Added export_for_udd.py script.

2 files changed, 80 insertions, 0 deletions

diff --git a/export_for_udd.py b/export_for_udd.py
new file mode 100755
index 0000000..4220e2f
--- /dev/null
+++ b/export_for_udd.py
@@ -0,0 +1,71 @@
+#!/usr/bin/python
+
+from patchtracker.DB import PatchTrackerDB, PackageWithoutDiffException
+from patchtracker.Util import print_used_memory
+import sys
+import json
+
+def enumerate_packages(db):
+	for packages_per_letter in db.findCollection().pkgs.itervalues():
+		for package_name, package_data in packages_per_letter.iteritems():
+			for suite, package in package_data.iteritems():
+				yield (package_name, suite, package.version)
+
+def main():
+	db = PatchTrackerDB()
+	packages = []
+	try:
+		i = 0
+		for package, suite, version in enumerate_packages(db):
+
+			# Just for testing and debugging...
+			#if package not in ['geogebra', 'wotsap', 'netrw', 'haskell-devscripts',
+			#	'haskell-filestore', 'openide-utils']:
+			#	continue
+
+			print >> sys.stderr, "%d %s %s %s" % (i, package, suite, version)
+			i += 1
+
+			try:
+				dh = db.makeDiffHandler(package, version)
+				series = dh.series(ghost=True)
+				if series:
+					series_type = series.style
+				else:
+					series_type = "no_series"
+				try:
+					nondebian_diff = dh.nondebiandir(ghost=True)
+					nondebian = nondebian_diff.diffstat().lines()
+				except AttributeError:
+					# This means that the handler is a DebTarHandler
+					nondebian = [0, 0, 0]
+			except PackageWithoutDiffException:
+				# The package is native
+				series_type = "native"
+				nondebian = [0, 0, 0]
+
+			patches = [(patch, patch_data.diffstat().lines()) for (patch, patch_data) in series]
+
+			packages.append({'package': package, 'suite': suite, 'version': version,
+				'series_type': series_type, 'nondebian': nondebian,
+				'patches': patches})
+
+	# This is mainly intended for debugging: if you want to interrupt the
+	# process, you still can obtain the JSON with the packages processed so far.
+	# Consistency is assured by the fact that append() is atomic in Python
+	except KeyboardInterrupt:
+		pass
+
+    # TODO We don't need to keep all the packages dictionary in memory; there are
+    # libraries to access JSON in a stream (DOM-like) fashion, but Python
+    # bindings don't appear to be available in Debian (libyajl)
+
+	# Pretty printing:
+	#json.dump(packages, sys.stdout, sort_keys=True, indent=4)
+
+	# Awful printing:
+	json.dump(packages, sys.stdout)
+
+if __name__ == '__main__':
+	main()
+
diff --git a/patchtracker/Util.py b/patchtracker/Util.py
new file mode 100644
index 0000000..31a200c
--- /dev/null
+++ b/patchtracker/Util.py
@@ -0,0 +1,9 @@
+
+import resource
+import sys
+
+def print_used_memory(msg):
+	"""Small utility function to find where RAM is used at most."""
+	r = resource.getrusage(resource.RUSAGE_SELF)
+	print >> sys.stderr,  "%s; used memory = %d" % (msg, r.ru_maxrss)
+