[StGit PATCH] Read several objects at once with git cat-file --batch

["Karl Hasselström" <kha@treskal.com>]

Instead of spawning a separate cat-file process for every blob and
commit we want to read. This speeds things up slightly: about 6-8%
when uncommitting and rebasing 1470 linux-kernel patches (perftest.py
rebase-newrebase-add-file-linux).

Signed-off-by: Karl Hasselström <kha@treskal.com>

---

 stgit/lib/git.py |   34 +++++++++++++++++++++++++++++++++-
 stgit/run.py     |   17 +++++++++++++++++
 2 files changed, 50 insertions(+), 1 deletions(-)


diff --git a/stgit/lib/git.py b/stgit/lib/git.py
index 648e190..95efd9a 100644
--- a/stgit/lib/git.py
+++ b/stgit/lib/git.py
@@ -520,6 +520,37 @@ class RunWithEnvCwd(RunWithEnv):
         @param args: Command and argument vector"""
         return RunWithEnv.run(self, args, self.env_in_cwd)
 
+class CatFileProcess(object):
+    def __init__(self, repo):
+        self.__repo = repo
+        self.__proc = None
+    def __get_process(self):
+        if not self.__proc:
+            self.__proc = self.__repo.run(['git', 'cat-file', '--batch']
+                                          ).run_background()
+        return self.__proc
+    def cat_file(self, sha1):
+        p = self.__get_process()
+        p.stdin.write('%s\n' % sha1)
+        p.stdin.flush()
+
+        # Read until we have the entire status line.
+        s = ''
+        while not '\n' in s:
+            s += os.read(p.stdout.fileno(), 4096)
+        h, b = s.split('\n', 1)
+        if h == '%s missing' % sha1:
+            raise SomeException()
+        hash, type, length = h.split()
+        assert hash == sha1
+        length = int(length)
+
+        # Read until we have the entire object plus the trailing
+        # newline.
+        while len(b) < length + 1:
+            b += os.read(p.stdout.fileno(), 4096)
+        return type, b[:-1]
+
 class Repository(RunWithEnv):
     """Represents a git repository."""
     def __init__(self, directory):
@@ -531,6 +562,7 @@ class Repository(RunWithEnv):
         self.__default_index = None
         self.__default_worktree = None
         self.__default_iw = None
+        self.__catfile = CatFileProcess(self)
     env = property(lambda self: { 'GIT_DIR': self.__git_dir })
     @classmethod
     def default(cls):
@@ -580,7 +612,7 @@ class Repository(RunWithEnv):
     directory = property(lambda self: self.__git_dir)
     refs = property(lambda self: self.__refs)
     def cat_object(self, sha1):
-        return self.run(['git', 'cat-file', '-p', sha1]).raw_output()
+        return self.__catfile.cat_file(sha1)[1]
     def rev_parse(self, rev):
         try:
             return self.get_commit(self.run(
diff --git a/stgit/run.py b/stgit/run.py
index 7493ed3..ccca059 100644
--- a/stgit/run.py
+++ b/stgit/run.py
@@ -130,6 +130,19 @@ class Run:
             raise self.exc('%s failed: %s' % (self.__cmd[0], e))
         self.__log_end(self.exitcode)
         self.__check_exitcode()
+    def __run_background(self):
+        """Run in background."""
+        assert self.__indata == None
+        try:
+            p = subprocess.Popen(self.__cmd, env = self.__env, cwd = self.__cwd,
+                                 stdin = subprocess.PIPE,
+                                 stdout = subprocess.PIPE,
+                                 stderr = subprocess.PIPE)
+        except OSError, e:
+            raise self.exc('%s failed: %s' % (self.__cmd[0], e))
+        self.stdin = p.stdin
+        self.stdout = p.stdout
+        self.stderr = p.stderr
     def returns(self, retvals):
         self.__good_retvals = retvals
         return self
@@ -181,6 +194,10 @@ class Run:
     def run(self):
         """Just run, with no IO redirection."""
         self.__run_noio()
+    def run_background(self):
+        """Run as a background process."""
+        self.__run_background()
+        return self
     def xargs(self, xargs):
         """Just run, with no IO redirection. The extra arguments are
         appended to the command line a few at a time; the command is