OLD | NEW |
---|---|
1 // Copyright 2014 The Chromium Authors. All rights reserved. | 1 // Copyright 2014 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 #include "net/filter/filter.h" | 5 #include "net/filter/filter.h" |
6 | 6 |
7 #include "base/files/file_path.h" | 7 #include "base/files/file_path.h" |
8 #include "base/strings/string_util.h" | 8 #include "base/strings/string_util.h" |
9 #include "net/base/filename_util_unsafe.h" | 9 #include "net/base/filename_util_unsafe.h" |
10 #include "net/base/io_buffer.h" | 10 #include "net/base/io_buffer.h" |
11 #include "net/base/mime_util.h" | 11 #include "net/base/mime_util.h" |
12 #include "net/base/sdch_net_log_params.h" | |
12 #include "net/filter/gzip_filter.h" | 13 #include "net/filter/gzip_filter.h" |
13 #include "net/filter/sdch_filter.h" | 14 #include "net/filter/sdch_filter.h" |
14 #include "net/url_request/url_request_context.h" | 15 #include "net/url_request/url_request_context.h" |
15 #include "url/gurl.h" | 16 #include "url/gurl.h" |
16 | 17 |
18 namespace net { | |
19 | |
17 namespace { | 20 namespace { |
18 | 21 |
19 // Filter types (using canonical lower case only): | 22 // Filter types (using canonical lower case only): |
20 const char kDeflate[] = "deflate"; | 23 const char kDeflate[] = "deflate"; |
21 const char kGZip[] = "gzip"; | 24 const char kGZip[] = "gzip"; |
22 const char kXGZip[] = "x-gzip"; | 25 const char kXGZip[] = "x-gzip"; |
23 const char kSdch[] = "sdch"; | 26 const char kSdch[] = "sdch"; |
24 // compress and x-compress are currently not supported. If we decide to support | 27 // compress and x-compress are currently not supported. If we decide to support |
25 // them, we'll need the same mime type compatibility hack we have for gzip. For | 28 // them, we'll need the same mime type compatibility hack we have for gzip. For |
26 // more information, see Firefox's nsHttpChannel::ProcessNormal. | 29 // more information, see Firefox's nsHttpChannel::ProcessNormal. |
27 | 30 |
28 // Mime types: | 31 // Mime types: |
29 const char kApplicationXGzip[] = "application/x-gzip"; | 32 const char kApplicationXGzip[] = "application/x-gzip"; |
30 const char kApplicationGzip[] = "application/gzip"; | 33 const char kApplicationGzip[] = "application/gzip"; |
31 const char kApplicationXGunzip[] = "application/x-gunzip"; | 34 const char kApplicationXGunzip[] = "application/x-gunzip"; |
32 const char kTextHtml[] = "text/html"; | 35 const char kTextHtml[] = "text/html"; |
33 | 36 |
34 // Buffer size allocated when de-compressing data. | 37 // Buffer size allocated when de-compressing data. |
35 const int kFilterBufSize = 32 * 1024; | 38 const int kFilterBufSize = 32 * 1024; |
36 | 39 |
40 void LogSdchProblem(const FilterContext& filter_context, | |
41 SdchManager::ProblemCodes problem) { | |
42 SdchManager::SdchErrorRecovery(problem); | |
43 filter_context.GetNetLog().AddEvent( | |
44 NetLog::TYPE_SDCH_RESOURCE_ERROR, | |
45 base::Bind(&NetLogSdchResourceProblemCallback, problem)); | |
46 } | |
47 | |
37 } // namespace | 48 } // namespace |
38 | 49 |
39 namespace net { | |
40 | |
41 FilterContext::~FilterContext() { | 50 FilterContext::~FilterContext() { |
42 } | 51 } |
43 | 52 |
44 Filter::~Filter() {} | 53 Filter::~Filter() {} |
45 | 54 |
46 // static | 55 // static |
47 Filter* Filter::Factory(const std::vector<FilterType>& filter_types, | 56 Filter* Filter::Factory(const std::vector<FilterType>& filter_types, |
48 const FilterContext& filter_context) { | 57 const FilterContext& filter_context) { |
49 if (filter_types.empty()) | 58 if (filter_types.empty()) |
50 return NULL; | 59 return NULL; |
(...skipping 103 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
154 } | 163 } |
155 | 164 |
156 // static | 165 // static |
157 void Filter::FixupEncodingTypes( | 166 void Filter::FixupEncodingTypes( |
158 const FilterContext& filter_context, | 167 const FilterContext& filter_context, |
159 std::vector<FilterType>* encoding_types) { | 168 std::vector<FilterType>* encoding_types) { |
160 std::string mime_type; | 169 std::string mime_type; |
161 bool success = filter_context.GetMimeType(&mime_type); | 170 bool success = filter_context.GetMimeType(&mime_type); |
162 DCHECK(success || mime_type.empty()); | 171 DCHECK(success || mime_type.empty()); |
163 | 172 |
173 GURL url; | |
174 success = filter_context.GetURL(&url); | |
Randy Smith (Not in Mondays)
2014/08/13 17:35:46
Why the change?
baranovich
2014/08/13 19:13:48
Done.
| |
175 | |
164 if ((1 == encoding_types->size()) && | 176 if ((1 == encoding_types->size()) && |
165 (FILTER_TYPE_GZIP == encoding_types->front())) { | 177 (FILTER_TYPE_GZIP == encoding_types->front())) { |
166 if (LowerCaseEqualsASCII(mime_type, kApplicationXGzip) || | 178 if (LowerCaseEqualsASCII(mime_type, kApplicationXGzip) || |
167 LowerCaseEqualsASCII(mime_type, kApplicationGzip) || | 179 LowerCaseEqualsASCII(mime_type, kApplicationGzip) || |
168 LowerCaseEqualsASCII(mime_type, kApplicationXGunzip)) | 180 LowerCaseEqualsASCII(mime_type, kApplicationXGunzip)) |
169 // The server has told us that it sent us gziped content with a gzip | 181 // The server has told us that it sent us gziped content with a gzip |
170 // content encoding. Sadly, Apache mistakenly sets these headers for all | 182 // content encoding. Sadly, Apache mistakenly sets these headers for all |
171 // .gz files. We match Firefox's nsHttpChannel::ProcessNormal and ignore | 183 // .gz files. We match Firefox's nsHttpChannel::ProcessNormal and ignore |
172 // the Content-Encoding here. | 184 // the Content-Encoding here. |
173 encoding_types->clear(); | 185 encoding_types->clear(); |
174 | 186 |
175 GURL url; | 187 DCHECK(success); |
176 std::string disposition; | 188 std::string disposition; |
177 success = filter_context.GetURL(&url); | |
178 DCHECK(success); | |
179 filter_context.GetContentDisposition(&disposition); | 189 filter_context.GetContentDisposition(&disposition); |
180 // Don't supply a MIME type here, since that may cause disk IO. | 190 // Don't supply a MIME type here, since that may cause disk IO. |
181 base::FilePath::StringType extension = | 191 base::FilePath::StringType extension = |
182 GenerateFileExtensionUnsafe(url, disposition, "UTF-8", "", "", ""); | 192 GenerateFileExtensionUnsafe(url, disposition, "UTF-8", "", "", ""); |
183 | 193 |
184 if (filter_context.IsDownload()) { | 194 if (filter_context.IsDownload()) { |
185 // We don't want to decompress gzipped files when the user explicitly | 195 // We don't want to decompress gzipped files when the user explicitly |
186 // asks to download them. | 196 // asks to download them. |
187 // For the case of svgz files, we use the extension to distinguish | 197 // For the case of svgz files, we use the extension to distinguish |
188 // between svgz files and svg files compressed with gzip by the server. | 198 // between svgz files and svg files compressed with gzip by the server. |
(...skipping 14 matching lines...) Expand all Loading... | |
203 !IsSupportedMimeType(mime_type)) | 213 !IsSupportedMimeType(mime_type)) |
204 encoding_types->clear(); | 214 encoding_types->clear(); |
205 } | 215 } |
206 } | 216 } |
207 | 217 |
208 // If the request was for SDCH content, then we might need additional fixups. | 218 // If the request was for SDCH content, then we might need additional fixups. |
209 if (!filter_context.IsSdchResponse()) { | 219 if (!filter_context.IsSdchResponse()) { |
210 // It was not an SDCH request, so we'll just record stats. | 220 // It was not an SDCH request, so we'll just record stats. |
211 if (1 < encoding_types->size()) { | 221 if (1 < encoding_types->size()) { |
212 // Multiple filters were intended to only be used for SDCH (thus far!) | 222 // Multiple filters were intended to only be used for SDCH (thus far!) |
213 SdchManager::SdchErrorRecovery( | 223 LogSdchProblem(filter_context, |
214 SdchManager::MULTIENCODING_FOR_NON_SDCH_REQUEST); | 224 SdchManager::MULTIENCODING_FOR_NON_SDCH_REQUEST); |
215 } | 225 } |
216 if ((1 == encoding_types->size()) && | 226 if ((1 == encoding_types->size()) && |
217 (FILTER_TYPE_SDCH == encoding_types->front())) { | 227 (FILTER_TYPE_SDCH == encoding_types->front())) { |
218 SdchManager::SdchErrorRecovery( | 228 LogSdchProblem(filter_context, |
219 SdchManager::SDCH_CONTENT_ENCODE_FOR_NON_SDCH_REQUEST); | 229 SdchManager::SDCH_CONTENT_ENCODE_FOR_NON_SDCH_REQUEST); |
220 } | 230 } |
221 return; | 231 return; |
222 } | 232 } |
223 | 233 |
224 // The request was tagged as an SDCH request, which means the server supplied | 234 // The request was tagged as an SDCH request, which means the server supplied |
225 // a dictionary, and we advertised it in the request. Some proxies will do | 235 // a dictionary, and we advertised it in the request. Some proxies will do |
226 // very strange things to the request, or the response, so we have to handle | 236 // very strange things to the request, or the response, so we have to handle |
227 // them gracefully. | 237 // them gracefully. |
228 | 238 |
229 // If content encoding included SDCH, then everything is "relatively" fine. | 239 // If content encoding included SDCH, then everything is "relatively" fine. |
230 if (!encoding_types->empty() && | 240 if (!encoding_types->empty() && |
231 (FILTER_TYPE_SDCH == encoding_types->front())) { | 241 (FILTER_TYPE_SDCH == encoding_types->front())) { |
232 // Some proxies (found currently in Argentina) strip the Content-Encoding | 242 // Some proxies (found currently in Argentina) strip the Content-Encoding |
233 // text from "sdch,gzip" to a mere "sdch" without modifying the compressed | 243 // text from "sdch,gzip" to a mere "sdch" without modifying the compressed |
234 // payload. To handle this gracefully, we simulate the "probably" deleted | 244 // payload. To handle this gracefully, we simulate the "probably" deleted |
235 // ",gzip" by appending a tentative gzip decode, which will default to a | 245 // ",gzip" by appending a tentative gzip decode, which will default to a |
236 // no-op pass through filter if it doesn't get gzip headers where expected. | 246 // no-op pass through filter if it doesn't get gzip headers where expected. |
237 if (1 == encoding_types->size()) { | 247 if (1 == encoding_types->size()) { |
238 encoding_types->push_back(FILTER_TYPE_GZIP_HELPING_SDCH); | 248 encoding_types->push_back(FILTER_TYPE_GZIP_HELPING_SDCH); |
239 SdchManager::SdchErrorRecovery( | 249 LogSdchProblem(filter_context, |
240 SdchManager::OPTIONAL_GUNZIP_ENCODING_ADDED); | 250 SdchManager::OPTIONAL_GUNZIP_ENCODING_ADDED); |
241 } | 251 } |
242 return; | 252 return; |
243 } | 253 } |
244 | 254 |
245 // There are now several cases to handle for an SDCH request. Foremost, if | 255 // There are now several cases to handle for an SDCH request. Foremost, if |
246 // the outbound request was stripped so as not to advertise support for | 256 // the outbound request was stripped so as not to advertise support for |
247 // encodings, we might get back content with no encoding, or (for example) | 257 // encodings, we might get back content with no encoding, or (for example) |
248 // just gzip. We have to be sure that any changes we make allow for such | 258 // just gzip. We have to be sure that any changes we make allow for such |
249 // minimal coding to work. That issue is why we use TENTATIVE filters if we | 259 // minimal coding to work. That issue is why we use TENTATIVE filters if we |
250 // add any, as those filters sniff the content, and act as pass-through | 260 // add any, as those filters sniff the content, and act as pass-through |
(...skipping 13 matching lines...) Expand all Loading... | |
264 // The one unresolved failure mode comes when we advertise a dictionary, and | 274 // The one unresolved failure mode comes when we advertise a dictionary, and |
265 // the server tries to *send* a gzipped file (not gzip encode content), and | 275 // the server tries to *send* a gzipped file (not gzip encode content), and |
266 // then we could do a gzip decode :-(. Since SDCH is only (currently) | 276 // then we could do a gzip decode :-(. Since SDCH is only (currently) |
267 // supported server side on paths that only send HTML content, this mode has | 277 // supported server side on paths that only send HTML content, this mode has |
268 // never surfaced in the wild (and is unlikely to). | 278 // never surfaced in the wild (and is unlikely to). |
269 // We will gather a lot of stats as we perform the fixups | 279 // We will gather a lot of stats as we perform the fixups |
270 if (StartsWithASCII(mime_type, kTextHtml, false)) { | 280 if (StartsWithASCII(mime_type, kTextHtml, false)) { |
271 // Suspicious case: Advertised dictionary, but server didn't use sdch, and | 281 // Suspicious case: Advertised dictionary, but server didn't use sdch, and |
272 // we're HTML tagged. | 282 // we're HTML tagged. |
273 if (encoding_types->empty()) { | 283 if (encoding_types->empty()) { |
274 SdchManager::SdchErrorRecovery( | 284 LogSdchProblem(filter_context, SdchManager::ADDED_CONTENT_ENCODING); |
275 SdchManager::ADDED_CONTENT_ENCODING); | |
276 } else if (1 == encoding_types->size()) { | 285 } else if (1 == encoding_types->size()) { |
277 SdchManager::SdchErrorRecovery( | 286 LogSdchProblem(filter_context, SdchManager::FIXED_CONTENT_ENCODING); |
278 SdchManager::FIXED_CONTENT_ENCODING); | |
279 } else { | 287 } else { |
280 SdchManager::SdchErrorRecovery( | 288 LogSdchProblem(filter_context, SdchManager::FIXED_CONTENT_ENCODINGS); |
281 SdchManager::FIXED_CONTENT_ENCODINGS); | |
282 } | 289 } |
283 } else { | 290 } else { |
284 // Remarkable case!?! We advertised an SDCH dictionary, content-encoding | 291 // Remarkable case!?! We advertised an SDCH dictionary, content-encoding |
285 // was not marked for SDCH processing: Why did the server suggest an SDCH | 292 // was not marked for SDCH processing: Why did the server suggest an SDCH |
286 // dictionary in the first place??. Also, the content isn't | 293 // dictionary in the first place??. Also, the content isn't |
287 // tagged as HTML, despite the fact that SDCH encoding is mostly likely for | 294 // tagged as HTML, despite the fact that SDCH encoding is mostly likely for |
288 // HTML: Did some anti-virus system strip this tag (sometimes they strip | 295 // HTML: Did some anti-virus system strip this tag (sometimes they strip |
289 // accept-encoding headers on the request)?? Does the content encoding not | 296 // accept-encoding headers on the request)?? Does the content encoding not |
290 // start with "text/html" for some other reason?? We'll report this as a | 297 // start with "text/html" for some other reason?? We'll report this as a |
291 // fixup to a binary file, but it probably really is text/html (some how). | 298 // fixup to a binary file, but it probably really is text/html (some how). |
292 if (encoding_types->empty()) { | 299 if (encoding_types->empty()) { |
293 SdchManager::SdchErrorRecovery( | 300 LogSdchProblem(filter_context, |
294 SdchManager::BINARY_ADDED_CONTENT_ENCODING); | 301 SdchManager::BINARY_ADDED_CONTENT_ENCODING); |
295 } else if (1 == encoding_types->size()) { | 302 } else if (1 == encoding_types->size()) { |
296 SdchManager::SdchErrorRecovery( | 303 LogSdchProblem(filter_context, |
297 SdchManager::BINARY_FIXED_CONTENT_ENCODING); | 304 SdchManager::BINARY_FIXED_CONTENT_ENCODING); |
298 } else { | 305 } else { |
299 SdchManager::SdchErrorRecovery( | 306 LogSdchProblem(filter_context, |
300 SdchManager::BINARY_FIXED_CONTENT_ENCODINGS); | 307 SdchManager::BINARY_FIXED_CONTENT_ENCODINGS); |
301 } | 308 } |
302 } | 309 } |
303 | 310 |
304 // Leave the existing encoding type to be processed first, and add our | 311 // Leave the existing encoding type to be processed first, and add our |
305 // tentative decodings to be done afterwards. Vodaphone UK reportedyl will | 312 // tentative decodings to be done afterwards. Vodaphone UK reportedyl will |
306 // perform a second layer of gzip encoding atop the server's sdch,gzip | 313 // perform a second layer of gzip encoding atop the server's sdch,gzip |
307 // encoding, and then claim that the content encoding is a mere gzip. As a | 314 // encoding, and then claim that the content encoding is a mere gzip. As a |
308 // result we'll need (in that case) to do the gunzip, plus our tentative | 315 // result we'll need (in that case) to do the gunzip, plus our tentative |
309 // gunzip and tentative SDCH decoding. | 316 // gunzip and tentative SDCH decoding. |
310 // This approach nicely handles the empty() list as well, and should work with | 317 // This approach nicely handles the empty() list as well, and should work with |
(...skipping 89 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... | |
400 | 407 |
401 void Filter::PushDataIntoNextFilter() { | 408 void Filter::PushDataIntoNextFilter() { |
402 IOBuffer* next_buffer = next_filter_->stream_buffer(); | 409 IOBuffer* next_buffer = next_filter_->stream_buffer(); |
403 int next_size = next_filter_->stream_buffer_size(); | 410 int next_size = next_filter_->stream_buffer_size(); |
404 last_status_ = ReadFilteredData(next_buffer->data(), &next_size); | 411 last_status_ = ReadFilteredData(next_buffer->data(), &next_size); |
405 if (FILTER_ERROR != last_status_) | 412 if (FILTER_ERROR != last_status_) |
406 next_filter_->FlushStreamBuffer(next_size); | 413 next_filter_->FlushStreamBuffer(next_size); |
407 } | 414 } |
408 | 415 |
409 } // namespace net | 416 } // namespace net |
OLD | NEW |