Mercurial > hg > extraction-interface

comparison develop/models/extractapp.php @ 39:8347776a44fc extractapp

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
embedded topic, taglist,book metadata in file
author Zoe Hong <zhong@mpiwg-berlin.mpg.de
date Sun, 08 Mar 2015 22:28:21 +0100
parents 4b3da4802998
children 2e938dc046db
comparison
equal deleted inserted replaced
38:4b3da4802998 39:8347776a44fc
6 6
7 public function Index() { 7 public function Index() {
8 return array("Index Value 1", "Value 2", "Value 3"); 8 return array("Index Value 1", "Value 2", "Value 3");
9 } 9 }
10 10
11 protected $section_id = 0, $data_path, $file_id = 0, $current_fileId=0, $branch_id = 0, $user_id = 0, $lg_text = "", $topic = 0; 11 protected $section_id = 0, $data_path, $file_id = 0, $current_fileId=0,
12 $branch_id = 0, $user_id = 0, $lg_text = "", $topic = 0, $taglist_infile = "", $book_meta = "";
12 public $messages = ""; 13 public $messages = "";
13 14
14 private function Initialize($_urlvalues) { 15 private function Initialize($_urlvalues) {
15 $this->SetSectionId($_urlvalues); 16 $this->SetSectionId($_urlvalues);
16 17
191 192
192 } 193 }
193 */ 194 */
194 195
195 196
196 197 private function Taglist_infileUpToDate($taglistArray) {
198 // compare $this->taglist_infile is the same as $taglistArray
199 $taglist_infile = $this->taglist_infile;
200 if (count($taglist_infile) != count($taglistArray)) {
201 return false;
202 }
203
204 foreach ($taglistArray as $row_indb) {
205 foreach ($taglist_infile as $row) {
206 //$taglistArray: array( $row['id'], $row['name'], $row['tag'], $row['color'] )
207 if ($row[0] == $row_indb[0] && $row[1] == $row_indb[1] && $row[2] == $row_indb[2] && $row[3] == $row_indb[3]) {
208
209 } else {
210 return false;
211 }
212 }
213 }
214 return true;
215 }
197 // === for tagging === 216 // === for tagging ===
198 public function StartTagging() { 217 public function StartTagging() {
199 218
200 // $this->Initialize($urlvalues);
201
202 $section_id = $this->section_id; 219 $section_id = $this->section_id;
203 $stringInput = $this->lg_text; 220 $stringInput = $this->lg_text;
204 221
222 $data = array(); // data array to be passed to view
205 223
206 //$taglistArray = $this->GetTaglistArray(); 224 //$taglistArray = $this->GetTaglistArray();
207 //for GetTaglistByTopicID: 225 //for GetTaglistByTopicID:
208 $taglistArray = $this->GetTaglistByTopicID($this->GetTopic()); 226 $taglistArray = $this->GetTaglistByTopicID($this->GetTopic());
227
228 // TODO: check if taglist_infile is up-to-date
229 if ( !$this->Taglist_infileUpToDate($taglistArray)) {
230 $data['taglist_infile'] = $this->taglist_infile;
231 }
232
233
209 // topic list 234 // topic list
210 $topiclistArray = $this->GetTopiclistArray(); 235 $topiclistArray = $this->GetTopiclistArray();
211 236
212 237
213 $wordlistArray = $this->GetWordlistArray(); 238 $wordlistArray = $this->GetWordlistArray();
214 239
215 $data = array();
216 $data['stringInput'] = $stringInput; 240 $data['stringInput'] = $stringInput;
217 $data['taglistArray'] = $taglistArray; 241 $data['taglistArray'] = $taglistArray;
218 $data['wordlistArray'] = $wordlistArray; 242 $data['wordlistArray'] = $wordlistArray;
219 $data['section_id'] = $section_id; 243 $data['section_id'] = $section_id;
220 $data['topiclistArray'] = $topiclistArray; 244 $data['topiclistArray'] = $topiclistArray;
1117 1141
1118 1142
1119 return $data; 1143 return $data;
1120 } 1144 }
1121 1145
1146 private function ParseMetaData($filename) {
1147 $text = file_get_contents($filename);
1148 $xml = simplexml_load_string($text) or die("Error: Cannot load from xml string");
1149
1150 $this->topic = $xml->topic; // set topic id
1151
1152 // get taglist in file
1153 $taglist_infile = $xml->taglist;
1154 $taglistArray = array();
1155 foreach ($taglist_infile as $row) {
1156 array_push($taglistArray, array((string)$row->id, (string)$row->name, (string)$row->tag, (string)$row->color));
1157 }
1158 $this->taglist_infile = $taglistArray;
1159
1160 // TODO: get meta data of book
1161 $this->book_meta = $xml->book;
1162
1163 // echo $taglist->name.", ".$taglist->tag."," .$taglist->color;
1164 // --- detect if the taglist set is up-to-date or not ---
1165
1166
1167 $contentString = $xml->text_content->asXML();
1168
1169 return $contentString;
1170 }
1122 private function GetSectionContent() { 1171 private function GetSectionContent() {
1123 $section_id = $this->GetSectionId(); 1172 $section_id = $this->GetSectionId();
1124 $section_info = $this->GetSectionInfo(); 1173 $section_info = $this->GetSectionInfo();
1125 1174
1126 $bookId = $section_info['bookId']; 1175 $bookId = $section_info['bookId'];
1128 $endPage = $section_info['endPage']; 1177 $endPage = $section_info['endPage'];
1129 1178
1130 $contentString=""; 1179 $contentString="";
1131 $data_path = $this->GetDataPath(); 1180 $data_path = $this->GetDataPath();
1132 if ( file_exists($data_path."parsing_files/".$section_id.".txt") ) { 1181 if ( file_exists($data_path."parsing_files/".$section_id.".txt") ) {
1133 $contentString=file_get_contents($data_path."parsing_files/".$section_id.".txt"); 1182 $filename = $data_path."parsing_files/".$section_id.".txt";
1134 $stringInput = $contentString; 1183 // --- parsing meta data
1184 $stringInput = $this->ParseMetaData($filename);
1185 // ----
1135 1186
1136 // if the text is from file system 1187 // if the text is from file system
1137 $stringInput = preg_replace("/ /u", "○", $stringInput); 1188 $stringInput = preg_replace("/ /u", "○", $stringInput);
1138 $stringInput = preg_replace("/\n/u", "<br>", $stringInput); 1189 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
1139 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput); 1190 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);