| 1 |
import importlib.util |
| 2 |
import io |
| 3 |
import os |
| 4 |
import pathlib |
| 5 |
import sys |
| 6 |
import textwrap |
| 7 |
import token |
| 8 |
import tokenize |
| 9 |
from typing import IO, Any, Final, cast |
| 10 |
|
| 11 |
from pegen.build import compile_c_extension |
| 12 |
from pegen.c_generator import CParserGenerator |
| 13 |
from pegen.grammar import Grammar |
| 14 |
from pegen.grammar_parser import GeneratedParser as GrammarParser |
| 15 |
from pegen.parser import Parser |
| 16 |
from pegen.python_generator import PythonParserGenerator |
| 17 |
from pegen.tokenizer import Tokenizer |
| 18 |
|
| 19 |
ALL_TOKENS = token.tok_name |
| 20 |
EXACT_TOKENS = token.EXACT_TOKEN_TYPES |
| 21 |
NON_EXACT_TOKENS = { |
| 22 |
name for index, name in token.tok_name.items() if index not in EXACT_TOKENS.values() |
| 23 |
} |
| 24 |
|
| 25 |
|
| 26 |
def generate_parser(grammar: Grammar) -> type[Parser]: |
| 27 |
# Generate a parser. |
| 28 |
out = io.StringIO() |
| 29 |
genr = PythonParserGenerator(grammar, out) |
| 30 |
genr.generate("<string>") |
| 31 |
|
| 32 |
# Load the generated parser class. |
| 33 |
ns: dict[str, Any] = {} |
| 34 |
exec(out.getvalue(), ns) |
| 35 |
return ns["GeneratedParser"] |
| 36 |
|
| 37 |
|
| 38 |
def run_parser(file: IO[bytes], parser_class: type[Parser], *, verbose: bool = False) -> Any: |
| 39 |
# Run a parser on a file (stream). |
| 40 |
tokenizer = Tokenizer(tokenize.generate_tokens(file.readline)) # type: ignore[arg-type] # typeshed issue #3515 |
| 41 |
parser = parser_class(tokenizer, verbose=verbose) |
| 42 |
result = parser.start() |
| 43 |
if result is None: |
| 44 |
raise parser.make_syntax_error("invalid syntax") |
| 45 |
return result |
| 46 |
|
| 47 |
|
| 48 |
def parse_string( |
| 49 |
source: str, parser_class: type[Parser], *, dedent: bool = True, verbose: bool = False |
| 50 |
) -> Any: |
| 51 |
# Run the parser on a string. |
| 52 |
if dedent: |
| 53 |
source = textwrap.dedent(source) |
| 54 |
file = io.StringIO(source) |
| 55 |
return run_parser(file, parser_class, verbose=verbose) # type: ignore[arg-type] # typeshed issue #3515 |
| 56 |
|
| 57 |
|
| 58 |
def make_parser(source: str) -> type[Parser]: |
| 59 |
# Combine parse_string() and generate_parser(). |
| 60 |
grammar = parse_string(source, GrammarParser) |
| 61 |
return generate_parser(grammar) |
| 62 |
|
| 63 |
|
| 64 |
def import_file(full_name: str, path: str) -> Any: |
| 65 |
"""Import a python module from a path""" |
| 66 |
|
| 67 |
spec = importlib.util.spec_from_file_location(full_name, path) |
| 68 |
assert spec is not None |
| 69 |
mod = importlib.util.module_from_spec(spec) |
| 70 |
|
| 71 |
# We assume this is not None and has an exec_module() method. |
| 72 |
# See https://docs.python.org/3/reference/import.html?highlight=exec_module#loading |
| 73 |
loader = cast(Any, spec.loader) |
| 74 |
loader.exec_module(mod) |
| 75 |
return mod |
| 76 |
|
| 77 |
|
| 78 |
def generate_c_parser_source(grammar: Grammar) -> str: |
| 79 |
out = io.StringIO() |
| 80 |
genr = CParserGenerator(grammar, ALL_TOKENS, EXACT_TOKENS, NON_EXACT_TOKENS, out) |
| 81 |
genr.generate("<string>") |
| 82 |
return out.getvalue() |
| 83 |
|
| 84 |
|
| 85 |
def generate_parser_c_extension( |
| 86 |
grammar: Grammar, |
| 87 |
path: pathlib.PurePath, |
| 88 |
debug: bool = False, |
| 89 |
library_dir: str | None = None, |
| 90 |
) -> Any: |
| 91 |
"""Generate a parser c extension for the given grammar in the given path |
| 92 |
|
| 93 |
Returns a module object with a parse_string() method. |
| 94 |
TODO: express that using a Protocol. |
| 95 |
""" |
| 96 |
# Make sure that the working directory is empty: reusing non-empty temporary |
| 97 |
# directories when generating extensions can lead to segmentation faults. |
| 98 |
# Check issue #95 (https://github.com/gvanrossum/pegen/issues/95) for more |
| 99 |
# context. |
| 100 |
assert not os.listdir(path) |
| 101 |
source = path / "parse.c" |
| 102 |
with open(source, "w", encoding="utf-8") as file: |
| 103 |
genr = CParserGenerator( |
| 104 |
grammar, ALL_TOKENS, EXACT_TOKENS, NON_EXACT_TOKENS, file, debug=debug |
| 105 |
) |
| 106 |
genr.generate("parse.c") |
| 107 |
compile_c_extension( |
| 108 |
str(source), |
| 109 |
build_dir=str(path), |
| 110 |
# Significant test_peg_generator speedups |
| 111 |
disable_optimization=True, |
| 112 |
library_dir=library_dir, |
| 113 |
) |
| 114 |
|
| 115 |
|
| 116 |
def print_memstats() -> bool: |
| 117 |
MiB: Final = 2**20 |
| 118 |
try: |
| 119 |
import psutil |
| 120 |
except ImportError: |
| 121 |
return False |
| 122 |
print("Memory stats:") |
| 123 |
process = psutil.Process() |
| 124 |
meminfo = process.memory_info() |
| 125 |
res = {} |
| 126 |
res["rss"] = meminfo.rss / MiB |
| 127 |
res["vms"] = meminfo.vms / MiB |
| 128 |
if sys.platform == "win32": |
| 129 |
res["maxrss"] = meminfo.peak_wset / MiB |
| 130 |
else: |
| 131 |
# See https://stackoverflow.com/questions/938733/total-memory-used-by-python-process |
| 132 |
import resource # Since it doesn't exist on Windows. |
| 133 |
|
| 134 |
rusage = resource.getrusage(resource.RUSAGE_SELF) |
| 135 |
if sys.platform == "darwin": |
| 136 |
factor = 1 |
| 137 |
else: |
| 138 |
factor = 1024 # Linux |
| 139 |
res["maxrss"] = rusage.ru_maxrss * factor / MiB |
| 140 |
for key, value in res.items(): |
| 141 |
print(f" {key:12.12s}: {value:10.0f} MiB") |
| 142 |
return True |