Added ghost patches support.

The ghost options disable the actual loading in memory of the patches.
When you just want to know the number of patches or their stats,
this saves quite a lot of memory and CPU time.

3 files changed, 59 insertions, 26 deletions

diff --git a/patchtracker/DebTarHandler.py b/patchtracker/DebTarHandler.py
index 311e39b..3564af2 100644
--- a/patchtracker/DebTarHandler.py
+++ b/patchtracker/DebTarHandler.py
@@ -10,8 +10,8 @@ class DebTarHandler:
     self.tarfile = fname
     self.size = os.stat(fname)[stat.ST_SIZE]
 
-  def series(self):
-    return Patch.Quilt30PatchSeries(self.tarfile)
+  def series(self, ghost=False):
+    return Patch.Quilt30PatchSeries(self.tarfile, ghost=ghost)
   
 if __name__ == "__main__":
   print "DebTarHandler testing"
diff --git a/patchtracker/DiffGzHandler.py b/patchtracker/DiffGzHandler.py
index 515b3ed..769ce18 100644
--- a/patchtracker/DiffGzHandler.py
+++ b/patchtracker/DiffGzHandler.py
@@ -14,7 +14,7 @@ class DiffGzHandler:
     self.diff = fname
     self.size = os.stat(fname)[stat.ST_SIZE]
 
-  def filterdiff(self, include=None, exclude=None):
+  def filterdiff(self, include=None, exclude=None, ghost=False):
     cmd = ["filterdiff","-z","-p","1"]
     if include:
       cmd += [ "-i", include]
@@ -24,7 +24,8 @@ class DiffGzHandler:
       raise Exception("DiffGzHandler.filterdiff called w/o include/exclude")
     i,o,e=os.popen3(cmd+[self.diff])
     i.close()
-    p = Patch(o)
+    p = Patch(o, ghost=ghost)
+    o.close()
     err = e.read()
     if len(err):
       raise DiffGzException("filterdiff gave errors: "+err)
@@ -33,10 +34,10 @@ class DiffGzHandler:
   def debiandir(self):
     return self.filterdiff(include='debian/*')
 
-  def nondebiandir(self):
-    return self.filterdiff(exclude='debian/*')
+  def nondebiandir(self, ghost=False):
+    return self.filterdiff(exclude='debian/*', ghost=ghost)
 
-  def series(self):
+  def series(self, ghost=False):
     patches = None
     embedded = self.filterdiff(include='debian/patches*')
 
@@ -50,7 +51,7 @@ class DiffGzHandler:
       err = e.read()
       if len(err):
         raise Exception("unable to extract series patches:\n"+err)
-      patches = PatchSeries(td)
+      patches = PatchSeries(td, ghost=ghost)
       os.system("rm -rf %s"%(td))
 
     return patches
diff --git a/patchtracker/Patch.py b/patchtracker/Patch.py
index 6533910..80ad8dc 100644
--- a/patchtracker/Patch.py
+++ b/patchtracker/Patch.py
@@ -3,21 +3,43 @@ import os
 import errno
 from glob import glob
 import tarfile
+from subprocess import Popen, PIPE
+from shutil import copyfileobj
 
 class Diffstat:
-  def __init__(self, patch):
+  def __init__(self, patch, ghost=False, patch_fh=None):
     self.patch = patch
-    i,o = os.popen2("diffstat -p1")
-    i.write(str(patch))
-    i.close()
-    self.output = o.readlines()
+    self.ghost = ghost
+    if not ghost:
+      i,o = os.popen2("diffstat -p1")
+      i.write(str(patch))
+      i.close()
+      self.output = o.readlines()
+    else:
+      popen = Popen("diffstat -p1 -t", shell=True, close_fds=True,
+        stdin=PIPE, stdout=PIPE, stderr=PIPE)
+      popen.stderr.close()
+      copyfileobj(patch_fh, popen.stdin)
+      popen.stdin.close()
+      popen.stdout.readline()
+      self._stats = map(lambda x: map(lambda y: y.strip(), x.split(',')), popen.stdout.readlines())
+      popen.wait()
 
   def stats(self):
-    i,o = os.popen2("diffstat -p1 -t")
-    i.write(str(self.patch))
-    i.close()
-    o.readline()
-    return [map(lambda x: x.strip(), l.split(",")) for l in o.readlines()]
+    if not self.ghost:
+      i,o = os.popen2("diffstat -p1 -t")
+      i.write(str(self.patch))
+      i.close()
+      o.readline()
+      return [map(lambda x: x.strip(), l.split(",")) for l in o.readlines()]
+    else:
+      return self._stats
+
+  def lines(self):
+    added, removed, modified = 0, 0, 0
+    for [added2, removed2, modified2, name] in self.stats():
+      added, removed, modified = added + int(added2), removed + int(removed2), modified + int(modified2)
+    return (added, removed, modified)
 
   def summary(self):
     return self.output[-1]
@@ -26,8 +48,13 @@ class Diffstat:
     return "".join(self.output)
 
 class Patch:
-  def __init__(self, fh, level=1):
-    self.p = fh.readlines()
+  def __init__(self, fh, level=1, ghost=False):
+    self.ghost = ghost
+    if not ghost:
+      self.p = fh.readlines()
+    else:
+      self.p = []
+      self._diffstat = Diffstat(self, ghost=True, patch_fh=fh)
     self.lvl = level
 
   def __str__(self):
@@ -37,7 +64,10 @@ class Patch:
     return len(self.p)
 
   def diffstat(self):
-    return Diffstat(self)
+    if not self.ghost:
+      return Diffstat(self)
+    else:
+      return self._diffstat
 
 class GenericPatchSeries (list):
   def blank(self):
@@ -68,7 +98,8 @@ class GenericPatchSeries (list):
 
 # XXX this entire __init__ stuff is way to ugly
 class PatchSeries (GenericPatchSeries):
-  def __init__(self, dir):
+  def __init__(self, dir, ghost=False):
+    self.ghost = ghost
     fd = None
     self.blank()
     self.style = "simple"
@@ -118,11 +149,11 @@ class PatchSeries (GenericPatchSeries):
     removelater=[]
     for p in self.names:
       try:
-        self.patches[p] = Patch(file(os.sep.join([dir, p])))
+        self.patches[p] = Patch(file(os.sep.join([dir, p])), ghost=ghost)
       except IOError, e:
         if e.errno == errno.ENOENT and self.style == "dpatch":
           try:
-            self.patches[p] = Patch(file(os.sep.join([dir, p+".dpatch"])))
+            self.patches[p] = Patch(file(os.sep.join([dir, p+".dpatch"])), ghost=ghost)
           except:
             #print "ERROR: could not find patch",p
             self.blank()
@@ -138,8 +169,9 @@ class PatchSeries (GenericPatchSeries):
       self.names.remove(p)
 
 class Quilt30PatchSeries (GenericPatchSeries):
-  def __init__(self, tarBall):
+  def __init__(self, tarBall, ghost=False):
     self.blank()
+    self.ghost = ghost
     self.style = "quilt (3.0)"
     self.tarfh = tarfile.open(tarBall, 'r:*')
     try:
@@ -162,7 +194,7 @@ class Quilt30PatchSeries (GenericPatchSeries):
 
     # XXX to lazy eval this might be better
     for name in self.names:
-      self.patches[name] = Patch(self.tarfh.extractfile("debian/patches/"+name))
+      self.patches[name] = Patch(self.tarfh.extractfile("debian/patches/"+name), ghost=ghost)
 
 if __name__ == "__main__":
   print "Patch.py testing"