github-lambda: fix directory index and allow to override content-type

this patch allows users to use files like `index.xml` as directory
index, by renaming it to `index.html`, because Amazon S3 only allows one
file name as directory index. it also allows users to override the
automatically guessed content-type with something else, e.g. to force a
charset.

1 files changed, 34 insertions, 9 deletions

diff --git a/src/blogc-github-lambda/lambda_function.py b/src/blogc-github-lambda/lambda_function.py
index e456397..5a3bf5e 100644
--- a/src/blogc-github-lambda/lambda_function.py
+++ b/src/blogc-github-lambda/lambda_function.py
@@ -60,7 +60,23 @@ def get_tarball(repo_name):
     return rootdir
 
 
-def sync_s3(src, dest):
+def translate_filename(filename):
+    f = filename.split('/')
+    if len(f) == 0:
+        return filename
+    basename = f[-1]
+
+    # replace any index.$EXT file with index.html, because s3 only allows
+    # users to declare one directory index file name.
+    p = basename.split('.')
+    if len(p) == 2 and p[0] == 'index':
+        f[-1] = 'index.html'
+        return '/'.join(f)
+
+    return filename
+
+
+def sync_s3(src, dest, settings_file):
     s3 = boto3.resource('s3')
     bucket = s3.Bucket(dest)
 
@@ -69,35 +85,43 @@ def sync_s3(src, dest):
         if not obj.key.endswith('/'):
             remote_files[obj.key] = obj
 
-    local_files = []
+    local_files = {}
     for root, dirs, files in os.walk(src):
         real_root = root[len(src):].lstrip('/')
         for file in files:
-            local_files.append(os.path.join(real_root, file))
+            f = os.path.join(real_root, file)
+            local_files[translate_filename(f)] = f
 
     to_upload = []
     for file in local_files:
         if file not in remote_files:
-            to_upload.append(file)
+            to_upload.append(local_files[file])
 
     to_delete = []
     for file in remote_files:
         if file in local_files:
-            with open(os.path.join(src, file)) as fp:
+            with open(os.path.join(src, local_files[file])) as fp:
                 l = hashlib.sha1(fp.read())
 
             with closing(remote_files[file].get()['Body']) as fp:
                 r = hashlib.sha1(fp.read())
 
             if l.hexdigest() != r.hexdigest():
-                to_upload.append(file)
+                to_upload.append(local_files[file])
         else:
             to_delete.append(file)
 
+    content_types = {}
+    if os.path.exists(settings_file):
+        with open(settings_file, 'r') as fp:
+            settings = json.load(fp)
+            content_types = settings.get('content-type', {})
+
     for file in to_upload:
         with open(os.path.join(src, file), 'rb') as fp:
-            print 'Uploading file:', file
-            mime, _ = mimetypes.guess_type(file)
+            mime = content_types.get(file, mimetypes.guess_type(file)[0])
+            file = translate_filename(file)
+            print 'Uploading file: %s; content-type: "%s"' % (file, mime)
             if mime is not None:
                 bucket.put_object(Key=file, Body=fp, ContentType=mime)
             else:
@@ -124,4 +148,5 @@ def lambda_handler(event, context):
                              stderr=None if debug else subprocess.PIPE)
         if rv != 0:
             raise RuntimeError('Failed to run make')
-        sync_s3(os.path.join(rootdir, '_build'), repo_name)
+        sync_s3(os.path.join(rootdir, '_build'), repo_name,
+                os.path.join(rootdir, 's3.json'))