Author: fijal
Branch: str-measure
Changeset: r91221:9923d9d8b6c5
Date: 2017-05-10 16:17 +0200
http://bitbucket.org/pypy/pypy/changeset/9923d9d8b6c5/
Log: a branch to start measuring how strings are used
diff --git a/pypy/interpreter/executioncontext.py
b/pypy/interpreter/executioncontext.py
--- a/pypy/interpreter/executioncontext.py
+++ b/pypy/interpreter/executioncontext.py
@@ -153,6 +153,7 @@
Like bytecode_trace() but doesn't invoke any other events besides the
trace function.
"""
+ frame._frame_counter += 1
if (frame.get_w_f_trace() is None or self.is_tracing or
self.gettrace() is None):
return
diff --git a/pypy/interpreter/pyframe.py b/pypy/interpreter/pyframe.py
--- a/pypy/interpreter/pyframe.py
+++ b/pypy/interpreter/pyframe.py
@@ -76,6 +76,9 @@
valuestackdepth = 0 # number of items on valuestack
lastblock = None
+ # XXX string debugging
+ _frame_counter = 0
+
# other fields:
# builtin - builtin cache, only if honor__builtins__ is True
diff --git a/pypy/module/__pypy__/__init__.py b/pypy/module/__pypy__/__init__.py
--- a/pypy/module/__pypy__/__init__.py
+++ b/pypy/module/__pypy__/__init__.py
@@ -98,6 +98,7 @@
'decode_long' : 'interp_magic.decode_long',
'_promote' : 'interp_magic._promote',
'stack_almost_full' : 'interp_magic.stack_almost_full',
+ 'set_str_debug_file' : 'interp_debug.set_str_debug_file',
}
if sys.platform == 'win32':
interpleveldefs['get_console_cp'] = 'interp_magic.get_console_cp'
diff --git a/pypy/module/__pypy__/interp_debug.py
b/pypy/module/__pypy__/interp_debug.py
--- a/pypy/module/__pypy__/interp_debug.py
+++ b/pypy/module/__pypy__/interp_debug.py
@@ -28,3 +28,15 @@
@jit.dont_look_inside
def debug_flush(space):
debug.debug_flush()
+
+class Cache(object):
+ def __init__(self, space):
+ self.w_debug_file = None
+
+def set_str_debug_file(space, w_debug_file):
+ if space.is_none(w_debug_file):
+ w_debug_file = None
+ space.fromcache(Cache).w_debug_file = w_debug_file
+
+def get_str_debug_file(space):
+ return space.fromcache(Cache).w_debug_file
\ No newline at end of file
diff --git a/pypy/objspace/std/bytearrayobject.py
b/pypy/objspace/std/bytearrayobject.py
--- a/pypy/objspace/std/bytearrayobject.py
+++ b/pypy/objspace/std/bytearrayobject.py
@@ -50,15 +50,15 @@
def nonmovable_carray(self, space):
return BytearrayBuffer(self.data, False).get_raw_address()
- def _new(self, value):
+ def _new(self, space, value):
if value is self.data:
value = value[:]
return W_BytearrayObject(value)
- def _new_from_buffer(self, buffer):
+ def _new_from_buffer(self, space, buffer):
return W_BytearrayObject([buffer[i] for i in range(len(buffer))])
- def _new_from_list(self, value):
+ def _new_from_list(self, space, value):
return W_BytearrayObject(value)
def _empty(self):
@@ -443,10 +443,10 @@
def descr_add(self, space, w_other):
if isinstance(w_other, W_BytearrayObject):
- return self._new(self.data + w_other.data)
+ return self._new(space, self.data + w_other.data)
if isinstance(w_other, W_BytesObject):
- return self._add(self._op_val(space, w_other))
+ return self._add(space, self._op_val(space, w_other))
try:
buffer = _get_buffer(space, w_other)
@@ -454,11 +454,11 @@
if e.match(space, space.w_TypeError):
return space.w_NotImplemented
raise
- return self._add(buffer)
+ return self._add(space, buffer)
- @specialize.argtype(1)
- def _add(self, other):
- return self._new(self.data + [other[i] for i in range(len(other))])
+ @specialize.argtype(2)
+ def _add(self, space, other):
+ return self._new(space, self.data + [other[i] for i in
range(len(other))])
def descr_reverse(self, space):
self.data.reverse()
diff --git a/pypy/objspace/std/bytesobject.py b/pypy/objspace/std/bytesobject.py
--- a/pypy/objspace/std/bytesobject.py
+++ b/pypy/objspace/std/bytesobject.py
@@ -481,10 +481,10 @@
"found", len(self._value))
return space.newint(ord(self._value[0]))
- def _new(self, value):
+ def _new(self, space, value):
return W_BytesObject(value)
- def _new_from_list(self, value):
+ def _new_from_list(self, space, value):
return W_BytesObject(''.join(value))
def _empty(self):
diff --git a/pypy/objspace/std/objspace.py b/pypy/objspace/std/objspace.py
--- a/pypy/objspace/std/objspace.py
+++ b/pypy/objspace/std/objspace.py
@@ -361,7 +361,7 @@
def newunicode(self, uni):
assert uni is not None
assert isinstance(uni, unicode)
- return W_UnicodeObject(uni)
+ return W_UnicodeObject(self, uni)
def type(self, w_obj):
jit.promote(w_obj.__class__)
diff --git a/pypy/objspace/std/stringmethods.py
b/pypy/objspace/std/stringmethods.py
--- a/pypy/objspace/std/stringmethods.py
+++ b/pypy/objspace/std/stringmethods.py
@@ -20,7 +20,7 @@
#if start == 0 and stop == len(s) and space.is_w(space.type(orig_obj),
# space.w_bytes):
# return orig_obj
- return self._new(s[start:stop])
+ return self._new(space, s[start:stop])
def _convert_idx_params(self, space, w_start, w_end):
value = self._val(space)
@@ -61,7 +61,7 @@
if e.match(space, space.w_TypeError):
return space.w_NotImplemented
raise
- return self._new(self._val(space) + other)
+ return self._new(space, self._val(space) + other)
# Bytearray overrides this method, CPython doesn't support contacting
# buffers and strs, and unicodes are always handled above
@@ -77,8 +77,8 @@
if times <= 0:
return self._empty()
if self._len() == 1:
- return self._new(self._multi_chr(self._val(space)[0]) * times)
- return self._new(self._val(space) * times)
+ return self._new(space, self._multi_chr(self._val(space)[0]) *
times)
+ return self._new(space, self._val(space) * times)
descr_rmul = descr_mul
@@ -94,7 +94,7 @@
return self._sliced(space, selfvalue, start, stop, self)
else:
ret = _descr_getslice_slowpath(selfvalue, start, step, sl)
- return self._new_from_list(ret)
+ return self._new_from_list(space, ret)
index = space.getindex_w(w_index, space.w_IndexError, "string index")
return self._getitem_result(space, index)
@@ -105,7 +105,7 @@
character = selfvalue[index]
except IndexError:
raise oefmt(space.w_IndexError, "string index out of range")
- return self._new(character)
+ return self._new(space, character)
def descr_getslice(self, space, w_start, w_stop):
selfvalue = self._val(space)
@@ -125,7 +125,7 @@
builder.append(self._upper(value[0]))
for i in range(1, len(value)):
builder.append(self._lower(value[i]))
- return self._new(builder.build())
+ return self._new(space, builder.build())
@unwrap_spec(width=int, w_fillchar=WrappedDefault(' '))
def descr_center(self, space, width, w_fillchar):
@@ -143,7 +143,7 @@
else:
centered = value
- return self._new(centered)
+ return self._new(space, centered)
def descr_count(self, space, w_sub, w_start=None, w_end=None):
value, start, end = self._convert_idx_params(space, w_start, w_end)
@@ -207,7 +207,7 @@
tabsize) + token
oldtoken = token
- return self._new(expanded)
+ return self._new(space, expanded)
def _tabindent(self, token, tabsize):
"""calculates distance behind the token to the next tabstop"""
@@ -442,7 +442,7 @@
if value and i != 0:
sb.append(value)
sb.append(unwrapped[i])
- return self._new(sb.build())
+ return self._new(space, sb.build())
def _join_autoconvert(self, space, list_w):
assert False, 'unreachable'
@@ -459,7 +459,7 @@
fillchar = self._multi_chr(fillchar[0])
value = value + fillchar * d
- return self._new(value)
+ return self._new(space, value)
@unwrap_spec(width=int, w_fillchar=WrappedDefault(' '))
def descr_rjust(self, space, width, w_fillchar):
@@ -473,14 +473,14 @@
fillchar = self._multi_chr(fillchar[0])
value = d * fillchar + value
- return self._new(value)
+ return self._new(space, value)
def descr_lower(self, space):
value = self._val(space)
builder = self._builder(len(value))
for i in range(len(value)):
builder.append(self._lower(value[i]))
- return self._new(builder.build())
+ return self._new(space, builder.build())
def descr_partition(self, space, w_sub):
from pypy.objspace.std.bytearrayobject import W_BytearrayObject
@@ -501,11 +501,11 @@
pos = find(value, sub, 0, len(value))
if pos != -1 and isinstance(self, W_BytearrayObject):
- w_sub = self._new_from_buffer(sub)
+ w_sub = self._new_from_buffer(space, sub)
if pos == -1:
if isinstance(self, W_BytearrayObject):
- self = self._new(value)
+ self = self._new(space, value)
return space.newtuple([self, self._empty(), self._empty()])
else:
return space.newtuple(
@@ -531,11 +531,11 @@
pos = rfind(value, sub, 0, len(value))
if pos != -1 and isinstance(self, W_BytearrayObject):
- w_sub = self._new_from_buffer(sub)
+ w_sub = self._new_from_buffer(space, sub)
if pos == -1:
if isinstance(self, W_BytearrayObject):
- self = self._new(value)
+ self = self._new(space, value)
return space.newtuple([self._empty(), self._empty(), self])
else:
return space.newtuple(
@@ -557,7 +557,7 @@
except OverflowError:
raise oefmt(space.w_OverflowError, "replace string is too long")
- return self._new(res)
+ return self._new(space, res)
@unwrap_spec(maxsplit=int)
def descr_split(self, space, w_sep=None, maxsplit=-1):
@@ -716,13 +716,13 @@
builder.append(self._upper(ch))
else:
builder.append(ch)
- return self._new(builder.build())
+ return self._new(space, builder.build())
def descr_title(self, space):
selfval = self._val(space)
if len(selfval) == 0:
return self
- return self._new(self.title(selfval))
+ return self._new(space, self.title(selfval))
@jit.elidable
def title(self, value):
@@ -764,24 +764,24 @@
for char in string:
if not deletion_table[ord(char)]:
buf.append(table[ord(char)])
- return self._new(buf.build())
+ return self._new(space, buf.build())
def descr_upper(self, space):
value = self._val(space)
builder = self._builder(len(value))
for i in range(len(value)):
builder.append(self._upper(value[i]))
- return self._new(builder.build())
+ return self._new(space, builder.build())
@unwrap_spec(width=int)
def descr_zfill(self, space, width):
selfval = self._val(space)
if len(selfval) == 0:
- return self._new(self._multi_chr(self._chr('0')) * width)
+ return self._new(space, self._multi_chr(self._chr('0')) * width)
num_zeros = width - len(selfval)
if num_zeros <= 0:
# cannot return self, in case it is a subclass of str
- return self._new(selfval)
+ return self._new(space, selfval)
builder = self._builder(width)
if len(selfval) > 0 and (selfval[0] == '+' or selfval[0] == '-'):
@@ -792,10 +792,10 @@
start = 0
builder.append_multiple_char(self._chr('0'), num_zeros)
builder.append_slice(selfval, start, len(selfval))
- return self._new(builder.build())
+ return self._new(space, builder.build())
def descr_getnewargs(self, space):
- return space.newtuple([self._new(self._val(space))])
+ return space.newtuple([self._new(space, self._val(space))])
# ____________________________________________________________
# helpers for slow paths, moved out because they contain loops
diff --git a/pypy/objspace/std/unicodeobject.py
b/pypy/objspace/std/unicodeobject.py
--- a/pypy/objspace/std/unicodeobject.py
+++ b/pypy/objspace/std/unicodeobject.py
@@ -1,5 +1,6 @@
"""The builtin unicode implementation"""
+import py
from rpython.rlib.objectmodel import (
compute_hash, compute_unique_id, import_from_mixin,
enforceargs)
@@ -29,11 +30,26 @@
class W_UnicodeObject(W_Root):
import_from_mixin(StringMethods)
_immutable_fields_ = ['_value']
+ _frame_counter = 0
+ _frame_id = 0
@enforceargs(uni=unicode)
- def __init__(self, unistr):
+ def __init__(self, space, unistr):
assert isinstance(unistr, unicode)
self._value = unistr
+ if space is None:
+ return
+ frame = space.getexecutioncontext().gettopframe()
+ if frame is None:
+ return
+ self._frame_counter = frame._frame_counter
+ self._frame_id = compute_unique_id(frame)
+ from pypy.module.__pypy__.interp_debug import get_str_debug_file
+ w_file = get_str_debug_file(space)
+ if w_file is None:
+ return
+ space.call_function(space.getattr(w_file, space.newtext("write")),
+ space.newtext("descr_new %s %s\n" % (self._frame_counter,
self._frame_id)))
def __repr__(self):
"""representation for debugging purposes"""
@@ -43,10 +59,10 @@
# for testing
return self._value
- def create_if_subclassed(self):
+ def create_if_subclassed(self, space):
if type(self) is W_UnicodeObject:
return self
- return W_UnicodeObject(self._value)
+ return W_UnicodeObject(space, self._value)
def is_w(self, space, w_other):
if not isinstance(w_other, W_UnicodeObject):
@@ -105,11 +121,11 @@
"found", len(self._value))
return space.newint(ord(self._value[0]))
- def _new(self, value):
- return W_UnicodeObject(value)
+ def _new(self, space, value):
+ return W_UnicodeObject(space, value)
- def _new_from_list(self, value):
- return W_UnicodeObject(u''.join(value))
+ def _new_from_list(self, space, value):
+ return W_UnicodeObject(space, u''.join(value))
def _empty(self):
return W_UnicodeObject.EMPTY
@@ -222,7 +238,7 @@
assert isinstance(w_value, W_UnicodeObject)
w_newobj = space.allocate_instance(W_UnicodeObject, w_unicodetype)
- W_UnicodeObject.__init__(w_newobj, w_value._value)
+ W_UnicodeObject.__init__(w_newobj, space, w_value._value)
return w_newobj
def descr_repr(self, space):
@@ -354,7 +370,7 @@
raise oefmt(space.w_TypeError,
"character mapping must return integer, None "
"or unicode")
- return W_UnicodeObject(u''.join(result))
+ return W_UnicodeObject(space, u''.join(result))
def descr_encode(self, space, w_encoding=None, w_errors=None):
encoding, errors = _get_encoding_and_errors(space, w_encoding,
@@ -420,7 +436,7 @@
def wrapunicode(space, uni):
- return W_UnicodeObject(uni)
+ return W_UnicodeObject(space, uni)
def plain_str2unicode(space, s):
@@ -562,7 +578,7 @@
return unicode_from_encoded_object(space, w_bytes, encoding, "strict")
s = space.bytes_w(w_bytes)
try:
- return W_UnicodeObject(s.decode("ascii"))
+ return W_UnicodeObject(space, s.decode("ascii"))
except UnicodeDecodeError:
# raising UnicodeDecodeError is messy, "please crash for me"
return unicode_from_encoded_object(space, w_bytes, "ascii", "strict")
@@ -967,6 +983,39 @@
of the specified width. The string S is never truncated.
"""
+def setup():
+ from pypy.module.__pypy__.interp_debug import get_str_debug_file
+
+ def wrap(func):
+ d = {'orig': func, 'get_str_debug_file': get_str_debug_file}
+ name = func.__name__
+ orig_args = list(func.__code__.co_varnames[:func.__code__.co_argcount])
+ args = orig_args[:]
+ if func.func_defaults:
+ i = func.__code__.co_argcount - len(func.func_defaults)
+ for j, default in enumerate(func.func_defaults):
+ args[i] = "%s = %r" % (args[i], func.func_defaults[j])
+ i += 1
+ func_args = ", ".join(args)
+ lines = ["def %s(%s):" % (name, func_args),
+ " w_file = get_str_debug_file(space)",
+ " if w_file is not None:",
+ " txt = '%s ' + str(self._frame_counter) + ' ' +
str(self._frame_id) + ' '+ '\\n'" % func.func_name,
+ " space.call_function(space.getattr(w_file,
space.newtext('write')), space.newtext(txt))",
+ " return orig(%s)" % (", ".join(orig_args),)]
+ exec "\n".join(lines) in d
+ if hasattr(func, 'unwrap_spec'):
+ d[name].unwrap_spec = func.unwrap_spec
+ # get it as an unbound method
+ return d[name]
+
+ for k, v in W_UnicodeObject.__dict__.iteritems():
+ if k == 'descr_new':
+ continue
+ if k.startswith('descr_'):
+ setattr(W_UnicodeObject, k, wrap(getattr(W_UnicodeObject, k)))
+
+setup()
W_UnicodeObject.typedef = TypeDef(
"unicode", basestring_typedef,
@@ -1112,7 +1161,7 @@
return [s for s in value]
-W_UnicodeObject.EMPTY = W_UnicodeObject(u'')
+W_UnicodeObject.EMPTY = W_UnicodeObject(None, u'')
# Helper for converting int/long
diff --git a/rpython/translator/platform/darwin.py
b/rpython/translator/platform/darwin.py
--- a/rpython/translator/platform/darwin.py
+++ b/rpython/translator/platform/darwin.py
@@ -16,6 +16,7 @@
standalone_only = ('-mdynamic-no-pic',)
shared_only = ()
+ accepts_flto = False
link_flags = (DARWIN_VERSION_MIN,)
cflags = ('-O3', '-fomit-frame-pointer', DARWIN_VERSION_MIN)
diff --git a/rpython/translator/platform/posix.py
b/rpython/translator/platform/posix.py
--- a/rpython/translator/platform/posix.py
+++ b/rpython/translator/platform/posix.py
@@ -11,6 +11,7 @@
exe_ext = ''
make_cmd = 'make'
so_prefixes = ('lib', '')
+ accepts_flto = True
relevant_environ = ('CPATH', 'LIBRARY_PATH', 'C_INCLUDE_PATH')
@@ -131,7 +132,7 @@
cflags = tuple(self.cflags) + tuple(self.standalone_only)
# xxx check which compilers accept this option or not
- if not config or config.translation.gcrootfinder != 'asmgcc':
+ if (not config or config.translation.gcrootfinder != 'asmgcc') and
self.accepts_flto:
cflags = ('-flto',) + cflags
m = GnuMakefile(path)
_______________________________________________
pypy-commit mailing list
pypy-commit@python.org
https://mail.python.org/mailman/listinfo/pypy-commit
