[pypy-svn] r31023 - in pypy/dist/pypy/module/bz2: . test
rhymes at codespeak.net
rhymes at codespeak.net
Sat Aug 5 18:42:00 CEST 2006
Author: rhymes
Date: Sat Aug 5 18:41:56 2006
New Revision: 31023
Modified:
pypy/dist/pypy/module/bz2/interp_bz2.py
pypy/dist/pypy/module/bz2/test/test_bz2.py
Log:
BZ2File.read() is in place too
Modified: pypy/dist/pypy/module/bz2/interp_bz2.py
==============================================================================
--- pypy/dist/pypy/module/bz2/interp_bz2.py (original)
+++ pypy/dist/pypy/module/bz2/interp_bz2.py Sat Aug 5 18:41:56 2006
@@ -78,6 +78,11 @@
else:
SMALLCHUNK = BUFSIZ
+if sizeof(c_int) > 4:
+ BIGCHUNK = 512 * 32
+else:
+ BIGCHUNK = 512 * 1024
+
MAXINT = sys.maxint
pythonapi.PyFile_FromString.argtypes = [c_char_p, c_char_p]
@@ -225,7 +230,7 @@
while True:
libbz2.BZ2_bzRead(byref(bzerror), obj.fp, byref(ch), 1)
obj.pos += 1
- if bzerror != BZ_OK or buf_pos == end_pos:
+ if bzerror.value != BZ_OK or buf_pos == end_pos:
break
if skipnextlf:
@@ -235,7 +240,7 @@
# skipnextlf true means we saw a \r before.
newlinetypes |= NEWLINE_CRLF
libbz2.BZ2_bzRead(byref(bzerror), obj.fp, byref(ch), 1)
- if bzerror != BZ_OK: break
+ if bzerror.value != BZ_OK: break
else:
newlinetypes |= NEWLINE_CR
@@ -248,7 +253,7 @@
buf_pos += 1
if ch.value == '\n': break
- if bzerror == BZ_STREAM_END and skipnextlf:
+ if bzerror.value == BZ_STREAM_END and skipnextlf:
newlinetypes |= NEWLINE_CR
else: # if not universal newlines use the normal loop
while True:
@@ -268,7 +273,7 @@
obj.mode = MODE_READ_EOF
break
elif bzerror.value != BZ_OK:
- _catch_bz2_error(space, bzerror)
+ _catch_bz2_error(space, bzerror.value)
if ch.value == '\n': break
# must be because buf_pos == end_pos
@@ -289,7 +294,17 @@
used_v_size = buf_pos
if used_v_size != total_v_size:
return "".join(buf_lst[:used_v_size])
- return "".join(buf_lst)
+ return "".join(buf_lst)
+
+def _new_buffer_size(current_size):
+ if current_size > SMALLCHUNK:
+ # keep doubling until we reach BIGCHUNK
+ # then keep adding BIGCHUNK
+ if current_size <= BIGCHUNK:
+ return current_size + current_size
+ else:
+ return current_size + BIGCHUNK
+ return current_size + SMALLCHUNK
class _BZ2File(Wrappable):
def __init__(self, space, filename, mode='r', buffering=-1, compresslevel=9):
@@ -362,8 +377,8 @@
self.fp = libbz2.BZ2_bzWriteOpen(byref(bzerror), self._file,
compresslevel, 0, 0)
- if bzerror != BZ_OK:
- _catch_bz2_error(self.space, bzerror)
+ if bzerror.value != BZ_OK:
+ _catch_bz2_error(self.space, bzerror.value)
self.mode = (MODE_WRITE, MODE_READ)[mode_char == 'r']
@@ -406,8 +421,8 @@
raise OperationError(self.space.w_IOError,
self.space.wrap(_get_error_msg()))
- if bzerror != BZ_OK:
- return _catch_bz2_error(self.space, bzerror)
+ if bzerror.value != BZ_OK:
+ return _catch_bz2_error(self.space, bzerror.value)
return ret
close.unwrap_spec = ['self']
@@ -454,10 +469,10 @@
self.pos += chunksize
bytesread += chunksize
- if bzerror == BZ_STREAM_END:
+ if bzerror.value == BZ_STREAM_END:
break
- elif bzerror != BZ_OK:
- _catch_bz2_error(self.space, bzerror)
+ elif bzerror.value != BZ_OK:
+ _catch_bz2_error(self.space, bzerror.value)
self.mode = MODE_READ_EOF
self.size = self.pos
@@ -474,8 +489,8 @@
else:
# we cannot move back, so rewind the stream
libbz2.BZ2_bzReadClose(byref(bzerror), self.fp)
- if bzerror != BZ_OK:
- _catch_bz2_error(self.space, bzerror)
+ if bzerror.value != BZ_OK:
+ _catch_bz2_error(self.space, bzerror.value)
ret = libc.fseek(self._file, 0, SEEK_SET)
if ret != 0:
@@ -485,8 +500,8 @@
self.pos = 0
self.fp = libbz2.BZ2_bzReadOpen(byref(bzerror), self._file,
0, 0, None, 0)
- if bzerror != BZ_OK:
- _catch_bz2_error(self.space, bzerror)
+ if bzerror.value != BZ_OK:
+ _catch_bz2_error(self.space, bzerror.value)
self.mode = MODE_READ
@@ -508,11 +523,11 @@
self.pos += chunksize
bytesread += chunksize
- if bzerror == BZ_STREAM_END:
+ if bzerror.value == BZ_STREAM_END:
self.size = self.pos
self.mode = MODE_READ_EOF
- elif bzerror != BZ_OK:
- _catch_bz2_error(self.space, bzerror)
+ elif bzerror.value != BZ_OK:
+ _catch_bz2_error(self.space, bzerror.value)
if bytesread == offset:
break
@@ -540,14 +555,63 @@
size = (size, 0)[size < 0]
return self.space.wrap(_getline(self.space, self, size))
readline.unwrap_spec = ['self', int]
-
+
+ def read(self, size=-1):
+ """read([size]) -> string
+
+ Read at most size uncompressed bytes, returned as a string. If the size
+ argument is negative or omitted, read until EOF is reached."""
+
+ self._check_if_close()
+
+ if self.mode == MODE_READ_EOF:
+ return self.space.wrap("")
+ elif not self.mode == MODE_READ:
+ raise OperationError(self.space.w_IOError,
+ self.space.wrap("file is not ready for reading"))
+
+ bufsize = (size, _new_buffer_size(0))[size < 0]
+
+ if bufsize > MAXINT:
+ raise OperationError(self.space.w_OverflowError,
+ self.space.wrap(
+ "requested number of bytes is more than a Python string can hold"))
+
+ bytesread = 0
+ buf = create_string_buffer(bufsize)
+ bzerror = c_int()
+ while True:
+ chunksize = _univ_newline_read(bzerror, self.fp, buf,
+ bufsize - bytesread, self)
+ self.pos += chunksize
+ bytesread += chunksize
+
+ if bzerror.value == BZ_STREAM_END:
+ self.size = self.pos
+ self.mode = MODE_READ_EOF
+ break
+ elif bzerror.value != BZ_OK:
+ _catch_bz2_error(self.space, bzerror.value)
+
+ if size < 0:
+ bufsize = _new_buffer_size(bufsize)
+ else:
+ break
+
+ buf_lst = list(buf.value)
+ if bytesread != bufsize:
+ return self.space.wrap("".join(buf_lst[:bytesread]))
+ return self.space.wrap("".join(buf_lst))
+ read.unwrap_spec = ['self', int]
+
_BZ2File.typedef = TypeDef("_BZ2File",
close = interp2app(_BZ2File.close, unwrap_spec=_BZ2File.close.unwrap_spec),
tell = interp2app(_BZ2File.tell, unwrap_spec=_BZ2File.tell.unwrap_spec),
seek = interp2app(_BZ2File.seek, unwrap_spec=_BZ2File.seek.unwrap_spec),
readline = interp2app(_BZ2File.readline,
- unwrap_spec=_BZ2File.readline.unwrap_spec)
+ unwrap_spec=_BZ2File.readline.unwrap_spec),
+ read = interp2app(_BZ2File.read, unwrap_spec=_BZ2File.read.unwrap_spec)
)
def BZ2File(space, filename, mode='r', buffering=-1, compresslevel=9):
Modified: pypy/dist/pypy/module/bz2/test/test_bz2.py
==============================================================================
--- pypy/dist/pypy/module/bz2/test/test_bz2.py (original)
+++ pypy/dist/pypy/module/bz2/test/test_bz2.py Sat Aug 5 18:41:56 2006
@@ -103,6 +103,29 @@
line_read = bz2f.readline()
assert line_read == line
bz2f.close()
+
+ def test_read(self):
+ def create_temp_file(crlf=False):
+ DATA = 'BZh91AY&SY.\xc8N\x18\x00\x01>_\x80\x00\x10@\x02\xff\xf0\x01\x07n\x00?\xe7\xff\xe00\x01\x99\xaa\x00\xc0\x03F\x86\x8c#&\x83F\x9a\x03\x06\xa6\xd0\xa6\x93M\x0fQ\xa7\xa8\x06\x804hh\x12$\x11\xa4i4\xf14S\xd2<Q\xb5\x0fH\xd3\xd4\xdd\xd5\x87\xbb\xf8\x94\r\x8f\xafI\x12\xe1\xc9\xf8/E\x00pu\x89\x12]\xc9\xbbDL\nQ\x0e\t1\x12\xdf\xa0\xc0\x97\xac2O9\x89\x13\x94\x0e\x1c7\x0ed\x95I\x0c\xaaJ\xa4\x18L\x10\x05#\x9c\xaf\xba\xbc/\x97\x8a#C\xc8\xe1\x8cW\xf9\xe2\xd0\xd6M\xa7\x8bXa<e\x84t\xcbL\xb3\xa7\xd9\xcd\xd1\xcb\x84.\xaf\xb3\xab\xab\xad`n}\xa0lh\tE,\x8eZ\x15\x17VH>\x88\xe5\xcd9gd6\x0b\n\xe9\x9b\xd5\x8a\x99\xf7\x08.K\x8ev\xfb\xf7xw\xbb\xdf\xa1\x92\xf1\xdd|/";\xa2\xba\x9f\xd5\xb1#A\xb6\xf6\xb3o\xc9\xc5y\\\xebO\xe7\x85\x9a\xbc\xb6f8\x952\xd5\xd7"%\x89>V,\xf7\xa6z\xe2\x9f\xa3\xdf\x11\x11"\xd6E)I\xa9\x13^\xca\xf3r\xd0\x03U\x922\xf26\xec\xb6\xed\x8b\xc3U\x13\x9d\xc5\x170\xa4\xfa^\x92\xacDF\x8a\x97\xd6\x19\xfe\xdd\xb8\xbd\x1a\x9a\x19\xa3\x80ankR\x8b\xe5\xd83]\xa9\xc6\x08\x82f\xf6\xb9"6l$\xb8j@\xc0\x8a\xb0l1..\xbak\x83ls\x15\xbc\xf4\xc1\x13\xbe\xf8E\xb8\x9d\r\xa8\x9dk\x84\xd3n\xfa\xacQ\x07\xb1%y\xaav\xb4\x08\xe0z\x1b\x16\xf5\x04\xe9\xcc\xb9\x08z\x1en7.G\xfc]\xc9\x14\xe1B@\xbb!8`'
+ DATA_CRLF = 'BZh91AY&SY\xaez\xbbN\x00\x01H\xdf\x80\x00\x12@\x02\xff\xf0\x01\x07n\x00?\xe7\xff\xe0@\x01\xbc\xc6`\x86*\x8d=M\xa9\x9a\x86\xd0L@\x0fI\xa6!\xa1\x13\xc8\x88jdi\x8d@\x03@\x1a\x1a\x0c\x0c\x83 \x00\xc4h2\x19\x01\x82D\x84e\t\xe8\x99\x89\x19\x1ah\x00\r\x1a\x11\xaf\x9b\x0fG\xf5(\x1b\x1f?\t\x12\xcf\xb5\xfc\x95E\x00ps\x89\x12^\xa4\xdd\xa2&\x05(\x87\x04\x98\x89u\xe40%\xb6\x19\'\x8c\xc4\x89\xca\x07\x0e\x1b!\x91UIFU%C\x994!DI\xd2\xfa\xf0\xf1N8W\xde\x13A\xf5\x9cr%?\x9f3;I45A\xd1\x8bT\xb1<l\xba\xcb_\xc00xY\x17r\x17\x88\x08\x08@\xa0\ry@\x10\x04$)`\xf2\xce\x89z\xb0s\xec\x9b.iW\x9d\x81\xb5-+t\x9f\x1a\'\x97dB\xf5x\xb5\xbe.[.\xd7\x0e\x81\xe7\x08\x1cN`\x88\x10\xca\x87\xc3!"\x80\x92R\xa1/\xd1\xc0\xe6mf\xac\xbd\x99\xcca\xb3\x8780>\xa4\xc7\x8d\x1a\\"\xad\xa1\xabyBg\x15\xb9l\x88\x88\x91k"\x94\xa4\xd4\x89\xae*\xa6\x0b\x10\x0c\xd6\xd4m\xe86\xec\xb5j\x8a\x86j\';\xca.\x01I\xf2\xaaJ\xe8\x88\x8cU+t3\xfb\x0c\n\xa33\x13r2\r\x16\xe0\xb3(\xbf\x1d\x83r\xe7M\xf0D\x1365\xd8\x88\xd3\xa4\x92\xcb2\x06\x04\\\xc1\xb0\xea//\xbek&\xd8\xe6+t\xe5\xa1\x13\xada\x16\xder5"w]\xa2i\xb7[\x97R \xe2IT\xcd;Z\x04dk4\xad\x8a\t\xd3\x81z\x10\xf1:^`\xab\x1f\xc5\xdc\x91N\x14$+\x9e\xae\xd3\x80'
+
+ f = open("foo", "wb")
+
+ data = (DATA, DATA_CRLF)[crlf]
+ f.write(data)
+ f.close()
+
+ from bz2 import BZ2File
+ create_temp_file()
+
+ TEXT = 'root:x:0:0:root:/root:/bin/bash\nbin:x:1:1:bin:/bin:\ndaemon:x:2:2:daemon:/sbin:\nadm:x:3:4:adm:/var/adm:\nlp:x:4:7:lp:/var/spool/lpd:\nsync:x:5:0:sync:/sbin:/bin/sync\nshutdown:x:6:0:shutdown:/sbin:/sbin/shutdown\nhalt:x:7:0:halt:/sbin:/sbin/halt\nmail:x:8:12:mail:/var/spool/mail:\nnews:x:9:13:news:/var/spool/news:\nuucp:x:10:14:uucp:/var/spool/uucp:\noperator:x:11:0:operator:/root:\ngames:x:12:100:games:/usr/games:\ngopher:x:13:30:gopher:/usr/lib/gopher-data:\nftp:x:14:50:FTP User:/var/ftp:/bin/bash\nnobody:x:65534:65534:Nobody:/home:\npostfix:x:100:101:postfix:/var/spool/postfix:\nniemeyer:x:500:500::/home/niemeyer:/bin/bash\npostgres:x:101:102:PostgreSQL Server:/var/lib/pgsql:/bin/bash\nmysql:x:102:103:MySQL server:/var/lib/mysql:/bin/bash\nwww:x:103:104::/var/www:/bin/false\n'
+
+ bz2f = BZ2File("foo")
+ raises(TypeError, bz2f.read, None)
+ text_read = bz2f.read()
+ assert text_read == TEXT
+ bz2f.close()
+
# #!/usr/bin/python
# from test import test_support
@@ -161,14 +184,7 @@
# f.write(data)
# f.close()
#
-# def testRead(self):
-# # "Test BZ2File.read()"
-# self.createTempFile()
-# bz2f = BZ2File(self.filename)
-# self.assertRaises(TypeError, bz2f.read, None)
-# self.assertEqual(bz2f.read(), self.TEXT)
-# bz2f.close()
-#
+
# def testReadChunk10(self):
# # "Test BZ2File.read() in chunks of 10 bytes"
# self.createTempFile()
More information about the Pypy-commit
mailing list