Workaround to avoid SCM repository names duplicates due to cvsanaly bug

Workaround to get SCM repository names using uri instead of name
due to bug MetricsGrimoire/CVSAnalY#100

vizgrimoire/SCM.py

@@ -249,7 +249,8 @@ def create_filter_report_top(filter_, period, startdate, enddate, destdir, npeop
             items = [items]
 
         fn = os.path.join(destdir, filter_.get_filename(SCM()))
-        createJSON(items, fn)
+        escaped_items = [i.replace('/','_') for i in items]
+        createJSON(escaped_items, fn)
 
         for item in items :
             item_name = "'"+ item+ "'"
@@ -276,12 +277,9 @@ def create_filter_report(filter_, period, startdate, enddate, destdir, npeople,
             items = [items]
 
         fn = os.path.join(destdir, filter_.get_filename(SCM()))
-        createJSON(items, fn)
-
-        if filter_name in ("domain", "company", "repository"):
-            items_list = {'name' : [], 'commits_365' : [], 'authors_365' : []}
-        else:
-            items_list = items
+        escaped_items = [i.replace('/','_') for i in items]
+        logging.info("D1")
+        createJSON(escaped_items, fn)
 
         for item in items :
             item_name = "'"+ item+ "'"
@@ -296,16 +294,8 @@ def create_filter_report(filter_, period, startdate, enddate, destdir, npeople,
             fn = os.path.join(destdir, filter_item.get_static_filename(SCM()))
             createJSON(agg, fn)
 
-            if filter_name in ("domain", "company", "repository"):
-                items_list['name'].append(item.replace('/', '_'))
-                items_list['commits_365'].append(agg['commits_365'])
-                items_list['authors_365'].append(agg['authors_365'])
-
         SCM.create_filter_report_top(filter_, period, startdate, enddate, destdir, npeople, identities_db)
 
-        fn = os.path.join(destdir, filter_.get_filename(SCM()))
-        createJSON(items_list, fn)
-
         if (filter_name == "company"):
             ds = SCM
             #summary =  SCM.get_filter_summary(filter_, period, startdate, enddate, identities_db, 10)

vizgrimoire/data_source.py

@@ -518,6 +518,9 @@ def convert_all_to_single(cls, data, filter_, destdir, evolutionary, period='mon
         if cls == ITS or cls == SCM:
             if filter_.get_name() in ["company+country","company+project"]:
                 data['name'] = data.pop('filter')
+                data['name'] = [item.replace('/', '_') for item in data['name']]
+        if cls == SCM:
+            data['name'] = [item.replace('/', '_') for item in data['name']]
 
         if not evolutionary:
             # First create the JSON with the list of items

vizgrimoire/metrics/query_builder.py

@@ -315,7 +315,7 @@ def get_group_field (ds_query, filter_type):
             if ds_query == MLSQuery:
                 field = "DISTINCT(SUBSTR(mp.email_address,LOCATE('@',mp.email_address)+1)) as name"
         elif analysis == "repository":
-            field = "r.name"
+            field = "r.uri AS name"
             if ds_query == ITSQuery: field = "t.url"
             elif ds_query == SCRQuery: field = "t.url"
             elif ds_query == MLSQuery: field = "ml.mailing_list_url"
@@ -500,11 +500,11 @@ def GetSQLRepositoriesWhere (self, repository):
         if isinstance(repository, list):
             repo_filter = "("
             for repo in repository:
-                repo_filter = repo_filter + "r.name = " + repo + " or "
+                repo_filter = repo_filter + "r.uri = " + repo + " or "
             repo_filter = repo_filter[:-4] + ")"
             fields.add(repo_filter)
         elif repository is not None:
-            fields.add("r.name = "+ repository)
+            fields.add("r.uri = "+ repository)
 
         return fields
 

vizgrimoire/metrics/scm_metrics.py

@@ -945,7 +945,7 @@ def _get_sql(self, evolutionary):
     def get_list(self):
         """Repositories list ordered by number of commits"""
         q = """
-            select count(distinct(sid)) as total, name
+            select count(distinct(sid)) as total, r.uri AS name
             from repositories r, (
               select distinct(s.id) as sid, repository_id from actions a, scmlog s
               where s.id = a.commit_id  and s.author_date >=%s and s.author_date < %s) t