lorry.zip-importer: Use Unix-format mtimes if available

The Zip file format is extensible, and there are several extensions to
file metadata that provide a Unix-format mtime.  These allow us to
avoid the local time ambiguity altogether.

1 files changed, 44 insertions, 6 deletions

diff --git a/lorry.zip-importer b/lorry.zip-importer
index 3b1ddd6..dfb73eb 100755
--- a/lorry.zip-importer
+++ b/lorry.zip-importer
@@ -12,6 +12,7 @@
 
 import calendar
 import os.path
+import struct
 import subprocess
 import sys
 import time
@@ -24,12 +25,52 @@ committer_name = 'Lorry Zip Importer'
 committer_email = 'lorry-zip-importer@lorry'
 
 
+# File header 'extra' field tags
+EXT_TAG_UNIX0 = 0x000d   # PKWARE Unix, aka Unix type 0
+EXT_TAG_TIME = 0x5455    # Extended Timestamp
+EXT_TIME_FLAG_MTIME = 1  # mtime present (and first)
+EXT_TAG_UNIX1 = 0x5855   # Info-ZIP Unix type 1
+
+
+# Iterate over fields within a file header 'extra' block
+def zip_extra_fields(extra):
+    pos = 0
+    while len(extra) >= pos + 4:
+        tag, size = struct.unpack('<HH', extra[pos : pos + 4])
+        pos += 4
+        if len(extra) < pos + size:
+            return
+        yield tag, extra[pos : pos + size]
+        pos += size
+
+
+# Make our best guess at the mtime of a zip file entry
+def zip_info_mtime(info):
+    # Look for Unix-format mtime in the 'extra' block
+    for tag, data in zip_extra_fields(info.extra):
+        format = None
+        if tag in [EXT_TAG_UNIX0, EXT_TAG_UNIX1]:
+            format = '<4xL'  # AcTime, ModTime
+        elif tag == EXT_TAG_TIME:
+            # First byte indicates which timestamps follow
+            if len(data) >= 1 and data[0] & EXT_TIME_FLAG_MTIME:
+                format = '<xL'  # Flags, ModTime
+        if format:
+            min_len = struct.calcsize(format)
+            if len(data) >= min_len:
+                return struct.unpack(format, data[:min_len])[0]
+
+    # Timestamps in the main header are in local time, but the time
+    # zone offset is unspecified.  We choose to interpret them as UTC.
+    return calendar.timegm(info.date_time + (0, 0, 0))
+
+
 def export(zipfile, fast_import):
     def printlines(list):
         for str in list:
             fast_import.write(str.encode('utf-8') + b"\n")
 
-    commit_time = (1970, 1, 1, 0, 0, 0)
+    commit_time = 0
     next_mark = 1
     common_prefix = None
     mark = dict()
@@ -40,8 +81,7 @@ def export(zipfile, fast_import):
             continue
         info = zip.getinfo(name)
 
-        if commit_time < info.date_time:
-            commit_time = info.date_time
+        commit_time = max(commit_time, zip_info_mtime(info))
         if common_prefix == None:
             common_prefix = name[:name.rfind('/') + 1]
         else:
@@ -56,10 +96,8 @@ def export(zipfile, fast_import):
                     'data ' + str(info.file_size)))
         fast_import.write(zip.read(name) + b"\n")
 
-    # Zip file timestamps are in local time, but the time zone offset
-    # is unspecified.  We choose to interpret them as UTC.
     committer = committer_name + ' <' + committer_email + '> %d +0000' % \
-        calendar.timegm(commit_time + (0, 0, 0))
+        commit_time
 
     zipfile_basename = os.path.basename(zipfile)
     printlines(('commit ' + branch_ref, 'committer ' + committer, \