blob: 1527d68a6c3b7d759595f6671168ccf6f306ced3 [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
Yurii Zubrytskyi1d6d8ac2024-07-03 16:10:17 -070017#include "xml/XmlPullParser.h"
18
19#include <algorithm>
Adam Lesinskice5e56e2016-10-21 17:56:45 -070020#include <string>
Yurii Zubrytskyi1d6d8ac2024-07-03 16:10:17 -070021#include <tuple>
Adam Lesinskice5e56e2016-10-21 17:56:45 -070022
Adam Lesinski1ab598f2015-08-14 14:26:04 -070023#include "util/Util.h"
Adam Lesinski467f1712015-11-16 17:35:44 -080024#include "xml/XmlUtil.h"
Adam Lesinski1ab598f2015-08-14 14:26:04 -070025
Jeremy Meyerb4f83ff2023-11-30 19:29:50 +000026using ::android::InputStream;
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070027using ::android::StringPiece;
Adam Lesinskid5083f62017-01-16 15:07:21 -080028
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080029namespace aapt {
Adam Lesinski467f1712015-11-16 17:35:44 -080030namespace xml {
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080031
32constexpr char kXmlNamespaceSep = 1;
33
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070034XmlPullParser::XmlPullParser(InputStream* in) : in_(in), empty_(), depth_(0) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -070035 parser_ = XML_ParserCreateNS(nullptr, kXmlNamespaceSep);
36 XML_SetUserData(parser_, this);
37 XML_SetElementHandler(parser_, StartElementHandler, EndElementHandler);
38 XML_SetNamespaceDeclHandler(parser_, StartNamespaceHandler,
39 EndNamespaceHandler);
40 XML_SetCharacterDataHandler(parser_, CharacterDataHandler);
41 XML_SetCommentHandler(parser_, CommentDataHandler);
Ryan Mitchellcb76d732018-06-05 10:15:04 -070042 XML_SetCdataSectionHandler(parser_, StartCdataSectionHandler, EndCdataSectionHandler);
Adam Lesinskice5e56e2016-10-21 17:56:45 -070043 event_queue_.push(EventData{Event::kStartDocument, 0, depth_++});
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080044}
45
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070046XmlPullParser::~XmlPullParser() {
47 XML_ParserFree(parser_);
48}
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080049
Adam Lesinskice5e56e2016-10-21 17:56:45 -070050XmlPullParser::Event XmlPullParser::Next() {
51 const Event currentEvent = event();
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070052 if (currentEvent == Event::kBadDocument || currentEvent == Event::kEndDocument) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -070053 return currentEvent;
54 }
55
56 event_queue_.pop();
57 while (event_queue_.empty()) {
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070058 const char* buffer = nullptr;
59 size_t buffer_size = 0;
60 bool done = false;
61 if (!in_->Next(reinterpret_cast<const void**>(&buffer), &buffer_size)) {
62 if (in_->HadError()) {
63 error_ = in_->GetError();
64 event_queue_.push(EventData{Event::kBadDocument});
65 break;
66 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -070067
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070068 done = true;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080069 }
70
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070071 if (XML_Parse(parser_, buffer, buffer_size, done) == XML_STATUS_ERROR) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -070072 error_ = XML_ErrorString(XML_GetErrorCode(parser_));
73 event_queue_.push(EventData{Event::kBadDocument});
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070074 break;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080075 }
76
Adam Lesinskice5e56e2016-10-21 17:56:45 -070077 if (done) {
78 event_queue_.push(EventData{Event::kEndDocument, 0, 0});
Adam Lesinski24aad162015-04-24 19:19:30 -070079 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -070080 }
Adam Lesinski24aad162015-04-24 19:19:30 -070081
Adam Lesinskice5e56e2016-10-21 17:56:45 -070082 Event next_event = event();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080083
Adam Lesinskice5e56e2016-10-21 17:56:45 -070084 // Record namespace prefixes and package names so that we can do our own
85 // handling of references that use namespace aliases.
86 if (next_event == Event::kStartNamespace ||
87 next_event == Event::kEndNamespace) {
Ryan Mitchell4382e442021-07-14 12:53:01 -070088 std::optional<ExtractedPackage> result = ExtractPackageFromNamespace(namespace_uri());
Adam Lesinskice5e56e2016-10-21 17:56:45 -070089 if (next_event == Event::kStartNamespace) {
90 if (result) {
91 package_aliases_.emplace_back(
92 PackageDecl{namespace_prefix(), std::move(result.value())});
93 }
94 } else {
95 if (result) {
96 package_aliases_.pop_back();
97 }
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080098 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -070099 }
100
101 return next_event;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800102}
103
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700104XmlPullParser::Event XmlPullParser::event() const {
105 return event_queue_.front().event;
106}
107
108const std::string& XmlPullParser::error() const { return error_; }
109
110const std::string& XmlPullParser::comment() const {
111 return event_queue_.front().data1;
112}
113
114size_t XmlPullParser::line_number() const {
115 return event_queue_.front().line_number;
116}
117
118size_t XmlPullParser::depth() const { return event_queue_.front().depth; }
119
120const std::string& XmlPullParser::text() const {
121 if (event() != Event::kText) {
122 return empty_;
123 }
124 return event_queue_.front().data1;
125}
126
127const std::string& XmlPullParser::namespace_prefix() const {
128 const Event current_event = event();
129 if (current_event != Event::kStartNamespace &&
130 current_event != Event::kEndNamespace) {
131 return empty_;
132 }
133 return event_queue_.front().data1;
134}
135
136const std::string& XmlPullParser::namespace_uri() const {
137 const Event current_event = event();
138 if (current_event != Event::kStartNamespace &&
139 current_event != Event::kEndNamespace) {
140 return empty_;
141 }
142 return event_queue_.front().data2;
143}
144
Yurii Zubrytskyia5775142022-11-02 17:49:49 -0700145std::optional<ExtractedPackage> XmlPullParser::TransformPackageAlias(StringPiece alias) const {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700146 if (alias.empty()) {
Adam Lesinski1ef0fa92017-08-15 21:32:49 -0700147 return ExtractedPackage{{}, false /*private*/};
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700148 }
149
150 const auto end_iter = package_aliases_.rend();
151 for (auto iter = package_aliases_.rbegin(); iter != end_iter; ++iter) {
152 if (alias == iter->prefix) {
153 if (iter->package.package.empty()) {
Adam Lesinski1ef0fa92017-08-15 21:32:49 -0700154 return ExtractedPackage{{}, iter->package.private_namespace};
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700155 }
156 return iter->package;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800157 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700158 }
159 return {};
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800160}
161
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700162const std::string& XmlPullParser::element_namespace() const {
163 const Event current_event = event();
164 if (current_event != Event::kStartElement &&
165 current_event != Event::kEndElement) {
166 return empty_;
167 }
168 return event_queue_.front().data1;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800169}
170
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700171const std::string& XmlPullParser::element_name() const {
172 const Event current_event = event();
173 if (current_event != Event::kStartElement &&
174 current_event != Event::kEndElement) {
175 return empty_;
176 }
177 return event_queue_.front().data2;
Adam Lesinski24aad162015-04-24 19:19:30 -0700178}
179
Ryan Mitchell326e35ff2021-04-12 07:50:42 -0700180const std::vector<XmlPullParser::PackageDecl>& XmlPullParser::package_decls() const {
181 return package_aliases_;
182}
183
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700184XmlPullParser::const_iterator XmlPullParser::begin_attributes() const {
185 return event_queue_.front().attributes.begin();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800186}
187
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700188XmlPullParser::const_iterator XmlPullParser::end_attributes() const {
189 return event_queue_.front().attributes.end();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800190}
191
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700192size_t XmlPullParser::attribute_count() const {
193 if (event() != Event::kStartElement) {
194 return 0;
195 }
196 return event_queue_.front().attributes.size();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800197}
198
199/**
200 * Extracts the namespace and name of an expanded element or attribute name.
201 */
Adam Lesinskid5083f62017-01-16 15:07:21 -0800202static void SplitName(const char* name, std::string* out_ns, std::string* out_name) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700203 const char* p = name;
204 while (*p != 0 && *p != kXmlNamespaceSep) {
205 p++;
206 }
207
208 if (*p == 0) {
Adam Lesinskid5083f62017-01-16 15:07:21 -0800209 out_ns->clear();
210 out_name->assign(name);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700211 } else {
Adam Lesinskid5083f62017-01-16 15:07:21 -0800212 out_ns->assign(name, (p - name));
213 out_name->assign(p + 1);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700214 }
215}
216
217void XMLCALL XmlPullParser::StartNamespaceHandler(void* user_data,
218 const char* prefix,
219 const char* uri) {
220 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
221 std::string namespace_uri = uri != nullptr ? uri : std::string();
222 parser->namespace_uris_.push(namespace_uri);
223 parser->event_queue_.push(
224 EventData{Event::kStartNamespace,
225 XML_GetCurrentLineNumber(parser->parser_), parser->depth_++,
226 prefix != nullptr ? prefix : std::string(), namespace_uri});
227}
228
229void XMLCALL XmlPullParser::StartElementHandler(void* user_data,
230 const char* name,
231 const char** attrs) {
232 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
233
234 EventData data = {Event::kStartElement,
235 XML_GetCurrentLineNumber(parser->parser_),
236 parser->depth_++};
Adam Lesinskid5083f62017-01-16 15:07:21 -0800237 SplitName(name, &data.data1, &data.data2);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700238
239 while (*attrs) {
240 Attribute attribute;
Adam Lesinskid5083f62017-01-16 15:07:21 -0800241 SplitName(*attrs++, &attribute.namespace_uri, &attribute.name);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700242 attribute.value = *attrs++;
243
244 // Insert in sorted order.
245 auto iter = std::lower_bound(data.attributes.begin(), data.attributes.end(),
246 attribute);
247 data.attributes.insert(iter, std::move(attribute));
248 }
249
250 // Move the structure into the queue (no copy).
251 parser->event_queue_.push(std::move(data));
252}
253
254void XMLCALL XmlPullParser::CharacterDataHandler(void* user_data, const char* s,
255 int len) {
256 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
257
Adam Lesinskid5083f62017-01-16 15:07:21 -0800258 parser->event_queue_.push(EventData{Event::kText, XML_GetCurrentLineNumber(parser->parser_),
259 parser->depth_, std::string(s, len)});
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700260}
261
262void XMLCALL XmlPullParser::EndElementHandler(void* user_data,
263 const char* name) {
264 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
265
266 EventData data = {Event::kEndElement,
267 XML_GetCurrentLineNumber(parser->parser_),
268 --(parser->depth_)};
Adam Lesinskid5083f62017-01-16 15:07:21 -0800269 SplitName(name, &data.data1, &data.data2);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700270
271 // Move the data into the queue (no copy).
272 parser->event_queue_.push(std::move(data));
273}
274
275void XMLCALL XmlPullParser::EndNamespaceHandler(void* user_data,
276 const char* prefix) {
277 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
278
279 parser->event_queue_.push(
280 EventData{Event::kEndNamespace, XML_GetCurrentLineNumber(parser->parser_),
281 --(parser->depth_), prefix != nullptr ? prefix : std::string(),
282 parser->namespace_uris_.top()});
283 parser->namespace_uris_.pop();
284}
285
286void XMLCALL XmlPullParser::CommentDataHandler(void* user_data,
287 const char* comment) {
288 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
289
290 parser->event_queue_.push(EventData{Event::kComment,
291 XML_GetCurrentLineNumber(parser->parser_),
292 parser->depth_, comment});
293}
294
Ryan Mitchellcb76d732018-06-05 10:15:04 -0700295void XMLCALL XmlPullParser::StartCdataSectionHandler(void* user_data) {
296 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
297
298 parser->event_queue_.push(EventData{Event::kCdataStart,
299 XML_GetCurrentLineNumber(parser->parser_),
300 parser->depth_ });
301}
302
303void XMLCALL XmlPullParser::EndCdataSectionHandler(void* user_data) {
304 XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data);
305
306 parser->event_queue_.push(EventData{Event::kCdataEnd,
307 XML_GetCurrentLineNumber(parser->parser_),
308 parser->depth_ });
309}
310
Yurii Zubrytskyia5775142022-11-02 17:49:49 -0700311std::optional<StringPiece> FindAttribute(const XmlPullParser* parser, StringPiece name) {
Jeremy Meyer211bec22024-06-04 14:22:03 -0700312 return FindAttribute(parser, "", name);
313}
314
315std::optional<android::StringPiece> FindAttribute(const XmlPullParser* parser,
316 android::StringPiece namespace_uri,
317 android::StringPiece name) {
318 auto iter = parser->FindAttribute(namespace_uri, name);
319
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700320 if (iter != parser->end_attributes()) {
321 return StringPiece(util::TrimWhitespace(iter->value));
322 }
323 return {};
324}
325
Yurii Zubrytskyia5775142022-11-02 17:49:49 -0700326std::optional<StringPiece> FindNonEmptyAttribute(const XmlPullParser* parser, StringPiece name) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700327 auto iter = parser->FindAttribute("", name);
328 if (iter != parser->end_attributes()) {
329 StringPiece trimmed = util::TrimWhitespace(iter->value);
330 if (!trimmed.empty()) {
331 return trimmed;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800332 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700333 }
334 return {};
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800335}
336
Yurii Zubrytskyi1d6d8ac2024-07-03 16:10:17 -0700337XmlPullParser::const_iterator XmlPullParser::FindAttribute(android::StringPiece namespace_uri,
338 android::StringPiece name) const {
339 const auto end_iter = end_attributes();
340 const auto iter = std::lower_bound(begin_attributes(), end_iter, std::tuple(namespace_uri, name),
341 [](const Attribute& attr, const auto& rhs) {
342 return std::tie(attr.namespace_uri, attr.name) < rhs;
343 });
344 if (iter != end_iter && namespace_uri == iter->namespace_uri && name == iter->name) {
345 return iter;
346 }
347 return end_iter;
348}
349
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700350} // namespace xml
351} // namespace aapt