Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 1 | /* |
| 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 Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 17 | #include <iostream> |
| 18 | #include <string> |
| 19 | |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 20 | #include "util/Maybe.h" |
| 21 | #include "util/Util.h" |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 22 | #include "xml/XmlPullParser.h" |
| 23 | #include "xml/XmlUtil.h" |
Adam Lesinski | 1ab598f | 2015-08-14 14:26:04 -0700 | [diff] [blame] | 24 | |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 25 | namespace aapt { |
Adam Lesinski | 467f171 | 2015-11-16 17:35:44 -0800 | [diff] [blame] | 26 | namespace xml { |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 27 | |
| 28 | constexpr char kXmlNamespaceSep = 1; |
| 29 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 30 | XmlPullParser::XmlPullParser(std::istream& in) : in_(in), empty_(), depth_(0) { |
| 31 | parser_ = XML_ParserCreateNS(nullptr, kXmlNamespaceSep); |
| 32 | XML_SetUserData(parser_, this); |
| 33 | XML_SetElementHandler(parser_, StartElementHandler, EndElementHandler); |
| 34 | XML_SetNamespaceDeclHandler(parser_, StartNamespaceHandler, |
| 35 | EndNamespaceHandler); |
| 36 | XML_SetCharacterDataHandler(parser_, CharacterDataHandler); |
| 37 | XML_SetCommentHandler(parser_, CommentDataHandler); |
| 38 | event_queue_.push(EventData{Event::kStartDocument, 0, depth_++}); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 39 | } |
| 40 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 41 | XmlPullParser::~XmlPullParser() { XML_ParserFree(parser_); } |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 42 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 43 | XmlPullParser::Event XmlPullParser::Next() { |
| 44 | const Event currentEvent = event(); |
| 45 | if (currentEvent == Event::kBadDocument || |
| 46 | currentEvent == Event::kEndDocument) { |
| 47 | return currentEvent; |
| 48 | } |
| 49 | |
| 50 | event_queue_.pop(); |
| 51 | while (event_queue_.empty()) { |
| 52 | in_.read(buffer_, sizeof(buffer_) / sizeof(*buffer_)); |
| 53 | |
| 54 | const bool done = in_.eof(); |
| 55 | if (in_.bad() && !done) { |
| 56 | error_ = strerror(errno); |
| 57 | event_queue_.push(EventData{Event::kBadDocument}); |
| 58 | continue; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 59 | } |
| 60 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 61 | if (XML_Parse(parser_, buffer_, in_.gcount(), done) == XML_STATUS_ERROR) { |
| 62 | error_ = XML_ErrorString(XML_GetErrorCode(parser_)); |
| 63 | event_queue_.push(EventData{Event::kBadDocument}); |
| 64 | continue; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 65 | } |
| 66 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 67 | if (done) { |
| 68 | event_queue_.push(EventData{Event::kEndDocument, 0, 0}); |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 69 | } |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 70 | } |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 71 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 72 | Event next_event = event(); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 73 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 74 | // Record namespace prefixes and package names so that we can do our own |
| 75 | // handling of references that use namespace aliases. |
| 76 | if (next_event == Event::kStartNamespace || |
| 77 | next_event == Event::kEndNamespace) { |
| 78 | Maybe<ExtractedPackage> result = |
| 79 | ExtractPackageFromNamespace(namespace_uri()); |
| 80 | if (next_event == Event::kStartNamespace) { |
| 81 | if (result) { |
| 82 | package_aliases_.emplace_back( |
| 83 | PackageDecl{namespace_prefix(), std::move(result.value())}); |
| 84 | } |
| 85 | } else { |
| 86 | if (result) { |
| 87 | package_aliases_.pop_back(); |
| 88 | } |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 89 | } |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 90 | } |
| 91 | |
| 92 | return next_event; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 93 | } |
| 94 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 95 | XmlPullParser::Event XmlPullParser::event() const { |
| 96 | return event_queue_.front().event; |
| 97 | } |
| 98 | |
| 99 | const std::string& XmlPullParser::error() const { return error_; } |
| 100 | |
| 101 | const std::string& XmlPullParser::comment() const { |
| 102 | return event_queue_.front().data1; |
| 103 | } |
| 104 | |
| 105 | size_t XmlPullParser::line_number() const { |
| 106 | return event_queue_.front().line_number; |
| 107 | } |
| 108 | |
| 109 | size_t XmlPullParser::depth() const { return event_queue_.front().depth; } |
| 110 | |
| 111 | const std::string& XmlPullParser::text() const { |
| 112 | if (event() != Event::kText) { |
| 113 | return empty_; |
| 114 | } |
| 115 | return event_queue_.front().data1; |
| 116 | } |
| 117 | |
| 118 | const std::string& XmlPullParser::namespace_prefix() const { |
| 119 | const Event current_event = event(); |
| 120 | if (current_event != Event::kStartNamespace && |
| 121 | current_event != Event::kEndNamespace) { |
| 122 | return empty_; |
| 123 | } |
| 124 | return event_queue_.front().data1; |
| 125 | } |
| 126 | |
| 127 | const std::string& XmlPullParser::namespace_uri() 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().data2; |
| 134 | } |
| 135 | |
| 136 | Maybe<ExtractedPackage> XmlPullParser::TransformPackageAlias( |
| 137 | const StringPiece& alias, const StringPiece& local_package) const { |
| 138 | if (alias.empty()) { |
| 139 | return ExtractedPackage{local_package.ToString(), false /* private */}; |
| 140 | } |
| 141 | |
| 142 | const auto end_iter = package_aliases_.rend(); |
| 143 | for (auto iter = package_aliases_.rbegin(); iter != end_iter; ++iter) { |
| 144 | if (alias == iter->prefix) { |
| 145 | if (iter->package.package.empty()) { |
| 146 | return ExtractedPackage{local_package.ToString(), |
| 147 | iter->package.private_namespace}; |
| 148 | } |
| 149 | return iter->package; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 150 | } |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 151 | } |
| 152 | return {}; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 153 | } |
| 154 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 155 | const std::string& XmlPullParser::element_namespace() const { |
| 156 | const Event current_event = event(); |
| 157 | if (current_event != Event::kStartElement && |
| 158 | current_event != Event::kEndElement) { |
| 159 | return empty_; |
| 160 | } |
| 161 | return event_queue_.front().data1; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 162 | } |
| 163 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 164 | const std::string& XmlPullParser::element_name() const { |
| 165 | const Event current_event = event(); |
| 166 | if (current_event != Event::kStartElement && |
| 167 | current_event != Event::kEndElement) { |
| 168 | return empty_; |
| 169 | } |
| 170 | return event_queue_.front().data2; |
Adam Lesinski | 24aad16 | 2015-04-24 19:19:30 -0700 | [diff] [blame] | 171 | } |
| 172 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 173 | XmlPullParser::const_iterator XmlPullParser::begin_attributes() const { |
| 174 | return event_queue_.front().attributes.begin(); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 175 | } |
| 176 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 177 | XmlPullParser::const_iterator XmlPullParser::end_attributes() const { |
| 178 | return event_queue_.front().attributes.end(); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 179 | } |
| 180 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 181 | size_t XmlPullParser::attribute_count() const { |
| 182 | if (event() != Event::kStartElement) { |
| 183 | return 0; |
| 184 | } |
| 185 | return event_queue_.front().attributes.size(); |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 186 | } |
| 187 | |
| 188 | /** |
| 189 | * Extracts the namespace and name of an expanded element or attribute name. |
| 190 | */ |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 191 | static void SplitName(const char* name, std::string& out_ns, |
| 192 | std::string& out_name) { |
| 193 | const char* p = name; |
| 194 | while (*p != 0 && *p != kXmlNamespaceSep) { |
| 195 | p++; |
| 196 | } |
| 197 | |
| 198 | if (*p == 0) { |
| 199 | out_ns = std::string(); |
| 200 | out_name = name; |
| 201 | } else { |
| 202 | out_ns = StringPiece(name, (p - name)).ToString(); |
| 203 | out_name = p + 1; |
| 204 | } |
| 205 | } |
| 206 | |
| 207 | void XMLCALL XmlPullParser::StartNamespaceHandler(void* user_data, |
| 208 | const char* prefix, |
| 209 | const char* uri) { |
| 210 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data); |
| 211 | std::string namespace_uri = uri != nullptr ? uri : std::string(); |
| 212 | parser->namespace_uris_.push(namespace_uri); |
| 213 | parser->event_queue_.push( |
| 214 | EventData{Event::kStartNamespace, |
| 215 | XML_GetCurrentLineNumber(parser->parser_), parser->depth_++, |
| 216 | prefix != nullptr ? prefix : std::string(), namespace_uri}); |
| 217 | } |
| 218 | |
| 219 | void XMLCALL XmlPullParser::StartElementHandler(void* user_data, |
| 220 | const char* name, |
| 221 | const char** attrs) { |
| 222 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data); |
| 223 | |
| 224 | EventData data = {Event::kStartElement, |
| 225 | XML_GetCurrentLineNumber(parser->parser_), |
| 226 | parser->depth_++}; |
| 227 | SplitName(name, data.data1, data.data2); |
| 228 | |
| 229 | while (*attrs) { |
| 230 | Attribute attribute; |
| 231 | SplitName(*attrs++, attribute.namespace_uri, attribute.name); |
| 232 | attribute.value = *attrs++; |
| 233 | |
| 234 | // Insert in sorted order. |
| 235 | auto iter = std::lower_bound(data.attributes.begin(), data.attributes.end(), |
| 236 | attribute); |
| 237 | data.attributes.insert(iter, std::move(attribute)); |
| 238 | } |
| 239 | |
| 240 | // Move the structure into the queue (no copy). |
| 241 | parser->event_queue_.push(std::move(data)); |
| 242 | } |
| 243 | |
| 244 | void XMLCALL XmlPullParser::CharacterDataHandler(void* user_data, const char* s, |
| 245 | int len) { |
| 246 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data); |
| 247 | |
| 248 | parser->event_queue_.push( |
| 249 | EventData{Event::kText, XML_GetCurrentLineNumber(parser->parser_), |
| 250 | parser->depth_, StringPiece(s, len).ToString()}); |
| 251 | } |
| 252 | |
| 253 | void XMLCALL XmlPullParser::EndElementHandler(void* user_data, |
| 254 | const char* name) { |
| 255 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data); |
| 256 | |
| 257 | EventData data = {Event::kEndElement, |
| 258 | XML_GetCurrentLineNumber(parser->parser_), |
| 259 | --(parser->depth_)}; |
| 260 | SplitName(name, data.data1, data.data2); |
| 261 | |
| 262 | // Move the data into the queue (no copy). |
| 263 | parser->event_queue_.push(std::move(data)); |
| 264 | } |
| 265 | |
| 266 | void XMLCALL XmlPullParser::EndNamespaceHandler(void* user_data, |
| 267 | const char* prefix) { |
| 268 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data); |
| 269 | |
| 270 | parser->event_queue_.push( |
| 271 | EventData{Event::kEndNamespace, XML_GetCurrentLineNumber(parser->parser_), |
| 272 | --(parser->depth_), prefix != nullptr ? prefix : std::string(), |
| 273 | parser->namespace_uris_.top()}); |
| 274 | parser->namespace_uris_.pop(); |
| 275 | } |
| 276 | |
| 277 | void XMLCALL XmlPullParser::CommentDataHandler(void* user_data, |
| 278 | const char* comment) { |
| 279 | XmlPullParser* parser = reinterpret_cast<XmlPullParser*>(user_data); |
| 280 | |
| 281 | parser->event_queue_.push(EventData{Event::kComment, |
| 282 | XML_GetCurrentLineNumber(parser->parser_), |
| 283 | parser->depth_, comment}); |
| 284 | } |
| 285 | |
| 286 | Maybe<StringPiece> FindAttribute(const XmlPullParser* parser, |
| 287 | const StringPiece& name) { |
| 288 | auto iter = parser->FindAttribute("", name); |
| 289 | if (iter != parser->end_attributes()) { |
| 290 | return StringPiece(util::TrimWhitespace(iter->value)); |
| 291 | } |
| 292 | return {}; |
| 293 | } |
| 294 | |
| 295 | Maybe<StringPiece> FindNonEmptyAttribute(const XmlPullParser* parser, |
| 296 | const StringPiece& name) { |
| 297 | auto iter = parser->FindAttribute("", name); |
| 298 | if (iter != parser->end_attributes()) { |
| 299 | StringPiece trimmed = util::TrimWhitespace(iter->value); |
| 300 | if (!trimmed.empty()) { |
| 301 | return trimmed; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 302 | } |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 303 | } |
| 304 | return {}; |
Adam Lesinski | 6f6ceb7 | 2014-11-14 14:48:12 -0800 | [diff] [blame] | 305 | } |
| 306 | |
Adam Lesinski | ce5e56e | 2016-10-21 17:56:45 -0700 | [diff] [blame^] | 307 | } // namespace xml |
| 308 | } // namespace aapt |