[Blends-commit] r3010 - /blends/trunk/team_analysis_tools/upload_history.py

[tille at users.alioth.debian.org]

Author: tille
Date: Tue Nov  1 17:17:29 2011
New Revision: 3010

URL: http://svn.debian.org/wsvn/blends/?sc=1&rev=3010
Log:
Adapted to PG 9.1

Modified:
    blends/trunk/team_analysis_tools/upload_history.py

Modified: blends/trunk/team_analysis_tools/upload_history.py
URL: http://svn.debian.org/wsvn/blends/blends/trunk/team_analysis_tools/upload_history.py?rev=3010&op=diff
==============================================================================
--- blends/trunk/team_analysis_tools/upload_history.py (original)
+++ blends/trunk/team_analysis_tools/upload_history.py Tue Nov  1 17:17:29 2011
@@ -2,8 +2,10 @@
 # Copyright 2011: Andreas Tille <tille at debian.org>
 # License: GPL
 
-# MAXUPLOADERS=20
-MAXUPLOADERS=10
+# PLOTUPLOADERS=20
+PLOTUPLOADERS=10
+
+MAXUPLOADERS=1000
 
 teams = { 
           'debian-med' :       'debian-med-packaging at lists.alioth.debian.org'      ,
@@ -40,6 +42,7 @@
 
 try:
     conn = psycopg2.connect(host="localhost",port=PORT,user="guest",database="udd")
+    conn.set_session(autocommit) # make sure it is possible to continue after a query error
 except psycopg2.OperationalError:
     try:
         conn = psycopg2.connect(host="localhost",port=DEFAULTPORT,user="guest",database="udd")
@@ -75,17 +78,21 @@
     return result
 
 for team in teams.keys():
+    # print team
     datafile='uploaders_'+team+'.txt'
     out = open(datafile, 'w')
-    query = "SELECT replace(uploader,' ','_') AS uploader FROM active_uploader_names_of_pkggroup('%s', 1000) AS (uploader text);" % (teams[team])
+    query = "SELECT replace(uploader,' ','_') AS uploader FROM active_uploader_names_of_pkggroup('%s', %i) AS (uploader text);" % (teams[team], MAXUPLOADERS)
     # print query
     curs.execute(query)
 
+    nuploaders = curs.rowcount
+    if nuploaders == 0:
+	print >>stderr, "No uploaders found for team %s" % team
+	continue
+
     print >>out, ' year',
-    nuploaders = 0
     for row in curs.fetchall():
         print >>out, '\t' + re.sub('^(.*_\w)[^_]*$', '\\1', row[0]),
-        nuploaders += 1
     print >>out, ''
 
     typestring = 'year text'
@@ -98,19 +105,38 @@
                      FROM active_uploader_per_year_of_pkggroup(''%s'', %i) AS (name text, year int, count int)',
              'SELECT * FROM active_uploader_names_of_pkggroup(''%s'', %i) AS (category text)'
         ) As (%s)
-""" % (teams[team], nuploaders, teams[team], nuploaders, typestring)
+""" % (teams[team], nuploaders, teams[team], MAXUPLOADERS, typestring)
 
     try:
 	# print query
         curs.execute(query)
     except psycopg2.ProgrammingError, err:
 	if crosstab_missing_re.match(str(err)):
-	    print >>stderr, """Please do
-	psql udd < /usr/share/postgresql/<pgversion>/contrib/tablefunc.sql
-before calling this program."""
+#	    print >>stderr, """Please do
+#	psql udd < /usr/share/postgresql/<pgversion>/contrib/tablefunc.sql
+#before calling this program."""
+	    print >>stderr, "Please do `psql udd -c 'CREATE EXTENSION tablefunc;'` before calling this program.\n", err, query, nuploaders
+	    exit(-1)
 	else:
-            print >>stderr, "To few uploaders in %s team.\n%s" % (team, err)
-	exit(-1)
+	    m = re.match(".*\n.*Query-specified return tuple has (\d+) columns but crosstab returns (\d+).*", str(err))
+	    nuploaders_calculated = int(m.group(1))
+	    nuploaders_returned   = int(m.group(2))
+	    # somehow the crosstable returns less columns
+	    print >>stderr, "Warning: team %s should have %d uploaders but crosstable returned only %d. Try again with this number." % (team, nuploaders_calculated, nuploaders_returned)
+	    typestring = 'year text'
+	    for i in range(nuploaders_returned - 1):
+    		typestring = typestring + ', upl' + str(i+1) + ' int'
+		query = """SELECT *
+	FROM 
+	crosstab(
+	     'SELECT year AS row_name, name AS bucket, count AS value
+                     FROM active_uploader_per_year_of_pkggroup(''%s'', %i) AS (name text, year int, count int)',
+             'SELECT * FROM active_uploader_names_of_pkggroup(''%s'', %i) AS (category text)'
+        ) As (%s)
+""" % (teams[team], nuploaders, teams[team], nuploaders, typestring)
+	    # print query
+	    conn.rollback()
+            curs.execute(query)
     for row in curs.fetchall():
         print >>out, ' ' + row[0] ,
         for v in row[1:]:
@@ -120,7 +146,7 @@
                 print >>out, '\t0',
         print >>out, ''
     out.close()
-    cmdstring='./author_stats_create_graph ' + datafile + ' ' + str(min(nuploaders, MAXUPLOADERS))
+    cmdstring='./author_stats_create_graph ' + datafile + ' ' + str(min(nuploaders, PLOTUPLOADERS))
     if len(argv) > 1 :
 	if argv[1].startswith('pdf'):
 	    cmdstring = cmdstring + ' "" pdf'