Rev 16: Simplify output_lines. in http://people.ubuntu.com/~robertc/baz2.0/plugins/groupcompress/trunk

[Robert Collins]

At http://people.ubuntu.com/~robertc/baz2.0/plugins/groupcompress/trunk

------------------------------------------------------------
revno: 16
revision-id: robertc at robertcollins.net-20080724023950-rl13uu60j66wks61
parent: robertc at robertcollins.net-20080724015151-vul2s7bru8fqytjl
committer: Robert Collins <robertc at robertcollins.net>
branch nick: trunk
timestamp: Thu 2008-07-24 12:39:50 +1000
message:
  Simplify output_lines.
modified:
  groupcompress.py               groupcompress.py-20080705181503-ccbxd6xuy1bdnrpu-8
=== modified file 'groupcompress.py'
--- a/groupcompress.py	2008-07-24 01:51:51 +0000
+++ b/groupcompress.py	2008-07-24 02:39:50 +0000
@@ -137,7 +137,6 @@
         line_locations = self.line_locations
         range_len = 0
         range_start = 0
-        flush_range = self.flush_range
         copy_ends = None
         # We either copy a range (while there are reusable lines) or we 
         # insert new lines. To find reusable lines we traverse 
@@ -152,7 +151,7 @@
                 else:
                     range_len += 1
             else:
-                locations, next = line_locations[line]
+                locations = line_locations[line]
                 if copy_ends:
                     next_locations = locations.intersection(copy_ends)
                     if len(next_locations):
@@ -164,7 +163,7 @@
                 if copy_ends:
                     result.append((min(copy_ends) - range_len, range_start, range_len))
                 range_len = 1
-                copy_ends = next
+                copy_ends = set(loc + 1 for loc in locations)
                 range_start = pos
             pos += 1
         if copy_ends:
@@ -197,7 +196,6 @@
         new_lines.append('sha1: %s\n' % sha1)
         index_lines = [False, False]
         pos = 0
-        line_locations = self.line_locations
         range_len = 0
         range_start = 0
         flush_range = self.flush_range
@@ -209,13 +207,13 @@
         for old_start, new_start, range_len in blocks:
             if new_start != current_pos:
                 # non-matching region
-                flush_range(False, current_pos, None, new_start - current_pos,
+                flush_range(current_pos, None, new_start - current_pos,
                     lines, new_lines, index_lines)
             current_pos = new_start + range_len
             if not range_len:
                 continue
-            flush_range(True, new_start, [old_start + range_len], range_len,
-                lines, new_lines, index_lines)
+            flush_range(new_start, old_start, range_len, lines,
+                new_lines, index_lines)
         delta_start = (self.endpoint, len(self.lines))
         self.output_lines(new_lines, index_lines)
         trim_encoding_newline(lines)
@@ -240,13 +238,12 @@
         sha1 = sha_strings(lines)
         return lines, sha1
 
-    def flush_range(self, copying, range_start, copy_ends, range_len, lines, new_lines, index_lines):
+    def flush_range(self, range_start, copy_start, range_len, lines, new_lines, index_lines):
         if not range_len:
             return
         insert_instruction = "i,%d\n" % range_len
-        if copying:
+        if copy_start is not None:
             # range stops, flush and start a new copy range
-            copy_start = min(copy_ends) - range_len
             stop_byte = self.line_offsets[copy_start + range_len - 1]
             if copy_start == 0:
                 start_byte = 0
@@ -263,7 +260,7 @@
         new_lines.append(insert_instruction)
         new_lines.extend(lines[range_start:range_start+range_len])
         index_lines.append(False)
-        index_lines.extend([not copying]*range_len)
+        index_lines.extend([copy_start is None]*range_len)
 
     def output_lines(self, new_lines, index_lines):
         """Output some lines.
@@ -274,15 +271,15 @@
         """
         endpoint = self.endpoint
         offset = len(self.lines)
+        line_append = self.line_offsets.append
+        setdefault = self.line_locations.setdefault
         for (pos, line), index in izip(enumerate(new_lines), index_lines):
-            self.lines.append(line)
             endpoint += len(line)
-            self.line_offsets.append(endpoint)
+            line_append(endpoint)
             if index:
-                indices, next_lines = self.line_locations.setdefault(line,
-                    (set(), set()))
+                indices = setdefault(line, set())
                 indices.add(pos + offset)
-                next_lines.add(pos + offset + 1)
+        self.lines.extend(new_lines)
         self.endpoint = endpoint
 
     def ratio(self):