optimized revisions lookups

tests/conftest.py

@@ -232,15 +232,20 @@ def result_set_stored(jm, initial_data, sample_resultset):
 
 @pytest.fixture(scope='function')
 def mock_get_resultset(monkeypatch, result_set_stored):
-
     from treeherder.etl import common
 
-    def _get_resultset(project, revision):
-        return {
+    def _get_resultset(params):
+        for k in params:
+            rev = params[k][0]
+            params[k] = {
+                rev: {
                     'id': 1,
                     'revision_hash': result_set_stored[0]['revision_hash']
                 }
-    monkeypatch.setattr(common, 'get_resultset', _get_resultset)
+            }
+        return params
+
+    monkeypatch.setattr(common, 'lookup_revisions', _get_resultset)
 
 @pytest.fixture()
 def refdata():

tests/etl/test_common.py

@@ -3,14 +3,13 @@ from django.conf import settings
 
 
 def test_get_revision_hash(jm, initial_data,
-                           result_set_stored, mock_get_resultset):
+                           result_set_stored, mock_get_remote_content):
     """That the correct revision_hash is retrieved is the revision exists"""
     from treeherder.etl import common
-    project = settings.DATABASES["default"]["TEST_NAME"]
+    project = result_set_stored[0]['revisions'][0]['repository']
     revision = result_set_stored[0]['revisions'][0]['revision']
-    resultset = common.get_resultset(project, revision)
-
-    assert resultset['revision_hash'] == result_set_stored[0]['revision_hash']
+    resultset = common.lookup_revisions({project: [revision]})
+    assert resultset[project][revision]['revision_hash'] == result_set_stored[0]['revision_hash']
 
 
 def test_get_revision_hash_none(jm, mock_get_remote_content,
@@ -19,5 +18,5 @@ def test_get_revision_hash_none(jm, mock_get_remote_content,
     from treeherder.etl import common
     project = settings.DATABASES["default"]["TEST_NAME"]
     revision = "fakerevision"
-    resultset = common.get_resultset(project, revision)
-    assert resultset == None
+    resultset = common.lookup_revisions({project: [revision]})
+    assert len(resultset) == 0

tests/model/derived/test_jobs_model.py

@@ -55,7 +55,7 @@ def test_ingest_all_sample_jobs(jm, refdata, sample_data, initial_data, sample_r
     job_data = sample_data.job_data
     test_utils.do_job_ingestion(jm, refdata, job_data, sample_resultset)
 
-def test_bad_date_value_ingestion(jm, initial_data):
+def test_bad_date_value_ingestion(jm, initial_data, mock_log_parser):
     """
     Test ingesting an blob with bad date value
 

tests/model/derived/test_objectstore_model.py

@@ -110,7 +110,7 @@ def test_process_objects_unknown_error(jm):
     assert row_id == 0
     assert response == exp_resp
 
-def test_ingest_sample_data(jm, sample_data, sample_resultset):
+def test_ingest_sample_data(jm, sample_data, sample_resultset, mock_log_parser):
     """Process all job structures in the job_data.txt file"""
 
     resultset_count = len(sample_resultset)

treeherder/etl/buildapi.py

@@ -1,5 +1,5 @@
 import logging
-
+from collections import defaultdict
 from django.conf import settings
 
 from . import buildbot
@@ -44,19 +44,27 @@ class Builds4hTransformerMixin(object):
         our restful api
         """
         job_list = []
+        revisions = defaultdict(list)
+        for build in data['builds']:
+            prop = build['properties']
+            prop['revision'] = prop.get('revision',
+                            prop.get('got_revision',
+                                prop.get('sourcestamp', None)))
+
+            if not prop['revision']:
+                continue
+            revisions[prop['branch']].append(prop['revision'][0:12])
+
+        revisions_lookup = common.lookup_revisions(revisions)
+
         for build in data['builds']:
 
             prop = build['properties']
-            revision = prop.get('revision',
-                            prop.get('got_revision',
-                                prop.get('sourcestamp', None)))
-            if not revision:
-                continue
 
-            try:
-                resultset = common.get_resultset(prop['branch'], revision[0:12])
-            except Exception:
-                resultset = None
+            if not prop['revision'] in revisions_lookup[prop['branch']]:
+                continue
+            # the resultset object is in the nested dict
+            resultset = revisions_lookup[prop['branch']][prop['revision']]
             if not resultset:
                 continue
 
@@ -133,14 +141,29 @@ class PendingTransformerMixin(object):
         job_list = []
 
         projects = set(x.project for x in Datasource.objects.cached())
+        revision_dict = defaultdict(list)
+
+        # loop to catch all the revisions
         for project, revisions in data['pending'].items():
+            # this skips those projects we don't care about
             if not project in projects:
                 continue
+
             for rev, jobs in revisions.items():
-                resultset = common.get_resultset(project, rev)
-                if not resultset:
-                    continue
-                for job in jobs:
+                revision_dict[project].append(rev)
+
+        # retrieving the revision->resultset lookups
+        revisions_lookup = common.lookup_revisions(revision_dict)
+
+        for project, revisions in revisions_lookup.items():
+
+            for revision in revisions:
+
+                resultset = revisions[revision]
+                # using project and revision form the revision lookups
+                # to filter those jobs with unmatched revision
+                for job in data['pending'][project][revision]:
+
                     treeherder_data = {
                         'revision_hash': resultset['revision_hash'],
                         'resultset_id': resultset['id'],
@@ -190,15 +213,28 @@ class RunningTransformerMixin(object):
         """
         job_list = []
         projects = set(x.project for x in Datasource.objects.cached())
+        revisions = defaultdict(list)
+
+        # loop to catch all the revisions
         for project, revisions in data['running'].items():
+            # this skips those projects we don't care about
             if not project in projects:
                 continue
-            for rev, jobs in revisions.items():
-                resultset = common.get_resultset(project, rev)
-                if not resultset:
-                    continue
 
-                for job in jobs:
+            for rev, jobs in revisions.items():
+                revisions[project].append(rev)
+
+        # retrieving the revision->resultset lookups
+        revisions_lookup = common.lookup_revisions(revisions)
+
+        for project, revisions in revisions_lookup.items():
+
+            for revision in revisions:
+
+                resultset = revisions[revision]
+                # using project and revision form the revision lookups
+                # to filter those jobs with unmatched revision
+                for job in data['running'][project][revision]:
                     treeherder_data = {
                         'revision_hash': resultset['revision_hash'],
                         'resultset_id': resultset['id'],

treeherder/etl/common.py

@@ -1,3 +1,4 @@
+from collections import defaultdict
 import hashlib
 import urllib2
 import simplejson as json
@@ -76,28 +77,25 @@ def get_remote_content(url):
     return content
 
 
-
-def get_resultset(project, revision):
-    """Retrieve a revision hash given a single revision"""
-
-    cache_key = "{0}:{1}".format(project, revision)
-
-    cached_resultset = cache.get(cache_key)
-
-    if not cached_resultset:
-        endpoint = reverse('resultset-list', kwargs={"project": project})
+def lookup_revisions(revision_dict):
+    """
+    Retrieve a list of revision->resultset lookups
+    """
+    lookup = dict()
+    for project, revisions in revision_dict.items():
+        revision_set = set(revisions)
+        endpoint = reverse('revision-lookup-list', kwargs={"project": project})
+        # build the query string as a comma separated list of revisions
+        q = ','.join(revision_set)
         url = "{0}/{1}/?revision={2}".format(
             settings.API_HOSTNAME.strip('/'),
             endpoint.strip('/'),
-            revision
+            q
         )
         content = get_remote_content(url)
         if content:
-            cached_resultset = content[0]
-            cache.set(cache_key, cached_resultset, 60 * 60 * 12)
-        else:
-            return None
-    return cached_resultset
+            lookup[project] = get_remote_content(url)
+    return lookup
 
 
 def generate_revision_hash(revisions):

treeherder/etl/pulse.py

@@ -357,9 +357,12 @@ class TreeherderPulseDataAdapter(PulseDataAdapter, ObjectstoreLoaderMixin):
     def adapt_data(self, data):
         """Adapts the PulseDataAdapter into the treeherder input data structure"""
 
-        resultset = common.get_resultset(data['branch'], data['revision'])
+        resultset = common.lookup_revisions({data['branch']: [data['revision']]})
+        if not resultset:
+            return {}
 
-        treeherder_data = resultset or {}
+        del resultset[data['branch']][data['revision']]['id']
+        treeherder_data = resultset[data['branch']][data['revision']]
         treeherder_data['project'] = data['branch']
         ####
         #TODO: This is a temporary fix, this data will not be located
@@ -378,7 +381,7 @@ class TreeherderPulseDataAdapter(PulseDataAdapter, ObjectstoreLoaderMixin):
                 #modified
                 request_id, data['request_times'][unicode(request_id)]
             ),
-            'revision_hash': resultset['revision_hash'],
+            'revision_hash': treeherder_data.pop('revision_hash'),
             'name': data['test_name'],
             'product_name': data['product'],
             'state': 'completed',

treeherder/model/derived/jobs.py

@@ -300,6 +300,26 @@ class JobsModel(TreeherderModelBase):
 
         return self.as_list(return_list, "result_set", **kwargs)
 
+    def get_revision_resultset_lookup(self, revision_list):
+        """
+        Create a list of revision->resultset lookups from a list of revision
+        """
+
+        replacement = ",".join(["%s"] * len(revision_list))
+        replacement = " AND revision IN ("+replacement+") "
+
+        proc = "jobs.selects.get_result_set_list"
+        lookups = self.get_jobs_dhub().execute(
+            proc=proc,
+            placeholders=revision_list+[0, len(revision_list)],
+            debug_show=self.DEBUG,
+            replace=[replacement],
+            return_type="dict",
+            key_column="revision"
+        )
+        return lookups
+
+
     def get_result_set_details(self, result_set_ids):
         """
         Retrieve all revisions associated with a set of ``result_set`` (also known as ``pushes``)

treeherder/model/sql/jobs.json

@@ -239,7 +239,9 @@
             "sql":"SELECT DISTINCT
                       rs.id,
                       rs.revision_hash,
-                      rs.push_timestamp
+                      rs.push_timestamp,
+                      revision.id as revision_id,
+                      revision.revision
                    FROM result_set AS rs
                    INNER JOIN revision_map
                       ON rs.id = revision_map.result_set_id

treeherder/webapp/api/urls.py

@@ -34,6 +34,13 @@ project_bound_router.register(
     base_name='note',
 )
 
+project_bound_router.register(
+    r'revision-lookup',
+    views.RevisionLookupSetViewSet,
+    base_name='revision-lookup',
+)
+
+
 # this is the default router for plain restful endpoints
 
 # refdata endpoints:

treeherder/webapp/api/views.py

@@ -6,8 +6,9 @@ from rest_framework import viewsets
 from rest_framework.response import Response
 from rest_framework.decorators import action
 from rest_framework.reverse import reverse
-from treeherder.model import models
+from rest_framework.exceptions import ParseError
 
+from treeherder.model import models
 from treeherder.model.derived import (JobsModel, DatasetNotFoundError,
                                       ObjectNotFoundException)
 
@@ -364,6 +365,22 @@ class ResultSetViewSet(viewsets.ViewSet):
         return Response({"message": "well-formed JSON stored"})
 
 
+
+class RevisionLookupSetViewSet(viewsets.ViewSet):
+
+    @with_jobs
+    def list(self, request, project, jm):
+
+        revision_filter = request.QUERY_PARAMS.get('revision', None)
+        if not revision_filter:
+            raise ParseError(detail="The revision parameter is mandatory for this endpoint")
+
+        revision_list = revision_filter.split(",")
+
+        return Response(jm.get_revision_resultset_lookup(revision_list))
+
+
+
 #####################
 # Refdata ViewSets
 #####################