rusteerio 0.0.0-stage → 0.1.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (206) hide show
  1. package/CONTRIBUTING.md +73 -0
  2. package/Cargo.lock +1051 -0
  3. package/Cargo.toml +36 -0
  4. package/LICENSE +21 -0
  5. package/README.md +252 -2
  6. package/RELEASING.md +138 -0
  7. package/THIRD_PARTY_NOTICES.md +7443 -0
  8. package/build.rs +3 -0
  9. package/dist/chunk-DKULX6VO.mjs +3185 -0
  10. package/dist/chunk-DKULX6VO.mjs.map +1 -0
  11. package/dist/chunk-OHPGEY3E.mjs +32 -0
  12. package/dist/chunk-OHPGEY3E.mjs.map +1 -0
  13. package/dist/index.cjs +3230 -0
  14. package/dist/index.cjs.map +1 -0
  15. package/dist/index.d.mts +113 -0
  16. package/dist/index.d.ts +113 -0
  17. package/dist/index.mjs +38 -0
  18. package/dist/index.mjs.map +1 -0
  19. package/dist/slim.cjs +2441 -0
  20. package/dist/slim.cjs.map +1 -0
  21. package/dist/slim.d.mts +11 -0
  22. package/dist/slim.d.ts +11 -0
  23. package/dist/slim.mjs +19 -0
  24. package/dist/slim.mjs.map +1 -0
  25. package/dist/types-BP6TpxE5.d.mts +532 -0
  26. package/dist/types-BP6TpxE5.d.ts +532 -0
  27. package/dist/utils.cjs +61 -0
  28. package/dist/utils.cjs.map +1 -0
  29. package/dist/utils.d.mts +13 -0
  30. package/dist/utils.d.ts +13 -0
  31. package/dist/utils.mjs +17 -0
  32. package/dist/utils.mjs.map +1 -0
  33. package/native/src/arena.rs +1781 -0
  34. package/native/src/attributes.rs +923 -0
  35. package/native/src/codec.rs +172 -0
  36. package/native/src/diagnostics.rs +1249 -0
  37. package/native/src/extraction.rs +226 -0
  38. package/native/src/gc.rs +166 -0
  39. package/native/src/hook_sink.rs +727 -0
  40. package/native/src/hook_stream.rs +513 -0
  41. package/native/src/lib.rs +2095 -0
  42. package/native/src/location.rs +392 -0
  43. package/native/src/runtime.rs +189 -0
  44. package/native/src/selector.rs +1287 -0
  45. package/native/src/sink.rs +261 -0
  46. package/native/src/stream.rs +334 -0
  47. package/native/src/tokenizer.rs +355 -0
  48. package/native/src/tracker.rs +232 -0
  49. package/native/vendor/README.md +25 -0
  50. package/native/vendor/html5ever-0.35.0/.cargo_vcs_info.json +6 -0
  51. package/native/vendor/html5ever-0.35.0/Cargo.lock +828 -0
  52. package/native/vendor/html5ever-0.35.0/Cargo.toml +85 -0
  53. package/native/vendor/html5ever-0.35.0/LICENSE-APACHE +201 -0
  54. package/native/vendor/html5ever-0.35.0/LICENSE-MIT +25 -0
  55. package/native/vendor/html5ever-0.35.0/README.md +60 -0
  56. package/native/vendor/html5ever-0.35.0/benches/html5ever.rs +86 -0
  57. package/native/vendor/html5ever-0.35.0/data/bench/lipsum-zh.html +19 -0
  58. package/native/vendor/html5ever-0.35.0/data/bench/lipsum.html +40 -0
  59. package/native/vendor/html5ever-0.35.0/data/bench/medium-fragment.html +24 -0
  60. package/native/vendor/html5ever-0.35.0/data/bench/small-fragment.html +7 -0
  61. package/native/vendor/html5ever-0.35.0/data/bench/strong.html +1 -0
  62. package/native/vendor/html5ever-0.35.0/data/bench/tiny-fragment.html +1 -0
  63. package/native/vendor/html5ever-0.35.0/examples/arena.rs +356 -0
  64. package/native/vendor/html5ever-0.35.0/examples/capi/tokenize.c +74 -0
  65. package/native/vendor/html5ever-0.35.0/examples/noop-tokenize.rs +47 -0
  66. package/native/vendor/html5ever-0.35.0/examples/noop-tree-builder.rs +132 -0
  67. package/native/vendor/html5ever-0.35.0/examples/print-tree-actions.rs +185 -0
  68. package/native/vendor/html5ever-0.35.0/examples/tokenize.rs +111 -0
  69. package/native/vendor/html5ever-0.35.0/src/driver.rs +148 -0
  70. package/native/vendor/html5ever-0.35.0/src/lib.rs +30 -0
  71. package/native/vendor/html5ever-0.35.0/src/macros.rs +34 -0
  72. package/native/vendor/html5ever-0.35.0/src/serialize/mod.rs +255 -0
  73. package/native/vendor/html5ever-0.35.0/src/tokenizer/char_ref/mod.rs +451 -0
  74. package/native/vendor/html5ever-0.35.0/src/tokenizer/interface.rs +99 -0
  75. package/native/vendor/html5ever-0.35.0/src/tokenizer/mod.rs +2305 -0
  76. package/native/vendor/html5ever-0.35.0/src/tokenizer/states.rs +97 -0
  77. package/native/vendor/html5ever-0.35.0/src/tree_builder/data.rs +170 -0
  78. package/native/vendor/html5ever-0.35.0/src/tree_builder/mod.rs +1848 -0
  79. package/native/vendor/html5ever-0.35.0/src/tree_builder/rules.rs +1494 -0
  80. package/native/vendor/html5ever-0.35.0/src/tree_builder/tag_sets.rs +114 -0
  81. package/native/vendor/html5ever-0.35.0/src/tree_builder/types.rs +91 -0
  82. package/native/vendor/html5ever-0.35.0/src/util/str.rs +52 -0
  83. package/native/vendor/html5gum-0.8.4/Cargo.lock +1006 -0
  84. package/native/vendor/html5gum-0.8.4/Cargo.toml +109 -0
  85. package/native/vendor/html5gum-0.8.4/LICENSE +19 -0
  86. package/native/vendor/html5gum-0.8.4/README.md +110 -0
  87. package/native/vendor/html5gum-0.8.4/benches/files/0050.html +2131 -0
  88. package/native/vendor/html5gum-0.8.4/benches/files/6D30.html +1004 -0
  89. package/native/vendor/html5gum-0.8.4/benches/files.rs +32 -0
  90. package/native/vendor/html5gum-0.8.4/benches/patterns.rs +31 -0
  91. package/native/vendor/html5gum-0.8.4/src/arrayvec.rs +29 -0
  92. package/native/vendor/html5gum-0.8.4/src/char_validator.rs +134 -0
  93. package/native/vendor/html5gum-0.8.4/src/emitters/callback.rs +676 -0
  94. package/native/vendor/html5gum-0.8.4/src/emitters/default.rs +207 -0
  95. package/native/vendor/html5gum-0.8.4/src/emitters/emitter.rs +344 -0
  96. package/native/vendor/html5gum-0.8.4/src/emitters/html5ever.rs +306 -0
  97. package/native/vendor/html5gum-0.8.4/src/emitters/mod.rs +25 -0
  98. package/native/vendor/html5gum-0.8.4/src/entities.rs +3342 -0
  99. package/native/vendor/html5gum-0.8.4/src/error.rs +96 -0
  100. package/native/vendor/html5gum-0.8.4/src/htmlstring.rs +130 -0
  101. package/native/vendor/html5gum-0.8.4/src/lib.rs +66 -0
  102. package/native/vendor/html5gum-0.8.4/src/machine.rs +2136 -0
  103. package/native/vendor/html5gum-0.8.4/src/machine_helper.rs +277 -0
  104. package/native/vendor/html5gum-0.8.4/src/read_helper.rs +244 -0
  105. package/native/vendor/html5gum-0.8.4/src/reader.rs +440 -0
  106. package/native/vendor/html5gum-0.8.4/src/span.rs +98 -0
  107. package/native/vendor/html5gum-0.8.4/src/state.rs +16 -0
  108. package/native/vendor/html5gum-0.8.4/src/testutils.rs +71 -0
  109. package/native/vendor/html5gum-0.8.4/src/tokenizer.rs +189 -0
  110. package/native/vendor/html5gum-0.8.4/src/utils.rs +88 -0
  111. package/native/vendor/markup5ever-0.35.0/.cargo_vcs_info.json +6 -0
  112. package/native/vendor/markup5ever-0.35.0/Cargo.toml +45 -0
  113. package/native/vendor/markup5ever-0.35.0/LICENSE-APACHE +201 -0
  114. package/native/vendor/markup5ever-0.35.0/LICENSE-MIT +25 -0
  115. package/native/vendor/markup5ever-0.35.0/interface/mod.rs +422 -0
  116. package/native/vendor/markup5ever-0.35.0/interface/tree_builder.rs +280 -0
  117. package/native/vendor/markup5ever-0.35.0/lib.rs +59 -0
  118. package/native/vendor/markup5ever-0.35.0/serialize.rs +77 -0
  119. package/native/vendor/markup5ever-0.35.0/util/buffer_queue.rs +333 -0
  120. package/native/vendor/markup5ever-0.35.0/util/smallcharset.rs +88 -0
  121. package/package.json +137 -4
  122. package/rusteerio.darwin-arm64.node +0 -0
  123. package/rusteerio.darwin-x64.node +0 -0
  124. package/rusteerio.linux-arm64-gnu.node +0 -0
  125. package/rusteerio.linux-arm64-musl.node +0 -0
  126. package/rusteerio.linux-x64-gnu.node +0 -0
  127. package/rusteerio.linux-x64-musl.node +0 -0
  128. package/rusteerio.win32-arm64-msvc.node +0 -0
  129. package/rusteerio.win32-x64-msvc.node +0 -0
  130. package/scripts/check-native.mjs +15 -0
  131. package/scripts/check-release.mjs +235 -0
  132. package/scripts/fix-declarations.mjs +27 -0
  133. package/scripts/generate-notices.mjs +176 -0
  134. package/scripts/notices/README.md +13 -0
  135. package/scripts/notices/cheerio-MIT.txt +21 -0
  136. package/scripts/notices/html5ever-AUTHORS.txt +20 -0
  137. package/scripts/notices/html5ever-COPYRIGHT.txt +8 -0
  138. package/scripts/notices/html5ever-LICENSE-APACHE.txt +201 -0
  139. package/scripts/notices/html5ever-LICENSE-MIT.txt +25 -0
  140. package/scripts/notices/napi-rs-MIT.txt +43 -0
  141. package/scripts/notices/neon-MIT.txt +19 -0
  142. package/scripts/notices/runtime-01c266bced4a434da005.txt +3 -0
  143. package/scripts/notices/runtime-0a54e647fe54104658b5.txt +14 -0
  144. package/scripts/notices/runtime-0b74dfa0bcee5c420c6b.txt +25 -0
  145. package/scripts/notices/runtime-0d542e0c8804e39aa7f3.txt +177 -0
  146. package/scripts/notices/runtime-0f96a83840e146e43c0e.txt +21 -0
  147. package/scripts/notices/runtime-123a331b5dbf04c30097.txt +25 -0
  148. package/scripts/notices/runtime-172020dbfd5b53a226df.txt +34 -0
  149. package/scripts/notices/runtime-209fbbe0ad52d9235e37.txt +26 -0
  150. package/scripts/notices/runtime-23f18e03dc49df91622f.txt +23 -0
  151. package/scripts/notices/runtime-268872b9816f90fd8e85.txt +220 -0
  152. package/scripts/notices/runtime-35242e7a83f69875e6ed.txt +176 -0
  153. package/scripts/notices/runtime-378f5840b258e2779c39.txt +25 -0
  154. package/scripts/notices/runtime-4108245a1f2df9d4e94d.txt +25 -0
  155. package/scripts/notices/runtime-4455bf75a91154108304.txt +21 -0
  156. package/scripts/notices/runtime-553fffcd9b1cb158bc3e.txt +13 -0
  157. package/scripts/notices/runtime-62c7a1e35f56406896d7.txt +176 -0
  158. package/scripts/notices/runtime-6485b8ed310d3f0340bf.txt +25 -0
  159. package/scripts/notices/runtime-69fef7b0f322a6555415.txt +201 -0
  160. package/scripts/notices/runtime-6df43f6f4b5d4587f3d8.txt +187 -0
  161. package/scripts/notices/runtime-799e9ca9d179295ef372.txt +24 -0
  162. package/scripts/notices/runtime-7b63ecd5f1902af1b637.txt +25 -0
  163. package/scripts/notices/runtime-7bc7d5de2a0b793d4e1a.txt +21 -0
  164. package/scripts/notices/runtime-7e12e5df4bae12cb2158.txt +24 -0
  165. package/scripts/notices/runtime-861399f8c21c042b1105.txt +12 -0
  166. package/scripts/notices/runtime-8ada45cd9f843acf64e4.txt +201 -0
  167. package/scripts/notices/runtime-90eb64f0279b0d9432ac.txt +12 -0
  168. package/scripts/notices/runtime-a60eea817514531668d7.txt +201 -0
  169. package/scripts/notices/runtime-a7c936ff1ed8fa340172.txt +25 -0
  170. package/scripts/notices/runtime-ab6eec6caf0fa5775e41.txt +275 -0
  171. package/scripts/notices/runtime-b1181a40b2a7b25cf66f.txt +19 -0
  172. package/scripts/notices/runtime-b71bd43a069ca0641a9e.txt +25 -0
  173. package/scripts/notices/runtime-c16f8dcf1a368b83be78.txt +201 -0
  174. package/scripts/notices/runtime-c2cfccb812fe482101a8.txt +21 -0
  175. package/scripts/notices/runtime-e99d88d232bf57d70f0f.txt +25 -0
  176. package/scripts/notices/runtime-f32fb3b417a194167cfa.txt +9 -0
  177. package/scripts/notices/runtime-f5062c9a188d81dfe66b.txt +39 -0
  178. package/scripts/notices/runtime-f9070a832833fa7488fd.txt +4 -0
  179. package/scripts/notices/runtime-fab3dd6bdab226f1c086.txt +373 -0
  180. package/scripts/notices/runtime-fba3318d0c600380177e.txt +62 -0
  181. package/scripts/notices/runtime-ff8f68cb076caf8cefe7.txt +25 -0
  182. package/scripts/notices/runtime-ff92bed461f50338dd70.txt +74 -0
  183. package/scripts/notices/runtime.json +1632 -0
  184. package/scripts/notices/sources.json +555 -0
  185. package/scripts/notices/xray-MIT.txt +21 -0
  186. package/src/attributes.ts +271 -0
  187. package/src/codec.ts +74 -0
  188. package/src/dom.ts +87 -0
  189. package/src/extract-plan.ts +97 -0
  190. package/src/extract.ts +130 -0
  191. package/src/hooks.ts +220 -0
  192. package/src/index.ts +223 -0
  193. package/src/io.ts +491 -0
  194. package/src/native.ts +135 -0
  195. package/src/nodes.ts +290 -0
  196. package/src/parse.ts +53 -0
  197. package/src/parser-types.ts +157 -0
  198. package/src/router.ts +327 -0
  199. package/src/selection.ts +566 -0
  200. package/src/slim.ts +12 -0
  201. package/src/transport.ts +104 -0
  202. package/src/types.ts +186 -0
  203. package/src/utils.ts +29 -0
  204. package/src/workers.ts +21 -0
  205. package/tsconfig.json +15 -0
  206. package/tsup.config.ts +16 -0
@@ -0,0 +1,513 @@
1
+ //! A blocking native tokenizer with live JavaScript adapter calls. Its input and
2
+ //! callback waits release the same CPU permit used by all other native jobs.
3
+ use crate::{diagnostics::DiagnosticContext, runtime, NativeDocument};
4
+ use napi::bindgen_prelude::*;
5
+ use napi::threadsafe_function::{ErrorStrategy, ThreadsafeFunction, ThreadsafeFunctionCallMode};
6
+ use napi::JsFunction;
7
+ use napi_derive::napi;
8
+ use serde_json::{json, Value};
9
+ use std::cell::RefCell;
10
+ use std::io;
11
+ use std::rc::Rc;
12
+ use std::sync::atomic::{AtomicBool, AtomicU64, Ordering};
13
+ use std::sync::{mpsc, Arc, Condvar, Mutex};
14
+
15
+ fn closed() -> Error {
16
+ Error::new(Status::GenericFailure, "Native hook stream is closed")
17
+ }
18
+ fn interrupted() -> io::Error {
19
+ io::Error::new(io::ErrorKind::Interrupted, "Native hook stream is closed")
20
+ }
21
+ type Reply = mpsc::Sender<Result<NativeDocument>>;
22
+ enum Message {
23
+ Chunk(String, mpsc::Sender<Result<()>>),
24
+ Finish(Reply),
25
+ }
26
+ struct State {
27
+ sender: Mutex<Option<mpsc::SyncSender<Message>>>,
28
+ cancelled: Arc<AtomicBool>,
29
+ ended: AtomicBool,
30
+ issued: AtomicU64,
31
+ turn: Mutex<u64>,
32
+ wake: Condvar,
33
+ }
34
+ impl State {
35
+ fn send(&self, sequence: u64, message: Message) -> Result<()> {
36
+ let mut turn = self.turn.lock().unwrap();
37
+ while *turn != sequence && !self.cancelled.load(Ordering::Acquire) {
38
+ turn = self.wake.wait(turn).unwrap();
39
+ }
40
+ if self.cancelled.load(Ordering::Acquire) {
41
+ return Err(closed());
42
+ }
43
+ let result = self
44
+ .sender
45
+ .lock()
46
+ .unwrap()
47
+ .as_ref()
48
+ .ok_or_else(closed)?
49
+ .send(message)
50
+ .map_err(|_| closed());
51
+ *turn += 1;
52
+ self.wake.notify_all();
53
+ result
54
+ }
55
+ }
56
+ impl Drop for State {
57
+ fn drop(&mut self) {
58
+ self.cancelled.store(true, Ordering::Release);
59
+ self.sender.get_mut().unwrap().take();
60
+ }
61
+ }
62
+ #[napi]
63
+ pub struct NativeHookStream {
64
+ state: Arc<State>,
65
+ }
66
+ #[napi]
67
+ impl NativeHookStream {
68
+ #[napi(constructor)]
69
+ pub fn new(
70
+ env: Env,
71
+ is_document: bool,
72
+ options_json: String,
73
+ bridge: JsFunction,
74
+ ) -> Result<Self> {
75
+ Self::create(env, is_document, options_json, bridge, false)
76
+ }
77
+
78
+ #[napi]
79
+ pub fn write(&self, chunk: String) -> Result<AsyncTask<HookWriteTask>> {
80
+ Ok(AsyncTask::new(self.write_task(chunk)?))
81
+ }
82
+ #[napi]
83
+ pub fn finish(&self) -> Result<AsyncTask<HookFinishTask>> {
84
+ Ok(AsyncTask::new(self.finish_task()?))
85
+ }
86
+ #[napi]
87
+ pub fn abort(&self) {
88
+ self.state.cancelled.store(true, Ordering::Release);
89
+ self.state.ended.store(true, Ordering::Release);
90
+ self.state.sender.lock().unwrap().take();
91
+ self.state.wake.notify_all();
92
+ }
93
+ }
94
+ impl NativeHookStream {
95
+ fn create(
96
+ env: Env,
97
+ is_document: bool,
98
+ options_json: String,
99
+ bridge: JsFunction,
100
+ native: bool,
101
+ ) -> Result<Self> {
102
+ let options: Value = serde_json::from_str(&options_json).map_err(crate::json_error)?;
103
+ let mut function: ThreadsafeFunction<String, ErrorStrategy::Fatal> = bridge
104
+ .create_threadsafe_function(
105
+ 0,
106
+ |context: napi::threadsafe_function::ThreadSafeCallContext<String>| {
107
+ Ok(vec![context.env.create_string(&context.value)?])
108
+ },
109
+ )?;
110
+ function.unref(&env)?;
111
+ let (tx, rx) = mpsc::sync_channel(2);
112
+ let cancelled = Arc::new(AtomicBool::new(false));
113
+ let worker_cancel = cancelled.clone();
114
+ std::thread::Builder::new()
115
+ .name("rusteerio-hook-stream".into())
116
+ .spawn(move || run(rx, worker_cancel, options, is_document, function, native))
117
+ .map_err(|error| Error::new(Status::GenericFailure, error.to_string()))?;
118
+ Ok(Self {
119
+ state: Arc::new(State {
120
+ sender: Mutex::new(Some(tx)),
121
+ cancelled,
122
+ ended: AtomicBool::new(false),
123
+ issued: AtomicU64::new(0),
124
+ turn: Mutex::new(0),
125
+ wake: Condvar::new(),
126
+ }),
127
+ })
128
+ }
129
+ pub(crate) fn native(
130
+ env: Env,
131
+ is_document: bool,
132
+ options_json: String,
133
+ bridge: JsFunction,
134
+ ) -> Result<Self> {
135
+ Self::create(env, is_document, options_json, bridge, true)
136
+ }
137
+ pub(crate) fn write_task(&self, chunk: String) -> Result<HookWriteTask> {
138
+ if self.state.ended.load(Ordering::Acquire) {
139
+ return Err(closed());
140
+ }
141
+ let sequence = self.state.issued.fetch_add(1, Ordering::Relaxed);
142
+ Ok(HookWriteTask {
143
+ state: self.state.clone(),
144
+ sequence,
145
+ chunk: Some(chunk),
146
+ })
147
+ }
148
+ pub(crate) fn finish_task(&self) -> Result<HookFinishTask> {
149
+ if self.state.ended.swap(true, Ordering::AcqRel) {
150
+ return Err(closed());
151
+ }
152
+ let sequence = self.state.issued.fetch_add(1, Ordering::Relaxed);
153
+ Ok(HookFinishTask {
154
+ state: self.state.clone(),
155
+ sequence,
156
+ })
157
+ }
158
+ }
159
+ pub struct HookWriteTask {
160
+ state: Arc<State>,
161
+ sequence: u64,
162
+ chunk: Option<String>,
163
+ }
164
+ #[napi]
165
+ impl Task for HookWriteTask {
166
+ type Output = ();
167
+ type JsValue = ();
168
+ fn compute(&mut self) -> Result<()> {
169
+ let (tx, rx) = mpsc::channel();
170
+ self.state.send(
171
+ self.sequence,
172
+ Message::Chunk(self.chunk.take().unwrap_or_default(), tx),
173
+ )?;
174
+ rx.recv().map_err(|_| closed())?
175
+ }
176
+ fn resolve(&mut self, _: Env, _: ()) -> Result<()> {
177
+ Ok(())
178
+ }
179
+ }
180
+ pub struct HookFinishTask {
181
+ state: Arc<State>,
182
+ sequence: u64,
183
+ }
184
+ #[napi]
185
+ impl Task for HookFinishTask {
186
+ type Output = NativeDocument;
187
+ type JsValue = NativeDocument;
188
+ fn compute(&mut self) -> Result<NativeDocument> {
189
+ let (tx, rx) = mpsc::channel();
190
+ self.state.send(self.sequence, Message::Finish(tx))?;
191
+ rx.recv().map_err(|_| closed())?
192
+ }
193
+ fn resolve(&mut self, _: Env, output: NativeDocument) -> Result<NativeDocument> {
194
+ Ok(output)
195
+ }
196
+ }
197
+
198
+ struct Budget(RefCell<Option<runtime::Permit>>);
199
+ impl Budget {
200
+ fn pause(&self) {
201
+ self.0.borrow_mut().take();
202
+ }
203
+ fn resume(&self) -> Result<()> {
204
+ if self.0.borrow().is_none() {
205
+ *self.0.borrow_mut() = Some(runtime::permit()?);
206
+ }
207
+ Ok(())
208
+ }
209
+ }
210
+ struct Input {
211
+ receiver: mpsc::Receiver<Message>,
212
+ buffer: Vec<u8>,
213
+ position: usize,
214
+ pending: Option<(String, usize)>,
215
+ ack: Option<mpsc::Sender<Result<()>>>,
216
+ finish: Rc<RefCell<Option<Reply>>>,
217
+ eof: bool,
218
+ budget: Rc<Budget>,
219
+ context: Rc<DiagnosticContext>,
220
+ cancelled: Arc<AtomicBool>,
221
+ }
222
+ impl Input {
223
+ fn available(&mut self, length: usize) -> io::Result<bool> {
224
+ if self.cancelled.load(Ordering::Acquire) {
225
+ return Err(interrupted());
226
+ }
227
+ while self.buffer.len() - self.position < length && !self.eof {
228
+ if self.position > 0 {
229
+ self.buffer.drain(..self.position);
230
+ self.position = 0;
231
+ }
232
+ if self.cancelled.load(Ordering::Acquire) {
233
+ if let Some(ack) = self.ack.take() {
234
+ let _ = ack.send(Err(closed()));
235
+ }
236
+ return Err(interrupted());
237
+ }
238
+ if let Some((chunk, position)) = &mut self.pending {
239
+ self.budget
240
+ .resume()
241
+ .map_err(|error| io::Error::other(error.to_string()))?;
242
+ let mut end = (*position + 16 * 1024).min(chunk.len());
243
+ while !chunk.is_char_boundary(end) {
244
+ end -= 1;
245
+ }
246
+ let part = &chunk[*position..end];
247
+ self.context.append_source(part);
248
+ self.buffer.extend_from_slice(part.as_bytes());
249
+ *position = end;
250
+ if end == chunk.len() {
251
+ self.pending = None;
252
+ }
253
+ continue;
254
+ }
255
+ self.budget.pause();
256
+ // Retain only bounded tokenizer lookahead across the acknowledgment.
257
+ if let Some(ack) = self.ack.take() {
258
+ let _ = ack.send(Ok(()));
259
+ }
260
+ let message = self.receiver.recv().map_err(|_| interrupted())?;
261
+ self.budget
262
+ .resume()
263
+ .map_err(|error| io::Error::other(error.to_string()))?;
264
+ match message {
265
+ Message::Chunk(chunk, ack) => {
266
+ self.pending = Some((chunk, 0));
267
+ self.ack = Some(ack);
268
+ }
269
+ Message::Finish(reply) => {
270
+ *self.finish.borrow_mut() = Some(reply);
271
+ self.eof = true;
272
+ }
273
+ }
274
+ }
275
+ Ok(self.buffer.len() - self.position >= length)
276
+ }
277
+ }
278
+ impl html5gum::Reader for Input {
279
+ type Error = io::Error;
280
+ fn read_byte(&mut self) -> io::Result<Option<u8>> {
281
+ if self.available(1)? {
282
+ let byte = self.buffer[self.position];
283
+ self.position += 1;
284
+ Ok(Some(byte))
285
+ } else {
286
+ Ok(None)
287
+ }
288
+ }
289
+ fn try_read_string(&mut self, source: &[u8], case_sensitive: bool) -> io::Result<bool> {
290
+ if !self.available(source.len())? {
291
+ return Ok(false);
292
+ }
293
+ let equal = self.buffer[self.position..]
294
+ .iter()
295
+ .copied()
296
+ .take(source.len())
297
+ .zip(source.iter().copied())
298
+ .all(|(a, b)| {
299
+ if case_sensitive {
300
+ a == b
301
+ } else {
302
+ a.eq_ignore_ascii_case(&b)
303
+ }
304
+ });
305
+ if equal {
306
+ self.position += source.len();
307
+ }
308
+ Ok(equal)
309
+ }
310
+ fn read_until<'a>(
311
+ &'a mut self,
312
+ needle: &[u8],
313
+ _: &'a mut [u8; 4],
314
+ ) -> io::Result<Option<&'a [u8]>> {
315
+ if !self.available(1)? {
316
+ return Ok(None);
317
+ }
318
+ let start = self.position;
319
+ let bytes = &self.buffer[start..];
320
+ let count = bytes
321
+ .iter()
322
+ .position(|byte| needle.contains(byte))
323
+ .map_or(bytes.len(), |at| if at == 0 { 1 } else { at });
324
+ self.position += count;
325
+ Ok(Some(&self.buffer[start..self.position]))
326
+ }
327
+ }
328
+ impl Drop for Input {
329
+ fn drop(&mut self) {
330
+ if let Some(ack) = self.ack.take() {
331
+ let _ = ack.send(if self.cancelled.load(Ordering::Acquire) {
332
+ Err(closed())
333
+ } else {
334
+ Ok(())
335
+ });
336
+ }
337
+ }
338
+ }
339
+ fn run(
340
+ receiver: mpsc::Receiver<Message>,
341
+ cancelled: Arc<AtomicBool>,
342
+ options: Value,
343
+ is_document: bool,
344
+ function: ThreadsafeFunction<String, ErrorStrategy::Fatal>,
345
+ native: bool,
346
+ ) {
347
+ let budget = Rc::new(Budget(RefCell::new(None)));
348
+ let keep_source = options.get("sourceCodeLocationInfo") == Some(&Value::Bool(true));
349
+ let track_positions =
350
+ keep_source || native || options.get("__collectParseErrors") == Some(&Value::Bool(true));
351
+ let context = Rc::new(DiagnosticContext::streaming_with_source_mode(
352
+ keep_source,
353
+ track_positions,
354
+ ));
355
+ let finish = Rc::new(RefCell::new(None));
356
+ let input = Input {
357
+ receiver,
358
+ buffer: Vec::new(),
359
+ position: 0,
360
+ pending: None,
361
+ ack: None,
362
+ finish: finish.clone(),
363
+ eof: false,
364
+ budget: budget.clone(),
365
+ context: context.clone(),
366
+ cancelled: cancelled.clone(),
367
+ };
368
+ let callback_budget = budget.clone();
369
+ let callback_cancel = cancelled.clone();
370
+ let call = Box::new(move |method: &str, args: Value| -> Result<Value> {
371
+ callback_budget.pause();
372
+ if callback_cancel.load(Ordering::Acquire) {
373
+ return Err(closed());
374
+ }
375
+ let (tx, rx) = mpsc::channel();
376
+ let status = function.call_with_return_value(
377
+ json!({"method":method,"args":args}).to_string(),
378
+ ThreadsafeFunctionCallMode::NonBlocking,
379
+ move |reply: String| {
380
+ let _ = tx.send(reply);
381
+ Ok(())
382
+ },
383
+ );
384
+ if status != Status::Ok {
385
+ return Err(Error::new(status, "Could not invoke the parser hook"));
386
+ }
387
+ let text = rx.recv().map_err(|_| closed())?;
388
+ callback_budget.resume()?;
389
+ let value: Value = serde_json::from_str(&text).map_err(crate::json_error)?;
390
+ if value.get("__rusteerioHookError") == Some(&Value::Bool(true)) {
391
+ callback_cancel.store(true, Ordering::Release);
392
+ return Err(Error::new(
393
+ Status::GenericFailure,
394
+ "User parser hook failed",
395
+ ));
396
+ }
397
+ Ok(value)
398
+ });
399
+ let result = (|| {
400
+ budget.resume()?;
401
+ if native {
402
+ let sink = crate::sink::CompatibleSink::with_errors(if is_document {
403
+ scraper::Html::new_document()
404
+ } else {
405
+ scraper::Html::new_fragment()
406
+ });
407
+ let errors = RefCell::new(Vec::new());
408
+ let failure = RefCell::new(None);
409
+ let html = crate::diagnostics::parse_reader_with_sink(
410
+ input,
411
+ is_document,
412
+ &options,
413
+ sink,
414
+ context.clone(),
415
+ |sink, context| {
416
+ for message in sink.take_errors() {
417
+ if let Some(error) = context.error(&message) {
418
+ errors.borrow_mut().push(error.clone());
419
+ if failure.borrow().is_none() {
420
+ if let Err(error) = call("__parseError", json!([error])) {
421
+ *failure.borrow_mut() = Some(error);
422
+ cancelled.store(true, Ordering::Release);
423
+ }
424
+ }
425
+ }
426
+ }
427
+ },
428
+ )
429
+ .map_err(|error| Error::new(Status::GenericFailure, error.to_string()))?;
430
+ if let Some(error) = failure.into_inner() {
431
+ return Err(error);
432
+ }
433
+ let source = (options.get("sourceCodeLocationInfo") == Some(&Value::Bool(true)))
434
+ .then(|| context.source_between(0, context.source_len()));
435
+ let mut parser_options = options;
436
+ parser_options["__collectParseErrors"] = Value::Bool(false);
437
+ let document = NativeDocument::from_html(html, is_document, parser_options, source);
438
+ document.set_diagnostic_records(errors.into_inner());
439
+ Ok(document)
440
+ } else {
441
+ crate::hook_sink::parse_reader(input, options, is_document, context, call)
442
+ }
443
+ })();
444
+ budget.pause();
445
+ cancelled.store(true, Ordering::Release);
446
+ if let Some(reply) = finish.borrow_mut().take() {
447
+ let _ = reply.send(result);
448
+ };
449
+ }
450
+
451
+ #[cfg(test)]
452
+ mod tests {
453
+ use super::*;
454
+ #[test]
455
+ fn blocking_reader_preserves_tokens_errors_and_backpressure_across_chunks() {
456
+ let source = "<!DOCTYPE html>😀\r\n<p a a='x'>&#0;&amp;<!-- c --></p><script>A</script>";
457
+ let expected =
458
+ crate::NativeDocument::parse_options(source, true, json!({}), None).outer_html(vec![0]);
459
+ let expected_errors = crate::diagnostics::collect(source, true, &json!({}));
460
+ let (tx, rx) = mpsc::sync_channel(2);
461
+ let worker = std::thread::spawn(move || {
462
+ let budget = Rc::new(Budget(RefCell::new(None)));
463
+ let context = Rc::new(DiagnosticContext::streaming(true));
464
+ let finish = Rc::new(RefCell::new(None));
465
+ let input = Input {
466
+ receiver: rx,
467
+ buffer: Vec::new(),
468
+ position: 0,
469
+ pending: None,
470
+ ack: None,
471
+ finish: finish.clone(),
472
+ eof: false,
473
+ budget: budget.clone(),
474
+ context: context.clone(),
475
+ cancelled: Arc::new(AtomicBool::new(false)),
476
+ };
477
+ budget.resume().unwrap();
478
+ let mut errors = Vec::new();
479
+ let html = crate::diagnostics::parse_reader_with_sink(
480
+ input,
481
+ true,
482
+ &json!({}),
483
+ crate::sink::CompatibleSink::with_errors(scraper::Html::new_document()),
484
+ context,
485
+ |sink, context| {
486
+ for message in sink.take_errors() {
487
+ if let Some(error) = context.error(&message) {
488
+ errors.push(error);
489
+ }
490
+ }
491
+ },
492
+ )
493
+ .unwrap();
494
+ budget.pause();
495
+ let document = NativeDocument::from_html(html, true, json!({}), None);
496
+ let output = document.outer_html(vec![0]);
497
+ let reply = finish.borrow_mut().take().unwrap();
498
+ reply.send(Ok(document)).unwrap();
499
+ (output, errors)
500
+ });
501
+ for ch in source.chars() {
502
+ let (ack, done) = mpsc::channel();
503
+ tx.send(Message::Chunk(ch.to_string(), ack)).unwrap();
504
+ done.recv().unwrap().unwrap();
505
+ }
506
+ let (reply, done) = mpsc::channel();
507
+ tx.send(Message::Finish(reply)).unwrap();
508
+ done.recv().unwrap().unwrap();
509
+ let (output, errors) = worker.join().unwrap();
510
+ assert_eq!(output, expected);
511
+ assert_eq!(errors, expected_errors);
512
+ }
513
+ }