1 <?php

     2 

     3 namespace CorpusParole\Libraries\Transcript;

     4 

     5 class LacitoTranscriptConverter extends TranscriptConverterBase {

     6 

     7     // get document title

     8     public function getSourceTitle() {

     9         $baseNode = $this->source->documentElement;

    10         $defaultLanguage = $baseNode->hasAttribute('xml:lang')?$baseNode->getAttribute('xml:lang'):'und';

    11         $resTitle = [];

    12         $xpath = new \DOMXPath($this->source);

    13         foreach($xpath->evaluate("/TEXT/HEADER/TITLE") as $headerTitle) {

    14             $resTitle[$headerTitle->hasAttribute("xml:lang")?$headerTitle->getAttribute('xml:lang'):$defaultLanguage] = (string)$headerTitle->textContent;

    15         }

    16         return $resTitle?:null;

    17     }

    18 

    19 

    20     // do nothing

    21     public function parseSource() {

    22         // do nothing

    23     }

    24 

    25 

    26     private function addAudioInfo($audioNode, &$annotation) {

    27         $annotation['begin'] = floatval($audioNode->getAttribute('start'))*1000;

    28         $annotation['end'] = floatval($audioNode->getAttribute('end'))*1000;

    29     }

    30 

    31     private function addFormInfo($formNode, &$annotation) {

    32         $annotation['content']['data']['content'] = $formNode->textContent;

    33     }

    34 

    35     private function addTranslInfo($translNode, &$annotation) {

    36         $annotation['content']['data']['transl']

    37             = $this->buildTextvalue($translNode->textContent, $translNode->getAttribute('xml:lang'));

    38     }

    39 

    40     private function addWordInfo($wNode, &$annotation) {

    41         if(!array_key_exists('words',$annotation['content']['data'])) {

    42             $annotation['content']['data']['words'] = [];

    43         }

    44         $content = "";

    45         $formNodes = $wNode->getElementsByTagName('FORM');

    46         if(count($formNodes) !== 0 ) {

    47             $content = $formNodes[0]->textContent;

    48         }

    49         $transl = "";

    50         $translNodes = $wNode->getElementsByTagName('TRANSL');

    51         if(count($translNodes) !== 0) {

    52             $transl = $this->buildTextvalue($translNodes[0]->textContent, $translNodes[0]->getAttribute('xml:lang'));

    53         }

    54         array_push($annotation['content']['data']['words'], ['content' => $content, 'transl' => $transl]);

    55     }

    56 

    57 

    58     public function buildAnnotations() {

    59         $xpath = new \DOMXPath($this->source);

    60 

    61         $annotationList = [];

    62 

    63         foreach($xpath->evaluate("/TEXT/S") as $i => $s) {

    64             $data = [];

    65             $speaker = $s->getAttribute('who');

    66             if($speaker) {

    67                 $data['speaker'] = $speaker;

    68             }

    69 

    70             $annotation = [

    71                 "id" => $this->document->getId()."_a".sprintf("%03d",$i+1),

    72                 "media" => $this->getMediaRefId(),

    73                 "content" => [

    74                     "mimetype" => "application/json",

    75                     "data" => $data

    76                 ]

    77             ];

    78 

    79             foreach($s->childNodes as $cnode) {

    80                 if($cnode->nodeType === XML_ELEMENT_NODE) {

    81                     switch($cnode->tagName) {

    82                     case "AUDIO":

    83                         $this->addAudioInfo($cnode, $annotation);

    84                         break;

    85                     case "FORM":

    86                         $this->addFormInfo($cnode, $annotation);

    87                         break;

    88                     case "TRANSL":

    89                         $this->addTranslInfo($cnode, $annotation);

    90                         break;

    91                     case "W":

    92                         $this->addWordInfo($cnode, $annotation);

    93                         break;

    94                     }

    95                 }

    96             }

    97 

    98             array_push($annotationList, $annotation);

    99         }

   100 

   101         return $annotationList;

   102     }

   103 

   104 

   105 }