blob: fe4cd018d80806e60fd41acf37242f033d68a664 [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
17#ifndef AAPT_XML_PULL_PARSER_H
18#define AAPT_XML_PULL_PARSER_H
19
Adam Lesinski1ab598f2015-08-14 14:26:04 -070020#include <expat.h>
Adam Lesinskice5e56e2016-10-21 17:56:45 -070021
Adam Lesinskicacb28f2016-10-19 12:18:14 -070022#include <algorithm>
Adam Lesinski1ab598f2015-08-14 14:26:04 -070023#include <istream>
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080024#include <ostream>
Adam Lesinski1ab598f2015-08-14 14:26:04 -070025#include <queue>
26#include <stack>
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080027#include <string>
28#include <vector>
29
Adam Lesinskice5e56e2016-10-21 17:56:45 -070030#include "android-base/macros.h"
Adam Lesinskid5083f62017-01-16 15:07:21 -080031#include "androidfw/StringPiece.h"
Adam Lesinskice5e56e2016-10-21 17:56:45 -070032
33#include "Resource.h"
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070034#include "io/Io.h"
Adam Lesinskice5e56e2016-10-21 17:56:45 -070035#include "process/IResourceTableConsumer.h"
Adam Lesinskice5e56e2016-10-21 17:56:45 -070036#include "xml/XmlUtil.h"
37
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080038namespace aapt {
Adam Lesinski467f1712015-11-16 17:35:44 -080039namespace xml {
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080040
Adam Lesinski1ab598f2015-08-14 14:26:04 -070041class XmlPullParser : public IPackageDeclStack {
Adam Lesinskicacb28f2016-10-19 12:18:14 -070042 public:
43 enum class Event {
44 kBadDocument,
45 kStartDocument,
46 kEndDocument,
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080047
Adam Lesinskicacb28f2016-10-19 12:18:14 -070048 kStartNamespace,
49 kEndNamespace,
50 kStartElement,
51 kEndElement,
52 kText,
53 kComment,
Ryan Mitchellcb76d732018-06-05 10:15:04 -070054 kCdataStart,
55 kCdataEnd,
Adam Lesinskicacb28f2016-10-19 12:18:14 -070056 };
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080057
Adam Lesinskicacb28f2016-10-19 12:18:14 -070058 /**
Adam Lesinskice5e56e2016-10-21 17:56:45 -070059 * Skips to the next direct descendant node of the given start_depth,
Adam Lesinskicacb28f2016-10-19 12:18:14 -070060 * skipping namespace nodes.
61 *
Adam Lesinskice5e56e2016-10-21 17:56:45 -070062 * When NextChildNode() returns true, you can expect Comments, Text, and
Adam Lesinskicacb28f2016-10-19 12:18:14 -070063 * StartElement events.
64 */
Adam Lesinskice5e56e2016-10-21 17:56:45 -070065 static bool NextChildNode(XmlPullParser* parser, size_t start_depth);
66 static bool SkipCurrentElement(XmlPullParser* parser);
67 static bool IsGoodEvent(Event event);
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080068
Adam Lesinskiefeb7af2017-08-02 14:57:43 -070069 explicit XmlPullParser(io::InputStream* in);
Adam Lesinskicacb28f2016-10-19 12:18:14 -070070 ~XmlPullParser();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080071
Adam Lesinskicacb28f2016-10-19 12:18:14 -070072 /**
73 * Returns the current event that is being processed.
74 */
Adam Lesinskice5e56e2016-10-21 17:56:45 -070075 Event event() const;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080076
Adam Lesinskice5e56e2016-10-21 17:56:45 -070077 const std::string& error() const;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080078
Adam Lesinskicacb28f2016-10-19 12:18:14 -070079 /**
80 * Note, unlike XmlPullParser, the first call to next() will return
81 * StartElement of the first element.
82 */
Adam Lesinskice5e56e2016-10-21 17:56:45 -070083 Event Next();
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080084
Adam Lesinskicacb28f2016-10-19 12:18:14 -070085 //
86 // These are available for all nodes.
87 //
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080088
Adam Lesinskice5e56e2016-10-21 17:56:45 -070089 const std::string& comment() const;
90 size_t line_number() const;
91 size_t depth() const;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080092
Adam Lesinskicacb28f2016-10-19 12:18:14 -070093 /**
94 * Returns the character data for a Text event.
95 */
Adam Lesinskice5e56e2016-10-21 17:56:45 -070096 const std::string& text() const;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -080097
Adam Lesinskicacb28f2016-10-19 12:18:14 -070098 //
99 // Namespace prefix and URI are available for StartNamespace and EndNamespace.
100 //
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800101
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700102 const std::string& namespace_prefix() const;
103 const std::string& namespace_uri() const;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800104
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700105 //
106 // These are available for StartElement and EndElement.
107 //
Adam Lesinski467f1712015-11-16 17:35:44 -0800108
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700109 const std::string& element_namespace() const;
110 const std::string& element_name() const;
Adam Lesinski467f1712015-11-16 17:35:44 -0800111
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700112 /*
113 * Uses the current stack of namespaces to resolve the package. Eg:
114 * xmlns:app = "http://schemas.android.com/apk/res/com.android.app"
115 * ...
116 * android:text="@app:string/message"
117 *
118 * In this case, 'app' will be converted to 'com.android.app'.
119 *
120 * If xmlns:app="http://schemas.android.com/apk/res-auto", then
121 * 'package' will be set to 'defaultPackage'.
122 */
Yurii Zubrytskyia5775142022-11-02 17:49:49 -0700123 std::optional<ExtractedPackage> TransformPackageAlias(android::StringPiece alias) const override;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800124
Ryan Mitchell326e35ff2021-04-12 07:50:42 -0700125 struct PackageDecl {
126 std::string prefix;
127 ExtractedPackage package;
128 };
129
130 const std::vector<PackageDecl>& package_decls() const;
131
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700132 //
133 // Remaining methods are for retrieving information about attributes
134 // associated with a StartElement.
135 //
136 // Attributes must be in sorted order (according to the less than operator
137 // of struct Attribute).
138 //
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800139
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700140 struct Attribute {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700141 std::string namespace_uri;
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700142 std::string name;
143 std::string value;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800144
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700145 int compare(const Attribute& rhs) const;
146 bool operator<(const Attribute& rhs) const;
147 bool operator==(const Attribute& rhs) const;
148 bool operator!=(const Attribute& rhs) const;
149 };
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800150
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700151 using const_iterator = std::vector<Attribute>::const_iterator;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800152
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700153 const_iterator begin_attributes() const;
154 const_iterator end_attributes() const;
155 size_t attribute_count() const;
Adam Lesinskid5083f62017-01-16 15:07:21 -0800156 const_iterator FindAttribute(android::StringPiece namespace_uri, android::StringPiece name) const;
Adam Lesinski1ab598f2015-08-14 14:26:04 -0700157
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700158 private:
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700159 DISALLOW_COPY_AND_ASSIGN(XmlPullParser);
160
161 static void XMLCALL StartNamespaceHandler(void* user_data, const char* prefix,
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700162 const char* uri);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700163 static void XMLCALL StartElementHandler(void* user_data, const char* name,
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700164 const char** attrs);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700165 static void XMLCALL CharacterDataHandler(void* user_data, const char* s,
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700166 int len);
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700167 static void XMLCALL EndElementHandler(void* user_data, const char* name);
168 static void XMLCALL EndNamespaceHandler(void* user_data, const char* prefix);
169 static void XMLCALL CommentDataHandler(void* user_data, const char* comment);
Ryan Mitchellcb76d732018-06-05 10:15:04 -0700170 static void XMLCALL StartCdataSectionHandler(void* user_data);
171 static void XMLCALL EndCdataSectionHandler(void* user_data);
Adam Lesinski1ab598f2015-08-14 14:26:04 -0700172
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700173 struct EventData {
174 Event event;
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700175 size_t line_number;
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700176 size_t depth;
177 std::string data1;
178 std::string data2;
179 std::vector<Attribute> attributes;
180 };
Adam Lesinski1ab598f2015-08-14 14:26:04 -0700181
Adam Lesinskiefeb7af2017-08-02 14:57:43 -0700182 io::InputStream* in_;
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700183 XML_Parser parser_;
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700184 std::queue<EventData> event_queue_;
185 std::string error_;
186 const std::string empty_;
187 size_t depth_;
188 std::stack<std::string> namespace_uris_;
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700189 std::vector<PackageDecl> package_aliases_;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800190};
191
Adam Lesinski467f1712015-11-16 17:35:44 -0800192/**
193 * Finds the attribute in the current element within the global namespace.
194 */
Ryan Mitchell4382e442021-07-14 12:53:01 -0700195std::optional<android::StringPiece> FindAttribute(const XmlPullParser* parser,
Yurii Zubrytskyia5775142022-11-02 17:49:49 -0700196 android::StringPiece name);
Adam Lesinski467f1712015-11-16 17:35:44 -0800197
198/**
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700199 * Finds the attribute in the current element within the global namespace. The
200 * attribute's value
Adam Lesinski467f1712015-11-16 17:35:44 -0800201 * must not be the empty string.
202 */
Ryan Mitchell4382e442021-07-14 12:53:01 -0700203std::optional<android::StringPiece> FindNonEmptyAttribute(const XmlPullParser* parser,
Yurii Zubrytskyia5775142022-11-02 17:49:49 -0700204 android::StringPiece name);
Adam Lesinski467f1712015-11-16 17:35:44 -0800205
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800206//
207// Implementation
208//
209
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700210inline ::std::ostream& operator<<(::std::ostream& out,
211 XmlPullParser::Event event) {
212 switch (event) {
213 case XmlPullParser::Event::kBadDocument:
214 return out << "BadDocument";
215 case XmlPullParser::Event::kStartDocument:
216 return out << "StartDocument";
217 case XmlPullParser::Event::kEndDocument:
218 return out << "EndDocument";
219 case XmlPullParser::Event::kStartNamespace:
220 return out << "StartNamespace";
221 case XmlPullParser::Event::kEndNamespace:
222 return out << "EndNamespace";
223 case XmlPullParser::Event::kStartElement:
224 return out << "StartElement";
225 case XmlPullParser::Event::kEndElement:
226 return out << "EndElement";
227 case XmlPullParser::Event::kText:
228 return out << "Text";
229 case XmlPullParser::Event::kComment:
230 return out << "Comment";
Ryan Mitchellcb76d732018-06-05 10:15:04 -0700231 case XmlPullParser::Event::kCdataStart:
232 return out << "CdataStart";
233 case XmlPullParser::Event::kCdataEnd:
234 return out << "CdataEnd";
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700235 }
236 return out;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800237}
238
Adam Lesinskiefeb7af2017-08-02 14:57:43 -0700239inline bool XmlPullParser::NextChildNode(XmlPullParser* parser, size_t start_depth) {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700240 Event event;
Adam Lesinski1ab598f2015-08-14 14:26:04 -0700241
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700242 // First get back to the start depth.
Adam Lesinskiefeb7af2017-08-02 14:57:43 -0700243 while (IsGoodEvent(event = parser->Next()) && parser->depth() > start_depth + 1) {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700244 }
Adam Lesinski1ab598f2015-08-14 14:26:04 -0700245
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700246 // Now look for the first good node.
Adam Lesinskiefeb7af2017-08-02 14:57:43 -0700247 while ((event != Event::kEndElement || parser->depth() > start_depth) && IsGoodEvent(event)) {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700248 switch (event) {
249 case Event::kText:
250 case Event::kComment:
251 case Event::kStartElement:
Ryan Mitchellcb76d732018-06-05 10:15:04 -0700252 case Event::kCdataStart:
253 case Event::kCdataEnd:
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700254 return true;
255 default:
256 break;
Adam Lesinski1ab598f2015-08-14 14:26:04 -0700257 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700258 event = parser->Next();
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700259 }
260 return false;
Adam Lesinski1ab598f2015-08-14 14:26:04 -0700261}
262
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700263inline bool XmlPullParser::SkipCurrentElement(XmlPullParser* parser) {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700264 int depth = 1;
265 while (depth > 0) {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700266 switch (parser->Next()) {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700267 case Event::kEndDocument:
268 return true;
269 case Event::kBadDocument:
270 return false;
271 case Event::kStartElement:
272 depth++;
273 break;
274 case Event::kEndElement:
275 depth--;
276 break;
277 default:
278 break;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800279 }
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700280 }
281 return true;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800282}
283
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700284inline bool XmlPullParser::IsGoodEvent(XmlPullParser::Event event) {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700285 return event != Event::kBadDocument && event != Event::kEndDocument;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800286}
287
288inline int XmlPullParser::Attribute::compare(const Attribute& rhs) const {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700289 int cmp = namespace_uri.compare(rhs.namespace_uri);
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700290 if (cmp != 0) return cmp;
291 return name.compare(rhs.name);
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800292}
293
294inline bool XmlPullParser::Attribute::operator<(const Attribute& rhs) const {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700295 return compare(rhs) < 0;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800296}
297
298inline bool XmlPullParser::Attribute::operator==(const Attribute& rhs) const {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700299 return compare(rhs) == 0;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800300}
301
302inline bool XmlPullParser::Attribute::operator!=(const Attribute& rhs) const {
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700303 return compare(rhs) != 0;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800304}
305
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700306inline XmlPullParser::const_iterator XmlPullParser::FindAttribute(
Adam Lesinskid5083f62017-01-16 15:07:21 -0800307 android::StringPiece namespace_uri, android::StringPiece name) const {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700308 const auto end_iter = end_attributes();
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700309 const auto iter = std::lower_bound(
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700310 begin_attributes(), end_iter,
Adam Lesinskid5083f62017-01-16 15:07:21 -0800311 std::pair<android::StringPiece, android::StringPiece>(namespace_uri, name),
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700312 [](const Attribute& attr,
Adam Lesinskid5083f62017-01-16 15:07:21 -0800313 const std::pair<android::StringPiece, android::StringPiece>& rhs) -> bool {
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700314 int cmp = attr.namespace_uri.compare(
315 0, attr.namespace_uri.size(), rhs.first.data(), rhs.first.size());
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700316 if (cmp < 0) return true;
317 if (cmp > 0) return false;
318 cmp = attr.name.compare(0, attr.name.size(), rhs.second.data(),
319 rhs.second.size());
320 if (cmp < 0) return true;
321 return false;
322 });
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800323
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700324 if (iter != end_iter && namespace_uri == iter->namespace_uri &&
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700325 name == iter->name) {
326 return iter;
327 }
Adam Lesinskice5e56e2016-10-21 17:56:45 -0700328 return end_iter;
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800329}
330
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700331} // namespace xml
332} // namespace aapt
Adam Lesinski6f6ceb72014-11-14 14:48:12 -0800333
Adam Lesinskicacb28f2016-10-19 12:18:14 -0700334#endif // AAPT_XML_PULL_PARSER_H