xml/reader/
error.rs

1use crate::reader::lexer::Token;
2use crate::Encoding;
3
4use std::borrow::Cow;
5use std::error::Error as _;
6use std::{error, fmt, io, str};
7
8use crate::common::{Position, TextPosition};
9use crate::util;
10
11/// Failure reason
12#[derive(Debug)]
13pub enum ErrorKind {
14    /// This is an ill-formed XML document
15    Syntax(Cow<'static, str>),
16    /// Reader/writer reported an error
17    Io(io::Error),
18    /// The document contains bytes that are not allowed in UTF-8 strings
19    Utf8(str::Utf8Error),
20    /// The document ended while they were elements/comments/etc. still open
21    UnexpectedEof,
22}
23
24#[derive(Debug, Clone, PartialEq)]
25#[non_exhaustive]
26pub(crate) enum SyntaxError {
27    CannotRedefineXmlnsPrefix,
28    CannotRedefineXmlPrefix,
29    /// Recursive custom entity expanded to too many chars, it could be DoS
30    EntityTooBig,
31    EmptyEntity,
32    NoRootElement,
33    ProcessingInstructionWithoutName,
34    UnbalancedRootElement,
35    UnexpectedEof,
36    UnexpectedOpeningTag,
37    /// Missing `]]>`
38    UnclosedCdata,
39    UnexpectedQualifiedName(Token),
40    UnexpectedTokenOutsideRoot(Token),
41    UnexpectedToken(Token),
42    UnexpectedTokenInEntity(Token),
43    UnexpectedTokenInClosingTag(Token),
44    UnexpectedTokenInOpeningTag(Token),
45    InvalidQualifiedName(Box<str>),
46    UnboundAttribute(Box<str>),
47    UnboundElementPrefix(Box<str>),
48    UnexpectedClosingTag(Box<str>),
49    UnexpectedName(Box<str>),
50    /// Found <?xml-like PI not at the beginning of a document,
51    /// which is an error, see section 2.6 of XML 1.1 spec
52    UnexpectedProcessingInstruction(Box<str>, Token),
53    CannotUndefinePrefix(Box<str>),
54    InvalidCharacterEntity(u32),
55    InvalidDefaultNamespace(Box<str>),
56    InvalidNamePrefix(Box<str>),
57    InvalidNumericEntity(Box<str>),
58    InvalidStandaloneDeclaration(Box<str>),
59    InvalidXmlProcessingInstruction(Box<str>),
60    RedefinedAttribute(Box<str>),
61    UndefinedEntity(Box<str>),
62    UnexpectedEntity(Box<str>),
63    UnexpectedNameInsideXml(Box<str>),
64    UnsupportedEncoding(Box<str>),
65    /// In DTD
66    UnknownMarkupDeclaration(Box<str>),
67    UnexpectedXmlVersion(Box<str>),
68    ConflictingEncoding(Encoding, Encoding),
69    UnexpectedTokenBefore(&'static str, char),
70    /// Document has more stuff than `ParserConfig` allows
71    ExceededConfiguredLimit,
72}
73
74impl fmt::Display for SyntaxError {
75    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
76        self.to_cow().fmt(f)
77    }
78}
79
80impl SyntaxError {
81    #[inline(never)]
82    #[cold]
83    pub(crate) fn to_cow(&self) -> Cow<'static, str> {
84        match *self {
85            Self::CannotRedefineXmlnsPrefix => "Cannot redefine XMLNS prefix".into(),
86            Self::CannotRedefineXmlPrefix => "Default XMLNS prefix cannot be rebound to another value".into(),
87            Self::EmptyEntity => "Encountered empty entity".into(),
88            Self::EntityTooBig => "Entity too big".into(),
89            Self::NoRootElement => "Unexpected end of stream: no root element found".into(),
90            Self::ProcessingInstructionWithoutName => "Encountered processing instruction without a name".into(),
91            Self::UnbalancedRootElement => "Unexpected end of stream: still inside the root element".into(),
92            Self::UnclosedCdata => "Unclosed <![CDATA[".into(),
93            Self::UnexpectedEof => "Unexpected end of stream".into(),
94            Self::UnexpectedOpeningTag => "'<' is not allowed in attributes".into(),
95            Self::CannotUndefinePrefix(ref ln) => format!("Cannot undefine prefix '{ln}'").into(),
96            Self::ConflictingEncoding(a, b) => format!("Declared encoding {a}, but uses {b}").into(),
97            Self::InvalidCharacterEntity(num) => format!("Invalid character U+{num:04X}").into(),
98            Self::InvalidDefaultNamespace(ref name) => format!("Namespace '{name}' cannot be default").into(),
99            Self::InvalidNamePrefix(ref prefix) => format!("'{prefix}' cannot be an element name prefix").into(),
100            Self::InvalidNumericEntity(ref v) => format!("Invalid numeric entity: {v}").into(),
101            Self::InvalidQualifiedName(ref e) => format!("Qualified name is invalid: {e}").into(),
102            Self::InvalidStandaloneDeclaration(ref value) => format!("Invalid standalone declaration value: {value}").into(),
103            Self::InvalidXmlProcessingInstruction(ref name) => format!("Invalid processing instruction: <?{name} - \"<?xml\"-like PI is only valid at the beginning of the document").into(),
104            Self::RedefinedAttribute(ref name) => format!("Attribute '{name}' is redefined").into(),
105            Self::UnboundAttribute(ref name) => format!("Attribute {name} prefix is unbound").into(),
106            Self::UnboundElementPrefix(ref name) => format!("Element {name} prefix is unbound").into(),
107            Self::UndefinedEntity(ref v) => format!("Undefined entity: {v}").into(),
108            Self::UnexpectedClosingTag(ref expected_got) => format!("Unexpected closing tag: {expected_got}").into(),
109            Self::UnexpectedEntity(ref name) => format!("Unexpected entity: {name}").into(),
110            Self::UnexpectedName(ref name) => format!("Unexpected name: {name}").into(),
111            Self::UnexpectedNameInsideXml(ref name) => format!("Unexpected name inside XML declaration: {name}").into(),
112            Self::UnexpectedProcessingInstruction(ref buf, token) => format!("Unexpected token inside processing instruction: <?{buf}{token}").into(),
113            Self::UnexpectedQualifiedName(e) => format!("Unexpected token inside qualified name: {e}").into(),
114            Self::UnexpectedToken(token) => format!("Unexpected token: {token}").into(),
115            Self::UnexpectedTokenBefore(before, c) => format!("Unexpected token '{before}' before '{c}'").into(),
116            Self::UnexpectedTokenInClosingTag(token) => format!("Unexpected token inside closing tag: {token}").into(),
117            Self::UnexpectedTokenInEntity(token) => format!("Unexpected token inside entity: {token}").into(),
118            Self::UnexpectedTokenInOpeningTag(token) => format!("Unexpected token inside opening tag: {token}").into(),
119            Self::UnexpectedTokenOutsideRoot(token) => format!("Unexpected characters outside the root element: {token}").into(),
120            Self::UnexpectedXmlVersion(ref version) => format!("Invalid XML version: {version}").into(),
121            Self::UnknownMarkupDeclaration(ref v) => format!("Unknown markup declaration: {v}").into(),
122            Self::UnsupportedEncoding(ref v) => format!("Unsupported encoding: {v}").into(),
123            Self::ExceededConfiguredLimit => "This document is larger/more complex than allowed by the parser's configuration".into(),
124        }
125    }
126}
127
128/// An XML parsing error.
129///
130/// Consists of a 2D position in a document and a textual message describing the error.
131#[derive(Clone, PartialEq, Eq, Debug)]
132pub struct Error {
133    pub(crate) pos: TextPosition,
134    pub(crate) kind: ErrorKind,
135}
136
137impl fmt::Display for Error {
138    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
139        use self::ErrorKind::{Io, Syntax, UnexpectedEof, Utf8};
140
141        write!(f, "{} ", self.pos)?;
142        match &self.kind {
143            Io(io_error) => io_error.fmt(f),
144            Utf8(reason) => reason.fmt(f),
145            Syntax(msg) => f.write_str(msg),
146            UnexpectedEof => f.write_str("Unexpected EOF"),
147        }
148    }
149}
150
151impl Position for Error {
152    #[inline]
153    fn position(&self) -> TextPosition { self.pos }
154}
155
156impl Error {
157    /// Returns a reference to a message which is contained inside this error.
158    #[cold]
159    #[doc(hidden)]
160    #[allow(deprecated)]
161    #[must_use]
162    pub fn msg(&self) -> &str {
163        use self::ErrorKind::{Io, Syntax, UnexpectedEof, Utf8};
164        match &self.kind {
165            Io(io_error) => io_error.description(),
166            Utf8(reason) => reason.description(),
167            Syntax(msg) => msg.as_ref(),
168            UnexpectedEof => "Unexpected EOF",
169        }
170    }
171
172    /// Failure reason
173    #[must_use]
174    #[inline]
175    pub fn kind(&self) -> &ErrorKind {
176        &self.kind
177    }
178}
179
180impl error::Error for Error {
181    #[allow(deprecated)]
182    #[cold]
183    fn description(&self) -> &str { self.msg() }
184}
185
186impl<'a, P, M> From<(&'a P, M)> for Error where P: Position, M: Into<Cow<'static, str>> {
187    #[cold]
188    fn from(orig: (&'a P, M)) -> Self {
189        Self {
190            pos: orig.0.position(),
191            kind: ErrorKind::Syntax(orig.1.into()),
192        }
193    }
194}
195
196impl From<util::CharReadError> for Error {
197    #[cold]
198    fn from(e: util::CharReadError) -> Self {
199        use crate::util::CharReadError::{Io, UnexpectedEof, Utf8};
200        Self {
201            pos: TextPosition::new(),
202            kind: match e {
203                UnexpectedEof => ErrorKind::UnexpectedEof,
204                Utf8(reason) => ErrorKind::Utf8(reason),
205                Io(io_error) => ErrorKind::Io(io_error),
206            },
207        }
208    }
209}
210
211impl From<io::Error> for Error {
212    #[cold]
213    fn from(e: io::Error) -> Self {
214        Self {
215            pos: TextPosition::new(),
216            kind: ErrorKind::Io(e),
217        }
218    }
219}
220
221impl Clone for ErrorKind {
222    #[cold]
223    fn clone(&self) -> Self {
224        use self::ErrorKind::{Io, Syntax, UnexpectedEof, Utf8};
225        match self {
226            UnexpectedEof => UnexpectedEof,
227            Utf8(reason) => Utf8(*reason),
228            Io(io_error) => Io(io::Error::new(io_error.kind(), io_error.to_string())),
229            Syntax(msg) => Syntax(msg.clone()),
230        }
231    }
232}
233impl PartialEq for ErrorKind {
234    #[allow(deprecated)]
235    fn eq(&self, other: &Self) -> bool {
236        use self::ErrorKind::{Io, Syntax, UnexpectedEof, Utf8};
237        match (self, other) {
238            (UnexpectedEof, UnexpectedEof) => true,
239            (Utf8(left), Utf8(right)) => left == right,
240            (Io(left), Io(right)) =>
241                left.kind() == right.kind() &&
242                left.description() == right.description(),
243            (Syntax(left), Syntax(right)) =>
244                left == right,
245
246            (_, _) => false,
247        }
248    }
249}
250impl Eq for ErrorKind {}
251
252#[test]
253fn err_size() {
254    assert!(std::mem::size_of::<SyntaxError>() <= 24);
255}