Make the untar API more intuitive

As discussed in #112.

2 files changed, 58 insertions, 24 deletions

diff --git a/scripts/support.py b/scripts/support.py
index 95d663fe0..7cb0e0963 100755
--- a/scripts/support.py
+++ b/scripts/support.py
@@ -22,29 +22,64 @@ import sys
 import tempfile
 
 
+def _open_archive(tarfile, tmp_dir):
+  with tempfile.TemporaryFile(mode='w+') as f:
+    try:
+      subprocess.check_call(['tar', '-xvf', tarfile], cwd=tmp_dir, stdout=f)
+    except:
+      f.seek(0)
+      sys.stderr.write(f.read())
+      raise
+  return os.listdir(tmp_dir)
+
+
+def _files_same(dir1, dir2, basenames):
+  diff = filecmp.cmpfiles(dir1, dir2, basenames)
+  return 0 == len(diff[1] + diff[2])
+
+
+def _dirs_same(dir1, dir2, basenames):
+  for d in basenames:
+    diff = filecmp.dircmp(os.path.join(dir1, d), os.path.join(dir2, d))
+    if 0 != len(diff.left_only + diff.right_only + diff.diff_files +
+                diff.common_funny + diff.funny_files):
+      return False
+  return True
+
+
+def _move_files(dirfrom, dirto, basenames):
+  for f in basenames:
+    from_file = os.path.join(dirfrom, f)
+    to_file = os.path.join(dirto, f)
+    if os.path.isfile(to_file):
+      os.path.remove(to_file)
+    shutil.move(from_file, to_file)
+
+
+def _move_dirs(dirfrom, dirto, basenames):
+    for d in basenames:
+      from_dir = os.path.join(dirfrom, d)
+      to_dir = os.path.join(dirto, d)
+      if os.path.isdir(to_dir):
+        shutil.rmtree(to_dir)
+      shutil.move(from_dir, to_dir)
+
+
 def untar(tarfile, outdir):
-  """Returns True if the untar dir differs from a pre-existing dir."""
-  tmp_dir = tempfile.mkdtemp()
+  """Returns True if untar content differs from pre-existing outdir content."""
+  tmpdir = tempfile.mkdtemp()
   try:
-    with tempfile.TemporaryFile(mode='w+') as f:
-      try:
-        subprocess.check_call(['tar', '-xvf', tarfile], cwd=tmp_dir, stdout=f)
-      except:
-        f.seek(0)
-        sys.stderr.write(f.read())
-        raise
-    untar_outdir = os.path.join(tmp_dir, os.path.basename(outdir))
-    if os.path.exists(outdir):
-      diff = filecmp.dircmp(untar_outdir, outdir)
-      if not (diff.left_only + diff.right_only + diff.diff_files +
-              diff.common_funny + diff.funny_files):
-        # outdir already existed with exactly the same content.
-        return False
-      shutil.rmtree(outdir)
-    # The untar files are different, or there was no previous outdir.
-    print 'Updating', outdir
-    shutil.move(untar_outdir, outdir)
+    untared = _open_archive(tarfile, tmpdir)
+    files = [f for f in untared if os.path.isfile(os.path.join(tmpdir, f))]
+    dirs = [d for d in untared if os.path.isdir(os.path.join(tmpdir, d))]
+    assert len(files) + len(dirs) == len(untared), 'Only files and directories'
+    if _files_same(tmpdir, outdir, files) and _dirs_same(tmpdir, outdir, dirs):
+      # Nothing new or different in the tarfile.
+      return False
+    # Some or all of the files / directories are new.
+    _move_files(tmpdir, outdir, files)
+    _move_dirs(tmpdir, outdir, dirs)
     return True
   finally:
-    if os.path.isdir(tmp_dir):
-      shutil.rmtree(tmp_dir)
+    if os.path.isdir(tmpdir):
+      shutil.rmtree(tmpdir)
diff --git a/update.py b/update.py
index d224c74b2..c0d14411c 100755
--- a/update.py
+++ b/update.py
@@ -25,7 +25,6 @@ import scripts.support
 BASE_DIR = os.path.abspath('test')
 REVISION_PATH = os.path.join(BASE_DIR, 'revision')
 TORTURE_TAR = 'wasm-torture-s-%s.tbz2'
-TORTURE_DIR = os.path.join(BASE_DIR, 'torture-s')
 
 
 def write_revision(revision):
@@ -43,7 +42,7 @@ def run(force_latest, override_hash):
   revision = (override_hash if override_hash else
               scripts.storage.download_revision(force_latest=force_latest))
   downloaded = scripts.storage.download_tar(TORTURE_TAR, BASE_DIR, revision)
-  updates += scripts.support.untar(downloaded, TORTURE_DIR)
+  updates += scripts.support.untar(downloaded, BASE_DIR)
   if updates:
     # Only update revision if the files it downloaded are different.
     print 'Updating revision to', revision