Merge lp:~brian-murray/errors/newbuckets into lp:errors

[Brian Murray]

I've added an API method for checking SourceVersionBuckets to see if there are new buckets in a specific version of a package compared to a previous version.

Jockey will return something:

http://10.0.3.182/api/1.0/package-version-new-buckets/?format=json&package=jockey&previous_version=0.9.7-0ubuntu7.7&new_version=0.9.7-0ubuntu7.8

[Evan (ev)]

Are we missing some back-populated data?

These return the empty set:

http://poppy-dev.local/api/1.0/package-version-new-buckets/?format=json&package=ubiquity&previous_version=2.10.24&new_version=2.14.2

http://poppy-dev.local/api/1.0/package-version-new-buckets/?format=json&package=nautilus&previous_version=1:3.4.1-0ubuntu1&new_version=1:3.6.3-0ubuntu14

But these work:

http://poppy-dev.local/api/1.0/package-version-new-buckets/?format=json&package=apport&previous_version=2.6.1-0ubuntu3&new_version=2.9.2-0ubuntu5

http://poppy-dev.local/api/1.0/package-version-new-buckets/?format=json&package=software-center&previous_version=5.4.1.4&new_version=5.6.0-0ubuntu2

http://poppy-dev.local/api/1.0/package-version-new-buckets/?format=json&package=update-manager&previous_version=1:0.174.3&new_version=1:0.185

[Evan (ev)]

Otherwise, this looks good!

Thanks

[Evan (ev)]

I've since noticed that the backfill RT didn't complete :)

=== modified file 'errors/api/resources.py'
--- errors/api/resources.py	2013-03-28 00:49:31 +0000
+++ errors/api/resources.py	2013-04-10 23:41:26 +0000
@@ -330,6 +330,23 @@
                 yield ResultObject(result)
         return wrapped()
 
+class PackageVersionNewBuckets(ErrorsResource):
+    function = fields.CharField(attribute='function', readonly=True)
+    class Meta(ErrorsMeta):
+        resource_name = 'package-version-new-buckets'
+
+    def obj_get_list(self, request):
+        src_package = request.GET.get('package', None)
+        previous_version = request.GET.get('previous_version', None)
+        new_version = request.GET.get('new_version', None)
+        class wrapped(list):
+            def __getslice__(self, start, finish):
+                results = cassandra.get_package_new_buckets(src_package,
+                            previous_version, new_version)
+                for bucket in results:
+                    yield ResultObject({'function': bucket })
+        return wrapped()
+
 class InstanceResource(ErrorsResource):
     instance = fields.DictField(attribute='instance', readonly=True)
     class Meta(ErrorsMeta):

=== modified file 'errors/api/urls.py'
--- errors/api/urls.py	2013-03-28 00:49:31 +0000
+++ errors/api/urls.py	2013-04-10 23:41:26 +0000
@@ -16,7 +16,8 @@
                         InstancesResource,
                         VersionsResource,
                         PackageVersionIsMostRecent,
-                        CrashSignaturesForBug)
+                        CrashSignaturesForBug,
+                        PackageVersionNewBuckets)
 
 v1_api = Api(api_name='1.0')
 v1_api.register(RetraceResultResource())
@@ -36,6 +37,7 @@
 v1_api.register(InstancesResource())
 v1_api.register(VersionsResource())
 v1_api.register(CrashSignaturesForBug())
+v1_api.register(PackageVersionNewBuckets())
 
 urlpatterns = patterns('',
     url(r'^', include(v1_api.urls)),

=== modified file 'errors/cassandra.py'
--- errors/cassandra.py	2013-04-06 17:45:55 +0000
+++ errors/cassandra.py	2013-04-10 23:41:26 +0000
@@ -467,6 +467,34 @@
         results['this_count'] = today_crashes
     return results
 
+def get_package_new_buckets(src_pkg, previous_version, new_version):
+    srcversionbuckets_cf = pycassa.ColumnFamily(pool, 'SourceVersionBuckets')
+    bucketsystems_cf = pycassa.ColumnFamily(pool, 'BucketSystems')
+    data = srcversionbuckets_cf.multiget([(src_pkg, previous_version),
+                (src_pkg, new_version)])
+    results = []
+    # new version has no buckets
+    try:
+        data[(src_pkg, new_version)]
+    except KeyError:
+        return results
+    # if previous version has no buckets return an empty list
+    try:
+        data[(src_pkg, previous_version)]
+    except KeyError:
+        data[(src_pkg, previous_version)] = []
+    new_buckets = set(data[src_pkg, new_version]).difference(set(data[src_pkg,
+                      previous_version]))
+    for bucket in new_buckets:
+        try:
+            count = len(bucketsystems_cf.get(bucket, column_count=4))
+        except NotFoundException:
+            continue
+        if count <= 3:
+            continue
+        results.append(bucket)
+    return results
+
 def record_bug_for_bucket(bucketid, bug):
     bucketmetadata_cf = pycassa.ColumnFamily(pool, 'BucketMetadata')
     bugtocrashsignatures_cf = pycassa.ColumnFamily(pool, 'BugToCrashSignatures')