Rietveld Code Review Tool
Help | Bug tracker | Discussion group | Source code

Side by Side Diff: compiled/RegExpFilter.cpp

Issue 29333474: Issue 4125 - [emscripten] Convert filter classes to C++ (Closed)
Patch Set: Rebased, addressed comments, changed StringMap::find() return value Created Feb. 18, 2016, 4:02 p.m.
Left:
Right:
Use n/p to move between diff chunks; N/P to move between comments.
Jump to:
View unified diff | Download patch
OLDNEW
(Empty)
1 #include <climits>
2
3 #include <emscripten.h>
4
5 #include "RegExpFilter.h"
6 #include "StringScanner.h"
7 #include "StringMap.h"
8
9 namespace
10 {
11 enum
12 {
13 TYPE_OTHER = 0x1,
14 TYPE_SCRIPT = 0x2,
15 TYPE_IMAGE = 0x4,
16 TYPE_STYLESHEET = 0x8,
17 TYPE_OBJECT = 0x10,
18 TYPE_SUBDOCUMENT = 0x20,
19 TYPE_DOCUMENT = 0x40,
20 TYPE_PING = 0x400,
21 TYPE_XMLHTTPREQUEST = 0x800,
22 TYPE_OBJECT_SUBREQUEST = 0x1000,
23 TYPE_MEDIA = 0x4000,
24 TYPE_FONT = 0x8000,
25 TYPE_POPUP = 0x8000000,
26 TYPE_GENERICBLOCK = 0x10000000,
27 TYPE_GENERICHIDE = 0x20000000,
28 TYPE_ELEMHIDE = 0x40000000,
29 };
30
31 StringMap<int> typeMap {
32 {u"other"_str, TYPE_OTHER},
33 {u"script"_str, TYPE_SCRIPT},
34 {u"image"_str, TYPE_IMAGE},
35 {u"stylesheet"_str, TYPE_STYLESHEET},
36 {u"object"_str, TYPE_OBJECT},
37 {u"subdocument"_str, TYPE_SUBDOCUMENT},
38 {u"document"_str, TYPE_DOCUMENT},
39 {u"xbl"_str, TYPE_OTHER}, // Backwards compat
40 {u"ping"_str, TYPE_PING},
41 {u"xmlhttprequest"_str, TYPE_XMLHTTPREQUEST},
42 {u"object-subrequest"_str, TYPE_OBJECT_SUBREQUEST},
43 {u"dtd"_str, TYPE_OTHER}, // Backwards compat
44 {u"media"_str, TYPE_MEDIA},
45 {u"font"_str, TYPE_FONT},
46 {u"background"_str, TYPE_IMAGE}, // Backwards compat
47
48 {u"popup"_str, TYPE_POPUP},
49 {u"genericblock"_str, TYPE_GENERICBLOCK},
50 {u"generichide"_str, TYPE_GENERICHIDE},
51 {u"elemhide"_str, TYPE_ELEMHIDE},
52 };
53
54 const int defaultTypeMask = INT_MAX & ~(TYPE_DOCUMENT | TYPE_ELEMHIDE | TYPE_P OPUP |
55 TYPE_GENERICBLOCK | TYPE_GENERICHIDE);
56
57 int GenerateRegExp(const String& regexp, bool matchCase)
58 {
59 return EM_ASM_INT(return regexps.create($0, $1), &regexp, matchCase);
60 }
61
62 void NormalizeWhitespace(DependentString& text)
63 {
64 // We want to remove all spaces but bail out early in the common scenario
65 // that the string contains no spaces.
66
67 // Look for the first space
68 String::size_type len = text.length();
69 String::size_type pos;
70 for (pos = 0; pos < len; pos++)
71 if (text[pos] == ' ')
72 break;
73
74 if (pos >= len)
75 return;
76
77 // Found spaces, move characters to remove them
78 String::size_type delta = 1;
79 for (pos = pos + 1; pos < len; pos++)
80 {
81 if (text[pos] == ' ')
82 delta++;
83 else
84 text[pos - delta] = text[pos];
85 }
86 text.reset(text, 0, len - delta);
87 }
88
89 void ParseOption(String& text, DependentString& error, RegExpFilterData& data,
90 int optionStart, int optionEnd, int valueStart, int valueEnd)
91 {
92 if (optionEnd <= optionStart)
93 return;
94
95 bool reverse = false;
96 if (text[optionStart] == u'~')
97 {
98 reverse = true;
99 optionStart++;
100 }
101
102 DependentString name(text, optionStart, optionEnd - optionStart);
103 for (size_t i = 0; i < name.length(); ++i)
104 {
105 char16_t currChar = name[i];
106 if (currChar >= u'A' && currChar <= u'Z')
107 name[i] = currChar + u'a' - u'A';
108 else if (currChar == u'_')
109 name[i] = u'-';
110 }
111
112 auto it = typeMap.find(name);
113 if (it)
114 {
115 if (data.mContentType < 0)
116 data.mContentType = reverse ? defaultTypeMask : 0;
117 if (reverse)
118 data.mContentType &= ~it->second;
119 else
120 data.mContentType |= it->second;
121 }
122 else if (name.equals(u"domain"_str))
123 {
124 if (valueStart >= 0 && valueEnd > valueStart)
125 {
126 data.mDomainsStart = valueStart;
127 data.mDomainsEnd = valueEnd;
128 ActiveFilter::ToLower(text, data.mDomainsStart, data.mDomainsEnd);
129 }
130 }
131 else if (name.equals(u"sitekey"_str))
132 {
133 if (valueStart >= 0 && valueEnd > valueStart)
134 {
135 data.mSitekeysStart = valueStart;
136 data.mSitekeysEnd = valueEnd;
137 }
138 }
139 else if (name.equals(u"match-case"_str))
140 data.mMatchCase = !reverse;
141 else if (name.equals(u"third-party"_str))
142 data.mThirdParty = reverse ? TrippleState::NO : TrippleState::YES;
143 else if (name.equals(u"collapse"_str))
144 data.mCollapse = reverse ? TrippleState::NO : TrippleState::YES;
145 else
146 error.reset(u"filter_unknown_option"_str);
147 }
148
149 void ParseOptions(String& text, DependentString& error, RegExpFilterData& data ,
150 String::size_type optionsStart)
151 {
152 data.mMatchCase = false;
153 data.mThirdParty = TrippleState::ANY;
154 data.mCollapse = TrippleState::ANY;
155 data.mDomainsStart = String::npos;
156 data.mSitekeysStart = String::npos;
157 if (optionsStart >= text.length())
158 {
159 data.mContentType = defaultTypeMask;
160 return;
161 }
162
163 data.mContentType = -1;
164
165 int optionStart = data.mPatternEnd + 1;
166 int optionEnd = -1;
167 int valueStart = -1;
168
169 StringScanner scanner(text, optionStart, u',');
170 bool done = false;
171 while (!done)
172 {
173 done = scanner.done();
174 switch (scanner.next())
175 {
176 case u'=':
177 if (optionEnd < 0)
178 {
179 optionEnd = scanner.position();
180 valueStart = optionEnd + 1;
181 }
182 break;
183 case u',':
184 if (optionEnd < 0)
185 optionEnd = scanner.position();
186 ParseOption(text, error, data, optionStart, optionEnd, valueStart,
187 scanner.position());
188 if (!error.empty())
189 return;
190
191 optionStart = scanner.position() + 1;
192 optionEnd = -1;
193 valueStart = -1;
194 break;
195 }
196 }
197
198 if (data.mContentType < 0)
199 data.mContentType = defaultTypeMask;
200 }
201 }
202
203 RegExpFilter::RegExpFilter(const String& text, const RegExpFilterData& data)
204 : ActiveFilter(text, true), RegExpFilterData(data)
205 {
206 }
207
208 RegExpFilter::~RegExpFilter()
209 {
210 if (HasRegExp())
211 EM_ASM_ARGS(regexps.delete($0), mRegexpId);
212 }
213
214 Filter::Type RegExpFilter::Parse(DependentString& text, DependentString& error,
215 RegExpFilterData& data)
216 {
217 NormalizeWhitespace(text);
218
219 bool blocking = true;
220
221 data.mPatternStart = 0;
222 if (text.length() >= 2 && text[0] == u'@' && text[1] == u'@')
223 {
224 blocking = false;
225 data.mPatternStart = 2;
226 }
227
228 data.mPatternEnd = text.find(u'$', data.mPatternStart);
229 if (data.mPatternEnd == text.npos)
230 data.mPatternEnd = text.length();
231
232 ParseOptions(text, error, data, data.mPatternEnd + 1);
233 if (!error.empty())
234 return Type::INVALID;
235
236 if (data.mPatternEnd - data.mPatternStart >= 2 &&
237 text[data.mPatternStart] == u'/' &&
238 text[data.mPatternEnd - 1] == u'/')
239 {
240 data.SetRegExp(GenerateRegExp(DependentString(text, data.mPatternStart + 1,
241 data.mPatternEnd - data.mPatternStart - 2), data.mMatchCase));
242 if (data.mRegexpId == -1)
243 {
244 error.reset(u"filter_invalid_regexp"_str);
245 return Type::INVALID;
246 }
247 }
248
249 if (blocking)
250 return Type::BLOCKING;
251 else
252 return Type::WHITELIST;
253 }
254
255 void RegExpFilter::ParseSitekeys(const String& sitekeys) const
256 {
257 StringScanner scanner(sitekeys, 0, u'|');
258 size_t start = 0;
259 bool done = false;
260 while (!done)
261 {
262 done = scanner.done();
263 if (scanner.next() == u'|')
264 {
265 if (scanner.position() > start)
266 AddSitekey(DependentString(sitekeys, start, scanner.position() - start)) ;
267 start = scanner.position() + 1;
268 }
269 }
270 }
271
272 void RegExpFilter::InitJSTypes()
273 {
274 EM_ASM(exports.RegExpFilter.typeMap = {};);
275 for (auto it = typeMap.begin(); it != typeMap.end(); ++it)
276 EM_ASM_ARGS(exports.RegExpFilter.typeMap[getStringData($0).replace("-", "_") .toUpperCase()] = $1, &(it->first), it->second);
277 }
278
279 OwnedString RegExpFilter::RegExpFromSource(const String& source)
280 {
281 /* TODO: this is very inefficient */
282
283 // Note: This doesn't remove trailing wildcards, otherwise the result should
284 // be identical to Filter.toRegExp().
285 OwnedString result;
286 String::value_type prevChar = u'*';
287 for (String::size_type i = 0; i < source.length(); ++i)
288 {
289 String::value_type currChar = source[i];
290 switch (currChar)
291 {
292 case u'*':
293 if (prevChar != u'*')
294 result.append(u".*"_str);
295 break;
296 case u'^':
297 result.append(u"(?:[\\x00-\\x24\\x26-\\x2C\\x2F\\x3A-\\x40\\x5B-\\x5E\\x 60\\x7B-\\x7F]|$)"_str);
298 break;
299 case u'|':
300 if (i == 0)
301 {
302 // Anchor at expression start, maybe extended anchor?
303 if (i + 1 < source.length() && source[i + 1] == u'|')
304 {
305 result.append(u"^[\\w\\-]+:\\/+(?!\\/)(?:[^\\/]+\\.)?"_str);
306 ++i;
307 }
308 else
309 result.append(u'^');
310 }
311 else if (i == source.length() - 1)
312 {
313 // Anchor at expression end, ignore if following separator placeholder
314 if (prevChar != u'^')
315 result.append(u'$');
316 }
317 else
318 {
319 // Not actually an anchor, escape it
320 result.append(u"\\|"_str);
321 }
322 break;
323 default:
324 if (!(currChar >= u'a' && currChar <= u'z') &&
325 !(currChar >= u'A' && currChar <= u'Z') &&
326 !(currChar >= u'0' && currChar <= u'9') &&
327 currChar < 128)
328 {
329 result.append(u'\\');
330 }
331 result.append(currChar);
332 }
333 prevChar = currChar;
334 }
335 return result;
336 }
337
338 RegExpFilter::DomainMap* RegExpFilter::GetDomains() const
339 {
340 if (!DomainsParsingDone())
341 {
342 ParseDomains(GetDomainsSource(mText), u'|');
343 SetDomainsParsingDone();
344 }
345 return ActiveFilter::GetDomains();
346 }
347
348 RegExpFilter::SitekeySet* RegExpFilter::GetSitekeys() const
349 {
350 if (!SitekeyParsingDone())
351 {
352 ParseSitekeys(GetSitekeysSource(mText));
353 SetSitekeysParsingDone();
354 }
355 return ActiveFilter::GetSitekeys();
356 }
357
358 bool RegExpFilter::Matches(const String& location, int typeMask,
359 DependentString& docDomain, bool thirdParty, const String& sitekey) const
360 {
361 if (!(mContentType & typeMask) ||
362 (mThirdParty == TrippleState::YES && !thirdParty) ||
363 (mThirdParty == TrippleState::NO && thirdParty) ||
364 !IsActiveOnDomain(docDomain, sitekey))
365 {
366 return false;
367 }
368
369 if (!RegExpParsingDone())
370 {
371 const OwnedString pattern(GetRegExpSource(mText));
372 SetRegExp(GenerateRegExp(RegExpFromSource(pattern), mMatchCase));
373 }
374 return EM_ASM_INT(return regexps.test($0, $1), mRegexpId, &location);
375 }
OLDNEW

Powered by Google App Engine
This is Rietveld