Better patch parser; support svn log --diff

Makefile

 .PHONY: dogfood test clean build dist-test dist
 
 dogfood:
-	./cdiff.py -s
-	git diff | ./cdiff.py
+	./cdiff.py
 	git diff | ./cdiff.py -s
 
 test:

cdiff

 import sys
 import cdiff
 
-try:
-    sys.exit(cdiff.main())
-except:
-    sys.exit(1)
+sys.exit(cdiff.main())
 
 # vim:set et sts=4 sw=4 tw=80:

cdiff.py

 
 class Hunk(object):
 
-    def __init__(self, hunk_header, old_addr, new_addr):
-        self._hunk_header = hunk_header
+    def __init__(self, hunk_headers, hunk_meta, old_addr, new_addr):
+        self._hunk_headers = hunk_headers
+        self._hunk_meta = hunk_meta
         self._old_addr = old_addr   # tuple (start, offset)
         self._new_addr = new_addr   # tuple (start, offset)
         self._hunk_list = []        # list of tuple (attr, line)
 
-    def get_header(self):
-        return self._hunk_header
+    def get_hunk_headers(self):
+        return self._hunk_headers
+
+    def get_hunk_meta(self):
+        return self._hunk_meta
 
     def get_old_addr(self):
         return self._old_addr
         self._new_path = new_path
         self._hunks = hunks
 
-    # Follow detector and the parse_hunk_header() are suppose to be overwritten
-    # by derived class
+    # Follow detector and the parse_hunk_meta() are suppose to be overwritten
+    # by derived class.  No is_header() anymore, all non-recognized lines are
+    # considered as headers
     #
     def is_old_path(self, line):
         return False
     def is_new_path(self, line):
         return False
 
-    def is_hunk_header(self, line):
+    def is_hunk_meta(self, line):
         return False
 
-    def parse_hunk_header(self, line):
+    def parse_hunk_meta(self, line):
         """Returns a 2-eliment tuple, each of them is a tuple in form of (start,
         offset)"""
         return False
     def is_eof(self, line):
         return False
 
-    def is_header(self, line):
-        return False
-
     def markup_traditional(self):
         """Returns a generator"""
         for line in self._headers:
         yield self._markup_new_path(self._new_path)
 
         for hunk in self._hunks:
-            yield self._markup_hunk_header(hunk.get_header())
+            for hunk_header in hunk.get_hunk_headers():
+                yield self._markup_hunk_header(hunk_header)
+            yield self._markup_hunk_meta(hunk.get_hunk_meta())
             for old, new, changed in hunk.mdiff():
                 if changed:
                     if not old[0]:
 
         # yield hunks
         for hunk in self._hunks:
-            yield self._markup_hunk_header(hunk.get_header())
+            for hunk_header in hunk.get_hunk_headers():
+                yield self._markup_hunk_header(hunk_header)
+            yield self._markup_hunk_meta(hunk.get_hunk_meta())
             for old, new, changed in hunk.mdiff():
                 if old[0]:
                     left_num = str(hunk.get_old_addr()[0] + int(old[0]) - 1)
         return colorize(line, 'yellow')
 
     def _markup_hunk_header(self, line):
+        return colorize(line, 'lightcyan')
+
+    def _markup_hunk_meta(self, line):
         return colorize(line, 'lightblue')
 
     def _markup_common(self, line):
     def is_new_path(self, line):
         return line.startswith('+++ ')
 
-    def is_hunk_header(self, line):
-        return line.startswith('@@ -')
+    def is_hunk_meta(self, line):
+        return line.startswith('@@ -') or line.startswith('## -')
 
-    def parse_hunk_header(self, hunk_header):
+    def parse_hunk_meta(self, hunk_meta):
         # @@ -3,7 +3,6 @@
-        a = hunk_header.split()[1].split(',')   # -3 7
+        a = hunk_meta.split()[1].split(',')   # -3 7
         if len(a) > 1:
             old_addr = (int(a[0][1:]), int(a[1]))
         else:
             # @@ -1 +1,2 @@
             old_addr = (int(a[0][1:]), 0)
 
-        b = hunk_header.split()[2].split(',')   # +3 6
+        b = hunk_meta.split()[2].split(',')   # +3 6
         if len(b) > 1:
             new_addr = (int(b[0][1:]), int(b[1]))
         else:
         return (old_addr, new_addr)
 
     def is_old(self, line):
-        return line.startswith('-') and not self.is_old_path(line)
+        """Exclude header line from svn log --diff output"""
+        return line.startswith('-') and not self.is_old_path(line) and \
+                not re.match(r'^-{4,}$', line.rstrip())
 
     def is_new(self, line):
         return line.startswith('+') and not self.is_new_path(line)
 
     def is_eof(self, line):
         # \ No newline at end of file
-        return line.startswith('\\')
-
-    def is_header(self, line):
-        return re.match(r'^[^+@\\ -]', line)
+        # \ No newline at end of property
+        return line.startswith(r'\ No newline at end of')
 
 
 class DiffParser(object):
 
     def __init__(self, stream):
-        """Detect Udiff with 3 conditions"""
+        """Detect Udiff with 3 conditions, '## ' uaually indicates svn property
+        changes in output from `svn log --diff`
+        """
         flag = 0
-        for line in stream[:20]:
+        for line in stream[:100]:
             if line.startswith('--- '):
                 flag |= 1
             elif line.startswith('+++ '):
                 flag |= 2
-            elif line.startswith('@@ '):
+            elif line.startswith('@@ ') or line.startswith('## '):
                 flag |= 4
-        if flag & 7:
-            self._type = 'udiff'
+            if flag & 7:
+                self._type = 'udiff'
+                break
         else:
             raise RuntimeError('unknown diff type')
 
 
         out_diffs = []
         headers = []
-        old_path = None
-        new_path = None
-        hunks = []
-        hunk = None
 
         while stream:
-            # 'common' line occurs before 'old_path' is considered as header
-            # too, this happens with `git log -p` and `git show <commit>`
-            #
-            if difflet.is_header(stream[0]) or \
-                    (difflet.is_common(stream[0]) and old_path is None):
-                if headers and old_path:
-                    # Encounter a new header
-                    assert new_path is not None
-                    assert hunk is not None
-                    hunks.append(hunk)
-                    out_diffs.append(Diff(headers, old_path, new_path, hunks))
-                    headers = []
-                    old_path = None
-                    new_path = None
-                    hunks = []
-                    hunk = None
-                else:
-                    headers.append(stream.pop(0))
-
-            elif difflet.is_old_path(stream[0]):
-                if old_path:
-                    # Encounter a new patch set
-                    assert new_path is not None
-                    assert hunk is not None
-                    hunks.append(hunk)
-                    out_diffs.append(Diff(headers, old_path, new_path, hunks))
-                    headers = []
-                    old_path = None
-                    new_path = None
-                    hunks = []
-                    hunk = None
-                else:
-                    old_path = stream.pop(0)
+            if difflet.is_old_path(stream[0]):
+                old_path = stream.pop(0)
+                out_diffs.append(Diff(headers, old_path, None, []))
+                headers = []
 
             elif difflet.is_new_path(stream[0]):
-                assert old_path is not None
-                assert new_path is None
                 new_path = stream.pop(0)
-
-            elif difflet.is_hunk_header(stream[0]):
-                assert old_path is not None
-                assert new_path is not None
-                if hunk:
-                    # Encounter a new hunk header
-                    hunks.append(hunk)
-                    hunk = None
-                else:
-                    hunk_header = stream.pop(0)
-                    old_addr, new_addr = difflet.parse_hunk_header(hunk_header)
-                    hunk = Hunk(hunk_header, old_addr, new_addr)
-
-            elif difflet.is_old(stream[0]) or difflet.is_new(stream[0]) or \
-                    difflet.is_common(stream[0]):
-                assert old_path is not None
-                assert new_path is not None
-                assert hunk is not None
+                out_diffs[-1]._new_path = new_path
+
+            elif difflet.is_hunk_meta(stream[0]):
+                hunk_meta = stream.pop(0)
+                old_addr, new_addr = difflet.parse_hunk_meta(hunk_meta)
+                hunk = Hunk(headers, hunk_meta, old_addr, new_addr)
+                headers = []
+                out_diffs[-1]._hunks.append(hunk)
+
+            elif out_diffs and out_diffs[-1]._hunks and \
+                    (difflet.is_old(stream[0]) or difflet.is_new(stream[0]) or \
+                    difflet.is_common(stream[0])):
                 hunk_line = stream.pop(0)
-                hunk.append(hunk_line[0], hunk_line[1:])
+                out_diffs[-1]._hunks[-1].append(hunk_line[0], hunk_line[1:])
 
             elif difflet.is_eof(stream[0]):
                 # ignore
                 stream.pop(0)
 
             else:
-                raise RuntimeError('unknown patch format: %s' % stream[0])
-
-        # The last patch
-        if hunk:
-            hunks.append(hunk)
-        if old_path:
-            if new_path:
-                out_diffs.append(Diff(headers, old_path, new_path, hunks))
-            else:
-                raise RuntimeError('unknown patch format after "%s"' % old_path)
-        elif headers:
-            raise RuntimeError('unknown patch format: %s' % \
-                    ('\n'.join(headers)))
+                # All other non-recognized lines are considered as headers or
+                # hunk headers respectively
+                #
+                headers.append(stream.pop(0))
+
+        if headers:
+            raise RuntimeError('dangling header(s):\n%s' % ''.join(headers))
+
+        # Validate the last patch set
+        if out_diffs:
+            assert out_diffs[-1]._old_path is not None
+            assert out_diffs[-1]._new_path is not None
+            assert len(out_diffs[-1]._hunks) > 0
+            assert len(out_diffs[-1]._hunks[-1]._hunk_meta) > 0
 
         return out_diffs
 

tests/README

-# To generate expected output, use following command, then review with `less -R`
+# To generate expected output, chdir to a subdir and use following command, then
+# review with `less -R`
+#
 ../../cdiff.py -c always in.diff > out.normal 
 ../../cdiff.py -c always -s in.diff > out.side-by-side
 ../../cdiff.py -c always -s in.diff -w70 > out.w70

tests/regression.sh

     local cdiff_opt=${3:-""}
 
     echo -n "Test option '$cdiff_opt' with input '$input' ... "
-    if $CDIFF $cdiff_opt $input | diff -ubq $expected_out - >& /dev/null; then
+    if $CDIFF $cdiff_opt $input 2>/dev/null \
+            | diff -ubq $expected_out - >& /dev/null; then
         pass
         return 0
     else

tests/svn-log/in.diff

+------------------------------------------------------------------------
+r1235 | ymattw | 2011-09-01 17:00:02 +0800 (Thu, 01 Sep 2011) | 3 lines
+
+Interpreter fix
+
+
+Index: src/share/example.sh
+===================================================================
+--- src/share/example.sh	(revision 1234)
++++ src/share/example.sh	(revision 1235)
+@@ -1,3 +1,3 @@
+-#!/bin/sh
++#!/bin/bash
+ #
+ # Test program, also try run me as yroot (sudo ./example.sh)
+
+------------------------------------------------------------------------
+r1234 | ymattw | 2011-09-01 16:00:02 +0800 (Thu, 01 Sep 2011) | 3 lines
+
+Implement common tool to run command as headless account
+
+
+Index: ChangeLog
+===================================================================
+--- ChangeLog	(revision 1233)
++++ ChangeLog	(revision 1234)
+@@ -1,3 +1,6 @@
++Version 0.0.4
++    * Add prototype of perl module
++
+ Version 0.0.3
+     * Implement '-d' option
+     * Add configfile to set global debug flag
+Index: src/share/example.sh
+===================================================================
+--- src/share/example.sh	(revision 0)
++++ src/share/example.sh	(revision 1234)
+@@ -0,0 +1,4 @@
++#!/bin/sh
++#
++# Test program, also try run me as yroot (sudo ./example.sh)
++#
+
+Property changes on: src/share/example.sh
+___________________________________________________________________
+Added: svn:executable
+## -0,0 +1 ##
++*
+\ No newline at end of property
+Added: svn:keywords
+## -0,0 +1 ##
++Id
+\ No newline at end of property

tests/svn-log/out.normal

+------------------------------------------------------------------------
+r1235 | ymattw | 2011-09-01 17:00:02 +0800 (Thu, 01 Sep 2011) | 3 lines
+
+Interpreter fix
+
+
+Index: src/share/example.sh
+===================================================================
+--- src/share/example.sh	(revision 1234)
++++ src/share/example.sh	(revision 1235)
+@@ -1,3 +1,3 @@
+-#!/bin/sh
++#!/bin/bash
+ #
+ # Test program, also try run me as yroot (sudo ./example.sh)
+
+------------------------------------------------------------------------
+r1234 | ymattw | 2011-09-01 16:00:02 +0800 (Thu, 01 Sep 2011) | 3 lines
+
+Implement common tool to run command as headless account
+
+
+Index: ChangeLog
+===================================================================
+--- ChangeLog	(revision 1233)
++++ ChangeLog	(revision 1234)
+@@ -1,3 +1,6 @@
++Version 0.0.4
++    * Add prototype of perl module
++
+ Version 0.0.3
+     * Implement '-d' option
+     * Add configfile to set global debug flag
+Index: src/share/example.sh
+===================================================================
+--- src/share/example.sh	(revision 0)
++++ src/share/example.sh	(revision 1234)
+@@ -0,0 +1,4 @@
++#!/bin/sh
++#
++# Test program, also try run me as yroot (sudo ./example.sh)
++#
+
+Property changes on: src/share/example.sh
+___________________________________________________________________
+Added: svn:executable
+## -0,0 +1 ##
++*
+Added: svn:keywords
+## -0,0 +1 ##
++Id
+

tests/svn-log/out.side-by-side

+------------------------------------------------------------------------
+r1235 | ymattw | 2011-09-01 17:00:02 +0800 (Thu, 01 Sep 2011) | 3 lines
+
+Interpreter fix
+
+
+Index: src/share/example.sh
+===================================================================
+--- src/share/example.sh	(revision 1234)
++++ src/share/example.sh	(revision 1235)
+@@ -1,3 +1,3 @@
+1 #!/bin/sh                                                                        1 #!/bin/bash
+2 #                                                                                2 #
+3 # Test program, also try run me as yroot (sudo ./example.sh)                     3 # Test program, also try run me as yroot (sudo ./example.sh)
+
+------------------------------------------------------------------------
+r1234 | ymattw | 2011-09-01 16:00:02 +0800 (Thu, 01 Sep 2011) | 3 lines
+
+Implement common tool to run command as headless account
+
+
+Index: ChangeLog
+===================================================================
+--- ChangeLog	(revision 1233)
++++ ChangeLog	(revision 1234)
+@@ -1,3 +1,6 @@
+                                                                                   1 Version 0.0.4
+                                                                                   2     * Add prototype of perl module
+                                                                                   3 
+1 Version 0.0.3                                                                    4 Version 0.0.3
+2     * Implement '-d' option                                                      5     * Implement '-d' option
+3     * Add configfile to set global debug flag                                    6     * Add configfile to set global debug flag
+Index: src/share/example.sh
+===================================================================
+--- src/share/example.sh	(revision 0)
++++ src/share/example.sh	(revision 1234)
+@@ -0,0 +1,4 @@
+                                                                                     1 #!/bin/sh
+                                                                                     2 #
+                                                                                     3 # Test program, also try run me as yroot (sudo ./example.sh)
+                                                                                     4 #
+
+Property changes on: src/share/example.sh
+___________________________________________________________________
+Added: svn:executable
+## -0,0 +1 ##
+                                                                                     1 *
+Added: svn:keywords
+## -0,0 +1 ##
+                                                                                     1 Id

tests/svn-log/out.w70

+------------------------------------------------------------------------
+r1235 | ymattw | 2011-09-01 17:00:02 +0800 (Thu, 01 Sep 2011) | 3 lines
+
+Interpreter fix
+
+
+Index: src/share/example.sh
+===================================================================
+--- src/share/example.sh	(revision 1234)
++++ src/share/example.sh	(revision 1235)
+@@ -1,3 +1,3 @@
+1 #!/bin/sh                                                              1 #!/bin/bash
+2 #                                                                      2 #
+3 # Test program, also try run me as yroot (sudo ./example.sh)           3 # Test program, also try run me as yroot (sudo ./example.sh)
+
+------------------------------------------------------------------------
+r1234 | ymattw | 2011-09-01 16:00:02 +0800 (Thu, 01 Sep 2011) | 3 lines
+
+Implement common tool to run command as headless account
+
+
+Index: ChangeLog
+===================================================================
+--- ChangeLog	(revision 1233)
++++ ChangeLog	(revision 1234)
+@@ -1,3 +1,6 @@
+                                                                         1 Version 0.0.4
+                                                                         2     * Add prototype of perl module
+                                                                         3 
+1 Version 0.0.3                                                          4 Version 0.0.3
+2     * Implement '-d' option                                            5     * Implement '-d' option
+3     * Add configfile to set global debug flag                          6     * Add configfile to set global debug flag
+Index: src/share/example.sh
+===================================================================
+--- src/share/example.sh	(revision 0)
++++ src/share/example.sh	(revision 1234)
+@@ -0,0 +1,4 @@
+                                                                           1 #!/bin/sh
+                                                                           2 #
+                                                                           3 # Test program, also try run me as yroot (sudo ./example.sh)
+                                                                           4 #
+
+Property changes on: src/share/example.sh
+___________________________________________________________________
+Added: svn:executable
+## -0,0 +1 ##
+                                                                           1 *
+Added: svn:keywords
+## -0,0 +1 ##
+                                                                           1 Id

tests/test_cdiff.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+"""Unit test for cdiff"""
+
+import sys
+if sys.hexversion < 0x02050000:
+    raise SystemExit("*** Requires python >= 2.5.0")
+
+sys.path.insert(0, '..')
+
+import unittest
+import cdiff
+
+class TestCdiff(unittest.TestCase):
+
+    def test_foo(self):
+        self.assertTrue(1)
+
+if __name__ == '__main__':
+    unittest.main()
+
+# vim:set et sts=4 sw=4 tw=80: