Make upload_to_google_storage.py and tests work with Python 3.

Call decode() and encode() appropriately to manipulate bytes and strings.

Bug: 984182, 1009819
Change-Id: I8a73788ab912d6c6682012308954e4ab4df5fa98
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/tools/depot_tools/+/1859998
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>
Auto-Submit: Raphael Kubo da Costa <raphael.kubo.da.costa@intel.com>
Reviewed-by: Edward Lesmes <ehmaldonado@chromium.org>

tests/upload_to_google_storage_unittest.py

@@ -71,7 +71,7 @@ class UploadTests(unittest.TestCase):
            '%s/%s' % (self.base_url, self.lorem_ipsum_sha1)))])
     self.assertTrue(os.path.exists(output_filename))
     self.assertEqual(
-        open(output_filename, 'rb').read(),
+        open(output_filename, 'rb').read().decode(),
         '7871c8e24da15bad8b0be2c36edc9dc77e37727f')
     os.remove(output_filename)
     self.assertEqual(code, 0)
@@ -103,9 +103,9 @@ class UploadTests(unittest.TestCase):
   def test_upload_single_file_remote_exists(self):
     filenames = [self.lorem_ipsum]
     output_filename = '%s.sha1'  % self.lorem_ipsum
-    etag_string = 'ETag: 634d7c1ed3545383837428f031840a1e'
-    self.gsutil.add_expected(0, '', '')
-    self.gsutil.add_expected(0, etag_string, '')
+    etag_string = b'ETag: 634d7c1ed3545383837428f031840a1e'
+    self.gsutil.add_expected(0, b'', b'')
+    self.gsutil.add_expected(0, etag_string, b'')
     code = upload_to_google_storage.upload_to_google_storage(
         filenames, self.base_url, self.gsutil, False, False, 1, False, None)
     self.assertEqual(
@@ -116,7 +116,7 @@ class UploadTests(unittest.TestCase):
           ('ls', '-L', '%s/%s' % (self.base_url, self.lorem_ipsum_sha1)))])
     self.assertTrue(os.path.exists(output_filename))
     self.assertEqual(
-        open(output_filename, 'rb').read(),
+        open(output_filename, 'rb').read().decode(),
         '7871c8e24da15bad8b0be2c36edc9dc77e37727f')
     os.remove(output_filename)
     self.assertEqual(code, 0)
@@ -150,7 +150,7 @@ class UploadTests(unittest.TestCase):
     output_filename = '%s.sha1' % self.lorem_ipsum
     fake_hash = '6871c8e24da15bad8b0be2c36edc9dc77e37727f'
     with open(output_filename, 'wb') as f:
-      f.write(fake_hash)  # Fake hash.
+      f.write(fake_hash.encode())  # Fake hash.
     code = upload_to_google_storage.upload_to_google_storage(
         filenames, self.base_url, self.gsutil, False, False, 1, True, None)
     self.assertEqual(
@@ -162,7 +162,7 @@ class UploadTests(unittest.TestCase):
          ('check_call',
           ('cp', filenames[0], '%s/%s' % (self.base_url, fake_hash)))])
     self.assertEqual(
-        open(output_filename, 'rb').read(), fake_hash)
+        open(output_filename, 'rb').read().decode(), fake_hash)
     os.remove(output_filename)
     self.assertEqual(code, 0)
 

upload_to_google_storage.py

@@ -63,13 +63,13 @@ def get_md5_cached(filename):
   # See if we can find an existing MD5 sum stored in a file.
   if os.path.exists('%s.md5' % filename):
     with open('%s.md5' % filename, 'rb') as f:
-      md5_match = re.search('([a-z0-9]{32})', f.read())
+      md5_match = re.search('([a-z0-9]{32})', f.read().decode())
       if md5_match:
         return md5_match.group(1)
   else:
     md5_hash = get_md5(filename)
     with open('%s.md5' % filename, 'wb') as f:
-      f.write(md5_hash)
+      f.write(md5_hash.encode())
     return md5_hash
 
 
@@ -84,7 +84,7 @@ def _upload_worker(
     if gsutil.check_call('ls', file_url)[0] == 0 and not force:
       # File exists, check MD5 hash.
       _, out, _ = gsutil.check_call_with_retries('ls', '-L', file_url)
-      etag_match = re.search(r'ETag:\s+([a-z0-9]{32})', out)
+      etag_match = re.search(r'ETag:\s+([a-z0-9]{32})', out.decode())
       if etag_match:
         remote_md5 = etag_match.group(1)
         # Calculate the MD5 checksum to match it to Google Storage's ETag.
@@ -176,15 +176,15 @@ def upload_to_google_storage(
           'Main> Found hash for %s, sha1 calculation skipped.' % filename)
       with open(filename + '.sha1', 'rb') as f:
         sha1_file = f.read(1024)
-      if not re.match('^([a-z0-9]{40})$', sha1_file):
+      if not re.match('^([a-z0-9]{40})$', sha1_file.decode()):
         print('Invalid sha1 hash file %s.sha1' % filename, file=sys.stderr)
         return 1
-      upload_queue.put((filename, sha1_file))
+      upload_queue.put((filename, sha1_file.decode()))
       continue
     stdout_queue.put('Main> Calculating hash for %s...' % filename)
     sha1_sum = get_sha1(filename)
     with open(filename + '.sha1', 'wb') as f:
-      f.write(sha1_sum)
+      f.write(sha1_sum.encode())
     stdout_queue.put('Main> Done calculating hash for %s.' % filename)
     upload_queue.put((filename, sha1_sum))
   hashing_duration = time.time() - hashing_start