* examine_package.py: Summarise duplicate copyright file entries

[dak.git] / dak / examine_package.py

diff --git a/dak/examine_package.py b/dak/examine_package.py
index e87cc281e802e1dbeb554e4a51972f793bd2d358..ba2855d4aacdc3f9f567aca92b5c8d4f17e9fe0e 100755 (executable)
--- a/dak/examine_package.py
+++ b/dak/examine_package.py
@@ -32,7 +32,7 @@
 
 ################################################################################
 
-import errno, os, pg, re, sys
+import errno, os, pg, re, sys, md5
 import apt_pkg, apt_inst
 import daklib.database, daklib.utils
 
@@ -82,6 +82,8 @@ Cnf = daklib.utils.get_conf()
 projectB = pg.connect(Cnf["DB::Name"], Cnf["DB::Host"], int(Cnf["DB::Port"]))
 daklib.database.init(Cnf, projectB)
 
+printed_copyrights = {}
+
 ################################################################################
 
 def usage (exit_code=0):
@@ -302,7 +304,15 @@ def print_copyright (deb_filename):
         return
 
     o = os.popen("ar p %s data.tar.gz | tar xzOf - %s" % (deb_filename, copyright))
-    print o.read()
+    copyright = o.read()
+    copyrightmd5 = md5.md5(copyright).hexdigest()
+
+    if printed_copyrights.has_key(copyrightmd5):
+        print "Copyright is the same as %s.\n" % \
+               (printed_copyrights[copyrightmd5])
+    else:
+       print copyright
+       printed_copyrights[copyrightmd5] = "%s (%s)" % (package, deb_filename)
 
 def check_dsc (dsc_filename):
     print "---- .dsc file for %s ----" % (dsc_filename)