languages/cpython
languages/cpython
1
0
Fork 0
mirror of https://github.com/python/cpython synced 2024-10-14 22:49:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

gh-105069: Add a readline-like callable to the tokenizer to consume input iteratively (#105070)

Browse source
This commit is contained in:
Pablo Galindo Salgado 2023-05-30 22:43:34 +01:00 committed by GitHub
parent 2ea34cfb3a
commit 9216e69a87
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
7 changed files with 276 additions and 98 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
Lib/inspect.py
View file

@ -2203,7 +2203,7 @@ def _signature_strip_non_python_syntax(signature):
add(string)
if (string == ','):
add(' ')
clean_signature = ''.join(text).strip()
clean_signature = ''.join(text).strip().replace("\n", "")
return clean_signature, self_parameter

141
Lib/test/test_tokenize.py
View file

@ -1,6 +1,6 @@
from test import support
from test.support import os_helper
from tokenize import (tokenize, _tokenize, untokenize, NUMBER, NAME, OP,
from tokenize import (tokenize, untokenize, NUMBER, NAME, OP,
STRING, ENDMARKER, ENCODING, tok_name, detect_encoding,
open as tokenize_open, Untokenizer, generate_tokens,
NEWLINE, _generate_tokens_from_c_tokenizer, DEDENT, TokenInfo)
@ -51,6 +51,25 @@ def check_tokenize(self, s, expected):
[" ENCODING 'utf-8' (0, 0) (0, 0)"] +
expected.rstrip().splitlines())
def test_invalid_readline(self):
def gen():
yield "sdfosdg"
yield "sdfosdg"
with self.assertRaises(TypeError):
list(tokenize(gen().__next__))
def gen():
yield b"sdfosdg"
yield b"sdfosdg"
with self.assertRaises(TypeError):
list(generate_tokens(gen().__next__))
def gen():
yield "sdfosdg"
1/0
with self.assertRaises(ZeroDivisionError):
list(generate_tokens(gen().__next__))
def test_implicit_newline(self):
# Make sure that the tokenizer puts in an implicit NEWLINE
# when the input lacks a trailing new line.
@ -1154,7 +1173,8 @@ class TestTokenizerAdheresToPep0263(TestCase):
def _testFile(self, filename):
path = os.path.join(os.path.dirname(__file__), filename)
TestRoundtrip.check_roundtrip(self, open(path, 'rb'))
with open(path, 'rb') as f:
TestRoundtrip.check_roundtrip(self, f)
def test_utf8_coding_cookie_and_no_utf8_bom(self):
f = 'tokenize_tests-utf8-coding-cookie-and-no-utf8-bom-sig.txt'
@ -1199,7 +1219,8 @@ def readline():
yield b''
# skip the initial encoding token and the end tokens
tokens = list(_tokenize(readline(), encoding='utf-8'))[:-2]
tokens = list(_generate_tokens_from_c_tokenizer(readline().__next__, encoding='utf-8',
extra_tokens=True))[:-2]
expected_tokens = [TokenInfo(3, '"ЉЊЈЁЂ"', (1, 0), (1, 7), '"ЉЊЈЁЂ"\n')]
self.assertEqual(tokens, expected_tokens,
"bytes not decoded with encoding")
@ -1468,13 +1489,13 @@ def test_tokenize(self):
def mock_detect_encoding(readline):
return encoding, [b'first', b'second']
def mock__tokenize(readline, encoding):
def mock__tokenize(readline, encoding, **kwargs):
nonlocal encoding_used
encoding_used = encoding
out = []
while True:
try:
next_line = next(readline)
next_line = readline()
except StopIteration:
return out
if next_line:
@ -1491,16 +1512,16 @@ def mock_readline():
return str(counter).encode()
orig_detect_encoding = tokenize_module.detect_encoding
orig__tokenize = tokenize_module._tokenize
orig_c_token = tokenize_module._generate_tokens_from_c_tokenizer
tokenize_module.detect_encoding = mock_detect_encoding
tokenize_module._tokenize = mock__tokenize
tokenize_module._generate_tokens_from_c_tokenizer = mock__tokenize
try:
results = tokenize(mock_readline)
self.assertEqual(list(results)[1:],
[b'first', b'second', b'1', b'2', b'3', b'4'])
finally:
tokenize_module.detect_encoding = orig_detect_encoding
tokenize_module._tokenize = orig__tokenize
tokenize_module._generate_tokens_from_c_tokenizer = orig_c_token
self.assertEqual(encoding_used, encoding)
@ -1827,12 +1848,33 @@ class CTokenizeTest(TestCase):
def check_tokenize(self, s, expected):
# Format the tokens in s in a table format.
# The ENDMARKER and final NEWLINE are omitted.
f = StringIO(s)
with self.subTest(source=s):
result = stringify_tokens_from_source(
_generate_tokens_from_c_tokenizer(s), s
_generate_tokens_from_c_tokenizer(f.readline), s
)
self.assertEqual(result, expected.rstrip().splitlines())
def test_encoding(self):
def readline(encoding):
yield "1+1".encode(encoding)
expected = [
TokenInfo(type=NUMBER, string='1', start=(1, 0), end=(1, 1), line='1+1\n'),
TokenInfo(type=OP, string='+', start=(1, 1), end=(1, 2), line='1+1\n'),
TokenInfo(type=NUMBER, string='1', start=(1, 2), end=(1, 3), line='1+1\n'),
TokenInfo(type=NEWLINE, string='\n', start=(1, 3), end=(1, 4), line='1+1\n'),
TokenInfo(type=ENDMARKER, string='', start=(2, 0), end=(2, 0), line='')
]
for encoding in ["utf-8", "latin-1", "utf-16"]:
with self.subTest(encoding=encoding):
tokens = list(_generate_tokens_from_c_tokenizer(
readline(encoding).__next__,
extra_tokens=True,
encoding=encoding,
))
self.assertEqual(tokens, expected)
def test_int(self):
self.check_tokenize('0xff <= 255', """\
@ -2668,43 +2710,44 @@ def test_unicode(self):
def test_invalid_syntax(self):
def get_tokens(string):
return list(_generate_tokens_from_c_tokenizer(string))
the_string = StringIO(string)
return list(_generate_tokens_from_c_tokenizer(the_string.readline))
self.assertRaises(SyntaxError, get_tokens, "(1+2]")
self.assertRaises(SyntaxError, get_tokens, "(1+2}")
self.assertRaises(SyntaxError, get_tokens, "{1+2]")
for case in [
"(1+2]",
"(1+2}",
"{1+2]",
"1_",
"1.2_",
"1e2_",
"1e+",
self.assertRaises(SyntaxError, get_tokens, "1_")
self.assertRaises(SyntaxError, get_tokens, "1.2_")
self.assertRaises(SyntaxError, get_tokens, "1e2_")
self.assertRaises(SyntaxError, get_tokens, "1e+")
self.assertRaises(SyntaxError, get_tokens, "\xa0")
self.assertRaises(SyntaxError, get_tokens, "€")
self.assertRaises(SyntaxError, get_tokens, "0b12")
self.assertRaises(SyntaxError, get_tokens, "0b1_2")
self.assertRaises(SyntaxError, get_tokens, "0b2")
self.assertRaises(SyntaxError, get_tokens, "0b1_")
self.assertRaises(SyntaxError, get_tokens, "0b")
self.assertRaises(SyntaxError, get_tokens, "0o18")
self.assertRaises(SyntaxError, get_tokens, "0o1_8")
self.assertRaises(SyntaxError, get_tokens, "0o8")
self.assertRaises(SyntaxError, get_tokens, "0o1_")
self.assertRaises(SyntaxError, get_tokens, "0o")
self.assertRaises(SyntaxError, get_tokens, "0x1_")
self.assertRaises(SyntaxError, get_tokens, "0x")
self.assertRaises(SyntaxError, get_tokens, "1_")
self.assertRaises(SyntaxError, get_tokens, "012")
self.assertRaises(SyntaxError, get_tokens, "1.2_")
self.assertRaises(SyntaxError, get_tokens, "1e2_")
self.assertRaises(SyntaxError, get_tokens, "1e+")
self.assertRaises(SyntaxError, get_tokens, "'sdfsdf")
self.assertRaises(SyntaxError, get_tokens, "'''sdfsdf''")
self.assertRaises(SyntaxError, get_tokens, "("*1000+"a"+")"*1000)
self.assertRaises(SyntaxError, get_tokens, "]")
"\xa0",
"€",
"0b12",
"0b1_2",
"0b2",
"0b1_",
"0b",
"0o18",
"0o1_8",
"0o8",
"0o1_",
"0o",
"0x1_",
"0x",
"1_",
"012",
"1.2_",
"1e2_",
"1e+",
"'sdfsdf",
"'''sdfsdf''",
"("*1000+"a"+")"*1000,
"]",
]:
with self.subTest(case=case):
self.assertRaises(SyntaxError, get_tokens, case)
def test_max_indent(self):
MAXINDENT = 100
@ -2715,20 +2758,24 @@ def generate_source(indents):
return source
valid = generate_source(MAXINDENT - 1)
tokens = list(_generate_tokens_from_c_tokenizer(valid))
the_input = StringIO(valid)
tokens = list(_generate_tokens_from_c_tokenizer(the_input.readline))
self.assertEqual(tokens[-2].type, DEDENT)
self.assertEqual(tokens[-1].type, ENDMARKER)
compile(valid, "<string>", "exec")
invalid = generate_source(MAXINDENT)
self.assertRaises(SyntaxError, lambda: list(_generate_tokens_from_c_tokenizer(invalid)))
the_input = StringIO(invalid)
self.assertRaises(SyntaxError, lambda: list(_generate_tokens_from_c_tokenizer(the_input.readline)))
self.assertRaises(
IndentationError, compile, invalid, "<string>", "exec"
)
def test_continuation_lines_indentation(self):
def get_tokens(string):
return [(kind, string) for (kind, string, *_) in _generate_tokens_from_c_tokenizer(string)]
the_string = StringIO(string)
return [(kind, string) for (kind, string, *_)
in _generate_tokens_from_c_tokenizer(the_string.readline)]
code = dedent("""
def fib(n):

32
Lib/tokenize.py
View file

@ -34,6 +34,7 @@
import sys
from token import *
from token import EXACT_TOKEN_TYPES
import _tokenize
cookie_re = re.compile(r'^[ \t\f]*#.*?coding[:=][ \t]*([-\w.]+)', re.ASCII)
blank_re = re.compile(br'^[ \t\f]*(?:[#\r\n]|$)', re.ASCII)
@ -443,12 +444,7 @@ def tokenize(readline):
# BOM will already have been stripped.
encoding = "utf-8"
yield TokenInfo(ENCODING, encoding, (0, 0), (0, 0), '')
yield from _tokenize(rl_gen, encoding)
def _tokenize(rl_gen, encoding):
source = b"".join(rl_gen).decode(encoding)
for token in _generate_tokens_from_c_tokenizer(source, extra_tokens=True):
yield token
yield from _generate_tokens_from_c_tokenizer(rl_gen.__next__, encoding, extra_tokens=True)
def generate_tokens(readline):
"""Tokenize a source reading Python code as unicode strings.
@ -456,16 +452,7 @@ def generate_tokens(readline):
This has the same API as tokenize(), except that it expects the *readline*
callable to return str objects instead of bytes.
"""
def _gen():
while True:
try:
line = readline()
except StopIteration:
return
if not line:
return
yield line.encode()
return _tokenize(_gen(), 'utf-8')
return _generate_tokens_from_c_tokenizer(readline, extra_tokens=True)
def main():
import argparse
@ -502,9 +489,9 @@ def error(message, filename=None, location=None):
tokens = list(tokenize(f.readline))
else:
filename = "<stdin>"
tokens = _tokenize(
tokens = _generate_tokens_from_c_tokenizer(
(x.encode('utf-8') for x in iter(sys.stdin.readline, "")
), "utf-8")
), "utf-8", extra_tokens=True)
# Output the tokenization
@ -531,10 +518,13 @@ def error(message, filename=None, location=None):
perror("unexpected error: %s" % err)
raise
def _generate_tokens_from_c_tokenizer(source, extra_tokens=False):
def _generate_tokens_from_c_tokenizer(source, encoding=None, extra_tokens=False):
"""Tokenize a source reading Python code as unicode strings using the internal C tokenizer"""
import _tokenize as c_tokenizer
for info in c_tokenizer.TokenizerIter(source, extra_tokens=extra_tokens):
if encoding is None:
it = _tokenize.TokenizerIter(source, extra_tokens=extra_tokens)
else:
it = _tokenize.TokenizerIter(source, encoding=encoding, extra_tokens=extra_tokens)
for info in it:
yield TokenInfo._make(info)

136
Parser/tokenizer.c
View file

@ -103,6 +103,7 @@ tok_new(void)
tok->filename = NULL;
tok->decoding_readline = NULL;
tok->decoding_buffer = NULL;
tok->readline = NULL;
tok->type_comments = 0;
tok->async_hacks = 0;
tok->async_def = 0;
@ -139,8 +140,9 @@ static char *
error_ret(struct tok_state *tok) /* XXX */
{
tok->decoding_erred = 1;
if (tok->fp != NULL && tok->buf != NULL) /* see _PyTokenizer_Free */
if ((tok->fp != NULL || tok->readline != NULL) && tok->buf != NULL) {/* see _PyTokenizer_Free */
PyMem_Free(tok->buf);
}
tok->buf = tok->cur = tok->inp = NULL;
tok->start = NULL;
tok->end = NULL;
@ -900,6 +902,33 @@ _PyTokenizer_FromString(const char *str, int exec_input, int preserve_crlf)
return tok;
}
struct tok_state *
_PyTokenizer_FromReadline(PyObject* readline, const char* enc,
int exec_input, int preserve_crlf)
{
struct tok_state *tok = tok_new();
if (tok == NULL)
return NULL;
if ((tok->buf = (char *)PyMem_Malloc(BUFSIZ)) == NULL) {
_PyTokenizer_Free(tok);
return NULL;
}
tok->cur = tok->inp = tok->buf;
tok->end = tok->buf + BUFSIZ;
tok->fp = NULL;
if (enc != NULL) {
tok->encoding = new_string(enc, strlen(enc), tok);
if (!tok->encoding) {
_PyTokenizer_Free(tok);
return NULL;
}
}
tok->decoding_state = STATE_NORMAL;
Py_INCREF(readline);
tok->readline = readline;
return tok;
}
/* Set up tokenizer for UTF-8 string */
struct tok_state *
@ -969,8 +998,9 @@ _PyTokenizer_Free(struct tok_state *tok)
}
Py_XDECREF(tok->decoding_readline);
Py_XDECREF(tok->decoding_buffer);
Py_XDECREF(tok->readline);
Py_XDECREF(tok->filename);
if (tok->fp != NULL && tok->buf != NULL) {
if ((tok->readline != NULL || tok->fp != NULL ) && tok->buf != NULL) {
PyMem_Free(tok->buf);
}
if (tok->input) {
@ -1021,6 +1051,71 @@ tok_readline_raw(struct tok_state *tok)
return 1;
}
static int
tok_readline_string(struct tok_state* tok) {
PyObject* line = NULL;
PyObject* raw_line = PyObject_CallNoArgs(tok->readline);
if (raw_line == NULL) {
if (PyErr_ExceptionMatches(PyExc_StopIteration)) {
PyErr_Clear();
return 1;
}
error_ret(tok);
goto error;
}
if(tok->encoding != NULL) {
if (!PyBytes_Check(raw_line)) {
PyErr_Format(PyExc_TypeError, "readline() returned a non-bytes object");
error_ret(tok);
goto error;
}
line = PyUnicode_Decode(PyBytes_AS_STRING(raw_line), PyBytes_GET_SIZE(raw_line),
tok->encoding, "replace");
Py_CLEAR(raw_line);
if (line == NULL) {
error_ret(tok);
goto error;
}
} else {
if(!PyUnicode_Check(raw_line)) {
PyErr_Format(PyExc_TypeError, "readline() returned a non-string object");
error_ret(tok);
goto error;
}
line = raw_line;
raw_line = NULL;
}
Py_ssize_t buflen;
const char* buf = PyUnicode_AsUTF8AndSize(line, &buflen);
if (buf == NULL) {
error_ret(tok);
goto error;
}
// Make room for the null terminator *and* potentially
// an extra newline character that we may need to artificially
// add.
size_t buffer_size = buflen + 2;
if (!tok_reserve_buf(tok, buffer_size)) {
goto error;
}
memcpy(tok->inp, buf, buflen);
tok->inp += buflen;
*tok->inp = '\0';
if (tok->start == NULL) {
tok->buf = tok->cur;
}
tok->line_start = tok->cur;
Py_DECREF(line);
return 1;
error:
Py_XDECREF(raw_line);
Py_XDECREF(line);
return 0;
}
static int
tok_underflow_string(struct tok_state *tok) {
char *end = strchr(tok->inp, '\n');
@ -1195,6 +1290,38 @@ tok_underflow_file(struct tok_state *tok) {
return tok->done == E_OK;
}
static int
tok_underflow_readline(struct tok_state* tok) {
assert(tok->decoding_state == STATE_NORMAL);
assert(tok->fp == NULL && tok->input == NULL && tok->decoding_readline == NULL);
if (tok->start == NULL && !INSIDE_FSTRING(tok)) {
tok->cur = tok->inp = tok->buf;
}
if (!tok_readline_string(tok)) {
return 0;
}
if (tok->inp == tok->cur) {
tok->done = E_EOF;
return 0;
}
if (tok->inp[-1] != '\n') {
assert(tok->inp + 1 < tok->end);
/* Last line does not end in \n, fake one */
*tok->inp++ = '\n';
*tok->inp = '\0';
}
ADVANCE_LINENO();
/* The default encoding is UTF-8, so make sure we don't have any
non-UTF-8 sequences in it. */
if (!tok->encoding && !ensure_utf8(tok->cur, tok)) {
error_ret(tok);
return 0;
}
assert(tok->done == E_OK);
return tok->done == E_OK;
}
#if defined(Py_DEBUG)
static void
print_escape(FILE *f, const char *s, Py_ssize_t size)
@ -1238,7 +1365,10 @@ tok_nextc(struct tok_state *tok)
if (tok->done != E_OK) {
return EOF;
}
if (tok->fp == NULL) {
if (tok->readline) {
rc = tok_underflow_readline(tok);
}
else if (tok->fp == NULL) {
rc = tok_underflow_string(tok);
}
else if (tok->prompt != NULL) {

2
Parser/tokenizer.h
View file

@ -109,6 +109,7 @@ struct tok_state {
expression (cf. issue 16806) */
PyObject *decoding_readline; /* open(...).readline */
PyObject *decoding_buffer;
PyObject *readline; /* readline() function */
const char* enc; /* Encoding for the current str. */
char* str; /* Source string being tokenized (if tokenizing from a string)*/
char* input; /* Tokenizer's newline translated copy of the string. */
@ -137,6 +138,7 @@ struct tok_state {
extern struct tok_state *_PyTokenizer_FromString(const char *, int, int);
extern struct tok_state *_PyTokenizer_FromUTF8(const char *, int, int);
extern struct tok_state *_PyTokenizer_FromReadline(PyObject*, const char*, int, int);
extern struct tok_state *_PyTokenizer_FromFile(FILE *, const char*,
const char *, const char *);
extern void _PyTokenizer_Free(struct tok_state *);

12
Python/Python-tokenize.c
View file

@ -37,15 +37,17 @@ typedef struct
@classmethod
_tokenizer.tokenizeriter.__new__ as tokenizeriter_new
source: str
readline: object
/
*
extra_tokens: bool
encoding: str(c_default="NULL") = 'utf-8'
[clinic start generated code]*/
static PyObject *
tokenizeriter_new_impl(PyTypeObject *type, const char *source,
int extra_tokens)
/*[clinic end generated code: output=f6f9d8b4beec8106 input=90dc5b6a5df180c2]*/
tokenizeriter_new_impl(PyTypeObject *type, PyObject *readline,
int extra_tokens, const char *encoding)
/*[clinic end generated code: output=7501a1211683ce16 input=f7dddf8a613ae8bd]*/
{
tokenizeriterobject *self = (tokenizeriterobject *)type->tp_alloc(type, 0);
if (self == NULL) {
@ -55,7 +57,7 @@ tokenizeriter_new_impl(PyTypeObject *type, const char *source,
if (filename == NULL) {
return NULL;
}
self->tok = _PyTokenizer_FromUTF8(source, 1, 1);
self->tok = _PyTokenizer_FromReadline(readline, encoding, 1, 1);
if (self->tok == NULL) {
Py_DECREF(filename);
return NULL;

49
Python/clinic/Python-tokenize.c.h generated
View file

@ -9,8 +9,8 @@ preserve
static PyObject *
tokenizeriter_new_impl(PyTypeObject *type, const char *source,
int extra_tokens);
tokenizeriter_new_impl(PyTypeObject *type, PyObject *readline,
int extra_tokens, const char *encoding);
static PyObject *
tokenizeriter_new(PyTypeObject *type, PyObject *args, PyObject *kwargs)
@ -25,7 +25,7 @@ tokenizeriter_new(PyTypeObject *type, PyObject *args, PyObject *kwargs)
PyObject *ob_item[NUM_KEYWORDS];
} _kwtuple = {
.ob_base = PyVarObject_HEAD_INIT(&PyTuple_Type, NUM_KEYWORDS)
.ob_item = { &_Py_ID(source), &_Py_ID(extra_tokens), },
.ob_item = { &_Py_ID(extra_tokens), &_Py_ID(encoding), },
};
#undef NUM_KEYWORDS
#define KWTUPLE (&_kwtuple.ob_base.ob_base)
@ -34,43 +34,50 @@ tokenizeriter_new(PyTypeObject *type, PyObject *args, PyObject *kwargs)
# define KWTUPLE NULL
#endif // !Py_BUILD_CORE
static const char * const _keywords[] = {"source", "extra_tokens", NULL};
static const char * const _keywords[] = {"", "extra_tokens", "encoding", NULL};
static _PyArg_Parser _parser = {
.keywords = _keywords,
.fname = "tokenizeriter",
.kwtuple = KWTUPLE,
};
#undef KWTUPLE
PyObject *argsbuf[2];
PyObject *argsbuf[3];
PyObject * const *fastargs;
Py_ssize_t nargs = PyTuple_GET_SIZE(args);
const char *source;
Py_ssize_t noptargs = nargs + (kwargs ? PyDict_GET_SIZE(kwargs) : 0) - 2;
PyObject *readline;
int extra_tokens;
const char *encoding = NULL;
fastargs = _PyArg_UnpackKeywords(_PyTuple_CAST(args)->ob_item, nargs, kwargs, NULL, &_parser, 1, 1, 1, argsbuf);
if (!fastargs) {
goto exit;
}
if (!PyUnicode_Check(fastargs[0])) {
_PyArg_BadArgument("tokenizeriter", "argument 'source'", "str", fastargs[0]);
goto exit;
}
Py_ssize_t source_length;
source = PyUnicode_AsUTF8AndSize(fastargs[0], &source_length);
if (source == NULL) {
goto exit;
}
if (strlen(source) != (size_t)source_length) {
PyErr_SetString(PyExc_ValueError, "embedded null character");
goto exit;
}
readline = fastargs[0];
extra_tokens = PyObject_IsTrue(fastargs[1]);
if (extra_tokens < 0) {
goto exit;
}
return_value = tokenizeriter_new_impl(type, source, extra_tokens);
if (!noptargs) {
goto skip_optional_kwonly;
}
if (!PyUnicode_Check(fastargs[2])) {
_PyArg_BadArgument("tokenizeriter", "argument 'encoding'", "str", fastargs[2]);
goto exit;
}
Py_ssize_t encoding_length;
encoding = PyUnicode_AsUTF8AndSize(fastargs[2], &encoding_length);
if (encoding == NULL) {
goto exit;
}
if (strlen(encoding) != (size_t)encoding_length) {
PyErr_SetString(PyExc_ValueError, "embedded null character");
goto exit;
}
skip_optional_kwonly:
return_value = tokenizeriter_new_impl(type, readline, extra_tokens, encoding);
exit:
return return_value;
}
/*[clinic end generated code: output=940b564c67f6e0e2 input=a9049054013a1b77]*/
/*[clinic end generated code: output=48be65a2808bdfa6 input=a9049054013a1b77]*/