/src/trafficserver/lib/yamlcpp/src/singledocparser.cpp
Line | Count | Source |
1 | | #include <algorithm> |
2 | | #include <cstdio> |
3 | | #include <sstream> |
4 | | |
5 | | #include "collectionstack.h" // IWYU pragma: keep |
6 | | #include "scanner.h" |
7 | | #include "singledocparser.h" |
8 | | #include "tag.h" |
9 | | #include "token.h" |
10 | | #include "yaml-cpp/depthguard.h" |
11 | | #include "yaml-cpp/emitterstyle.h" |
12 | | #include "yaml-cpp/eventhandler.h" |
13 | | #include "yaml-cpp/exceptions.h" // IWYU pragma: keep |
14 | | #include "yaml-cpp/mark.h" |
15 | | #include "yaml-cpp/null.h" |
16 | | |
17 | | namespace YAML { |
18 | | SingleDocParser::SingleDocParser(Scanner& scanner, const Directives& directives) |
19 | 3.43k | : m_scanner(scanner), |
20 | 3.43k | m_directives(directives), |
21 | 3.43k | m_pCollectionStack(new CollectionStack), |
22 | 3.43k | m_anchors{}, |
23 | 3.43k | m_curAnchor(0) {} |
24 | | |
25 | 3.43k | SingleDocParser::~SingleDocParser() = default; |
26 | | |
27 | | // HandleDocument |
28 | | // . Handles the next document |
29 | | // . Throws a ParserException on error. |
30 | 3.43k | void SingleDocParser::HandleDocument(EventHandler& eventHandler) { |
31 | 3.43k | assert(!m_scanner.empty()); // guaranteed that there are tokens |
32 | 3.43k | assert(!m_curAnchor); |
33 | | |
34 | 3.43k | eventHandler.OnDocumentStart(m_scanner.peek().mark); |
35 | | |
36 | | // eat doc start |
37 | 3.43k | if (m_scanner.peek().type == Token::DOC_START) |
38 | 2 | m_scanner.pop(); |
39 | | |
40 | | // recurse! |
41 | 3.43k | HandleNode(eventHandler); |
42 | | |
43 | 3.43k | eventHandler.OnDocumentEnd(); |
44 | | |
45 | | // and finally eat any doc ends we see |
46 | 3.75k | while (!m_scanner.empty() && m_scanner.peek().type == Token::DOC_END) |
47 | 318 | m_scanner.pop(); |
48 | 3.43k | } |
49 | | |
50 | 95.3k | void SingleDocParser::HandleNode(EventHandler& eventHandler) { |
51 | 95.3k | DepthGuard<500> depthguard(depth, m_scanner.mark(), ErrorMsg::BAD_FILE); |
52 | | |
53 | | // an empty node *is* a possibility |
54 | 95.3k | if (m_scanner.empty()) { |
55 | 20 | eventHandler.OnNull(m_scanner.mark(), NullAnchor); |
56 | 20 | return; |
57 | 20 | } |
58 | | |
59 | | // save location |
60 | 95.3k | Mark mark = m_scanner.peek().mark; |
61 | | |
62 | | // special case: a value node by itself must be a map, with no header |
63 | 95.3k | if (m_scanner.peek().type == Token::VALUE) { |
64 | 440 | eventHandler.OnMapStart(mark, "?", NullAnchor, EmitterStyle::Default); |
65 | 440 | HandleMap(eventHandler); |
66 | 440 | eventHandler.OnMapEnd(); |
67 | 440 | return; |
68 | 440 | } |
69 | | |
70 | | // special case: an alias node |
71 | 94.9k | if (m_scanner.peek().type == Token::ALIAS) { |
72 | 1.05k | eventHandler.OnAlias(mark, LookupAnchor(mark, m_scanner.peek().value)); |
73 | 1.05k | m_scanner.pop(); |
74 | 1.05k | return; |
75 | 1.05k | } |
76 | | |
77 | 93.8k | std::string tag; |
78 | 93.8k | std::string anchor_name; |
79 | 93.8k | anchor_t anchor; |
80 | 93.8k | ParseProperties(tag, anchor, anchor_name); |
81 | | |
82 | 93.8k | if (!anchor_name.empty()) |
83 | 3.78k | eventHandler.OnAnchor(mark, anchor_name); |
84 | | |
85 | | // after parsing properties, an empty node is again a possibility |
86 | 93.8k | if (m_scanner.empty()) { |
87 | 605 | eventHandler.OnNull(mark, anchor); |
88 | 605 | return; |
89 | 605 | } |
90 | | |
91 | 93.2k | const Token& token = m_scanner.peek(); |
92 | | |
93 | | // add non-specific tags |
94 | 93.2k | if (tag.empty()) |
95 | 86.9k | tag = (token.type == Token::NON_PLAIN_SCALAR ? "!" : "?"); |
96 | | |
97 | 93.2k | if (token.type == Token::PLAIN_SCALAR |
98 | 9.59k | && tag.compare("?") == 0 && IsNullString(token.value)) { |
99 | 2.24k | eventHandler.OnNull(mark, anchor); |
100 | 2.24k | m_scanner.pop(); |
101 | 2.24k | return; |
102 | 2.24k | } |
103 | | |
104 | | // now split based on what kind of node we should be |
105 | 91.0k | switch (token.type) { |
106 | 7.35k | case Token::PLAIN_SCALAR: |
107 | 8.17k | case Token::NON_PLAIN_SCALAR: |
108 | 8.17k | eventHandler.OnScalar(mark, tag, anchor, token.value); |
109 | 8.17k | m_scanner.pop(); |
110 | 8.17k | return; |
111 | 21.2k | case Token::FLOW_SEQ_START: |
112 | 21.2k | eventHandler.OnSequenceStart(mark, tag, anchor, EmitterStyle::Flow); |
113 | 21.2k | HandleSequence(eventHandler); |
114 | 21.2k | eventHandler.OnSequenceEnd(); |
115 | 21.2k | return; |
116 | 8.16k | case Token::BLOCK_SEQ_START: |
117 | 8.16k | eventHandler.OnSequenceStart(mark, tag, anchor, EmitterStyle::Block); |
118 | 8.16k | HandleSequence(eventHandler); |
119 | 8.16k | eventHandler.OnSequenceEnd(); |
120 | 8.16k | return; |
121 | 12.7k | case Token::FLOW_MAP_START: |
122 | 12.7k | eventHandler.OnMapStart(mark, tag, anchor, EmitterStyle::Flow); |
123 | 12.7k | HandleMap(eventHandler); |
124 | 12.7k | eventHandler.OnMapEnd(); |
125 | 12.7k | return; |
126 | 6.57k | case Token::BLOCK_MAP_START: |
127 | 6.57k | eventHandler.OnMapStart(mark, tag, anchor, EmitterStyle::Block); |
128 | 6.57k | HandleMap(eventHandler); |
129 | 6.57k | eventHandler.OnMapEnd(); |
130 | 6.57k | return; |
131 | 21.0k | case Token::KEY: |
132 | | // compact maps can only go in a flow sequence |
133 | 21.0k | if (m_pCollectionStack->GetCurCollectionType() == |
134 | 21.0k | CollectionType::FlowSeq) { |
135 | 20.5k | eventHandler.OnMapStart(mark, tag, anchor, EmitterStyle::Flow); |
136 | 20.5k | HandleMap(eventHandler); |
137 | 20.5k | eventHandler.OnMapEnd(); |
138 | 20.5k | return; |
139 | 20.5k | } |
140 | 458 | break; |
141 | 12.9k | default: |
142 | 12.9k | break; |
143 | 91.0k | } |
144 | | |
145 | 13.4k | if (tag == "?") |
146 | 11.9k | eventHandler.OnNull(mark, anchor); |
147 | 1.49k | else |
148 | 1.49k | eventHandler.OnScalar(mark, tag, anchor, ""); |
149 | 13.4k | } |
150 | | |
151 | 29.4k | void SingleDocParser::HandleSequence(EventHandler& eventHandler) { |
152 | | // split based on start token |
153 | 29.4k | switch (m_scanner.peek().type) { |
154 | 8.16k | case Token::BLOCK_SEQ_START: |
155 | 8.16k | HandleBlockSequence(eventHandler); |
156 | 8.16k | break; |
157 | 21.2k | case Token::FLOW_SEQ_START: |
158 | 21.2k | HandleFlowSequence(eventHandler); |
159 | 21.2k | break; |
160 | 0 | default: |
161 | 0 | break; |
162 | 29.4k | } |
163 | 29.4k | } |
164 | | |
165 | 8.16k | void SingleDocParser::HandleBlockSequence(EventHandler& eventHandler) { |
166 | | // eat start token |
167 | 8.16k | m_scanner.pop(); |
168 | 8.16k | m_pCollectionStack->PushCollectionType(CollectionType::BlockSeq); |
169 | | |
170 | 16.9k | while (true) { |
171 | 12.1k | if (m_scanner.empty()) |
172 | 0 | throw ParserException(m_scanner.mark(), ErrorMsg::END_OF_SEQ); |
173 | | |
174 | 12.1k | Token token = m_scanner.peek(); |
175 | 12.1k | if (token.type != Token::BLOCK_ENTRY && token.type != Token::BLOCK_SEQ_END) |
176 | 16 | throw ParserException(token.mark, ErrorMsg::END_OF_SEQ); |
177 | | |
178 | 12.1k | m_scanner.pop(); |
179 | 12.1k | if (token.type == Token::BLOCK_SEQ_END) |
180 | 3.27k | break; |
181 | | |
182 | | // check for null |
183 | 8.83k | if (!m_scanner.empty()) { |
184 | 8.81k | const Token& nextToken = m_scanner.peek(); |
185 | 8.81k | if (nextToken.type == Token::BLOCK_ENTRY || |
186 | 8.21k | nextToken.type == Token::BLOCK_SEQ_END) { |
187 | 978 | eventHandler.OnNull(nextToken.mark, NullAnchor); |
188 | 978 | continue; |
189 | 978 | } |
190 | 8.81k | } |
191 | | |
192 | 7.85k | HandleNode(eventHandler); |
193 | 7.85k | } |
194 | | |
195 | 8.14k | m_pCollectionStack->PopCollectionType(CollectionType::BlockSeq); |
196 | 8.14k | } |
197 | | |
198 | 21.2k | void SingleDocParser::HandleFlowSequence(EventHandler& eventHandler) { |
199 | | // eat start token |
200 | 21.2k | m_scanner.pop(); |
201 | 21.2k | m_pCollectionStack->PushCollectionType(CollectionType::FlowSeq); |
202 | | |
203 | 49.6k | while (true) { |
204 | 30.2k | if (m_scanner.empty()) |
205 | 135 | throw ParserException(m_scanner.mark(), ErrorMsg::END_OF_SEQ_FLOW); |
206 | | |
207 | | // first check for end |
208 | 30.1k | if (m_scanner.peek().type == Token::FLOW_SEQ_END) { |
209 | 1.10k | m_scanner.pop(); |
210 | 1.10k | break; |
211 | 1.10k | } |
212 | | |
213 | | // then read the node |
214 | 29.0k | HandleNode(eventHandler); |
215 | | |
216 | 29.0k | if (m_scanner.empty()) |
217 | 463 | throw ParserException(m_scanner.mark(), ErrorMsg::END_OF_SEQ_FLOW); |
218 | | |
219 | | // now eat the separator (or could be a sequence end, which we ignore - but |
220 | | // if it's neither, then it's a bad node) |
221 | 28.5k | Token& token = m_scanner.peek(); |
222 | 28.5k | if (token.type == Token::FLOW_ENTRY) |
223 | 7.92k | m_scanner.pop(); |
224 | 20.6k | else if (token.type != Token::FLOW_SEQ_END) |
225 | 83 | throw ParserException(token.mark, ErrorMsg::END_OF_SEQ_FLOW); |
226 | 28.5k | } |
227 | | |
228 | 20.5k | m_pCollectionStack->PopCollectionType(CollectionType::FlowSeq); |
229 | 20.5k | } |
230 | | |
231 | 40.3k | void SingleDocParser::HandleMap(EventHandler& eventHandler) { |
232 | | // split based on start token |
233 | 40.3k | switch (m_scanner.peek().type) { |
234 | 6.57k | case Token::BLOCK_MAP_START: |
235 | 6.57k | HandleBlockMap(eventHandler); |
236 | 6.57k | break; |
237 | 12.7k | case Token::FLOW_MAP_START: |
238 | 12.7k | HandleFlowMap(eventHandler); |
239 | 12.7k | break; |
240 | 20.5k | case Token::KEY: |
241 | 20.5k | HandleCompactMap(eventHandler); |
242 | 20.5k | break; |
243 | 440 | case Token::VALUE: |
244 | 440 | HandleCompactMapWithNoKey(eventHandler); |
245 | 440 | break; |
246 | 0 | default: |
247 | 0 | break; |
248 | 40.3k | } |
249 | 40.3k | } |
250 | | |
251 | 6.57k | void SingleDocParser::HandleBlockMap(EventHandler& eventHandler) { |
252 | | // eat start token |
253 | 6.57k | m_scanner.pop(); |
254 | 6.57k | m_pCollectionStack->PushCollectionType(CollectionType::BlockMap); |
255 | | |
256 | 14.1k | while (true) { |
257 | 11.3k | if (m_scanner.empty()) |
258 | 0 | throw ParserException(m_scanner.mark(), ErrorMsg::END_OF_MAP); |
259 | | |
260 | 11.3k | Token token = m_scanner.peek(); |
261 | 11.3k | if (token.type != Token::KEY && token.type != Token::VALUE && |
262 | 3.76k | token.type != Token::BLOCK_MAP_END) |
263 | 50 | throw ParserException(token.mark, ErrorMsg::END_OF_MAP); |
264 | | |
265 | 11.2k | if (token.type == Token::BLOCK_MAP_END) { |
266 | 3.71k | m_scanner.pop(); |
267 | 3.71k | break; |
268 | 3.71k | } |
269 | | |
270 | | // grab key (if non-null) |
271 | 7.55k | if (token.type == Token::KEY) { |
272 | 5.11k | m_scanner.pop(); |
273 | 5.11k | HandleNode(eventHandler); |
274 | 5.11k | } else { |
275 | 2.44k | eventHandler.OnNull(token.mark, NullAnchor); |
276 | 2.44k | } |
277 | | |
278 | | // now grab value (optional) |
279 | 7.55k | if (!m_scanner.empty() && m_scanner.peek().type == Token::VALUE) { |
280 | 4.63k | m_scanner.pop(); |
281 | 4.63k | HandleNode(eventHandler); |
282 | 4.63k | } else { |
283 | 2.92k | eventHandler.OnNull(token.mark, NullAnchor); |
284 | 2.92k | } |
285 | 7.55k | } |
286 | | |
287 | 6.52k | m_pCollectionStack->PopCollectionType(CollectionType::BlockMap); |
288 | 6.52k | } |
289 | | |
290 | 12.7k | void SingleDocParser::HandleFlowMap(EventHandler& eventHandler) { |
291 | | // eat start token |
292 | 12.7k | m_scanner.pop(); |
293 | 12.7k | m_pCollectionStack->PushCollectionType(CollectionType::FlowMap); |
294 | | |
295 | 34.0k | while (true) { |
296 | 22.6k | if (m_scanner.empty()) |
297 | 121 | throw ParserException(m_scanner.mark(), ErrorMsg::END_OF_MAP_FLOW); |
298 | | |
299 | 22.5k | Token& token = m_scanner.peek(); |
300 | 22.5k | const Mark mark = token.mark; |
301 | | // first check for end |
302 | 22.5k | if (token.type == Token::FLOW_MAP_END) { |
303 | 993 | m_scanner.pop(); |
304 | 993 | break; |
305 | 993 | } |
306 | | |
307 | | // grab key (if non-null) |
308 | 21.5k | if (token.type == Token::KEY) { |
309 | 17.7k | m_scanner.pop(); |
310 | 17.7k | HandleNode(eventHandler); |
311 | 17.7k | } else { |
312 | 3.80k | eventHandler.OnNull(mark, NullAnchor); |
313 | 3.80k | } |
314 | | |
315 | | // now grab value (optional) |
316 | 21.5k | if (!m_scanner.empty() && m_scanner.peek().type == Token::VALUE) { |
317 | 6.12k | m_scanner.pop(); |
318 | 6.12k | HandleNode(eventHandler); |
319 | 15.4k | } else { |
320 | 15.4k | eventHandler.OnNull(mark, NullAnchor); |
321 | 15.4k | } |
322 | | |
323 | 21.5k | if (m_scanner.empty()) |
324 | 273 | throw ParserException(m_scanner.mark(), ErrorMsg::END_OF_MAP_FLOW); |
325 | | |
326 | | // now eat the separator (or could be a map end, which we ignore - but if |
327 | | // it's neither, then it's a bad node) |
328 | 21.3k | Token& nextToken = m_scanner.peek(); |
329 | 21.3k | if (nextToken.type == Token::FLOW_ENTRY) |
330 | 8.94k | m_scanner.pop(); |
331 | 12.3k | else if (nextToken.type != Token::FLOW_MAP_END) |
332 | 74 | throw ParserException(nextToken.mark, ErrorMsg::END_OF_MAP_FLOW); |
333 | 21.3k | } |
334 | | |
335 | 12.3k | m_pCollectionStack->PopCollectionType(CollectionType::FlowMap); |
336 | 12.3k | } |
337 | | |
338 | | // . Single "key: value" pair in a flow sequence |
339 | 20.5k | void SingleDocParser::HandleCompactMap(EventHandler& eventHandler) { |
340 | 20.5k | m_pCollectionStack->PushCollectionType(CollectionType::CompactMap); |
341 | | |
342 | | // grab key |
343 | 20.5k | Mark mark = m_scanner.peek().mark; |
344 | 20.5k | m_scanner.pop(); |
345 | 20.5k | HandleNode(eventHandler); |
346 | | |
347 | | // now grab value (optional) |
348 | 20.5k | if (!m_scanner.empty() && m_scanner.peek().type == Token::VALUE) { |
349 | 453 | m_scanner.pop(); |
350 | 453 | HandleNode(eventHandler); |
351 | 20.1k | } else { |
352 | 20.1k | eventHandler.OnNull(mark, NullAnchor); |
353 | 20.1k | } |
354 | | |
355 | 20.5k | m_pCollectionStack->PopCollectionType(CollectionType::CompactMap); |
356 | 20.5k | } |
357 | | |
358 | | // . Single ": value" pair in a flow sequence |
359 | 440 | void SingleDocParser::HandleCompactMapWithNoKey(EventHandler& eventHandler) { |
360 | 440 | m_pCollectionStack->PushCollectionType(CollectionType::CompactMap); |
361 | | |
362 | | // null key |
363 | 440 | eventHandler.OnNull(m_scanner.peek().mark, NullAnchor); |
364 | | |
365 | | // grab value |
366 | 440 | m_scanner.pop(); |
367 | 440 | HandleNode(eventHandler); |
368 | | |
369 | 440 | m_pCollectionStack->PopCollectionType(CollectionType::CompactMap); |
370 | 440 | } |
371 | | |
372 | | // ParseProperties |
373 | | // . Grabs any tag or anchor tokens and deals with them. |
374 | | void SingleDocParser::ParseProperties(std::string& tag, anchor_t& anchor, |
375 | 93.7k | std::string& anchor_name) { |
376 | 93.7k | tag.clear(); |
377 | 93.7k | anchor_name.clear(); |
378 | 93.7k | anchor = NullAnchor; |
379 | | |
380 | 104k | while (true) { |
381 | 104k | if (m_scanner.empty()) |
382 | 605 | return; |
383 | | |
384 | 103k | switch (m_scanner.peek().type) { |
385 | 6.76k | case Token::TAG: |
386 | 6.76k | ParseTag(tag); |
387 | 6.76k | break; |
388 | 3.80k | case Token::ANCHOR: |
389 | 3.80k | ParseAnchor(anchor, anchor_name); |
390 | 3.80k | break; |
391 | 93.1k | default: |
392 | 93.1k | return; |
393 | 103k | } |
394 | 103k | } |
395 | 93.7k | } |
396 | | |
397 | 6.76k | void SingleDocParser::ParseTag(std::string& tag) { |
398 | 6.76k | Token& token = m_scanner.peek(); |
399 | 6.76k | if (!tag.empty()) |
400 | 33 | throw ParserException(token.mark, ErrorMsg::MULTIPLE_TAGS); |
401 | | |
402 | 6.73k | Tag tagInfo(token); |
403 | 6.73k | tag = tagInfo.Translate(m_directives); |
404 | 6.73k | m_scanner.pop(); |
405 | 6.73k | } |
406 | | |
407 | 3.80k | void SingleDocParser::ParseAnchor(anchor_t& anchor, std::string& anchor_name) { |
408 | 3.80k | Token& token = m_scanner.peek(); |
409 | 3.80k | if (anchor) |
410 | 7 | throw ParserException(token.mark, ErrorMsg::MULTIPLE_ANCHORS); |
411 | | |
412 | 3.79k | anchor_name = token.value; |
413 | 3.79k | anchor = RegisterAnchor(token.value); |
414 | 3.79k | m_scanner.pop(); |
415 | 3.79k | } |
416 | | |
417 | 3.79k | anchor_t SingleDocParser::RegisterAnchor(const std::string& name) { |
418 | 3.79k | if (name.empty()) |
419 | 0 | return NullAnchor; |
420 | | |
421 | 3.79k | return m_anchors[name] = ++m_curAnchor; |
422 | 3.79k | } |
423 | | |
424 | | anchor_t SingleDocParser::LookupAnchor(const Mark& mark, |
425 | 1.05k | const std::string& name) const { |
426 | 1.05k | auto it = m_anchors.find(name); |
427 | 1.05k | if (it == m_anchors.end()) { |
428 | 245 | std::stringstream ss; |
429 | 245 | ss << ErrorMsg::UNKNOWN_ANCHOR << name; |
430 | 245 | throw ParserException(mark, ss.str()); |
431 | 245 | } |
432 | | |
433 | 812 | return it->second; |
434 | 1.05k | } |
435 | | } // namespace YAML |