extraction-interface: models/extractapp.php comparison

comparison models/extractapp.php @ 78:960ba96efce1 extractapp

Update: click to popup remove-tag-window; select to popup tag-window

author	Zoe Hong <zhong@mpiwg-berlin.mpg.de>
date	Mon, 20 Apr 2015 15:44:54 +0200
parents	97c1e5102a22
children	e6e213b26454

comparison

equal deleted inserted replaced

-:97c1e5102a22
+:960ba96efce1
 $this->lg_text = $lg_text;
 }
-public function SetBookMetaDataBySectionId() {
+public function SetBookMetadataBySectionId() {
 // get book_meta from $$this->get_section_metadata_by_sectionId_url
 $section_meta_url = $this->get_section_metadata_by_sectionId_url.$this->section_id;
 $section_meta = json_decode(file_get_contents($section_meta_url), true);
 $stringInput = $this->lg_text;
 $data = array();    // data array to be passed to view
 //$taglistArray = $this->GetTaglistArray();
-//for GetTaglistByTopicID:
+//for GetTaglistByTopicId:
-$taglistArray = $this->GetTaglistByTopicID($this->GetTopic());
+$taglistArray = $this->GetTaglistByTopicId($this->GetTopic());
 // $this->taglist_infile is set (1) from file or (2) from _postdata['taglistArray'] which comes from frontend that user decided
 // $this->taglist_infile is the most up-to-date taglist decided by user. Should be written into file.
 if( $this->TaglistSubsetIn($this->taglist_infile, $taglistArray) ) {    // TaglistSubsetIn($l1,$l2): $l1 is a subset of $l2 or not
 $this->taglist_infile = "";
 file_put_contents("data/parsing_files/".$postdata['filename'].".txt", $require);
 }
 }
-public function UpdateInfoByResponseFromLGService($response) {
+public function UpdateInfoResponsedFromLGService($response) {
 if (isset($response["file"])) {
 $response_file = $response["file"];
 }
 if (isset($response["branch"])) {
 $this->current_fileId = 0;
 }
 }
 // This is only called by SaveFullTextToLGService()
-private function AppendMetaData($text_content) {
+private function AppendMetadata($text_content) {
 $text = '<?xml version="1.0" encoding="UTF-8"?>';
 $text .= "\n<text>\n";
 // --- topic ---
 $text .= "<topic>".$this->topic."</topic>\n";
 // --- book meta data ---
 $text .= "</section>\n";
 // ---- taglist ---
 // $taglist = $this->taglist_infile;
 // obtain the latest taglist from db
-$taglist = $this->GetTaglistByTopicID($this->GetTopic());
+$taglist = $this->GetTaglistByTopicId($this->GetTopic());
 foreach ($taglist as $tagitem) {
 $text .= "<tagitem>\n";
 $text .= "<id>".$tagitem[0]."</id>\n";
 $text .= "<name>".$tagitem[1]."</name>\n";
 $text .= "</text>";
 // ----
 return $text;
 }
 public function SaveFullTextToLGService($_postdata) {
 // save tagged text (full text) by Jorge's API to lg service
 // --------
 global $AT_LOCAL;
 // -- new branch case
 $require = "<text_content>".$require."</text_content>\n";
 }
 */
-$require = $this->AppendMetaData($require);
+$require = $this->AppendMetadata($require);
 //saving in my local machine in developing phrase
 if ($AT_LOCAL) {
 file_put_contents("data/parsing_files/".$_postdata['sectionId'].".txt", $require);
 }
 return $response;
 }
 // --- for regex ----
-public function GetRegexFilenameById($topic_id) {
+private function GetRegexFilenameById($topic_id) {
 $query = sprintf("SELECT regex_filename FROM `topic_regex_relation` WHERE  topic_id=\"%s\"", $topic_id);
 $result = mysql_query($query);
 if (!$result) {
 return json_encode("Failed during selecting topic_regex_relation table.");
 }
 array_push($filenames, $row['regex_filename']);
 }
 return $filenames;
 }
-public function SmartRegexLoad($topic_id) {
+public function LoadSmartRegex($topic_id) {
 // Load regex file based on current topic. Only shows the regex in this topic --
 $filenames = $this->GetRegexFilenameById($topic_id);
 // Get regex file from filesystem ----
 echo json_encode($returnArray);
 return;
 }
-public function SmartRegexSave($_postdata) {
+public function SaveSmartRegex($_postdata) {
 if ($_postdata['text']){
 // --- update topic_regex_relation table ---
 $topic_id = $_postdata['topic_id'];
 $filename = $_postdata['filename'].'.txt';
 }
 $outputTableArray[0]["other"] = "其他";
 $outputTableArray[0]["page"] = "頁數";
 $outputTableArray[0]["full"] = "全文";
+// id, name, tag, color in _taglistArray
 foreach ( $_taglistArray as $tagValue ) {
-$content = preg_replace("/<\/".$tagValue[2].">○*<".$tagValue[2].">/u", "", $_content);
+$tag_name = $tagValue[2];   // $tagValue[2] is tag_name
-$content = preg_replace("/<".$tagValue[2].">[ ]*<\/".$tagValue[2].">/u", "", $_content);
+$content = preg_replace("/<\/".$tag_name.">○*<".$tag_name.">/u", "", $_content);
+$content = preg_replace("/<".$tag_name.">[ ]*<\/".$tag_name.">/u", "", $_content);
 }
 $contentLineArray = explode( "<br>", $content );
 $count=0;
 foreach ( $contentLineArray as $value ) {
 $count++;
 $recordString = $value;
 $otherString = $recordString;
 //echo $recordString."<br>\n";
+// find hyper link in pattern with <a>...</a>
 if ( preg_match("/【<a(.*?)>(.*?)<\/a>】/u", $recordString, $matches) ) {
 $pageNow = $matches[2];
 }
 foreach ( $_taglistArray as $tagValue ) {
-if ( preg_match_all("/<".$tagValue[2].">(.*?)<\/".$tagValue[2].">/u", $recordString, $matches, PREG_SET_ORDER) ) {
+$tag_name = $tagValue[2];   // $tagValue[2] is tag_name
+if ( preg_match_all("/<".$tag_name.">(.*?)<\/".$tag_name.">/u", $recordString, $matches, PREG_SET_ORDER) ) {
 foreach ( $matches as $matchesValue ) {
 $matchesValue[1] = preg_replace("/○/u", "", $matchesValue[1]);
 if ( preg_match_all("/〈(.*?)〉/u", $matchesValue[1], $matches2, PREG_SET_ORDER) ) {
 foreach ( $matches2 as $matches2Value ) {
-if ( isset($outputTableArray[$count][0][$tagValue[2]]) ) {
+if ( isset($outputTableArray[$count][0][$tag_name]) ) {
-$outputTableArray[$count][0][$tagValue[2]] .= ";".$matches2Value[1];
+$outputTableArray[$count][0][$tag_name] .= ";".$matches2Value[1];
 } else {
-$outputTableArray[$count][0][$tagValue[2]] = $matches2Value[1];
+$outputTableArray[$count][0][$tag_name] = $matches2Value[1];
 }
 }
 } else {
-if ( isset($outputTableArray[$count][0][$tagValue[2]]) ) {
+if ( isset($outputTableArray[$count][0][$tag_name]) ) {
-$outputTableArray[$count][0][$tagValue[2]] .= ";".$matchesValue[1];
+$outputTableArray[$count][0][$tag_name] .= ";".$matchesValue[1];
 } else {
-$outputTableArray[$count][0][$tagValue[2]] = $matchesValue[1];
+$outputTableArray[$count][0][$tag_name] = $matchesValue[1];
 }
 }
 }
-$otherString = preg_replace("/<".$tagValue[2].">(.*?)<\/".$tagValue[2].">/u", "　", $otherString);
+$otherString = preg_replace("/<".$tag_name.">(.*?)<\/".$tag_name.">/u", "　", $otherString);
 }
 }
 $otherString = preg_replace("/○/u", "", $otherString);
 $outputTableArray[$count]["other"] = $otherString;
 $outputTableArray[$count]["page"] = $pageNow;
 $sectionName = $postdata['sectionName'];
 $bookId = $postdata['bookId'];
 $bookName = $postdata['bookName'];
 //$taglistArray = $this->GetTaglistArray();
-$taglistArray = $this->GetTaglistByTopicID($topic);
+$taglistArray = $this->GetTaglistByTopicId($topic);
 }
 // =====
 $topic_tag = $this->GetTopicTag($topic);
 $row = mysql_fetch_assoc($result);
 $largest_id = $row['AUTO_INCREMENT']-1;
 $topic_id = $_postdata['topic_id'];
-$result = $this->GetTaglistByTopicID($topic_id);
+$result = $this->GetTaglistByTopicId($topic_id);
 $taglistArray = array();
 foreach ($result as $row) {
 $taglistArray[$row[0]] = array($row[1], $row[2], $row[3]);
 /**
 *
 */
 $topic = $postdata['topic'];
-$result = $this->GetTopicByID($topic);
+$result = $this->GetTopicById($topic);
 $row = mysql_fetch_assoc($result);
 $topic_name_en = $row['name_en'];
 $topic_name_ch = $row['name_ch'];
 $topic_name_pinyin = $row['name_pinyin'];
 */
 }
-private function GetTaglistByTopicID($topic_id) {
+private function GetTaglistByTopicId($topic_id) {
 $taglistArray = array();
 // select taglist ids from topic_tag_relation table
 $query = sprintf("SELECT * FROM `topic_tag_relation` WHERE `topic_id`='%s'", $topic_id);
 $result = mysql_query($query);
 if (!$result) {
 public function SetTopic($topic) {
 $this->topic = $topic;
 }
-public function GetTopic() {
+private function GetTopic() {
 return $this->topic;
 }
 private function GetTopiclistArray() {
 $topiclistArray = array();
-$result = $this->GetTopiclist();
+$result = $this->GetTopicList();
 while ($row = mysql_fetch_assoc($result)) {
 //array_push($topiclistArray, array('id'=>$row['id'],'name'=>$row['name'],'tag'=>$row['tag']));
 array_push($topiclistArray, array('id'=>$row['id'],'tag'=>$row['tag'],'name_en'=>$row['name_en'],'name_ch'=>$row['name_ch'],'name_pinyin'=>$row['name_pinyin'],));
 }
 return $topiclistArray;
 }
 private function GetTopicTag($topic_id) {
-$result = $this->GetTopicByID($topic_id);
+$result = $this->GetTopicById($topic_id);
 $row = mysql_fetch_assoc($result);
 $tag = $row['tag'];
 return $tag;
 }
 private function GetTopicName($topic_id) {
-$result = $this->GetTopicByID($topic_id);
+$result = $this->GetTopicById($topic_id);
 $row = mysql_fetch_assoc($result);
 //$name = $row['name'];
 $name = array('name_en'=>$row['name_en'], 'name_ch'=>$row['name_ch'], 'name_pinyin'=>$row['name_pinyin']);
 return $name;
 */
 }
 // =======================================
-public function sortFunction($a,$b) {
+private function sortFunction($a,$b) {
 return strlen($b)-strlen($a);
 }
 private function GetSectionId() {
 $section_id = $this->GetSectionId();
 if (!is_numeric($section_id)){
 return $section_id;
 }
-$result = $this->GetSectionsByID($section_id);
+$result = $this->GetSectionsById($section_id);
 while ($row = mysql_fetch_assoc($result)) {
 $bookId=$row['books_id'];
 $startPage=$row['start_page'];
 }
 /*
 // get book_meta from books table in db on localhost
 $book_meta = array();
-$books_result = $this->GetBooksByID($bookId);
+$books_result = $this->GetBooksById($bookId);
 while ($row = mysql_fetch_assoc($books_result)) {
 array_push($book_meta, array($row['id'],$row['name'],$row['author'],(string)$row['start_year'],(string)$row['line'],(string)$row['dynasty']));
 // use 'start_year' as year, 'line' is pagenumber
 }
 $this->book_meta = $book_meta;
 return $wordlistArray;
 }
-protected function GetBooksInfo($bookId) {
+private function GetBooksInfo($bookId) {
-$result = $this->GetSectionsByID($bookId);
+$result = $this->GetSectionsById($bookId);
 while ($row = mysql_fetch_assoc($result)) {
 $bookName = $row['name'];
 }
 $data = array();

Mercurial > hg > extraction-interface

comparison models/extractapp.php @ 78:960ba96efce1 extractapp