Mercurial > hg > extraction-interface

comparison develop/models/extractapp.php @ 6:63e08b98032f

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
rewrite extraction interface into PHP MVC architecture. (Although js hasn't been rewritten into MVC, it's fitted into the current PHP MVC architecture.) - The root of the new PHP MVC is at 'develop/'. - extraction interface is called "Extractapp" with several action, eg TaggingText, EditWordlist, EditTaglist, ExportTable.
author Zoe Hong <zhong@mpiwg-berlin.mpg.de>
date Thu, 05 Feb 2015 16:07:53 +0100
parents
children 584b1623e9ef
comparison
equal deleted inserted replaced
5:cbbb7ef22394 6:63e08b98032f
1
2
3 <?php
4
5 class ExtractappModel extends BaseModel{
6
7 public function Index() {
8 return array("Index Value 1", "Value 2", "Value 3");
9 }
10
11 protected $section_id, $data_path;
12
13
14 private function Initialize($_urlvalues) {
15 $this->SetSectionId($_urlvalues);
16 }
17
18 // === for tagging ===
19 public function StartTagging($urlvalues) {
20
21 $this->Initialize($urlvalues);
22
23 $section_id = $this->GetSectionId();
24
25 $stringInput = $this->GetSectionContent();
26
27 $taglistArray = $this->GetTaglistArray();
28
29 $wordlistArray = $this->GetWordlistArray();
30
31 $data = array();
32 $data['stringInput'] = $stringInput;
33 $data['taglistArray'] = $taglistArray;
34 $data['wordlistArray'] = $wordlistArray;
35 $data['section_id'] = $section_id;
36
37
38 return $data;
39
40 }
41
42 public function SaveFullText($postdata) {
43 if ($postdata['text']){
44 $date = date('Y_m_d_H_i_s', time());
45 if ( file_exists("data/parsing_files/".$postdata['filename'].".txt") ) {
46 $oldFile = file_get_contents("data/parsing_files/".$postdata['filename'].".txt");
47 file_put_contents("data/parsing_files/".$postdata['filename']."_".$date.".txt", $oldFile);
48 }
49
50 if (get_magic_quotes_gpc()) {
51 $require = stripslashes($postdata['text']);
52 } else {
53 $require = $postdata['text'];
54 }
55
56 $require = preg_replace("/【<a(.*?)>(.*?)<\/a>】/u", "【\\2】", $require);
57 $require = preg_replace('/&amp;/u', "&", $require);
58 $require = preg_replace("/○/u", " ", $require);
59 $require = preg_replace("/<br>/u", "\n", $require);
60 $require = preg_replace("/<br>/u", "\n", $require);
61 file_put_contents("data/parsing_files/".$postdata['filename'].".txt", $require);
62 }
63
64 }
65
66
67
68 // === for export table ===
69 public function ExportTable($urlvalues, $postdata) {
70 $this->Initialize($urlvalues);
71 $content = $postdata['content'];
72
73
74 // outputTableArray:
75 $section_id = $this->GetSectionId();
76
77 $section_info = $this->GetSectionInfo();
78
79 $sectionName = $section_info['sectionName'];
80 $bookId = $section_info['bookId'];
81
82 $books_info = $this->GetBooksInfo($bookId);
83 $bookName = $books_info['bookName'];
84
85 $outputTableArray = array();
86
87 $taglistArray = $this->GetTaglistArray();
88
89 $outputTableArray[0]=array();
90 $outputTableArray[0][0]=array();
91 $outputTableArray[0][1]=array();
92 foreach ( $taglistArray as $value ) {
93 $outputTableArray[0][0][$value[2]] = $value[1];
94 $outputTableArray[0][1][$value[2]] = $value[1]."(Title)";
95 }
96 $outputTableArray[0]["other"] = "其他";
97 $outputTableArray[0]["page"] = "頁數";
98 $outputTableArray[0]["full"] = "全文";
99
100 foreach ( $taglistArray as $tagValue ) {
101 $content = preg_replace("/<\/".$tagValue[2].">○*<".$tagValue[2].">/u", "", $content);
102 $content = preg_replace("/<".$tagValue[2].">[ ]*<\/".$tagValue[2].">/u", "", $content);
103 }
104
105 $contentLineArray = explode( "<br>", $content );
106
107 $count=0;
108 $pageNow=NULL;
109 foreach ( $contentLineArray as $value ) {
110 $count++;
111 $recordString = $value;
112 $otherString = $recordString;
113 //echo $recordString."<br>\n";
114 if ( preg_match("/【<a(.*?)>(.*?)<\/a>】/u", $recordString, $matches) ) {
115 $pageNow = $matches[2];
116 }
117 foreach ( $taglistArray as $tagValue ) {
118 if ( preg_match_all("/<".$tagValue[2].">(.*?)<\/".$tagValue[2].">/u", $recordString, $matches, PREG_SET_ORDER) ) {
119 foreach ( $matches as $matchesValue ) {
120 $matchesValue[1] = preg_replace("/○/u", "", $matchesValue[1]);
121 if ( preg_match_all("/〈(.*?)〉/u", $matchesValue[1], $matches2, PREG_SET_ORDER) ) {
122 foreach ( $matches2 as $matches2Value ) {
123 if ( isset($outputTableArray[$count][0][$tagValue[2]]) ) {
124 $outputTableArray[$count][0][$tagValue[2]] .= ";".$matches2Value[1];
125 } else {
126 $outputTableArray[$count][0][$tagValue[2]] = $matches2Value[1];
127 }
128 }
129 } else {
130 if ( isset($outputTableArray[$count][0][$tagValue[2]]) ) {
131 $outputTableArray[$count][0][$tagValue[2]] .= ";".$matchesValue[1];
132 } else {
133 $outputTableArray[$count][0][$tagValue[2]] = $matchesValue[1];
134 }
135 }
136 }
137 $otherString = preg_replace("/<".$tagValue[2].">(.*?)<\/".$tagValue[2].">/u", "　", $otherString);
138 }
139 }
140 $otherString = preg_replace("/○/u", "", $otherString);
141 $outputTableArray[$count]["other"] = $otherString;
142 $outputTableArray[$count]["page"] = $pageNow;
143 $value = preg_replace("/>/u", "&gt;", $value);
144 $value = preg_replace("/</u", "&lt;", $value);
145 $outputTableArray[$count]["full"] = $value;
146 }
147
148 foreach ( $outputTableArray as $arrayIndex => $arrayValue ) {
149 if ( !isset($arrayValue[0]["person"]) ) {
150 unset($outputTableArray[$arrayIndex]);
151 }
152 }
153
154
155 $data = array();
156
157 $data['outputTableArray'] = $outputTableArray;
158 $data['bookId'] = $bookId;
159 $data['section_id'] = $section_id;
160 $data['bookName'] = $bookName;
161 $data['sectionName'] = $sectionName;
162
163 return $data;
164
165 }
166
167 // === for manage tag list ===
168 public function EditTaglist($urlvalues) {
169 $this->Initialize($urlvalues);
170 $result = $this->GetTaglist();
171 $taglistArray = array();
172 while ($row = mysql_fetch_assoc($result)) {
173 $taglistArray[$row['id']] = array($row['name'], $row['tag'], $row['color']);
174 }
175
176 $data = array();
177 $data['taglistArray'] = $taglistArray;
178 return $data;
179
180 }
181
182 private function _GetTag($_postdata) {
183 if (get_magic_quotes_gpc()) {
184 $id = stripslashes($_postdata['id']);
185 $name = stripslashes($_postdata['name']);
186 $tag = stripslashes($_postdata['tag']);
187 $color = stripslashes($_postdata['color']);
188 } else {
189 $id = $_postdata['id'];
190 $name = $_postdata['name'];
191 $tag = $_postdata['tag'];
192 $color = $_postdata['color'];
193 }
194 return array($id, $name, $tag, $color);
195 }
196
197 public function NewTagElement($postdata) {
198 if ($postdata['id']){
199 list($id, $name, $tag, $color) = $this->_GetTag($postdata);
200
201 $query1 = sprintf("INSERT INTO `taglist` (`id`, `name`, `tag`, `color`, `systemName`) VALUES (%s, %s, %s, %s, %s)",
202 $this->GetSQLValueString($id, "int"),
203 $this->GetSQLValueString($name, "text"),
204 $this->GetSQLValueString($tag, "text"),
205 $this->GetSQLValueString($color, "text"),
206 $this->GetSQLValueString($this->systemNAME, "text"));
207 $result1 = mysql_query($query1);
208 }
209
210 }
211
212 public function SaveTagElement($postdata) {
213 if ($postdata['id']){
214 list($id, $name, $tag, $color) = $this->_GetTag($postdata);
215
216 $queryInsert = sprintf("UPDATE taglist SET `name`=%s, `tag`=%s, `color`=%s WHERE `id`=%s",
217 $this->GetSQLValueString($name, "text"),
218 $this->GetSQLValueString($tag, "text"),
219 $this->GetSQLValueString($color, "text"),
220 $this->GetSQLValueString($id, "int"));
221 $resultInsert = mysql_query($queryInsert);
222
223 }
224
225 }
226
227 public function DeleteTag($postdata) {
228 if ($postdata['id']) {
229 $queryInsert = sprintf("DELETE FROM `taglist` WHERE `id` = %s", stripslashes($postdata['id']));
230 $resultInsert = mysql_query($queryInsert);
231 }
232
233 }
234 // ===========================
235
236 // === for manage wordlist ===
237 public function EditWordlist($urlvalues) {
238 $this->Initialize($urlvalues);
239 $result = $this->GetWordlist();
240 $wordlistArray = array();
241 while ($row = mysql_fetch_assoc($result)) {
242 $wordlistArray[$row['id']] = $row['name'];
243 }
244
245 $data = array();
246 $data['wordlistArray'] = $wordlistArray;
247 return $data;
248 }
249
250
251 public function AddNewList($postdata) {
252 if ($postdata['text']){
253 if (get_magic_quotes_gpc()) {
254 $name = stripslashes($postdata['text']);
255 } else {
256 $name = $postdata['text'];
257 }
258
259 $query1 = sprintf("INSERT INTO `wordlist` (`name`, `systemName`) VALUES (%s, %s)",
260 $this->GetSQLValueString($name, "text"),
261 $this->GetSQLValueString($this->systemNAME, "text"));
262 $result1 = mysql_query($query1);
263 file_put_contents( "data/wordlist/".mysql_insert_id().".txt", "(empty now)");
264 }
265 }
266
267
268 public function SaveWordlist($postdata) {
269
270 if ($postdata['text']){
271 $date = date('Y_m_d_H_i_s', time());
272 if ( file_exists("data/wordlist/".$postdata['filename'].".txt") ) {
273 $filename = "data/wordlist/".$postdata['filename']."_".$date.".txt";
274 $oldFile = file_get_contents("data/wordlist/".$postdata['filename'].".txt");
275 file_put_contents($filename, $oldFile);
276 } else {
277 $filename = "data/wordlist/".$postdata['filename'].".txt";
278 }
279
280
281 if (get_magic_quotes_gpc()) {
282 $require = stripslashes($postdata['text']);
283 } else {
284 $require = $postdata['text'];
285 }
286
287 $require = preg_replace("/<br>/u", "<br>", $require);
288 file_put_contents($filename, $require);
289 }
290 /*
291 if ($postdata['text']){
292 $date = date('Y_m_d_H_i_s', time());
293 if ( file_exists("data/wordlist/".$postdata['filename'].".txt") ) {
294 $oldFile = file_get_contents("data/wordlist/".$postdata['filename'].".txt");
295 file_put_contents("data/wordlist/".$postdata['filename']."_".$date.".txt", $oldFile);
296 }
297
298 if (get_magic_quotes_gpc()) {
299 $require = stripslashes($postdata['text']);
300 } else {
301 $require = $postdata['text'];
302 }
303
304 // $require = preg_replace("/<br>/u", "", $require);
305 file_put_contents("data/wordlist/".$postdata['filename'].".txt", $require);
306 }
307 */
308 }
309
310
311 // =======================================
312
313 public function sortFunction($a,$b) {
314 return strlen($b)-strlen($a);
315 }
316
317
318 private function SetSectionId($_urlvalues) {
319 // TODO: maybe get user info also
320 // get book id from url
321 if ($_urlvalues['id'] != "") {
322 $section_id = $_urlvalues['id'];
323 } else {
324 return json_encode("Error: No section id");
325 /* ???? */
326 /*
327 $get_book_id = $urlvalues['book'];
328 $get_start = $urlvalues['start'];
329 $get_end = $urlvalues['end'];
330 */
331 }
332 $this->section_id = $section_id;
333
334 }
335
336 private function GetSectionId() {
337 if ($this->section_id) {
338 return $this->section_id;
339 } else {
340 return json_encode("Error: No section id");
341 }
342 }
343
344 private function GetSectionInfo() {
345 $section_id = $this->GetSectionId();
346 if (!is_numeric($section_id)){
347 return $section_id;
348 }
349
350 $result = $this->GetSectionsByID($section_id);
351
352
353 while ($row = mysql_fetch_assoc($result)) {
354 $bookId=$row['books_id'];
355 $startPage=$row['start_page'];
356 $endPage=$row['end_page'];
357 $sectionName = $row['name'];
358 }
359
360 $data = array();
361 $data['bookId'] = $bookId;
362 $data['startPage'] = $startPage;
363 $data['endPage'] = $endPage;
364 $data['sectionName'] = $sectionName;
365
366 return $data;
367 }
368
369 private function GetSectionContent() {
370 $section_id = $this->GetSectionId();
371 $section_info = $this->GetSectionInfo();
372
373 $bookId = $section_info['bookId'];
374 $startPage = $section_info['startPage'];
375 $endPage = $section_id['endPage'];
376
377
378 $contentString="";
379 $data_path = $this->GetDataPath();
380 if ( file_exists($data_path."parsing_files/".$section_id.".txt") ) {
381 $contentString=file_get_contents($data_path."parsing_files/".$section_id.".txt");
382 $stringInput = $contentString;
383 $stringInput = preg_replace("/ /u", "○", $stringInput);
384 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
385 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
386 } else {
387 $query = sprintf("SELECT `content`, `line`, `books_id` FROM `contents` WHERE `books_id`=\"%s\" AND `line`>=%d AND `line`<=%d", $bookId, $startPage, $endPage);
388 $result = mysql_query($query);
389 while ($row = mysql_fetch_assoc($result)) {
390 $contentString.="【".$row['line']."】".$row['content']."\n";
391 }
392 $stringInput = $contentString;
393 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput);
394 $stringInput = preg_replace("/ /u", "○", $stringInput);
395 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
396 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
397 }
398
399 return $stringInput;
400 }
401
402 private function GetDataPath() {
403 return getcwd()."/data/"; // get the current file path, which is getcwd(), and concatenate with "/data/"
404 }
405
406 private function GetTaglistArray() {
407 $taglistArray="";
408 $result = $this->GetTaglist();
409 while ($row = mysql_fetch_assoc($result)) {
410 $taglistArray[] = array( $row['id'], $row['name'], $row['tag'], $row['color'] );
411
412 }
413
414 return $taglistArray;
415
416 }
417
418 private function GetWordlistArray() {
419 $wordlistArray="";
420 $result = $this->GetWordlist();
421 while ($row = mysql_fetch_assoc($result)) {
422 $listString = file_get_contents("wordlist/".$row['id'].".txt");
423 $listString = preg_replace("/<div>/u", "\n", $listString);
424 $listString = preg_replace("/<\/div>/u", "", $listString);
425 $listString = preg_replace("/<span(.*?)>/u", "", $listString);
426 $listString = preg_replace("/<\/span>/u", "", $listString);
427 //$listString = preg_replace("/\n/u", "|", $listString);
428
429 $wordlistArray2 = explode( "\n", $listString );
430 usort($wordlistArray2,'sortFunction');
431 foreach ( $wordlistArray2 as $index=>$value ) {
432 $wordlistArray2[$index] = implode("○?", preg_split("/(?<!^)(?!$)/u", $value));
433 }
434 foreach ( $wordlistArray2 as $index=>$value ) {
435 if ($value=="") unset($wordlistArray2[$index]);
436
437 }
438 $listString = implode("|", $wordlistArray2);
439
440 if ( $listString[0]=="|" ) $listString = substr($listString, 1);
441 $wordlistArray[] = array( $row['id'], $row['name'], $listString );
442 }
443
444
445 return $wordlistArray;
446
447 }
448
449 protected function GetBooksInfo($bookId) {
450 $result = $this->GetSectionsByID($bookId);
451 while ($row = mysql_fetch_assoc($result)) {
452 $bookName = $row['name'];
453 }
454
455 $data = array();
456 $data['bookName'] = $bookName;
457
458 return $data;
459 }
460
461
462 }
463
464
465
466 ?>