Author: Armin Rigo <ar...@tunes.org>
Branch: unicode-utf8-re
Changeset: r93339:ebe0641d78a5
Date: 2017-12-09 20:44 +0100
http://bitbucket.org/pypy/pypy/changeset/ebe0641d78a5/
Log: Must not call ctx.next() when the type of ctx is not exactly known.
Workaround for interp_sre, where the few calls are not performance-
sensitive.
diff --git a/pypy/module/_sre/interp_sre.py b/pypy/module/_sre/interp_sre.py
--- a/pypy/module/_sre/interp_sre.py
+++ b/pypy/module/_sre/interp_sre.py
@@ -198,7 +198,7 @@
matchlist_w.append(w_item)
reset_at = ctx.match_end
if ctx.match_start == ctx.match_end:
- reset_at = ctx.next(reset_at)
+ reset_at = ctx.next_indirect(reset_at)
ctx.reset(reset_at)
return space.newlist(matchlist_w)
@@ -223,7 +223,7 @@
if ctx.match_start == ctx.match_end: # zero-width match
if ctx.match_start == ctx.end: # or end of string
break
- ctx.reset(ctx.next(ctx.match_end))
+ ctx.reset(ctx.next_indirect(ctx.match_end))
continue
splitlist.append(slice_w(space, ctx, last, ctx.match_start,
space.w_None))
@@ -320,7 +320,7 @@
strbuilder, last_pos, ctx.match_start)
start = ctx.match_end
if start == ctx.match_start:
- start = ctx.next(start)
+ start = ctx.next_indirect(start)
if not (last_pos == ctx.match_start
== ctx.match_end and n > 0):
# the above ignores empty matches on latest position
@@ -682,13 +682,13 @@
ctx = self.ctx
nextstart = ctx.match_end
if ctx.match_start == nextstart:
- nextstart = ctx.next(nextstart)
+ nextstart = ctx.next_indirect(nextstart)
self.ctx = ctx.fresh_copy(nextstart)
match = W_SRE_Match(self.srepat, ctx)
return match
else:
# obscure corner case
- self.ctx.match_start = self.ctx.next(self.ctx.match_start)
+ self.ctx.match_start = self.ctx.next_indirect(self.ctx.match_start)
return None
W_SRE_Scanner.typedef = TypeDef(
diff --git a/rpython/rlib/rsre/rsre_core.py b/rpython/rlib/rsre/rsre_core.py
--- a/rpython/rlib/rsre/rsre_core.py
+++ b/rpython/rlib/rsre/rsre_core.py
@@ -172,6 +172,8 @@
return position1 - position2
def go_forward_by_bytes(self, base_position, index):
return base_position + index
+ def next_indirect(self, position):
+ return position + 1 # like next(), but can be called indirectly
def get_mark(self, gid):
return find_mark(self.match_marks, gid)
diff --git a/rpython/rlib/rsre/rsre_utf8.py b/rpython/rlib/rsre/rsre_utf8.py
--- a/rpython/rlib/rsre/rsre_utf8.py
+++ b/rpython/rlib/rsre/rsre_utf8.py
@@ -34,6 +34,7 @@
def next(self, position):
return rutf8.next_codepoint_pos(self._utf8, position)
+ next_indirect = next
def prev(self, position):
if position <= 0:
diff --git a/rpython/rlib/rsre/test/support.py
b/rpython/rlib/rsre/test/support.py
--- a/rpython/rlib/rsre/test/support.py
+++ b/rpython/rlib/rsre/test/support.py
@@ -29,6 +29,7 @@
def next(self, position):
assert isinstance(position, Position)
return Position(position._p + 1)
+ next_indirect = next
def prev(self, position):
assert isinstance(position, Position)
_______________________________________________
pypy-commit mailing list
pypy-commit@python.org
https://mail.python.org/mailman/listinfo/pypy-commit
