1//! Contains parser configuration structure.
2use std::collections::HashMap;
3use std::io::Read;
4
5use crate::reader::EventReader;
6use crate::util::Encoding;
7
8/// Limits to defend from billion laughs attack
9const DEFAULT_MAX_ENTITY_EXPANSION_LENGTH: usize = 1_000_000;
10const DEFAULT_MAX_ENTITY_EXPANSION_DEPTH: u8 = 10;
11
12/// Parser configuration structure. **There are more config methods than public fileds — see methods below**.
13///
14/// This structure contains various configuration options which affect
15/// behavior of the parser.
16#[derive(Clone, PartialEq, Eq, Debug)]
17pub struct ParserConfig {
18 /// Whether or not should whitespace in textual events be removed. Default is false.
19 ///
20 /// When true, all standalone whitespace will be removed (this means no
21 /// `Whitespace` events will be emitted), and leading and trailing whitespace
22 /// from `Character` events will be deleted. If after trimming `Characters`
23 /// event will be empty, it will also be omitted from output stream. This is
24 /// possible, however, only if `whitespace_to_characters` or
25 /// `cdata_to_characters` options are set.
26 ///
27 /// This option does not affect CDATA events, unless `cdata_to_characters`
28 /// option is also set. In that case CDATA content will also be trimmed.
29 pub trim_whitespace: bool,
30
31 /// Whether or not should whitespace be converted to characters.
32 /// Default is false.
33 ///
34 /// If true, instead of `Whitespace` events `Characters` events with the
35 /// same content will be emitted. If `trim_whitespace` is also true, these
36 /// events will be trimmed to nothing and, consequently, not emitted.
37 pub whitespace_to_characters: bool,
38
39 /// Whether or not should CDATA be converted to characters.
40 /// Default is false.
41 ///
42 /// If true, instead of `CData` events `Characters` events with the same
43 /// content will be emitted. If `trim_whitespace` is also true, these events
44 /// will be trimmed. If corresponding CDATA contained nothing but whitespace,
45 /// this event will be omitted from the stream.
46 pub cdata_to_characters: bool,
47
48 /// Whether or not should comments be omitted. Default is true.
49 ///
50 /// If true, `Comment` events will not be emitted at all.
51 pub ignore_comments: bool,
52
53 /// Whether or not should sequential `Characters` events be merged.
54 /// Default is true.
55 ///
56 /// If true, multiple sequential `Characters` events will be merged into
57 /// a single event, that is, their data will be concatenated.
58 ///
59 /// Multiple sequential `Characters` events are only possible if either
60 /// `cdata_to_characters` or `ignore_comments` are set. Otherwise character
61 /// events will always be separated by other events.
62 pub coalesce_characters: bool,
63
64 /// A map of extra entities recognized by the parser. Default is an empty map.
65 ///
66 /// By default the XML parser recognizes the entities defined in the XML spec. Sometimes,
67 /// however, it is convenient to make the parser recognize additional entities which
68 /// are also not available through the DTD definitions (especially given that at the moment
69 /// DTD parsing is not supported).
70 pub extra_entities: HashMap<String, String>,
71
72 /// Whether or not the parser should ignore the end of stream. Default is false.
73 ///
74 /// By default the parser will either error out when it encounters a premature end of
75 /// stream or complete normally if the end of stream was expected. If you want to continue
76 /// reading from a stream whose input is supplied progressively, you can set this option to true.
77 /// In this case the parser will allow you to invoke the `next()` method even if a supposed end
78 /// of stream has happened.
79 ///
80 /// Note that support for this functionality is incomplete; for example, the parser will fail if
81 /// the premature end of stream happens inside PCDATA. Therefore, use this option at your own risk.
82 pub ignore_end_of_stream: bool,
83
84 /// Whether or not non-unicode entity references get replaced with the replacement character
85 ///
86 /// When true, any decimal or hexadecimal character reference that cannot be converted from a
87 /// u32 to a char using [std::char::from_u32](https://doc.rust-lang.org/std/char/fn.from_u32.html)
88 /// will be converted into the unicode REPLACEMENT CHARACTER (U+FFFD).
89 pub replace_unknown_entity_references: bool,
90
91 /// Whether or not whitespace at the root level of the document is ignored. Default is true.
92 ///
93 /// By default any whitespace that is not enclosed within at least one level of elements will be
94 /// ignored. Setting this value to false will cause root level whitespace events to be emitted.
95 ///
96 /// **There are configuration options – see methods below**
97 pub ignore_root_level_whitespace: bool,
98}
99
100impl ParserConfig {
101 /// Returns a new config with default values.
102 ///
103 /// You can tweak default values using builder-like pattern:
104 ///
105 /// ```rust
106 /// use xml::reader::ParserConfig;
107 ///
108 /// let config = ParserConfig::new()
109 /// .trim_whitespace(true)
110 /// .ignore_comments(true)
111 /// .coalesce_characters(false);
112 /// ```
113 #[must_use]
114 #[inline]
115 pub fn new() -> Self {
116 Self {
117 trim_whitespace: false,
118 whitespace_to_characters: false,
119 cdata_to_characters: false,
120 ignore_comments: true,
121 coalesce_characters: true,
122 extra_entities: HashMap::new(),
123 ignore_end_of_stream: false,
124 replace_unknown_entity_references: false,
125 ignore_root_level_whitespace: true,
126 }
127 }
128
129 /// Creates an XML reader with this configuration. The reader should be wrapped in a `BufReader`, otherwise parsing may be very slow.
130 ///
131 /// This is a convenience method for configuring and creating a reader at the same time:
132 ///
133 /// ```rust
134 /// use xml::reader::ParserConfig;
135 ///
136 /// let mut source: &[u8] = b"...";
137 ///
138 /// let reader = ParserConfig::new()
139 /// .trim_whitespace(true)
140 /// .ignore_comments(true)
141 /// .coalesce_characters(false)
142 /// .create_reader(&mut source);
143 /// ```
144 ///
145 /// This method is exactly equivalent to calling `EventReader::new_with_config()` with
146 /// this configuration object.
147 #[inline]
148 pub fn create_reader<R: Read>(self, source: R) -> EventReader<R> {
149 EventReader::new_with_config(source, self)
150 }
151
152 /// Adds a new entity mapping and returns an updated config object.
153 ///
154 /// This is a convenience method for adding external entities mappings to the XML parser.
155 /// An example:
156 ///
157 /// ```rust
158 /// use xml::reader::ParserConfig;
159 ///
160 /// let mut source: &[u8] = b"...";
161 ///
162 /// let reader = ParserConfig::new()
163 /// .add_entity("nbsp", " ")
164 /// .add_entity("copy", "©")
165 /// .add_entity("reg", "®")
166 /// .create_reader(&mut source);
167 /// ```
168 #[must_use]
169 pub fn add_entity<S: Into<String>, T: Into<String>>(mut self, entity: S, value: T) -> Self {
170 self.extra_entities.insert(entity.into(), value.into());
171 self
172 }
173}
174
175impl Default for ParserConfig {
176 #[inline]
177 fn default() -> Self {
178 Self::new()
179 }
180}
181
182gen_setters! { ParserConfig,
183 trim_whitespace: val bool,
184 whitespace_to_characters: val bool,
185 cdata_to_characters: val bool,
186 ignore_comments: val bool,
187 coalesce_characters: val bool,
188 ignore_end_of_stream: val bool,
189 replace_unknown_entity_references: val bool,
190 ignore_root_level_whitespace: val bool
191}
192
193/// Backwards-compatible extension of `ParserConfig`, which will eventually be merged into the original `ParserConfig` struct
194#[derive(Clone, PartialEq, Eq, Debug)]
195#[non_exhaustive]
196pub struct ParserConfig2 {
197 pub(crate) c: ParserConfig,
198
199 /// Use this encoding as the default. Necessary for UTF-16 files without BOM.
200 pub override_encoding: Option<Encoding>,
201
202 /// Allow `<?xml encoding="…">` to contain unsupported encoding names,
203 /// and interpret them as Latin1 instead. This will mangle non-ASCII characters, but usually it won't fail parsing.
204 pub ignore_invalid_encoding_declarations: bool,
205
206 /// Documents with multiple root elements are ill-formed
207 pub allow_multiple_root_elements: bool,
208
209 /// Abort if custom entities create a string longer than this
210 pub max_entity_expansion_length: usize,
211 /// Entities can expand into other entities this many times (be careful about exponential cost!)
212 pub max_entity_expansion_depth: u8,
213
214 /// Maximum length of tag name or attribute name
215 pub max_name_length: usize,
216
217 /// Max number of attributes per element
218 pub max_attributes: usize,
219
220 /// Max number of bytes in each attribute
221 pub max_attribute_length: usize,
222
223 /// Maximum length of strings reprsenting characters, comments, and processing instructions
224 pub max_data_length: usize,
225}
226
227impl Default for ParserConfig2 {
228 fn default() -> Self {
229 Self {
230 c: ParserConfig::default(),
231 override_encoding: None,
232 ignore_invalid_encoding_declarations: false,
233 allow_multiple_root_elements: true,
234 max_entity_expansion_length: DEFAULT_MAX_ENTITY_EXPANSION_LENGTH,
235 max_entity_expansion_depth: DEFAULT_MAX_ENTITY_EXPANSION_DEPTH,
236 max_attributes: 1 << 16,
237 max_attribute_length: 1 << 30,
238 max_data_length: 1 << 30,
239 max_name_length: 1 << 18,
240 }
241 }
242}
243
244impl ParserConfig2 {
245 /// Create extended configuration struct
246 #[inline]
247 #[must_use]
248 pub fn new() -> Self {
249 Self::default()
250 }
251
252 /// Read character encoding from `Content-Type` header.
253 /// Set this when parsing XML documents fetched over HTTP.
254 ///
255 /// `text/*` MIME types do *not* imply latin1. UTF-8 is always the default fallback.
256 #[must_use] pub fn content_type(mut self, mime_type: &str) -> Self {
257 let charset = mime_type.split_once(';')
258 .and_then(|(_, args)| args.split_once("charset"))
259 .and_then(|(_, args)| args.split_once('='));
260 if let Some((_, charset)) = charset {
261 let name = charset.trim().trim_matches('"');
262 if let Ok(enc) = name.parse() {
263 self.override_encoding = Some(enc);
264 }
265 }
266 self
267 }
268
269 /// Creates an XML reader with this configuration. The reader should be wrapped in a `BufReader`, otherwise parsing may be very slow.
270 ///
271 /// This is a convenience method for configuring and creating a reader at the same time:
272 ///
273 /// ```rust
274 /// use xml::reader::ParserConfig;
275 ///
276 /// let mut source: &[u8] = b"...";
277 ///
278 /// let reader = ParserConfig::new()
279 /// .trim_whitespace(true)
280 /// .ignore_comments(true)
281 /// .coalesce_characters(false)
282 /// .create_reader(&mut source);
283 /// ```
284 ///
285 /// This method is exactly equivalent to calling `EventReader::new_with_config()` with
286 /// this configuration object.
287 #[inline]
288 pub fn create_reader<R: Read>(self, source: R) -> EventReader<R> {
289 EventReader::new_with_config(source, self)
290 }
291}
292
293impl From<ParserConfig> for ParserConfig2 {
294 #[inline]
295 fn from(c: ParserConfig) -> Self {
296 Self { c, ..Default::default() }
297 }
298}
299
300gen_setters! { ParserConfig2,
301 /// Set if you got one in the HTTP header
302 override_encoding: val Option<Encoding>,
303 /// Allows invalid documents. There should be only a single root element in XML.
304 allow_multiple_root_elements: val bool,
305 /// Abort if custom entities create a string longer than this
306 max_entity_expansion_length: val usize,
307 /// Entities can expand into other entities this many times (be careful about exponential cost!)
308 max_entity_expansion_depth: val u8,
309 /// Max number of attributes per element
310 max_attributes: val usize,
311 /// Maximum length of tag name or attribute name
312 max_name_length: val usize,
313 /// Max number of bytes in each attribute
314 max_attribute_length: val usize,
315 /// Maximum length of strings reprsenting characters, comments, and processing instructions
316 max_data_length: val usize,
317 /// Allow `<?xml encoding="bogus"?>`
318 ignore_invalid_encoding_declarations: val bool
319}
320
321gen_setters! { ParserConfig,
322 /// Set if you got one in the HTTP header (see `content_type`)
323 override_encoding: c2 Option<Encoding>,
324 /// Allow `<?xml encoding="bogus"?>`
325 ignore_invalid_encoding_declarations: c2 bool,
326 /// Allows invalid documents. There should be only a single root element in XML.
327 allow_multiple_root_elements: c2 bool,
328
329 /// Abort if custom entities create a string longer than this
330 max_entity_expansion_length: c2 usize,
331 /// Entities can expand into other entities this many times (be careful about exponential cost!)
332 max_entity_expansion_depth: c2 u8,
333 /// Max number of attributes per element
334 max_attributes: c2 usize,
335 /// Maximum length of tag name or attribute name
336 max_name_length: c2 usize,
337 /// Max number of bytes in each attribute
338 max_attribute_length: c2 usize,
339 /// Maximum length of strings reprsenting characters, comments, and processing instructions
340 max_data_length: c2 usize,
341
342 /// Set encoding from the MIME type. Important for HTTP compatibility.
343 content_type: c2 &str
344}
345
346gen_setters! { ParserConfig2,
347 trim_whitespace: delegate bool,
348 whitespace_to_characters: delegate bool,
349 cdata_to_characters: delegate bool,
350 ignore_comments: delegate bool,
351 coalesce_characters: delegate bool,
352 ignore_end_of_stream: delegate bool,
353 replace_unknown_entity_references: delegate bool,
354 /// Whether or not whitespace at the root level of the document is ignored. Default is true.
355 ignore_root_level_whitespace: delegate bool
356}
357
358#[test]
359fn mime_parse() {
360 let c = ParserConfig2::new().content_type("text/xml;charset=Us-AScii").max_entity_expansion_length(1000);
361 assert_eq!(c.override_encoding, Some(Encoding::Ascii));
362
363 let c = ParserConfig2::new().max_entity_expansion_depth(3).content_type("text/xml;charset = \"UTF-16\"");
364 assert_eq!(c.override_encoding, Some(Encoding::Utf16));
365}
366