1 <?php

     2 

     3 namespace CorpusParole\Libraries\Transcript;

     4 

     5 class TranscriberTranscriptConverter extends TranscriptConverterBase {

     6 

     7     private $topics = [];

     8     private $topicIds = [];

     9     private $speakers = [];

    10     private $speakerIds = [];

    11     private $lists = [];

    12     private $annotationTypes = [];

    13     private $annotations = [];

    14     private $turnCounter = 1;

    15     private $annotationCounter = 1;

    16 

    17     private function parseTopics() {

    18 

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

    20         foreach($xpath->evaluate("/Trans/Topics/Topic") as $i=>$topicNode) {

    21             $topicId = $this->document->getId()."_tpc".sprintf("%03d",$i+1);

    22             $this->topicIds[$topicNode->getAttribute('id')] = $topicId;

    23             array_push($this->topics,[

    24                 'id' => $topicId,

    25                 'desc' => $topicNode->getAttribute('desc')

    26             ]);

    27         }

    28     }

    29 

    30     private function parseSpeakers() {

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

    32         foreach($xpath->evaluate("/Trans/Speakers/Speaker") as $i=>$speakerNode) {

    33             $speakerId = $this->document->getId()."_spkr".sprintf("%03d",$i+1);

    34             $this->speakerIds[$speakerNode->getAttribute('id')] = $speakerId;

    35             array_push($this->speakers,[

    36                 'id' => $speakerId,

    37                 'name' => $speakerNode->getAttribute('name')

    38             ]);

    39         }

    40     }

    41 

    42     private function buildTurnAnnotations($turnNode, $turnId, $begin, $end) {

    43 

    44         $currentAnnotations = [];

    45         $currentBegin = $begin;

    46         $currentSpeaker = null;

    47         $turnSpeakers = array_reduce(

    48             explode(" ", $turnNode->getAttribute('speaker')),

    49             function($res, $spk) {

    50                 array_push(

    51                     $res,

    52                     array_key_exists($spk,$this->speakerIds)?['id-ref' => $this->speakerIds[$spk]]:$spk

    53                 );

    54                 return $res;

    55             },

    56             []

    57         );

    58         if(count($turnSpeakers) == 1) {

    59             $currentSpeaker = $turnSpeakers[0];

    60         }

    61 

    62         foreach($turnNode->childNodes as $cnode) {

    63             if($cnode->nodeType === XML_TEXT_NODE) {

    64                 $textContent = trim($cnode->textContent);

    65                 if(empty($textContent))

    66                     continue;

    67                 $aData = $currentSpeaker?["speaker" => $currentSpeaker]:[];

    68                 $aData['content'] = $textContent;

    69                 $newAnnotation = [

    70                     'id' => $this->document->getId()."_a".sprintf("%04d", $this->annotationCounter++),

    71                     'begin' => $currentBegin,

    72                     'end' => $end,

    73                     'media' => $this->getMediaRefId(),

    74                     'type' => $turnId,

    75                     'content' => [ "mimetype" => "application/json", "data" => $aData],

    76                     'meta' => [ 'id-ref' => $turnId ]

    77                 ];

    78                 array_push($currentAnnotations, $newAnnotation);

    79 

    80             } elseif($cnode->nodeType === XML_ELEMENT_NODE && $cnode->tagName === "Who") {

    81 

    82                 $currentSpeaker = $turnSpeakers[intval($cnode->getAttribute('nb'))-1];

    83 

    84             } elseif($cnode->nodeType === XML_ELEMENT_NODE && $cnode->tagName === "Sync") {

    85 

    86                 $currentBegin = floatval($cnode->getAttribute('time')) * 1000;

    87                 foreach($currentAnnotations as &$nAnnot) {

    88                     $nAnnot['end'] = $currentBegin;

    89                 }

    90                 $this->annotations = array_merge($this->annotations, $currentAnnotations);

    91                 $currentAnnotations = [];

    92             }

    93         }

    94         $this->annotations = array_merge($this->annotations, $currentAnnotations);

    95 

    96     }

    97 

    98     private function parseTurn($turnNode) {

    99 

   100         $turnIndex = $this->turnCounter++;

   101         $turnId = $this->document->getId()."_trn".sprintf("%04d", $turnIndex);

   102         $begin = floatval($turnNode->getAttribute("startTime")) * 1000;

   103         $end =  floatval($turnNode->getAttribute("endTime")) * 1000;

   104         $turn = [

   105             'id' => $turnId,

   106             'dc:title' => "Turn $turnIndex",

   107             'corpus:begin' => $begin,

   108             'corpus:end' => $end

   109         ];

   110         array_push($this->annotationTypes, $turn);

   111 

   112         $this->buildTurnAnnotations($turnNode, $turnId, $begin, $end);

   113 

   114         return $turnId;

   115     }

   116 

   117     private function parseSections() {

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

   119         foreach($xpath->evaluate("/Trans/Episode/Section") as $sectionIndex=>$sectionNode) {

   120             $sectionItems = [];

   121             foreach($sectionNode->childNodes as $turnNode) {

   122                 if($turnNode->nodeType === XML_ELEMENT_NODE && $turnNode->tagName === 'Turn') {

   123                     array_push($sectionItems, [ "id-ref" => $this->parseTurn($turnNode)]);

   124                 }

   125             }

   126             $section = [

   127                 'id' => $this->document->getId()."_sctn".sprintf("%03d", $sectionIndex+1),

   128                 'items' => $sectionItems,

   129                 'meta' => [

   130                     'corpus:topic'=> ["id-ref" => $this->topicIds[$sectionNode->getAttribute('topic')]],

   131                     'corpus:begin' => floatval($sectionNode->getAttribute('startTime'))*1000,

   132                     'corpus:end' => floatval($sectionNode->getAttribute('endTime'))*1000,

   133                 ]

   134             ];

   135 

   136             array_push($this->lists, $section);

   137         }

   138     }

   139 

   140     public function parseSource() {

   141         // do nothing

   142         $this->parseTopics();

   143         $this->parseSpeakers();

   144         $this->parseSections();

   145     }

   146 

   147         // add resources

   148     public function buildResources() {

   149         return [

   150             ['id' => "topics"  , "content" => ['mimetype' => 'application/json', 'data' => $this->topics]],

   151             ['id' => "speakers", "content" => ['mimetype' => 'application/json', 'data' => $this->speakers]],

   152         ];

   153     }

   154 

   155     // add lists

   156     public function buildLists() {

   157         return $this->lists;

   158     }

   159 

   160     // add annotation types

   161     public function buildAnnotationTypes() {

   162         return $this->annotationTypes;

   163     }

   164 

   165 

   166     public function buildAnnotations() {

   167         return $this->annotations;

   168     }

   169 

   170 

   171 }