| 1 | /*
|
| 2 | * Fast lexer using re2c.
|
| 3 | */
|
| 4 |
|
| 5 | #include <stdarg.h> // va_list, etc.
|
| 6 | #include <stdio.h> // printf
|
| 7 |
|
| 8 | #include <Python.h>
|
| 9 |
|
| 10 | #include "_gen/frontend/id_kind.asdl_c.h"
|
| 11 | #include "_gen/frontend/types.asdl_c.h" // for lex_mode_e
|
| 12 | #include "_gen/frontend/match.re2c.h"
|
| 13 |
|
| 14 | // TODO: Should this be shared among all extensions?
|
| 15 | // Log messages to stderr.
|
| 16 | #if 0
|
| 17 | static void debug(const char* fmt, ...) {
|
| 18 | va_list args;
|
| 19 | va_start(args, fmt);
|
| 20 | vfprintf(stderr, fmt, args);
|
| 21 | va_end(args);
|
| 22 | fprintf(stderr, "\n");
|
| 23 | }
|
| 24 | #endif
|
| 25 |
|
| 26 | static PyObject *
|
| 27 | fastlex_MatchOshToken(PyObject *self, PyObject *args) {
|
| 28 | int lex_mode;
|
| 29 |
|
| 30 | unsigned char* line;
|
| 31 | int line_len;
|
| 32 |
|
| 33 | int start_pos;
|
| 34 | if (!PyArg_ParseTuple(args, "is#i",
|
| 35 | &lex_mode, &line, &line_len, &start_pos)) {
|
| 36 | return NULL;
|
| 37 | }
|
| 38 |
|
| 39 | // Bounds checking. It's OK to be called with a start_pos looking at \0.
|
| 40 | // Eol_Tok is inserted everywhere.
|
| 41 | if (start_pos > line_len) {
|
| 42 | PyErr_Format(PyExc_ValueError,
|
| 43 | "Invalid MatchOshToken call (start_pos = %d, line_len = %d)",
|
| 44 | start_pos, line_len);
|
| 45 | return NULL;
|
| 46 | }
|
| 47 |
|
| 48 | int id;
|
| 49 | int end_pos;
|
| 50 | MatchOshToken(lex_mode, line, line_len, start_pos, &id, &end_pos);
|
| 51 | return Py_BuildValue("(ii)", id, end_pos);
|
| 52 | }
|
| 53 |
|
| 54 | static PyObject *
|
| 55 | fastlex_MatchEchoToken(PyObject *self, PyObject *args) {
|
| 56 | unsigned char* line;
|
| 57 | int line_len;
|
| 58 |
|
| 59 | int start_pos;
|
| 60 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 61 | return NULL;
|
| 62 | }
|
| 63 |
|
| 64 | // Bounds checking.
|
| 65 | if (start_pos > line_len) {
|
| 66 | PyErr_Format(PyExc_ValueError,
|
| 67 | "Invalid MatchEchoToken call (start_pos = %d, line_len = %d)",
|
| 68 | start_pos, line_len);
|
| 69 | return NULL;
|
| 70 | }
|
| 71 |
|
| 72 | int id;
|
| 73 | int end_pos;
|
| 74 | MatchEchoToken(line, line_len, start_pos, &id, &end_pos);
|
| 75 | return Py_BuildValue("(ii)", id, end_pos);
|
| 76 | }
|
| 77 |
|
| 78 | static PyObject *
|
| 79 | fastlex_MatchGlobToken(PyObject *self, PyObject *args) {
|
| 80 | unsigned char* line;
|
| 81 | int line_len;
|
| 82 |
|
| 83 | int start_pos;
|
| 84 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 85 | return NULL;
|
| 86 | }
|
| 87 |
|
| 88 | // Bounds checking.
|
| 89 | if (start_pos > line_len) {
|
| 90 | PyErr_Format(PyExc_ValueError,
|
| 91 | "Invalid MatchGlobToken call (start_pos = %d, line_len = %d)",
|
| 92 | start_pos, line_len);
|
| 93 | return NULL;
|
| 94 | }
|
| 95 |
|
| 96 | int id;
|
| 97 | int end_pos;
|
| 98 | MatchGlobToken(line, line_len, start_pos, &id, &end_pos);
|
| 99 | return Py_BuildValue("(ii)", id, end_pos);
|
| 100 | }
|
| 101 |
|
| 102 | static PyObject *
|
| 103 | fastlex_MatchPS1Token(PyObject *self, PyObject *args) {
|
| 104 | unsigned char* line;
|
| 105 | int line_len;
|
| 106 |
|
| 107 | int start_pos;
|
| 108 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 109 | return NULL;
|
| 110 | }
|
| 111 |
|
| 112 | // Bounds checking.
|
| 113 | if (start_pos > line_len) {
|
| 114 | PyErr_Format(PyExc_ValueError,
|
| 115 | "Invalid MatchPS1Token call (start_pos = %d, line_len = %d)",
|
| 116 | start_pos, line_len);
|
| 117 | return NULL;
|
| 118 | }
|
| 119 |
|
| 120 | int id;
|
| 121 | int end_pos;
|
| 122 | MatchPS1Token(line, line_len, start_pos, &id, &end_pos);
|
| 123 | return Py_BuildValue("(ii)", id, end_pos);
|
| 124 | }
|
| 125 |
|
| 126 | static PyObject *
|
| 127 | fastlex_MatchHistoryToken(PyObject *self, PyObject *args) {
|
| 128 | unsigned char* line;
|
| 129 | int line_len;
|
| 130 |
|
| 131 | int start_pos;
|
| 132 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 133 | return NULL;
|
| 134 | }
|
| 135 |
|
| 136 | // Bounds checking.
|
| 137 | if (start_pos > line_len) {
|
| 138 | PyErr_Format(PyExc_ValueError,
|
| 139 | "Invalid MatchHistoryToken call (start_pos = %d, line_len = %d)",
|
| 140 | start_pos, line_len);
|
| 141 | return NULL;
|
| 142 | }
|
| 143 |
|
| 144 | int id;
|
| 145 | int end_pos;
|
| 146 | MatchHistoryToken(line, line_len, start_pos, &id, &end_pos);
|
| 147 | return Py_BuildValue("(ii)", id, end_pos);
|
| 148 | }
|
| 149 |
|
| 150 | static PyObject *
|
| 151 | fastlex_MatchBraceRangeToken(PyObject *self, PyObject *args) {
|
| 152 | unsigned char* line;
|
| 153 | int line_len;
|
| 154 |
|
| 155 | int start_pos;
|
| 156 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 157 | return NULL;
|
| 158 | }
|
| 159 |
|
| 160 | // Bounds checking.
|
| 161 | if (start_pos > line_len) {
|
| 162 | PyErr_Format(PyExc_ValueError,
|
| 163 | "Invalid MatchBraceRangeToken call (start_pos = %d, line_len = %d)",
|
| 164 | start_pos, line_len);
|
| 165 | return NULL;
|
| 166 | }
|
| 167 |
|
| 168 | int id;
|
| 169 | int end_pos;
|
| 170 | MatchBraceRangeToken(line, line_len, start_pos, &id, &end_pos);
|
| 171 | return Py_BuildValue("(ii)", id, end_pos);
|
| 172 | }
|
| 173 |
|
| 174 | static PyObject *
|
| 175 | fastlex_MatchJ8Token(PyObject *self, PyObject *args) {
|
| 176 | unsigned char* line;
|
| 177 | int line_len;
|
| 178 |
|
| 179 | int start_pos;
|
| 180 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 181 | return NULL;
|
| 182 | }
|
| 183 |
|
| 184 | // Bounds checking.
|
| 185 | if (start_pos > line_len) {
|
| 186 | PyErr_Format(PyExc_ValueError,
|
| 187 | "Invalid MatchJ8Token call (start_pos = %d, line_len = %d)",
|
| 188 | start_pos, line_len);
|
| 189 | return NULL;
|
| 190 | }
|
| 191 |
|
| 192 | int id;
|
| 193 | int end_pos;
|
| 194 | MatchJ8Token(line, line_len, start_pos, &id, &end_pos);
|
| 195 | return Py_BuildValue("(ii)", id, end_pos);
|
| 196 | }
|
| 197 |
|
| 198 | static PyObject *
|
| 199 | fastlex_MatchJ8LinesToken(PyObject *self, PyObject *args) {
|
| 200 | unsigned char* line;
|
| 201 | int line_len;
|
| 202 |
|
| 203 | int start_pos;
|
| 204 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 205 | return NULL;
|
| 206 | }
|
| 207 |
|
| 208 | // Bounds checking.
|
| 209 | if (start_pos > line_len) {
|
| 210 | PyErr_Format(PyExc_ValueError,
|
| 211 | "Invalid MatchJ8LinesToken call (start_pos = %d, line_len = %d)",
|
| 212 | start_pos, line_len);
|
| 213 | return NULL;
|
| 214 | }
|
| 215 |
|
| 216 | int id;
|
| 217 | int end_pos;
|
| 218 | MatchJ8LinesToken(line, line_len, start_pos, &id, &end_pos);
|
| 219 | return Py_BuildValue("(ii)", id, end_pos);
|
| 220 | }
|
| 221 |
|
| 222 |
|
| 223 | static PyObject *
|
| 224 | fastlex_MatchJ8StrToken(PyObject *self, PyObject *args) {
|
| 225 | unsigned char* line;
|
| 226 | int line_len;
|
| 227 |
|
| 228 | int start_pos;
|
| 229 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 230 | return NULL;
|
| 231 | }
|
| 232 |
|
| 233 | // Bounds checking.
|
| 234 | if (start_pos > line_len) {
|
| 235 | PyErr_Format(PyExc_ValueError,
|
| 236 | "Invalid MatchJ8StrToken call (start_pos = %d, line_len = %d)",
|
| 237 | start_pos, line_len);
|
| 238 | return NULL;
|
| 239 | }
|
| 240 |
|
| 241 | int id;
|
| 242 | int end_pos;
|
| 243 | MatchJ8StrToken(line, line_len, start_pos, &id, &end_pos);
|
| 244 | return Py_BuildValue("(ii)", id, end_pos);
|
| 245 | }
|
| 246 |
|
| 247 | static PyObject *
|
| 248 | fastlex_MatchJsonStrToken(PyObject *self, PyObject *args) {
|
| 249 | unsigned char* line;
|
| 250 | int line_len;
|
| 251 |
|
| 252 | int start_pos;
|
| 253 | if (!PyArg_ParseTuple(args, "s#i", &line, &line_len, &start_pos)) {
|
| 254 | return NULL;
|
| 255 | }
|
| 256 |
|
| 257 | // Bounds checking.
|
| 258 | if (start_pos > line_len) {
|
| 259 | PyErr_Format(PyExc_ValueError,
|
| 260 | "Invalid MatchJsonStrToken call (start_pos = %d, line_len = %d)",
|
| 261 | start_pos, line_len);
|
| 262 | return NULL;
|
| 263 | }
|
| 264 |
|
| 265 | int id;
|
| 266 | int end_pos;
|
| 267 | MatchJsonStrToken(line, line_len, start_pos, &id, &end_pos);
|
| 268 | return Py_BuildValue("(ii)", id, end_pos);
|
| 269 | }
|
| 270 |
|
| 271 | static PyObject *
|
| 272 | fastlex_IsValidVarName(PyObject *self, PyObject *args) {
|
| 273 | unsigned char *name;
|
| 274 | int len;
|
| 275 |
|
| 276 | if (!PyArg_ParseTuple(args, "s#", &name, &len)) {
|
| 277 | return NULL;
|
| 278 | }
|
| 279 | return PyBool_FromLong(IsValidVarName(name, len));
|
| 280 | }
|
| 281 |
|
| 282 | static PyObject *
|
| 283 | fastlex_ShouldHijack(PyObject *self, PyObject *args) {
|
| 284 | unsigned char *name;
|
| 285 | int len;
|
| 286 |
|
| 287 | if (!PyArg_ParseTuple(args, "s#", &name, &len)) {
|
| 288 | return NULL;
|
| 289 | }
|
| 290 | return PyBool_FromLong(ShouldHijack(name, len));
|
| 291 | }
|
| 292 |
|
| 293 | static PyObject *
|
| 294 | fastlex_LooksLikeInteger(PyObject *self, PyObject *args) {
|
| 295 | unsigned char *name;
|
| 296 | int len;
|
| 297 |
|
| 298 | if (!PyArg_ParseTuple(args, "s#", &name, &len)) {
|
| 299 | return NULL;
|
| 300 | }
|
| 301 | return PyBool_FromLong(LooksLikeInteger(name, len));
|
| 302 | }
|
| 303 |
|
| 304 | static PyObject *
|
| 305 | fastlex_LooksLikeFloat(PyObject *self, PyObject *args) {
|
| 306 | unsigned char *name;
|
| 307 | int len;
|
| 308 |
|
| 309 | if (!PyArg_ParseTuple(args, "s#", &name, &len)) {
|
| 310 | return NULL;
|
| 311 | }
|
| 312 | return PyBool_FromLong(LooksLikeFloat(name, len));
|
| 313 | }
|
| 314 |
|
| 315 | #ifdef OVM_MAIN
|
| 316 | #include "pyext/fastlex.c/methods.def"
|
| 317 | #else
|
| 318 | static PyMethodDef methods[] = {
|
| 319 | {"MatchOshToken", fastlex_MatchOshToken, METH_VARARGS,
|
| 320 | "(lexer mode, line, start_pos) -> (id, end_pos)."},
|
| 321 | {"MatchEchoToken", fastlex_MatchEchoToken, METH_VARARGS,
|
| 322 | "(line, start_pos) -> (id, end_pos)."},
|
| 323 | {"MatchGlobToken", fastlex_MatchGlobToken, METH_VARARGS,
|
| 324 | "(line, start_pos) -> (id, end_pos)."},
|
| 325 | {"MatchPS1Token", fastlex_MatchPS1Token, METH_VARARGS,
|
| 326 | "(line, start_pos) -> (id, end_pos)."},
|
| 327 | {"MatchHistoryToken", fastlex_MatchHistoryToken, METH_VARARGS,
|
| 328 | "(line, start_pos) -> (id, end_pos)."},
|
| 329 | {"MatchBraceRangeToken", fastlex_MatchBraceRangeToken, METH_VARARGS,
|
| 330 | "(line, start_pos) -> (id, end_pos)."},
|
| 331 | {"MatchJ8Token", fastlex_MatchJ8Token, METH_VARARGS,
|
| 332 | "(line, start_pos) -> (id, end_pos)."},
|
| 333 | {"MatchJ8LinesToken", fastlex_MatchJ8LinesToken, METH_VARARGS,
|
| 334 | "(line, start_pos) -> (id, end_pos)."},
|
| 335 | {"MatchJ8StrToken", fastlex_MatchJ8StrToken, METH_VARARGS,
|
| 336 | "(line, start_pos) -> (id, end_pos)."},
|
| 337 | {"MatchJsonStrToken", fastlex_MatchJsonStrToken, METH_VARARGS,
|
| 338 | "(line, start_pos) -> (id, end_pos)."},
|
| 339 | {"IsValidVarName", fastlex_IsValidVarName, METH_VARARGS,
|
| 340 | "Is it a valid var name?"},
|
| 341 | // Should we hijack this shebang line?
|
| 342 | {"ShouldHijack", fastlex_ShouldHijack, METH_VARARGS, ""},
|
| 343 | {"LooksLikeInteger", fastlex_LooksLikeInteger, METH_VARARGS, ""},
|
| 344 | {"LooksLikeFloat", fastlex_LooksLikeFloat, METH_VARARGS, ""},
|
| 345 | {NULL, NULL},
|
| 346 | };
|
| 347 | #endif
|
| 348 |
|
| 349 | void initfastlex(void) {
|
| 350 | Py_InitModule("fastlex", methods);
|
| 351 | }
|