blob: bfa07490b9c06f15d3626d33d6006025e49972c1 [file] [log] [blame]
Adam Lesinski6f6ceb72014-11-14 14:48:12 -08001/*
2 * Copyright (C) 2015 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
Adam Lesinskice5e56e2016-10-21 17:56:45 -070017#include <iostream>
18#include <string>
19
Adam Lesinski1ab598f2015-08-14 14:26:04 -070020#include "util/Util.h"
Adam Lesinski467f1712015-11-16 17:35:44 -080021#include "xml/XmlPullParser.h"
22#include "xml/XmlUtil.h"
Adam Lesinski1ab598f2015-08-14 14:26:04 -070023
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070024using ::aapt::io::InputStream;
25using ::android::StringPiece;
Adam Lesinskid5083f62017-01-16 15:07:21 -080026
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080027namespace aapt {
Adam Lesinski467f1712015-11-16 17:35:44 -080028namespace xml {
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080029
30constexpr char kXmlNamespaceSep = 1;
31
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070032XmlPullParser::XmlPullParser(InputStream* in) : in_(in), empty_(), depth_(0) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -070033 parser_ = XML_ParserCreateNS(nullptr, kXmlNamespaceSep);
34 XML_SetUserData(parser_, this);
35 XML_SetElementHandler(parser_, StartElementHandler, EndElementHandler);
36 XML_SetNamespaceDeclHandler(parser_, StartNamespaceHandler,
37 EndNamespaceHandler);
38 XML_SetCharacterDataHandler(parser_, CharacterDataHandler);
39 XML_SetCommentHandler(parser_, CommentDataHandler);
Ryan Mitchellcb76d732018-06-05 10:15:04 -070040 XML_SetCdataSectionHandler(parser_, StartCdataSectionHandler, EndCdataSectionHandler);
Adam Lesinskice5e56e2016-10-21 17:56:45 -070041 event_queue_.push(EventData{Event::kStartDocument, 0, depth_++});
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080042}
43
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070044XmlPullParser::~XmlPullParser() {
45 XML_ParserFree(parser_);
46}
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080047
Adam Lesinskice5e56e2016-10-21 17:56:45 -070048XmlPullParser::Event XmlPullParser::Next() {
49 const Event currentEvent = event();
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070050 if (currentEvent == Event::kBadDocument || currentEvent == Event::kEndDocument) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -070051 return currentEvent;
52 }
53
54 event_queue_.pop();
55 while (event_queue_.empty()) {
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070056 const char* buffer = nullptr;
57 size_t buffer_size = 0;
58 bool done = false;
59 if (!in_->Next(reinterpret_cast<const void**>(&buffer), &buffer_size)) {
60 if (in_->HadError()) {
61 error_ = in_->GetError();
62 event_queue_.push(EventData{Event::kBadDocument});
63 break;
64 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -070065
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070066 done = true;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080067 }
68
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070069 if (XML_Parse(parser_, buffer, buffer_size, done) == XML_STATUS_ERROR) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -070070 error_ = XML_ErrorString(XML_GetErrorCode(parser_));
71 event_queue_.push(EventData{Event::kBadDocument});
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070072 break;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080073 }
74
Adam Lesinskice5e56e2016-10-21 17:56:45 -070075 if (done) {
76 event_queue_.push(EventData{Event::kEndDocument, 0, 0});
Adam Lesinski24aad162015-04-24 19:19:30 -070077 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -070078 }
Adam Lesinski24aad162015-04-24 19:19:30 -070079
Adam Lesinskice5e56e2016-10-21 17:56:45 -070080 Event next_event = event();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080081
Adam Lesinskice5e56e2016-10-21 17:56:45 -070082 // Record namespace prefixes and package names so that we can do our own
83 // handling of references that use namespace aliases.
84 if (next_event == Event::kStartNamespace ||
85 next_event == Event::kEndNamespace) {
Ryan Mitchell4382e442021-07-14 12:53:01 -070086 std::optional<ExtractedPackage> result = ExtractPackageFromNamespace(namespace_uri());
Adam Lesinskice5e56e2016-10-21 17:56:45 -070087 if (next_event == Event::kStartNamespace) {
88 if (result) {
89 package_aliases_.emplace_back(
90 PackageDecl{namespace_prefix(), std::move(result.value())});
91 }
92 } else {
93 if (result) {
94 package_aliases_.pop_back();
95 }
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080096 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -070097 }
98
99 return next_event;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800100}
101
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700102XmlPullParser::Event XmlPullParser::event() const {
103 return event_queue_.front().event;
104}
105
106const std::string& XmlPullParser::error() const { return error_; }
107
108const std::string& XmlPullParser::comment() const {
109 return event_queue_.front().data1;
110}
111
112size_t XmlPullParser::line_number() const {
113 return event_queue_.front().line_number;
114}
115
116size_t XmlPullParser::depth() const { return event_queue_.front().depth; }
117
118const std::string& XmlPullParser::text() const {
119 if (event() != Event::kText) {
120 return empty_;
121 }
122 return event_queue_.front().data1;
123}
124
125const std::string& XmlPullParser::namespace_prefix() const {
126 const Event current_event = event();
127 if (current_event != Event::kStartNamespace &&
128 current_event != Event::kEndNamespace) {
129 return empty_;
130 }
131 return event_queue_.front().data1;
132}
133
134const std::string& XmlPullParser::namespace_uri() const {
135 const Event current_event = event();
136 if (current_event != Event::kStartNamespace &&
137 current_event != Event::kEndNamespace) {
138 return empty_;
139 }
140 return event_queue_.front().data2;
141}
142
Ryan Mitchell4382e442021-07-14 12:53:01 -0700143std::optional<ExtractedPackage> XmlPullParser::TransformPackageAlias(
144 const StringPiece& alias) const {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700145 if (alias.empty()) {
Adam Lesinski1ef0fa92017-08-15 21:32:49 -0700146 return ExtractedPackage{{}, false /*private*/};
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700147 }
148
149 const auto end_iter = package_aliases_.rend();
150 for (auto iter = package_aliases_.rbegin(); iter != end_iter; ++iter) {
151 if (alias == iter->prefix) {
152 if (iter->package.package.empty()) {
Adam Lesinski1ef0fa92017-08-15 21:32:49 -0700153 return ExtractedPackage{{}, iter->package.private_namespace};
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700154 }
155 return iter->package;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800156 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700157 }
158 return {};
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800159}
160
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700161const std::string& XmlPullParser::element_namespace() const {
162 const Event current_event = event();
163 if (current_event != Event::kStartElement &&
164 current_event != Event::kEndElement) {
165 return empty_;
166 }
167 return event_queue_.front().data1;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800168}
169
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700170const std::string& XmlPullParser::element_name() const {
171 const Event current_event = event();
172 if (current_event != Event::kStartElement &&
173 current_event != Event::kEndElement) {
174 return empty_;
175 }
176 return event_queue_.front().data2;
Adam Lesinski24aad162015-04-24 19:19:30 -0700177}
178
Ryan Mitchell326e35ff2021-04-12 07:50:42 -0700179const std::vector<XmlPullParser::PackageDecl>& XmlPullParser::package_decls() const {
180 return package_aliases_;
181}
182
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700183XmlPullParser::const_iterator XmlPullParser::begin_attributes() const {
184 return event_queue_.front().attributes.begin();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800185}
186
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700187XmlPullParser::const_iterator XmlPullParser::end_attributes() const {
188 return event_queue_.front().attributes.end();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800189}
190
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700191size_t XmlPullParser::attribute_count() const {
192 if (event() != Event::kStartElement) {
193 return 0;
194 }
195 return event_queue_.front().attributes.size();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800196}
197
198/**
199 * Extracts the namespace and name of an expanded element or attribute name.
200 */
Adam Lesinskid5083f62017-01-16 15:07:21 -0800201static void SplitName(const char* name, std::string* out_ns, std::string* out_name) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700202 const char* p = name;
203 while (*p != 0 && *p != kXmlNamespaceSep) {
204 p++;
205 }
206
207 if (*p == 0) {
Adam Lesinskid5083f62017-01-16 15:07:21 -0800208 out_ns->clear();
209 out_name->assign(name);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700210 } else {
Adam Lesinskid5083f62017-01-16 15:07:21 -0800211 out_ns->assign(name, (p - name));
212 out_name->assign(p + 1);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700213 }
214}
215
216void XMLCALL XmlPullParser::StartNamespaceHandler(void* user_data,
217 const char* prefix,
218 const char* uri) {
219 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
220 std::string namespace_uri = uri != nullptr ? uri : std::string();
221 parser->namespace_uris_.push(namespace_uri);
222 parser->event_queue_.push(
223 EventData{Event::kStartNamespace,
224 XML_GetCurrentLineNumber(parser->parser_), parser->depth_++,
225 prefix != nullptr ? prefix : std::string(), namespace_uri});
226}
227
228void XMLCALL XmlPullParser::StartElementHandler(void* user_data,
229 const char* name,
230 const char** attrs) {
231 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
232
233 EventData data = {Event::kStartElement,
234 XML_GetCurrentLineNumber(parser->parser_),
235 parser->depth_++};
Adam Lesinskid5083f62017-01-16 15:07:21 -0800236 SplitName(name, &data.data1, &data.data2);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700237
238 while (*attrs) {
239 Attribute attribute;
Adam Lesinskid5083f62017-01-16 15:07:21 -0800240 SplitName(*attrs++, &attribute.namespace_uri, &attribute.name);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700241 attribute.value = *attrs++;
242
243 // Insert in sorted order.
244 auto iter = std::lower_bound(data.attributes.begin(), data.attributes.end(),
245 attribute);
246 data.attributes.insert(iter, std::move(attribute));
247 }
248
249 // Move the structure into the queue (no copy).
250 parser->event_queue_.push(std::move(data));
251}
252
253void XMLCALL XmlPullParser::CharacterDataHandler(void* user_data, const char* s,
254 int len) {
255 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
256
Adam Lesinskid5083f62017-01-16 15:07:21 -0800257 parser->event_queue_.push(EventData{Event::kText, XML_GetCurrentLineNumber(parser->parser_),
258 parser->depth_, std::string(s, len)});
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700259}
260
261void XMLCALL XmlPullParser::EndElementHandler(void* user_data,
262 const char* name) {
263 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
264
265 EventData data = {Event::kEndElement,
266 XML_GetCurrentLineNumber(parser->parser_),
267 --(parser->depth_)};
Adam Lesinskid5083f62017-01-16 15:07:21 -0800268 SplitName(name, &data.data1, &data.data2);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700269
270 // Move the data into the queue (no copy).
271 parser->event_queue_.push(std::move(data));
272}
273
274void XMLCALL XmlPullParser::EndNamespaceHandler(void* user_data,
275 const char* prefix) {
276 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
277
278 parser->event_queue_.push(
279 EventData{Event::kEndNamespace, XML_GetCurrentLineNumber(parser->parser_),
280 --(parser->depth_), prefix != nullptr ? prefix : std::string(),
281 parser->namespace_uris_.top()});
282 parser->namespace_uris_.pop();
283}
284
285void XMLCALL XmlPullParser::CommentDataHandler(void* user_data,
286 const char* comment) {
287 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
288
289 parser->event_queue_.push(EventData{Event::kComment,
290 XML_GetCurrentLineNumber(parser->parser_),
291 parser->depth_, comment});
292}
293
Ryan Mitchellcb76d732018-06-05 10:15:04 -0700294void XMLCALL XmlPullParser::StartCdataSectionHandler(void* user_data) {
295 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
296
297 parser->event_queue_.push(EventData{Event::kCdataStart,
298 XML_GetCurrentLineNumber(parser->parser_),
299 parser->depth_ });
300}
301
302void XMLCALL XmlPullParser::EndCdataSectionHandler(void* user_data) {
303 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
304
305 parser->event_queue_.push(EventData{Event::kCdataEnd,
306 XML_GetCurrentLineNumber(parser->parser_),
307 parser->depth_ });
308}
309
Ryan Mitchell4382e442021-07-14 12:53:01 -0700310std::optional<StringPiece> FindAttribute(const XmlPullParser* parser, const StringPiece& name) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700311 auto iter = parser->FindAttribute("", name);
312 if (iter != parser->end_attributes()) {
313 return StringPiece(util::TrimWhitespace(iter->value));
314 }
315 return {};
316}
317
Ryan Mitchell4382e442021-07-14 12:53:01 -0700318std::optional<StringPiece> FindNonEmptyAttribute(const XmlPullParser* parser,
319 const StringPiece& name) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700320 auto iter = parser->FindAttribute("", name);
321 if (iter != parser->end_attributes()) {
322 StringPiece trimmed = util::TrimWhitespace(iter->value);
323 if (!trimmed.empty()) {
324 return trimmed;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800325 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700326 }
327 return {};
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800328}
329
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700330} // namespace xml
331} // namespace aapt