iri_string/parser/
validate.rs

1//! Validating parsers for non-trusted (possibly invalid) input.
2
3mod authority;
4mod path;
5
6use crate::parser::char;
7use crate::parser::str::{
8    find_split, find_split2_hole, find_split_hole, satisfy_chars_with_pct_encoded,
9};
10use crate::spec::Spec;
11use crate::validate::Error;
12
13use self::authority::validate_authority;
14pub(crate) use self::authority::{validate_host, validate_userinfo};
15pub(crate) use self::path::validate_path;
16use self::path::{
17    validate_path_abempty, validate_path_absolute_authority_absent,
18    validate_path_relative_authority_absent,
19};
20
21/// Returns `Ok(_)` if the string matches `scheme`.
22pub(crate) fn validate_scheme(i: &str) -> Result<(), Error> {
23    debug_assert!(!i.is_empty());
24    let bytes = i.as_bytes();
25    if bytes[0].is_ascii_alphabetic()
26        && bytes[1..]
27            .iter()
28            .all(|&b| b.is_ascii() && char::is_ascii_scheme_continue(b))
29    {
30        Ok(())
31    } else {
32        Err(Error::new())
33    }
34}
35
36/// Returns `Ok(_)` if the string matches `query` or `iquery`.
37pub(crate) fn validate_query<S: Spec>(i: &str) -> Result<(), Error> {
38    let is_valid =
39        satisfy_chars_with_pct_encoded(i, char::is_ascii_frag_query, char::is_nonascii_query::<S>);
40    if is_valid {
41        Ok(())
42    } else {
43        Err(Error::new())
44    }
45}
46
47/// Returns `Ok(_)` if the string matches `authority path-abempty` rule sequence.
48fn validate_authority_path_abempty<S: Spec>(i: &str) -> Result<(), Error> {
49    let (maybe_authority, maybe_path) = match find_split(i, b'/') {
50        Some(v) => v,
51        None => (i, ""),
52    };
53    validate_authority::<S>(maybe_authority)?;
54    validate_path_abempty::<S>(maybe_path)
55}
56
57/// Returns `Ok(_)` if the string matches `URI`/`IRI` rules.
58#[inline]
59pub(crate) fn validate_uri<S: Spec>(i: &str) -> Result<(), Error> {
60    validate_uri_reference_common::<S>(i, UriReferenceRule::Absolute)
61}
62
63/// Returns `Ok(_)` if the string matches `URI-reference`/`IRI-reference` rules.
64#[inline]
65pub(crate) fn validate_uri_reference<S: Spec>(i: &str) -> Result<(), Error> {
66    validate_uri_reference_common::<S>(i, UriReferenceRule::Any)
67}
68
69/// Returns `Ok(_)` if the string matches `absolute-URI`/`absolute-IRI` rules.
70#[inline]
71pub(crate) fn validate_absolute_uri<S: Spec>(i: &str) -> Result<(), Error> {
72    validate_uri_reference_common::<S>(i, UriReferenceRule::AbsoluteWithoutFragment)
73}
74
75/// Syntax rule for URI/IRI references.
76#[derive(Clone, Copy, PartialEq, Eq, Hash)]
77enum UriReferenceRule {
78    /// `URI` and `IRI`.
79    ///
80    /// This can have a fragment.
81    Absolute,
82    /// `absolute-URI` and `absolute-IRI`.
83    ///
84    /// This cannot have a fragment.
85    AbsoluteWithoutFragment,
86    /// `URI-reference` and `IRI-reference`.
87    ///
88    /// This can be relative.
89    Any,
90}
91
92impl UriReferenceRule {
93    /// Returns `true` is the relative reference is allowed.
94    #[inline]
95    #[must_use]
96    fn is_relative_allowed(self) -> bool {
97        self == Self::Any
98    }
99
100    /// Returns `true` is the fragment part is allowed.
101    #[inline]
102    #[must_use]
103    fn is_fragment_allowed(self) -> bool {
104        matches!(self, Self::Absolute | Self::Any)
105    }
106}
107
108/// Returns `Ok(_)` if the string matches `URI-reference`/`IRI-reference` rules.
109fn validate_uri_reference_common<S: Spec>(
110    i: &str,
111    ref_rule: UriReferenceRule,
112) -> Result<(), Error> {
113    // Validate `scheme ":"`.
114    let (i, _scheme) = match find_split_hole(i, b':') {
115        None => {
116            if ref_rule.is_relative_allowed() {
117                return validate_relative_ref::<S>(i);
118            } else {
119                return Err(Error::new());
120            }
121        }
122        Some(("", _)) => return Err(Error::new()),
123        Some((maybe_scheme, rest)) => {
124            if validate_scheme(maybe_scheme).is_err() {
125                // The string before the first colon is not a scheme.
126                // Falling back to `relative-ref` parsing.
127                if ref_rule.is_relative_allowed() {
128                    return validate_relative_ref::<S>(i);
129                } else {
130                    return Err(Error::new());
131                }
132            }
133            (rest, maybe_scheme)
134        }
135    };
136
137    // Validate `hier-part`.
138    let after_path = match i.strip_prefix("//") {
139        Some(i) => {
140            let (maybe_authority_path, after_path) = match find_split2_hole(i, b'?', b'#') {
141                Some((maybe_authority_path, c, rest)) => (maybe_authority_path, Some((c, rest))),
142                None => (i, None),
143            };
144            validate_authority_path_abempty::<S>(maybe_authority_path)?;
145            after_path
146        }
147        None => {
148            let (maybe_path, after_path) = match find_split2_hole(i, b'?', b'#') {
149                Some((maybe_path, c, rest)) => (maybe_path, Some((c, rest))),
150                None => (i, None),
151            };
152            // Authority is absent.
153            validate_path_absolute_authority_absent::<S>(maybe_path)?;
154            after_path
155        }
156    };
157
158    // Validate `[ "?" query ] [ "#" fragment ]`.
159    if let Some((first, rest)) = after_path {
160        validate_after_path::<S>(first, rest, ref_rule.is_fragment_allowed())?;
161    }
162    Ok(())
163}
164
165/// Returns `Ok(_)` if the string matches `relative-ref`/`irelative-ref` rules.
166pub(crate) fn validate_relative_ref<S: Spec>(i: &str) -> Result<(), Error> {
167    // Validate `relative-part`.
168    let after_path = match i.strip_prefix("//") {
169        Some(i) => {
170            let (maybe_authority_path, after_path) = match find_split2_hole(i, b'?', b'#') {
171                Some((maybe_authority_path, c, rest)) => (maybe_authority_path, Some((c, rest))),
172                None => (i, None),
173            };
174            validate_authority_path_abempty::<S>(maybe_authority_path)?;
175            after_path
176        }
177        None => {
178            let (maybe_path, after_path) = match find_split2_hole(i, b'?', b'#') {
179                Some((maybe_path, c, rest)) => (maybe_path, Some((c, rest))),
180                None => (i, None),
181            };
182            // Authority is absent.
183            validate_path_relative_authority_absent::<S>(maybe_path)?;
184            after_path
185        }
186    };
187
188    // Validate `[ "?" query ] [ "#" fragment ]`.
189    if let Some((first, rest)) = after_path {
190        validate_after_path::<S>(first, rest, true)?;
191    }
192    Ok(())
193}
194
195/// Returns `Ok(_)` if the string matches `[ "?" query ] [ "#" fragment ]` (or IRI version).
196fn validate_after_path<S: Spec>(first: u8, rest: &str, accept_fragment: bool) -> Result<(), Error> {
197    let (maybe_query, maybe_fragment) = if first == b'?' {
198        match find_split_hole(rest, b'#') {
199            Some(v) => v,
200            None => (rest, ""),
201        }
202    } else {
203        debug_assert_eq!(first, b'#');
204        ("", rest)
205    };
206    validate_query::<S>(maybe_query)?;
207    if !accept_fragment && !maybe_fragment.is_empty() {
208        return Err(Error::new());
209    }
210    validate_fragment::<S>(maybe_fragment)
211}
212
213/// Returns `Ok(_)` if the string matches `fragment`/`ifragment` rules.
214pub(crate) fn validate_fragment<S: Spec>(i: &str) -> Result<(), Error> {
215    let is_valid = satisfy_chars_with_pct_encoded(
216        i,
217        char::is_ascii_frag_query,
218        char::is_nonascii_fragment::<S>,
219    );
220    if is_valid {
221        Ok(())
222    } else {
223        Err(Error::new())
224    }
225}
iri_string/parser/validate.rs

iri_string/parser/
validate.rs
