Mercurial > hg > extraction-interface

comparison develop/models/extractapp.php @ 12:e82ca7375e93

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Integration with LocalGazetteerService (LGService), getting text from LGService and saving text to LGService. Adjust the control flow (controller).
author Zoe Hong <zhong@mpiwg-berlin.mpg.de>
date Mon, 16 Feb 2015 16:51:39 +0100
parents 3d6fba07bfbd
children cc36a20a68ab
comparison
equal deleted inserted replaced
11:3d6fba07bfbd 12:e82ca7375e93
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, $data_path; 11 protected $section_id, $data_path, $file_id, $branch_id, $user_id, $lg_tex;
12
13 12
14 private function Initialize($_urlvalues) { 13 private function Initialize($_urlvalues) {
15 $this->SetSectionId($_urlvalues); 14 $this->SetSectionId($_urlvalues);
16 15
17 } 16 }
18 17
18 public function GetTextFromFileId($_postdata) {
19 $file_id = $_postdata['fileId'];
20 $branch_id = $_postdata['branchId'];
21 $section_id = $_postdata['sectionId'];
22
23 $this->branch_id = $branch_id;
24 $this->file_id = $file_id;
25 $this->user_id = $_postdata['userId'];
26 $this->section_id = $section_id;
27
28 // get from URL with file_id
29 $lg_text_url = $this->get_text_from_fileId_url.$file_id;
30 $lg_text = file_get_contents($lg_text_url);
31
32 $stringInput = $lg_text;
33 $stringInput = preg_replace("/ /u", "○", $stringInput);
34 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
35 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
36 $lg_text = $stringInput;
37
38 $this->lg_text = $lg_text;
39
40 }
41
42 public function GetTextFromSectionId($_postdata) {
43
44 $section_id = $_postdata['sectionId'];
45 $this->section_id = $section_id;
46 $this->user_id = $_postdata['userId'];
47
48 // get from URL with file_id
49 $lg_text_url = $this->get_text_from_sectionId_url.$section_id;
50 $lg_text = file_get_contents($lg_text_url);
51
52 $stringInput = $lg_text;
53 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput);
54 $stringInput = preg_replace("/ /u", "○", $stringInput);
55 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
56 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
57 $lg_text = $stringInput;
58
59 $this->lg_text = $lg_text;
60 }
61
62 public function GetTextFromLocal($_id){
63 $this->section_id = $_id;
64 echo "DEBUG: from my local"."<br>";
65 $this->lg_text = $this->GetSectionContent();
66
67 }
68
69 public function GetInfoFromPreviousPage($_postdata) {
70
71 if ($_postdata['fileId']) {
72 $this->file_id = $_postdata['fileId'];
73 }
74 if ($_postdata['sectionId']) {
75 $this->section_id = $_postdata['sectionId'];
76 }
77 if ($_postdata['text']) {
78 $this->lg_text = $_postdata['text'];
79 }
80 if ($_postdata['branchId']) {
81 $this->branch_id = $_postdata['branchId'];
82 }
83 if ($_postdata['userId']) {
84 $this->user_id = $_postdata['userId'];
85 }
86
87
88 }
89 public function InitData($_postdata) {
90 $file_id = $_postdata['fileId'];
91 $branch_id = $_postdata['branchId'];
92 $section_id = $_postdata['sectionId'];
93
94 $this->branch_id = $branch_id;
95 $this->file_id = $file_id;
96 $this->user_id = $_postdata['userId'];
97 $this->section_id = $section_id;
98
99 if ($file_id != 0 && $branch_id != 0) {
100 // get from URL with file_id
101 $lg_text_url = $this->get_text_from_fileId_url.$file_id;
102 } else if ($section_id != 0) {
103 // get from URL with section_id
104 $lg_text_url = $this->get_text_from_sectionId_url.$section_id;
105 } else {
106 echo "wrong url!!";
107 return;
108 }
109
110
111 $lg_text = file_get_contents($lg_text_url);
112
113 /*
114 $stringInput = preg_replace("/ /u", "○", $stringInput);
115 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
116 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
117 */
118
119 // the text is from database
120 $stringInput = $lg_text;
121 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput);
122 $stringInput = preg_replace("/ /u", "○", $stringInput);
123 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
124 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
125 $lg_text = $stringInput;
126
127 $this->lg_text = $lg_text;
128
129
130 }
131
19 // === for tagging === 132 // === for tagging ===
20 public function StartTagging($urlvalues) { 133 public function StartTagging() {
21 134
22 $this->Initialize($urlvalues); 135 // $this->Initialize($urlvalues);
23 136
24 $section_id = $this->GetSectionId(); 137 $section_id = $this->section_id;
25 138 $stringInput = $this->lg_text;
26 $stringInput = $this->GetSectionContent(); 139
27 140
28 //$taglistArray = $this->GetTaglistArray(); 141 //$taglistArray = $this->GetTaglistArray();
29 //for GetTaglistByTopicID: 142 //for GetTaglistByTopicID:
30
31 $taglistArray = $this->GetTaglistByTopicID($this->GetTopic()); 143 $taglistArray = $this->GetTaglistByTopicID($this->GetTopic());
32 144 // topic list
33 // topic list?
34 $topiclistArray = $this->GetTopiclistArray(); 145 $topiclistArray = $this->GetTopiclistArray();
35 146
36 147
37 $wordlistArray = $this->GetWordlistArray(); 148 $wordlistArray = $this->GetWordlistArray();
38 149
43 $data['section_id'] = $section_id; 154 $data['section_id'] = $section_id;
44 $data['topiclistArray'] = $topiclistArray; 155 $data['topiclistArray'] = $topiclistArray;
45 $data['default_topic_id'] = $this->GetTopic(); 156 $data['default_topic_id'] = $this->GetTopic();
46 $data['topic_tag'] = $this->GetTopicTag($this->GetTopic()); 157 $data['topic_tag'] = $this->GetTopicTag($this->GetTopic());
47 158
159 $data['info'] = array('file_id'=>$this->file_id, 'user_id'=>$this->user_id, 'branch_id'=>$this->branch_id, 'section_id'=>$this->section_id);
160
48 return $data; 161 return $data;
49 162
50 } 163 }
51 164
52 public function SaveFullText($postdata) { 165 public function SaveFullText($postdata) {
71 file_put_contents("data/parsing_files/".$postdata['filename'].".txt", $require); 184 file_put_contents("data/parsing_files/".$postdata['filename'].".txt", $require);
72 } 185 }
73 186
74 } 187 }
75 188
189 public function UpdateInfoByResonseFromLGService($response) {
190
191 $response_file = $response["file"];
192 $response_branch = $response["branch"];
193
194 $this->file_id = (string)$response_file["id"]; // should updated
195 echo "file_id: ".$this->file_id;
196 echo "<br>";
197 $this->branch_id = (string)$response_branch["id"];
198 echo "branch_id: ".$this->branch_id;
199 echo "<br>";
200
201 }
202 public function SaveFullTextToLGService($_postdata) {
203 // save tagged text (full text) by Jorge's API to lg service
204
205 // --------
206 // saving in my local machine in developing phrase
207 if ($_postdata['text']){
208 $date = date('Y_m_d_H_i_s', time());
209 if ( file_exists("data/parsing_files/".$_postdata['sectionId'].".txt") ) {
210 $oldFile = file_get_contents("data/parsing_files/".$_postdata['sectionId'].".txt");
211 file_put_contents("data/parsing_files/".$_postdata['sectionId']."_".$date.".txt", $oldFile);
212 }
213
214 if (get_magic_quotes_gpc()) {
215 $require = stripslashes($_postdata['text']);
216 } else {
217 $require = $_postdata['text'];
218 }
219
220 $require = preg_replace("/【<a(.*?)>(.*?)<\/a>】/u", "【\\2】", $require);
221 $require = preg_replace('/&amp;/u', "&", $require);
222 $require = preg_replace("/○/u", " ", $require);
223 $require = preg_replace("/<br>/u", "\n", $require);
224 $require = preg_replace("/<br>/u", "\n", $require);
225 file_put_contents("data/parsing_files/".$_postdata['sectionId'].".txt", $require);
226 }
227
228 echo getcwd()."<br>";
229 // ------
230
231
232 if ($_postdata['branchId'] != 0) {
233 // exiting branch case
234 $postfields = array(
235 //"text" => $_postdata['text'],
236 "text" => $require,
237 "branchId" => $_postdata['branchId'],
238 "userId" => $_postdata['userId'],
239 "userPreviousFileId" => $_postdata['fileId'],
240 );
241 $save_url = $this->save_to_LGService_url;
242
243 } else {
244 // new branch case
245 echo "saveNew!";
246 $postfields = array(
247 //"text" => $_postdata['text'],
248 "text" => $require,
249 "sectionId" => $_postdata['sectionId'],
250 //"userId" => $_postdata['userId'],
251 // TODO: change userId when we can query by sectionId from LGService using search
252 "userId" => "12",
253 // "label" => $_postdata['label'],
254 // TODO: ask user for lable
255 "label" => "label for section ".$_postdata['sectionId'],
256 );
257 $save_url = $this->save_new_to_LGService_url;
258 }
259
260
261
262
263 // set up the curl resource
264 $ch = curl_init();
265 curl_setopt($ch, CURLOPT_URL, $save_url);
266 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
267 curl_setopt($ch, CURLOPT_CUSTOMREQUEST, "POST");
268 curl_setopt($ch, CURLOPT_POST, true);
269 curl_setopt($ch, CURLOPT_POSTFIELDS, $postfields);
270 //curl_setopt($ch, CURLOPT_HEADER, true);
271 curl_setopt($ch, CURLOPT_HTTPHEADER, array(
272 'Content-type: multipart/form-data;charset=utf-8'
273 ));
274
275 // execute the request
276
277 $output = curl_exec($ch);
278
279 // output the profile information - includes the header
280 //echo($output) . PHP_EOL;
281
282 // close curl resource to free up system resources
283 curl_close($ch);
284
285 $response = json_decode($output, true);
286 return $response;
287
288 }
76 289
77 // === for export table === 290 // === for export table ===
78 public function ExportTable($urlvalues, $postdata) { 291 public function ExportTable($urlvalues, $postdata) {
79 $this->Initialize($urlvalues); 292 // $this->Initialize($urlvalues);
80 $content = $postdata['content']; 293 $content = $postdata['content'];
81 $topic = $postdata['topic']; 294 $topic = $postdata['topic'];
82 295
83 // outputTableArray: 296 // outputTableArray:
84 $section_id = $this->GetSectionId(); 297 // $section_id = $this->GetSectionId();
298 $section_id = $this->section_id;
299 if (!$section_id) {
300 //$this->Initialize($urlvalues);
301 $section_id = $this->GetSectionId();
302 }
85 303
86 $section_info = $this->GetSectionInfo(); 304 $section_info = $this->GetSectionInfo();
87 305
88 $sectionName = $section_info['sectionName']; 306 $sectionName = $section_info['sectionName'];
89 $bookId = $section_info['bookId']; 307 $bookId = $section_info['bookId'];
193 411
194 } 412 }
195 413
196 // === for manage tag list === 414 // === for manage tag list ===
197 public function EditTaglist($urlvalues) { 415 public function EditTaglist($urlvalues) {
198 $this->Initialize($urlvalues); 416 // $this->Initialize($urlvalues);
199 $result = $this->GetTaglist(); 417 $result = $this->GetTaglist();
200 $taglistArray = array(); 418 $taglistArray = array();
201 while ($row = mysql_fetch_assoc($result)) { 419 while ($row = mysql_fetch_assoc($result)) {
202 $taglistArray[$row['id']] = array($row['name'], $row['tag'], $row['color']); 420 $taglistArray[$row['id']] = array($row['name'], $row['tag'], $row['color']);
203 } 421 }
260 } 478 }
261 479
262 } 480 }
263 481
264 // === for config topic === 482 // === for config topic ===
265 public function ConfigTagsInTopic($urlvalues, $postdata) { 483 public function ConfigTagsInTopic($postdata) {
266 484
485 $query = "SELECT taglist.*, topic_tag_relation.topic_id FROM taglist LEFT JOIN topic_tag_relation ON taglist.id = topic_tag_relation.tag_id ORDER BY `topic_id`";
486 $result = mysql_query($query);
487 if (!$result) {
488 return json_encode("Failed during selecting/joining taglist and topic_tag_relation table.");
489 }
490 $taglistArray = array();
491 while ($row = mysql_fetch_assoc($result)) {
492 array_push($taglistArray, array('id'=>$row['id'],'name'=>$row['name'], 'tag'=>$row['tag'], 'color'=>$row['color'], 'topic_id'=>$row['topic_id']) );
493 }
494
495 $topic = $postdata['topic'];
496 $result = $this->GetTopicByID($topic);
497 $row = mysql_fetch_assoc($result);
498 $topic_name = $row['name'];
499
500 $data = array();
501 $data['taglistArray'] = $taglistArray;
502 $data['topic'] = $topic;
503 $data['topic_name'] = $topic_name;
504
505 return $data;
267 } 506 }
268 507
269 private function GetTaglistByTopicID($topic_id) { 508 private function GetTaglistByTopicID($topic_id) {
270 $taglistArray = array(); 509 $taglistArray = array();
271 // select taglist ids from topic_tag_relation table 510 // select taglist ids from topic_tag_relation table
324 563
325 // =========================== 564 // ===========================
326 565
327 // === for manage wordlist === 566 // === for manage wordlist ===
328 public function EditWordlist($urlvalues) { 567 public function EditWordlist($urlvalues) {
329 $this->Initialize($urlvalues); 568 //$this->Initialize($urlvalues);
330 $result = $this->GetWordlist(); 569 $result = $this->GetWordlist();
331 $wordlistArray = array(); 570 $wordlistArray = array();
332 while ($row = mysql_fetch_assoc($result)) { 571 while ($row = mysql_fetch_assoc($result)) {
333 $wordlistArray[$row['id']] = $row['name']; 572 $wordlistArray[$row['id']] = $row['name'];
334 } 573 }
403 642
404 public function sortFunction($a,$b) { 643 public function sortFunction($a,$b) {
405 return strlen($b)-strlen($a); 644 return strlen($b)-strlen($a);
406 } 645 }
407 646
408 647 // TODO: delete this
409 private function SetSectionId($_urlvalues) { 648 private function SetSectionId($_urlvalues) {
410 // TODO: maybe get user info also 649 // TODO: maybe get user info also
411 // get book id from url 650 // get book id from url
412 if ($_urlvalues['id'] != "") { 651 if ($_urlvalues['id'] != "") {
413 $section_id = $_urlvalues['id']; 652 $section_id = $_urlvalues['id'];
423 $this->section_id = $section_id; 662 $this->section_id = $section_id;
424 663
425 } 664 }
426 665
427 private function GetSectionId() { 666 private function GetSectionId() {
428 if ($this->section_id) { 667
668 if (is_numeric($this->section_id)) {
429 return $this->section_id; 669 return $this->section_id;
430 } else { 670 } else {
431 return json_encode("Error: No section id"); 671 return json_encode("Error: No section id");
432 } 672 }
433 } 673 }
452 $data['bookId'] = $bookId; 692 $data['bookId'] = $bookId;
453 $data['startPage'] = $startPage; 693 $data['startPage'] = $startPage;
454 $data['endPage'] = $endPage; 694 $data['endPage'] = $endPage;
455 $data['sectionName'] = $sectionName; 695 $data['sectionName'] = $sectionName;
456 696
697
457 return $data; 698 return $data;
458 } 699 }
459 700
460 private function GetSectionContent() { 701 private function GetSectionContent() {
461 $section_id = $this->GetSectionId(); 702 $section_id = $this->GetSectionId();
462 $section_info = $this->GetSectionInfo(); 703 $section_info = $this->GetSectionInfo();
463 704
705 var_dump($section_id);
706 var_dump($section_info);
707
464 $bookId = $section_info['bookId']; 708 $bookId = $section_info['bookId'];
465 $startPage = $section_info['startPage']; 709 $startPage = $section_info['startPage'];
466 $endPage = $section_id['endPage']; 710 $endPage = $section_info['endPage'];
467
468 711
469 $contentString=""; 712 $contentString="";
470 $data_path = $this->GetDataPath(); 713 $data_path = $this->GetDataPath();
471 if ( file_exists($data_path."parsing_files/".$section_id.".txt") ) { 714 if ( file_exists($data_path."parsing_files/".$section_id.".txt") ) {
472 $contentString=file_get_contents($data_path."parsing_files/".$section_id.".txt"); 715 $contentString=file_get_contents($data_path."parsing_files/".$section_id.".txt");
473 $stringInput = $contentString; 716 $stringInput = $contentString;
717
718 // if the text is from file system
474 $stringInput = preg_replace("/ /u", "○", $stringInput); 719 $stringInput = preg_replace("/ /u", "○", $stringInput);
475 $stringInput = preg_replace("/\n/u", "<br>", $stringInput); 720 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
476 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput); 721 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
477 } else { 722 } else {
478 $query = sprintf("SELECT `content`, `line`, `books_id` FROM `contents` WHERE `books_id`=\"%s\" AND `line`>=%d AND `line`<=%d", $bookId, $startPage, $endPage); 723 $query = sprintf("SELECT `content`, `line`, `books_id` FROM `contents` WHERE `books_id`=\"%s\" AND `line`>=%d AND `line`<=%d", $bookId, $startPage, $endPage);
479 $result = mysql_query($query); 724 $result = mysql_query($query);
725 if (!$result) {
726 return json_encode("Failed during selecting content table.");
727 }
480 while ($row = mysql_fetch_assoc($result)) { 728 while ($row = mysql_fetch_assoc($result)) {
481 $contentString.="【".$row['line']."】".$row['content']."\n"; 729 $contentString.="【".$row['line']."】".$row['content']."\n";
482 } 730 }
731
732 // the text is from database
483 $stringInput = $contentString; 733 $stringInput = $contentString;
484 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput); 734 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput);
485 $stringInput = preg_replace("/ /u", "○", $stringInput); 735 $stringInput = preg_replace("/ /u", "○", $stringInput);
486 $stringInput = preg_replace("/\n/u", "<br>", $stringInput); 736 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
487 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput); 737 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);