, vec![]);
3818 assert_eq!(
3819 de.write,
3820 vec![
3821 Start(BytesStart::new("skipped-1")),
3822 End(BytesEnd::new("skipped-1")),
3823 Start(BytesStart::new("skipped-2")),
3824 End(BytesEnd::new("skipped-2")),
3825 ]
3826 );
3827
3828 ////////////////////////////////////////////////////////////////////////////////////////
3829
3830 assert_eq!(de.next().unwrap(), Start(BytesStart::new("inner")));
3831 assert_eq!(de.peek().unwrap(), &Start(BytesStart::new("skipped-3")));
3832 assert_eq!(
3833 de.read,
3834 vec![
3835 // This comment here to keep the same formatting of both arrays
3836 // otherwise rustfmt suggest one-line it
3837 Start(BytesStart::new("skipped-3")),
3838 ]
3839 );
3840 assert_eq!(
3841 de.write,
3842 vec![
3843 Start(BytesStart::new("skipped-1")),
3844 End(BytesEnd::new("skipped-1")),
3845 Start(BytesStart::new("skipped-2")),
3846 End(BytesEnd::new("skipped-2")),
3847 ]
3848 );
3849
3850 // start_replay() should start replay from this point
3851 let checkpoint2 = de.skip_checkpoint();
3852 assert_eq!(checkpoint2, 4);
3853
3854 // Should skip third and forth <skipped-N/> elements
3855 de.skip().unwrap(); // skipped-3
3856 de.skip().unwrap(); // skipped-4
3857 assert_eq!(de.read, vec![]);
3858 assert_eq!(
3859 de.write,
3860 vec![
3861 // checkpoint 1
3862 Start(BytesStart::new("skipped-1")),
3863 End(BytesEnd::new("skipped-1")),
3864 Start(BytesStart::new("skipped-2")),
3865 End(BytesEnd::new("skipped-2")),
3866 // checkpoint 2
3867 Start(BytesStart::new("skipped-3")),
3868 End(BytesEnd::new("skipped-3")),
3869 Start(BytesStart::new("skipped-4")),
3870 End(BytesEnd::new("skipped-4")),
3871 ]
3872 );
3873 assert_eq!(de.next().unwrap(), Start(BytesStart::new("target-2")));
3874 assert_eq!(de.next().unwrap(), End(BytesEnd::new("target-2")));
3875 assert_eq!(de.peek().unwrap(), &End(BytesEnd::new("inner")));
3876 assert_eq!(
3877 de.read,
3878 vec![
3879 // This comment here to keep the same formatting of both arrays
3880 // otherwise rustfmt suggest one-line it
3881 End(BytesEnd::new("inner")),
3882 ]
3883 );
3884 assert_eq!(
3885 de.write,
3886 vec![
3887 // checkpoint 1
3888 Start(BytesStart::new("skipped-1")),
3889 End(BytesEnd::new("skipped-1")),
3890 Start(BytesStart::new("skipped-2")),
3891 End(BytesEnd::new("skipped-2")),
3892 // checkpoint 2
3893 Start(BytesStart::new("skipped-3")),
3894 End(BytesEnd::new("skipped-3")),
3895 Start(BytesStart::new("skipped-4")),
3896 End(BytesEnd::new("skipped-4")),
3897 ]
3898 );
3899
3900 // Start replay events from checkpoint 2
3901 de.start_replay(checkpoint2);
3902 assert_eq!(
3903 de.read,
3904 vec![
3905 Start(BytesStart::new("skipped-3")),
3906 End(BytesEnd::new("skipped-3")),
3907 Start(BytesStart::new("skipped-4")),
3908 End(BytesEnd::new("skipped-4")),
3909 End(BytesEnd::new("inner")),
3910 ]
3911 );
3912 assert_eq!(
3913 de.write,
3914 vec![
3915 Start(BytesStart::new("skipped-1")),
3916 End(BytesEnd::new("skipped-1")),
3917 Start(BytesStart::new("skipped-2")),
3918 End(BytesEnd::new("skipped-2")),
3919 ]
3920 );
3921
3922 // Replayed events
3923 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skipped-3")));
3924 assert_eq!(de.next().unwrap(), End(BytesEnd::new("skipped-3")));
3925 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skipped-4")));
3926 assert_eq!(de.next().unwrap(), End(BytesEnd::new("skipped-4")));
3927
3928 assert_eq!(de.next().unwrap(), End(BytesEnd::new("inner")));
3929 assert_eq!(de.read, vec![]);
3930 assert_eq!(
3931 de.write,
3932 vec![
3933 Start(BytesStart::new("skipped-1")),
3934 End(BytesEnd::new("skipped-1")),
3935 Start(BytesStart::new("skipped-2")),
3936 End(BytesEnd::new("skipped-2")),
3937 ]
3938 );
3939
3940 ////////////////////////////////////////////////////////////////////////////////////////
3941
3942 // New events
3943 assert_eq!(de.next().unwrap(), Start(BytesStart::new("target-1")));
3944 assert_eq!(de.next().unwrap(), End(BytesEnd::new("target-1")));
3945
3946 assert_eq!(de.read, vec![]);
3947 assert_eq!(
3948 de.write,
3949 vec![
3950 Start(BytesStart::new("skipped-1")),
3951 End(BytesEnd::new("skipped-1")),
3952 Start(BytesStart::new("skipped-2")),
3953 End(BytesEnd::new("skipped-2")),
3954 ]
3955 );
3956
3957 // Start replay events from checkpoint 1
3958 de.start_replay(checkpoint1);
3959 assert_eq!(
3960 de.read,
3961 vec![
3962 Start(BytesStart::new("skipped-1")),
3963 End(BytesEnd::new("skipped-1")),
3964 Start(BytesStart::new("skipped-2")),
3965 End(BytesEnd::new("skipped-2")),
3966 ]
3967 );
3968 assert_eq!(de.write, vec![]);
3969
3970 // Replayed events
3971 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skipped-1")));
3972 assert_eq!(de.next().unwrap(), End(BytesEnd::new("skipped-1")));
3973 assert_eq!(de.next().unwrap(), Start(BytesStart::new("skipped-2")));
3974 assert_eq!(de.next().unwrap(), End(BytesEnd::new("skipped-2")));
3975
3976 assert_eq!(de.read, vec![]);
3977 assert_eq!(de.write, vec![]);
3978
3979 // New events
3980 assert_eq!(de.next().unwrap(), End(BytesEnd::new("root")));
3981 assert_eq!(de.next().unwrap(), Eof);
3982 }
3983
3984 /// Checks that limiting buffer size works correctly
3985 #[test]
3986 fn limit() {
3987 use serde::Deserialize;
3988
3989 #[derive(Debug, Deserialize)]
3990 #[allow(unused)]
3991 struct List {
3992 item: Vec<()>,
3993 }
3994
3995 let mut de = make_de(
3996 "\
3997 <any-name>\
3998 <item/>\
3999 <another-item>\
4000 <some-element>with text</some-element>\
4001 <yet-another-element/>\
4002 </another-item>\
4003 <item/>\
4004 <item/>\
4005 </any-name>\
4006 ",
4007 );
4008 de.event_buffer_size(NonZeroUsize::new(3));
4009
4010 match List::deserialize(&mut de) {
4011 Err(DeError::TooManyEvents(count)) => assert_eq!(count.get(), 3),
4012 e => panic!("Expected `Err(TooManyEvents(3))`, but got `{:?}`", e),
4013 }
4014 }
4015
4016 /// Without handling Eof in `skip` this test failed with memory allocation
4017 #[test]
4018 fn invalid_xml() {
4019 use crate::de::DeEvent::*;
4020
4021 let mut de = make_de("<root>");
4022
4023 // Cache all events
4024 let checkpoint = de.skip_checkpoint();
4025 de.skip().unwrap();
4026 de.start_replay(checkpoint);
4027 assert_eq!(de.read, vec![Start(BytesStart::new("root")), Eof]);
4028 }
4029 }
4030
4031 mod read_to_end {
4032 use super::*;
4033 use crate::de::DeEvent::*;
4034 use pretty_assertions::assert_eq;
4035
4036 #[test]
4037 fn complex() {
4038 let mut de = make_de(
4039 r#"
4040 <root>
4041 <tag a="1"><tag>text</tag>content</tag>
4042 <tag a="2"><![CDATA[cdata content]]></tag>
4043 <self-closed/>
4044 </root>
4045 "#,
4046 );
4047
4048 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4049 assert_eq!(de.next().unwrap(), Start(BytesStart::new("root")));
4050
4051 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4052 assert_eq!(
4053 de.next().unwrap(),
4054 Start(BytesStart::from_content(r#"tag a="1""#, 3))
4055 );
4056 assert_eq!(de.read_to_end(QName(b"tag")).unwrap(), ());
4057
4058 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4059 assert_eq!(
4060 de.next().unwrap(),
4061 Start(BytesStart::from_content(r#"tag a="2""#, 3))
4062 );
4063 assert_eq!(de.next().unwrap(), Text("cdata content".into()));
4064 assert_eq!(de.next().unwrap(), End(BytesEnd::new("tag")));
4065
4066 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4067 assert_eq!(de.next().unwrap(), Start(BytesStart::new("self-closed")));
4068 assert_eq!(de.read_to_end(QName(b"self-closed")).unwrap(), ());
4069
4070 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4071 assert_eq!(de.next().unwrap(), End(BytesEnd::new("root")));
4072 assert_eq!(de.next().unwrap(), Text("\n ".into()));
4073 assert_eq!(de.next().unwrap(), Eof);
4074 }
4075
4076 #[test]
4077 fn invalid_xml1() {
4078 let mut de = make_de("<tag><tag></tag>");
4079
4080 assert_eq!(de.next().unwrap(), Start(BytesStart::new("tag")));
4081 assert_eq!(de.peek().unwrap(), &Start(BytesStart::new("tag")));
4082
4083 match de.read_to_end(QName(b"tag")) {
4084 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4085 assert_eq!(cause, IllFormedError::MissingEndTag("tag".into()))
4086 }
4087 x => panic!(
4088 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4089 x
4090 ),
4091 }
4092 assert_eq!(de.next().unwrap(), Eof);
4093 }
4094
4095 #[test]
4096 fn invalid_xml2() {
4097 let mut de = make_de("<tag><![CDATA[]]><tag></tag>");
4098
4099 assert_eq!(de.next().unwrap(), Start(BytesStart::new("tag")));
4100 assert_eq!(de.peek().unwrap(), &Text("".into()));
4101
4102 match de.read_to_end(QName(b"tag")) {
4103 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4104 assert_eq!(cause, IllFormedError::MissingEndTag("tag".into()))
4105 }
4106 x => panic!(
4107 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4108 x
4109 ),
4110 }
4111 assert_eq!(de.next().unwrap(), Eof);
4112 }
4113 }
4114
4115 #[test]
4116 fn borrowing_reader_parity() {
4117 let s = r#"
4118 <item name="hello" source="world.rs">Some text</item>
4119 <item2/>
4120 <item3 value="world" />
4121 "#;
4122
4123 let mut reader1 = IoReader {
4124 reader: NsReader::from_reader(s.as_bytes()),
4125 buf: Vec::new(),
4126 };
4127 let mut reader2 = SliceReader {
4128 reader: NsReader::from_str(s),
4129 };
4130
4131 loop {
4132 let event1 = reader1.next().unwrap();
4133 let event2 = reader2.next().unwrap();
4134
4135 if let (PayloadEvent::Eof, PayloadEvent::Eof) = (&event1, &event2) {
4136 break;
4137 }
4138
4139 assert_eq!(event1, event2);
4140 }
4141 }
4142
4143 #[test]
4144 fn borrowing_reader_events() {
4145 let s = r#"
4146 <item name="hello" source="world.rs">Some text</item>
4147 <item2></item2>
4148 <item3/>
4149 <item4 value="world" />
4150 "#;
4151
4152 let mut reader = SliceReader {
4153 reader: NsReader::from_str(s),
4154 };
4155
4156 let config = reader.reader.config_mut();
4157 config.expand_empty_elements = true;
4158
4159 let mut events = Vec::new();
4160
4161 loop {
4162 let event = reader.next().unwrap();
4163 if let PayloadEvent::Eof = event {
4164 break;
4165 }
4166 events.push(event);
4167 }
4168
4169 use crate::de::PayloadEvent::*;
4170
4171 assert_eq!(
4172 events,
4173 vec![
4174 Text(BytesText::from_escaped("\n ")),
4175 Start(BytesStart::from_content(
4176 r#"item name="hello" source="world.rs""#,
4177 4
4178 )),
4179 Text(BytesText::from_escaped("Some text")),
4180 End(BytesEnd::new("item")),
4181 Text(BytesText::from_escaped("\n ")),
4182 Start(BytesStart::from_content("item2", 5)),
4183 End(BytesEnd::new("item2")),
4184 Text(BytesText::from_escaped("\n ")),
4185 Start(BytesStart::from_content("item3", 5)),
4186 End(BytesEnd::new("item3")),
4187 Text(BytesText::from_escaped("\n ")),
4188 Start(BytesStart::from_content(r#"item4 value="world" "#, 5)),
4189 End(BytesEnd::new("item4")),
4190 Text(BytesText::from_escaped("\n ")),
4191 ]
4192 )
4193 }
4194
4195 /// Ensures, that [`Deserializer::read_string()`] never can get an `End` event,
4196 /// because parser reports error early
4197 #[test]
4198 fn read_string() {
4199 match from_str::<String>(r#"</root>"#) {
4200 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4201 assert_eq!(cause, IllFormedError::UnmatchedEndTag("root".into()));
4202 }
4203 x => panic!(
4204 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4205 x
4206 ),
4207 }
4208
4209 let s: String = from_str(r#"<root></root>"#).unwrap();
4210 assert_eq!(s, "");
4211
4212 match from_str::<String>(r#"<root></other>"#) {
4213 Err(DeError::InvalidXml(Error::IllFormed(cause))) => assert_eq!(
4214 cause,
4215 IllFormedError::MismatchedEndTag {
4216 expected: "root".into(),
4217 found: "other".into(),
4218 }
4219 ),
4220 x => panic!("Expected `Err(InvalidXml(IllFormed(_))`, but got `{:?}`", x),
4221 }
4222 }
4223
4224 /// Tests for https://github.com/tafia/quick-xml/issues/474.
4225 ///
4226 /// That tests ensures that comments and processed instructions is ignored
4227 /// and can split one logical string in pieces.
4228 mod merge_text {
4229 use super::*;
4230 use pretty_assertions::assert_eq;
4231
4232 #[test]
4233 fn text() {
4234 let mut de = make_de("text");
4235 assert_eq!(de.next().unwrap(), DeEvent::Text("text".into()));
4236 }
4237
4238 #[test]
4239 fn cdata() {
4240 let mut de = make_de("<![CDATA[cdata]]>");
4241 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata".into()));
4242 }
4243
4244 #[test]
4245 fn text_and_cdata() {
4246 let mut de = make_de("text and <![CDATA[cdata]]>");
4247 assert_eq!(de.next().unwrap(), DeEvent::Text("text and cdata".into()));
4248 }
4249
4250 #[test]
4251 fn text_and_empty_cdata() {
4252 let mut de = make_de("text and <![CDATA[]]>");
4253 assert_eq!(de.next().unwrap(), DeEvent::Text("text and ".into()));
4254 }
4255
4256 #[test]
4257 fn cdata_and_text() {
4258 let mut de = make_de("<![CDATA[cdata]]> and text");
4259 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata and text".into()));
4260 }
4261
4262 #[test]
4263 fn empty_cdata_and_text() {
4264 let mut de = make_de("<![CDATA[]]> and text");
4265 assert_eq!(de.next().unwrap(), DeEvent::Text(" and text".into()));
4266 }
4267
4268 #[test]
4269 fn cdata_and_cdata() {
4270 let mut de = make_de(
4271 "\
4272 <![CDATA[cdata]]]]>\
4273 <![CDATA[>cdata]]>\
4274 ",
4275 );
4276 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4277 }
4278
4279 mod comment_between {
4280 use super::*;
4281 use pretty_assertions::assert_eq;
4282
4283 #[test]
4284 fn text() {
4285 let mut de = make_de(
4286 "\
4287 text \
4288 <!--comment 1--><!--comment 2--> \
4289 text\
4290 ",
4291 );
4292 assert_eq!(de.next().unwrap(), DeEvent::Text("text text".into()));
4293 }
4294
4295 #[test]
4296 fn cdata() {
4297 let mut de = make_de(
4298 "\
4299 <![CDATA[cdata]]]]>\
4300 <!--comment 1--><!--comment 2-->\
4301 <![CDATA[>cdata]]>\
4302 ",
4303 );
4304 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4305 }
4306
4307 #[test]
4308 fn text_and_cdata() {
4309 let mut de = make_de(
4310 "\
4311 text \
4312 <!--comment 1--><!--comment 2-->\
4313 <![CDATA[ cdata]]>\
4314 ",
4315 );
4316 assert_eq!(de.next().unwrap(), DeEvent::Text("text cdata".into()));
4317 }
4318
4319 #[test]
4320 fn text_and_empty_cdata() {
4321 let mut de = make_de(
4322 "\
4323 text \
4324 <!--comment 1--><!--comment 2-->\
4325 <![CDATA[]]>\
4326 ",
4327 );
4328 assert_eq!(de.next().unwrap(), DeEvent::Text("text ".into()));
4329 }
4330
4331 #[test]
4332 fn cdata_and_text() {
4333 let mut de = make_de(
4334 "\
4335 <![CDATA[cdata ]]>\
4336 <!--comment 1--><!--comment 2--> \
4337 text \
4338 ",
4339 );
4340 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata text ".into()));
4341 }
4342
4343 #[test]
4344 fn empty_cdata_and_text() {
4345 let mut de = make_de(
4346 "\
4347 <![CDATA[]]>\
4348 <!--comment 1--><!--comment 2--> \
4349 text \
4350 ",
4351 );
4352 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4353 }
4354
4355 #[test]
4356 fn cdata_and_cdata() {
4357 let mut de = make_de(
4358 "\
4359 <![CDATA[cdata]]]>\
4360 <!--comment 1--><!--comment 2-->\
4361 <![CDATA[]>cdata]]>\
4362 ",
4363 );
4364 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4365 }
4366 }
4367
4368 mod pi_between {
4369 use super::*;
4370 use pretty_assertions::assert_eq;
4371
4372 #[test]
4373 fn text() {
4374 let mut de = make_de(
4375 "\
4376 text \
4377 <?pi 1?><?pi 2?> \
4378 text\
4379 ",
4380 );
4381 assert_eq!(de.next().unwrap(), DeEvent::Text("text text".into()));
4382 }
4383
4384 #[test]
4385 fn cdata() {
4386 let mut de = make_de(
4387 "\
4388 <![CDATA[cdata]]]]>\
4389 <?pi 1?><?pi 2?>\
4390 <![CDATA[>cdata]]>\
4391 ",
4392 );
4393 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4394 }
4395
4396 #[test]
4397 fn text_and_cdata() {
4398 let mut de = make_de(
4399 "\
4400 text \
4401 <?pi 1?><?pi 2?>\
4402 <![CDATA[ cdata]]>\
4403 ",
4404 );
4405 assert_eq!(de.next().unwrap(), DeEvent::Text("text cdata".into()));
4406 }
4407
4408 #[test]
4409 fn text_and_empty_cdata() {
4410 let mut de = make_de(
4411 "\
4412 text \
4413 <?pi 1?><?pi 2?>\
4414 <![CDATA[]]>\
4415 ",
4416 );
4417 assert_eq!(de.next().unwrap(), DeEvent::Text("text ".into()));
4418 }
4419
4420 #[test]
4421 fn cdata_and_text() {
4422 let mut de = make_de(
4423 "\
4424 <![CDATA[cdata ]]>\
4425 <?pi 1?><?pi 2?> \
4426 text \
4427 ",
4428 );
4429 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata text ".into()));
4430 }
4431
4432 #[test]
4433 fn empty_cdata_and_text() {
4434 let mut de = make_de(
4435 "\
4436 <![CDATA[]]>\
4437 <?pi 1?><?pi 2?> \
4438 text \
4439 ",
4440 );
4441 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4442 }
4443
4444 #[test]
4445 fn cdata_and_cdata() {
4446 let mut de = make_de(
4447 "\
4448 <![CDATA[cdata]]]>\
4449 <?pi 1?><?pi 2?>\
4450 <![CDATA[]>cdata]]>\
4451 ",
4452 );
4453 assert_eq!(de.next().unwrap(), DeEvent::Text("cdata]]>cdata".into()));
4454 }
4455 }
4456 }
4457
4458 /// Tests for https://github.com/tafia/quick-xml/issues/474.
4459 ///
4460 /// This tests ensures that any combination of payload data is processed
4461 /// as expected.
4462 mod triples {
4463 use super::*;
4464 use pretty_assertions::assert_eq;
4465
4466 mod start {
4467 use super::*;
4468
4469 /// <tag1><tag2>...
4470 // The same name is intentional
4471 #[allow(clippy::module_inception)]
4472 mod start {
4473 use super::*;
4474 use pretty_assertions::assert_eq;
4475
4476 #[test]
4477 fn start() {
4478 let mut de = make_de("<tag1><tag2><tag3>");
4479 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4480 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4481 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag3")));
4482 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4483 }
4484
4485 /// Not matching end tag will result to error
4486 #[test]
4487 fn end() {
4488 let mut de = make_de("<tag1><tag2></tag2>");
4489 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4490 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4491 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag2")));
4492 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4493 }
4494
4495 #[test]
4496 fn text() {
4497 let mut de = make_de("<tag1><tag2> text ");
4498 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4499 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4500 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4501 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4502 }
4503
4504 #[test]
4505 fn cdata() {
4506 let mut de = make_de("<tag1><tag2><![CDATA[ cdata ]]>");
4507 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4508 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4509 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4510 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4511 }
4512
4513 #[test]
4514 fn eof() {
4515 let mut de = make_de("<tag1><tag2>");
4516 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4517 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4518 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4519 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4520 }
4521 }
4522
4523 /// <tag></tag>...
4524 mod end {
4525 use super::*;
4526 use pretty_assertions::assert_eq;
4527
4528 #[test]
4529 fn start() {
4530 let mut de = make_de("<tag></tag><tag2>");
4531 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4532 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4533 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4534 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4535 }
4536
4537 #[test]
4538 fn end() {
4539 let mut de = make_de("<tag></tag></tag2>");
4540 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4541 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4542 match de.next() {
4543 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4544 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag2".into()));
4545 }
4546 x => panic!(
4547 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4548 x
4549 ),
4550 }
4551 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4552 }
4553
4554 #[test]
4555 fn text() {
4556 let mut de = make_de("<tag></tag> text ");
4557 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4558 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4559 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4560 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4561 }
4562
4563 #[test]
4564 fn cdata() {
4565 let mut de = make_de("<tag></tag><![CDATA[ cdata ]]>");
4566 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4567 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4568 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4569 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4570 }
4571
4572 #[test]
4573 fn eof() {
4574 let mut de = make_de("<tag></tag>");
4575 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4576 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4577 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4578 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4579 }
4580 }
4581
4582 /// <tag> text ...
4583 mod text {
4584 use super::*;
4585 use pretty_assertions::assert_eq;
4586
4587 #[test]
4588 fn start() {
4589 let mut de = make_de("<tag> text <tag2>");
4590 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4591 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4592 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4593 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4594 }
4595
4596 #[test]
4597 fn end() {
4598 let mut de = make_de("<tag> text </tag>");
4599 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4600 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4601 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4602 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4603 }
4604
4605 // start::text::text has no difference from start::text
4606
4607 #[test]
4608 fn cdata() {
4609 let mut de = make_de("<tag> text <![CDATA[ cdata ]]>");
4610 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4611 assert_eq!(de.next().unwrap(), DeEvent::Text(" text cdata ".into()));
4612 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4613 }
4614
4615 #[test]
4616 fn eof() {
4617 let mut de = make_de("<tag> text ");
4618 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4619 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4620 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4621 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4622 }
4623 }
4624
4625 /// <tag><![CDATA[ cdata ]]>...
4626 mod cdata {
4627 use super::*;
4628 use pretty_assertions::assert_eq;
4629
4630 #[test]
4631 fn start() {
4632 let mut de = make_de("<tag><![CDATA[ cdata ]]><tag2>");
4633 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4634 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4635 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4636 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4637 }
4638
4639 #[test]
4640 fn end() {
4641 let mut de = make_de("<tag><![CDATA[ cdata ]]></tag>");
4642 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4643 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4644 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4645 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4646 }
4647
4648 #[test]
4649 fn text() {
4650 let mut de = make_de("<tag><![CDATA[ cdata ]]> text ");
4651 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4652 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata text ".into()));
4653 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4654 }
4655
4656 #[test]
4657 fn cdata() {
4658 let mut de = make_de("<tag><![CDATA[ cdata ]]><![CDATA[ cdata2 ]]>");
4659 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4660 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata cdata2 ".into()));
4661 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4662 }
4663
4664 #[test]
4665 fn eof() {
4666 let mut de = make_de("<tag><![CDATA[ cdata ]]>");
4667 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4668 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4669 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4670 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4671 }
4672 }
4673 }
4674
4675 /// Start from End event will always generate an error
4676 #[test]
4677 fn end() {
4678 let mut de = make_de("</tag>");
4679 match de.next() {
4680 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4681 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4682 }
4683 x => panic!(
4684 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4685 x
4686 ),
4687 }
4688 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4689 }
4690
4691 mod text {
4692 use super::*;
4693 use pretty_assertions::assert_eq;
4694
4695 mod start {
4696 use super::*;
4697 use pretty_assertions::assert_eq;
4698
4699 #[test]
4700 fn start() {
4701 let mut de = make_de(" text <tag1><tag2>");
4702 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4703 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4704 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4705 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4706 }
4707
4708 /// Not matching end tag will result in error
4709 #[test]
4710 fn end() {
4711 let mut de = make_de(" text <tag></tag>");
4712 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4713 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4714 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4715 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4716 }
4717
4718 #[test]
4719 fn text() {
4720 let mut de = make_de(" text <tag> text2 ");
4721 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4722 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4723 assert_eq!(de.next().unwrap(), DeEvent::Text(" text2 ".into()));
4724 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4725 }
4726
4727 #[test]
4728 fn cdata() {
4729 let mut de = make_de(" text <tag><![CDATA[ cdata ]]>");
4730 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4731 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4732 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4733 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4734 }
4735
4736 #[test]
4737 fn eof() {
4738 let mut de = make_de(" text <tag>");
4739 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4740 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4741 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4742 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4743 }
4744 }
4745
4746 /// End event without corresponding start event will always generate an error
4747 #[test]
4748 fn end() {
4749 let mut de = make_de(" text </tag>");
4750 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4751 match de.next() {
4752 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4753 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4754 }
4755 x => panic!(
4756 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4757 x
4758 ),
4759 }
4760 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4761 }
4762
4763 // text::text::something is equivalent to text::something
4764
4765 mod cdata {
4766 use super::*;
4767 use pretty_assertions::assert_eq;
4768
4769 #[test]
4770 fn start() {
4771 let mut de = make_de(" text <![CDATA[ cdata ]]><tag>");
4772 assert_eq!(de.next().unwrap(), DeEvent::Text(" text cdata ".into()));
4773 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4774 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4775 }
4776
4777 #[test]
4778 fn end() {
4779 let mut de = make_de(" text <![CDATA[ cdata ]]></tag>");
4780 assert_eq!(de.next().unwrap(), DeEvent::Text(" text cdata ".into()));
4781 match de.next() {
4782 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4783 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4784 }
4785 x => panic!(
4786 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4787 x
4788 ),
4789 }
4790 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4791 }
4792
4793 #[test]
4794 fn text() {
4795 let mut de = make_de(" text <![CDATA[ cdata ]]> text2 ");
4796 assert_eq!(
4797 de.next().unwrap(),
4798 DeEvent::Text(" text cdata text2 ".into())
4799 );
4800 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4801 }
4802
4803 #[test]
4804 fn cdata() {
4805 let mut de = make_de(" text <![CDATA[ cdata ]]><![CDATA[ cdata2 ]]>");
4806 assert_eq!(
4807 de.next().unwrap(),
4808 DeEvent::Text(" text cdata cdata2 ".into())
4809 );
4810 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4811 }
4812
4813 #[test]
4814 fn eof() {
4815 let mut de = make_de(" text <![CDATA[ cdata ]]>");
4816 assert_eq!(de.next().unwrap(), DeEvent::Text(" text cdata ".into()));
4817 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4818 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4819 }
4820 }
4821 }
4822
4823 mod cdata {
4824 use super::*;
4825 use pretty_assertions::assert_eq;
4826
4827 mod start {
4828 use super::*;
4829 use pretty_assertions::assert_eq;
4830
4831 #[test]
4832 fn start() {
4833 let mut de = make_de("<![CDATA[ cdata ]]><tag1><tag2>");
4834 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4835 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag1")));
4836 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag2")));
4837 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4838 }
4839
4840 /// Not matching end tag will result in error
4841 #[test]
4842 fn end() {
4843 let mut de = make_de("<![CDATA[ cdata ]]><tag></tag>");
4844 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4845 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4846 assert_eq!(de.next().unwrap(), DeEvent::End(BytesEnd::new("tag")));
4847 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4848 }
4849
4850 #[test]
4851 fn text() {
4852 let mut de = make_de("<![CDATA[ cdata ]]><tag> text ");
4853 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4854 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4855 assert_eq!(de.next().unwrap(), DeEvent::Text(" text ".into()));
4856 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4857 }
4858
4859 #[test]
4860 fn cdata() {
4861 let mut de = make_de("<![CDATA[ cdata ]]><tag><![CDATA[ cdata2 ]]>");
4862 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4863 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4864 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata2 ".into()));
4865 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4866 }
4867
4868 #[test]
4869 fn eof() {
4870 let mut de = make_de("<![CDATA[ cdata ]]><tag>");
4871 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4872 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4873 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4874 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4875 }
4876 }
4877
4878 /// End event without corresponding start event will always generate an error
4879 #[test]
4880 fn end() {
4881 let mut de = make_de("<![CDATA[ cdata ]]></tag>");
4882 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata ".into()));
4883 match de.next() {
4884 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4885 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4886 }
4887 x => panic!(
4888 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4889 x
4890 ),
4891 }
4892 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4893 }
4894
4895 mod text {
4896 use super::*;
4897 use pretty_assertions::assert_eq;
4898
4899 #[test]
4900 fn start() {
4901 let mut de = make_de("<![CDATA[ cdata ]]> text <tag>");
4902 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata text ".into()));
4903 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4904 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4905 }
4906
4907 #[test]
4908 fn end() {
4909 let mut de = make_de("<![CDATA[ cdata ]]> text </tag>");
4910 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata text ".into()));
4911 match de.next() {
4912 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4913 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4914 }
4915 x => panic!(
4916 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4917 x
4918 ),
4919 }
4920 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4921 }
4922
4923 // cdata::text::text is equivalent to cdata::text
4924
4925 #[test]
4926 fn cdata() {
4927 let mut de = make_de("<![CDATA[ cdata ]]> text <![CDATA[ cdata2 ]]>");
4928 assert_eq!(
4929 de.next().unwrap(),
4930 DeEvent::Text(" cdata text cdata2 ".into())
4931 );
4932 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4933 }
4934
4935 #[test]
4936 fn eof() {
4937 let mut de = make_de("<![CDATA[ cdata ]]> text ");
4938 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata text ".into()));
4939 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4940 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4941 }
4942 }
4943
4944 // The same name is intentional
4945 #[allow(clippy::module_inception)]
4946 mod cdata {
4947 use super::*;
4948 use pretty_assertions::assert_eq;
4949
4950 #[test]
4951 fn start() {
4952 let mut de = make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]><tag>");
4953 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata cdata2 ".into()));
4954 assert_eq!(de.next().unwrap(), DeEvent::Start(BytesStart::new("tag")));
4955 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4956 }
4957
4958 #[test]
4959 fn end() {
4960 let mut de = make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]></tag>");
4961 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata cdata2 ".into()));
4962 match de.next() {
4963 Err(DeError::InvalidXml(Error::IllFormed(cause))) => {
4964 assert_eq!(cause, IllFormedError::UnmatchedEndTag("tag".into()));
4965 }
4966 x => panic!(
4967 "Expected `Err(InvalidXml(IllFormed(_)))`, but got `{:?}`",
4968 x
4969 ),
4970 }
4971 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4972 }
4973
4974 #[test]
4975 fn text() {
4976 let mut de = make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]> text ");
4977 assert_eq!(
4978 de.next().unwrap(),
4979 DeEvent::Text(" cdata cdata2 text ".into())
4980 );
4981 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4982 }
4983
4984 #[test]
4985 fn cdata() {
4986 let mut de =
4987 make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]><![CDATA[ cdata3 ]]>");
4988 assert_eq!(
4989 de.next().unwrap(),
4990 DeEvent::Text(" cdata cdata2 cdata3 ".into())
4991 );
4992 assert_eq!(de.next().unwrap(), DeEvent::Eof);
4993 }
4994
4995 #[test]
4996 fn eof() {
4997 let mut de = make_de("<![CDATA[ cdata ]]><![CDATA[ cdata2 ]]>");
4998 assert_eq!(de.next().unwrap(), DeEvent::Text(" cdata cdata2 ".into()));
4999 assert_eq!(de.next().unwrap(), DeEvent::Eof);
5000 assert_eq!(de.next().unwrap(), DeEvent::Eof);
5001 }
5002 }
5003 }
5004 }
5005}