From 99eeba1f44b8452e23457f77b86134972672bdf8 Mon Sep 17 00:00:00 2001
From: Sam Thursfield <sam.thursfield@codethink.co.uk>
Date: Mon, 11 Apr 2016 12:10:11 +0100
Subject: migrations/008: Make it work for chunks whose ref is not a commit
 SHA1

Reported here: <https://storyboard.baserock.org/#!/story/93>

Change-Id: I08aa2f2f205f805661dfd2043eb100baaf2eb72b
---
 migrations/008-submodules-in-strata.py | 51 ++++++++++++++++++++++++++++++----
 1 file changed, 45 insertions(+), 6 deletions(-)

diff --git a/migrations/008-submodules-in-strata.py b/migrations/008-submodules-in-strata.py
index 33404f2..73e57e0 100755
--- a/migrations/008-submodules-in-strata.py
+++ b/migrations/008-submodules-in-strata.py
@@ -81,14 +81,47 @@ def get_repo_name(repo):
 
 ## End of code based on ybd repos.py 
 
-def get_toplevel_file_list_from_repo(url, ref):
+
+# This function is taken from morphlib/git.py.
+def is_valid_sha1(ref):
+    '''Checks whether a string is a valid SHA1.'''
+    return len(ref) == 40 and all(x in string.hexdigits for x in ref)
+
+
+def resolve_ref_to_commit(url, ref):
+    '''Resolve ref to a SHA1, using the remote morph-cache server.
+
+    This is only useful for repos hosted on a Baserock Trove, but it does work
+    without needing to clone the whole repo.
+
+    '''
+    response = requests.get(
+        GIT_CACHE_SERVER_URL + '1.0/sha1s',
+        params={'repo': url, 'ref': ref},
+        headers={'Accept': 'application/json'},
+        timeout=9)
+    logging.debug("Got response: %s" % response)
+    try:
+        response.raise_for_status()
+        commit = response.json()['sha1']
+    except Exception as e:
+        raise RuntimeError(
+            "Unexpected response from server %s for repo %s: %s" %
+            (GIT_CACHE_SERVER_URL, url, e.message))
+    return commit
+
+
+def get_toplevel_file_list_from_repo(url, commit):
     '''Try to list the set of files in the root directory of the repo at 'url'.
 
+    This is only useful for repos hosted on a Baserock Trove, but it does work
+    without needing to clone the whole repo.
+
     '''
     try:
         response = requests.get(
             GIT_CACHE_SERVER_URL + '1.0/trees',
-            params={'repo': url, 'ref': ref},
+            params={'repo': url, 'ref': commit},
             headers={'Accept': 'application/json'},
             timeout=9)
         logging.debug("Got response: %s" % response)
@@ -185,12 +218,18 @@ def add_submodules_to_strata(contents, filename):
         if 'submodules' in chunk_ref:
             continue
         try:
+            if is_valid_sha1(chunk_git_ref):
+                chunk_git_commit = chunk_git_ref
+            else:
+                chunk_git_commit = resolve_ref_to_commit (chunk_git_url, chunk_git_ref)
+
             toplevel_file_list = get_toplevel_file_list_from_repo(
-                chunk_git_url, chunk_git_ref)
+                chunk_git_url, chunk_git_commit)
         except Exception as e:
+            logging.debug(e)
             message = (
-                "Unable to look up repo %s on remote Git server %s. Check that "
-                "the repo URL is correct." % (chunk_git_url, TROVE_HOST))
+                "Unable to look up repo %s on remote Git server %s. Check "
+                "that the repo URL is correct." % (chunk_git_url, TROVE_HOST))
             warning = (
                 "If you are using a Trove that is not %s, please edit the "
                 "TROVE_HOST constant in this script and run it again." %
@@ -206,7 +245,7 @@ def add_submodules_to_strata(contents, filename):
             "%s: got file list %s", chunk_git_url, toplevel_file_list)
 
         if u'.gitmodules' in toplevel_file_list:
-            submodules = submodules_to_dict(chunk_git_url, chunk_git_ref)
+            submodules = submodules_to_dict(chunk_git_url, chunk_git_commit)
             if submodules:
                 chunk_ref['submodules'] = submodules
                 changed = True
-- 
cgit v1.2.1