lorry.{g,}zip-importer: Read and write in binary mode

gzip and zip files are not in text format, and the file contents
that we send to git may not be either.  So:

* Open gzip files in binary mode.
* Open the pipe to 'git fast-import' in binary mode.  This requires
  using subprocess.Popen instead of os.popen.
* Change printlines functions to encode text in UTF-8.
* Use printlines functions in the remaining places where we need
  to write text.

2 files changed, 24 insertions, 20 deletions

diff --git a/lorry.gzip-importer b/lorry.gzip-importer
index 8d482b7..d3d78b4 100755
--- a/lorry.gzip-importer
+++ b/lorry.gzip-importer
@@ -3,9 +3,9 @@
 # gzip archive frontend for git-fast-import
 
 from gzip import GzipFile
-import os
 import os.path
 import struct
+import subprocess
 import sys
 
 
@@ -17,7 +17,7 @@ committer_email = 'lorry-gzip-importer@lorry'
 
 # The size of a gzip file is stored in the last 4 bytes
 def uncompressedsize(filename):
-    with open(filename) as f:
+    with open(filename, 'rb') as f:
         f.seek(-4, 2)
         return struct.unpack('I', f.read(4))[0]
 
@@ -25,7 +25,7 @@ def uncompressedsize(filename):
 def export(zipfile, fast_import):
     def printlines(list):
         for str in list:
-            fast_import.write(str + "\n")
+            fast_import.write(str.encode('utf-8') + b"\n")
 
     # Gzip does have an encoded mtime, however Python's GzipFile
     # just ignores it, so we just yank the mtime of the zip file itself.
@@ -34,7 +34,7 @@ def export(zipfile, fast_import):
     zip = GzipFile(zipfile, 'rb')
 
     printlines(('blob', 'mark :1', 'data ' + str(file_size)))
-    fast_import.write(zip.read() + "\n")
+    fast_import.write(zip.read() + b"\n")
 
     committer = committer_name + ' <' + committer_email + '> %d +0000' % \
         mtime
@@ -46,7 +46,7 @@ def export(zipfile, fast_import):
 
     last_dot = zipfile_basename[:-1].rfind('.')
     unzipped_file = zipfile_basename[:last_dot]
-    fast_import.write('M 100644 :1 ' + unzipped_file + '\n');
+    printlines(('M 100644 :1 ' + unzipped_file,));
 
     zipname, _ = os.path.splitext(zipfile_basename)
     printlines(('',  'tag ' + zipname, \
@@ -59,11 +59,13 @@ def main():
         print('usage:', sys.argv[0], '<gzipfile>...')
         sys.exit(1)
 
-    fast_import = os.popen('git fast-import --quiet', 'w')
-    for zipfile in sys.argv[1:]:
-        export(zipfile, fast_import)
-    if fast_import.close():
-        sys.exit(1)
+    with subprocess.Popen('git fast-import --quiet', shell=True,
+                          stdin=subprocess.PIPE) as import_proc:
+        for zipfile in sys.argv[1:]:
+            export(zipfile, import_proc.stdin)
+        import_proc.stdin.close()
+        if import_proc.wait() != 0:
+            sys.exit(1)
 
 
 main()
diff --git a/lorry.zip-importer b/lorry.zip-importer
index 7eca8f6..10b3b2b 100755
--- a/lorry.zip-importer
+++ b/lorry.zip-importer
@@ -10,8 +10,8 @@
 ##  python import-zips.py *.zip
 ##  git log --stat import-zips
 
-import os
 import os.path
+import subprocess
 import sys
 import time
 from zipfile import ZipFile
@@ -26,7 +26,7 @@ committer_email = 'lorry-zip-importer@lorry'
 def export(zipfile, fast_import):
     def printlines(list):
         for str in list:
-            fast_import.write(str + "\n")
+            fast_import.write(str.encode('utf-8') + b"\n")
 
     commit_time = 0
     next_mark = 1
@@ -53,7 +53,7 @@ def export(zipfile, fast_import):
 
         printlines(('blob', 'mark ' + mark[name], \
                     'data ' + str(info.file_size)))
-        fast_import.write(zip.read(name) + "\n")
+        fast_import.write(zip.read(name) + b"\n")
 
     committer = committer_name + ' <' + committer_email + '> %d +0000' % \
         time.mktime(commit_time + (0, 0, 0))
@@ -64,8 +64,8 @@ def export(zipfile, fast_import):
         '', 'deleteall'))
 
     for name in mark.keys():
-        fast_import.write('M 100644 ' + mark[name] + ' ' +
-            name[len(common_prefix):] + "\n")
+        printlines(('M 100644 ' + mark[name] + ' ' +
+                    name[len(common_prefix):],))
 
     zipname, _ = os.path.splitext(zipfile_basename)
 
@@ -79,11 +79,13 @@ def main():
         print('usage:', sys.argv[0], '<zipfile>...')
         sys.exit(1)
 
-    fast_import = os.popen('git fast-import --quiet', 'w')
-    for zipfile in sys.argv[1:]:
-        export(zipfile, fast_import)
-    if fast_import.close():
-        sys.exit(1)
+    with subprocess.Popen('git fast-import --quiet', shell=True,
+                          stdin=subprocess.PIPE) as import_proc:
+        for zipfile in sys.argv[1:]:
+            export(zipfile, import_proc.stdin)
+        import_proc.stdin.close()
+        if import_proc.wait() != 0:
+            sys.exit(1)
 
 
 main()