/home/runner/work/MathCAT/MathCAT/src/interface.rs
Line | Count | Source |
1 | | //! The interface module provides functionality both for calling from an API and also running the code from `main`. |
2 | | //! |
3 | | #![allow(non_snake_case)] |
4 | | #![allow(clippy::needless_return)] |
5 | | use std::cell::RefCell; |
6 | | use std::sync::LazyLock; |
7 | | |
8 | | use crate::canonicalize::{as_text, create_mathml_element}; |
9 | | use crate::errors::*; |
10 | | use phf::phf_map; |
11 | | use regex::{Captures, Regex}; |
12 | | use sxd_document::dom::{Element, Document, ChildOfRoot, ChildOfElement, Attribute}; |
13 | | use sxd_document::parser; |
14 | | use sxd_document::Package; |
15 | | |
16 | | use crate::canonicalize::{as_element, name}; |
17 | | use crate::shim_filesystem::{find_all_dirs_shim, find_files_in_dir_that_ends_with_shim}; |
18 | | use log::{debug, error}; |
19 | | |
20 | | use crate::navigate::*; |
21 | | use crate::pretty_print::mml_to_string; |
22 | | use crate::xpath_functions::{is_leaf, IsNode}; |
23 | | use std::panic::{catch_unwind, AssertUnwindSafe}; |
24 | | |
25 | | /// Maximum depth to prevent stack overflow on deeply nested MathML |
26 | | pub const MAX_DEPTH: usize = 512; |
27 | | |
28 | | #[cfg(feature = "enable-logs")] |
29 | | use std::sync::Once; |
30 | | #[cfg(feature = "enable-logs")] |
31 | | static INIT: Once = Once::new(); |
32 | | |
33 | 45.0k | fn enable_logs() { |
34 | | #[cfg(feature = "enable-logs")] |
35 | | INIT.call_once(||{ |
36 | | #[cfg(target_os = "android")] |
37 | | { |
38 | | use log::*; |
39 | | use android_logger::*; |
40 | | |
41 | | android_logger::init_once( |
42 | | Config::default() |
43 | | .with_max_level(LevelFilter::Trace) |
44 | | .with_tag("MathCat") |
45 | | ); |
46 | | trace!("Activated Android logger!"); |
47 | | } |
48 | | }); |
49 | 45.0k | } |
50 | | |
51 | | // For getting a message from a panic |
52 | | thread_local! { |
53 | | // Stores (Message, File, Line) |
54 | | static PANIC_INFO: RefCell<Option<(String, String, u32)>> = const { RefCell::new(None) }; |
55 | | } |
56 | | |
57 | | /// Initialize the panic handler to catch panics and store the message, file, and line number in `PANIC_INFO`. |
58 | 13.5k | pub fn init_panic_handler() { |
59 | | use std::panic; |
60 | | |
61 | 13.5k | panic::set_hook(Box::new(|info| {1 |
62 | 1 | let location = info.location() |
63 | 1 | .map(|l| format!("{}:{}", l.file(), l.line())) |
64 | 1 | .unwrap_or_else(|| "unknown"0 .to_string0 ()); |
65 | | |
66 | 1 | let payload = info.payload(); |
67 | 1 | let msg = if let Some(s0 ) = payload.downcast_ref::<&'static str>() { |
68 | 0 | s.to_string() |
69 | 1 | } else if let Some(s) = payload.downcast_ref::<String>() { |
70 | 1 | s.clone() |
71 | | } else { |
72 | 0 | "Unknown panic payload".to_string() |
73 | | }; |
74 | | |
75 | | // Use try_with/try_borrow_mut to ensure the hook never panics itself |
76 | 1 | let _ = PANIC_INFO.try_with(|cell| { |
77 | 1 | if let Ok(mut slot) = cell.try_borrow_mut() { |
78 | 1 | *slot = Some((msg, location, 0)); |
79 | 1 | }0 |
80 | 1 | }); |
81 | 1 | })); |
82 | 13.5k | } |
83 | | |
84 | 41.1k | pub fn report_any_panic<T>(result: Result<Result<T, Error>, Box<dyn std::any::Any + Send>>) -> Result<T, Error> { |
85 | 41.1k | match result { |
86 | 41.1k | Ok(val) => val, |
87 | | Err(_) => { |
88 | | // Retrieve the smuggled info |
89 | 1 | let details = PANIC_INFO.with(|cell| cell.borrow_mut().take()); |
90 | | |
91 | 1 | if let Some((msg, file, line)) = details { |
92 | 1 | Err(anyhow::anyhow!( |
93 | 1 | "MathCAT crash! Please report the following information: '{}' at {}:{}", |
94 | 1 | msg, file, line |
95 | 1 | )) |
96 | | } else { |
97 | 0 | Err(anyhow::anyhow!("MathCAT crash! -- please report")) |
98 | | } |
99 | | } |
100 | | } |
101 | 41.1k | } |
102 | | |
103 | | // wrap up some common functionality between the call from 'main' and AT |
104 | 4.91k | fn cleanup_mathml(mathml: Element) -> Result<Element> { |
105 | 4.91k | trim_element(mathml, false); |
106 | 4.91k | let mathml4.91k = crate::canonicalize::canonicalize(mathml)?1 ; |
107 | 4.91k | let mathml = add_ids(mathml); |
108 | 4.91k | return Ok(mathml); |
109 | 4.91k | } |
110 | | |
111 | | thread_local! { |
112 | | /// The current node being navigated (also spoken and brailled) is stored in `MATHML_INSTANCE`. |
113 | | pub static MATHML_INSTANCE: RefCell<Package> = init_mathml_instance(); |
114 | | } |
115 | | |
116 | 3.92k | fn init_mathml_instance() -> RefCell<Package> { |
117 | 3.92k | let package = parser::parse("<math></math>") |
118 | 3.92k | .expect("Internal error in 'init_mathml_instance;: didn't parse initializer string"); |
119 | 3.92k | return RefCell::new(package); |
120 | 3.92k | } |
121 | | |
122 | | /// Set the Rules directory |
123 | | /// IMPORTANT: this should be the very first call to MathCAT. If 'dir' is an empty string, the environment var 'MathCATRulesDir' is tried. |
124 | 5.08k | pub fn set_rules_dir(dir: impl AsRef<str>) -> Result<()> { |
125 | 5.08k | enable_logs(); |
126 | 5.08k | init_panic_handler(); |
127 | 5.08k | let dir = dir.as_ref().to_string(); |
128 | 5.08k | let result = catch_unwind(AssertUnwindSafe(|| { |
129 | | use std::path::PathBuf; |
130 | 5.08k | let dir_os = if dir.is_empty() { |
131 | 0 | std::env::var_os("MathCATRulesDir").unwrap_or_default() |
132 | | } else { |
133 | 5.08k | std::ffi::OsString::from(&dir) |
134 | | }; |
135 | 5.08k | let pref_manager = crate::prefs::PreferenceManager::get(); |
136 | 5.08k | pref_manager.borrow_mut().initialize(PathBuf::from(dir_os)) |
137 | 5.08k | })); |
138 | 5.08k | return report_any_panic(result); |
139 | 5.08k | } |
140 | | |
141 | | /// Returns the version number (from Cargo.toml) of the build |
142 | 0 | pub fn get_version() -> String { |
143 | 0 | enable_logs(); |
144 | | const VERSION: &str = env!("CARGO_PKG_VERSION"); |
145 | 0 | return VERSION.to_string(); |
146 | 0 | } |
147 | | |
148 | | /// This will override any previous MathML that was set. |
149 | | /// This returns canonical MathML with 'id's set on any node that doesn't have an id. |
150 | | /// The ids can be used for sync highlighting if the `Bookmark` API preference is true. |
151 | 4.88k | pub fn set_mathml(mathml_str: impl AsRef<str>) -> Result<String> { |
152 | 4.88k | enable_logs(); |
153 | | // if these are present when resent to MathJaX, MathJaX crashes (https://github.com/mathjax/MathJax/issues/2822) |
154 | 3 | static MATHJAX_V2: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"class *= *['"]MJX-.*?['"]"#).unwrap()); |
155 | 3 | static MATHJAX_V3: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"class *= *['"]data-mjx-.*?['"]"#).unwrap()); |
156 | | |
157 | | // Strip out processing instructions and comments -- these are not MathML and can cause DOS problems in the parser |
158 | 3 | static PROCESSING_INSTRUCTION: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"<\?[\s\S]{1,2048}\?>"#).unwrap()); |
159 | 3 | static XML_COMMENT: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"(?s)"#).unwrap()); |
160 | | |
161 | | // These have some length limits to avoid DOS attacks via long strings |
162 | 3 | static NAMESPACE_DECL: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"xmlns:[[:alpha:]]{1,32}"#).unwrap()); |
163 | 3 | static PREFIX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"(</?)[[:alpha:]]{1,32}:"#).unwrap()); |
164 | 3 | static HTML_ENTITIES: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"&([a-zA-Z]{2,10});"#).unwrap()); |
165 | 4.88k | let result = catch_unwind(AssertUnwindSafe(|| { |
166 | 4.88k | NAVIGATION_STATE.with(|nav_stack| { |
167 | 4.88k | nav_stack.borrow_mut().reset(); |
168 | 4.88k | }); |
169 | | |
170 | | // We need the main definitions files to be read in so canonicalize can work. |
171 | | // This call reads all of them for the current preferences, but that's ok since they will likely be used |
172 | 4.88k | crate::speech::SPEECH_RULES.with(|rules| rules.borrow_mut().read_files())?0 ; |
173 | | |
174 | 4.88k | let mathml_str = mathml_str.as_ref(); |
175 | | // Safety guard: Reject strings > 1MB to prevent DoS/Stack issues |
176 | 4.88k | if mathml_str.len() > 1024 * 1024 { |
177 | 0 | bail!("MathML string of size {} bytes exceeds length limit of 1MB", mathml_str.len()); |
178 | 4.88k | } |
179 | | |
180 | 4.88k | return MATHML_INSTANCE.with(|old_package| { |
181 | | static HTML_ENTITIES_MAPPING: phf::Map<&str, &str> = include!("entities.in"); |
182 | | |
183 | 4.88k | let mut error_message = "".to_string(); // can't return a result inside the replace_all, so we do this hack of setting the message and then returning the error |
184 | | |
185 | 4.88k | let mathml_str = XML_COMMENT.replace_all(mathml_str, ""); |
186 | 4.88k | let mathml_str = PROCESSING_INSTRUCTION.replace_all(&mathml_str, ""); |
187 | | // FIX: need to deal with character data and convert to something the parser knows |
188 | 4.88k | let mathml_str = HTML_ENTITIES.replace_all(&mathml_str, |cap: &Captures| match HTML_ENTITIES_MAPPING96 .get96 (&cap[1]) { |
189 | | None => { |
190 | 1 | error_message = format!("No entity named '{}'", &cap[0]); |
191 | 1 | cap[0].to_string() |
192 | | } |
193 | 95 | Some(&ch) => ch.to_string(), |
194 | 96 | }); |
195 | | |
196 | 4.88k | if !error_message.is_empty() { |
197 | | // Clear stale state so subsequent API calls do not return previous user's data (security issue) |
198 | 1 | old_package.replace(parser::parse("<math></math>").unwrap()); |
199 | 1 | bail!(error_message); |
200 | 4.88k | } |
201 | 4.88k | let mathml_str = MATHJAX_V2.replace_all(&mathml_str, ""); |
202 | 4.88k | let mathml_str = MATHJAX_V3.replace_all(&mathml_str, ""); |
203 | | |
204 | | // the speech rules use the xpath "name" function and that includes the prefix |
205 | | // getting rid of the prefix properly probably involves a recursive replacement in the tree |
206 | | // if the prefix is used, it is almost certainly something like "m" or "mml", so this cheat will work. |
207 | 4.88k | let mathml_str = NAMESPACE_DECL.replace(&mathml_str, "xmlns"); // do this before the PREFIX replace! |
208 | 4.88k | let mathml_str = PREFIX.replace_all(&mathml_str, "$1"); |
209 | | |
210 | 4.88k | let new_package = parser::parse(&mathml_str); |
211 | 4.88k | if let Err(e1 ) = new_package { |
212 | | // Clear stale state so subsequent API calls do not return previous user's data (security issue) |
213 | 1 | old_package.replace(parser::parse("<math></math>").unwrap()); |
214 | 1 | bail!("Invalid MathML input:\n{}\nError is: {}", &mathml_str, &e.to_string()); |
215 | 4.88k | } |
216 | | |
217 | 4.88k | let new_package = new_package.unwrap(); |
218 | 4.88k | let mathml = get_element(&new_package); |
219 | 4.88k | let mathml4.88k = cleanup_mathml(mathml)?1 ; |
220 | 4.88k | let mathml_string = mml_to_string(mathml); |
221 | 4.88k | old_package.replace(new_package); |
222 | | |
223 | 4.88k | return Ok(mathml_string); |
224 | 4.88k | }); |
225 | 4.88k | })); |
226 | | |
227 | 4.88k | return report_any_panic(result); |
228 | 4.88k | } |
229 | | |
230 | | /// Get the spoken text of the MathML that was set. |
231 | | /// The speech takes into account any AT or user preferences. |
232 | 3.46k | pub fn get_spoken_text() -> Result<String> { |
233 | 3.46k | enable_logs(); |
234 | 3.46k | let result = catch_unwind(AssertUnwindSafe(|| { |
235 | 3.46k | MATHML_INSTANCE.with(|package_instance| { |
236 | 3.46k | let package_instance = package_instance.borrow(); |
237 | 3.46k | let mathml = get_element(&package_instance); |
238 | 3.46k | let new_package = Package::new(); |
239 | 3.46k | let intent = crate::speech::intent_from_mathml(mathml, new_package.as_document())?0 ; |
240 | 3.46k | debug!("Intent tree:\n{}", mml_to_string0 (intent0 )); |
241 | 3.46k | let speech = crate::speech::speak_mathml(intent, "", 0)?0 ; |
242 | 3.46k | return Ok(speech); |
243 | 3.46k | }) |
244 | 3.46k | })); |
245 | 3.46k | return report_any_panic(result); |
246 | 3.46k | } |
247 | | |
248 | | /// Get the spoken text for an overview of the MathML that was set. |
249 | | /// The speech takes into account any AT or user preferences. |
250 | | /// Note: this implementation for is currently minimal and should not be used. |
251 | 0 | pub fn get_overview_text() -> Result<String> { |
252 | 0 | enable_logs(); |
253 | 0 | let result = catch_unwind(AssertUnwindSafe(|| { |
254 | 0 | MATHML_INSTANCE.with(|package_instance| { |
255 | 0 | let package_instance = package_instance.borrow(); |
256 | 0 | let mathml = get_element(&package_instance); |
257 | 0 | let speech = crate::speech::overview_mathml(mathml, "", 0)?; |
258 | 0 | return Ok(speech); |
259 | 0 | }) |
260 | 0 | })); |
261 | 0 | return report_any_panic(result); |
262 | 0 | } |
263 | | |
264 | | /// Get the value of the named preference. |
265 | | /// None is returned if `name` is not a known preference. |
266 | 100 | pub fn get_preference(name: impl AsRef<str>) -> Result<String> { |
267 | 100 | enable_logs(); |
268 | 100 | let name = name.as_ref().to_string(); |
269 | 100 | let result = catch_unwind(AssertUnwindSafe(|| { |
270 | | use crate::prefs::NO_PREFERENCE; |
271 | 100 | crate::speech::SPEECH_RULES.with(|rules| { |
272 | 100 | let rules = rules.borrow(); |
273 | 100 | let pref_manager = rules.pref_manager.borrow(); |
274 | 100 | let mut value = pref_manager.pref_to_string(&name); |
275 | 100 | if value == NO_PREFERENCE { |
276 | 1 | value = pref_manager.pref_to_string(&name); |
277 | 99 | } |
278 | 100 | if value == NO_PREFERENCE { |
279 | 1 | bail!("No preference named '{}'", name); |
280 | | } else { |
281 | 99 | return Ok(value); |
282 | | } |
283 | 100 | }) |
284 | 100 | })); |
285 | 100 | return report_any_panic(result); |
286 | 100 | } |
287 | | |
288 | | /// Set a MathCAT preference. The preference name should be a known preference name. |
289 | | /// The value should either be a string or a number (depending upon the preference being set) |
290 | | /// The list of known user preferences is in the MathCAT user documentation. |
291 | | /// Here are common preferences set by programs (not settable by the user): |
292 | | /// * TTS -- SSML, SAPI5, None |
293 | | /// * Pitch -- normalized at '1.0' |
294 | | /// * Rate -- words per minute (should match current speech rate). |
295 | | /// There is a separate "MathRate" that is user settable that causes a relative percentage change from this rate. |
296 | | /// * Volume -- default 100 |
297 | | /// * Voice -- set a voice to use (not implemented) |
298 | | /// * Gender -- set pick any voice of the given gender (not implemented) |
299 | | /// * Bookmark -- set to `true` if a `mark`/`bookmark` should be part of the returned speech (used for sync highlighting) |
300 | | /// |
301 | | /// Important: both the preference name and value are case-sensitive |
302 | | /// |
303 | | /// This function can be called multiple times to set different values. |
304 | | /// The values are persistent and extend beyond calls to [`set_mathml`]. |
305 | | /// A value can be overwritten by calling this function again with a different value. |
306 | | /// |
307 | | /// Be careful setting preferences -- these potentially override user settings, so only preferences that really need setting should be set. |
308 | 17.7k | pub fn set_preference(name: impl AsRef<str>, value: impl AsRef<str>) -> Result<()> { |
309 | 17.7k | enable_logs(); |
310 | 17.7k | let name = name.as_ref().to_string(); |
311 | 17.7k | let value = value.as_ref().to_string(); |
312 | 17.7k | let result = catch_unwind(AssertUnwindSafe(|| { |
313 | 17.7k | set_preference_impl(&name, &value) |
314 | 17.7k | })); |
315 | 17.7k | return report_any_panic(result); |
316 | 17.7k | } |
317 | | |
318 | 17.7k | fn set_preference_impl(name: &str, value: &str) -> Result<()> { |
319 | 17.7k | let mut value = value.to_string(); |
320 | 17.7k | if name == "Language" || name == "LanguageAuto"12.7k { |
321 | | // check the format |
322 | 5.03k | if value != "Auto" { |
323 | | // could get es, es-419, or en-us-nyc ... we only care about the first two parts so we clean it up a little |
324 | 5.03k | let mut lang_country_split = value.split('-'); |
325 | 5.03k | let language = lang_country_split.next().unwrap_or(""); |
326 | 5.03k | let country = lang_country_split.next().unwrap_or(""); |
327 | 5.03k | if language.len() != 2 { |
328 | 0 | bail!( |
329 | | "Improper format for 'Language' preference '{}'. Should be of form 'en' or 'en-gb'", |
330 | | value |
331 | | ); |
332 | 5.03k | } |
333 | 5.03k | let mut new_lang_country = language.to_string(); // need a temp value because 'country' is borrowed from 'value' above |
334 | 5.03k | if !country.is_empty() { |
335 | 321 | new_lang_country.push('-'); |
336 | 321 | new_lang_country.push_str(country); |
337 | 4.70k | } |
338 | 5.03k | value = new_lang_country; |
339 | 0 | } |
340 | 5.03k | if name == "LanguageAuto" && value == "Auto"0 { |
341 | 0 | bail!("'LanguageAuto' can not have the value 'Auto'"); |
342 | 5.03k | } |
343 | 12.7k | } |
344 | | |
345 | 17.7k | crate::speech::SPEECH_RULES.with(|rules| -> Result<()> { |
346 | 17.7k | if let Some(error_string0 ) = rules.borrow().get_error() { |
347 | 0 | bail!("{}", error_string); |
348 | 17.7k | } |
349 | 17.7k | Ok(()) |
350 | 17.7k | })?0 ; |
351 | | |
352 | | // Do not hold a SpeechRules borrow while updating preferences: invalidation clears rule caches. |
353 | 17.7k | let pref_manager = crate::prefs::PreferenceManager::get(); |
354 | 17.7k | let mut pref_manager = pref_manager.borrow_mut(); |
355 | 17.7k | if name == "LanguageAuto" { |
356 | 0 | let language_pref = pref_manager.pref_to_string("Language"); |
357 | 0 | if language_pref != "Auto" { |
358 | 0 | bail!( |
359 | | "'LanguageAuto' can only be used when 'Language' has the value 'Auto'; Language={}", |
360 | | language_pref |
361 | | ); |
362 | 0 | } |
363 | 17.7k | } |
364 | 17.7k | let lower_case_value = value.to_lowercase(); |
365 | 17.7k | if lower_case_value == "true" || lower_case_value == "false"17.6k { |
366 | 1.50k | pref_manager.set_api_boolean_pref(name, value.to_lowercase() == "true"); |
367 | 1.50k | } else { |
368 | 16.2k | match name { |
369 | 16.2k | "Pitch" | "Rate" | "Volume" | "CapitalLetters_Pitch" | "MathRate" | "PauseFactor" => { |
370 | 0 | pref_manager.set_api_float_pref(name, to_float(name, &value)?) |
371 | | } |
372 | | _ => { |
373 | 16.2k | pref_manager.set_string_pref(name, &value)?0 ; |
374 | | } |
375 | | } |
376 | | }; |
377 | | |
378 | 17.7k | return Ok(()); |
379 | 17.7k | } |
380 | | |
381 | 0 | fn to_float(name: &str, value: &str) -> Result<f64> { |
382 | 0 | return match value.parse::<f64>() { |
383 | 0 | Ok(val) => Ok(val), |
384 | 0 | Err(_) => bail!("SetPreference: preference'{}'s value '{}' must be a float", name, value), |
385 | | }; |
386 | 0 | } |
387 | | |
388 | | /// Get the braille associated with the MathML that was set by [`set_mathml`]. |
389 | | /// The braille returned depends upon the preference for the `code` preference (default `Nemeth`). |
390 | | /// If 'nav_node_id' is given, it is highlighted based on the value of `BrailleNavHighlight` (default: `EndPoints`) |
391 | 1.36k | pub fn get_braille(nav_node_id: impl AsRef<str>) -> Result<String> { |
392 | 1.36k | enable_logs(); |
393 | 1.36k | let nav_node_id = nav_node_id.as_ref().to_string(); |
394 | 1.36k | let result = catch_unwind(AssertUnwindSafe(|| { |
395 | 1.36k | MATHML_INSTANCE.with(|package_instance| { |
396 | 1.36k | let package_instance = package_instance.borrow(); |
397 | 1.36k | let mathml = get_element(&package_instance); |
398 | 1.36k | let braille = crate::braille::braille_mathml(mathml, &nav_node_id)?0 .0; |
399 | 1.36k | return Ok(braille); |
400 | 1.36k | }) |
401 | 1.36k | })); |
402 | 1.36k | return report_any_panic(result); |
403 | 1.36k | } |
404 | | |
405 | | /// Get the braille associated with the current navigation focus of the MathML that was set by [`set_mathml`]. |
406 | | /// The braille returned depends upon the preference for the `code` preference (default `Nemeth`). |
407 | | /// The returned braille is brailled as if the current navigation focus is the entire expression to be brailled. |
408 | 0 | pub fn get_navigation_braille() -> Result<String> { |
409 | 0 | enable_logs(); |
410 | 0 | let result = catch_unwind(AssertUnwindSafe(|| { |
411 | 0 | MATHML_INSTANCE.with(|package_instance| { |
412 | 0 | let package_instance = package_instance.borrow(); |
413 | 0 | let mathml = get_element(&package_instance); |
414 | 0 | let new_package = Package::new(); // used if we need to create a new tree |
415 | 0 | let new_doc = new_package.as_document(); |
416 | 0 | let nav_mathml = NAVIGATION_STATE.with(|nav_stack| { |
417 | 0 | return match nav_stack.borrow_mut().get_navigation_mathml(mathml) { |
418 | 0 | Err(e) => Err(e), |
419 | 0 | Ok((found, offset)) => { |
420 | | // get the MathML node and wrap it inside of a <math> element |
421 | | // if the offset is given, we need to get the character it references |
422 | 0 | if offset == 0 { |
423 | 0 | if name(found) == "math" { |
424 | 0 | Ok(found) |
425 | | } else { |
426 | 0 | let new_mathml = create_mathml_element(&new_doc, "math"); |
427 | 0 | new_mathml.append_child(copy_mathml(found)); |
428 | 0 | new_doc.root().append_child(new_mathml); |
429 | 0 | Ok(new_mathml) |
430 | | } |
431 | 0 | } else if !is_leaf(found) { |
432 | 0 | bail!( |
433 | | "Internal error: non-zero offset '{}' on a non-leaf element '{}'", |
434 | | offset, |
435 | 0 | name(found) |
436 | | ); |
437 | 0 | } else if let Some(ch) = as_text(found).chars().nth(offset) { |
438 | 0 | let internal_mathml = create_mathml_element(&new_doc, name(found)); |
439 | 0 | internal_mathml.set_text(&ch.to_string()); |
440 | 0 | let new_mathml = create_mathml_element(&new_doc, "math"); |
441 | 0 | new_mathml.append_child(internal_mathml); |
442 | 0 | new_doc.root().append_child(new_mathml); |
443 | 0 | Ok(new_mathml) |
444 | | } else { |
445 | 0 | bail!( |
446 | | "Internal error: offset '{}' on leaf element '{}' doesn't exist", |
447 | | offset, |
448 | 0 | mml_to_string(found) |
449 | | ); |
450 | | } |
451 | | } |
452 | | }; |
453 | 0 | })?; |
454 | | |
455 | 0 | let braille = crate::braille::braille_mathml(nav_mathml, "")?.0; |
456 | 0 | return Ok(braille); |
457 | 0 | }) |
458 | 0 | })); |
459 | 0 | return report_any_panic(result); |
460 | 0 | } |
461 | | |
462 | | /// Given a key code along with the modifier keys, the current node is moved accordingly (or value reported in some cases). |
463 | | /// `key` is the [keycode](https://developer.mozilla.org/en-US/docs/Web/API/KeyboardEvent/keyCode#constants_for_keycode_value) for the key (in JavaScript, `ev.key_code`) |
464 | | /// The spoken text for the new current node is returned. |
465 | 0 | pub fn do_navigate_keypress( |
466 | 0 | key: usize, |
467 | 0 | shift_key: bool, |
468 | 0 | control_key: bool, |
469 | 0 | alt_key: bool, |
470 | 0 | meta_key: bool, |
471 | 0 | ) -> Result<String> { |
472 | 0 | enable_logs(); |
473 | 0 | let result = catch_unwind(AssertUnwindSafe(|| { |
474 | 0 | MATHML_INSTANCE.with(|package_instance| { |
475 | 0 | let package_instance = package_instance.borrow(); |
476 | 0 | let mathml = get_element(&package_instance); |
477 | 0 | return do_mathml_navigate_key_press(mathml, key, shift_key, control_key, alt_key, meta_key); |
478 | 0 | }) |
479 | 0 | })); |
480 | 0 | return report_any_panic(result); |
481 | 0 | } |
482 | | |
483 | | /// Given a navigation command, the current node is moved accordingly. |
484 | | /// This is a higher level interface than `do_navigate_keypress` for applications that want to interpret the keys themselves. |
485 | | /// The valid commands are: |
486 | | /// * Standard move commands: |
487 | | /// `MovePrevious`, `MoveNext`, `MoveStart`, `MoveEnd`, `MoveLineStart`, `MoveLineEnd` |
488 | | /// * Movement in a table or elementary math: |
489 | | /// `MoveCellPrevious`, `MoveCellNext`, `MoveCellUp`, `MoveCellDown`, `MoveColumnStart`, `MoveColumnEnd` |
490 | | /// * Moving into children or out to parents: |
491 | | /// `ZoomIn`, `ZoomOut`, `ZoomOutAll`, `ZoomInAll` |
492 | | /// * Undo the last movement command: |
493 | | /// `MoveLastLocation` |
494 | | /// * Read commands (standard speech): |
495 | | /// `ReadPrevious`, `ReadNext`, `ReadCurrent`, `ReadCellCurrent`, `ReadStart`, `ReadEnd`, `ReadLineStart`, `ReadLineEnd` |
496 | | /// * Describe commands (overview): |
497 | | /// `DescribePrevious`, `DescribeNext`, `DescribeCurrent` |
498 | | /// * Location information: |
499 | | /// `WhereAmI`, `WhereAmIAll` |
500 | | /// * Change navigation modes (circle up/down): |
501 | | /// `ToggleZoomLockUp`, `ToggleZoomLockDown` |
502 | | /// * Speak the current navigation mode |
503 | | /// `ToggleSpeakMode` |
504 | | /// |
505 | | /// There are 10 place markers that can be set/read/described or moved to. |
506 | | /// * Setting: |
507 | | /// `SetPlacemarker0`, `SetPlacemarker1`, `SetPlacemarker2`, `SetPlacemarker3`, `SetPlacemarker4`, `SetPlacemarker5`, `SetPlacemarker6`, `SetPlacemarker7`, `SetPlacemarker8`, `SetPlacemarker9` |
508 | | /// * Reading: |
509 | | /// `Read0`, `Read1`, `Read2`, `Read3`, `Read4`, `Read5`, `Read6`, `Read7`, `Read8`, `Read9` |
510 | | /// * Describing: |
511 | | /// `Describe0`, `Describe1`, `Describe2`, `Describe3`, `Describe4`, `Describe5`, `Describe6`, `Describe7`, `Describe8`, `Describe9` |
512 | | /// * Moving: |
513 | | /// `MoveTo0`, `MoveTo1`, `MoveTo2`, `MoveTo3`, `MoveTo4`, `MoveTo5`, `MoveTo6`, `MoveTo7`, `MoveTo8`, `MoveTo9` |
514 | | /// |
515 | | /// When done with Navigation, call with `Exit` |
516 | 0 | pub fn do_navigate_command(command: impl AsRef<str>) -> Result<String> { |
517 | 0 | enable_logs(); |
518 | 0 | let command = command.as_ref().to_string(); |
519 | 0 | let result = catch_unwind(AssertUnwindSafe(|| { |
520 | 0 | let cmd = NAV_COMMANDS.get_key(&command); // gets a &'static version of the command |
521 | 0 | if cmd.is_none() { |
522 | 0 | bail!("Unknown command in call to DoNavigateCommand()"); |
523 | 0 | }; |
524 | 0 | let cmd = *cmd.unwrap(); |
525 | 0 | MATHML_INSTANCE.with(|package_instance| { |
526 | 0 | let package_instance = package_instance.borrow(); |
527 | 0 | let mathml = get_element(&package_instance); |
528 | 0 | return do_navigate_command_string(mathml, cmd); |
529 | 0 | }) |
530 | 0 | })); |
531 | 0 | return report_any_panic(result); |
532 | 0 | } |
533 | | |
534 | | /// Given an 'id' and an offset (for tokens), set the navigation node to that id. |
535 | | /// An error is returned if the 'id' doesn't exist |
536 | 2 | pub fn set_navigation_node(id: impl AsRef<str>, offset: usize) -> Result<()> { |
537 | 2 | enable_logs(); |
538 | 2 | let id = id.as_ref().to_string(); |
539 | 2 | let result = catch_unwind(AssertUnwindSafe(|| { |
540 | 2 | MATHML_INSTANCE.with(|package_instance| { |
541 | 2 | let package_instance = package_instance.borrow(); |
542 | 2 | let mathml = get_element(&package_instance); |
543 | 2 | return set_navigation_node_from_id(mathml, &id, offset); |
544 | 2 | }) |
545 | 2 | })); |
546 | 2 | return report_any_panic(result); |
547 | 2 | } |
548 | | |
549 | | /// Return the MathML associated with the current (navigation) node and the offset (0-based) from that mathml (not yet implemented) |
550 | | /// The offset is needed for token elements that have multiple characters. |
551 | 0 | pub fn get_navigation_mathml() -> Result<(String, usize)> { |
552 | 0 | enable_logs(); |
553 | 0 | let result = catch_unwind(AssertUnwindSafe(|| { |
554 | 0 | MATHML_INSTANCE.with(|package_instance| { |
555 | 0 | let package_instance = package_instance.borrow(); |
556 | 0 | let mathml = get_element(&package_instance); |
557 | 0 | return NAVIGATION_STATE.with(|nav_stack| { |
558 | 0 | return match nav_stack.borrow_mut().get_navigation_mathml(mathml) { |
559 | 0 | Err(e) => Err(e), |
560 | 0 | Ok((found, offset)) => Ok((mml_to_string(found), offset)), |
561 | | }; |
562 | 0 | }); |
563 | 0 | }) |
564 | 0 | })); |
565 | 0 | return report_any_panic(result); |
566 | 0 | } |
567 | | |
568 | | /// Return the `id` and `offset` (0-based) associated with the current (navigation) node. |
569 | | /// `offset` (not yet implemented) |
570 | | /// The offset is needed for token elements that have multiple characters. |
571 | 2 | pub fn get_navigation_mathml_id() -> Result<(String, usize)> { |
572 | 2 | enable_logs(); |
573 | 2 | let result = catch_unwind(AssertUnwindSafe(|| { |
574 | 2 | MATHML_INSTANCE.with(|package_instance| { |
575 | 2 | let package_instance = package_instance.borrow(); |
576 | 2 | let mathml = get_element(&package_instance); |
577 | 2 | return Ok(NAVIGATION_STATE.with(|nav_stack| { |
578 | 2 | return nav_stack.borrow().get_navigation_mathml_id(mathml); |
579 | 2 | })); |
580 | 2 | }) |
581 | 2 | })); |
582 | 2 | return report_any_panic(result); |
583 | 2 | } |
584 | | |
585 | | /// Return the start and end braille character positions associated with the current (navigation) node. |
586 | 2 | pub fn get_braille_position() -> Result<(usize, usize)> { |
587 | 2 | enable_logs(); |
588 | 2 | let result = catch_unwind(AssertUnwindSafe(|| { |
589 | 2 | MATHML_INSTANCE.with(|package_instance| { |
590 | 2 | let package_instance = package_instance.borrow(); |
591 | 2 | let mathml = get_element(&package_instance); |
592 | 2 | let nav_node = get_navigation_mathml_id()?0 ; |
593 | 2 | let (_, start, end) = crate::braille::braille_mathml(mathml, &nav_node.0)?0 ; |
594 | 2 | return Ok((start, end)); |
595 | 2 | }) |
596 | 2 | })); |
597 | 2 | return report_any_panic(result); |
598 | 2 | } |
599 | | |
600 | | /// Given a 0-based braille position, return the smallest MathML node enclosing it. |
601 | | /// This node might be a leaf with an offset. |
602 | 91 | pub fn get_navigation_node_from_braille_position(position: usize) -> Result<(String, usize)> { |
603 | 91 | enable_logs(); |
604 | 91 | let result = catch_unwind(AssertUnwindSafe(|| { |
605 | 91 | MATHML_INSTANCE.with(|package_instance| { |
606 | 91 | let package_instance = package_instance.borrow(); |
607 | 91 | let mathml = get_element(&package_instance); |
608 | 91 | return crate::braille::get_navigation_node_from_braille_position(mathml, position); |
609 | 91 | }) |
610 | 91 | })); |
611 | 91 | return report_any_panic(result); |
612 | 91 | } |
613 | | |
614 | 0 | pub fn get_supported_braille_codes() -> Result<Vec<String>> { |
615 | 0 | enable_logs(); |
616 | 0 | let result = catch_unwind(AssertUnwindSafe(|| { |
617 | 0 | let rules_dir = crate::prefs::PreferenceManager::get().borrow().get_rules_dir(); |
618 | 0 | let braille_dir = rules_dir.join("Braille"); |
619 | 0 | let mut braille_code_paths = Vec::new(); |
620 | | |
621 | 0 | find_all_dirs_shim(&braille_dir, &mut braille_code_paths); |
622 | 0 | let mut braille_code_paths = braille_code_paths.iter() |
623 | 0 | .map(|path| path.strip_prefix(&braille_dir).unwrap().to_string_lossy().to_string()) |
624 | 0 | .filter(|string_path| !string_path.is_empty() ) |
625 | 0 | .collect::<Vec<String>>(); |
626 | 0 | braille_code_paths.sort(); |
627 | | |
628 | 0 | Ok(braille_code_paths) |
629 | 0 | })); |
630 | 0 | return report_any_panic(result); |
631 | 0 | } |
632 | | |
633 | | /// Returns a Vec of all supported languages ("en", "es", ...) |
634 | 1 | pub fn get_supported_languages() -> Result<Vec<String>> { |
635 | 1 | enable_logs(); |
636 | 1 | let result = catch_unwind(AssertUnwindSafe(|| { |
637 | 1 | let rules_dir = crate::prefs::PreferenceManager::get().borrow().get_rules_dir(); |
638 | 1 | let lang_dir = rules_dir.join("Languages"); |
639 | 1 | let mut lang_paths = Vec::new(); |
640 | | |
641 | 1 | find_all_dirs_shim(&lang_dir, &mut lang_paths); |
642 | 1 | let mut language_paths = lang_paths.iter() |
643 | 13 | .map1 (|path| path.strip_prefix(&lang_dir).unwrap() |
644 | 13 | .to_string_lossy() |
645 | 13 | .replace(std::path::MAIN_SEPARATOR, "-") |
646 | 13 | .to_string()) |
647 | 13 | .filter1 (|string_path| !string_path.is_empty() ) |
648 | 1 | .collect::<Vec<String>>(); |
649 | | |
650 | | // make sure the 'zz' test dir isn't included (build.rs removes it, but for debugging is there) |
651 | 13 | language_paths1 .retain1 (|s| !s.starts_with("zz")); |
652 | 1 | language_paths.sort(); |
653 | 1 | Ok(language_paths) |
654 | 1 | })); |
655 | 1 | return report_any_panic(result); |
656 | 1 | } |
657 | | |
658 | 0 | pub fn get_supported_speech_styles(lang: impl AsRef<str>) -> Result<Vec<String>> { |
659 | 0 | enable_logs(); |
660 | 0 | let lang = lang.as_ref().to_string(); |
661 | 0 | let result = catch_unwind(AssertUnwindSafe(|| { |
662 | 0 | let rules_dir = crate::prefs::PreferenceManager::get().borrow().get_rules_dir(); |
663 | 0 | let lang_dir = rules_dir.join("Languages").join(&lang); |
664 | 0 | let mut speech_styles = find_files_in_dir_that_ends_with_shim(&lang_dir, "_Rules.yaml"); |
665 | 0 | for file_name in &mut speech_styles { |
666 | 0 | file_name.truncate(file_name.len() - "_Rules.yaml".len()) |
667 | | } |
668 | 0 | speech_styles.sort(); |
669 | 0 | speech_styles.dedup(); // remove duplicates -- shouldn't be any, but just in case |
670 | 0 | Ok(speech_styles) |
671 | 0 | })); |
672 | 0 | return report_any_panic(result); |
673 | 0 | } |
674 | | |
675 | | // utility functions |
676 | | |
677 | | /// Copy (recursively) the (MathML) element and return the new one. |
678 | | /// The Element type does not copy and modifying the structure of an element's child will modify the element, so we need a copy |
679 | | /// Convert the returned error from set_mathml, etc., to a useful string for display |
680 | 363 | pub fn copy_mathml(mathml: Element) -> Element { |
681 | 363 | return copy_mathml_recursive(mathml, 0); |
682 | 363 | } |
683 | | |
684 | 4.53k | fn copy_mathml_recursive(mathml: Element, depth: usize) -> Element { |
685 | | // Safety: Prevent stack overflow on deeply nested MathML |
686 | 4.53k | if depth > MAX_DEPTH { |
687 | | // Return the element as a leaf if it's too deep to prevent crash |
688 | 0 | return create_mathml_element(&mathml.document(), name(mathml)); |
689 | 4.53k | } |
690 | | |
691 | | // If it represents MathML, the 'Element' can only have Text and Element children along with attributes |
692 | 4.53k | let children = mathml.children(); |
693 | 4.53k | let new_mathml = create_mathml_element(&mathml.document(), name(mathml)); |
694 | 9.52k | mathml.attributes().iter()4.53k .for_each4.53k (|attr| { |
695 | 9.52k | new_mathml.set_attribute_value(attr.name(), attr.value()); |
696 | 9.52k | }); |
697 | | |
698 | | // can't use is_leaf/as_text because this is also used with the intent tree |
699 | 4.53k | if children.len() == 1 && |
700 | 3.26k | let Some(text2.59k ) = children[0].text() { |
701 | 2.59k | new_mathml.set_text(text.text()); |
702 | 2.59k | return new_mathml; |
703 | 1.93k | } |
704 | | |
705 | 1.93k | let mut new_children = Vec::with_capacity(children.len()); |
706 | 4.17k | for child in children1.93k { |
707 | 4.17k | let child = as_element(child); |
708 | 4.17k | let new_child = copy_mathml_recursive(child, depth + 1); |
709 | 4.17k | new_children.push(new_child); |
710 | 4.17k | } |
711 | 1.93k | new_mathml.append_children(new_children); |
712 | 1.93k | return new_mathml; |
713 | 4.53k | } |
714 | | |
715 | 0 | pub fn errors_to_string(e: &Error) -> String { |
716 | 0 | enable_logs(); |
717 | 0 | let mut result = format!("{e}\n"); |
718 | 0 | for cause in e.chain().skip(1) { // skips original error |
719 | 0 | result += &format!("caused by: {cause}\n"); |
720 | 0 | } |
721 | 0 | result |
722 | 0 | } |
723 | | |
724 | 4.91k | fn add_ids(mathml: Element) -> Element { |
725 | | use std::time::SystemTime; |
726 | 4.91k | let time = if cfg!(target_family = "wasm") { |
727 | 0 | fastrand::usize(..) |
728 | | } else { |
729 | 4.91k | SystemTime::now() |
730 | 4.91k | .duration_since(SystemTime::UNIX_EPOCH) |
731 | 4.91k | .unwrap() |
732 | 4.91k | .as_millis() as usize |
733 | | }; |
734 | 4.91k | let mut time_part = radix_fmt::radix(time, 36).to_string(); |
735 | 4.91k | if time_part.len() < 3 { |
736 | 0 | time_part.push_str("a2c"); // needs to be at least three chars |
737 | 4.91k | } |
738 | 4.91k | let mut random_part = radix_fmt::radix(fastrand::u32(..), 36).to_string(); |
739 | 4.91k | if random_part.len() < 4 { |
740 | 0 | random_part.push_str("a1b2"); // needs to be at least four chars |
741 | 4.91k | } |
742 | 4.91k | let prefix = "M".to_string() + &time_part[time_part.len() - 3..] + &random_part[random_part.len() - 4..] + "-"; // begin with letter |
743 | 4.91k | add_ids_to_all(mathml, &prefix, 0); |
744 | 4.91k | return mathml; |
745 | | |
746 | 57.8k | fn add_ids_to_all(mathml: Element, id_prefix: &str, count: usize) -> usize { |
747 | 57.8k | let mut count = count; |
748 | 57.8k | if mathml.attribute("id").is_none() { |
749 | 57.1k | mathml.set_attribute_value("id", (id_prefix.to_string() + &count.to_string()).as_str()); |
750 | 57.1k | mathml.set_attribute_value("data-id-added", "true"); |
751 | 57.1k | count += 1; |
752 | 57.1k | }707 ; |
753 | | |
754 | 57.8k | if crate::xpath_functions::is_leaf(mathml) { |
755 | 35.8k | return count; |
756 | 22.0k | } |
757 | | |
758 | 52.9k | for child in mathml22.0k .children22.0k () { |
759 | 52.9k | let child = as_element(child); |
760 | 52.9k | count = add_ids_to_all(child, id_prefix, count); |
761 | 52.9k | } |
762 | 22.0k | return count; |
763 | 57.8k | } |
764 | 4.91k | } |
765 | | |
766 | 10.3k | pub fn get_element(package: &Package) -> Element<'_> { |
767 | 10.3k | enable_logs(); |
768 | 10.3k | let doc = package.as_document(); |
769 | 10.3k | let mut result = None; |
770 | 10.3k | for root_child in doc.root().children() { |
771 | 10.3k | if let ChildOfRoot::Element(e) = root_child { |
772 | 10.3k | assert!(result.is_none()); |
773 | 10.3k | result = Some(e); |
774 | 0 | } |
775 | | } |
776 | 10.3k | return result.unwrap(); |
777 | 10.3k | } |
778 | | |
779 | | /// Get the intent after setting the MathML |
780 | | /// Used in testing |
781 | | #[allow(dead_code)] |
782 | 32 | pub fn get_intent<'a>(mathml: Element<'a>, doc: Document<'a>) -> Result<Element<'a>> { |
783 | 32 | crate::speech::SPEECH_RULES.with(|rules| rules.borrow_mut().read_files().unwrap()); |
784 | 32 | let mathml = cleanup_mathml(mathml)?0 ; |
785 | 32 | return crate::speech::intent_from_mathml(mathml, doc); |
786 | 32 | } |
787 | | |
788 | | #[allow(dead_code)] |
789 | 22 | fn trim_doc(doc: &Document) { |
790 | 22 | for root_child in doc.root().children() { |
791 | 22 | if let ChildOfRoot::Element(e) = root_child { |
792 | 22 | trim_element(e, false); |
793 | 22 | } else { |
794 | 0 | doc.root().remove_child(root_child); // comment or processing instruction |
795 | 0 | } |
796 | | } |
797 | 22 | } |
798 | | |
799 | | /// Not really meant to be public -- used by tests in some packages |
800 | 55.5k | pub fn trim_element(e: Element, allow_structure_in_leaves: bool) { |
801 | | // "<mtext>this is text</mtext" results in 3 text children |
802 | | // these are combined into one child as it makes code downstream simpler |
803 | | |
804 | | // space, tab, newline, carriage return all get collapsed to a single space |
805 | | const WHITESPACE: &[char] = &[' ', '\u{0009}', '\u{000A}','\u{000C}', '\u{000D}']; |
806 | 3 | static WHITESPACE_MATCH: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"[ \u{0009}\u{000A}\u{00C}\u{000D}]+"#).unwrap()); |
807 | | |
808 | 55.5k | if is_leaf(e) && (!allow_structure_in_leaves34.7k || IsNode::is_mathml230 (e230 )) { |
809 | | // Assume it is HTML inside of the leaf -- turn the HTML into a string |
810 | 34.7k | make_leaf_element(e); |
811 | 34.7k | return; |
812 | 20.7k | } |
813 | | |
814 | 20.7k | let mut single_text = "".to_string(); |
815 | 87.6k | for child in e20.7k .children20.7k () { |
816 | 87.6k | match child { |
817 | 50.1k | ChildOfElement::Element(c) => { |
818 | 50.1k | trim_element(c, allow_structure_in_leaves); |
819 | 50.1k | } |
820 | 37.4k | ChildOfElement::Text(t) => { |
821 | 37.4k | single_text += t.text(); |
822 | 37.4k | e.remove_child(child); |
823 | 37.4k | } |
824 | 21 | _ => { |
825 | 21 | e.remove_child(child); |
826 | 21 | } |
827 | | } |
828 | | } |
829 | | |
830 | | // CSS considers only space, tab, linefeed, and carriage return as collapsable whitespace |
831 | 20.7k | if !(is_leaf(e) || name(e) == "intent-literal" || single_text.is_empty()) { |
832 | | // intent-literal comes from testing intent |
833 | | // FIX: we have a problem -- what should happen??? |
834 | | // FIX: For now, just keep the children and ignore the text and log an error -- shouldn't panic/crash |
835 | 14.3k | if !single_text.trim_matches(WHITESPACE).is_empty() { |
836 | 20 | error!( |
837 | | "trim_element: both element and textual children which shouldn't happen -- ignoring text '{single_text}'" |
838 | | ); |
839 | 14.2k | } |
840 | 14.3k | return; |
841 | 6.44k | } |
842 | 6.44k | if e.children().is_empty() && !single_text.is_empty()276 { |
843 | 0 | // debug!("Combining text in {}: '{}' -> '{}'", e.name().local_part(), single_text, trimmed_text); |
844 | 0 | e.set_text(&WHITESPACE_MATCH.replace_all(&single_text, " ")); |
845 | 6.44k | } |
846 | | |
847 | 34.7k | fn make_leaf_element(mathml_leaf: Element) { |
848 | | // MathML leaves like <mn> really shouldn't have non-textual content, but you could have embedded HTML |
849 | | // Here, we convert them to leaves by grabbing up all the text and making that the content |
850 | | // Potentially, we leave them and let (default) rules do something, but it makes other parts of the code |
851 | | // messier because checking the text of a leaf becomes Option<&str> rather than just &str |
852 | 34.7k | let children = mathml_leaf.children(); |
853 | 34.7k | if children.is_empty() { |
854 | 503 | return; |
855 | 34.2k | } |
856 | | |
857 | 34.2k | if rewrite_and_flatten_embedded_mathml(mathml_leaf) { |
858 | 1 | return; |
859 | 34.2k | } |
860 | | |
861 | | // gather up the text |
862 | 34.2k | let mut text = "".to_string(); |
863 | 34.6k | for child in children34.2k { |
864 | 34.6k | let child_text = match child { |
865 | 8 | ChildOfElement::Element(child) => { |
866 | 8 | if name(child) == "mglyph" { |
867 | 3 | child.attribute_value("alt").unwrap_or("").to_string() |
868 | | } else { |
869 | 5 | gather_text(child) |
870 | | } |
871 | | } |
872 | 34.4k | ChildOfElement::Text(t) => { |
873 | | // debug!("ChildOfElement::Text: '{}'", t.text()); |
874 | 34.4k | t.text().to_string() |
875 | | } |
876 | 222 | _ => "".to_string(), |
877 | | }; |
878 | 34.6k | if !child_text.is_empty() { |
879 | 34.4k | text += &child_text; |
880 | 34.4k | }223 |
881 | | } |
882 | | |
883 | | // get rid of the old children and replace with the text we just built |
884 | 34.2k | mathml_leaf.clear_children(); |
885 | 34.2k | mathml_leaf.set_text(WHITESPACE_MATCH.replace_all(&text, " ").trim_matches(WHITESPACE)); |
886 | | // debug!("make_leaf_element: text is '{}'", crate::canonicalize::as_text(mathml_leaf)); |
887 | | |
888 | | /// gather up all the contents of the element and return them with a leading space |
889 | 7 | fn gather_text(html: Element) -> String { |
890 | 7 | let mut text = "".to_string(); // since we are throwing out the element tag, add a space between the contents |
891 | 7 | for child in html.children() { |
892 | 7 | match child { |
893 | 2 | ChildOfElement::Element(child) => { |
894 | 2 | text += &gather_text(child); |
895 | 2 | } |
896 | 5 | ChildOfElement::Text(t) => text += t.text(), |
897 | 0 | _ => (), |
898 | | } |
899 | | } |
900 | | // debug!("gather_text: '{}'", text); |
901 | 7 | return text; |
902 | 7 | } |
903 | 34.7k | } |
904 | | |
905 | 34.2k | fn rewrite_and_flatten_embedded_mathml(mathml_leaf: Element) -> bool { |
906 | | // first see if it can or needs to be rewritten |
907 | | // this is likely rare, so we do a check and if true, to a second pass building the result |
908 | 34.2k | let mut needs_rewrite = false; |
909 | 34.6k | for child in mathml_leaf34.2k .children34.2k () { |
910 | 34.6k | if let Some(element8 ) = child.element() { |
911 | 8 | if name(element) != "math" { |
912 | 7 | return false; // something other than MathML as a child -- can't rewrite |
913 | 1 | } |
914 | 1 | needs_rewrite = true; |
915 | 34.6k | } |
916 | | }; |
917 | | |
918 | 34.2k | if !needs_rewrite { |
919 | 34.2k | return false; |
920 | 1 | } |
921 | | |
922 | | // now do the rewrite, flatting out the mathml and returning an mrow with the children |
923 | 1 | let leaf_name = name(mathml_leaf); |
924 | 1 | let doc = mathml_leaf.document(); |
925 | 1 | let mut new_children = Vec::new(); |
926 | 1 | let mut is_last_mtext = false; |
927 | 5 | for child in mathml_leaf1 .children1 () { |
928 | 5 | if let Some(element1 ) = child.element() { |
929 | 1 | trim_element(element, true); |
930 | 1 | new_children.append(&mut element.children()); // don't want 'math' wrapper |
931 | 1 | is_last_mtext = false; |
932 | 4 | } else if let Some(text) = child.text() { |
933 | | // combine adjacent text nodes into single nodes |
934 | 4 | if is_last_mtext { |
935 | 2 | let last_child = new_children.last_mut().unwrap().element().unwrap(); |
936 | 2 | let new_text = as_text(last_child).to_string() + text.text(); |
937 | 2 | last_child.set_text(&new_text); |
938 | 2 | } else { |
939 | 2 | let new_leaf_node = create_mathml_element(&doc, leaf_name); |
940 | 2 | new_leaf_node.set_text(text.text()); |
941 | 2 | new_children.push(ChildOfElement::Element(new_leaf_node)); |
942 | 2 | is_last_mtext = true; |
943 | 2 | } |
944 | 0 | } |
945 | | }; |
946 | | |
947 | | // clean up whitespace in text nodes |
948 | 3 | for child in &mut new_children1 { |
949 | 3 | if let Some(element) = child.element() && is_leaf(element) { |
950 | 2 | let text = as_text(element); |
951 | 2 | let cleaned_text = WHITESPACE_MATCH.replace_all(text, " ").trim_matches(WHITESPACE).to_string(); |
952 | 2 | element.set_text(&cleaned_text); |
953 | 2 | }1 |
954 | | } |
955 | | |
956 | 1 | crate::canonicalize::set_mathml_name(mathml_leaf, "mrow"); |
957 | 1 | mathml_leaf.clear_children(); |
958 | 1 | mathml_leaf.append_children(new_children); |
959 | | |
960 | | // debug!("rewrite_and_flatten_embedded_mathml: flattened\n'{}'", mml_to_string(mathml_leaf)); |
961 | 1 | return true; |
962 | 34.2k | } |
963 | 55.5k | } |
964 | | |
965 | | // used for testing trim |
966 | | /// returns Ok() if two Documents are equal or some info where they differ in the Err |
967 | | #[allow(dead_code)] |
968 | 11 | fn is_same_doc(doc1: &Document, doc2: &Document) -> Result<()> { |
969 | | // assume 'e' doesn't have element children until proven otherwise |
970 | | // this means we keep Text children until we are proven they aren't needed |
971 | 11 | if doc1.root().children().len() != doc2.root().children().len() { |
972 | 0 | bail!( |
973 | | "Children of docs have {} != {} children", |
974 | 0 | doc1.root().children().len(), |
975 | 0 | doc2.root().children().len() |
976 | | ); |
977 | 11 | } |
978 | | |
979 | 11 | for (i, (c1, c2)) in doc1 |
980 | 11 | .root() |
981 | 11 | .children() |
982 | 11 | .iter() |
983 | 11 | .zip(doc2.root().children().iter()) |
984 | 11 | .enumerate() |
985 | | { |
986 | 11 | match c1 { |
987 | 11 | ChildOfRoot::Element(e1) => { |
988 | 11 | if let ChildOfRoot::Element(e2) = c2 { |
989 | 11 | is_same_element(*e1, *e2, &[])?1 ; |
990 | | } else { |
991 | 0 | bail!("child #{}, first is element, second is something else", i); |
992 | | } |
993 | | } |
994 | 0 | ChildOfRoot::Comment(com1) => { |
995 | 0 | if let ChildOfRoot::Comment(com2) = c2 { |
996 | 0 | if com1.text() != com2.text() { |
997 | 0 | bail!("child #{} -- comment text differs", i); |
998 | 0 | } |
999 | | } else { |
1000 | 0 | bail!("child #{}, first is comment, second is something else", i); |
1001 | | } |
1002 | | } |
1003 | 0 | ChildOfRoot::ProcessingInstruction(p1) => { |
1004 | 0 | if let ChildOfRoot::ProcessingInstruction(p2) = c2 { |
1005 | 0 | if p1.target() != p2.target() || p1.value() != p2.value() { |
1006 | 0 | bail!("child #{} -- processing instruction differs", i); |
1007 | 0 | } |
1008 | | } else { |
1009 | 0 | bail!( |
1010 | | "child #{}, first is processing instruction, second is something else", |
1011 | | i |
1012 | | ); |
1013 | | } |
1014 | | } |
1015 | | } |
1016 | | } |
1017 | 10 | return Ok(()); |
1018 | 11 | } |
1019 | | |
1020 | | /// returns Ok() if two Documents are equal or some info where they differ in the Err |
1021 | | // Not really meant to be public -- used by tests in some packages |
1022 | | #[allow(dead_code)] |
1023 | 1.92k | pub fn is_same_element(e1: Element, e2: Element, ignore_attrs: &[&str]) -> Result<()> { |
1024 | 1.92k | enable_logs(); |
1025 | 1.92k | if name(e1) != name(e2) { |
1026 | 0 | bail!("Names not the same: {}, {}", name(e1), name(e2)); |
1027 | 1.92k | } |
1028 | | |
1029 | | // assume 'e' doesn't have element children until proven otherwise |
1030 | | // this means we keep Text children until we are proven they aren't needed |
1031 | 1.92k | if e1.children().len() != e2.children().len() { |
1032 | 0 | bail!( |
1033 | | "Children of {} have {} != {} children", |
1034 | 0 | name(e1), |
1035 | 0 | e1.children().len(), |
1036 | 0 | e2.children().len() |
1037 | | ); |
1038 | 1.92k | } |
1039 | | |
1040 | 1.92k | if let Err(e0 ) = attrs_are_same(e1.attributes(), e2.attributes(), ignore_attrs) { |
1041 | 0 | bail!("In element {}, {}", name(e1), e); |
1042 | 1.92k | } |
1043 | | |
1044 | 2.86k | for (i, (c1, c2)) in e1.children().iter()1.92k .zip1.92k (e2.children().iter()1.92k ).enumerate1.92k () { |
1045 | 2.86k | match c1 { |
1046 | 1.72k | ChildOfElement::Element(child1) => { |
1047 | 1.72k | if let ChildOfElement::Element(child2) = c2 { |
1048 | 1.72k | is_same_element(*child1, *child2, ignore_attrs)?2 ; |
1049 | | } else { |
1050 | 0 | bail!("{} child #{}, first is element, second is something else", name(e1), i); |
1051 | | } |
1052 | | } |
1053 | 0 | ChildOfElement::Comment(com1) => { |
1054 | 0 | if let ChildOfElement::Comment(com2) = c2 { |
1055 | 0 | if com1.text() != com2.text() { |
1056 | 0 | bail!("{} child #{} -- comment text differs", name(e1), i); |
1057 | 0 | } |
1058 | | } else { |
1059 | 0 | bail!("{} child #{}, first is comment, second is something else", name(e1), i); |
1060 | | } |
1061 | | } |
1062 | 0 | ChildOfElement::ProcessingInstruction(p1) => { |
1063 | 0 | if let ChildOfElement::ProcessingInstruction(p2) = c2 { |
1064 | 0 | if p1.target() != p2.target() || p1.value() != p2.value() { |
1065 | 0 | bail!("{} child #{} -- processing instruction differs", name(e1), i); |
1066 | 0 | } |
1067 | | } else { |
1068 | 0 | bail!( |
1069 | | "{} child #{}, first is processing instruction, second is something else", |
1070 | 0 | name(e1), |
1071 | | i |
1072 | | ); |
1073 | | } |
1074 | | } |
1075 | 1.14k | ChildOfElement::Text(t1) => { |
1076 | 1.14k | if let ChildOfElement::Text(t2) = c2 { |
1077 | 1.14k | if t1.text() != t2.text() { |
1078 | 1 | bail!("{} child #{} -- text differs", name(e1), i); |
1079 | 1.14k | } |
1080 | | } else { |
1081 | 0 | bail!("{} child #{}, first is text, second is something else", name(e1), i); |
1082 | | } |
1083 | | } |
1084 | | } |
1085 | | } |
1086 | 1.91k | return Ok(()); |
1087 | | |
1088 | | /// compares attributes -- '==' didn't seems to work |
1089 | 1.92k | fn attrs_are_same(attrs1: Vec<Attribute>, attrs2: Vec<Attribute>, ignore: &[&str]) -> Result<()> { |
1090 | 1.92k | let attrs1 = attrs1.iter() |
1091 | 1.92k | .filter(|a| !ignore1.40k .contains1.40k (&a.name().local_part()1.40k )).cloned() |
1092 | 1.92k | .collect::<Vec<Attribute>>(); |
1093 | 1.92k | let attrs2 = attrs2.iter() |
1094 | 1.92k | .filter(|a| !ignore1.40k .contains1.40k (&a.name().local_part()1.40k )).cloned() |
1095 | 1.92k | .collect::<Vec<Attribute>>(); |
1096 | 1.92k | if attrs1.len() != attrs2.len() { |
1097 | 0 | bail!("Attributes have different length: {:?} != {:?}", attrs1, attrs2); |
1098 | 1.92k | } |
1099 | | // can't guarantee attrs are in the same order |
1100 | 1.92k | for attr11.40k in attrs1 { |
1101 | 1.40k | if let Some(found_attr2) = attrs2 |
1102 | 1.40k | .iter() |
1103 | 1.88k | .find1.40k (|&attr2| attr1.name().local_part() == attr2.name().local_part()) |
1104 | | { |
1105 | 1.40k | if attr1.value() == found_attr2.value() { |
1106 | 1.40k | continue; |
1107 | | } else { |
1108 | 0 | bail!( |
1109 | | "Attribute named {} has differing values:\n '{}'\n '{}'", |
1110 | 0 | attr1.name().local_part(), |
1111 | 0 | attr1.value(), |
1112 | 0 | found_attr2.value() |
1113 | | ); |
1114 | | } |
1115 | | } else { |
1116 | 0 | bail!( |
1117 | | "Attribute name {} not in [{}]", |
1118 | 0 | print_attr(&attr1), |
1119 | 0 | print_attrs(&attrs2) |
1120 | | ); |
1121 | | } |
1122 | | } |
1123 | 1.92k | return Ok(()); |
1124 | | |
1125 | 0 | fn print_attr(attr: &Attribute) -> String { |
1126 | 0 | return format!("@{}='{}'", attr.name().local_part(), attr.value()); |
1127 | 0 | } |
1128 | 0 | fn print_attrs(attrs: &[Attribute]) -> String { |
1129 | 0 | return attrs.iter().map(print_attr).collect::<Vec<String>>().join(", "); |
1130 | 0 | } |
1131 | 1.92k | } |
1132 | 1.92k | } |
1133 | | |
1134 | | #[cfg(test)] |
1135 | | mod tests { |
1136 | | #[allow(unused_imports)] |
1137 | | use super::super::init_logger; |
1138 | | use super::*; |
1139 | | |
1140 | 10 | fn are_parsed_strs_equal(test: &str, target: &str) -> bool { |
1141 | 10 | let test_package = &parser::parse(test).expect("Failed to parse input"); |
1142 | 10 | let test_doc = test_package.as_document(); |
1143 | 10 | trim_doc(&test_doc); |
1144 | 10 | debug!("test:\n{}", mml_to_string0 (get_element0 (test_package0 ))); |
1145 | | |
1146 | 10 | let target_package = &parser::parse(target).expect("Failed to parse input"); |
1147 | 10 | let target_doc = target_package.as_document(); |
1148 | 10 | trim_doc(&target_doc); |
1149 | 10 | debug!("target:\n{}", mml_to_string0 (get_element0 (target_package0 ))); |
1150 | | |
1151 | 10 | match is_same_doc(&test_doc, &target_doc) { |
1152 | 10 | Ok(_) => return true, |
1153 | 0 | Err(e) => panic!("{}", e), |
1154 | | } |
1155 | 10 | } |
1156 | | |
1157 | | #[test] |
1158 | 1 | fn trim_same() { |
1159 | 1 | let trimmed_str = "<math><mrow><mo>-</mo><mi>a</mi></mrow></math>"; |
1160 | 1 | assert!(are_parsed_strs_equal(trimmed_str, trimmed_str)); |
1161 | 1 | } |
1162 | | |
1163 | | #[test] |
1164 | 1 | fn trim_whitespace() { |
1165 | 1 | let trimmed_str = "<math><mrow><mo>-</mo><mi> a </mi></mrow></math>"; |
1166 | 1 | let whitespace_str = "<math> <mrow ><mo>-</mo><mi> a </mi></mrow ></math>"; |
1167 | 1 | assert!(are_parsed_strs_equal(trimmed_str, whitespace_str)); |
1168 | 1 | } |
1169 | | |
1170 | | #[test] |
1171 | 1 | fn no_trim_whitespace_nbsp() { |
1172 | 1 | let trimmed_str = "<math><mrow><mo>-</mo><mtext>  a </mtext></mrow></math>"; |
1173 | 1 | let whitespace_str = "<math> <mrow ><mo>-</mo><mtext>  a </mtext></mrow ></math>"; |
1174 | 1 | assert!(are_parsed_strs_equal(trimmed_str, whitespace_str)); |
1175 | 1 | } |
1176 | | |
1177 | | #[test] |
1178 | 1 | fn trim_comment() { |
1179 | 1 | let whitespace_str = "<math> <mrow ><mo>-</mo><mi> a </mi></mrow ></math>"; |
1180 | 1 | let comment_str = "<math><mrow><mo>-</mo><!--a comment --><mi> a </mi></mrow></math>"; |
1181 | 1 | assert!(are_parsed_strs_equal(comment_str, whitespace_str)); |
1182 | 1 | } |
1183 | | |
1184 | | #[test] |
1185 | 1 | fn replace_mglyph() { |
1186 | 1 | let mglyph_str = "<math> |
1187 | 1 | <mrow> |
1188 | 1 | <mi>X<mglyph fontfamily='my-braid-font' index='2' alt='23braid' /></mi> |
1189 | 1 | <mo>+</mo> |
1190 | 1 | <mi> |
1191 | 1 | <mglyph fontfamily='my-braid-font' index='5' alt='132braid' />Y |
1192 | 1 | </mi> |
1193 | 1 | <mo>=</mo> |
1194 | 1 | <mi> |
1195 | 1 | <mglyph fontfamily='my-braid-font' index='3' alt='13braid' /> |
1196 | 1 | </mi> |
1197 | 1 | </mrow> |
1198 | 1 | </math>"; |
1199 | 1 | let result_str = "<math> |
1200 | 1 | <mrow> |
1201 | 1 | <mi>X23braid</mi> |
1202 | 1 | <mo>+</mo> |
1203 | 1 | <mi>132braidY</mi> |
1204 | 1 | <mo>=</mo> |
1205 | 1 | <mi>13braid</mi> |
1206 | 1 | </mrow> |
1207 | 1 | </math>"; |
1208 | 1 | assert!(are_parsed_strs_equal(mglyph_str, result_str)); |
1209 | 1 | } |
1210 | | |
1211 | | #[test] |
1212 | 1 | fn trim_differs() { |
1213 | 1 | let whitespace_str = "<math> <mrow ><mo>-</mo><mi> a </mi></mrow ></math>"; |
1214 | 1 | let different_str = "<math> <mrow ><mo>-</mo><mi> b </mi></mrow ></math>"; |
1215 | | |
1216 | | // need to manually do this since failure shouldn't be a panic |
1217 | 1 | let package1 = &parser::parse(whitespace_str).expect("Failed to parse input"); |
1218 | 1 | let doc1 = package1.as_document(); |
1219 | 1 | trim_doc(&doc1); |
1220 | 1 | debug!("doc1:\n{}", mml_to_string0 (get_element0 (package10 ))); |
1221 | | |
1222 | 1 | let package2 = parser::parse(different_str).expect("Failed to parse input"); |
1223 | 1 | let doc2 = package2.as_document(); |
1224 | 1 | trim_doc(&doc2); |
1225 | 1 | debug!("doc2:\n{}", mml_to_string0 (get_element0 (&package20 ))); |
1226 | | |
1227 | 1 | assert!(is_same_doc(&doc1, &doc2).is_err()); |
1228 | 1 | } |
1229 | | |
1230 | | #[test] |
1231 | 1 | fn test_entities() { |
1232 | | // this forces initialization |
1233 | 1 | set_rules_dir(super::super::abs_rules_dir_path()).unwrap(); |
1234 | | |
1235 | 1 | let entity_str = set_mathml("<math><mrow><mo>−</mo><mi>𝕞</mi></mrow></math>").unwrap(); |
1236 | 1 | let converted_str = |
1237 | 1 | set_mathml("<math><mrow><mo>−</mo><mi>𝕞</mi></mrow></math>").unwrap(); |
1238 | | |
1239 | | // need to remove unique ids |
1240 | 1 | static ID_MATCH: LazyLock<Regex> = LazyLock::new(|| Regex::new(r#"id='.+?' "#).unwrap()); |
1241 | 1 | let entity_str = ID_MATCH.replace_all(&entity_str, ""); |
1242 | 1 | let converted_str = ID_MATCH.replace_all(&converted_str, ""); |
1243 | 1 | assert_eq!(entity_str, converted_str, "normal entity test failed"); |
1244 | | |
1245 | 1 | let entity_str = set_mathml( |
1246 | | "<math data-quot=\""value"\" data-apos=''value''><mi>XXX</mi></math>", |
1247 | | ) |
1248 | 1 | .unwrap(); |
1249 | 1 | let converted_str = |
1250 | 1 | set_mathml("<math data-quot='\"value\"' data-apos=\"'value'\"><mi>XXX</mi></math>").unwrap(); |
1251 | 1 | let entity_str = ID_MATCH.replace_all(&entity_str, ""); |
1252 | 1 | let converted_str = ID_MATCH.replace_all(&converted_str, ""); |
1253 | 1 | assert_eq!(entity_str, converted_str, "special entities quote test failed"); |
1254 | | |
1255 | 1 | let entity_str = |
1256 | 1 | set_mathml("<math><mo><</mo><mo>></mo><mtext>&lt;</mtext></math>").unwrap(); |
1257 | 1 | let converted_str = |
1258 | 1 | set_mathml("<math><mo><</mo><mo>></mo><mtext>&lt;</mtext></math>") |
1259 | 1 | .unwrap(); |
1260 | 1 | let entity_str = ID_MATCH.replace_all(&entity_str, ""); |
1261 | 1 | let converted_str = ID_MATCH.replace_all(&converted_str, ""); |
1262 | 1 | assert_eq!(entity_str, converted_str, "special entities <,>,& test failed"); |
1263 | 1 | } |
1264 | | |
1265 | | #[test] |
1266 | 1 | fn can_recover_from_invalid_set_rules_dir() { |
1267 | | use std::env; |
1268 | | // MathCAT will check the env var "MathCATRulesDir" as an override, so the following test might succeed if we don't override the env var |
1269 | 1 | unsafe { env::set_var("MathCATRulesDir", "MathCATRulesDir"); } // safe because we are single threaded |
1270 | 1 | assert!(set_rules_dir("someInvalidRulesDir").is_err()); |
1271 | 1 | assert!( |
1272 | 1 | set_rules_dir(super::super::abs_rules_dir_path()).is_ok(), |
1273 | | "\nset_rules_dir to '{}' failed", |
1274 | 0 | super::super::abs_rules_dir_path() |
1275 | | ); |
1276 | 1 | assert!(set_mathml("<math><mn>1</mn></math>").is_ok()); |
1277 | 1 | } |
1278 | | |
1279 | | #[test] |
1280 | 1 | fn single_html_in_mtext() { |
1281 | 1 | let test = "<math><mn>1</mn> <mtext>a<p> para 1</p>bc</mtext> <mi>y</mi></math>"; |
1282 | 1 | let target = "<math><mn>1</mn> <mtext>a para 1bc</mtext> <mi>y</mi></math>"; |
1283 | 1 | assert!(are_parsed_strs_equal(test, target)); |
1284 | 1 | } |
1285 | | |
1286 | | #[test] |
1287 | 1 | fn multiple_html_in_mtext() { |
1288 | 1 | let test = "<math><mn>1</mn> <mtext>a<p>para 1</p> <p>para 2</p>bc </mtext> <mi>y</mi></math>"; |
1289 | 1 | let target = "<math><mn>1</mn> <mtext>apara 1 para 2bc</mtext> <mi>y</mi></math>"; |
1290 | 1 | assert!(are_parsed_strs_equal(test, target)); |
1291 | 1 | } |
1292 | | |
1293 | | #[test] |
1294 | 1 | fn nested_html_in_mtext() { |
1295 | 1 | let test = "<math><mn>1</mn> <mtext>a <ol><li>first</li><li>second</li></ol> bc</mtext> <mi>y</mi></math>"; |
1296 | 1 | let target = "<math><mn>1</mn> <mtext>a firstsecond bc</mtext> <mi>y</mi></math>"; |
1297 | 1 | assert!(are_parsed_strs_equal(test, target)); |
1298 | 1 | } |
1299 | | |
1300 | | #[test] |
1301 | 1 | fn empty_html_in_mtext() { |
1302 | 1 | let test = "<math><mn>1</mn> <mtext>a<br/>bc</mtext> <mi>y</mi></math>"; |
1303 | 1 | let target = "<math><mn>1</mn> <mtext>abc</mtext> <mi>y</mi></math>"; |
1304 | 1 | assert!(are_parsed_strs_equal(test, target)); |
1305 | 1 | } |
1306 | | |
1307 | | #[test] |
1308 | 1 | fn mathml_in_mtext() { |
1309 | 1 | let test = "<math><mtext>if <math> <msup><mi>n</mi><mn>2</mn></msup></math> is real</mtext></math>"; |
1310 | 1 | let target = "<math><mrow><mtext>if </mtext><msup><mi>n</mi><mn>2</mn></msup><mtext> is real</mtext></mrow></math>"; |
1311 | 1 | assert!(are_parsed_strs_equal(test, target)); |
1312 | 1 | } |
1313 | | |
1314 | | #[test] |
1315 | 1 | fn stack_overflow_protection() { |
1316 | 1 | set_rules_dir(super::super::abs_rules_dir_path()).unwrap(); |
1317 | 1 | let mut bad_mathml = String::from("<math>"); |
1318 | 513 | for _ in 0..MAX_DEPTH+11 { |
1319 | 513 | bad_mathml.push_str("<msqrt><mi>n</mi>"); |
1320 | 513 | } |
1321 | 513 | for _ in 0..MAX_DEPTH+11 { |
1322 | 513 | bad_mathml.push_str("</msqrt>"); |
1323 | 513 | } |
1324 | 1 | bad_mathml.push_str("</math>"); |
1325 | 1 | assert_eq!(set_mathml(bad_mathml).unwrap_err().to_string(), "MathML is too deeply nested to process"); |
1326 | 1 | } |
1327 | | |
1328 | | #[test] |
1329 | 1 | fn old_mathml_cleared_on_error() { |
1330 | 1 | set_rules_dir(super::super::abs_rules_dir_path()).unwrap(); |
1331 | 1 | let good_mathml = "<math><mn>3</mn></math>"; |
1332 | 1 | set_mathml(good_mathml).unwrap(); |
1333 | 1 | let bad_mathml = "<math><mi>&xabc;</mi></math>"; |
1334 | 1 | assert!(set_mathml(bad_mathml).is_err()); |
1335 | 1 | assert!(get_spoken_text().unwrap() == ""); |
1336 | 1 | set_mathml(good_mathml).unwrap(); |
1337 | 1 | let bad_mathml = "<math>garbage"; |
1338 | 1 | assert!(set_mathml(bad_mathml).is_err()); |
1339 | 1 | assert!(get_spoken_text().unwrap() == ""); |
1340 | 1 | } |
1341 | | } |