<?php

//Setting to get korpus info and location for schema validation
$korp_info_url = "http://demosb.spraakdata.gu.se/cgi-bin/korp/korp.cgi?command=info&corpus=FSBBLOGGVUXNA%2CFSBSKONLIT%2CFSBESSAISTIK%2CFSBSAKPROSA%2CLAGTEXTER%2CMYNDIGHET%2CHANKEN%2CSVENSKBYGDEN%2CSTUDENTBLADET%2CJAKOBSTADSTIDNING1999%2CJAKOBSTADSTIDNING2000%2CSWEACHUM%2CDN1987%2CGP1994%2CGP2001%2CGP2002%2CGP2003%2CGP2004%2CGP2005%2CGP2006%2CGP2007%2CGP2008%2CGP2009%2CGP2010%2CGP2011%2CGP2D%2CORDAT%2CFOF%2CPRESS65%2CPRESS76%2CPRESS95%2CPRESS96%2CPRESS97%2CPRESS98%2CSTRINDBERGBREV%2CBLOGGMIX%2CDRAMA%2CLASBART%2CPAROLE%2CPSALMBOKEN%2CSNP7879%2CSUC2%2CSTORSUC%2CSALTNLD_SWE%2CEUROPARLDA_SV%2CDIABETOLOG%2CLT1996%2CLT1997%2CLT1998%2CLT1999%2CLT2000%2CLT2001%2CSMITTSKYDD%2CVIVILL%2CSTRINDBERGROMANER%2CROMI%2CROMII%2CROMG%2CROM99%2CSFS%2CWIKIPEDIA%2CSWEWAC%2CASTRANOVA%2CKALLAN%2CMEDDELANDEN%2CNYAARGUS%2CPARGASKUNGORELSER%2CVASABLADET%2CFNB1999%2CFNB2000%2CHBL1991%2CHBL1998%2CHBL1999&log=1";
$metnord_schema = 'C:\\Users\\olof\\Documents\\NetBeansProjects\\metanord-schemas\\META-SHARE-Resource.xsd';

//get the stats from korp
$info = file_get_contents($korp_info_url);
$korp_info = json_decode($info, true);


ini_set("display_errors", "2");
ERROR_REPORTING(E_ALL);
// Enable user error handling
libxml_use_internal_errors(true);

//set the timezone
date_default_timezone_set('Europe/Paris');

//run batch convertions
print "\n--Lexicon--\n";
$legacyIDsLexicon = getLegacyIDs("../../metanord-ugot/LexicalConceptual");
batchConvert("../lexikon-dc", 'C:\\Users\olof\\Documents\\metanord\\lexicon\\', $metnord_schema, $legacyIDsLexicon);

print "--Corpora--\n";
$legacyIDsCorpora = getLegacyIDs("../../metanord-ugot/Corpus");
batchConvert("../corpus-dc", 'C:\\Users\olof\\Documents\\metanord\\corpus\\', $metnord_schema, $legacyIDsLexicon);



/********************************
 * runs a batch convert from a folder with dublin core metadata to a targeted 
 * metanord folder and validates the covertion
 * @form_dir source dir containing dublin core files
 * @taget dir for validated metanord
 * @metnord_schema path to the metanord schema
 * @legacyIDs array containing legacy ids
 */
function batchConvert($from_dir, $to_dir, $metnord_schema, $legacyIDs) {
    if ($handle = opendir($from_dir)) {
        while (false !== ($entry = readdir($handle))) {
            if ($entry != "." && $entry != ".." && $entry != ".svn") {
                $xml = simplexml_load_file($from_dir . '/' . $entry);
                $desc = _getBaseDescriptionArray($xml);

                if (!empty($legacyIDs[$desc['identifier']]['handle']))
                    $desc['handle'] = $legacyIDs[$desc['identifier']]['handle'];

                $metanord = dcToMetanord($desc);
                $metanord_xml = simplexml_load_string($metanord);
                //print_r($desc);
                $xmlv = new DOMDocument();
                $xmlv->loadXML($metanord);

                //validate the document
                if (!$xmlv->schemaValidate($metnord_schema)) {
                    echo "error\t $entry \n";
                    libxml_display_errors();
                } else {
                    echo "valid\t $entry \n";

                    //save the document (format with DOM)
                    $dom = new DOMDocument('1.0');
                    $dom->preserveWhiteSpace = false;
                    $dom->formatOutput = true;
                    $dom->loadXML($metanord);
                    $dom->save($to_dir . $entry);
                }
            }
        }
        closedir($handle);
    } else {
        print "Error loading $dir";
    }
}


/*******
 * takes a set of base descriptions and outputs metanord xml
 * @desc array of descriptions
 */
function dcToMetanord($desc) {
    //print_r($desc);
    //get the corpora info
    global $korp_info;

    //fix the licence-element
    if (stristr($desc['license'], "sb-info@svenska.gu.se") || stristr($desc['license'], "Egen:")) {
        $desc['license'] = 'CC_BY-SA_3.0';
    }

    $licences = '';
    if (stristr($desc['license'], ',')) {
        $tmp = explode(',', $desc['license']);

        foreach ($tmp as $l) {
            if (stristr($l, 'GFDL')) {
                $licences .= '<licence>GFDL</licence>' . "\n";
            } else if ($l == "CC-BY-SA 3.0") {
                $licences .= '<licence>CC_BY-SA_3.0</licence>' . "\n";
            } else if (stristr($l, 'LGPL 3.0')) {
                $licences .= '<licence>LGPL</licence>' . "\n";
            } else {
                $licences .= '<licence>' . trim($l) . '</licence>' . "\n";
            }
        }
    } else {
        $licences = '<licence>' . $desc['license'] . '</licence>';
    }


    $tokens = $desc['extent'];
    if (array_key_exists(strtoupper($desc['identifier']), $korp_info['corpora']) && is_numeric($korp_info['corpora'][strtoupper($desc['identifier'])]['info']['Size'])) {
        $tokens = $korp_info['corpora'][strtoupper($desc['identifier'])]['info']['Size'];
    }


    //begin building the output
    $output = '<?xml version="1.0" encoding="UTF-8"?>
<resourceInfo xmlns="http://www.ilsp.gr/META-XMLSchema">
  <identificationInfo>
    <resourceName>' . htmlspecialchars($desc['title']['en']) . '</resourceName>
    <resourceName lang="swe">' . htmlspecialchars($desc['title']['sv']) . '</resourceName>
    <description>' . htmlspecialchars($desc['description']['en']) . '</description>
    <description lang="swe">' . htmlspecialchars($desc['description']['sv']) . '</description>
    <resourceShortName>' . $desc['identifier'] . '</resourceShortName>
    <resourceShortName lang="swe">' . $desc['identifier'] . '</resourceShortName>
    <metaShareId></metaShareId>';

    if (!empty($desc['handle']))
        $output .='<identifier>' . $desc['handle'] . '</identifier>';
    else
        $output .='<identifier>hdl:10794/' . uniqid("", true) . '</identifier>';

    $output .='<identifier>' . $desc['identifier'] . '</identifier>
  </identificationInfo>
  <distributionInfo>';

    $output .= '<availability>available-restrictedUse</availability>';

    if (array_key_exists('relation', $desc)) {
        if (array_key_exists('downloads', $desc['relation'])) {
            foreach ($desc['relation']['downloads'] as $d) {
                $output .= '
            <licenceInfo>
                ' . $licences . '
                <restrictionsOfUse>shareAlike</restrictionsOfUse>
                <distributionAccessMedium>downloadable</distributionAccessMedium>
                <downloadLocation>' . $d['url'] . '</downloadLocation> 
            </licenceInfo>';
            }
        }

        if (array_key_exists('accessibleThroughInterface', $desc['relation'])) {
            foreach ($desc['relation']['accessibleThroughInterface'] as $d) {
                $output .= '
            <licenceInfo>
                ' . $licences . '
                <restrictionsOfUse>shareAlike</restrictionsOfUse>
                <distributionAccessMedium>accessibleThroughInterface</distributionAccessMedium>
                <executionLocation>' . $d['url'] . '</executionLocation> 
            </licenceInfo>';
            }
        }

        if (array_key_exists('other', $desc['relation'])) {
            foreach ($desc['relation']['other'] as $o) {
                $output .= '
                <licenceInfo>
                    ' . $licences . '
                    <restrictionsOfUse>shareAlike</restrictionsOfUse>
                    <distributionAccessMedium>other</distributionAccessMedium>
                    <downloadLocation>' . $o['url'] . '</downloadLocation> 
                </licenceInfo>';
            }
        }
    }


    $output .=
            '
    <iprHolder>
      <personInfo>
	<surname>Forsberg</surname>
	<givenName>Markus</givenName>
	<communicationInfo>
	  <email>sb-info@svenska.gu.se</email>
	</communicationInfo>
	<affiliation>
	  <organizationName>Språkbanken, University of Gothenburg</organizationName>
	  <organizationShortName>Språkbanken</organizationShortName>
	  <communicationInfo>
	    <email>sb-info@svenska.gu.se</email>
	  </communicationInfo>
	</affiliation>
      </personInfo>
    </iprHolder>
  </distributionInfo>
  <contactPerson>
    <surname>Forsberg</surname>
    <givenName>Markus</givenName>
    <communicationInfo>
      <email>sb-info@svenska.gu.se</email>
    </communicationInfo>
    <affiliation>
      <organizationName>Språkbanken</organizationName>
      <communicationInfo>
	<email>sb-info@svenska.gu.se</email>
      </communicationInfo>
    </affiliation>
  </contactPerson>
  <metadataInfo>
    <metadataCreationDate>' . date("Y-m-d") . '</metadataCreationDate>
    <metadataCreator>
      <surname>Olsson</surname>
      <givenName>Olof</givenName>
      <communicationInfo>
	<email>sb-info@svenska.gu.se</email>
      </communicationInfo>
      <affiliation>
	<organizationName>Språkbanken</organizationName>
	<communicationInfo>
	  <email>sb-info@svenska.gu.se</email>
	</communicationInfo>
      </affiliation>
    </metadataCreator>
    <source>META-NORD</source>
  </metadataInfo>  
  <validationInfo>
    <validated>true</validated>
    <validationType>formal</validationType>
    <validationMode>automatic</validationMode>
    <validationExtent>full</validationExtent>
  </validationInfo>';

    //Documentation (manuals)  
    $output .= '
    <resourceDocumentationInfo>
        <samplesLocation>http://spraakbanken.gu.se/eng/resource/' . $desc['identifier'] . '</samplesLocation>
        <toolDocumentationType>manual</toolDocumentationType>
        <toolDocumentationType>online</toolDocumentationType>
    </resourceDocumentationInfo>';

    $output .='
  <!--
  <relationInfo>
    <relationType></relationType>
    <relatedResource>
      <targetResourceNameURI></targetResourceNameURI>
    </relatedResource>
  </relationInfo>
  -->
  <resourceComponentType>';


    //corpus
    if ($desc['type'] == 'corpus') {
        $output .='
    <corpusInfo>
      <resourceType>corpus</resourceType>
      <corpusMediaType>
	<corpusTextInfo>
	  <mediaType>text</mediaType>
	  <lingualityInfo>
	    <lingualityType>monolingual</lingualityType>
	  </lingualityInfo>
	  <languageInfo>
	    <languageId>swe</languageId>
	    <languageName>Swedish</languageName>
	    <!--languageCoding>ISO-639-3</languageCoding-->
	    <languageScript>Latn</languageScript>
	    <languageVarietyInfo>
	      <languageVarietyType>other</languageVarietyType>
	      <languageVarietyName></languageVarietyName>
	      <sizePerLanguageVariety>
		<size>' . $tokens . '</size>
		<sizeUnit>tokens</sizeUnit>
		</sizePerLanguageVariety>
	      </languageVarietyInfo>
	  </languageInfo>';


        $output .='
        <sizeInfo>
            <size>' . $tokens . '</size>
            <sizeUnit>tokens</sizeUnit>
        </sizeInfo>';


        if (array_key_exists(strtoupper($desc['identifier']), $korp_info['corpora']) && is_numeric($korp_info['corpora'][strtoupper($desc['identifier'])]['info']['Sentences'])) {
            $output .='
        <sizeInfo>
            <size>' . $korp_info['corpora'][strtoupper($desc['identifier'])]['info']['Sentences'] . '</size>
            <sizeUnit>sentences</sizeUnit>
        </sizeInfo>';
        }

        $output .='
          <!--
	  <annotationInfo>
            <annotationType>segmentation</annotationType>
            <annotationType>lemmatization</annotationType>
            <annotationType>morphosyntacticAnnotation-posTagging</annotationType>
	    
	    <annotationType>structuralAnnotation</annotationType>
	    <annotationStandoff>true</annotationStandoff>
	    <segmentationLevel>sentence</segmentationLevel>
	    <segmentationLevel>word</segmentationLevel>
	    <annotationFormat>text/xml</annotationFormat>
	    <annotationMode>automatic</annotationMode>
	  </annotationInfo>
          -->
	</corpusTextInfo>
      </corpusMediaType>
    </corpusInfo>';
    } else if ($desc['type'] == 'lex') {
        $output .=
                '<lexicalConceptualResourceInfo>
                <resourceType>lexicalConceptualResource</resourceType>
                <lexicalConceptualResourceType>computationalLexicon</lexicalConceptualResourceType>
                <lexicalConceptualResourceEncodingInfo>
                    <encodingLevel>morphology</encodingLevel>
                    <conformanceToStandardsBestPractices>LMF</conformanceToStandardsBestPractices>
                </lexicalConceptualResourceEncodingInfo>
                <creationInfo></creationInfo>
                <lexicalConceptualResourceMediaType>
                    <lexicalConceptualResourceTextInfo>
                    <mediaType>text</mediaType>
                    <lingualityInfo>
                        <lingualityType>monolingual</lingualityType>
                    </lingualityInfo>
                    <languageInfo>
                        <!-- languageCoding>ISO-639-3</languageCoding -->
                        <languageId>swe</languageId>
                        <languageName>Swedish</languageName>
                    </languageInfo>
                    <modalityInfo>
                        <modalityType>writtenLanguage</modalityType>
                    </modalityInfo>
                    <!--
                    <sizeInfo>
                        <size>4242424242</size>
                        <sizeUnit>bytes</sizeUnit>
                    </sizeInfo>
                    -->
                    <sizeInfo>
                        <size>' . $desc['extent'] . '</size>
                        <sizeUnit>entries</sizeUnit>
                        <!-- sizeUnitMultiplier>unit</sizeUnitMultiplier -->
                    </sizeInfo>
                    <textFormatInfo>
                        <mimeType>other</mimeType>
                        <sizePerTextFormat>
                        <size>' . $desc['extent'] . '</size>
                        <sizeUnit>entries</sizeUnit>
                        <!--sizeUnitMultiplier>unit</sizeUnitMultiplier-->
                        </sizePerTextFormat>
                    </textFormatInfo>
                    <!--
                    <characterEncodingInfo>
                        <characterEncoding>UTF-8</characterEncoding>
                        <sizePerCharacterEncoding>
                        <size>23300000</size>
                        <sizeUnit>bytes</sizeUnit>
                        
                        </sizePerCharacterEncoding>
                    </characterEncodingInfo>
                    -->
                    <domainInfo>
                        <domain>general</domain>
                        <sizePerDomain>
                        <size>' . $desc['extent'] . '</size>
                        <sizeUnit>entries</sizeUnit>
                        <!--sizeUnitMultiplier>unit</sizeUnitMultiplier-->
                        </sizePerDomain>
                    </domainInfo>
                    <timeCoverageInfo>
                        <timeCoverage>other</timeCoverage>
                        <sizePerTimeCoverage>
                        <size>' . $desc['extent'] . '</size>
                        <sizeUnit>entries</sizeUnit>
                        <!--sizeUnitMultiplier>unit</sizeUnitMultiplier-->
                        </sizePerTimeCoverage>
                    </timeCoverageInfo>
                    <geographicCoverageInfo>
                        <geographicCoverage>other</geographicCoverage>
                        <sizePerGeographicCoverage>
                        <size>' . $desc['extent'] . '</size>
                        <sizeUnit>entries</sizeUnit>
                        <!--sizeUnitMultiplier>unit</sizeUnitMultiplier-->
                        </sizePerGeographicCoverage>
                    </geographicCoverageInfo>
                    </lexicalConceptualResourceTextInfo>
                </lexicalConceptualResourceMediaType>            
                </lexicalConceptualResourceInfo>';
    }
    $output .= '
  </resourceComponentType>
</resourceInfo>';

    return $output;
}

/**********
 * extracts the descriptions from the Dublin Core elements
 * @xml xml for the dublin core
 * returns array of the elements
 */
function _getBaseDescriptionArray($xml) {
    $title_sv = $xml->xpath('//dc:title[@xml:lang = "sv"]');
    $title_en = $xml->xpath('//dc:title[@xml:lang = "en"]');

    $description_sv = $xml->xpath('//dc:description[@xml:lang = "sv"]');
    $description_en = $xml->xpath('//dc:description[@xml:lang = "en"]');

    $extent = $xml->xpath('//dcterms:extent');
    $identifier = $xml->xpath('//dc:identifier');
    $license = $xml->xpath('//dc:license');
    $type = $xml->xpath('//dc:type');


    $tmp['identifier'] = (string) $identifier[0];
    $tmp['extent'] = (string) $extent[0];
    $tmp['type'] = (string) $type[0];
    $tmp['license'] = (string) $license[0];
    $tmp['title']['sv'] = (string) $title_sv[0];
    $tmp['title']['en'] = (string) $title_en[0];
    $tmp['description']['sv'] = (string) $description_sv[0];
    $tmp['description']['en'] = (string) $description_en[0];

    foreach ($xml->xpath('//dc:relation') as $rel) {
        $resource = array();

        $url = (string) $rel[0];
        $resource['url'] = trim($url);

        $xsi = $rel->attributes("xsi", 1);
        $format = (string) $xsi['format'];
        $type = (string) $xsi['type']; //"dcterms:URI"
        $resource['label'] = $format;

        if (!empty($resource['url'])) {
            if (strpos(strtolower($format), 'xml') !== false ||
                    strpos(strtolower($format), 'lmf') !== false ||
                    strpos(strtolower($format), 'html') !== false ||
                    strpos(strtolower($format), 'bz2') !== false ||
                    strpos(strtolower($format), 'txt') !== false ||
                    strpos(strtolower($format), 'xls') !== false ||
                    strpos($url, 'pub/reskit/parole.zip') !== false) {
                $tmp['relation']['downloads'][] = $resource;
            } else if (strpos($url, 'korp/#corpus=') !== false ||
                    strpos($url, 'glossa/html') !== false ||
                    strpos($url, 'FTS/search.phtml') !== false ||
                    strpos($url, 'http://www.medicinskkorpus.se/login.phtml') !== false ||
                    strpos($url, 'http://www.dramawebben.se/') !== false ||
                    strpos($url, 'http://sv.wikipedia.org/') !== false ||
                    strpos($url, 'FTS/search.phtml') !== false ||
                    strpos($url, 'FTS/search.phtml') !== false ||
                    strpos($url, 'karp') !== false) {
                $tmp['relation']['accessibleThroughInterface'][] = $resource;
            } else if (strpos($url, 'korp/ws') !== false ||
                    strpos($url, 'simple_parole_index.html') !== false ||
                    strpos($url, 'http://spraakbanken.gu.se/strindberg/') !== false ||
                    strpos($url, 'korp/ws') !== false ||
                    strpos($url, 'korp/ws') !== false) {
                $tmp['relation']['resourceDocumentationInfo']['manual'][] = $resource;
            } else {
                $tmp['relation']['other'][] = $resource;
            }
            //print $resource['url'] . "\n";
        }
    }

    return $tmp;
}

/******
 * get the identifiers from a folder containing existing metanord-records
 * @path path to the folder
 * return array of identifiers
 */
function getLegacyIDs($path) {
    $result = array();
    if ($handle = opendir($path)) {
        while (false !== ($entry = readdir($handle))) {
            if ($entry != "." && $entry != ".." && $entry != ".svn") {
                $xml = simplexml_load_file($path . '/' . $entry);

                $pid = $xml->IdentificationInfo->pid;
                $resourceShortName = $xml->IdentificationInfo->resourceShortName;
                $identifier = $xml->IdentificationInfo->identifier;

                $resourceShortName = strtolower($resourceShortName);

                if (strpos($resourceShortName, 'derwall supplement'))
                    $resourceShortName = "soederwall_supp";
                else if (strpos($resourceShortName, 'derwall'))
                    $resourceShortName = "soederwall";

                if (!empty($resourceShortName)) {
                    $result[$resourceShortName]['name'] = $resourceShortName;
                    $result[$resourceShortName]['pid'] = (string) $pid;


                    $result[$resourceShortName]['handle'] = 'hdl:10794/' . $result[$resourceShortName]['pid'];
                }
            }
        }
    }
    return $result;
}

////////////////////////////////////////////////////////////////////////////////
function libxml_display_error($error) {
    $return = "\t";
    switch ($error->level) {
        case LIBXML_ERR_WARNING:
            $return .= "Warning $error->code : ";
            break;
        case LIBXML_ERR_ERROR:
            $return .= "Error $error->code : ";
            break;
        case LIBXML_ERR_FATAL:
            $return .= "Fatal Error $error->code : ";
            break;
    }
    $return .= trim($error->message);
    if ($error->file) {
        $return .= " in $error->file";
    }
    $return .= " on line $error->line\t";

    return $return . "\n";
}

function libxml_display_errors() {
    $errors = libxml_get_errors();
    foreach ($errors as $error) {
        print libxml_display_error($error);
    }
    libxml_clear_errors();
}

?>