From 1964df62df9acf92a45ff28abd95276344769411 Mon Sep 17 00:00:00 2001
From: Hammad Akhtar <hammad13060@iiitd.ac.in>
Date: Tue, 13 Dec 2016 11:51:53 +0530
Subject: [PATCH] Bug 1323633 - Ensure that task graphs do not contain
 duplicate Treeherder symbols, also consider treeherder.collection.keys();
 r=dustin

MozReview-Commit-ID: 3F0BHQQmSOg

--HG--
extra : rebase_source : 52ba5d93d689e5ca64cee04dd013dcd79919049d
---
 taskcluster/taskgraph/generator.py            |  3 ++-
 taskcluster/taskgraph/taskgraph.py            |  5 ++++
 .../util/{verifydoc.py => verify.py}          | 27 +++++++++++++++++++
 3 files changed, 34 insertions(+), 1 deletion(-)
 rename taskcluster/taskgraph/util/{verifydoc.py => verify.py} (61%)

diff --git a/taskcluster/taskgraph/generator.py b/taskcluster/taskgraph/generator.py
index dcd35523ecdd..926b007592af 100644
--- a/taskcluster/taskgraph/generator.py
+++ b/taskcluster/taskgraph/generator.py
@@ -12,7 +12,7 @@ from .graph import Graph
 from .taskgraph import TaskGraph
 from .optimize import optimize_task_graph
 from .util.python_path import find_object
-from .util.verifydoc import verify_docs
+from .util.verify import verify_docs, verify_task_graph_symbol
 
 logger = logging.getLogger(__name__)
 
@@ -217,6 +217,7 @@ class TaskGraphGenerator(object):
         target_task_graph = TaskGraph(
             {l: all_tasks[l] for l in target_graph.nodes},
             target_graph)
+        target_task_graph.for_each_task(verify_task_graph_symbol, scratch_pad={})
         yield 'target_task_graph', target_task_graph
 
         logger.info("Generating optimized task graph")
diff --git a/taskcluster/taskgraph/taskgraph.py b/taskcluster/taskgraph/taskgraph.py
index 7736745ef50d..606ea209868a 100644
--- a/taskcluster/taskgraph/taskgraph.py
+++ b/taskcluster/taskgraph/taskgraph.py
@@ -46,6 +46,11 @@ class TaskGraph(object):
             tasks[key] = task_json
         return tasks
 
+    def for_each_task(self, f, *args, **kwargs):
+        for task_label in self.graph.visit_postorder():
+            task = self.tasks[task_label]
+            f(task, self, *args, **kwargs)
+
     def __getitem__(self, label):
         "Get a task by label"
         return self.tasks[label]
diff --git a/taskcluster/taskgraph/util/verifydoc.py b/taskcluster/taskgraph/util/verify.py
similarity index 61%
rename from taskcluster/taskgraph/util/verifydoc.py
rename to taskcluster/taskgraph/util/verify.py
index 17ca13eeb199..8810968c7d1d 100644
--- a/taskcluster/taskgraph/util/verifydoc.py
+++ b/taskcluster/taskgraph/util/verify.py
@@ -38,3 +38,30 @@ def verify_docs(filename, identifiers, appearing_as):
                     "{}: `{}` missing from doc file: `{}`"
                     .format(appearing_as, identifier, filename)
                 )
+
+
+def verify_task_graph_symbol(task, taskgraph, scratch_pad):
+    """
+        This function verifies that tuple
+        (collection.keys(), machine.platform, groupSymbol, symbol) is unique
+        for a target task graph.
+    """
+    task_dict = task.task
+    if "extra" in task_dict:
+        extra = task_dict["extra"]
+        if "treeherder" in extra:
+            treeherder = extra["treeherder"]
+
+            collection_keys = tuple(sorted(treeherder.get('collection', {}).keys()))
+            platform = treeherder.get('machine', {}).get('platform')
+            group_symbol = treeherder.get('groupSymbol')
+            symbol = treeherder.get('symbol')
+
+            key = (collection_keys, platform, group_symbol, symbol)
+            if key in scratch_pad:
+                raise Exception(
+                    "conflict between `{}`:`{}` for values `{}`"
+                    .format(task.label, scratch_pad[key], key)
+                )
+            else:
+                scratch_pad[key] = task.label