# HG changeset patch
# User Chloe Laisne <chloe.laisne@gmail.com>
# Date 1476948384 -19800
# Node ID c622fa18eb32278392ed4618baa8eb91024977cd
# Parent  e952c8a31a2bdc1e01abc98ece62dee9b455f52e# Parent  796ebdbf6a258b8f8f3d7cf04ad5bbae595bee2f
Merge

diff -r e952c8a31a2b -r c622fa18eb32 build/build.sh
--- a/build/build.sh	Thu Oct 20 12:54:07 2016 +0530
+++ b/build/build.sh	Thu Oct 20 12:56:24 2016 +0530
@@ -20,7 +20,7 @@
 
     echoblue "---> preparing bo client"
     pushd ../server/bo_client
-    npm install
+    /usr/local/bin/npm install
     ./node_modules/.bin/bower install
     popd
     echoblue "---> preparing bo client done"
@@ -28,14 +28,14 @@
     echoblue "---> preparing back"
     pushd ../server/src
     php composer.phar install
-    npm install
+    /usr/local/bin/npm install
     ./node_modules/.bin/bower install
     popd
     echoblue "---> preparing back done"
 
     echoblue "---> preparing app-client"
     pushd ../cms/app-client
-    npm install
+    /usr/local/bin/npm install
     ./node_modules/.bin/bower install
     popd
     echoblue "---> preparing app-client done"
@@ -48,8 +48,14 @@
     echoblue "---> checking vagrant"
     if vagrant status | grep -q "running"; then
         echoblue "---> starting vagrant"
+        if [ type workon ]; then
+            workon ansible
+        fi
         vagrant up
         VAGRANT_STARTED=true
+        if [ type deactivate ]; then
+            deactivate
+        fi
     fi
 
     echoblue "---> done"
diff -r e952c8a31a2b -r c622fa18eb32 build/post_upgrade.sh
--- a/build/post_upgrade.sh	Thu Oct 20 12:54:07 2016 +0530
+++ b/build/post_upgrade.sh	Thu Oct 20 12:56:24 2016 +0530
@@ -3,9 +3,9 @@
 
 if [ -x /usr/local/bin/drush ]; then
     echo "Reactivate corpus module"
-    /opt/remi/php56/root/usr/bin/php /usr/local/bin/drush -r /var/www/corpusdelaparole/drupal pm-enable -y corpus
+    /usr/local/bin/drush -r /var/www/corpusdelaparole/drupal pm-enable -y corpus
     echo "Clear drupal cache"
-    /opt/remi/php56/root/usr/bin/php /usr/local/bin/drush -r /var/www/corpusdelaparole/drupal cc all
+    /usr/local/bin/drush -r /var/www/corpusdelaparole/drupal cc all
 fi
 
 echo "Restart httpd"
diff -r e952c8a31a2b -r c622fa18eb32 build/pre_upgrade.sh
--- a/build/pre_upgrade.sh	Thu Oct 20 12:54:07 2016 +0530
+++ b/build/pre_upgrade.sh	Thu Oct 20 12:56:24 2016 +0530
@@ -3,5 +3,5 @@
 
 if [ -x /usr/local/bin/drush ]; then
     echo "Deactivate corpus module"
-    /opt/remi/php56/root/usr/bin/php /usr/local/bin/drush -r /var/www/corpusdelaparole/drupal pm-disable -y corpus
+    /usr/local/bin/drush -r /var/www/corpusdelaparole/drupal pm-disable -y corpus
 fi
diff -r e952c8a31a2b -r c622fa18eb32 build/set-version.sh
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/build/set-version.sh	Thu Oct 20 12:56:24 2016 +0530
@@ -0,0 +1,67 @@
+#!/usr/bin/env bash
+
+set -e
+
+function usage() {
+    cat <<EOF
+Usage: $0 "x.y.z"
+    "x.y.z": version number, following the semver convention
+EOF
+}
+
+if [ $# -ne 1 ]; then
+    echo "Bad number of parameters"
+    usage >&2
+    exit 1
+fi
+
+if [[ "$1" == "-h" || "$1" == "--help" ]]; then
+    usage
+    exit 0
+fi
+
+version=$1
+
+echo "Version: ${version}"
+
+[[ "${version}" =~ ^[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+$ ]] || {
+    echo "Bad format for version."
+    usage >&2
+    exit 1
+}
+
+DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+green="\x1B[0;32m" # '\e[1;32m' is too bright for white bg.
+blue="\x1B[1;34m"
+endColor="\x1B[0m"
+
+function echoblue() {
+    echo -e "${blue}$1${endColor}"
+}
+
+pushd "$DIR"
+
+pushd ../server/bo_client
+echoblue "Setting bo_client version"
+/usr/local/bin/npm run set-version -- "$version"
+popd
+
+pushd ../server/src
+echoblue "Setting back version"
+sed -i "" -e "s/\([:space:]*\'version\'[[:space:]]*\=\>[[:space:]]*\'\)[\.0-9]*\(\'.*\)/\1${version}\2/" config/version.php
+popd
+
+pushd ../cms
+echoblue "Setting cms version"
+/usr/local/bin/npm run set-version -- "$version"
+popd
+
+pushd ../cms/app-client
+echoblue "Setting app-client version"
+/usr/local/bin/npm run set-version -- "$version"
+popd
+
+echoblue "Version has been changed, please commit and push if necessary."
+
+
+popd
diff -r e952c8a31a2b -r c622fa18eb32 cms/app-client/package.json
--- a/cms/app-client/package.json	Thu Oct 20 12:54:07 2016 +0530
+++ b/cms/app-client/package.json	Thu Oct 20 12:56:24 2016 +0530
@@ -1,6 +1,6 @@
 {
   "name": "app-client",
-  "version": "0.0.2",
+  "version": "0.0.3",
   "description": "Small description for app-client goes here",
   "private": true,
   "directories": {
@@ -15,7 +15,8 @@
     "dl-transcripts": "ember dl-fixtures -u http://127.0.0.1:8000/api/v1/documents/ -d mirage/fixtures/transcripts.js -t transcripts -e \"11280.100/crdo-UVE_MOCIKA_SOUND,11280.100/crdo-CFPP2000_11_SOUND,11280.100/crdo-FRA_PK_IV_10_SOUND,11280.100/crdo-FSL-CUC023_SOUND,11280.100/crdo-NEE_KHIAAK_KO_AK_SOUND,11280.100/crdo-ESLO1_ENT_047\" -p 2 -f es6",
     "dl-themes": "ember dl-fixtures -u http://127.0.0.1:8000/api/v1/stats/themes/ -d mirage/fixtures/themes.js -t themes -f es6",
     "dl-discourses": "ember dl-fixtures -u http://127.0.0.1:8000/api/v1/stats/discourses/ -d mirage/fixtures/discourses.js -t discourses -f es6",
-    "prepare-maps": "node lib/scripts/prepare_maps.js"
+    "prepare-maps": "node lib/scripts/prepare_maps.js",
+    "set-version": "bash -c 'sed -i \"\" -e \"s/\\([:space:]*\\\"version\\\"[[:space:]]*\\:[[:space:]]*\\\"\\)[\\.0-9]*\\(\\\".*\\)/\\1${1}\\2/\" package.json' 0"
   },
   "repository": "",
   "engines": {
diff -r e952c8a31a2b -r c622fa18eb32 cms/package.json
--- a/cms/package.json	Thu Oct 20 12:54:07 2016 +0530
+++ b/cms/package.json	Thu Oct 20 12:56:24 2016 +0530
@@ -1,9 +1,10 @@
 {
   "name": "corpus-cms",
-  "version": "0.0.1",
+  "version": "0.0.3",
   "description": "drupal module to embed the Ember app \"app-client\"",
   "scripts": {
-    "build": "npm run build --prefix app-client/ && gulp copy-module"
+    "build": "npm run build --prefix app-client/ && gulp copy-module",
+    "set-version": "bash -c 'sed -i \"\" -e \"s/\\([:space:]*\\\"version\\\"[[:space:]]*\\:[[:space:]]*\\\"\\)[\\.0-9]*\\(\\\".*\\)/\\1${1}\\2/\" package.json' 0"
   },
   "author": "IRI",
   "devDependencies": {
diff -r e952c8a31a2b -r c622fa18eb32 readme.md
--- a/readme.md	Thu Oct 20 12:54:07 2016 +0530
+++ b/readme.md	Thu Oct 20 12:56:24 2016 +0530
@@ -1,9 +1,23 @@
 # Projet Corpus de la parole
 
 ## Prérequis
-- Vagrant
-- Ansible
-- node
+La plateforme de dévelopement visée est MacOS et Linux.
+Les éléments suivant sont des prérequis:
+  - php >= 5.6
+  - Vagrant
+  - Ansible.
+  - node
+
+### php
+
+### node
+
+### Vagrant
+Vagrant est utilisé pour le développement du back-office et la contruction des pacquets RPM.
+
+### Ansible
+Ansible n'est nécessaire que pour lancer l'environement de contruction des RPM.
+Ansible sera soit installé globalement, soit installé dans un environement virtuel. dans ce cas, `virtualenv-wrapper` doit être installé et l'environement virtuel doit avoir le nom `ansible`.
 
 ## Lancement VM dev
 - `cd /dev`
@@ -22,22 +36,28 @@
 ## Initialisation env de dev
 - `cd /server/bo_client`
 - `npm install`
-- `bower install`
+- `./node_modules/.bin/bower install`
 - `cd /server/src`
 - télécharger composer.phar (https://getcomposer.org/) et le mettre dans le répertoire server/src
 - `php composer.phar install`
 - `npm install`
-- `bower install`
+- `./node_modules/.bin/bowerbower install`
 
 ## Commandes bo_client
 - `ember serve`
 - `ember build` -> résultat dans `server/bo_client/dist`
 
+
 ## Commandes serveur php
 - Compilation css/js : `gulp` -> cette commande lance `ember build` de `/server/bo_client`
 - `php artisan serve`
 
 ## Creation de version
+- mettre à jour le numéro de version. Pour cela lancer le script `build/set-version.sh "x.y.z"`, avec "x.y.z" le numéro de version
+- lancer le script `build/build.sh`. options:
+    - `-d`: build debug
+    - `-p`: build de production
+
 - mettre à jour le fichier version `server/src/config/config/version.php`
 - mettre à jour le fichier version `cms/app-client/package.json`
 - `cd /server/src`
diff -r e952c8a31a2b -r c622fa18eb32 server/bo_client/package.json
--- a/server/bo_client/package.json	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/bo_client/package.json	Thu Oct 20 12:56:24 2016 +0530
@@ -1,6 +1,6 @@
 {
   "name": "bo-client",
-  "version": "0.0.2",
+  "version": "0.0.3",
   "description": "Bo client for Corpus Parole",
   "private": true,
   "directories": {
@@ -11,7 +11,8 @@
     "build": "ember build",
     "start": "ember server",
     "test": "ember test",
-    "dl-documents": "ember dl-fixtures -u http://127.0.0.1:8000/api/v1/documents/ -d server/fixtures/documents.js -p 2 -t documents -f require"
+    "dl-documents": "ember dl-fixtures -u http://127.0.0.1:8000/api/v1/documents/ -d server/fixtures/documents.js -p 2 -t documents -f require",
+    "set-version": "bash -c 'sed -i \"\" -e \"s/\\([:space:]*\\\"version\\\"[[:space:]]*\\:[[:space:]]*\\\"\\)[\\.0-9]*\\(\\\".*\\)/\\1${1}\\2/\" package.json' 0"
   },
   "repository": "",
   "engines": {
diff -r e952c8a31a2b -r c622fa18eb32 server/src/app/Console/Commands/IndexDocuments.php
--- a/server/src/app/Console/Commands/IndexDocuments.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/app/Console/Commands/IndexDocuments.php	Thu Oct 20 12:56:24 2016 +0530
@@ -104,7 +104,7 @@
                             ]
                         ],
                         'date' => [ 'type' => 'date', 'index' => 'not_analyzed'],
-                        'geonames_hyerarchy' => [ 'type' => 'string', 'index' => 'not_analyzed'],
+                        'geonames_hierarchy' => [ 'type' => 'string', 'index' => 'not_analyzed'],
                         'location' => [ 'type' => 'geo_point'],
                         'creation_date' => ['type' => 'date', 'index' => 'not_analyzed'],
                         'language' => ['type' => 'string', 'index' => 'not_analyzed'],
diff -r e952c8a31a2b -r c622fa18eb32 server/src/app/Http/Controllers/Api/DocumentController.php
--- a/server/src/app/Http/Controllers/Api/DocumentController.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/app/Http/Controllers/Api/DocumentController.php	Thu Oct 20 12:56:24 2016 +0530
@@ -31,10 +31,27 @@
         $perPage = intval($request->input('perpage', config('corpusparole.documents_per_page')));
 
         $filters = [];
-        $languages = $request->input('language', []);
+
+        $languages = CorpusFilterManager::prepareLanguages($request->input('language', []));
         if(!empty($languages)) {
-            $filters['language'] = CorpusFilterManager::prepareLanguages($languages);
+            $filters['language'] = $languages;
+        }
+
+        $location = CorpusFilterManager::prepareLocation($request->input('location', ''));
+        if(!empty($location)) {
+            $filters['location'] = $location;
         }
+
+        $themes = CorpusFilterManager::prepareTheme($request->input('theme', []));
+        if(!empty($themes)) {
+            $filters['themes'] = $themes;
+        }
+
+        $discourses = CorpusFilterManager::prepareDiscourse($request->input('discourse', []));
+        if(!empty($discourses)) {
+            $filters['discourses'] = $discourses;
+        }
+
         $sort = $request->input('sort', null);
 
         $paginator = $this->documentRepository->paginate($filters, $perPage, config('corpusparole.pagination_page_param'), null, $sort);
diff -r e952c8a31a2b -r c622fa18eb32 server/src/app/Libraries/Filters/CorpusFilterManager.php
--- a/server/src/app/Libraries/Filters/CorpusFilterManager.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/app/Libraries/Filters/CorpusFilterManager.php	Thu Oct 20 12:56:24 2016 +0530
@@ -9,7 +9,7 @@
     public static function getLanguageNode($nodeId) {
         $node = null;
         $currentNodes = [config('corpusparole.languages_treemap'),];
-        while(is_null($node) || !empty($currentNodes)) {
+        while(is_null($node) && !empty($currentNodes)) {
             $currentNode = array_pop($currentNodes);
             foreach($currentNode['children'] as $n) {
                 if($n['id'] === $nodeId) {
@@ -40,6 +40,9 @@
 
     public static function prepareLanguages($languages) {
 
+        if(is_null($languages)) {
+            return [];
+        }
         $resLanguage = [];
         if(is_string($languages)) {
             $languages = [ $languages, ];
@@ -71,4 +74,97 @@
         ];
     }
 
+    public static function prepareLocation($location) {
+        if(empty($location)) {
+            return null;
+        }
+        if(is_array($location)) {
+            $location = $location[0]; //from test above we know ther is at least one element
+        }
+        if(preg_match(config('corpusparole.geonames_url_regexp'), $location, $m)) {
+            $location = $m[1];
+        }
+        // for the moment we do not make this check. Incorrect loaction format will return an empty result
+        /*if(!preg_match('/\d+/', $location)) {
+            return null;
+        }*/
+
+        return $location;
+    }
+
+    /**
+     * Search in geonames_hierarchy (values are bae geonames ids)
+     */
+    public static function getLocationFilterPart($location) {
+        return [
+            'term' => [ 'geonames_hierarchy' => "$location" ]
+        ];
+    }
+
+    public static function prepareTheme($entities) {
+        if(empty($entities)) {
+            return [];
+        }
+        if(is_string($entities)) {
+            $entities = [$entities,];
+        }
+        return array_reduce($entities, function($res, $e) {
+            if(preg_match(config('corpusparole.bnf_ark_url_regexp'), $e, $m)) {
+                array_push($res, $m[1]);
+            } elseif(Utils::startsWith($e, config('corpusparole.bnf_ark_id_base'))) {
+                array_push($res, $e);
+            }
+
+            return $res;
+        }, []);
+    }
+
+    public static function getThemeFilterPart($themes) {
+        $nestedQueries = [];
+        foreach($themes as $t) {
+            $nestedQueries[] = [
+                'nested' => [
+                    'path' => "subject",
+                    'query' => [ 'term' => ['subject.code' => "$t" ] ]
+                ]
+            ];
+        }
+        return [
+            'bool' => [
+                    'must' => $nestedQueries
+            ]
+        ];
+    }
+
+    public static function prepareDiscourse($entities) {
+        if(empty($entities)) {
+            return [];
+        }
+        if(is_string($entities)) {
+            $entities = [$entities,];
+        }
+
+        return array_reduce($entities, function($res, $e) {
+            if(preg_match(config('corpusparole.bnf_ark_url_regexp'), $e)) {
+                array_push($res, $e);
+            } elseif(Utils::startsWith($e, config('corpusparole.bnf_ark_id_base'))) {
+                array_push($res, config('corpusparole.bnf_ark_base_url').$e);
+            }
+            return $res;
+        }, []);
+
+    }
+
+    public static function getDiscourseFilterPart($discourses) {
+        $subDiscQueries=array_map(function($d) {
+            return [ 'term' => ['discourse_types' => "$d" ]];
+        }, $discourses);
+
+        return [
+            'bool' => [
+                'must' => $subDiscQueries
+            ]
+        ];
+    }
+
 }
diff -r e952c8a31a2b -r c622fa18eb32 server/src/app/Models/Document.php
--- a/server/src/app/Models/Document.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/app/Models/Document.php	Thu Oct 20 12:56:24 2016 +0530
@@ -100,13 +100,13 @@
 
     public function getDiscourseTypes() {
         return array_values(array_filter($this->getTypes(), function($v) {
-            return $v instanceof Literal && $v->getDatatypeUri() === Config::get('corpusparole.olac_discourse_type')['uri'];
+            return $v instanceof Resource && preg_match(config('corpusparole.bnf_ark_url_regexp'), $v->getUri());
         }));
     }
 
     public function getOtherTypes() {
         $res = array_values(array_filter($this->getTypes(), function($v) {
-            return $v instanceof Resource || $v->getDatatypeUri() !== Config::get('corpusparole.olac_discourse_type')['uri'];
+            return !($v instanceof Resource) || !preg_match(config('corpusparole.bnf_ark_url_regexp'), $v->getUri());
         }));
         return $res;
     }
@@ -190,15 +190,14 @@
 
         //delete
         foreach($this->getDiscourseTypes() as $discourseType) {
-            $literalValue = new Literal($discourseType, null, Config::get('corpusparole.olac_discourse_type')['uri']);
-            $this->getProvidedCHO()->delete('dc11:type', $literalValue);
-            $this->currentDelta->getDeletedGraph()->add($this->getProvidedCHO(), 'dc11:type', new Literal($discourseType, null, Config::get('corpusparole.olac_discourse_type')['uri']));
+            $this->getProvidedCHO()->delete('dc11:type', $discourseType);
+            $this->currentDelta->getDeletedGraph()->addResource($this->getProvidedCHO(), 'dc11:type', $discourseType->getUri());
         }
 
         // and re-add them
         foreach($discoursesTypes as $dType) {
-            $this->getProvidedCHO()->add('dc11:type', new Literal($dType, null, Config::get('corpusparole.olac_discourse_type')['uri']));
-            $this->currentDelta->getAddedGraph()->add($this->getProvidedCHO(), 'dc11:type', new Literal($dType, null, Config::get('corpusparole.olac_discourse_type')['uri']));
+            $this->getProvidedCHO()->addResource('dc11:type', $dType);
+            $this->currentDelta->getAddedGraph()->addResource($this->getProvidedCHO(), 'dc11:type', $dType);
         }
 
         $this->clearMemoizationCache();
diff -r e952c8a31a2b -r c622fa18eb32 server/src/app/Repositories/RdfDocumentRepository.php
--- a/server/src/app/Repositories/RdfDocumentRepository.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/app/Repositories/RdfDocumentRepository.php	Thu Oct 20 12:56:24 2016 +0530
@@ -154,6 +154,29 @@
             }
             $qFilterParts[] = CorpusFilterManager::getLanguagesFilterPart($languages);
         }
+        if(array_key_exists('location', $filters) && !empty($filters['location'])) {
+            $location = $filters['location'];
+            if(is_array($location)) {
+                $location = $location[0]; // we know it is not empty
+            }
+            $qFilterParts[] = CorpusFilterManager::getLocationFilterPart($location);
+        }
+        if(array_key_exists('themes', $filters) && !empty($filters['themes'])) {
+            $themes = $filters['themes'];
+            if(is_string($themes)) {
+                $themes = [$themes,]; // we know it is not empty
+            }
+            $qFilterParts[] = CorpusFilterManager::getThemeFilterPart($themes);
+        }
+        if(array_key_exists('discourses', $filters) && !empty($filters['discourses'])) {
+            $discourses = $filters['discourses'];
+            if(is_string($discourses)) {
+                $discourses = [$discourses,]; // we know it is not empty
+            }
+            $qFilterParts[] = CorpusFilterManager::getDiscourseFilterPart($discourses);
+        }
+
+        Log::debug("queries : ".print_r($qFilterParts, true));
 
         $query = [
             'index' => config('corpusparole.elasticsearch_index'),
diff -r e952c8a31a2b -r c622fa18eb32 server/src/config/corpusparole.php
--- a/server/src/config/corpusparole.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/config/corpusparole.php	Thu Oct 20 12:56:24 2016 +0530
@@ -132,6 +132,7 @@
     'bnf_base_url' => 'http://data.bnf.fr/',
     'bnf_ark_base_url' => 'http://ark.bnf.fr/',
     'bnf_ark_url_regexp' => '/http[s]?\:\/\/(?:data|ark)\.bnf\.fr\/(ark\:\/12148\/[[:alnum:]]+)\/?/',
+    'bnf_ark_id_base' => "ark:/12148/",
     'bnf_cache_expiration' => 60*24*30,
     'bnf_max_ids' => 5,
     'bnf_query_url' => 'http://data.bnf.fr/sparql',
diff -r e952c8a31a2b -r c622fa18eb32 server/src/config/version.php
--- a/server/src/config/version.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/config/version.php	Thu Oct 20 12:56:24 2016 +0530
@@ -8,5 +8,5 @@
         | Must be incremented as needed.
         |
         */
-        'version' => '0.0.2'
+        'version' => '0.0.3'
     ];
diff -r e952c8a31a2b -r c622fa18eb32 server/src/tests/Libraries/Filters/CorpusFilterManagerTest.php
--- a/server/src/tests/Libraries/Filters/CorpusFilterManagerTest.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/tests/Libraries/Filters/CorpusFilterManagerTest.php	Thu Oct 20 12:56:24 2016 +0530
@@ -17,6 +17,18 @@
     }
 
     /**
+     * test prepare languages unknown
+     *
+     * @return void
+     */
+    public function testPrepareLanguagesUnkown()
+    {
+        $languagesInput = ['foobar'];
+        $languagesOutput = CorpusFilterManager::prepareLanguages($languagesInput);
+        $this->assertEquals($languagesOutput, ['http://lexvo.org/id/iso639-3/foobar']);
+    }
+
+    /**
      * test prepare languages addPrefix
      *
      * @return void
@@ -59,7 +71,7 @@
      */
     public function testPrepareLanguagesRecusionComplex()
     {
-        $languagesInput = ['fra', 'http://lexvo.org/id/iso639-3/gsw', 'corpus-oil', 'corpus-regional'];
+        $languagesInput = ['fra', 'http://lexvo.org/id/iso639-3/gsw', 'corpus-oil', 'corpus-regionals'];
         $languagesOutput = CorpusFilterManager::prepareLanguages($languagesInput);
         sort($languagesOutput);
         $languagesExpected = ['http://lexvo.org/id/iso639-3/fra', 'http://lexvo.org/id/iso639-3/gsw', 'http://lexvo.org/id/iso639-3/pcd',
@@ -69,5 +81,117 @@
         $this->assertEquals($languagesOutput, $languagesExpected);
     }
 
+    /**
+     * test prepare location no operation
+     *
+     * @return void
+     */
+    public function testPrepareLocationNoOp()
+    {
+        $locationInput = '3030293';
+        $locationOutput = CorpusFilterManager::prepareLocation($locationInput);
+        $this->assertEquals('3030293', $locationOutput);
+    }
+
+    /**
+     * test prepare location array
+     *
+     * @return void
+     */
+    public function testPrepareLocationArray()
+    {
+        $locationInput = ['3030293', 'foobar'];
+        $locationOutput = CorpusFilterManager::prepareLocation($locationInput);
+        $this->assertEquals('3030293', $locationOutput);
+    }
+
+
+    /**
+     * test prepare location unknown
+     *
+     * @return void
+     */
+    public function testPrepareLocationUnknown()
+    {
+        $locationInput = 'foobar';
+        $locationOutput = CorpusFilterManager::prepareLocation($locationInput);
+        $this->assertEquals('foobar', $locationOutput);
+    }
+
+    /**
+     * test prepare location geonames
+     *
+     * @return void
+     */
+    public function testPrepareLocationGeonames()
+    {
+        $locationInput = 'http://sws.geonames.org/3030293';
+        $locationOutput = CorpusFilterManager::prepareLocation($locationInput);
+        $this->assertEquals('3030293', $locationOutput);
+    }
+
+    /**
+     * test prepare location geonames array
+     *
+     * @return void
+     */
+    public function testPrepareLocationGeonamesArray()
+    {
+        $locationInput = ['http://sws.geonames.org/3030293', 'http://www.geonames.org/3017382'];
+        $locationOutput = CorpusFilterManager::prepareLocation($locationInput);
+        $this->assertEquals('3030293', $locationOutput);
+    }
+
+
+    /**
+     * test prepare themes noop
+     *
+     * @return void
+     */
+    public function testPrepareThemesNoOp()
+    {
+        $themesInput = ['ark:/12148/cb11937931x', 'ark:/12148/cb11946662b', 'ark:/12148/cb13318415c'];
+        $themesOutput = CorpusFilterManager::prepareTheme($themesInput);
+        $this->assertEquals($themesOutput, $themesInput);
+    }
+
+    /**
+     * test prepare themes full url
+     *
+     * @return void
+     */
+    public function testPrepareThemesFullUrl()
+    {
+        $themesInput = ['http://ark.bnf.fr/ark:/12148/cb11937931x', 'http://data.bnf.fr/ark:/12148/cb11946662b', 'https://ark.bnf.fr/ark:/12148/cb13318415c'];
+        $themesOutput = CorpusFilterManager::prepareTheme($themesInput);
+        $this->assertEquals($themesOutput, ['ark:/12148/cb11937931x', 'ark:/12148/cb11946662b', 'ark:/12148/cb13318415c']);
+    }
+
+
+    /**
+     * test prepare themes full url
+     *
+     * @return void
+     */
+    public function testPrepareThemesUnknown()
+    {
+        $themesInput = ['ark:/12148/cb11937931x', 'foo', 'ark:/12148/cb11946662b', 'ark:/12148/cb13318415c', 'bar'];
+        $themesOutput = CorpusFilterManager::prepareTheme($themesInput);
+        $this->assertEquals($themesOutput, ['ark:/12148/cb11937931x', 'ark:/12148/cb11946662b', 'ark:/12148/cb13318415c']);
+    }
+
+    /**
+     * test prepare themes mix
+     *
+     * @return void
+     */
+    public function testPrepareThemesMix()
+    {
+        $themesInput = ['ark:/12148/cb11937931x', 'foo', 'http://data.bnf.fr/ark:/12148/cb11946662b', 'ark:/12148/cb13318415c', 'bar'];
+        $themesOutput = CorpusFilterManager::prepareTheme($themesInput);
+        $this->assertEquals($themesOutput, ['ark:/12148/cb11937931x', 'ark:/12148/cb11946662b', 'ark:/12148/cb13318415c']);
+    }
+
+
 
 }
diff -r e952c8a31a2b -r c622fa18eb32 server/src/tests/Models/DocumentTest.php
--- a/server/src/tests/Models/DocumentTest.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/tests/Models/DocumentTest.php	Thu Oct 20 12:56:24 2016 +0530
@@ -140,7 +140,7 @@
     public function testGetOtherTypes() {
         $doc = new Document(config('corpusparole.corpus_doc_id_base_uri')."crdo-CFPP2000_35_SOUND", $this->inputGraphes['TEST']);
 
-        $this->assertCount(5, $doc->getOtherTypes(), "types array must be of size 5");
+        $this->assertCount(6, $doc->getOtherTypes(), "types array must be of size 5");
 
         foreach($doc->getTypes() as $type) {
             $this->assertThat(
@@ -156,12 +156,11 @@
     public function testGetDiscourseTypes() {
         $doc = new Document(config('corpusparole.corpus_doc_id_base_uri')."crdo-CFPP2000_35_SOUND", $this->inputGraphes['TEST']);
 
-        $this->assertCount(3, $doc->getDiscourseTypes(), "types array must be of size 3");
+        $this->assertCount(2, $doc->getDiscourseTypes(), "types array must be of size 3");
 
-        $this->assertContainsOnlyInstancesOf("EasyRdf\Literal", $doc->getDiscourseTypes(), "Result contains only literals");
+        $this->assertContainsOnlyInstancesOf("EasyRdf\Resource", $doc->getDiscourseTypes(), "Result contains only Resources");
         $type = $doc->getDiscourseTypes()[0];
-        $this->assertEquals("narrative", $type->getValue(), "discourse type is narrative");
-        $this->assertEquals("http://www.language-archives.org/OLAC/1.1/discourse-type", $type->getDatatypeUri(), "discourse type url");
+        $this->assertEquals("http://ark.bnf.fr/ark:/12148/cb11932135w", $type->getUri(), "discourse type is cb11932135w");
     }
 
     public function testCloneDocument() {
@@ -188,14 +187,14 @@
 
     public function testUpdateDiscourseTypes() {
 
-        $newDiscourseTypes = ['oratory','dialogue','narrative', 'formulaic', 'ludic'];
+        $newDiscourseTypes = ["http://ark.bnf.fr/ark:/12148/cb11932135w" , "http://ark.bnf.fr/ark:/12148/cb13319048g", "http://ark.bnf.fr/ark:/12148/cb11949715t" , "http://ark.bnf.fr/ark:/12148/cb119783362"];
 
         $doc = new Document(config('corpusparole.corpus_doc_id_base_uri')."crdo-CFPP2000_35_SOUND", $this->inputGraphes['TEST']);
-        $this->assertCount(3, $doc->getDiscourseTypes(), "types array must be of size 3");
+        $this->assertCount(2, $doc->getDiscourseTypes(), "types array must be of size 2");
 
         $doc->updateDiscourseTypes($newDiscourseTypes);
 
-        $this->assertCount(5, $doc->getDiscourseTypes(), "types array must be of size 5");
+        $this->assertCount(4, $doc->getDiscourseTypes(), "types array must be of size 4");
 
         $discourseTypes = $doc->getDiscourseTypes();
         foreach($newDiscourseTypes as $dt) {
@@ -206,7 +205,7 @@
 
     public function testUpdateDiscourseTypesDelta() {
 
-        $newDiscourseTypes = ['oratory','dialogue','narrative', 'formulaic', 'ludic'];
+        $newDiscourseTypes = ["http://ark.bnf.fr/ark:/12148/cb11932135w" , "http://ark.bnf.fr/ark:/12148/cb13319048g", "http://ark.bnf.fr/ark:/12148/cb11949715t" , "http://ark.bnf.fr/ark:/12148/cb119783362"];
 
         $doc = new Document(config('corpusparole.corpus_doc_id_base_uri')."crdo-CFPP2000_35_SOUND", $this->inputGraphes['TEST']);
 
@@ -221,13 +220,13 @@
         $this->assertEquals(config('corpusparole.corpus_doc_id_base_uri')."crdo-CFPP2000_35_SOUND", $delta->getDeletedGraph()->getUri(), "uri of deleted graph must be ok");
         $this->assertEquals(config('corpusparole.corpus_doc_id_base_uri')."crdo-CFPP2000_35_SOUND", $delta->getAddedGraph()->getUri(), "uri of added graph must be ok");
 
-        $this->assertEquals(3, $delta->getDeletedGraph()->countTriples(), "deleted graph must have only 3 triples");
-        $this->assertEquals(5, $delta->getAddedGraph()->countTriples(), "deleted graph must have only 5 triples");
+        $this->assertEquals(2, $delta->getDeletedGraph()->countTriples(), "deleted graph must have only 2 triples");
+        $this->assertEquals(4, $delta->getAddedGraph()->countTriples(), "deleted graph must have only 4 triples");
 
-        $resQueryDiscourseType = $delta->getAddedGraph()->allLiterals($doc->getProvidedCHO(), 'dc11:type');
+        $resQueryDiscourseType = $delta->getAddedGraph()->allResources($doc->getProvidedCHO(), 'dc11:type');
         foreach($resQueryDiscourseType as $dt) {
-            $this->assertInstanceOf(EasyRdf\Literal::class, $dt, "This must be a litteral");
-            $this->assertEquals('http://www.language-archives.org/OLAC/1.1/discourse-type', $dt->getDatatypeUri(), "The type of the Litteral must be correct");
+            $this->assertInstanceOf(EasyRdf\Resource::class, $dt, "This must be a Resource");
+            $this->assertRegexp(config('corpusparole.bnf_ark_url_regexp'), $dt->getUri(), "The resource uri must be bnf ark");
         }
         foreach($newDiscourseTypes as $dt) {
             $this->assertContains($dt, $resQueryDiscourseType, "all discourse types must be in result list");
diff -r e952c8a31a2b -r c622fa18eb32 server/src/tests/Repositories/DocumentRepositoryIntegrationTest.php
--- a/server/src/tests/Repositories/DocumentRepositoryIntegrationTest.php	Thu Oct 20 12:54:07 2016 +0530
+++ b/server/src/tests/Repositories/DocumentRepositoryIntegrationTest.php	Thu Oct 20 12:56:24 2016 +0530
@@ -210,7 +210,8 @@
     public function testSave() {
 
         $doc = new Document(config('corpusparole.corpus_doc_id_base_uri')."crdo-ALA_738", $this->graph);
-        $doc->updateDiscourseTypes(['oratory','drama','narrative']);
+        $updatedDiscourseTypes = ["http://ark.bnf.fr/ark:/12148/cb13319048g", "http://ark.bnf.fr/ark:/12148/cb11949715t" , "http://ark.bnf.fr/ark:/12148/cb119783362"];
+        $doc->updateDiscourseTypes($updatedDiscourseTypes);
 
         $res = $this->documentRepository->save($doc);
 
@@ -220,18 +221,19 @@
 
         $discoursesTypes = $res->getDiscourseTypes();
 
-        $this->assertCount(3, $discoursesTypes, "types array must be of size 1");
+        $this->assertCount(3, $discoursesTypes, "types array must be of size 3");
 
-        $this->assertContainsOnlyInstancesOf("EasyRdf\Literal", $discoursesTypes, "Result contains only literals");
+        $this->assertContainsOnlyInstancesOf("EasyRdf\Resource", $discoursesTypes, "Result contains only resources");
 
         $newDiscoursesTypes = [];
         foreach($discoursesTypes as $dt) {
-            array_push($newDiscoursesTypes, $dt->getValue());
-            $this->assertContains($dt->getValue(),['oratory','drama','narrative'],'Value in [oratory,drama,narrative]');
-            $this->assertEquals("http://www.language-archives.org/OLAC/1.1/discourse-type", $dt->getDatatypeUri(), "discourse type url");
+            array_push($newDiscoursesTypes, $dt->getUri());
+            $this->assertContains($dt->getUri(), $updatedDiscourseTypes,"Value in ".print_r($updatedDiscourseTypes, true));
         }
         sort($newDiscoursesTypes);
-        $this->assertEquals(['drama', 'narrative', 'oratory'], $newDiscoursesTypes, "array type must the same");
+        sort($updatedDiscourseTypes);
+
+        $this->assertEquals($updatedDiscourseTypes, $newDiscoursesTypes, "array type must the same");
     }
 
     public function testCount() {