Ajout librairies dépendantes - cf #150

PnX-SI · Dec 6, 2017 · d5ff62f · d5ff62f
1 parent 4ce58a5
commit d5ff62f
Show file tree

Hide file tree

Showing 2 changed files with 16 additions and 9 deletions.
diff --git a/data/scripts/import_wikimedia_commons/functions.py b/data/scripts/import_wikimedia_commons/functions.py
@@ -1,6 +1,8 @@
 
 import requests
 import psycopg2
+from lxml import etree
+import xmltodict
 
 from SPARQLWrapper import SPARQLWrapper, JSON
 
@@ -24,7 +26,6 @@ def main(dbconnexion, cd_refs, refreshAtlas=True, simulate=True):
       ?item wdt:P3186 '%s'
      SERVICE wikibase:label { bd:serviceParam wikibase:language "fr" }
     } LIMIT 200"""
-
     sparql = SPARQLWrapper("https://query.wikidata.org/sparql")
 
     sqlI = """INSERT INTO taxonomie.t_medias
@@ -42,11 +43,9 @@ def main(dbconnexion, cd_refs, refreshAtlas=True, simulate=True):
             for result in results["results"]["bindings"]:
                 if (result['image']['value']):
                     print(' -- INSERT IMAGE')
-                    from lxml import etree
                     # Recuperation des donnees sur commons
                     url = "https://tools.wmflabs.org/magnus-toolserver/commonsapi.php?image=%s" % result['image']['value'].split('Special:FilePath/', 1 )[1]
                     r = requests.get(url)
-                    import xmltodict
                     a = xmltodict.parse(r.content)
                     try:
                         aut = 'Commons'

diff --git a/data/scripts/import_wikimedia_commons/run_import.py b/data/scripts/import_wikimedia_commons/run_import.py
@@ -20,6 +20,14 @@
         ALTER TABLE taxonomie.t_medias ADD COLUMN source varchar(25);
         ALTER TABLE taxonomie.t_medias ADD COLUMN licence varchar(100);
         ALTER TABLE taxonomie.t_medias ALTER COLUMN auteur TYPE character varying(1000);
+
+    Librairie requises (à installer via pip dans un virtualenv de préférence)
+    lxml
+    psycopg2
+    requests
+    SPARQLWrapper
+    xmltodict
+
 '''
 try:
     conn = psycopg2.connect(SQLALCHEMY_DATABASE_URI)
@@ -28,12 +36,12 @@
 
 try:
     cur = conn.cursor()
-    sql = """SELECT DISTINCT cd_ref
-        FROM taxonomie.bib_noms
-        LEFT OUTER JOIN taxonomie.t_medias USING(cd_ref)
-        WHERE id_media IS NULL
-    """
-    # sql = """SELECT cd_ref from taxonomie.bib_noms LIMIT 10"""
+    # sql = """SELECT DISTINCT cd_ref
+    #     FROM taxonomie.bib_noms
+    #     LEFT OUTER JOIN taxonomie.t_medias USING(cd_ref)
+    #     WHERE id_media IS NULL
+    # """
+    sql = """SELECT cd_ref from taxonomie.bib_noms LIMIT 10"""
     # sql = """SELECT cd_ref from atlas.vm_taxons_plus_observes LIMIT 100"""
     cur.execute(sql)
     rows = cur.fetchall()