extraction-interface: models/extractapp.php comparison

comparison models/extractapp.php @ 112:7bdbc7fc7936 extractapp

Replace   to space

author	Calvin Yeh <cyeh@mpiwg-berlin.mpg.de>
date	Thu, 04 May 2017 16:57:16 +0200
parents	025530e37dcd
children	507070df62e6

comparison

equal deleted inserted replaced

-:3eacb630b9eb
+:7bdbc7fc7936
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with Extraction-interface.  If not, see <http://www.gnu.org/licenses/>.
 */
 /**
 * ExtractappModel class is extended from the BaseModel class.
 * This is where we implement the functions/models to prepare data that been sent to views.
 * ExtractappModel is notified by Extractapp controller when there has been a change in its state,
 * and produces the output/data to the corresponding view in "views/Extractapp/".
 *
 */
 class ExtractappModel extends BaseModel{
 protected $section_id = 0, $data_path, $file_id = 0, $current_fileId=0,
 $branch_id = 0, $user_id = 0, $lg_text = "", $topic, $taglist_infile = "",
 $book_meta = "", $book_dynasty = "";
 public $messages = array('info'=>"",'error'=>"",'warning'=>"",'debug'=>"");
 public function SetTextFromFileId() {   // remove $_postdata as input
 /**
 * Get text content from LGService by $this->file_id.
 * After preprocessing the text string (replace space to circle symbol, link for page number, etc),
 * set $this->lg_text to the processed text gotten from LGService.
 */
 /*
 // 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);
 $b = $section_meta['section']['book'];
 $this->book_name = $b['name'];
 $this->book_id = $b['id'];
 /*
 $book_meta = array();
 array_push($book_meta, array($b['id'],$b['name'],$b['level1'],$b['level2'],
 }
 public function SetTextFromSectionId() {  // remove $_postdata as input
 /**
 * Get text content from LGService by $this->section_id.
 * After preprocessing the text string (replace space to circle symbol, link for page number, etc),
 * set $this->lg_text to the processed text gotten from LGService.
 */
 /*
 $this->lg_text = $lg_text;
 }
 // This is only used on localhost
 public function SetTextFromLocal($_id){
 $this->section_id = $_id;
 //$this->branch_id = 1;   // local test sets branch_id to 1
 $this->messages['debug'] .= "[Debug] from my local"."<br>";
 $this->lg_text = $this->GetSectionContent();
 }
 public function SetInfoFromPreviousPage($_postdata) {
 /**
 * Get information from post data passsed by the previous page, and set variables in $this.
 * The previous page could be:
 *
 * (1) LGService: the first time user enters Extraction-interface, which should linked from LGService.
 *
 * (2) Extraction-interface itself: changing topic then continue to tag, saving text to LGService, reload the text from LGService.
 *
 */
 if (isset($_postdata['fileId'])) {
 $this->file_id = $_postdata['fileId'];
 }
 $this->user_id = $_postdata['userId'];
 }
 if (isset($_postdata['topic'])) {
 $this->topic = $_postdata['topic'];
 }
 if (isset($_postdata['sectionName'])) {
 $this->section_name = $_postdata['sectionName'];
 }
 if (isset($_postdata['bookName'])) {
 foreach ($list1 as $v1) {
 $cnt = 0;
 foreach ($list2 as $v2) {
 $num_field = count($v2);
 $matched = true;
 for ($i=0; $i < $num_field; $i++) {
 if ($v1[$i] != $v2[$i]) {
 $matched = false;
 break;
 }
 }
 break;
 }
 }
 if ($cnt == $cnt_list2) {
 return false;
 }
 }
 return true;
 }
 // === for tagging ===
 private function ReplaceLgTextWithLatestTag() {
 // modify $this->lg_text with the latest tags from database
 // can do it by getting the tag's id using tag's information in the file, then by the id to get latest tag (in db)
 $lg_text = $this->lg_text;
 $taglist_indb = $this->GetTaglistByTopicId($this->GetTopic());
 $taglist_infile = $this->taglist_infile;
 if ($taglist_infile == "") return;
 $count_tag_indb = count($taglist_indb);
 foreach ($taglist_infile as $t1) {
 $id = $t1[0];
 $tag = $t1[2];
 $cnt = 0;
 foreach ($taglist_indb as $t2) {
 if ($t2[0] == $id) {
 $latest_tag = $t2[2];
 }
 if ($tag != $t2[2]) {
 // not match with this t2
 $cnt += 1;
 } else {
 break;
 public function StartTagging() {
 /**
 * This is the main method for tagging text. It passes all the information to "views/Extractapp/TaggingText.php" view.
 * The information contain the text string, taglist array, wordlis array, topic, etc.
 */
 $section_id = $this->section_id;
 $data = array();    // data array to be passed to view
 //for GetTaglistByTopicId:
 $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 = "";
 } else {
 // taglist_infile is out-to-date
 // replace all the old tag with new ones in database
 // replace old tags for $this->lg_text with the latest tags from db
 $this->ReplaceLgTextWithLatestTag();
 }
 $stringInput = $this->lg_text;
 $data['taglist_infile'] = $this->taglist_infile;    // if taglist_infile == "" means taglist in file is up-to-date (will use taglist in db), otherwise means the taglist in file
 // book_meta
 $data['book_meta'] = $this->book_meta;
 // topic list
 $topiclistArray = $this->GetTopiclistArray();
 $data['topic_tag'] = $this->GetTopicTag($topic_id);
 $data['topic_name'] = $this->GetTopicName($topic_id);
 $data['topic_tag_ch'] = $this->GetTopicTagName($topic_id);
 $data['info'] = array('file_id'=>$this->file_id, 'user_id'=>$this->user_id,
 'branch_id'=>$this->branch_id, 'section_id'=>$this->section_id, 'book_id'=>$this->book_id,
 'book_name'=>$this->book_name,'section_name'=>$this->section_name, 'period'=>$this->book_meta->period,
 'current_fileId'=>$this->current_fileId);
 $this->messages['debug'] .= "[Debug] ";
 $this->messages['debug'] .= "file_id=".$this->file_id.", section_id=".$this->section_id;
 $this->messages['debug'] .= ", user_id=".$this->user_id.", branch_id=".$this->branch_id.", topic_id=".$this->topic;
 $data['messages'] = $this->messages;
 return $data;
 }
 public function SaveFullText($postdata) {
 // locally
 if ($postdata['text']){
 $date = date('Y_m_d_H_i_s', time());
 if ( file_exists("data/parsing_files/".$postdata['filename'].".txt") ) {
 $oldFile = file_get_contents("data/parsing_files/".$postdata['filename'].".txt");
 file_put_contents("data/parsing_files/".$postdata['filename']."_".$date.".txt", $oldFile);
 }
 if (get_magic_quotes_gpc()) {
 $require = stripslashes($postdata['text']);
 } else {
 $require = $postdata['text'];
 }
 $require = preg_replace("/【<a(.*?)>(.*?)<\/a>】/u", "【\\2】", $require);
 $require = preg_replace('/&amp;/u', "&", $require);
-$require = preg_replace("/○/u", " ", $require);
+$require = preg_replace('/&nbsp;/u', " ", $require); //avoiding invaild xml format
+$require = preg_replace("/○/u", " ", $require);
 $require = preg_replace("/<br>/u", "\n", $require);
 $require = preg_replace("/<br>/u", "\n", $require);
 file_put_contents("data/parsing_files/".$postdata['filename'].".txt", $require);
 }
 }
 public function UpdateInfoResponsedFromLGService($response) {
 /**
 *
 */
 if (isset($response["file"])) {
 $response_file = $response["file"];
 }
 }
 $status = (string)$response["status"];
 if ($status == "ok") {
 $this->messages['info'] .= "[Info] Saving success. <br>";
 $this->file_id = (string)$response_file["id"];
 $this->branch_id = (string)$response_branch["id"];
 } else if ($status == "error") {
 // saving not success
 $this->messages['error'] .= "[Error] Saving Failed! ".(string)$response["message"]."<br>";
 // pass the orignal text?
 $response_currentFile = $response["currentFile"];
 $this->current_fileId = (string)$response_currentFile["id"];
 }
 }
 public function ReloadSetting() {
 if ($this->current_fileId != 0) {
 $text .= "\n<text>\n";
 // --- topic ---
 $text .= "<topic>".$this->topic."</topic>\n";
 // --- book meta data ---
 $book = $this->book_meta;
 $text .= "<book>\n";
 $text .= "<id>".$book->id."</id>\n";
 $text .= "<name>".$book->name."</name>\n";
 $text .= "<level1>".$book->level1."</level1>\n";
 $text .= "<level2>".$book->level2."</level2>\n";
 $text .= "<volume>".$book->volume."</volume>\n";
 $text .= "<author>".$book->author."</author>\n";
 $text .= "<edition>".$book->edition."</edition>\n";
 $text .= "<in_jibengujiku>".$book->in_jibengujiku."</in_jibengujiku>\n";
 $text .= "<admin_type>".$book->admin_type."</admin_type>\n";
 $text .= "<place_name>".$book->place_name."</place_name>\n";
 $text .= "<x>".$book->x."</x>\n";
 $text .= "<y>".$book->y."</y>\n";
 $text .= "</book>\n";
 // --- section info ---
 $text .= "<section>\n";
 $text .= "<id>".$this->section_id."</id>\n";
 $text .= "<name>".$this->section_name."</name>\n";
 }
 public function SaveFullTextToLGService($_postdata) {
 /**
 * It saves the full tagged text string in postdata to LGService.
 * The responses returned by LGService will later be handled in
 */
 // --------
 global $AT_LOCAL;
 //saving in my local machine in developing phrase
 if ($AT_LOCAL) {
 file_put_contents("data/parsing_files/".$_postdata['sectionId']."_".$date.".txt", $oldFile);
 }
 }
 if (get_magic_quotes_gpc()) {
 $require = stripslashes($_postdata['text']);
 } else {
 $require = $_postdata['text'];
 }
 $require = preg_replace("/【<a(.*?)>(.*?)<\/a>】/u", "【\\2】", $require);
 $require = preg_replace('/&amp;/u', "&", $require);
-$require = preg_replace("/○/u", " ", $require);
+$require = preg_replace('/&nbsp;/u', " ", $require);   //avoiding invaild xml format
+$require = preg_replace("/○/u", " ", $require);
 $require = preg_replace("/<br(.*?)>/u", "\n", $require);
 //$require = preg_replace("/<br>/u", "\n", $require);
 $require = "<text_content>".$require."</text_content>\n";
 /*
 if ($_postdata['branchId'] == 0) {
 // -- new branch case
 $require = "<text_content>".$require."</text_content>\n";
 }
 */
 $require = $this->AppendMetadata($require);
 //saving in my local machine in developing phrase
 if ($AT_LOCAL) {
 file_put_contents("data/parsing_files/".$_postdata['sectionId'].".txt", $require);
 }
 }
 // ------
 if ($_postdata['branchId'] != 0) {
 // -- exiting branch case
 $postfields = array(
 "text" => $require,
 "branchId" => $_postdata['branchId'],
 "userId" => $_postdata['userId'],
 "userPreviousFileId" => $_postdata['fileId'],
 );
 $save_url = $this->save_to_LGService_url;
 } else {
 // -- new branch case
 // echo "saveNew!";
 $user_id = "0";
 if ($_postdata['userId']) {
 $user_id = $_postdata['userId'];
 }
 $postfields = array(
 "text" => $require,
 "sectionId" => $_postdata['sectionId'],
 "userId" => $user_id,
 "label" => $_postdata['label'],
 );
 $save_url = $this->save_new_to_LGService_url;
 }
 // set up the curl resource
 $ch = curl_init();
 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
 curl_setopt($ch, CURLOPT_URL, $save_url);
 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 curl_setopt($ch, CURLOPT_CUSTOMREQUEST, "POST");
 curl_setopt($ch, CURLOPT_POST, true);
 curl_setopt($ch, CURLOPT_POSTFIELDS, $postfields);
 //curl_setopt($ch, CURLOPT_HEADER, true);
 curl_setopt($ch, CURLOPT_HTTPHEADER, array(
 'Content-type: multipart/form-data;charset=utf-8'
 ));
 //if (!$AT_LOCAL) {
 // execute the request
 $output = curl_exec($ch);
 //}
 // output the profile information - includes the header
 //echo($output) . PHP_EOL;
 // close curl resource to free up system resources
 curl_close($ch);
 $response = json_decode($output, true);
 return $response;
 }
 }
 public function LoadSmartRegex($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 ----
 $data_path = "./data/"; // get the current file path, which is getcwd(), and concatenate with "/data/"
 $returnArray = array();
 $files1 = scandir($data_path."regex_files");
 foreach ( $files1 as $file ) {
 if ($_postdata['text']){
 // --- update topic_regex_relation table ---
 $topic_id = $_postdata['topic_id'];
 $filename = $_postdata['filename'].'.txt';
 // check if already existing a record for this regex file
 $query = "SELECT * FROM TopicRegexRelation WHERE topic_id=".$topic_id." AND regex_filename='".$filename."'";
 $result = mysql_query($query);
 if (!$result) {
 echo json_encode("ErrorDB");
 if ($row && !$_postdata['forcesave']) {
 // promt user a comfirm to force saving or not
 echo json_encode("ForceSave");
 return;
 }
 }
 if (!$_postdata['forcesave']) {
 $query = "INSERT INTO TopicRegexRelation (topic_id, regex_filename) VALUES (".$topic_id.",'".$filename."')";
 $result = mysql_query($query);
 if (!$result) {
 echo json_encode("ErrorDB");
 }
 }
 // --- write to filesystem ---
 /*
 $date = date('Y_m_d_H_i_s', time());
 if ( file_exists("regex_files/".$_POST['filename'].".txt") ) {
 $oldFile = file_get_contents("regex_files/".$_POST['filename'].".txt");
 file_put_contents("regex_files/".$_POST['filename']."_".$date.".txt", $oldFile);
 }
 */
 $data_path = "./data/"; // get the current file path, which is getcwd(), and concatenate with "/data/"
 if (get_magic_quotes_gpc()) {
 $require = stripslashes($_postdata['text']);
 } else {
 $require = $_postdata['text'];
 }
 file_put_contents( $data_path."regex_files/".$_postdata['filename'].".txt", $require);
 }
 }
 } else {
 $outputTableArray[$count][0][$tag_name] = $matchesValue[1];
 }
 }
 }
 $otherString = preg_replace("/<".$tag_name.">(.*?)<\/".$tag_name.">/u", "　", $otherString);
 }
 }
 $otherString = preg_replace("/○/u", "", $otherString);
 $outputTableArray[$count]["other"] = $otherString;
 $outputTableArray[$count]["page"] = $pageNow;
 $value = preg_replace("/>/u", "&gt;", $value);
 $value = preg_replace("/</u", "&lt;", $value);
 $outputTableArray[$count]["full"] = $value;
 }
 foreach ( $outputTableArray as $arrayIndex => $arrayValue ) {
 // output each row which the topic tag
 // e.g. the original version is for 'person'
 if ( !isset($arrayValue[0][$_topic_tag]) ) {
 unset($outputTableArray[$arrayIndex]);
 }
 }
 // -----
 return $outputTableArray;
 }
 // === for export table ===
 $data['largest_id'] = $largest_id;
 $data['topic_tag_id'] = $topic_tag_id;
 return $data;
 }
 private function GetTag($_postdata) {
 if (get_magic_quotes_gpc()) {
 $id = stripslashes($_postdata['id']);
 }
 return array($id, $name, $tag, $color);
 }
 public function NewTagElement($postdata) {
 if ($postdata['id']){
 list($id, $name, $tag, $color) = $this->GetTag($postdata);
 // for creationDate and lastChangeDate
 $date = date('Y-m-d H:i:s', time());
 $this->GetSQLValueString($tag, "text"),
 $this->GetSQLValueString($color, "text"),
 $this->GetSQLValueString($date, "date"),
 $this->GetSQLValueString($date, "date"));
 //$this->GetSQLValueString($this->systemNAME, "text"));
 $result1 = mysql_query($query1);
 if (!$result1) {
 //echo json_encode("error when insert into Taglist table");
 echo 0;
 return;
 }
 // add it to topic_tag_relation table
 $topic_id = $postdata['topic_id'];
 // tag_id is $id;
 $result = mysql_query($query);
 if (!$result) {
 echo 1;
 //echo json_encode("error when insert into TopicTagRelation table");
 }
 }
 }
 public function SaveTagElement($postdata) {
 if ($postdata['id']){
 list($id, $name, $tag, $color) = $this->GetTag($postdata);
 // for lastChangeDate
 $date = date('Y-m-d H:i:s', time());
 $query = sprintf("UPDATE Taglist SET `name`=%s, `tag`=%s, `color`=%s , `lastChangeDate`=%s WHERE `id`=%s",
 $this->GetSQLValueString($name, "text"),
 $this->GetSQLValueString($tag, "text"),
 $this->GetSQLValueString($color, "text"),
 $this->GetSQLValueString($date, "date"),
 $this->GetSQLValueString($id, "int"));
 $result = mysql_query($query);
 }
 }
 public function DeleteTag($postdata) {
 $topic_id = $postdata['topic_id'];
 $query = sprintf("DELETE FROM `TopicTagRelation` WHERE `tagId` = %s and `topicId` = %s", stripslashes($tag_id), stripcslashes($topic_id));
 $result = mysql_query($query);
 }
 }
 // === for config topic ===
 public function CreateNewTopic($_postdata) {
 $new_topic_id = "2";  // default topic_id=1 for 'person' topic
 // get user input data from postdata
 $new_topic_name_en = $_postdata['new_topic_name_en'];
 $new_topic_name_ch = $_postdata['new_topic_name_ch'];
 $new_topic_name_pi = $_postdata['new_topic_name_pi'];
 $new_topic_tag = $_postdata['new_topic_tag'];
 $new_topic_tag_name = $_postdata['new_topic_tag_name'];
 // -- Start Transaction --
 mysql_query("BEGIN");
 // create records for the new topic in database
 /* In table:
 taglist: name, tag, color (default, user can change later in the editing page)
 topics: tag, name_en, name_ch, name_pinyin
 topic_tag_relation: topic_id (new), tag_id (new)
 - need transaction for the following actions:
 taglist.name: topic_tag name in chinese
 taglist.tag: topic_tag name in eng
 topics.tag: (the same a taglist.tag)
 topics.name_en:
 topics.name_ch:
 topics.name_pinyin:
 topic_tag_relation.topic_id = new_topic_id
 topic_tag_relation.tag_id = new_tag_id
 which are the new record that just been created in the 'taglist' and 'topics' table
 */
 // new record in topics table
 $query="INSERT INTO Topic (tag, name_en, name_ch, name_pinyin) VALUES ('".$new_topic_tag."','".$new_topic_name_en."','".$new_topic_name_ch."','".$new_topic_name_pi."')";
 $result_topics=mysql_query($query);
 if($result_topics){
 $new_topic_id = mysql_insert_id();
 } else {
 //echo mysql_error();
 }
 // new record in taglist table
 $query="INSERT INTO Taglist (name, tag, color) VALUES ('".$new_topic_tag_name."','".$new_topic_tag."','rgb(255, 0, 174)')";
 $result_taglist=mysql_query($query);
 if($result_taglist){
 $new_tag_id = mysql_insert_id();
 } else {
 //echo mysql_error();
 }
 // new record in topic_tag_relation
 $query="INSERT INTO TopicTagRelation (topicId, tagId) VALUES ('".$new_topic_id."','".$new_tag_id."')";
 $result_topic_tag_relation=mysql_query($query);
 if($result_topic_tag_relation){
 } else {
 //echo mysql_error();
 }
 // --- End Transaction ---
 if ($result_taglist && $result_topics && $result_topic_tag_relation) {
 mysql_query("COMMIT");
 } else {
 mysql_query("ROLLBACK");
 return "error";
 }
 $_postdata['topic'] = $new_topic_id;
 return $_postdata;
 }
 public function ConfigTagsInTopic($_postdata) {
 /**
 *
 */
 $topic = $_postdata['topic'];
 $result = $this->GetTopicById($topic);
 $row = mysql_fetch_assoc($result);
 $query = "SELECT Taglist.*, TopicTagRelation.topicId FROM Taglist LEFT JOIN TopicTagRelation ON Taglist.id = TopicTagRelation.tagId ORDER BY `topicId`";
 $result = mysql_query($query);
 if (!$result) {
 return json_encode("Failed during selecting/joining Taglist and TopicTagRelation table.");
 }
 $tag_intopic = array();
 $tag_others = array();
 $tag_tmp_others = array();
 while ($row = mysql_fetch_assoc($result)) {
 // should exclude the topic tag, which is in either $topic_intopic nor $tag_tmp_others
 }
 }
 // --- check if there's any duplicated tags in tag_tmp_others. remove the duplicated tags
 $num_others = count($tag_tmp_others);
 for ($i=0; $i < $num_others; $i++) {
 $cnt = 0;
 for ($j=$i+1; $j < $num_others; $j++) {
 if ($tag_tmp_others[$i]['tag'] == $tag_tmp_others[$j]['tag']) {
 break;
 } else {
 $cnt++;
 }
 }
 // --- remove duplicated tags in tag_others that is duplicated with tags in tag_intopic
 $tag_tmp_others = $tag_others;
 $tag_others = array();
 foreach ($tag_tmp_others as $tmp) {
 $cnt = 0;
 foreach ($tag_intopic as $intopic) {
 if ($tmp['tag'] == $intopic['tag']) {
 break;
 if ($cnt == count($tag_intopic)) {
 // not appear in $tag_intopic
 array_push($tag_others, $tmp);
 }
 }
 // -----
 $data = array();
 //$data['taglistArray'] = $taglistArray;
 $topic_id = $_postdata['topic_id'];
 $tag_ids = json_decode(str_replace('\\', '', $_postdata['ids']));
 // update topic_tag_relation by tags_ids array as `tag_id` and topic_id as `topic_id`
 // --- add new topic_tag_relation ---
 foreach ($tag_ids as $tag_id) {
 $query = "SELECT * FROM TopicTagRelation WHERE tagId=".$tag_id;
 $result = mysql_query($query);
 if (!$result) {
 }
 if (!$flag) {
 // insert a row into topic_tag_relation table
 // $queryUpdate = "INSERT INTO TopicTagRelation (topicId, tagId) VALUES (".$topic_id.",".$tag_id.")";
 $date = date('Y-m-d H:i:s', time());
 $queryUpdate = sprintf("INSERT INTO TopicTagRelation (`topicId`, `tagId`, `creationDate`, `lastChangeDate`) VALUES (%s, %s, %s, %s)",
 $this->GetSQLValueString($topic_id, "int"),
 $this->GetSQLValueString($tag_id, "int"),
 $this->GetSQLValueString($date, "date"),
 $this->GetSQLValueString($date, "date"));
 $resultUpdate = mysql_query($queryUpdate);
 if (!$resultUpdate) {
 return json_encode("error when insert TopicTagRelation table");
 }
 }
 }
 $row = mysql_fetch_assoc($result);
 $tag = $row['tag'];
 // TODO need to reconstruct the structure for topic tag considering user may use Uppercase but only lowercase accpetable in html tag that why we store only lowercase in Taglist.tag field in database table.
 // Temporate solution is we replace " " with "_" and .lowercase(), which are as the same as we do in LGServices when create a topic and also create the correpsonding topic tag
 $tag = str_replace(' ', '_', $tag);
 return strtolower($tag);
 //return $tag;
 }
 }
 // ===========================
 // === for manage wordlist ===
 public function EditWordlist() {
 /**
 * EditWordlist shows the wordlist in a new window.
 return $data;
 }
 public function AddNewList($postdata) {
 if ($postdata['text']){
 if (get_magic_quotes_gpc()) {
 $name = stripslashes($postdata['text']);
 } else {
 $name = $postdata['text'];
 }
 $this->GetSQLValueString($name, "text"));
 //$this->GetSQLValueString($this->systemNAME, "text"));
 $result1 = mysql_query($query1);
 file_put_contents( "data/wordlist/".mysql_insert_id().".txt", "(empty now)");
 }
 }
 public function SaveWordlist($postdata) {
 if ($postdata['text']){
 $date = date('Y_m_d_H_i_s', time());
 if ( file_exists("data/wordlist/".$postdata['filename'].".txt") ) {
 $filename = "data/wordlist/".$postdata['filename']."_".$date.".txt";
 $oldFile = file_get_contents("data/wordlist/".$postdata['filename'].".txt");
 file_put_contents($filename, $oldFile);
 }
 /*else {
 $filename = "data/wordlist/".$postdata['filename'].".txt";
 }
 */
 $date = date('Y_m_d_H_i_s', time());
 if ( file_exists("data/wordlist/".$postdata['filename'].".txt") ) {
 $oldFile = file_get_contents("data/wordlist/".$postdata['filename'].".txt");
 file_put_contents("data/wordlist/".$postdata['filename']."_".$date.".txt", $oldFile);
 }
 if (get_magic_quotes_gpc()) {
 $require = stripslashes($postdata['text']);
 } else {
 $require = $postdata['text'];
 }
 return $this->section_id;
 } else {
 return json_encode("Error: No section id");
 }
 }
 /* only used locally */
 /*
 private function GetSectionInfo() {
 $section_id = $this->GetSectionId();
 return $section_id;
 }
 $result = $this->GetSectionsById($section_id);
 while ($row = mysql_fetch_assoc($result)) {
 $bookId=$row['books_id'];
 $startPage=$row['start_page'];
 $endPage=$row['end_page'];
 $sectionName = $row['name'];
 $taglistArray = array();
 foreach ($taglist_infile as $row) {
 array_push($taglistArray, array((string)$row->id,(string)$row->name,(string)$row->tag,(string)$row->color ));
 }
 if ($taglistArray) {
 $this->taglist_infile = $taglistArray;
 }
 // --- get book meta data from file, replaced by SetBookMetaDataBySectionId(), which retrives book meta data from LGService
 /*
 $book_meta = $xml->book;
 return $new_contentString;
 }
 /** This is only been called by SetTextFromLocal().
 */
 /*
 private function GetSectionContent() {
 $section_id = $this->GetSectionId();
 $section_info = $this->GetSectionInfo();
 $bookId = $section_info['bookId'];
 $contentString="";
 $data_path = $this->GetDataPath();
 if ( file_exists($data_path."parsing_files/".$section_id.".txt") ) {
 $filename = $data_path."parsing_files/".$section_id.".txt";
 // --- parsing meta data
 $stringInput = $this->ParseDataInFile($filename);
 // ----
 // if the text is from file system
 $stringInput = preg_replace("/ /u", "○", $stringInput);
 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
 $stringInput = preg_replace("/【(.*?)】/u", "【<a href=\"review_index_xml_images.php?books_id=".$bookId."&pages=\\1&entry=0\" target=\"_bookImg\">\\1</a>】", $stringInput);
 $this->branch_id = 1;   // testing at local
 } else {
 $query = sprintf("SELECT `content`, `line`, `books_id` FROM `contents` WHERE `books_id`=\"%s\" AND `line`>=%d AND `line`<=%d", $bookId, $startPage, $endPage);
 $result = mysql_query($query);
 return json_encode("Failed during selecting content table.");
 }
 while ($row = mysql_fetch_assoc($result)) {
 $contentString.="【".$row['line']."】".$row['content']."\n";
 }
 // the text is from database
 $stringInput = $contentString;
 $stringInput = preg_replace("/<(.*?)>/u", "○", $stringInput);
 $stringInput = preg_replace("/ /u", "○", $stringInput);
 $stringInput = preg_replace("/\n/u", "<br>", $stringInput);
 return $stringInput;
 }
 */
 private function GetDataPath() {
 return getcwd()."/data/"; // get the current file path, which is getcwd(), and concatenate with "/data/"
 }
 private function GetTaglistArray() {
 $listString = preg_replace("/<div>/u", "\n", $listString);
 $listString = preg_replace("/<\/div>/u", "", $listString);
 $listString = preg_replace("/<span(.*?)>/u", "", $listString);
 $listString = preg_replace("/<\/span>/u", "", $listString);
 //$listString = preg_replace("/\n/u", "|", $listString);
 $wordlistArray2 = explode( "\n", $listString );
 //usort($wordlistArray2,'sortFunction');
 foreach ( $wordlistArray2 as $index=>$value ) {
 $wordlistArray2[$index] = implode("○?", preg_split("/(?<!^)(?!$)/u", $value));
 }
 foreach ( $wordlistArray2 as $index=>$value ) {
 if ($value=="") unset($wordlistArray2[$index]);
 }
 $listString = implode("|", $wordlistArray2);
 if ( $listString[0]=="|" ) $listString = substr($listString, 1);
 $wordlistArray[] = array( $row['id'], $row['name'], $listString );
 }
 return $wordlistArray;
 }
 }
 ?>

Mercurial > hg > extraction-interface

comparison models/extractapp.php @ 112:7bdbc7fc7936 extractapp