Fix parsing a numeric literal immediately (without spaces) followed by
"not in" keywords, like in "1not in x". Now the parser only emits
a warning, not a syntax error.
We're no longer using _Py_IDENTIFIER() (or _Py_static_string()) in any core CPython code. It is still used in a number of non-builtin stdlib modules.
The replacement is: PyUnicodeObject (not pointer) fields under _PyRuntimeState, statically initialized as part of _PyRuntime. A new _Py_GET_GLOBAL_IDENTIFIER() macro facilitates lookup of the fields (along with _Py_GET_GLOBAL_STRING() for non-identifier strings).
https://bugs.python.org/issue46541#msg411799 explains the rationale for this change.
The core of the change is in:
* (new) Include/internal/pycore_global_strings.h - the declarations for the global strings, along with the macros
* Include/internal/pycore_runtime_init.h - added the static initializers for the global strings
* Include/internal/pycore_global_objects.h - where the struct in pycore_global_strings.h is hooked into _PyRuntimeState
* Tools/scripts/generate_global_objects.py - added generation of the global string declarations and static initializers
I've also added a --check flag to generate_global_objects.py (along with make check-global-objects) to check for unused global strings. That check is added to the PR CI config.
The remainder of this change updates the core code to use _Py_GET_GLOBAL_IDENTIFIER() instead of _Py_IDENTIFIER() and the related _Py*Id functions (likewise for _Py_GET_GLOBAL_STRING() instead of _Py_static_string()). This includes adding a few functions where there wasn't already an alternative to _Py*Id(), replacing the _Py_Identifier * parameter with PyObject *.
The following are not changed (yet):
* stop using _Py_IDENTIFIER() in the stdlib modules
* (maybe) get rid of _Py_IDENTIFIER(), etc. entirely -- this may not be doable as at least one package on PyPI using this (private) API
* (maybe) intern the strings during runtime init
https://bugs.python.org/issue46541
@pablogsal, sorry i failed to rebase to main, so i recreated https://github.com/python/cpython/pull/22190#issuecomment-1024633392
> PyRun_InteractiveOne\*() functions allow to explicitily set fd instead of stdin.
but stdin was hardcoded in readline call.
> This patch does not fix target file for prompt unlike original bpo one : prompt fd is unrelated to tokenizer source which could be read only. It is more of a bugfix regarding the docs : actual documentation say "prompt the user" so one would expect prompt to go on stdout not a file for both PyRun_InteractiveOne\*() and PyRun_InteractiveLoop\*().
Automerge-Triggered-By: GH:pablogsal
There are two errors that this commit fixes:
* The parser was not correctly computing the offset and the string
source for E_LINECONT errors due to the incorrect usage of strtok().
* The parser was not correctly unwinding the call stack when a tokenizer
exception happened in rules involving optionals ('?', [...]) as we
always make them return valid results by using the comma operator. We
need to check first if we don't have an error before continuing.
They support now splitting escape sequences between input chunks.
Add the third parameter "final" in codecs.unicode_escape_decode().
It is True by default to match the former behavior.
* Move _PyObject_CallNoArgs() to pycore_call.h (internal C API).
* _ssl, _sqlite and _testcapi extensions now call the public
PyObject_CallNoArgs() function, rather than _PyObject_CallNoArgs().
* _lsprof extension is now built with Py_BUILD_CORE_MODULE macro
defined to get access to internal _PyObject_CallNoArgs().
Fix typo in the private _PyObject_CallNoArg() function name: rename
it to _PyObject_CallNoArgs() to be consistent with the public
function PyObject_CallNoArgs().
The traceback.c and traceback.py mechanisms now utilize the newly added code.co_positions and PyCode_Addr2Location
to print carets on the specific expressions involved in a traceback.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
Co-authored-by: Ammar Askar <ammar@ammaraskar.com>
Co-authored-by: Batuhan Taskaya <batuhanosmantaskaya@gmail.com>
This is something I noticed while (now discontinued) experimenting
with the idea of annotating operators with location information. Unfortunately
without this addition, adding any `attributes` to stuff like `unaryop`
doesn't change anything since the code assumes they are singletons and
caches all instances. This patch fixes this assumption with including
the attributes as well as constructor fields.
ASDL Generator was lack of proper annotation related to generated
module. This patch implements a MetadataVisitor that produces a
metadata object to pass to other visitors that are visiting that
same module. For the inital patch, it dynamically retrieves int
sequences (like cmpop), that was previously hardcoded. It offers
an interface that is easy to extend.
Emit a deprecation warning if the numeric literal is immediately followed by
one of keywords: and, else, for, if, in, is, or. Raise a syntax error with
more informative message if it is immediately followed by other keyword or
identifier.
Automerge-Triggered-By: GH:pablogsal
When compiling an AST object with a direct / indirect reference
cycles, on the conversion phase because of exceeding amount of
calls, a segfault was raised. This patch adds recursion guards to
places for preventing user inputs to not to crash AST but instead
raise a RecursionError.
When the parser does a second pass to check for errors, these rules can
have some small side-effects as they may advance the parser more than
the point reached in the first pass. This can cause the tokenizer to ask
for extra tokens in interactive mode causing the tokenizer to show the
prompt instead of failing instantly.
To avoid this, add a new mode to the tokenizer that is activated in the
second pass and deactivates asking for new tokens when the interactive
line is finished. As the parsing should have reached the last line in
the first pass, the second pass should not need to ask for more tokens.
The invalid assignment rules are very delicate since the parser can
easily raise an invalid assignment when a keyword argument is provided.
As they are very deep into the grammar tree, is very difficult to
specify in which contexts these rules can be used and in which don't.
For that, we need to use a different version of the rule that doesn't do
error checking in those situations where we don't want the rule to raise
(keyword arguments and generator expressions).
We also need to check if we are in left-recursive rule, as those can try
to eagerly advance the parser even if the parse will fail at the end of
the expression. Failing to do this allows the parser to start parsing a
call as a tuple and incorrectly identify a keyword argument as an
invalid assignment, before it realizes that it was not a tuple after all.
This works by not caching the handle and instead getting the handle from
the file descriptor each time, so that if the actual handle changes by
fd redirection closing/opening the console handle beneath our feet, we
will keep working correctly.
To improve the user experience understanding what part of the error messages associated with SyntaxErrors is wrong, we can highlight the whole error range and not only place the caret at the first character. In this way:
>>> foo(x, z for z in range(10), t, w)
File "<stdin>", line 1
foo(x, z for z in range(10), t, w)
^
SyntaxError: Generator expression must be parenthesized
becomes
>>> foo(x, z for z in range(10), t, w)
File "<stdin>", line 1
foo(x, z for z in range(10), t, w)
^^^^^^^^^^^^^^^^^^^^
SyntaxError: Generator expression must be parenthesized
* Add source location attributes to alias.
* Move alias star construction to pegen helper.
Co-authored-by: blurb-it[bot] <43283697+blurb-it[bot]@users.noreply.github.com>
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
Rename AST functions of pycore_ast.h to use the "_PyAST_" prefix.
Remove macros creating aliases without prefix. For example, Module()
becomes _PyAST_Module(). Update Grammar/python.gram to use
_PyAST_xxx() functions.
* pycore_ast.h no longer defines the Yield macro.
* Fix a compiler warning on Windows: "warning C4005: 'Yield': macro
redefinition".
* Python-ast.c now defines directly functions with their real
_Py_xxx() name, rather than xxx().
* Remove "#undef Yield" in C files including pycore_ast.h.
Remove the pyarena.h header file with functions:
* PyArena_New()
* PyArena_Free()
* PyArena_Malloc()
* PyArena_AddPyObject()
These functions were undocumented, excluded from the limited C API,
and were only used internally by the compiler.
Add pycore_pyarena.h header. Rename functions:
* PyArena_New() => _PyArena_New()
* PyArena_Free() => _PyArena_Free()
* PyArena_Malloc() => _PyArena_Malloc()
* PyArena_AddPyObject() => _PyArena_AddPyObject()
Remove parser functions using the "struct _mod" type, because the
AST C API was removed:
* PyParser_ASTFromFile()
* PyParser_ASTFromFileObject()
* PyParser_ASTFromFilename()
* PyParser_ASTFromString()
* PyParser_ASTFromStringObject()
These functions were undocumented and excluded from the limited C
API.
Add pycore_parser.h internal header file. Rename functions:
* PyParser_ASTFromFileObject() => _PyParser_ASTFromFile()
* PyParser_ASTFromStringObject() => _PyParser_ASTFromString()
These functions are no longer exported (replace PyAPI_FUNC() with
extern).
Remove also _PyPegen_run_parser_from_file() function. Update
test_peg_generator to use _PyPegen_run_parser_from_file_pointer()
instead.
These functions were undocumented and excluded from the limited C
API.
Most names defined by these header files were not prefixed by "Py"
and so could create names conflicts. For example, Python-ast.h
defined a "Yield" macro which was conflict with the "Yield" name used
by the Windows <winbase.h> header.
Use the Python ast module instead.
* Move Include/asdl.h to Include/internal/pycore_asdl.h.
* Move Include/Python-ast.h to Include/internal/pycore_ast.h.
* Remove ast.h header file.
* pycore_symtable.h no longer includes Python-ast.h.
Remove the PyAST_Validate() function. It is no longer possible to
build a AST object (mod_ty type) with the public C API. The function
was already excluded from the limited C API (PEP 384).
Rename PyAST_Validate() function to _PyAST_Validate(), move it to the
internal C API, and don't export it anymore (replace PyAPI_FUNC with
extern).
The function was added in bpo-12575 by
the commit 832bfe2ebd.
test_peg_generator now defines _Py_TEST_PEGEN macro when building C
code to not call PyAST_Validate() in Parser/pegen.c. Moreover, it
defines Py_BUILD_CORE_MODULE macro to get access to the internal
C API.
Remove "global_ast_state" from Python-ast.c when it's built by
test_peg_generator: always get the AST state from the current interpreter.
Include/{odictobject.h,parser_interface.h,picklebufobject.h,pydebug.h,pyfpe.h}
into Include/cpython/.
Parser: peg_api: include Python.h instead of parser_interface.h.
* Add to the peg generator a new directive ('&&') that allows to expect
a token and hard fail the parsing if the token is not found. This
allows to quickly emmit syntax errors for missing tokens.
* Use the new grammar element to hard-fail if the ':' is missing before
suites.
When trying to extract the error line for the error message there
are two distinct cases:
1. The input comes from a file, which means that we can extract the
error line by using `PyErr_ProgramTextObject` and which we already
do.
2. The input does not come from a file, at which point we need to get
the source code from the tokenizer:
* If the tokenizer's current line number is the same with the line
of the error, we get the line from `tok->buf` and we're ready.
* Else, we can extract the error line from the source code in the
following two ways:
* If the input comes from a string we have all the input
in `tok->str` and we can extract the error line from it.
* If the input comes from stdin, i.e. the interactive prompt, we
do not have access to the previous line. That's why a new
field `tok->stdin_content` is added which holds the whole input for the
current (multiline) statement or expression. We can then extract the
error line from `tok->stdin_content` like we do in the string case above.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
This is only there so that alternative implementations written in statically-typed languages can use this grammar without
having type errors in the way.
Automerge-Triggered-By: GH:lysnikolaou
No longer use deprecated aliases to functions:
* Replace PyMem_MALLOC() with PyMem_Malloc()
* Replace PyMem_REALLOC() with PyMem_Realloc()
* Replace PyMem_FREE() with PyMem_Free()
* Replace PyMem_Del() with PyMem_Free()
* Replace PyMem_DEL() with PyMem_Free()
Modify also the PyMem_DEL() macro to use directly PyMem_Free().
Currently walruses are not allowerd in set literals and set comprehensions:
>>> {y := 4, 4**2, 3**3}
File "<stdin>", line 1
{y := 4, 4**2, 3**3}
^
SyntaxError: invalid syntax
but they should be allowed as well per PEP 572
Call _PyAST_Fini() on all interpreters, not only on the main
interpreter. Also, call it ealier to fix a reference leak.
Python types contain a reference to themselves in in their
PyTypeObject.tp_mro member. _PyAST_Fini() must called before the last
GC collection to destroy AST types.
_PyInterpreterState_Clear() now calls _PyAST_Fini(). It now also
calls _PyWarnings_Fini() on subinterpeters, not only on the main
interpreter.
Add an assertion in AST init_types() to ensure that the _ast module
is no longer used after _PyAST_Fini() has been called.
The ast module internal state is now per interpreter.
* Rename "astmodulestate" to "struct ast_state"
* Add pycore_ast.h internal header: the ast_state structure is now
declared in pycore_ast.h.
* Add PyInterpreterState.ast (struct ast_state)
* Remove get_ast_state()
* Rename get_global_ast_state() to get_ast_state()
* PyAST_obj2mod() now handles get_ast_state() failures
Left-recursive rules need to check for errors explicitly, since
even if the rule returns NULL, the parsing might continue and lead
to long-distance failures.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
* Implement running the parser a second time for the errors messages
The first parser run is only responsible for detecting whether
there is a `SyntaxError` or not. If there isn't the AST gets returned.
Otherwise, the parser is run a second time with all the `invalid_*`
rules enabled so that all the customized error messages get produced.
- Use the proper asdl sequence when creating empty arguments
- Remove reduntant casts (thanks to new typed asdl_sequences)
- Remove MarshalPrototypeVisitor and some utilities from asdl generator
- Fix the header of `Python/ast.c` (kept from pgen times)
Automerge-Triggered-By: @pablogsal
* Add new capability to the PEG parser to type variable assignments. For instance:
```
| a[asdl_stmt_seq*]=';'.small_stmt+ [';'] NEWLINE { a }
```
* Add new sequence types from the asdl definition (automatically generated)
* Make `asdl_seq` type a generic aliasing pointer type.
* Create a new `asdl_generic_seq` for the generic case using `void*`.
* The old `asdl_seq_GET`/`ast_seq_SET` macros now are typed.
* New `asdl_seq_GET_UNTYPED`/`ast_seq_SET_UNTYPED` macros for dealing with generic sequences.
* Changes all possible `asdl_seq` types to use specific versions everywhere.
Partially revert commit ac46eb4ad6:
"bpo-38113: Update the Python-ast.c generator to PEP384 (gh-15957)".
Using a module state per module instance is causing subtle practical
problems.
For example, the Mercurial project replaces the __import__() function
to implement lazy import, whereas Python expected that "import _ast"
always return a fully initialized _ast module.
Add _PyAST_Fini() to clear the state at exit.
The _ast module has no state (set _astmodule.m_size to 0). Remove
astmodule_traverse(), astmodule_clear() and astmodule_free()
functions.
This program can segfault the parser by stack overflow:
```
import ast
code = "f(" + ",".join(['a' for _ in range(100000)]) + ")"
print("Ready!")
ast.parse(code)
```
the reason is that the rule for arguments has a simple recursion when collecting args:
args[expr_ty]:
[...]
| a=named_expression b=[',' c=args { c }] {
[...] }
This consolidates the handling of my_fgets return values, so that interrupts are always handled, even if they come after EOF.
I believe PyOS_StdioReadline is still buggy in that I/O errors will not result in a proper Python exception being set. However, that is a separate issue.
This will improve the debug experience if something fails in the produced AST. Previously, errors in the produced AST can be felt much later like in the garbage collector or the compiler, making debugging them much more difficult.
GCC says
```
../cpython/Parser/string_parser.c: In function ‘fstring_find_expr’:
../cpython/Parser/string_parser.c:404:93: warning: ‘cols’ may be used uninitialized in this function [-Wmaybe-uninitialized]
404 | p2->starting_col_offset = p->tok->first_lineno == p->tok->lineno ? t->col_offset + cols : cols;
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~
../cpython/Parser/string_parser.c:384:16: note: ‘cols’ was declared here
384 | int lines, cols;
| ^~~~
../cpython/Parser/string_parser.c:403:45: warning: ‘lines’ may be used uninitialized in this function [-Wmaybe-uninitialized]
403 | p2->starting_lineno = t->lineno + lines - 1;
| ~~~~~~~~~~~~~~~~~~^~~
../cpython/Parser/string_parser.c:384:9: note: ‘lines’ was declared here
384 | int lines, cols;
| ^~~~~
```
and, indeed, if `PyBytes_AsString` somehow fails, lines & cols will not be initialized.
Fix a crash in the _ast module: it can no longer be loaded more than
once. It now uses a global state rather than a module state.
* Move _ast module state: use a global state instead.
* Set _astmodule.m_size to -1, so the extension cannot be loaded more
than once.
Rework asdl_c.py to pass the module state to functions in
Python-ast.c, instead of using astmodulestate_global.
Handle also PyState_AddModule() failure in init_types().
This commit changes the parsing of f-string expressions with the new parser. The parser gets pre-fed with the location of the expression itself (not the f-string, which was what we were doing before). This allows us to completely skip the shifting of the AST nodes after the parsing is completed.
`GET_INVALID_TARGET` might unexpectedly return `NULL`, which if not
caught will cause a SEGFAULT. Therefore, this commit introduces a new
inline function `RAISE_SYNTAX_ERROR_INVALID_TARGET` that always
checks for `GET_INVALID_TARGET` returning NULL and can be used in
the grammar, replacing the long C ternary operation used till now.
The following error messages get produced:
- `cannot delete ...` for invalid `del` targets
- `... is an illegal 'for' target` for invalid targets in for
statements
- `... is an illegal 'with' target` for invalid targets in
with statements
Additionally, a few `cut`s were added in various places before the
invocation of the `invalid_*` rule, in order to speed things
up.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
On Windows, #include "pyerrors.h" no longer defines "snprintf" and
"vsnprintf" macros.
PyOS_snprintf() and PyOS_vsnprintf() should be used to get portable
behavior.
Replace snprintf() calls with PyOS_snprintf() and replace vsnprintf()
calls with PyOS_vsnprintf().
This commit removes the old parser, the deprecated parser module, the old parser compatibility flags and environment variables and all associated support code and documentation.
It no longer serves a purpose (there's only one parser) and having "new" in any name will eventually look odd. Also, it impinges on a potential sub-namespace, `__new_...__`.
A line with only a line continuation character should be considered
a blank line at tokenizer level so that only a single NEWLINE token
gets emitted. The old parser was working around the issue, but the
new parser threw a `SyntaxError` for valid input. For example,
an empty line following a line continuation character was interpreted
as a `SyntaxError`.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
my_fgets() now calls _PyOS_InterruptOccurred(tstate) to check for
pending signals, rather calling PyOS_InterruptOccurred().
my_fgets() is called with the GIL released, whereas
PyOS_InterruptOccurred() must be called with the GIL held.
test_repl: use text=True and avoid SuppressCrashReport in
test_multiline_string_parsing().
Fix my_fgets() on Windows: fgets(fp) does crash if fileno(fp) is closed.
Fix GIL usage in PyOS_Readline(): lock the GIL to set an exception.
Pass tstate to my_fgets() and _PyOS_WindowsConsoleReadline(). Cleanup
these functions.
These are like keywords but they only work in context; they are not reserved except when there is an exact match.
This would enable things like match statements without reserving `match` (which would be bad for the `re.match()` function and probably lots of other places).
Automerge-Triggered-By: @gvanrossum
When a `SyntaxError` in the expression part of a fstring is found,
the filename attribute of the `SyntaxError` is always `<fstring>`.
With this commit, it gets changed to always have the name of the file
the fstring resides in.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
The error message, generated for a non-parenthesized generator expression
in function calls, was still the generic `invalid syntax`, when the generator expression wasn't appearing as the first argument in the call. With this patch, even on input like `f(a, b, c for c in d, e)`, the correct error message gets produced.
- Switch from getopt to argparse.
- Removed the limitation of not being able to produce both C and H simultaneously.
This will make it run faster since it parses the asdl definition once and uses the generated tree to generate both the header and the C source.
The following improvements are implemented in this commit:
- `p->error_indicator` is set, in case malloc or realloc fail.
- Avoid memory leaks in the case that realloc fails.
- Call `PyErr_NoMemory()` instead of `PyErr_Format()`, because it requires no memory.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
This commit fixes the new parser to disallow invalid targets in the
following scenarios:
- Augmented assignments must only accept a single target (Name,
Attribute or Subscript), but no tuples or lists.
- `except` clauses should only accept a single `Name` as a target.
Co-authored-by: Pablo Galindo <Pablogsal@gmail.com>
This commit fixes SyntaxError locations when the caret is not displayed,
by doing the following:
- `col_number` always gets set to the location of the offending
node/expr. When no caret is to be displayed, this gets achieved
by setting the object holding the error line to None.
- Introduce a new function `_PyPegen_raise_error_known_location`,
which can be called, when an arbitrary `lineno`/`col_offset`
needs to be passed. This function then gets used in the grammar
(through some new macros and inline functions) so that SyntaxError
locations of the new parser match that of the old.
With the new parser, the error message contains always the trailing
newlines, causing the comparison of the repr of the error messages
in codeop to fail. This commit makes the new parser mirror the old parser's
behaviour regarding trailing newlines.
This is for the C generator:
- Disallow rule and variable names starting with `_`
- Rename most local variable names generated by the parser to start with `_`
Exceptions:
- Renaming `p` to `_p` will be a separate PR
- There are still some names that might clash, e.g.
- anything starting with `Py`
- C reserved words (`if` etc.)
- Macros like `EXTRA` and `CHECK`
When parsing something like `f(g()=2)`, where the name of a default arg
is not a NAME, but an arbitrary expression, a specialised error message
is emitted.
When parsing a string with an invalid escape, the old parser used to
point to the beginning of the invalid string. This commit changes the new
parser to match that behaviour, since it's currently pointing to the
end of the string (or to be more precise, to the beginning of the next
token).
Due to backwards compatibility concerns regarding keywords immediately followed by a string without whitespace between them (like in `bg="#d00" if clear else"#fca"`) will fail to parse,
commit 41d5b94af4 has to be reverted.
