Mercurial > hg > extraction-interface

comparison models/extractapp.php @ 51:840cdb52f476 extractapp

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
add book_meta into file
author Zoe Hong <zhong@mpiwg-berlin.mpg.de>
date Thu, 26 Mar 2015 15:42:50 +0100
parents c98a0c6d7eb4
children 94064f625650
comparison
equal deleted inserted replaced
50:3e42a63de0ea 51:840cdb52f476
31 $this->book_name = $_postdata['bookName']; 31 $this->book_name = $_postdata['bookName'];
32 32
33 // get from URL with file_id 33 // get from URL with file_id
34 $lg_text_url = $this->get_text_from_fileId_url.$this->file_id; 34 $lg_text_url = $this->get_text_from_fileId_url.$this->file_id;
35 //$lg_text = file_get_contents($lg_text_url); 35 //$lg_text = file_get_contents($lg_text_url);
36 // --- parsing meta data 36 // --- get/set text and parsing meta data from text file
37 $lg_text = $this->ParseMetaData($lg_text_url); 37 $lg_text = $this->ParseMetaData($lg_text_url);
38 // TODO: maybe need to check if the book_meta is updated. compare the book_meta in text file and from _post (most up-to-date)
38 39
39 // ---- 40 // ----
40 41
41 42
42 $stringInput = $lg_text; 43 $stringInput = $lg_text;
59 $this->book_name = $_postdata['bookName']; 60 $this->book_name = $_postdata['bookName'];
60 61
61 // get from URL with file_id 62 // get from URL with file_id
62 $lg_text_url = $this->get_text_from_sectionId_url.$section_id; 63 $lg_text_url = $this->get_text_from_sectionId_url.$section_id;
63 $lg_text = file_get_contents($lg_text_url); 64 $lg_text = file_get_contents($lg_text_url);
65
66 // TODO: get book_meta from $_postdata and set $this->book_meta: book_id,book_name,author,year,pagenumber
67 $book_meta = array();
68 array_push($book_meta, array($this->book_id,$this->book_name,"","","")); // missing author,year,pagenumber
69
70 $this->book_meta = $book_meta;
71
72
64 73
65 $stringInput = $lg_text; 74 $stringInput = $lg_text;
66 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput); 75 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput);
67 $stringInput = preg_replace("/ /u", "○", $stringInput); 76 $stringInput = preg_replace("/ /u", "○", $stringInput);
68 $stringInput = preg_replace("/\n/u", "<br>", $stringInput); 77 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
355 $text .= "<topic>".$this->topic."</topic>\n"; 364 $text .= "<topic>".$this->topic."</topic>\n";
356 // book meta data 365 // book meta data
357 $book = $this->book_meta; 366 $book = $this->book_meta;
358 foreach ($book as $b) { 367 foreach ($book as $b) {
359 $text .= "<book>\n"; 368 $text .= "<book>\n";
360 $text .= "<title>".$b[0]."</title>\n"; 369 $text .= "<id>".$b[0]."</id>\n";
361 $text .= "<author>".$b[1]."</author>\n"; 370 $text .= "<name>".$b[1]."</name>\n";
362 $text .= "<year>".$b[2]."</year>\n"; 371 $text .= "<author>".$b[2]."</author>\n";
363 $text .= "<pagenumber>".$b[3]."</pagenumber>\n"; 372 $text .= "<year>".$b[3]."</year>\n";
373 $text .= "<pagenumber>".$b[4]."</pagenumber>\n";
364 $text .= "</book>\n"; 374 $text .= "</book>\n";
365 } 375 }
366 // taglist 376 // taglist
367 $taglist = $this->taglist_infile; 377 $taglist = $this->taglist_infile;
368 foreach ($taglist as $tagitem) { 378 foreach ($taglist as $tagitem) {
380 390
381 return $text; 391 return $text;
382 } 392 }
383 public function SaveFullTextToLGService($_postdata) { 393 public function SaveFullTextToLGService($_postdata) {
384 // save tagged text (full text) by Jorge's API to lg service 394 // save tagged text (full text) by Jorge's API to lg service
385
386 // -------- 395 // --------
387 if ($_postdata['text']){ 396 if ($_postdata['text']){
388 $date = date('Y_m_d_H_i_s', time()); 397 $date = date('Y_m_d_H_i_s', time());
389 if ( file_exists("data/parsing_files/".$_postdata['sectionId'].".txt") ) { 398 if ( file_exists("data/parsing_files/".$_postdata['sectionId'].".txt") ) {
390 $oldFile = file_get_contents("data/parsing_files/".$_postdata['sectionId'].".txt"); 399 $oldFile = file_get_contents("data/parsing_files/".$_postdata['sectionId'].".txt");
706 $result = mysql_query($query); 715 $result = mysql_query($query);
707 if (!$result) { 716 if (!$result) {
708 echo json_encode(mysql_error()); 717 echo json_encode(mysql_error());
709 } 718 }
710 $row = mysql_fetch_assoc($result); 719 $row = mysql_fetch_assoc($result);
711 $largest_id = $row['AUTO_INCREMENT']; 720 $largest_id = $row['AUTO_INCREMENT']-1;
712 721
713 722
714 $topic_id = $_postdata['topic_id']; 723 $topic_id = $_postdata['topic_id'];
715 $result = $this->GetTaglistByTopicID($topic_id); 724 $result = $this->GetTaglistByTopicID($topic_id);
716 725
1215 } 1224 }
1216 if ($taglistArray) { 1225 if ($taglistArray) {
1217 $this->taglist_infile = $taglistArray; 1226 $this->taglist_infile = $taglistArray;
1218 } 1227 }
1219 1228
1220 // get book meta data 1229 // get book meta data from file
1221 $book_meta = $xml->book; 1230 $book_meta = $xml->book;
1222 $book_metaArray = array(); 1231 $book_metaArray = array();
1223 foreach ($book_meta as $row) { 1232 foreach ($book_meta as $row) {
1224 //array_push($book_metaArray, array('title'=>(string)$row->title,'author'=>(string)$row->author,'year'=>(string)$row->year,'pagenumber'=>(string)$row->pagenumber )); 1233 array_push($book_metaArray, array((string)$row->id,(string)$row->name,(string)$row->author,(string)$row->year,(string)$row->pagenumber ));
1225 array_push($book_metaArray, array((string)$row->title,(string)$row->author,(string)$row->year,(string)$row->pagenumber ));
1226 } 1234 }
1227 if ($book_metaArray) { 1235 if ($book_metaArray) {
1228 $this->book_meta = $book_metaArray; 1236 $this->book_meta = $book_metaArray;
1229 } 1237 }
1230 1238
1272 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput); 1280 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput);
1273 $stringInput = preg_replace("/ /u", "○", $stringInput); 1281 $stringInput = preg_replace("/ /u", "○", $stringInput);
1274 $stringInput = preg_replace("/\n/u", "<br>", $stringInput); 1282 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
1275 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput); 1283 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
1276 } 1284 }
1285
1286 // get book_meta from books table
1287 $book_meta = array();
1288 $books_result = $this->GetBooksByID($bookId);
1289 while ($row = mysql_fetch_assoc($books_result)) {
1290 array_push($book_meta, array($row['id'],$row['name'],$row['author'],(string)$row['start_year'],(string)$row['line']));
1291 // use 'start_year' as year, 'line' is pagenumber
1292 }
1293
1294 $this->book_meta = $book_meta;
1295
1277 1296
1278 return $stringInput; 1297 return $stringInput;
1279 } 1298 }
1280 1299
1281 private function GetDataPath() { 1300 private function GetDataPath() {