Merge pull request #15 from jasonmacgowan/jasonmacgowan/metadata

fetch data for _InnerSourceMetadata

Author: zkoppert

crawler.py

@@ -2,6 +2,7 @@
 
 import json
 import os
+from base64 import b64decode
 from os.path import dirname, join
 
 import github3
@@ -21,17 +22,50 @@
     organization = os.getenv("ORGANIZATION")
 
     # Get all repos from organization
-    search_string = "org:" + organization + " topic:" + topic
+    search_string = "org:{} topic:{}".format(organization, topic)
     all_repos = gh.search_repositories(search_string)
     repo_list = []
 
     for repo in all_repos:
         if repo is not None:
-            # TODO: #7 For each resulting project add a key _InnerSourceMetadata
             print("{0}".format(repo.repository))
             full_repository = repo.repository.refresh()
-            # Add stuff here about innersource.json data before appending to list
-            repo_list.append(repo.as_dict())
+
+            innersource_repo = repo.as_dict()
+            innersource_repo["_InnerSourceMetadata"] = {}
+
+            # fetch innersource.json
+            try:
+                content = repo.repository.file_contents("/innersource.json").content
+                metadata = json.loads(b64decode(content))
+
+                innersource_repo["_InnerSourceMetadata"] = metadata
+            except github3.exceptions.NotFoundError:
+                # innersource.json not found in repository, but it's not required
+                pass
+
+            # fetch repository participation
+            participation = repo.repository.weekly_commit_count()
+            innersource_repo["_InnerSourceMetadata"]["participation"] = participation[
+                "all"
+            ]
+
+            # fetch contributing guidelines
+            try:
+                # if CONTRIBUTING.md exists in the repository, link to that instead of repo root
+                content = repo.repository.file_contents("/CONTRIBUTING.md").content
+                innersource_repo["_InnerSourceMetadata"][
+                    "guidelines"
+                ] = "CONTRIBUTING.md"
+            except github3.exceptions.NotFoundError:
+                # CONTRIBUTING.md not found in repository, but it's not required
+                pass
+
+            # fetch repository topics
+            topics = repo.repository.topics()
+            innersource_repo["_InnerSourceMetadata"]["topics"] = topics.names
+
+            repo_list.append(innersource_repo)
 
     # Write each repository to a repos.json file
     with open("repos.json", "w") as f: