| 1 | // Copyright 2014-2017 The html5ever Project Developers. See the |
| 2 | // COPYRIGHT file at the top-level directory of this distribution. |
| 3 | // |
| 4 | // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or |
| 5 | // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license |
| 6 | // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your |
| 7 | // option. This file may not be copied, modified, or distributed |
| 8 | // except according to those terms. |
| 9 | //! Types for tag and attribute names, and tree-builder functionality. |
| 10 | |
| 11 | use std::cell::Ref; |
| 12 | use std::fmt; |
| 13 | use tendril::StrTendril; |
| 14 | |
| 15 | pub use self::tree_builder::{create_element, AppendNode, AppendText, ElementFlags, NodeOrText}; |
| 16 | pub use self::tree_builder::{ElemName, NextParserState, Tracer, TreeSink}; |
| 17 | pub use self::tree_builder::{LimitedQuirks, NoQuirks, Quirks, QuirksMode}; |
| 18 | use super::{LocalName, Namespace, Prefix}; |
| 19 | |
| 20 | /// An [expanded name], containing the tag and the namespace. |
| 21 | /// |
| 22 | /// [expanded name]: https://www.w3.org/TR/REC-xml-names/#dt-expname |
| 23 | #[derive (Copy, Clone, Eq, Hash, PartialEq)] |
| 24 | pub struct ExpandedName<'a> { |
| 25 | pub ns: &'a Namespace, |
| 26 | pub local: &'a LocalName, |
| 27 | } |
| 28 | |
| 29 | impl ElemName for ExpandedName<'_> { |
| 30 | #[inline (always)] |
| 31 | fn ns(&self) -> &Namespace { |
| 32 | self.ns |
| 33 | } |
| 34 | |
| 35 | #[inline (always)] |
| 36 | fn local_name(&self) -> &LocalName { |
| 37 | self.local |
| 38 | } |
| 39 | } |
| 40 | |
| 41 | impl<'a> ElemName for Ref<'a, ExpandedName<'a>> { |
| 42 | #[inline (always)] |
| 43 | fn ns(&self) -> &Namespace { |
| 44 | self.ns |
| 45 | } |
| 46 | |
| 47 | #[inline (always)] |
| 48 | fn local_name(&self) -> &LocalName { |
| 49 | self.local |
| 50 | } |
| 51 | } |
| 52 | |
| 53 | impl fmt::Debug for ExpandedName<'_> { |
| 54 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 55 | if self.ns.is_empty() { |
| 56 | write!(f, " {}" , self.local) |
| 57 | } else { |
| 58 | write!(f, " {{{}}}: {}" , self.ns, self.local) |
| 59 | } |
| 60 | } |
| 61 | } |
| 62 | |
| 63 | /// Helper to quickly create an expanded name. |
| 64 | /// |
| 65 | /// Can be used with no namespace as `expanded_name!("", "some_name")` |
| 66 | /// or with a namespace as `expanded_name!(ns "some_name")`. In the |
| 67 | /// latter case, `ns` is one of the symbols which the [`ns!`][ns] |
| 68 | /// macro accepts; note the lack of a comma between the `ns` and |
| 69 | /// `"some_name"`. |
| 70 | /// |
| 71 | /// [ns]: macro.ns.html |
| 72 | /// |
| 73 | /// # Examples |
| 74 | /// |
| 75 | /// ``` |
| 76 | /// # #[macro_use ] extern crate markup5ever; |
| 77 | /// |
| 78 | /// # fn main() { |
| 79 | /// use markup5ever::ExpandedName; |
| 80 | /// |
| 81 | /// assert_eq!( |
| 82 | /// expanded_name!("" , "div" ), |
| 83 | /// ExpandedName { |
| 84 | /// ns: &ns!(), |
| 85 | /// local: &local_name!("div" ) |
| 86 | /// } |
| 87 | /// ); |
| 88 | /// |
| 89 | /// assert_eq!( |
| 90 | /// expanded_name!(html "div" ), |
| 91 | /// ExpandedName { |
| 92 | /// ns: &ns!(html), |
| 93 | /// local: &local_name!("div" ) |
| 94 | /// } |
| 95 | /// ); |
| 96 | /// # } |
| 97 | #[macro_export ] |
| 98 | macro_rules! expanded_name { |
| 99 | ("" , $local: tt) => { |
| 100 | $crate::interface::ExpandedName { |
| 101 | ns: &ns!(), |
| 102 | local: &local_name!($local), |
| 103 | } |
| 104 | }; |
| 105 | ($ns: ident $local: tt) => { |
| 106 | $crate::interface::ExpandedName { |
| 107 | ns: &ns!($ns), |
| 108 | local: &local_name!($local), |
| 109 | } |
| 110 | }; |
| 111 | } |
| 112 | |
| 113 | pub mod tree_builder; |
| 114 | |
| 115 | /// A fully qualified name (with a namespace), used to depict names of tags and attributes. |
| 116 | /// |
| 117 | /// Namespaces can be used to differentiate between similar XML fragments. For example: |
| 118 | /// |
| 119 | /// ```text |
| 120 | /// // HTML |
| 121 | /// <table> |
| 122 | /// <tr> |
| 123 | /// <td>Apples</td> |
| 124 | /// <td>Bananas</td> |
| 125 | /// </tr> |
| 126 | /// </table> |
| 127 | /// |
| 128 | /// // Furniture XML |
| 129 | /// <table> |
| 130 | /// <name>African Coffee Table</name> |
| 131 | /// <width>80</width> |
| 132 | /// <length>120</length> |
| 133 | /// </table> |
| 134 | /// ``` |
| 135 | /// |
| 136 | /// Without XML namespaces, we can't use those two fragments in the same document |
| 137 | /// at the same time. However if we declare a namespace we could instead say: |
| 138 | /// |
| 139 | /// ```text |
| 140 | /// |
| 141 | /// // Furniture XML |
| 142 | /// <furn:table xmlns:furn="https://furniture.rs"> |
| 143 | /// <furn:name>African Coffee Table</furn:name> |
| 144 | /// <furn:width>80</furn:width> |
| 145 | /// <furn:length>120</furn:length> |
| 146 | /// </furn:table> |
| 147 | /// ``` |
| 148 | /// |
| 149 | /// and bind the prefix `furn` to a different namespace. |
| 150 | /// |
| 151 | /// For this reason we parse names that contain a colon in the following way: |
| 152 | /// |
| 153 | /// ```text |
| 154 | /// <furn:table> |
| 155 | /// | | |
| 156 | /// | +- local name |
| 157 | /// | |
| 158 | /// prefix (when resolved gives namespace_url `https://furniture.rs`) |
| 159 | /// ``` |
| 160 | /// |
| 161 | /// NOTE: `Prefix`, `LocalName` and `Prefix` are all derivative of |
| 162 | /// `string_cache::atom::Atom` and `Atom` implements `Deref<str>`. |
| 163 | /// |
| 164 | #[derive (PartialEq, Eq, PartialOrd, Ord, Hash, Debug, Clone)] |
| 165 | #[cfg_attr (feature = "heap_size" , derive(HeapSizeOf))] |
| 166 | pub struct QualName { |
| 167 | /// The prefix of qualified (e.g. `furn` in `<furn:table>` above). |
| 168 | /// Optional (since some namespaces can be empty or inferred), and |
| 169 | /// only useful for namespace resolution (since different prefix |
| 170 | /// can still resolve to same namespace) |
| 171 | /// |
| 172 | /// ``` |
| 173 | /// |
| 174 | /// # fn main() { |
| 175 | /// use markup5ever::{QualName, Namespace, LocalName, Prefix}; |
| 176 | /// |
| 177 | /// let qual = QualName::new( |
| 178 | /// Some(Prefix::from("furn" )), |
| 179 | /// Namespace::from("https://furniture.rs" ), |
| 180 | /// LocalName::from("table" ), |
| 181 | /// ); |
| 182 | /// |
| 183 | /// assert_eq!("furn" , &qual.prefix.unwrap()); |
| 184 | /// |
| 185 | /// # } |
| 186 | /// ``` |
| 187 | pub prefix: Option<Prefix>, |
| 188 | /// The namespace after resolution (e.g. `https://furniture.rs` in example above). |
| 189 | /// |
| 190 | /// ``` |
| 191 | /// # use markup5ever::{QualName, Namespace, LocalName, Prefix}; |
| 192 | /// |
| 193 | /// # fn main() { |
| 194 | /// # let qual = QualName::new( |
| 195 | /// # Some(Prefix::from("furn" )), |
| 196 | /// # Namespace::from("https://furniture.rs" ), |
| 197 | /// # LocalName::from("table" ), |
| 198 | /// # ); |
| 199 | /// |
| 200 | /// assert_eq!("https://furniture.rs" , &qual.ns); |
| 201 | /// # } |
| 202 | /// ``` |
| 203 | /// |
| 204 | /// When matching namespaces used by HTML we can use `ns!` macro. |
| 205 | /// Although keep in mind that ns! macro only works with namespaces |
| 206 | /// that are present in HTML spec (like `html`, `xmlns`, `svg`, etc.). |
| 207 | /// |
| 208 | /// ``` |
| 209 | /// #[macro_use] extern crate markup5ever; |
| 210 | /// |
| 211 | /// # use markup5ever::{QualName, Namespace, LocalName, Prefix}; |
| 212 | /// |
| 213 | /// let html_table = QualName::new( |
| 214 | /// None, |
| 215 | /// ns!(html), |
| 216 | /// LocalName::from("table" ), |
| 217 | /// ); |
| 218 | /// |
| 219 | /// assert!( |
| 220 | /// match html_table.ns { |
| 221 | /// ns!(html) => true, |
| 222 | /// _ => false, |
| 223 | /// } |
| 224 | /// ); |
| 225 | /// |
| 226 | /// ``` |
| 227 | pub ns: Namespace, |
| 228 | /// The local name (e.g. `table` in `<furn:table>` above). |
| 229 | /// |
| 230 | /// ``` |
| 231 | /// # use markup5ever::{QualName, Namespace, LocalName, Prefix}; |
| 232 | /// |
| 233 | /// # fn main() { |
| 234 | /// # let qual = QualName::new( |
| 235 | /// # Some(Prefix::from("furn" )), |
| 236 | /// # Namespace::from("https://furniture.rs" ), |
| 237 | /// # LocalName::from("table" ), |
| 238 | /// # ); |
| 239 | /// |
| 240 | /// assert_eq!("table" , &qual.local); |
| 241 | /// # } |
| 242 | /// ``` |
| 243 | /// When matching local name we can also use the `local_name!` macro: |
| 244 | /// |
| 245 | /// ``` |
| 246 | /// #[macro_use] extern crate markup5ever; |
| 247 | /// |
| 248 | /// # use markup5ever::{QualName, Namespace, LocalName, Prefix}; |
| 249 | /// |
| 250 | /// # let qual = QualName::new( |
| 251 | /// # Some(Prefix::from("furn" )), |
| 252 | /// # Namespace::from("https://furniture.rs" ), |
| 253 | /// # LocalName::from("table" ), |
| 254 | /// # ); |
| 255 | /// |
| 256 | /// // Initialize qual to furniture example |
| 257 | /// |
| 258 | /// assert!( |
| 259 | /// match qual.local { |
| 260 | /// local_name!("table" ) => true, |
| 261 | /// _ => false, |
| 262 | /// } |
| 263 | /// ); |
| 264 | /// |
| 265 | /// ``` |
| 266 | pub local: LocalName, |
| 267 | } |
| 268 | |
| 269 | impl ElemName for Ref<'_, QualName> { |
| 270 | #[inline (always)] |
| 271 | fn ns(&self) -> &Namespace { |
| 272 | &self.ns |
| 273 | } |
| 274 | |
| 275 | #[inline (always)] |
| 276 | fn local_name(&self) -> &LocalName { |
| 277 | &self.local |
| 278 | } |
| 279 | } |
| 280 | |
| 281 | impl ElemName for &QualName { |
| 282 | #[inline (always)] |
| 283 | fn ns(&self) -> &Namespace { |
| 284 | &self.ns |
| 285 | } |
| 286 | |
| 287 | #[inline (always)] |
| 288 | fn local_name(&self) -> &LocalName { |
| 289 | &self.local |
| 290 | } |
| 291 | } |
| 292 | |
| 293 | impl QualName { |
| 294 | /// Basic constructor function. |
| 295 | /// |
| 296 | /// First let's try it for the following example where `QualName` |
| 297 | /// is defined as: |
| 298 | /// ```text |
| 299 | /// <furn:table> <!-- namespace url is https://furniture.rs --> |
| 300 | /// ``` |
| 301 | /// |
| 302 | /// Given this definition, we can define `QualName` using strings. |
| 303 | /// |
| 304 | /// ``` |
| 305 | /// use markup5ever::{QualName, Namespace, LocalName, Prefix}; |
| 306 | /// |
| 307 | /// # fn main() { |
| 308 | /// let qual_name = QualName::new( |
| 309 | /// Some(Prefix::from("furn" )), |
| 310 | /// Namespace::from("https://furniture.rs" ), |
| 311 | /// LocalName::from("table" ), |
| 312 | /// ); |
| 313 | /// # } |
| 314 | /// ``` |
| 315 | /// |
| 316 | /// If we were instead to construct this element instead: |
| 317 | /// |
| 318 | /// ```text |
| 319 | /// |
| 320 | /// <table> |
| 321 | /// ^^^^^---- no prefix and thus default html namespace |
| 322 | /// |
| 323 | /// ``` |
| 324 | /// |
| 325 | /// Or could define it using macros, like so: |
| 326 | /// |
| 327 | /// ``` |
| 328 | /// #[macro_use] extern crate markup5ever; |
| 329 | /// use markup5ever::{QualName, Namespace, LocalName, Prefix}; |
| 330 | /// |
| 331 | /// # fn main() { |
| 332 | /// let qual_name = QualName::new( |
| 333 | /// None, |
| 334 | /// ns!(html), |
| 335 | /// local_name!("table" ) |
| 336 | /// ); |
| 337 | /// # } |
| 338 | /// ``` |
| 339 | /// |
| 340 | /// Let's analyse the above example. |
| 341 | /// Since we have no prefix its value is None. Second we have html namespace. |
| 342 | /// In html5ever html namespaces are supported out of the box, |
| 343 | /// we can write `ns!(html)` instead of typing `Namespace::from("http://www.w3.org/1999/xhtml")`. |
| 344 | /// Local name is also one of the HTML elements local names, so can |
| 345 | /// use `local_name!("table")` macro. |
| 346 | /// |
| 347 | #[inline ] |
| 348 | pub fn new(prefix: Option<Prefix>, ns: Namespace, local: LocalName) -> QualName { |
| 349 | QualName { prefix, ns, local } |
| 350 | } |
| 351 | |
| 352 | /// Take a reference of `self` as an `ExpandedName`, dropping the unresolved prefix. |
| 353 | /// |
| 354 | /// In XML and HTML prefixes are only used to extract the relevant namespace URI. |
| 355 | /// Expanded name only contains resolved namespace and tag name, which are only |
| 356 | /// relevant parts of an XML or HTML tag and attribute name respectively. |
| 357 | /// |
| 358 | /// In lieu of our XML Namespace example |
| 359 | /// |
| 360 | /// ```text |
| 361 | /// <furn:table> <!-- namespace url is https://furniture.rs --> |
| 362 | /// ``` |
| 363 | /// For it the expanded name would become roughly equivalent to: |
| 364 | /// |
| 365 | /// ```text |
| 366 | /// ExpandedName { |
| 367 | /// ns: "https://furniture.rs", |
| 368 | /// local: "table", |
| 369 | /// } |
| 370 | /// ``` |
| 371 | /// |
| 372 | #[inline ] |
| 373 | pub fn expanded(&self) -> ExpandedName { |
| 374 | ExpandedName { |
| 375 | ns: &self.ns, |
| 376 | local: &self.local, |
| 377 | } |
| 378 | } |
| 379 | } |
| 380 | |
| 381 | /// A tag attribute, e.g. `class="test"` in `<div class="test" ...>`. |
| 382 | /// |
| 383 | /// The namespace on the attribute name is almost always ns!(""). |
| 384 | /// The tokenizer creates all attributes this way, but the tree |
| 385 | /// builder will adjust certain attribute names inside foreign |
| 386 | /// content (MathML, SVG). |
| 387 | #[derive (PartialEq, Eq, PartialOrd, Ord, Clone, Debug)] |
| 388 | pub struct Attribute { |
| 389 | /// The name of the attribute (e.g. the `class` in `<div class="test">`) |
| 390 | pub name: QualName, |
| 391 | /// The value of the attribute (e.g. the `"test"` in `<div class="test">`) |
| 392 | pub value: StrTendril, |
| 393 | } |
| 394 | |
| 395 | #[cfg (test)] |
| 396 | mod tests { |
| 397 | use super::Namespace; |
| 398 | |
| 399 | #[test ] |
| 400 | fn ns_macro() { |
| 401 | assert_eq!(ns!(), Namespace::from("" )); |
| 402 | |
| 403 | assert_eq!(ns!(html), Namespace::from("http://www.w3.org/1999/xhtml" )); |
| 404 | assert_eq!( |
| 405 | ns!(xml), |
| 406 | Namespace::from("http://www.w3.org/XML/1998/namespace" ) |
| 407 | ); |
| 408 | assert_eq!(ns!(xmlns), Namespace::from("http://www.w3.org/2000/xmlns/" )); |
| 409 | assert_eq!(ns!(xlink), Namespace::from("http://www.w3.org/1999/xlink" )); |
| 410 | assert_eq!(ns!(svg), Namespace::from("http://www.w3.org/2000/svg" )); |
| 411 | assert_eq!( |
| 412 | ns!(mathml), |
| 413 | Namespace::from("http://www.w3.org/1998/Math/MathML" ) |
| 414 | ); |
| 415 | } |
| 416 | } |
| 417 | |