| 1 |
#! /usr/bin/env python3 |
| 2 |
# This script generates token related files from Grammar/Tokens: |
| 3 |
# |
| 4 |
# make_rst: |
| 5 |
# Doc/library/token-list.inc |
| 6 |
# Doc/library/token.rst (checked, not generated) |
| 7 |
# make_h: |
| 8 |
# Include/token.h |
| 9 |
# make_c: |
| 10 |
# Parser/token.c |
| 11 |
# make_py: |
| 12 |
# Lib/token.py |
| 13 |
|
| 14 |
import re |
| 15 |
|
| 16 |
SCRIPT_NAME = 'Tools/build/generate_token.py' |
| 17 |
AUTO_GENERATED_BY_SCRIPT = f'Auto-generated by {SCRIPT_NAME}' |
| 18 |
NT_OFFSET = 256 |
| 19 |
|
| 20 |
def load_tokens(path): |
| 21 |
tok_names = [] |
| 22 |
string_to_tok = {} |
| 23 |
ERRORTOKEN = None |
| 24 |
with open(path) as fp: |
| 25 |
for line in fp: |
| 26 |
line = line.strip() |
| 27 |
# strip comments |
| 28 |
i = line.find('#') |
| 29 |
if i >= 0: |
| 30 |
line = line[:i].strip() |
| 31 |
if not line: |
| 32 |
continue |
| 33 |
fields = line.split() |
| 34 |
name = fields[0] |
| 35 |
value = len(tok_names) |
| 36 |
if name == 'ERRORTOKEN': |
| 37 |
ERRORTOKEN = value |
| 38 |
string = fields[1] if len(fields) > 1 else None |
| 39 |
if string: |
| 40 |
string = eval(string) |
| 41 |
string_to_tok[string] = value |
| 42 |
tok_names.append(name) |
| 43 |
return tok_names, ERRORTOKEN, string_to_tok |
| 44 |
|
| 45 |
|
| 46 |
def update_file(file, content): |
| 47 |
try: |
| 48 |
with open(file) as fobj: |
| 49 |
if fobj.read() == content: |
| 50 |
return False |
| 51 |
except (OSError, ValueError): |
| 52 |
pass |
| 53 |
with open(file, 'w') as fobj: |
| 54 |
fobj.write(content) |
| 55 |
return True |
| 56 |
|
| 57 |
|
| 58 |
token_h_template = f"""\ |
| 59 |
// {AUTO_GENERATED_BY_SCRIPT} |
| 60 |
""" |
| 61 |
token_h_template += """\ |
| 62 |
|
| 63 |
/* Token types */ |
| 64 |
#ifndef Py_INTERNAL_TOKEN_H |
| 65 |
#define Py_INTERNAL_TOKEN_H |
| 66 |
#ifdef __cplusplus |
| 67 |
extern "C" { |
| 68 |
#endif |
| 69 |
|
| 70 |
#ifndef Py_BUILD_CORE |
| 71 |
# error "this header requires Py_BUILD_CORE define" |
| 72 |
#endif |
| 73 |
|
| 74 |
#undef TILDE /* Prevent clash of our definition with system macro. Ex AIX, ioctl.h */ |
| 75 |
|
| 76 |
%s\ |
| 77 |
#define N_TOKENS %d |
| 78 |
#define NT_OFFSET %d |
| 79 |
|
| 80 |
/* Special definitions for cooperation with parser */ |
| 81 |
|
| 82 |
#define ISTERMINAL(x) ((x) < NT_OFFSET) |
| 83 |
#define ISNONTERMINAL(x) ((x) >= NT_OFFSET) |
| 84 |
#define ISEOF(x) ((x) == ENDMARKER) |
| 85 |
#define ISWHITESPACE(x) ((x) == ENDMARKER || \\ |
| 86 |
(x) == NEWLINE || \\ |
| 87 |
(x) == INDENT || \\ |
| 88 |
(x) == DEDENT) |
| 89 |
#define ISSTRINGLIT(x) ((x) == STRING || \\ |
| 90 |
(x) == FSTRING_MIDDLE || \\ |
| 91 |
(x) == TSTRING_MIDDLE) |
| 92 |
|
| 93 |
|
| 94 |
// Export these 4 symbols for 'test_peg_generator' |
| 95 |
PyAPI_DATA(const char * const) _PyParser_TokenNames[]; /* Token names */ |
| 96 |
PyAPI_FUNC(int) _PyToken_OneChar(int); |
| 97 |
PyAPI_FUNC(int) _PyToken_TwoChars(int, int); |
| 98 |
PyAPI_FUNC(int) _PyToken_ThreeChars(int, int, int); |
| 99 |
|
| 100 |
#ifdef __cplusplus |
| 101 |
} |
| 102 |
#endif |
| 103 |
#endif // !Py_INTERNAL_TOKEN_H |
| 104 |
""" |
| 105 |
|
| 106 |
def make_h(infile, outfile='Include/internal/pycore_token.h'): |
| 107 |
tok_names, ERRORTOKEN, string_to_tok = load_tokens(infile) |
| 108 |
|
| 109 |
defines = [] |
| 110 |
for value, name in enumerate(tok_names[:ERRORTOKEN + 1]): |
| 111 |
defines.append("#define %-15s %d\n" % (name, value)) |
| 112 |
|
| 113 |
if update_file(outfile, token_h_template % ( |
| 114 |
''.join(defines), |
| 115 |
len(tok_names), |
| 116 |
NT_OFFSET |
| 117 |
)): |
| 118 |
print("%s regenerated from %s" % (outfile, infile)) |
| 119 |
|
| 120 |
|
| 121 |
token_c_template = f"""\ |
| 122 |
/* {AUTO_GENERATED_BY_SCRIPT} */ |
| 123 |
""" |
| 124 |
token_c_template += """\ |
| 125 |
|
| 126 |
#include "Python.h" |
| 127 |
#include "pycore_token.h" |
| 128 |
|
| 129 |
/* Token names */ |
| 130 |
|
| 131 |
const char * const _PyParser_TokenNames[] = { |
| 132 |
%s\ |
| 133 |
}; |
| 134 |
|
| 135 |
/* Return the token corresponding to a single character */ |
| 136 |
|
| 137 |
int |
| 138 |
_PyToken_OneChar(int c1) |
| 139 |
{ |
| 140 |
%s\ |
| 141 |
return OP; |
| 142 |
} |
| 143 |
|
| 144 |
int |
| 145 |
_PyToken_TwoChars(int c1, int c2) |
| 146 |
{ |
| 147 |
%s\ |
| 148 |
return OP; |
| 149 |
} |
| 150 |
|
| 151 |
int |
| 152 |
_PyToken_ThreeChars(int c1, int c2, int c3) |
| 153 |
{ |
| 154 |
%s\ |
| 155 |
return OP; |
| 156 |
} |
| 157 |
""" |
| 158 |
|
| 159 |
def generate_chars_to_token(mapping, n=1): |
| 160 |
result = [] |
| 161 |
write = result.append |
| 162 |
indent = ' ' * n |
| 163 |
write(indent) |
| 164 |
write('switch (c%d) {\n' % (n,)) |
| 165 |
for c in sorted(mapping): |
| 166 |
write(indent) |
| 167 |
value = mapping[c] |
| 168 |
if isinstance(value, dict): |
| 169 |
write("case '%s':\n" % (c,)) |
| 170 |
write(generate_chars_to_token(value, n + 1)) |
| 171 |
write(indent) |
| 172 |
write(' break;\n') |
| 173 |
else: |
| 174 |
write("case '%s': return %s;\n" % (c, value)) |
| 175 |
write(indent) |
| 176 |
write('}\n') |
| 177 |
return ''.join(result) |
| 178 |
|
| 179 |
def make_c(infile, outfile='Parser/token.c'): |
| 180 |
tok_names, ERRORTOKEN, string_to_tok = load_tokens(infile) |
| 181 |
string_to_tok['<>'] = string_to_tok['!='] |
| 182 |
chars_to_token = {} |
| 183 |
for string, value in string_to_tok.items(): |
| 184 |
assert 1 <= len(string) <= 3 |
| 185 |
name = tok_names[value] |
| 186 |
m = chars_to_token.setdefault(len(string), {}) |
| 187 |
for c in string[:-1]: |
| 188 |
m = m.setdefault(c, {}) |
| 189 |
m[string[-1]] = name |
| 190 |
|
| 191 |
names = [] |
| 192 |
for value, name in enumerate(tok_names): |
| 193 |
if value >= ERRORTOKEN: |
| 194 |
name = '<%s>' % name |
| 195 |
names.append(' "%s",\n' % name) |
| 196 |
names.append(' "<N_TOKENS>",\n') |
| 197 |
|
| 198 |
if update_file(outfile, token_c_template % ( |
| 199 |
''.join(names), |
| 200 |
generate_chars_to_token(chars_to_token[1]), |
| 201 |
generate_chars_to_token(chars_to_token[2]), |
| 202 |
generate_chars_to_token(chars_to_token[3]) |
| 203 |
)): |
| 204 |
print("%s regenerated from %s" % (outfile, infile)) |
| 205 |
|
| 206 |
|
| 207 |
token_inc_template = f"""\ |
| 208 |
.. {AUTO_GENERATED_BY_SCRIPT} |
| 209 |
|
| 210 |
.. list-table:: |
| 211 |
:align: left |
| 212 |
:header-rows: 1 |
| 213 |
|
| 214 |
* - Token |
| 215 |
- Value |
| 216 |
%s |
| 217 |
""" |
| 218 |
|
| 219 |
def make_rst(infile, outfile='Doc/library/token-list.inc', |
| 220 |
rstfile='Doc/library/token.rst'): |
| 221 |
tok_names, ERRORTOKEN, string_to_tok = load_tokens(infile) |
| 222 |
tok_to_string = {value: s for s, value in string_to_tok.items()} |
| 223 |
|
| 224 |
needs_handwritten_doc = set() |
| 225 |
|
| 226 |
names = [] |
| 227 |
for value, name in enumerate(tok_names): |
| 228 |
if value in tok_to_string: |
| 229 |
assert name.isupper() |
| 230 |
names.append(f' * - .. data:: {name}') |
| 231 |
names.append(f' - ``"{tok_to_string[value]}"``') |
| 232 |
else: |
| 233 |
needs_handwritten_doc.add(name) |
| 234 |
|
| 235 |
has_handwritten_doc = set() |
| 236 |
with open(rstfile) as fileobj: |
| 237 |
tokendef_re = re.compile(r'.. data:: ([0-9A-Z_]+)\s*') |
| 238 |
for line in fileobj: |
| 239 |
if match := tokendef_re.fullmatch(line): |
| 240 |
has_handwritten_doc.add(match[1]) |
| 241 |
|
| 242 |
# Exclude non-token constants in token.py |
| 243 |
has_handwritten_doc -= {'N_TOKENS', 'NT_OFFSET', 'EXACT_TOKEN_TYPES'} |
| 244 |
|
| 245 |
if needs_handwritten_doc != has_handwritten_doc: |
| 246 |
message_parts = [f'ERROR: {rstfile} does not document all tokens!'] |
| 247 |
undocumented = needs_handwritten_doc - has_handwritten_doc |
| 248 |
extra = has_handwritten_doc - needs_handwritten_doc |
| 249 |
if undocumented: |
| 250 |
message_parts.append(f'Undocumented tokens: {undocumented}') |
| 251 |
if extra: |
| 252 |
message_parts.append(f'Documented nonexistent tokens: {extra}') |
| 253 |
exit('\n'.join(message_parts)) |
| 254 |
|
| 255 |
if update_file(outfile, token_inc_template % '\n'.join(names)): |
| 256 |
print("%s regenerated from %s" % (outfile, infile)) |
| 257 |
|
| 258 |
|
| 259 |
token_py_template = f'''\ |
| 260 |
"""Token constants.""" |
| 261 |
# {AUTO_GENERATED_BY_SCRIPT} |
| 262 |
''' |
| 263 |
token_py_template += ''' |
| 264 |
__all__ = ['tok_name', 'ISTERMINAL', 'ISNONTERMINAL', 'ISEOF', |
| 265 |
'EXACT_TOKEN_TYPES'] |
| 266 |
|
| 267 |
%s |
| 268 |
N_TOKENS = %d |
| 269 |
# Special definitions for cooperation with parser |
| 270 |
NT_OFFSET = %d |
| 271 |
|
| 272 |
tok_name = {value: name |
| 273 |
for name, value in globals().items() |
| 274 |
if isinstance(value, int) and not name.startswith('_')} |
| 275 |
__all__.extend(tok_name.values()) |
| 276 |
|
| 277 |
EXACT_TOKEN_TYPES = { |
| 278 |
%s |
| 279 |
} |
| 280 |
|
| 281 |
def ISTERMINAL(x: int) -> bool: |
| 282 |
return x < NT_OFFSET |
| 283 |
|
| 284 |
def ISNONTERMINAL(x: int) -> bool: |
| 285 |
return x >= NT_OFFSET |
| 286 |
|
| 287 |
def ISEOF(x: int) -> bool: |
| 288 |
return x == ENDMARKER |
| 289 |
''' |
| 290 |
|
| 291 |
def make_py(infile, outfile='Lib/token.py'): |
| 292 |
tok_names, ERRORTOKEN, string_to_tok = load_tokens(infile) |
| 293 |
|
| 294 |
constants = [] |
| 295 |
for value, name in enumerate(tok_names): |
| 296 |
constants.append('%s = %d' % (name, value)) |
| 297 |
constants.insert(ERRORTOKEN, |
| 298 |
"# These aren't used by the C tokenizer but are needed for tokenize.py") |
| 299 |
|
| 300 |
token_types = [] |
| 301 |
for s, value in sorted(string_to_tok.items()): |
| 302 |
token_types.append(' %r: %s,' % (s, tok_names[value])) |
| 303 |
|
| 304 |
if update_file(outfile, token_py_template % ( |
| 305 |
'\n'.join(constants), |
| 306 |
len(tok_names), |
| 307 |
NT_OFFSET, |
| 308 |
'\n'.join(token_types), |
| 309 |
)): |
| 310 |
print("%s regenerated from %s" % (outfile, infile)) |
| 311 |
|
| 312 |
|
| 313 |
def main(op, infile='Grammar/Tokens', *args): |
| 314 |
make = globals()['make_' + op] |
| 315 |
make(infile, *args) |
| 316 |
|
| 317 |
|
| 318 |
if __name__ == '__main__': |
| 319 |
import sys |
| 320 |
main(*sys.argv[1:]) |