From: olivier <olivier@3bf09e05-f825-4182-b9bc-eedd7160adf0>
Date: Thu, 11 Jun 2009 18:00:02 +0000 (+0000)
Subject: migration: add item keywords mapper
X-Git-Url: https://git.parisson.com/?a=commitdiff_plain;h=4ac37a09174affb39968edc75ef24af719ddf51b;p=telemeta-data.git

migration: add item keywords mapper

git-svn-id: http://svn.parisson.org/svn/crem@104 3bf09e05-f825-4182-b9bc-eedd7160adf0
---

diff --git a/trunk/import/migration/tasks/enums.py b/trunk/import/migration/tasks/enums.py
index e08932b..c701a46 100644
--- a/trunk/import/migration/tasks/enums.py
+++ b/trunk/import/migration/tasks/enums.py
@@ -50,8 +50,7 @@ class SimpleEnumMigrator(DataMigrator):
         'Terrain_ou_Autr':  'recording_contexts',
         'Numerisation':     'ad_conversions',
         'Form':             'vernacular_styles',
-        'FormStyl generi':  'generic_styles',
-        'Mot_Clef':         'context_keywords'
+        'FormStyl generi':  'generic_styles'
     }
 
     def get_name(self):
@@ -63,8 +62,6 @@ class SimpleEnumMigrator(DataMigrator):
             dest = self.map[src]
             if src == 'Form':
                 src_field = 'Form'
-            elif src == 'Mot_Clef':
-                src_field = 'Mot_Clef'
             else:
                 src_field = 'value'
 
diff --git a/trunk/import/migration/tasks/items.py b/trunk/import/migration/tasks/items.py
index d576f82..4b0f73d 100644
--- a/trunk/import/migration/tasks/items.py
+++ b/trunk/import/migration/tasks/items.py
@@ -116,3 +116,70 @@ class ItemsEnumMapper(EnumMapper):
     def process(self):
         EnumMapper.process(self, 'Phono', 'Cote_Phono', 'media_items', self.map)
 
+class ItemsKeywordsMapper(DataMigrator):
+    """Map many-to-many context keywords to items"""
+
+    implements(IDataMigrator)
+
+    def get_name(self):
+        return "items:keywords"
+
+    def process(self):
+        self.step()
+        self.target_cursor.execute("DELETE FROM context_keywords")
+        self.target_cursor.execute("INSERT INTO context_keywords (value) "
+                                   "SELECT Mot_Clef FROM %s.Mot_Clef WHERE Mot_Clef <> ''" 
+                                   % self.src_db_name)
+        self.step()
+        self.target_cursor.execute("INSERT INTO context_keywords (value) "
+                                   "SELECT DISTINCT(Mot_Clef) FROM %s.Fonction_Usage " 
+                                   "WHERE Mot_Clef <> '' AND Mot_Clef NOT IN (SELECT value FROM context_keywords)" 
+                                   % self.src_db_name)
+
+        self.step()
+        self.src_cursor.execute("SELECT * FROM Fonction_Usage GROUP BY Cote_Phono, Mot_Clef")
+        self.stats = {'relations': self.src_cursor.rowcount, 'converted': 0, 'nosuchitem': 0, 
+                      'nosuchkeyword': 0, 'empty': 0}
+        self.src_cursor.execute("SELECT * FROM Fonction_Usage WHERE Mot_Clef = '' GROUP BY Cote_Phono, Mot_Clef")
+        self.stats['empty'] = self.src_cursor.rowcount
+
+        self.step()
+        self.target_cursor.execute("DELETE from media_item_keywords")
+
+        self.step()
+        query = "INSERT INTO media_item_keywords (item_id, keyword_id) " \
+                "SELECT i.id, k.id FROM %s.Fonction_Usage AS f " \
+                "INNER JOIN media_items AS i ON f.Cote_Phono = i.old_code " \
+                "INNER JOIN context_keywords AS k ON f.Mot_Clef = k.value " \
+                "GROUP BY i.id, k.id"
+        self.target_cursor.execute(query % self.src_db_name)
+        self.stats['converted'] = self.target_cursor.rowcount
+
+        self.step()
+        query = "SELECT * FROM %s.Fonction_Usage AS f LEFT JOIN media_items AS i " \
+                "ON f.Cote_Phono = i.old_code WHERE i.old_code IS NULL " \
+                "GROUP BY f.Cote_Phono, f.Mot_Clef"
+        self.target_cursor.execute(query % self.src_db_name)
+        self.stats['nosuchitem'] = self.target_cursor.rowcount
+
+        self.step()
+        query = "SELECT * FROM %s.Fonction_Usage AS f " \
+                "LEFT JOIN context_keywords AS k ON f.Mot_Clef = k.value " \
+                "WHERE k.value IS NULL AND f.Mot_Clef <> '' GROUP BY f.Cote_Phono, f.Mot_Clef"
+        self.target_cursor.execute(query % self.src_db_name)
+        self.stats['nosuchkeyword'] = self.target_cursor.rowcount
+        if self.target_cursor.rowcount:
+            print "Unknown keywords:"
+            query = "SELECT f.Mot_Clef, COUNT(*) FROM %s.Fonction_Usage AS f " \
+                    "LEFT JOIN context_keywords AS k ON f.Mot_Clef = k.value " \
+                    "WHERE k.value IS NULL AND f.Mot_Clef <> '' GROUP BY f.Mot_Clef"
+            self.target_cursor.execute(query % self.src_db_name)
+            while True:
+                row = self.target_cursor.fetchone()
+                if not row:
+                    break
+                print "  %s: count=%d" % row
+                
+            
+            
+