This reverts r63675 based on the discussion in this thread:
[python.git] / Modules / parsermodule.c
blob6e52343e0db188669208650fc03013bcae7ebe37
1 /* parsermodule.c
3 * Copyright 1995-1996 by Fred L. Drake, Jr. and Virginia Polytechnic
4 * Institute and State University, Blacksburg, Virginia, USA.
5 * Portions copyright 1991-1995 by Stichting Mathematisch Centrum,
6 * Amsterdam, The Netherlands. Copying is permitted under the terms
7 * associated with the main Python distribution, with the additional
8 * restriction that this additional notice be included and maintained
9 * on all distributed copies.
11 * This module serves to replace the original parser module written
12 * by Guido. The functionality is not matched precisely, but the
13 * original may be implemented on top of this. This is desirable
14 * since the source of the text to be parsed is now divorced from
15 * this interface.
17 * Unlike the prior interface, the ability to give a parse tree
18 * produced by Python code as a tuple to the compiler is enabled by
19 * this module. See the documentation for more details.
21 * I've added some annotations that help with the lint code-checking
22 * program, but they're not complete by a long shot. The real errors
23 * that lint detects are gone, but there are still warnings with
24 * Py_[X]DECREF() and Py_[X]INCREF() macros. The lint annotations
25 * look like "NOTE(...)".
28 #include "Python.h" /* general Python API */
29 #include "graminit.h" /* symbols defined in the grammar */
30 #include "node.h" /* internal parser structure */
31 #include "errcode.h" /* error codes for PyNode_*() */
32 #include "token.h" /* token definitions */
33 /* ISTERMINAL() / ISNONTERMINAL() */
34 #include "compile.h" /* PyNode_Compile() */
36 #ifdef lint
37 #include <note.h>
38 #else
39 #define NOTE(x)
40 #endif
42 /* String constants used to initialize module attributes.
45 static char parser_copyright_string[] =
46 "Copyright 1995-1996 by Virginia Polytechnic Institute & State\n\
47 University, Blacksburg, Virginia, USA, and Fred L. Drake, Jr., Reston,\n\
48 Virginia, USA. Portions copyright 1991-1995 by Stichting Mathematisch\n\
49 Centrum, Amsterdam, The Netherlands.";
52 PyDoc_STRVAR(parser_doc_string,
53 "This is an interface to Python's internal parser.");
55 static char parser_version_string[] = "0.5";
58 typedef PyObject* (*SeqMaker) (Py_ssize_t length);
59 typedef int (*SeqInserter) (PyObject* sequence,
60 Py_ssize_t index,
61 PyObject* element);
63 /* The function below is copyrighted by Stichting Mathematisch Centrum. The
64 * original copyright statement is included below, and continues to apply
65 * in full to the function immediately following. All other material is
66 * original, copyrighted by Fred L. Drake, Jr. and Virginia Polytechnic
67 * Institute and State University. Changes were made to comply with the
68 * new naming conventions. Added arguments to provide support for creating
69 * lists as well as tuples, and optionally including the line numbers.
73 static PyObject*
74 node2tuple(node *n, /* node to convert */
75 SeqMaker mkseq, /* create sequence */
76 SeqInserter addelem, /* func. to add elem. in seq. */
77 int lineno, /* include line numbers? */
78 int col_offset) /* include column offsets? */
80 if (n == NULL) {
81 Py_INCREF(Py_None);
82 return (Py_None);
84 if (ISNONTERMINAL(TYPE(n))) {
85 int i;
86 PyObject *v;
87 PyObject *w;
89 v = mkseq(1 + NCH(n) + (TYPE(n) == encoding_decl));
90 if (v == NULL)
91 return (v);
92 w = PyInt_FromLong(TYPE(n));
93 if (w == NULL) {
94 Py_DECREF(v);
95 return ((PyObject*) NULL);
97 (void) addelem(v, 0, w);
98 for (i = 0; i < NCH(n); i++) {
99 w = node2tuple(CHILD(n, i), mkseq, addelem, lineno, col_offset);
100 if (w == NULL) {
101 Py_DECREF(v);
102 return ((PyObject*) NULL);
104 (void) addelem(v, i+1, w);
107 if (TYPE(n) == encoding_decl)
108 (void) addelem(v, i+1, PyString_FromString(STR(n)));
109 return (v);
111 else if (ISTERMINAL(TYPE(n))) {
112 PyObject *result = mkseq(2 + lineno + col_offset);
113 if (result != NULL) {
114 (void) addelem(result, 0, PyInt_FromLong(TYPE(n)));
115 (void) addelem(result, 1, PyString_FromString(STR(n)));
116 if (lineno == 1)
117 (void) addelem(result, 2, PyInt_FromLong(n->n_lineno));
118 if (col_offset == 1)
119 (void) addelem(result, 3, PyInt_FromLong(n->n_col_offset));
121 return (result);
123 else {
124 PyErr_SetString(PyExc_SystemError,
125 "unrecognized parse tree node type");
126 return ((PyObject*) NULL);
130 * End of material copyrighted by Stichting Mathematisch Centrum.
135 /* There are two types of intermediate objects we're interested in:
136 * 'eval' and 'exec' types. These constants can be used in the st_type
137 * field of the object type to identify which any given object represents.
138 * These should probably go in an external header to allow other extensions
139 * to use them, but then, we really should be using C++ too. ;-)
142 #define PyST_EXPR 1
143 #define PyST_SUITE 2
146 /* These are the internal objects and definitions required to implement the
147 * ST type. Most of the internal names are more reminiscent of the 'old'
148 * naming style, but the code uses the new naming convention.
151 static PyObject*
152 parser_error = 0;
155 typedef struct {
156 PyObject_HEAD /* standard object header */
157 node* st_node; /* the node* returned by the parser */
158 int st_type; /* EXPR or SUITE ? */
159 } PyST_Object;
162 static void parser_free(PyST_Object *st);
163 static int parser_compare(PyST_Object *left, PyST_Object *right);
164 static PyObject *parser_getattr(PyObject *self, char *name);
167 static
168 PyTypeObject PyST_Type = {
169 PyVarObject_HEAD_INIT(NULL, 0)
170 "parser.st", /* tp_name */
171 (int) sizeof(PyST_Object), /* tp_basicsize */
172 0, /* tp_itemsize */
173 (destructor)parser_free, /* tp_dealloc */
174 0, /* tp_print */
175 parser_getattr, /* tp_getattr */
176 0, /* tp_setattr */
177 (cmpfunc)parser_compare, /* tp_compare */
178 0, /* tp_repr */
179 0, /* tp_as_number */
180 0, /* tp_as_sequence */
181 0, /* tp_as_mapping */
182 0, /* tp_hash */
183 0, /* tp_call */
184 0, /* tp_str */
185 0, /* tp_getattro */
186 0, /* tp_setattro */
188 /* Functions to access object as input/output buffer */
189 0, /* tp_as_buffer */
191 Py_TPFLAGS_DEFAULT, /* tp_flags */
193 /* __doc__ */
194 "Intermediate representation of a Python parse tree."
195 }; /* PyST_Type */
198 static int
199 parser_compare_nodes(node *left, node *right)
201 int j;
203 if (TYPE(left) < TYPE(right))
204 return (-1);
206 if (TYPE(right) < TYPE(left))
207 return (1);
209 if (ISTERMINAL(TYPE(left)))
210 return (strcmp(STR(left), STR(right)));
212 if (NCH(left) < NCH(right))
213 return (-1);
215 if (NCH(right) < NCH(left))
216 return (1);
218 for (j = 0; j < NCH(left); ++j) {
219 int v = parser_compare_nodes(CHILD(left, j), CHILD(right, j));
221 if (v != 0)
222 return (v);
224 return (0);
228 /* int parser_compare(PyST_Object* left, PyST_Object* right)
230 * Comparison function used by the Python operators ==, !=, <, >, <=, >=
231 * This really just wraps a call to parser_compare_nodes() with some easy
232 * checks and protection code.
235 static int
236 parser_compare(PyST_Object *left, PyST_Object *right)
238 if (left == right)
239 return (0);
241 if ((left == 0) || (right == 0))
242 return (-1);
244 return (parser_compare_nodes(left->st_node, right->st_node));
248 /* parser_newstobject(node* st)
250 * Allocates a new Python object representing an ST. This is simply the
251 * 'wrapper' object that holds a node* and allows it to be passed around in
252 * Python code.
255 static PyObject*
256 parser_newstobject(node *st, int type)
258 PyST_Object* o = PyObject_New(PyST_Object, &PyST_Type);
260 if (o != 0) {
261 o->st_node = st;
262 o->st_type = type;
264 else {
265 PyNode_Free(st);
267 return ((PyObject*)o);
271 /* void parser_free(PyST_Object* st)
273 * This is called by a del statement that reduces the reference count to 0.
276 static void
277 parser_free(PyST_Object *st)
279 PyNode_Free(st->st_node);
280 PyObject_Del(st);
284 /* parser_st2tuple(PyObject* self, PyObject* args, PyObject* kw)
286 * This provides conversion from a node* to a tuple object that can be
287 * returned to the Python-level caller. The ST object is not modified.
290 static PyObject*
291 parser_st2tuple(PyST_Object *self, PyObject *args, PyObject *kw)
293 PyObject *line_option = 0;
294 PyObject *col_option = 0;
295 PyObject *res = 0;
296 int ok;
298 static char *keywords[] = {"ast", "line_info", "col_info", NULL};
300 if (self == NULL) {
301 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|OO:st2tuple", keywords,
302 &PyST_Type, &self, &line_option,
303 &col_option);
305 else
306 ok = PyArg_ParseTupleAndKeywords(args, kw, "|OO:totuple", &keywords[1],
307 &line_option, &col_option);
308 if (ok != 0) {
309 int lineno = 0;
310 int col_offset = 0;
311 if (line_option != NULL) {
312 lineno = (PyObject_IsTrue(line_option) != 0) ? 1 : 0;
314 if (col_option != NULL) {
315 col_offset = (PyObject_IsTrue(col_option) != 0) ? 1 : 0;
318 * Convert ST into a tuple representation. Use Guido's function,
319 * since it's known to work already.
321 res = node2tuple(((PyST_Object*)self)->st_node,
322 PyTuple_New, PyTuple_SetItem, lineno, col_offset);
324 return (res);
328 /* parser_st2list(PyObject* self, PyObject* args, PyObject* kw)
330 * This provides conversion from a node* to a list object that can be
331 * returned to the Python-level caller. The ST object is not modified.
334 static PyObject*
335 parser_st2list(PyST_Object *self, PyObject *args, PyObject *kw)
337 PyObject *line_option = 0;
338 PyObject *col_option = 0;
339 PyObject *res = 0;
340 int ok;
342 static char *keywords[] = {"ast", "line_info", "col_info", NULL};
344 if (self == NULL)
345 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|OO:st2list", keywords,
346 &PyST_Type, &self, &line_option,
347 &col_option);
348 else
349 ok = PyArg_ParseTupleAndKeywords(args, kw, "|OO:tolist", &keywords[1],
350 &line_option, &col_option);
351 if (ok) {
352 int lineno = 0;
353 int col_offset = 0;
354 if (line_option != 0) {
355 lineno = PyObject_IsTrue(line_option) ? 1 : 0;
357 if (col_option != NULL) {
358 col_offset = (PyObject_IsTrue(col_option) != 0) ? 1 : 0;
361 * Convert ST into a tuple representation. Use Guido's function,
362 * since it's known to work already.
364 res = node2tuple(self->st_node,
365 PyList_New, PyList_SetItem, lineno, col_offset);
367 return (res);
371 /* parser_compilest(PyObject* self, PyObject* args)
373 * This function creates code objects from the parse tree represented by
374 * the passed-in data object. An optional file name is passed in as well.
377 static PyObject*
378 parser_compilest(PyST_Object *self, PyObject *args, PyObject *kw)
380 PyObject* res = 0;
381 char* str = "<syntax-tree>";
382 int ok;
384 static char *keywords[] = {"ast", "filename", NULL};
386 if (self == NULL)
387 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|s:compilest", keywords,
388 &PyST_Type, &self, &str);
389 else
390 ok = PyArg_ParseTupleAndKeywords(args, kw, "|s:compile", &keywords[1],
391 &str);
393 if (ok)
394 res = (PyObject *)PyNode_Compile(self->st_node, str);
396 return (res);
400 /* PyObject* parser_isexpr(PyObject* self, PyObject* args)
401 * PyObject* parser_issuite(PyObject* self, PyObject* args)
403 * Checks the passed-in ST object to determine if it is an expression or
404 * a statement suite, respectively. The return is a Python truth value.
407 static PyObject*
408 parser_isexpr(PyST_Object *self, PyObject *args, PyObject *kw)
410 PyObject* res = 0;
411 int ok;
413 static char *keywords[] = {"ast", NULL};
415 if (self == NULL)
416 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!:isexpr", keywords,
417 &PyST_Type, &self);
418 else
419 ok = PyArg_ParseTupleAndKeywords(args, kw, ":isexpr", &keywords[1]);
421 if (ok) {
422 /* Check to see if the ST represents an expression or not. */
423 res = (self->st_type == PyST_EXPR) ? Py_True : Py_False;
424 Py_INCREF(res);
426 return (res);
430 static PyObject*
431 parser_issuite(PyST_Object *self, PyObject *args, PyObject *kw)
433 PyObject* res = 0;
434 int ok;
436 static char *keywords[] = {"ast", NULL};
438 if (self == NULL)
439 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!:issuite", keywords,
440 &PyST_Type, &self);
441 else
442 ok = PyArg_ParseTupleAndKeywords(args, kw, ":issuite", &keywords[1]);
444 if (ok) {
445 /* Check to see if the ST represents an expression or not. */
446 res = (self->st_type == PyST_EXPR) ? Py_False : Py_True;
447 Py_INCREF(res);
449 return (res);
453 #define PUBLIC_METHOD_TYPE (METH_VARARGS|METH_KEYWORDS)
455 static PyMethodDef
456 parser_methods[] = {
457 {"compile", (PyCFunction)parser_compilest, PUBLIC_METHOD_TYPE,
458 PyDoc_STR("Compile this ST object into a code object.")},
459 {"isexpr", (PyCFunction)parser_isexpr, PUBLIC_METHOD_TYPE,
460 PyDoc_STR("Determines if this ST object was created from an expression.")},
461 {"issuite", (PyCFunction)parser_issuite, PUBLIC_METHOD_TYPE,
462 PyDoc_STR("Determines if this ST object was created from a suite.")},
463 {"tolist", (PyCFunction)parser_st2list, PUBLIC_METHOD_TYPE,
464 PyDoc_STR("Creates a list-tree representation of this ST.")},
465 {"totuple", (PyCFunction)parser_st2tuple, PUBLIC_METHOD_TYPE,
466 PyDoc_STR("Creates a tuple-tree representation of this ST.")},
468 {NULL, NULL, 0, NULL}
472 static PyObject*
473 parser_getattr(PyObject *self, char *name)
475 return (Py_FindMethod(parser_methods, self, name));
479 /* err_string(char* message)
481 * Sets the error string for an exception of type ParserError.
484 static void
485 err_string(char *message)
487 PyErr_SetString(parser_error, message);
491 /* PyObject* parser_do_parse(PyObject* args, int type)
493 * Internal function to actually execute the parse and return the result if
494 * successful or set an exception if not.
497 static PyObject*
498 parser_do_parse(PyObject *args, PyObject *kw, char *argspec, int type)
500 char* string = 0;
501 PyObject* res = 0;
503 static char *keywords[] = {"source", NULL};
505 if (PyArg_ParseTupleAndKeywords(args, kw, argspec, keywords, &string)) {
506 node* n = PyParser_SimpleParseString(string,
507 (type == PyST_EXPR)
508 ? eval_input : file_input);
510 if (n)
511 res = parser_newstobject(n, type);
513 return (res);
517 /* PyObject* parser_expr(PyObject* self, PyObject* args)
518 * PyObject* parser_suite(PyObject* self, PyObject* args)
520 * External interfaces to the parser itself. Which is called determines if
521 * the parser attempts to recognize an expression ('eval' form) or statement
522 * suite ('exec' form). The real work is done by parser_do_parse() above.
525 static PyObject*
526 parser_expr(PyST_Object *self, PyObject *args, PyObject *kw)
528 NOTE(ARGUNUSED(self))
529 return (parser_do_parse(args, kw, "s:expr", PyST_EXPR));
533 static PyObject*
534 parser_suite(PyST_Object *self, PyObject *args, PyObject *kw)
536 NOTE(ARGUNUSED(self))
537 return (parser_do_parse(args, kw, "s:suite", PyST_SUITE));
542 /* This is the messy part of the code. Conversion from a tuple to an ST
543 * object requires that the input tuple be valid without having to rely on
544 * catching an exception from the compiler. This is done to allow the
545 * compiler itself to remain fast, since most of its input will come from
546 * the parser directly, and therefore be known to be syntactically correct.
547 * This validation is done to ensure that we don't core dump the compile
548 * phase, returning an exception instead.
550 * Two aspects can be broken out in this code: creating a node tree from
551 * the tuple passed in, and verifying that it is indeed valid. It may be
552 * advantageous to expand the number of ST types to include funcdefs and
553 * lambdadefs to take advantage of the optimizer, recognizing those STs
554 * here. They are not necessary, and not quite as useful in a raw form.
555 * For now, let's get expressions and suites working reliably.
559 static node* build_node_tree(PyObject *tuple);
560 static int validate_expr_tree(node *tree);
561 static int validate_file_input(node *tree);
562 static int validate_encoding_decl(node *tree);
564 /* PyObject* parser_tuple2st(PyObject* self, PyObject* args)
566 * This is the public function, called from the Python code. It receives a
567 * single tuple object from the caller, and creates an ST object if the
568 * tuple can be validated. It does this by checking the first code of the
569 * tuple, and, if acceptable, builds the internal representation. If this
570 * step succeeds, the internal representation is validated as fully as
571 * possible with the various validate_*() routines defined below.
573 * This function must be changed if support is to be added for PyST_FRAGMENT
574 * ST objects.
577 static PyObject*
578 parser_tuple2st(PyST_Object *self, PyObject *args, PyObject *kw)
580 NOTE(ARGUNUSED(self))
581 PyObject *st = 0;
582 PyObject *tuple;
583 node *tree;
585 static char *keywords[] = {"sequence", NULL};
587 if (!PyArg_ParseTupleAndKeywords(args, kw, "O:sequence2st", keywords,
588 &tuple))
589 return (0);
590 if (!PySequence_Check(tuple)) {
591 PyErr_SetString(PyExc_ValueError,
592 "sequence2st() requires a single sequence argument");
593 return (0);
596 * Convert the tree to the internal form before checking it.
598 tree = build_node_tree(tuple);
599 if (tree != 0) {
600 int start_sym = TYPE(tree);
601 if (start_sym == eval_input) {
602 /* Might be an eval form. */
603 if (validate_expr_tree(tree))
604 st = parser_newstobject(tree, PyST_EXPR);
605 else
606 PyNode_Free(tree);
608 else if (start_sym == file_input) {
609 /* This looks like an exec form so far. */
610 if (validate_file_input(tree))
611 st = parser_newstobject(tree, PyST_SUITE);
612 else
613 PyNode_Free(tree);
615 else if (start_sym == encoding_decl) {
616 /* This looks like an encoding_decl so far. */
617 if (validate_encoding_decl(tree))
618 st = parser_newstobject(tree, PyST_SUITE);
619 else
620 PyNode_Free(tree);
622 else {
623 /* This is a fragment, at best. */
624 PyNode_Free(tree);
625 err_string("parse tree does not use a valid start symbol");
628 /* Make sure we throw an exception on all errors. We should never
629 * get this, but we'd do well to be sure something is done.
631 if (st == NULL && !PyErr_Occurred())
632 err_string("unspecified ST error occurred");
634 return st;
638 /* node* build_node_children()
640 * Iterate across the children of the current non-terminal node and build
641 * their structures. If successful, return the root of this portion of
642 * the tree, otherwise, 0. Any required exception will be specified already,
643 * and no memory will have been deallocated.
646 static node*
647 build_node_children(PyObject *tuple, node *root, int *line_num)
649 Py_ssize_t len = PyObject_Size(tuple);
650 Py_ssize_t i;
651 int err;
653 for (i = 1; i < len; ++i) {
654 /* elem must always be a sequence, however simple */
655 PyObject* elem = PySequence_GetItem(tuple, i);
656 int ok = elem != NULL;
657 long type = 0;
658 char *strn = 0;
660 if (ok)
661 ok = PySequence_Check(elem);
662 if (ok) {
663 PyObject *temp = PySequence_GetItem(elem, 0);
664 if (temp == NULL)
665 ok = 0;
666 else {
667 ok = PyInt_Check(temp);
668 if (ok)
669 type = PyInt_AS_LONG(temp);
670 Py_DECREF(temp);
673 if (!ok) {
674 PyObject *err = Py_BuildValue("os", elem,
675 "Illegal node construct.");
676 PyErr_SetObject(parser_error, err);
677 Py_XDECREF(err);
678 Py_XDECREF(elem);
679 return (0);
681 if (ISTERMINAL(type)) {
682 Py_ssize_t len = PyObject_Size(elem);
683 PyObject *temp;
685 if ((len != 2) && (len != 3)) {
686 err_string("terminal nodes must have 2 or 3 entries");
687 return 0;
689 temp = PySequence_GetItem(elem, 1);
690 if (temp == NULL)
691 return 0;
692 if (!PyString_Check(temp)) {
693 PyErr_Format(parser_error,
694 "second item in terminal node must be a string,"
695 " found %s",
696 Py_TYPE(temp)->tp_name);
697 Py_DECREF(temp);
698 return 0;
700 if (len == 3) {
701 PyObject *o = PySequence_GetItem(elem, 2);
702 if (o != NULL) {
703 if (PyInt_Check(o))
704 *line_num = PyInt_AS_LONG(o);
705 else {
706 PyErr_Format(parser_error,
707 "third item in terminal node must be an"
708 " integer, found %s",
709 Py_TYPE(temp)->tp_name);
710 Py_DECREF(o);
711 Py_DECREF(temp);
712 return 0;
714 Py_DECREF(o);
717 len = PyString_GET_SIZE(temp) + 1;
718 strn = (char *)PyObject_MALLOC(len);
719 if (strn != NULL)
720 (void) memcpy(strn, PyString_AS_STRING(temp), len);
721 Py_DECREF(temp);
723 else if (!ISNONTERMINAL(type)) {
725 * It has to be one or the other; this is an error.
726 * Throw an exception.
728 PyObject *err = Py_BuildValue("os", elem, "unknown node type.");
729 PyErr_SetObject(parser_error, err);
730 Py_XDECREF(err);
731 Py_XDECREF(elem);
732 return (0);
734 err = PyNode_AddChild(root, type, strn, *line_num, 0);
735 if (err == E_NOMEM) {
736 PyObject_FREE(strn);
737 return (node *) PyErr_NoMemory();
739 if (err == E_OVERFLOW) {
740 PyObject_FREE(strn);
741 PyErr_SetString(PyExc_ValueError,
742 "unsupported number of child nodes");
743 return NULL;
746 if (ISNONTERMINAL(type)) {
747 node* new_child = CHILD(root, i - 1);
749 if (new_child != build_node_children(elem, new_child, line_num)) {
750 Py_XDECREF(elem);
751 return (0);
754 else if (type == NEWLINE) { /* It's true: we increment the */
755 ++(*line_num); /* line number *after* the newline! */
757 Py_XDECREF(elem);
759 return root;
763 static node*
764 build_node_tree(PyObject *tuple)
766 node* res = 0;
767 PyObject *temp = PySequence_GetItem(tuple, 0);
768 long num = -1;
770 if (temp != NULL)
771 num = PyInt_AsLong(temp);
772 Py_XDECREF(temp);
773 if (ISTERMINAL(num)) {
775 * The tuple is simple, but it doesn't start with a start symbol.
776 * Throw an exception now and be done with it.
778 tuple = Py_BuildValue("os", tuple,
779 "Illegal syntax-tree; cannot start with terminal symbol.");
780 PyErr_SetObject(parser_error, tuple);
781 Py_XDECREF(tuple);
783 else if (ISNONTERMINAL(num)) {
785 * Not efficient, but that can be handled later.
787 int line_num = 0;
788 PyObject *encoding = NULL;
790 if (num == encoding_decl) {
791 encoding = PySequence_GetItem(tuple, 2);
792 /* tuple isn't borrowed anymore here, need to DECREF */
793 tuple = PySequence_GetSlice(tuple, 0, 2);
795 res = PyNode_New(num);
796 if (res != NULL) {
797 if (res != build_node_children(tuple, res, &line_num)) {
798 PyNode_Free(res);
799 res = NULL;
801 if (res && encoding) {
802 Py_ssize_t len;
803 len = PyString_GET_SIZE(encoding) + 1;
804 res->n_str = (char *)PyObject_MALLOC(len);
805 if (res->n_str != NULL)
806 (void) memcpy(res->n_str, PyString_AS_STRING(encoding), len);
807 Py_DECREF(encoding);
808 Py_DECREF(tuple);
812 else {
813 /* The tuple is illegal -- if the number is neither TERMINAL nor
814 * NONTERMINAL, we can't use it. Not sure the implementation
815 * allows this condition, but the API doesn't preclude it.
817 PyObject *err = Py_BuildValue("os", tuple,
818 "Illegal component tuple.");
819 PyErr_SetObject(parser_error, err);
820 Py_XDECREF(err);
823 return (res);
828 * Validation routines used within the validation section:
830 static int validate_terminal(node *terminal, int type, char *string);
832 #define validate_ampersand(ch) validate_terminal(ch, AMPER, "&")
833 #define validate_circumflex(ch) validate_terminal(ch, CIRCUMFLEX, "^")
834 #define validate_colon(ch) validate_terminal(ch, COLON, ":")
835 #define validate_comma(ch) validate_terminal(ch, COMMA, ",")
836 #define validate_dedent(ch) validate_terminal(ch, DEDENT, "")
837 #define validate_equal(ch) validate_terminal(ch, EQUAL, "=")
838 #define validate_indent(ch) validate_terminal(ch, INDENT, (char*)NULL)
839 #define validate_lparen(ch) validate_terminal(ch, LPAR, "(")
840 #define validate_newline(ch) validate_terminal(ch, NEWLINE, (char*)NULL)
841 #define validate_rparen(ch) validate_terminal(ch, RPAR, ")")
842 #define validate_semi(ch) validate_terminal(ch, SEMI, ";")
843 #define validate_star(ch) validate_terminal(ch, STAR, "*")
844 #define validate_vbar(ch) validate_terminal(ch, VBAR, "|")
845 #define validate_doublestar(ch) validate_terminal(ch, DOUBLESTAR, "**")
846 #define validate_dot(ch) validate_terminal(ch, DOT, ".")
847 #define validate_at(ch) validate_terminal(ch, AT, "@")
848 #define validate_name(ch, str) validate_terminal(ch, NAME, str)
850 #define VALIDATER(n) static int validate_##n(node *tree)
852 VALIDATER(node); VALIDATER(small_stmt);
853 VALIDATER(class); VALIDATER(node);
854 VALIDATER(parameters); VALIDATER(suite);
855 VALIDATER(testlist); VALIDATER(varargslist);
856 VALIDATER(fpdef); VALIDATER(fplist);
857 VALIDATER(stmt); VALIDATER(simple_stmt);
858 VALIDATER(expr_stmt); VALIDATER(power);
859 VALIDATER(print_stmt); VALIDATER(del_stmt);
860 VALIDATER(return_stmt); VALIDATER(list_iter);
861 VALIDATER(raise_stmt); VALIDATER(import_stmt);
862 VALIDATER(import_name); VALIDATER(import_from);
863 VALIDATER(global_stmt); VALIDATER(list_if);
864 VALIDATER(assert_stmt); VALIDATER(list_for);
865 VALIDATER(exec_stmt); VALIDATER(compound_stmt);
866 VALIDATER(while); VALIDATER(for);
867 VALIDATER(try); VALIDATER(except_clause);
868 VALIDATER(test); VALIDATER(and_test);
869 VALIDATER(not_test); VALIDATER(comparison);
870 VALIDATER(comp_op); VALIDATER(expr);
871 VALIDATER(xor_expr); VALIDATER(and_expr);
872 VALIDATER(shift_expr); VALIDATER(arith_expr);
873 VALIDATER(term); VALIDATER(factor);
874 VALIDATER(atom); VALIDATER(lambdef);
875 VALIDATER(trailer); VALIDATER(subscript);
876 VALIDATER(subscriptlist); VALIDATER(sliceop);
877 VALIDATER(exprlist); VALIDATER(dictmaker);
878 VALIDATER(arglist); VALIDATER(argument);
879 VALIDATER(listmaker); VALIDATER(yield_stmt);
880 VALIDATER(testlist1); VALIDATER(gen_for);
881 VALIDATER(gen_iter); VALIDATER(gen_if);
882 VALIDATER(testlist_gexp); VALIDATER(yield_expr);
883 VALIDATER(yield_or_testlist); VALIDATER(or_test);
884 VALIDATER(old_test); VALIDATER(old_lambdef);
886 #undef VALIDATER
888 #define is_even(n) (((n) & 1) == 0)
889 #define is_odd(n) (((n) & 1) == 1)
892 static int
893 validate_ntype(node *n, int t)
895 if (TYPE(n) != t) {
896 PyErr_Format(parser_error, "Expected node type %d, got %d.",
897 t, TYPE(n));
898 return 0;
900 return 1;
904 /* Verifies that the number of child nodes is exactly 'num', raising
905 * an exception if it isn't. The exception message does not indicate
906 * the exact number of nodes, allowing this to be used to raise the
907 * "right" exception when the wrong number of nodes is present in a
908 * specific variant of a statement's syntax. This is commonly used
909 * in that fashion.
911 static int
912 validate_numnodes(node *n, int num, const char *const name)
914 if (NCH(n) != num) {
915 PyErr_Format(parser_error,
916 "Illegal number of children for %s node.", name);
917 return 0;
919 return 1;
923 static int
924 validate_terminal(node *terminal, int type, char *string)
926 int res = (validate_ntype(terminal, type)
927 && ((string == 0) || (strcmp(string, STR(terminal)) == 0)));
929 if (!res && !PyErr_Occurred()) {
930 PyErr_Format(parser_error,
931 "Illegal terminal: expected \"%s\"", string);
933 return (res);
937 /* X (',' X) [',']
939 static int
940 validate_repeating_list(node *tree, int ntype, int (*vfunc)(node *),
941 const char *const name)
943 int nch = NCH(tree);
944 int res = (nch && validate_ntype(tree, ntype)
945 && vfunc(CHILD(tree, 0)));
947 if (!res && !PyErr_Occurred())
948 (void) validate_numnodes(tree, 1, name);
949 else {
950 if (is_even(nch))
951 res = validate_comma(CHILD(tree, --nch));
952 if (res && nch > 1) {
953 int pos = 1;
954 for ( ; res && pos < nch; pos += 2)
955 res = (validate_comma(CHILD(tree, pos))
956 && vfunc(CHILD(tree, pos + 1)));
959 return (res);
963 /* validate_class()
965 * classdef:
966 * 'class' NAME ['(' testlist ')'] ':' suite
968 static int
969 validate_class(node *tree)
971 int nch = NCH(tree);
972 int res = (validate_ntype(tree, classdef) &&
973 ((nch == 4) || (nch == 6) || (nch == 7)));
975 if (res) {
976 res = (validate_name(CHILD(tree, 0), "class")
977 && validate_ntype(CHILD(tree, 1), NAME)
978 && validate_colon(CHILD(tree, nch - 2))
979 && validate_suite(CHILD(tree, nch - 1)));
981 else {
982 (void) validate_numnodes(tree, 4, "class");
985 if (res) {
986 if (nch == 7) {
987 res = ((validate_lparen(CHILD(tree, 2)) &&
988 validate_testlist(CHILD(tree, 3)) &&
989 validate_rparen(CHILD(tree, 4))));
991 else if (nch == 6) {
992 res = (validate_lparen(CHILD(tree,2)) &&
993 validate_rparen(CHILD(tree,3)));
996 return (res);
1000 /* if_stmt:
1001 * 'if' test ':' suite ('elif' test ':' suite)* ['else' ':' suite]
1003 static int
1004 validate_if(node *tree)
1006 int nch = NCH(tree);
1007 int res = (validate_ntype(tree, if_stmt)
1008 && (nch >= 4)
1009 && validate_name(CHILD(tree, 0), "if")
1010 && validate_test(CHILD(tree, 1))
1011 && validate_colon(CHILD(tree, 2))
1012 && validate_suite(CHILD(tree, 3)));
1014 if (res && ((nch % 4) == 3)) {
1015 /* ... 'else' ':' suite */
1016 res = (validate_name(CHILD(tree, nch - 3), "else")
1017 && validate_colon(CHILD(tree, nch - 2))
1018 && validate_suite(CHILD(tree, nch - 1)));
1019 nch -= 3;
1021 else if (!res && !PyErr_Occurred())
1022 (void) validate_numnodes(tree, 4, "if");
1023 if ((nch % 4) != 0)
1024 /* Will catch the case for nch < 4 */
1025 res = validate_numnodes(tree, 0, "if");
1026 else if (res && (nch > 4)) {
1027 /* ... ('elif' test ':' suite)+ ... */
1028 int j = 4;
1029 while ((j < nch) && res) {
1030 res = (validate_name(CHILD(tree, j), "elif")
1031 && validate_colon(CHILD(tree, j + 2))
1032 && validate_test(CHILD(tree, j + 1))
1033 && validate_suite(CHILD(tree, j + 3)));
1034 j += 4;
1037 return (res);
1041 /* parameters:
1042 * '(' [varargslist] ')'
1045 static int
1046 validate_parameters(node *tree)
1048 int nch = NCH(tree);
1049 int res = validate_ntype(tree, parameters) && ((nch == 2) || (nch == 3));
1051 if (res) {
1052 res = (validate_lparen(CHILD(tree, 0))
1053 && validate_rparen(CHILD(tree, nch - 1)));
1054 if (res && (nch == 3))
1055 res = validate_varargslist(CHILD(tree, 1));
1057 else {
1058 (void) validate_numnodes(tree, 2, "parameters");
1060 return (res);
1064 /* validate_suite()
1066 * suite:
1067 * simple_stmt
1068 * | NEWLINE INDENT stmt+ DEDENT
1070 static int
1071 validate_suite(node *tree)
1073 int nch = NCH(tree);
1074 int res = (validate_ntype(tree, suite) && ((nch == 1) || (nch >= 4)));
1076 if (res && (nch == 1))
1077 res = validate_simple_stmt(CHILD(tree, 0));
1078 else if (res) {
1079 /* NEWLINE INDENT stmt+ DEDENT */
1080 res = (validate_newline(CHILD(tree, 0))
1081 && validate_indent(CHILD(tree, 1))
1082 && validate_stmt(CHILD(tree, 2))
1083 && validate_dedent(CHILD(tree, nch - 1)));
1085 if (res && (nch > 4)) {
1086 int i = 3;
1087 --nch; /* forget the DEDENT */
1088 for ( ; res && (i < nch); ++i)
1089 res = validate_stmt(CHILD(tree, i));
1091 else if (nch < 4)
1092 res = validate_numnodes(tree, 4, "suite");
1094 return (res);
1098 static int
1099 validate_testlist(node *tree)
1101 return (validate_repeating_list(tree, testlist,
1102 validate_test, "testlist"));
1106 static int
1107 validate_testlist1(node *tree)
1109 return (validate_repeating_list(tree, testlist1,
1110 validate_test, "testlist1"));
1114 static int
1115 validate_testlist_safe(node *tree)
1117 return (validate_repeating_list(tree, testlist_safe,
1118 validate_old_test, "testlist_safe"));
1122 /* '*' NAME [',' '**' NAME] | '**' NAME
1124 static int
1125 validate_varargslist_trailer(node *tree, int start)
1127 int nch = NCH(tree);
1128 int res = 0;
1129 int sym;
1131 if (nch <= start) {
1132 err_string("expected variable argument trailer for varargslist");
1133 return 0;
1135 sym = TYPE(CHILD(tree, start));
1136 if (sym == STAR) {
1138 * ('*' NAME [',' '**' NAME]
1140 if (nch-start == 2)
1141 res = validate_name(CHILD(tree, start+1), NULL);
1142 else if (nch-start == 5)
1143 res = (validate_name(CHILD(tree, start+1), NULL)
1144 && validate_comma(CHILD(tree, start+2))
1145 && validate_doublestar(CHILD(tree, start+3))
1146 && validate_name(CHILD(tree, start+4), NULL));
1148 else if (sym == DOUBLESTAR) {
1150 * '**' NAME
1152 if (nch-start == 2)
1153 res = validate_name(CHILD(tree, start+1), NULL);
1155 if (!res)
1156 err_string("illegal variable argument trailer for varargslist");
1157 return res;
1161 /* validate_varargslist()
1163 * varargslist:
1164 * (fpdef ['=' test] ',')*
1165 * ('*' NAME [',' '**' NAME]
1166 * | '**' NAME)
1167 * | fpdef ['=' test] (',' fpdef ['=' test])* [',']
1170 static int
1171 validate_varargslist(node *tree)
1173 int nch = NCH(tree);
1174 int res = validate_ntype(tree, varargslist) && (nch != 0);
1175 int sym;
1177 if (!res)
1178 return 0;
1179 if (nch < 1) {
1180 err_string("varargslist missing child nodes");
1181 return 0;
1183 sym = TYPE(CHILD(tree, 0));
1184 if (sym == STAR || sym == DOUBLESTAR)
1185 /* whole thing matches:
1186 * '*' NAME [',' '**' NAME] | '**' NAME
1188 res = validate_varargslist_trailer(tree, 0);
1189 else if (sym == fpdef) {
1190 int i = 0;
1192 sym = TYPE(CHILD(tree, nch-1));
1193 if (sym == NAME) {
1195 * (fpdef ['=' test] ',')+
1196 * ('*' NAME [',' '**' NAME]
1197 * | '**' NAME)
1199 /* skip over (fpdef ['=' test] ',')+ */
1200 while (res && (i+2 <= nch)) {
1201 res = validate_fpdef(CHILD(tree, i));
1202 ++i;
1203 if (res && TYPE(CHILD(tree, i)) == EQUAL && (i+2 <= nch)) {
1204 res = (validate_equal(CHILD(tree, i))
1205 && validate_test(CHILD(tree, i+1)));
1206 if (res)
1207 i += 2;
1209 if (res && i < nch) {
1210 res = validate_comma(CHILD(tree, i));
1211 ++i;
1212 if (res && i < nch
1213 && (TYPE(CHILD(tree, i)) == DOUBLESTAR
1214 || TYPE(CHILD(tree, i)) == STAR))
1215 break;
1218 /* ... '*' NAME [',' '**' NAME] | '**' NAME
1219 * i --^^^
1221 if (res)
1222 res = validate_varargslist_trailer(tree, i);
1224 else {
1226 * fpdef ['=' test] (',' fpdef ['=' test])* [',']
1228 /* strip trailing comma node */
1229 if (sym == COMMA) {
1230 res = validate_comma(CHILD(tree, nch-1));
1231 if (!res)
1232 return 0;
1233 --nch;
1236 * fpdef ['=' test] (',' fpdef ['=' test])*
1238 res = validate_fpdef(CHILD(tree, 0));
1239 ++i;
1240 if (res && (i+2 <= nch) && TYPE(CHILD(tree, i)) == EQUAL) {
1241 res = (validate_equal(CHILD(tree, i))
1242 && validate_test(CHILD(tree, i+1)));
1243 i += 2;
1246 * ... (',' fpdef ['=' test])*
1247 * i ---^^^
1249 while (res && (nch - i) >= 2) {
1250 res = (validate_comma(CHILD(tree, i))
1251 && validate_fpdef(CHILD(tree, i+1)));
1252 i += 2;
1253 if (res && (nch - i) >= 2 && TYPE(CHILD(tree, i)) == EQUAL) {
1254 res = (validate_equal(CHILD(tree, i))
1255 && validate_test(CHILD(tree, i+1)));
1256 i += 2;
1259 if (res && nch - i != 0) {
1260 res = 0;
1261 err_string("illegal formation for varargslist");
1265 return res;
1269 /* list_iter: list_for | list_if
1271 static int
1272 validate_list_iter(node *tree)
1274 int res = (validate_ntype(tree, list_iter)
1275 && validate_numnodes(tree, 1, "list_iter"));
1276 if (res && TYPE(CHILD(tree, 0)) == list_for)
1277 res = validate_list_for(CHILD(tree, 0));
1278 else
1279 res = validate_list_if(CHILD(tree, 0));
1281 return res;
1284 /* gen_iter: gen_for | gen_if
1286 static int
1287 validate_gen_iter(node *tree)
1289 int res = (validate_ntype(tree, gen_iter)
1290 && validate_numnodes(tree, 1, "gen_iter"));
1291 if (res && TYPE(CHILD(tree, 0)) == gen_for)
1292 res = validate_gen_for(CHILD(tree, 0));
1293 else
1294 res = validate_gen_if(CHILD(tree, 0));
1296 return res;
1299 /* list_for: 'for' exprlist 'in' testlist [list_iter]
1301 static int
1302 validate_list_for(node *tree)
1304 int nch = NCH(tree);
1305 int res;
1307 if (nch == 5)
1308 res = validate_list_iter(CHILD(tree, 4));
1309 else
1310 res = validate_numnodes(tree, 4, "list_for");
1312 if (res)
1313 res = (validate_name(CHILD(tree, 0), "for")
1314 && validate_exprlist(CHILD(tree, 1))
1315 && validate_name(CHILD(tree, 2), "in")
1316 && validate_testlist_safe(CHILD(tree, 3)));
1318 return res;
1321 /* gen_for: 'for' exprlist 'in' test [gen_iter]
1323 static int
1324 validate_gen_for(node *tree)
1326 int nch = NCH(tree);
1327 int res;
1329 if (nch == 5)
1330 res = validate_gen_iter(CHILD(tree, 4));
1331 else
1332 res = validate_numnodes(tree, 4, "gen_for");
1334 if (res)
1335 res = (validate_name(CHILD(tree, 0), "for")
1336 && validate_exprlist(CHILD(tree, 1))
1337 && validate_name(CHILD(tree, 2), "in")
1338 && validate_or_test(CHILD(tree, 3)));
1340 return res;
1343 /* list_if: 'if' old_test [list_iter]
1345 static int
1346 validate_list_if(node *tree)
1348 int nch = NCH(tree);
1349 int res;
1351 if (nch == 3)
1352 res = validate_list_iter(CHILD(tree, 2));
1353 else
1354 res = validate_numnodes(tree, 2, "list_if");
1356 if (res)
1357 res = (validate_name(CHILD(tree, 0), "if")
1358 && validate_old_test(CHILD(tree, 1)));
1360 return res;
1363 /* gen_if: 'if' old_test [gen_iter]
1365 static int
1366 validate_gen_if(node *tree)
1368 int nch = NCH(tree);
1369 int res;
1371 if (nch == 3)
1372 res = validate_gen_iter(CHILD(tree, 2));
1373 else
1374 res = validate_numnodes(tree, 2, "gen_if");
1376 if (res)
1377 res = (validate_name(CHILD(tree, 0), "if")
1378 && validate_old_test(CHILD(tree, 1)));
1380 return res;
1383 /* validate_fpdef()
1385 * fpdef:
1386 * NAME
1387 * | '(' fplist ')'
1389 static int
1390 validate_fpdef(node *tree)
1392 int nch = NCH(tree);
1393 int res = validate_ntype(tree, fpdef);
1395 if (res) {
1396 if (nch == 1)
1397 res = validate_ntype(CHILD(tree, 0), NAME);
1398 else if (nch == 3)
1399 res = (validate_lparen(CHILD(tree, 0))
1400 && validate_fplist(CHILD(tree, 1))
1401 && validate_rparen(CHILD(tree, 2)));
1402 else
1403 res = validate_numnodes(tree, 1, "fpdef");
1405 return (res);
1409 static int
1410 validate_fplist(node *tree)
1412 return (validate_repeating_list(tree, fplist,
1413 validate_fpdef, "fplist"));
1417 /* simple_stmt | compound_stmt
1420 static int
1421 validate_stmt(node *tree)
1423 int res = (validate_ntype(tree, stmt)
1424 && validate_numnodes(tree, 1, "stmt"));
1426 if (res) {
1427 tree = CHILD(tree, 0);
1429 if (TYPE(tree) == simple_stmt)
1430 res = validate_simple_stmt(tree);
1431 else
1432 res = validate_compound_stmt(tree);
1434 return (res);
1438 /* small_stmt (';' small_stmt)* [';'] NEWLINE
1441 static int
1442 validate_simple_stmt(node *tree)
1444 int nch = NCH(tree);
1445 int res = (validate_ntype(tree, simple_stmt)
1446 && (nch >= 2)
1447 && validate_small_stmt(CHILD(tree, 0))
1448 && validate_newline(CHILD(tree, nch - 1)));
1450 if (nch < 2)
1451 res = validate_numnodes(tree, 2, "simple_stmt");
1452 --nch; /* forget the NEWLINE */
1453 if (res && is_even(nch))
1454 res = validate_semi(CHILD(tree, --nch));
1455 if (res && (nch > 2)) {
1456 int i;
1458 for (i = 1; res && (i < nch); i += 2)
1459 res = (validate_semi(CHILD(tree, i))
1460 && validate_small_stmt(CHILD(tree, i + 1)));
1462 return (res);
1466 static int
1467 validate_small_stmt(node *tree)
1469 int nch = NCH(tree);
1470 int res = validate_numnodes(tree, 1, "small_stmt");
1472 if (res) {
1473 int ntype = TYPE(CHILD(tree, 0));
1475 if ( (ntype == expr_stmt)
1476 || (ntype == print_stmt)
1477 || (ntype == del_stmt)
1478 || (ntype == pass_stmt)
1479 || (ntype == flow_stmt)
1480 || (ntype == import_stmt)
1481 || (ntype == global_stmt)
1482 || (ntype == assert_stmt)
1483 || (ntype == exec_stmt))
1484 res = validate_node(CHILD(tree, 0));
1485 else {
1486 res = 0;
1487 err_string("illegal small_stmt child type");
1490 else if (nch == 1) {
1491 res = 0;
1492 PyErr_Format(parser_error,
1493 "Unrecognized child node of small_stmt: %d.",
1494 TYPE(CHILD(tree, 0)));
1496 return (res);
1500 /* compound_stmt:
1501 * if_stmt | while_stmt | for_stmt | try_stmt | funcdef | classdef | decorated
1503 static int
1504 validate_compound_stmt(node *tree)
1506 int res = (validate_ntype(tree, compound_stmt)
1507 && validate_numnodes(tree, 1, "compound_stmt"));
1508 int ntype;
1510 if (!res)
1511 return (0);
1513 tree = CHILD(tree, 0);
1514 ntype = TYPE(tree);
1515 if ( (ntype == if_stmt)
1516 || (ntype == while_stmt)
1517 || (ntype == for_stmt)
1518 || (ntype == try_stmt)
1519 || (ntype == funcdef)
1520 || (ntype == classdef)
1521 || (ntype == decorated))
1522 res = validate_node(tree);
1523 else {
1524 res = 0;
1525 PyErr_Format(parser_error,
1526 "Illegal compound statement type: %d.", TYPE(tree));
1528 return (res);
1531 static int
1532 validate_yield_or_testlist(node *tree)
1534 if (TYPE(tree) == yield_expr)
1535 return validate_yield_expr(tree);
1536 else
1537 return validate_testlist(tree);
1540 static int
1541 validate_expr_stmt(node *tree)
1543 int j;
1544 int nch = NCH(tree);
1545 int res = (validate_ntype(tree, expr_stmt)
1546 && is_odd(nch)
1547 && validate_testlist(CHILD(tree, 0)));
1549 if (res && nch == 3
1550 && TYPE(CHILD(tree, 1)) == augassign) {
1551 res = validate_numnodes(CHILD(tree, 1), 1, "augassign")
1552 && validate_yield_or_testlist(CHILD(tree, 2));
1554 if (res) {
1555 char *s = STR(CHILD(CHILD(tree, 1), 0));
1557 res = (strcmp(s, "+=") == 0
1558 || strcmp(s, "-=") == 0
1559 || strcmp(s, "*=") == 0
1560 || strcmp(s, "/=") == 0
1561 || strcmp(s, "//=") == 0
1562 || strcmp(s, "%=") == 0
1563 || strcmp(s, "&=") == 0
1564 || strcmp(s, "|=") == 0
1565 || strcmp(s, "^=") == 0
1566 || strcmp(s, "<<=") == 0
1567 || strcmp(s, ">>=") == 0
1568 || strcmp(s, "**=") == 0);
1569 if (!res)
1570 err_string("illegal augmmented assignment operator");
1573 else {
1574 for (j = 1; res && (j < nch); j += 2)
1575 res = validate_equal(CHILD(tree, j))
1576 && validate_yield_or_testlist(CHILD(tree, j + 1));
1578 return (res);
1582 /* print_stmt:
1584 * 'print' ( [ test (',' test)* [','] ]
1585 * | '>>' test [ (',' test)+ [','] ] )
1587 static int
1588 validate_print_stmt(node *tree)
1590 int nch = NCH(tree);
1591 int res = (validate_ntype(tree, print_stmt)
1592 && (nch > 0)
1593 && validate_name(CHILD(tree, 0), "print"));
1595 if (res && nch > 1) {
1596 int sym = TYPE(CHILD(tree, 1));
1597 int i = 1;
1598 int allow_trailing_comma = 1;
1600 if (sym == test)
1601 res = validate_test(CHILD(tree, i++));
1602 else {
1603 if (nch < 3)
1604 res = validate_numnodes(tree, 3, "print_stmt");
1605 else {
1606 res = (validate_ntype(CHILD(tree, i), RIGHTSHIFT)
1607 && validate_test(CHILD(tree, i+1)));
1608 i += 2;
1609 allow_trailing_comma = 0;
1612 if (res) {
1613 /* ... (',' test)* [','] */
1614 while (res && i+2 <= nch) {
1615 res = (validate_comma(CHILD(tree, i))
1616 && validate_test(CHILD(tree, i+1)));
1617 allow_trailing_comma = 1;
1618 i += 2;
1620 if (res && !allow_trailing_comma)
1621 res = validate_numnodes(tree, i, "print_stmt");
1622 else if (res && i < nch)
1623 res = validate_comma(CHILD(tree, i));
1626 return (res);
1630 static int
1631 validate_del_stmt(node *tree)
1633 return (validate_numnodes(tree, 2, "del_stmt")
1634 && validate_name(CHILD(tree, 0), "del")
1635 && validate_exprlist(CHILD(tree, 1)));
1639 static int
1640 validate_return_stmt(node *tree)
1642 int nch = NCH(tree);
1643 int res = (validate_ntype(tree, return_stmt)
1644 && ((nch == 1) || (nch == 2))
1645 && validate_name(CHILD(tree, 0), "return"));
1647 if (res && (nch == 2))
1648 res = validate_testlist(CHILD(tree, 1));
1650 return (res);
1654 static int
1655 validate_raise_stmt(node *tree)
1657 int nch = NCH(tree);
1658 int res = (validate_ntype(tree, raise_stmt)
1659 && ((nch == 1) || (nch == 2) || (nch == 4) || (nch == 6)));
1661 if (res) {
1662 res = validate_name(CHILD(tree, 0), "raise");
1663 if (res && (nch >= 2))
1664 res = validate_test(CHILD(tree, 1));
1665 if (res && nch > 2) {
1666 res = (validate_comma(CHILD(tree, 2))
1667 && validate_test(CHILD(tree, 3)));
1668 if (res && (nch > 4))
1669 res = (validate_comma(CHILD(tree, 4))
1670 && validate_test(CHILD(tree, 5)));
1673 else
1674 (void) validate_numnodes(tree, 2, "raise");
1675 if (res && (nch == 4))
1676 res = (validate_comma(CHILD(tree, 2))
1677 && validate_test(CHILD(tree, 3)));
1679 return (res);
1683 /* yield_expr: 'yield' [testlist]
1685 static int
1686 validate_yield_expr(node *tree)
1688 int nch = NCH(tree);
1689 int res = (validate_ntype(tree, yield_expr)
1690 && ((nch == 1) || (nch == 2))
1691 && validate_name(CHILD(tree, 0), "yield"));
1693 if (res && (nch == 2))
1694 res = validate_testlist(CHILD(tree, 1));
1696 return (res);
1700 /* yield_stmt: yield_expr
1702 static int
1703 validate_yield_stmt(node *tree)
1705 return (validate_ntype(tree, yield_stmt)
1706 && validate_numnodes(tree, 1, "yield_stmt")
1707 && validate_yield_expr(CHILD(tree, 0)));
1711 static int
1712 validate_import_as_name(node *tree)
1714 int nch = NCH(tree);
1715 int ok = validate_ntype(tree, import_as_name);
1717 if (ok) {
1718 if (nch == 1)
1719 ok = validate_name(CHILD(tree, 0), NULL);
1720 else if (nch == 3)
1721 ok = (validate_name(CHILD(tree, 0), NULL)
1722 && validate_name(CHILD(tree, 1), "as")
1723 && validate_name(CHILD(tree, 2), NULL));
1724 else
1725 ok = validate_numnodes(tree, 3, "import_as_name");
1727 return ok;
1731 /* dotted_name: NAME ("." NAME)*
1733 static int
1734 validate_dotted_name(node *tree)
1736 int nch = NCH(tree);
1737 int res = (validate_ntype(tree, dotted_name)
1738 && is_odd(nch)
1739 && validate_name(CHILD(tree, 0), NULL));
1740 int i;
1742 for (i = 1; res && (i < nch); i += 2) {
1743 res = (validate_dot(CHILD(tree, i))
1744 && validate_name(CHILD(tree, i+1), NULL));
1746 return res;
1750 /* dotted_as_name: dotted_name [NAME NAME]
1752 static int
1753 validate_dotted_as_name(node *tree)
1755 int nch = NCH(tree);
1756 int res = validate_ntype(tree, dotted_as_name);
1758 if (res) {
1759 if (nch == 1)
1760 res = validate_dotted_name(CHILD(tree, 0));
1761 else if (nch == 3)
1762 res = (validate_dotted_name(CHILD(tree, 0))
1763 && validate_name(CHILD(tree, 1), "as")
1764 && validate_name(CHILD(tree, 2), NULL));
1765 else {
1766 res = 0;
1767 err_string("illegal number of children for dotted_as_name");
1770 return res;
1774 /* dotted_as_name (',' dotted_as_name)* */
1775 static int
1776 validate_dotted_as_names(node *tree)
1778 int nch = NCH(tree);
1779 int res = is_odd(nch) && validate_dotted_as_name(CHILD(tree, 0));
1780 int i;
1782 for (i = 1; res && (i < nch); i += 2)
1783 res = (validate_comma(CHILD(tree, i))
1784 && validate_dotted_as_name(CHILD(tree, i + 1)));
1785 return (res);
1789 /* import_as_name (',' import_as_name)* [','] */
1790 static int
1791 validate_import_as_names(node *tree)
1793 int nch = NCH(tree);
1794 int res = validate_import_as_name(CHILD(tree, 0));
1795 int i;
1797 for (i = 1; res && (i + 1 < nch); i += 2)
1798 res = (validate_comma(CHILD(tree, i))
1799 && validate_import_as_name(CHILD(tree, i + 1)));
1800 return (res);
1804 /* 'import' dotted_as_names */
1805 static int
1806 validate_import_name(node *tree)
1808 return (validate_ntype(tree, import_name)
1809 && validate_numnodes(tree, 2, "import_name")
1810 && validate_name(CHILD(tree, 0), "import")
1811 && validate_dotted_as_names(CHILD(tree, 1)));
1814 /* Helper function to count the number of leading dots in
1815 * 'from ...module import name'
1817 static int
1818 count_from_dots(node *tree)
1820 int i;
1821 for (i = 0; i < NCH(tree); i++)
1822 if (TYPE(CHILD(tree, i)) != DOT)
1823 break;
1824 return i;
1827 /* 'from' ('.'* dotted_name | '.') 'import' ('*' | '(' import_as_names ')' |
1828 * import_as_names
1830 static int
1831 validate_import_from(node *tree)
1833 int nch = NCH(tree);
1834 int ndots = count_from_dots(tree);
1835 int havename = (TYPE(CHILD(tree, ndots + 1)) == dotted_name);
1836 int offset = ndots + havename;
1837 int res = validate_ntype(tree, import_from)
1838 && (nch >= 4 + ndots)
1839 && validate_name(CHILD(tree, 0), "from")
1840 && (!havename || validate_dotted_name(CHILD(tree, ndots + 1)))
1841 && validate_name(CHILD(tree, offset + 1), "import");
1843 if (res && TYPE(CHILD(tree, offset + 2)) == LPAR)
1844 res = ((nch == offset + 5)
1845 && validate_lparen(CHILD(tree, offset + 2))
1846 && validate_import_as_names(CHILD(tree, offset + 3))
1847 && validate_rparen(CHILD(tree, offset + 4)));
1848 else if (res && TYPE(CHILD(tree, offset + 2)) != STAR)
1849 res = validate_import_as_names(CHILD(tree, offset + 2));
1850 return (res);
1854 /* import_stmt: import_name | import_from */
1855 static int
1856 validate_import_stmt(node *tree)
1858 int nch = NCH(tree);
1859 int res = validate_numnodes(tree, 1, "import_stmt");
1861 if (res) {
1862 int ntype = TYPE(CHILD(tree, 0));
1864 if (ntype == import_name || ntype == import_from)
1865 res = validate_node(CHILD(tree, 0));
1866 else {
1867 res = 0;
1868 err_string("illegal import_stmt child type");
1871 else if (nch == 1) {
1872 res = 0;
1873 PyErr_Format(parser_error,
1874 "Unrecognized child node of import_stmt: %d.",
1875 TYPE(CHILD(tree, 0)));
1877 return (res);
1883 static int
1884 validate_global_stmt(node *tree)
1886 int j;
1887 int nch = NCH(tree);
1888 int res = (validate_ntype(tree, global_stmt)
1889 && is_even(nch) && (nch >= 2));
1891 if (!res && !PyErr_Occurred())
1892 err_string("illegal global statement");
1894 if (res)
1895 res = (validate_name(CHILD(tree, 0), "global")
1896 && validate_ntype(CHILD(tree, 1), NAME));
1897 for (j = 2; res && (j < nch); j += 2)
1898 res = (validate_comma(CHILD(tree, j))
1899 && validate_ntype(CHILD(tree, j + 1), NAME));
1901 return (res);
1905 /* exec_stmt:
1907 * 'exec' expr ['in' test [',' test]]
1909 static int
1910 validate_exec_stmt(node *tree)
1912 int nch = NCH(tree);
1913 int res = (validate_ntype(tree, exec_stmt)
1914 && ((nch == 2) || (nch == 4) || (nch == 6))
1915 && validate_name(CHILD(tree, 0), "exec")
1916 && validate_expr(CHILD(tree, 1)));
1918 if (!res && !PyErr_Occurred())
1919 err_string("illegal exec statement");
1920 if (res && (nch > 2))
1921 res = (validate_name(CHILD(tree, 2), "in")
1922 && validate_test(CHILD(tree, 3)));
1923 if (res && (nch == 6))
1924 res = (validate_comma(CHILD(tree, 4))
1925 && validate_test(CHILD(tree, 5)));
1927 return (res);
1931 /* assert_stmt:
1933 * 'assert' test [',' test]
1935 static int
1936 validate_assert_stmt(node *tree)
1938 int nch = NCH(tree);
1939 int res = (validate_ntype(tree, assert_stmt)
1940 && ((nch == 2) || (nch == 4))
1941 && (validate_name(CHILD(tree, 0), "assert"))
1942 && validate_test(CHILD(tree, 1)));
1944 if (!res && !PyErr_Occurred())
1945 err_string("illegal assert statement");
1946 if (res && (nch > 2))
1947 res = (validate_comma(CHILD(tree, 2))
1948 && validate_test(CHILD(tree, 3)));
1950 return (res);
1954 static int
1955 validate_while(node *tree)
1957 int nch = NCH(tree);
1958 int res = (validate_ntype(tree, while_stmt)
1959 && ((nch == 4) || (nch == 7))
1960 && validate_name(CHILD(tree, 0), "while")
1961 && validate_test(CHILD(tree, 1))
1962 && validate_colon(CHILD(tree, 2))
1963 && validate_suite(CHILD(tree, 3)));
1965 if (res && (nch == 7))
1966 res = (validate_name(CHILD(tree, 4), "else")
1967 && validate_colon(CHILD(tree, 5))
1968 && validate_suite(CHILD(tree, 6)));
1970 return (res);
1974 static int
1975 validate_for(node *tree)
1977 int nch = NCH(tree);
1978 int res = (validate_ntype(tree, for_stmt)
1979 && ((nch == 6) || (nch == 9))
1980 && validate_name(CHILD(tree, 0), "for")
1981 && validate_exprlist(CHILD(tree, 1))
1982 && validate_name(CHILD(tree, 2), "in")
1983 && validate_testlist(CHILD(tree, 3))
1984 && validate_colon(CHILD(tree, 4))
1985 && validate_suite(CHILD(tree, 5)));
1987 if (res && (nch == 9))
1988 res = (validate_name(CHILD(tree, 6), "else")
1989 && validate_colon(CHILD(tree, 7))
1990 && validate_suite(CHILD(tree, 8)));
1992 return (res);
1996 /* try_stmt:
1997 * 'try' ':' suite (except_clause ':' suite)+ ['else' ':' suite]
1998 * | 'try' ':' suite 'finally' ':' suite
2001 static int
2002 validate_try(node *tree)
2004 int nch = NCH(tree);
2005 int pos = 3;
2006 int res = (validate_ntype(tree, try_stmt)
2007 && (nch >= 6) && ((nch % 3) == 0));
2009 if (res)
2010 res = (validate_name(CHILD(tree, 0), "try")
2011 && validate_colon(CHILD(tree, 1))
2012 && validate_suite(CHILD(tree, 2))
2013 && validate_colon(CHILD(tree, nch - 2))
2014 && validate_suite(CHILD(tree, nch - 1)));
2015 else if (!PyErr_Occurred()) {
2016 const char* name = "except";
2017 if (TYPE(CHILD(tree, nch - 3)) != except_clause)
2018 name = STR(CHILD(tree, nch - 3));
2020 PyErr_Format(parser_error,
2021 "Illegal number of children for try/%s node.", name);
2023 /* Skip past except_clause sections: */
2024 while (res && (TYPE(CHILD(tree, pos)) == except_clause)) {
2025 res = (validate_except_clause(CHILD(tree, pos))
2026 && validate_colon(CHILD(tree, pos + 1))
2027 && validate_suite(CHILD(tree, pos + 2)));
2028 pos += 3;
2030 if (res && (pos < nch)) {
2031 res = validate_ntype(CHILD(tree, pos), NAME);
2032 if (res && (strcmp(STR(CHILD(tree, pos)), "finally") == 0))
2033 res = (validate_numnodes(tree, 6, "try/finally")
2034 && validate_colon(CHILD(tree, 4))
2035 && validate_suite(CHILD(tree, 5)));
2036 else if (res) {
2037 if (nch == (pos + 3)) {
2038 res = ((strcmp(STR(CHILD(tree, pos)), "except") == 0)
2039 || (strcmp(STR(CHILD(tree, pos)), "else") == 0));
2040 if (!res)
2041 err_string("illegal trailing triple in try statement");
2043 else if (nch == (pos + 6)) {
2044 res = (validate_name(CHILD(tree, pos), "except")
2045 && validate_colon(CHILD(tree, pos + 1))
2046 && validate_suite(CHILD(tree, pos + 2))
2047 && validate_name(CHILD(tree, pos + 3), "else"));
2049 else
2050 res = validate_numnodes(tree, pos + 3, "try/except");
2053 return (res);
2057 static int
2058 validate_except_clause(node *tree)
2060 int nch = NCH(tree);
2061 int res = (validate_ntype(tree, except_clause)
2062 && ((nch == 1) || (nch == 2) || (nch == 4))
2063 && validate_name(CHILD(tree, 0), "except"));
2065 if (res && (nch > 1))
2066 res = validate_test(CHILD(tree, 1));
2067 if (res && (nch == 4))
2068 res = (validate_comma(CHILD(tree, 2))
2069 && validate_test(CHILD(tree, 3)));
2071 return (res);
2075 static int
2076 validate_test(node *tree)
2078 int nch = NCH(tree);
2079 int res = validate_ntype(tree, test) && is_odd(nch);
2081 if (res && (TYPE(CHILD(tree, 0)) == lambdef))
2082 res = ((nch == 1)
2083 && validate_lambdef(CHILD(tree, 0)));
2084 else if (res) {
2085 res = validate_or_test(CHILD(tree, 0));
2086 res = (res && (nch == 1 || (nch == 5 &&
2087 validate_name(CHILD(tree, 1), "if") &&
2088 validate_or_test(CHILD(tree, 2)) &&
2089 validate_name(CHILD(tree, 3), "else") &&
2090 validate_test(CHILD(tree, 4)))));
2092 return (res);
2095 static int
2096 validate_old_test(node *tree)
2098 int nch = NCH(tree);
2099 int res = validate_ntype(tree, old_test) && (nch == 1);
2101 if (res && (TYPE(CHILD(tree, 0)) == old_lambdef))
2102 res = (validate_old_lambdef(CHILD(tree, 0)));
2103 else if (res) {
2104 res = (validate_or_test(CHILD(tree, 0)));
2106 return (res);
2109 static int
2110 validate_or_test(node *tree)
2112 int nch = NCH(tree);
2113 int res = validate_ntype(tree, or_test) && is_odd(nch);
2115 if (res) {
2116 int pos;
2117 res = validate_and_test(CHILD(tree, 0));
2118 for (pos = 1; res && (pos < nch); pos += 2)
2119 res = (validate_name(CHILD(tree, pos), "or")
2120 && validate_and_test(CHILD(tree, pos + 1)));
2122 return (res);
2126 static int
2127 validate_and_test(node *tree)
2129 int pos;
2130 int nch = NCH(tree);
2131 int res = (validate_ntype(tree, and_test)
2132 && is_odd(nch)
2133 && validate_not_test(CHILD(tree, 0)));
2135 for (pos = 1; res && (pos < nch); pos += 2)
2136 res = (validate_name(CHILD(tree, pos), "and")
2137 && validate_not_test(CHILD(tree, 0)));
2139 return (res);
2143 static int
2144 validate_not_test(node *tree)
2146 int nch = NCH(tree);
2147 int res = validate_ntype(tree, not_test) && ((nch == 1) || (nch == 2));
2149 if (res) {
2150 if (nch == 2)
2151 res = (validate_name(CHILD(tree, 0), "not")
2152 && validate_not_test(CHILD(tree, 1)));
2153 else if (nch == 1)
2154 res = validate_comparison(CHILD(tree, 0));
2156 return (res);
2160 static int
2161 validate_comparison(node *tree)
2163 int pos;
2164 int nch = NCH(tree);
2165 int res = (validate_ntype(tree, comparison)
2166 && is_odd(nch)
2167 && validate_expr(CHILD(tree, 0)));
2169 for (pos = 1; res && (pos < nch); pos += 2)
2170 res = (validate_comp_op(CHILD(tree, pos))
2171 && validate_expr(CHILD(tree, pos + 1)));
2173 return (res);
2177 static int
2178 validate_comp_op(node *tree)
2180 int res = 0;
2181 int nch = NCH(tree);
2183 if (!validate_ntype(tree, comp_op))
2184 return (0);
2185 if (nch == 1) {
2187 * Only child will be a terminal with a well-defined symbolic name
2188 * or a NAME with a string of either 'is' or 'in'
2190 tree = CHILD(tree, 0);
2191 switch (TYPE(tree)) {
2192 case LESS:
2193 case GREATER:
2194 case EQEQUAL:
2195 case EQUAL:
2196 case LESSEQUAL:
2197 case GREATEREQUAL:
2198 case NOTEQUAL:
2199 res = 1;
2200 break;
2201 case NAME:
2202 res = ((strcmp(STR(tree), "in") == 0)
2203 || (strcmp(STR(tree), "is") == 0));
2204 if (!res) {
2205 PyErr_Format(parser_error,
2206 "illegal operator '%s'", STR(tree));
2208 break;
2209 default:
2210 err_string("illegal comparison operator type");
2211 break;
2214 else if ((res = validate_numnodes(tree, 2, "comp_op")) != 0) {
2215 res = (validate_ntype(CHILD(tree, 0), NAME)
2216 && validate_ntype(CHILD(tree, 1), NAME)
2217 && (((strcmp(STR(CHILD(tree, 0)), "is") == 0)
2218 && (strcmp(STR(CHILD(tree, 1)), "not") == 0))
2219 || ((strcmp(STR(CHILD(tree, 0)), "not") == 0)
2220 && (strcmp(STR(CHILD(tree, 1)), "in") == 0))));
2221 if (!res && !PyErr_Occurred())
2222 err_string("unknown comparison operator");
2224 return (res);
2228 static int
2229 validate_expr(node *tree)
2231 int j;
2232 int nch = NCH(tree);
2233 int res = (validate_ntype(tree, expr)
2234 && is_odd(nch)
2235 && validate_xor_expr(CHILD(tree, 0)));
2237 for (j = 2; res && (j < nch); j += 2)
2238 res = (validate_xor_expr(CHILD(tree, j))
2239 && validate_vbar(CHILD(tree, j - 1)));
2241 return (res);
2245 static int
2246 validate_xor_expr(node *tree)
2248 int j;
2249 int nch = NCH(tree);
2250 int res = (validate_ntype(tree, xor_expr)
2251 && is_odd(nch)
2252 && validate_and_expr(CHILD(tree, 0)));
2254 for (j = 2; res && (j < nch); j += 2)
2255 res = (validate_circumflex(CHILD(tree, j - 1))
2256 && validate_and_expr(CHILD(tree, j)));
2258 return (res);
2262 static int
2263 validate_and_expr(node *tree)
2265 int pos;
2266 int nch = NCH(tree);
2267 int res = (validate_ntype(tree, and_expr)
2268 && is_odd(nch)
2269 && validate_shift_expr(CHILD(tree, 0)));
2271 for (pos = 1; res && (pos < nch); pos += 2)
2272 res = (validate_ampersand(CHILD(tree, pos))
2273 && validate_shift_expr(CHILD(tree, pos + 1)));
2275 return (res);
2279 static int
2280 validate_chain_two_ops(node *tree, int (*termvalid)(node *), int op1, int op2)
2282 int pos = 1;
2283 int nch = NCH(tree);
2284 int res = (is_odd(nch)
2285 && (*termvalid)(CHILD(tree, 0)));
2287 for ( ; res && (pos < nch); pos += 2) {
2288 if (TYPE(CHILD(tree, pos)) != op1)
2289 res = validate_ntype(CHILD(tree, pos), op2);
2290 if (res)
2291 res = (*termvalid)(CHILD(tree, pos + 1));
2293 return (res);
2297 static int
2298 validate_shift_expr(node *tree)
2300 return (validate_ntype(tree, shift_expr)
2301 && validate_chain_two_ops(tree, validate_arith_expr,
2302 LEFTSHIFT, RIGHTSHIFT));
2306 static int
2307 validate_arith_expr(node *tree)
2309 return (validate_ntype(tree, arith_expr)
2310 && validate_chain_two_ops(tree, validate_term, PLUS, MINUS));
2314 static int
2315 validate_term(node *tree)
2317 int pos = 1;
2318 int nch = NCH(tree);
2319 int res = (validate_ntype(tree, term)
2320 && is_odd(nch)
2321 && validate_factor(CHILD(tree, 0)));
2323 for ( ; res && (pos < nch); pos += 2)
2324 res = (((TYPE(CHILD(tree, pos)) == STAR)
2325 || (TYPE(CHILD(tree, pos)) == SLASH)
2326 || (TYPE(CHILD(tree, pos)) == DOUBLESLASH)
2327 || (TYPE(CHILD(tree, pos)) == PERCENT))
2328 && validate_factor(CHILD(tree, pos + 1)));
2330 return (res);
2334 /* factor:
2336 * factor: ('+'|'-'|'~') factor | power
2338 static int
2339 validate_factor(node *tree)
2341 int nch = NCH(tree);
2342 int res = (validate_ntype(tree, factor)
2343 && (((nch == 2)
2344 && ((TYPE(CHILD(tree, 0)) == PLUS)
2345 || (TYPE(CHILD(tree, 0)) == MINUS)
2346 || (TYPE(CHILD(tree, 0)) == TILDE))
2347 && validate_factor(CHILD(tree, 1)))
2348 || ((nch == 1)
2349 && validate_power(CHILD(tree, 0)))));
2350 return (res);
2354 /* power:
2356 * power: atom trailer* ('**' factor)*
2358 static int
2359 validate_power(node *tree)
2361 int pos = 1;
2362 int nch = NCH(tree);
2363 int res = (validate_ntype(tree, power) && (nch >= 1)
2364 && validate_atom(CHILD(tree, 0)));
2366 while (res && (pos < nch) && (TYPE(CHILD(tree, pos)) == trailer))
2367 res = validate_trailer(CHILD(tree, pos++));
2368 if (res && (pos < nch)) {
2369 if (!is_even(nch - pos)) {
2370 err_string("illegal number of nodes for 'power'");
2371 return (0);
2373 for ( ; res && (pos < (nch - 1)); pos += 2)
2374 res = (validate_doublestar(CHILD(tree, pos))
2375 && validate_factor(CHILD(tree, pos + 1)));
2377 return (res);
2381 static int
2382 validate_atom(node *tree)
2384 int pos;
2385 int nch = NCH(tree);
2386 int res = validate_ntype(tree, atom);
2388 if (res && nch < 1)
2389 res = validate_numnodes(tree, nch+1, "atom");
2390 if (res) {
2391 switch (TYPE(CHILD(tree, 0))) {
2392 case LPAR:
2393 res = ((nch <= 3)
2394 && (validate_rparen(CHILD(tree, nch - 1))));
2396 if (res && (nch == 3)) {
2397 if (TYPE(CHILD(tree, 1))==yield_expr)
2398 res = validate_yield_expr(CHILD(tree, 1));
2399 else
2400 res = validate_testlist_gexp(CHILD(tree, 1));
2402 break;
2403 case LSQB:
2404 if (nch == 2)
2405 res = validate_ntype(CHILD(tree, 1), RSQB);
2406 else if (nch == 3)
2407 res = (validate_listmaker(CHILD(tree, 1))
2408 && validate_ntype(CHILD(tree, 2), RSQB));
2409 else {
2410 res = 0;
2411 err_string("illegal list display atom");
2413 break;
2414 case LBRACE:
2415 res = ((nch <= 3)
2416 && validate_ntype(CHILD(tree, nch - 1), RBRACE));
2418 if (res && (nch == 3))
2419 res = validate_dictmaker(CHILD(tree, 1));
2420 break;
2421 case BACKQUOTE:
2422 res = ((nch == 3)
2423 && validate_testlist1(CHILD(tree, 1))
2424 && validate_ntype(CHILD(tree, 2), BACKQUOTE));
2425 break;
2426 case NAME:
2427 case NUMBER:
2428 res = (nch == 1);
2429 break;
2430 case STRING:
2431 for (pos = 1; res && (pos < nch); ++pos)
2432 res = validate_ntype(CHILD(tree, pos), STRING);
2433 break;
2434 default:
2435 res = 0;
2436 break;
2439 return (res);
2443 /* listmaker:
2444 * test ( list_for | (',' test)* [','] )
2446 static int
2447 validate_listmaker(node *tree)
2449 int nch = NCH(tree);
2450 int ok = nch;
2452 if (nch == 0)
2453 err_string("missing child nodes of listmaker");
2454 else
2455 ok = validate_test(CHILD(tree, 0));
2458 * list_for | (',' test)* [',']
2460 if (nch == 2 && TYPE(CHILD(tree, 1)) == list_for)
2461 ok = validate_list_for(CHILD(tree, 1));
2462 else {
2463 /* (',' test)* [','] */
2464 int i = 1;
2465 while (ok && nch - i >= 2) {
2466 ok = (validate_comma(CHILD(tree, i))
2467 && validate_test(CHILD(tree, i+1)));
2468 i += 2;
2470 if (ok && i == nch-1)
2471 ok = validate_comma(CHILD(tree, i));
2472 else if (i != nch) {
2473 ok = 0;
2474 err_string("illegal trailing nodes for listmaker");
2477 return ok;
2480 /* testlist_gexp:
2481 * test ( gen_for | (',' test)* [','] )
2483 static int
2484 validate_testlist_gexp(node *tree)
2486 int nch = NCH(tree);
2487 int ok = nch;
2489 if (nch == 0)
2490 err_string("missing child nodes of testlist_gexp");
2491 else {
2492 ok = validate_test(CHILD(tree, 0));
2496 * gen_for | (',' test)* [',']
2498 if (nch == 2 && TYPE(CHILD(tree, 1)) == gen_for)
2499 ok = validate_gen_for(CHILD(tree, 1));
2500 else {
2501 /* (',' test)* [','] */
2502 int i = 1;
2503 while (ok && nch - i >= 2) {
2504 ok = (validate_comma(CHILD(tree, i))
2505 && validate_test(CHILD(tree, i+1)));
2506 i += 2;
2508 if (ok && i == nch-1)
2509 ok = validate_comma(CHILD(tree, i));
2510 else if (i != nch) {
2511 ok = 0;
2512 err_string("illegal trailing nodes for testlist_gexp");
2515 return ok;
2518 /* decorator:
2519 * '@' dotted_name [ '(' [arglist] ')' ] NEWLINE
2521 static int
2522 validate_decorator(node *tree)
2524 int ok;
2525 int nch = NCH(tree);
2526 ok = (validate_ntype(tree, decorator) &&
2527 (nch == 3 || nch == 5 || nch == 6) &&
2528 validate_at(CHILD(tree, 0)) &&
2529 validate_dotted_name(CHILD(tree, 1)) &&
2530 validate_newline(RCHILD(tree, -1)));
2532 if (ok && nch != 3) {
2533 ok = (validate_lparen(CHILD(tree, 2)) &&
2534 validate_rparen(RCHILD(tree, -2)));
2536 if (ok && nch == 6)
2537 ok = validate_arglist(CHILD(tree, 3));
2540 return ok;
2543 /* decorators:
2544 * decorator+
2546 static int
2547 validate_decorators(node *tree)
2549 int i, nch, ok;
2550 nch = NCH(tree);
2551 ok = validate_ntype(tree, decorators) && nch >= 1;
2553 for (i = 0; ok && i < nch; ++i)
2554 ok = validate_decorator(CHILD(tree, i));
2556 return ok;
2559 /* funcdef:
2561 * -5 -4 -3 -2 -1
2562 * 'def' NAME parameters ':' suite
2564 static int
2565 validate_funcdef(node *tree)
2567 int nch = NCH(tree);
2568 int ok = (validate_ntype(tree, funcdef)
2569 && (nch == 5)
2570 && validate_name(RCHILD(tree, -5), "def")
2571 && validate_ntype(RCHILD(tree, -4), NAME)
2572 && validate_colon(RCHILD(tree, -2))
2573 && validate_parameters(RCHILD(tree, -3))
2574 && validate_suite(RCHILD(tree, -1)));
2575 return ok;
2579 /* decorated
2580 * decorators (classdef | funcdef)
2582 static int
2583 validate_decorated(node *tree)
2585 int nch = NCH(tree);
2586 int ok = (validate_ntype(tree, decorated)
2587 && (nch == 2)
2588 && validate_decorators(RCHILD(tree, -2))
2589 && (validate_funcdef(RCHILD(tree, -1))
2590 || validate_class(RCHILD(tree, -1)))
2592 return ok;
2595 static int
2596 validate_lambdef(node *tree)
2598 int nch = NCH(tree);
2599 int res = (validate_ntype(tree, lambdef)
2600 && ((nch == 3) || (nch == 4))
2601 && validate_name(CHILD(tree, 0), "lambda")
2602 && validate_colon(CHILD(tree, nch - 2))
2603 && validate_test(CHILD(tree, nch - 1)));
2605 if (res && (nch == 4))
2606 res = validate_varargslist(CHILD(tree, 1));
2607 else if (!res && !PyErr_Occurred())
2608 (void) validate_numnodes(tree, 3, "lambdef");
2610 return (res);
2614 static int
2615 validate_old_lambdef(node *tree)
2617 int nch = NCH(tree);
2618 int res = (validate_ntype(tree, old_lambdef)
2619 && ((nch == 3) || (nch == 4))
2620 && validate_name(CHILD(tree, 0), "lambda")
2621 && validate_colon(CHILD(tree, nch - 2))
2622 && validate_test(CHILD(tree, nch - 1)));
2624 if (res && (nch == 4))
2625 res = validate_varargslist(CHILD(tree, 1));
2626 else if (!res && !PyErr_Occurred())
2627 (void) validate_numnodes(tree, 3, "old_lambdef");
2629 return (res);
2633 /* arglist:
2635 * (argument ',')* (argument [','] | '*' test [',' '**' test] | '**' test)
2637 static int
2638 validate_arglist(node *tree)
2640 int nch = NCH(tree);
2641 int i = 0;
2642 int ok = 1;
2644 if (nch <= 0)
2645 /* raise the right error from having an invalid number of children */
2646 return validate_numnodes(tree, nch + 1, "arglist");
2648 if (nch > 1) {
2649 for (i=0; i<nch; i++) {
2650 if (TYPE(CHILD(tree, i)) == argument) {
2651 node *ch = CHILD(tree, i);
2652 if (NCH(ch) == 2 && TYPE(CHILD(ch, 1)) == gen_for) {
2653 err_string("need '(', ')' for generator expression");
2654 return 0;
2660 while (ok && nch-i >= 2) {
2661 /* skip leading (argument ',') */
2662 ok = (validate_argument(CHILD(tree, i))
2663 && validate_comma(CHILD(tree, i+1)));
2664 if (ok)
2665 i += 2;
2666 else
2667 PyErr_Clear();
2669 ok = 1;
2670 if (nch-i > 0) {
2672 * argument | '*' test [',' '**' test] | '**' test
2674 int sym = TYPE(CHILD(tree, i));
2676 if (sym == argument) {
2677 ok = validate_argument(CHILD(tree, i));
2678 if (ok && i+1 != nch) {
2679 err_string("illegal arglist specification"
2680 " (extra stuff on end)");
2681 ok = 0;
2684 else if (sym == STAR) {
2685 ok = validate_star(CHILD(tree, i));
2686 if (ok && (nch-i == 2))
2687 ok = validate_test(CHILD(tree, i+1));
2688 else if (ok && (nch-i == 5))
2689 ok = (validate_test(CHILD(tree, i+1))
2690 && validate_comma(CHILD(tree, i+2))
2691 && validate_doublestar(CHILD(tree, i+3))
2692 && validate_test(CHILD(tree, i+4)));
2693 else {
2694 err_string("illegal use of '*' in arglist");
2695 ok = 0;
2698 else if (sym == DOUBLESTAR) {
2699 if (nch-i == 2)
2700 ok = (validate_doublestar(CHILD(tree, i))
2701 && validate_test(CHILD(tree, i+1)));
2702 else {
2703 err_string("illegal use of '**' in arglist");
2704 ok = 0;
2707 else {
2708 err_string("illegal arglist specification");
2709 ok = 0;
2712 return (ok);
2717 /* argument:
2719 * [test '='] test [gen_for]
2721 static int
2722 validate_argument(node *tree)
2724 int nch = NCH(tree);
2725 int res = (validate_ntype(tree, argument)
2726 && ((nch == 1) || (nch == 2) || (nch == 3))
2727 && validate_test(CHILD(tree, 0)));
2729 if (res && (nch == 2))
2730 res = validate_gen_for(CHILD(tree, 1));
2731 else if (res && (nch == 3))
2732 res = (validate_equal(CHILD(tree, 1))
2733 && validate_test(CHILD(tree, 2)));
2735 return (res);
2740 /* trailer:
2742 * '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
2744 static int
2745 validate_trailer(node *tree)
2747 int nch = NCH(tree);
2748 int res = validate_ntype(tree, trailer) && ((nch == 2) || (nch == 3));
2750 if (res) {
2751 switch (TYPE(CHILD(tree, 0))) {
2752 case LPAR:
2753 res = validate_rparen(CHILD(tree, nch - 1));
2754 if (res && (nch == 3))
2755 res = validate_arglist(CHILD(tree, 1));
2756 break;
2757 case LSQB:
2758 res = (validate_numnodes(tree, 3, "trailer")
2759 && validate_subscriptlist(CHILD(tree, 1))
2760 && validate_ntype(CHILD(tree, 2), RSQB));
2761 break;
2762 case DOT:
2763 res = (validate_numnodes(tree, 2, "trailer")
2764 && validate_ntype(CHILD(tree, 1), NAME));
2765 break;
2766 default:
2767 res = 0;
2768 break;
2771 else {
2772 (void) validate_numnodes(tree, 2, "trailer");
2774 return (res);
2778 /* subscriptlist:
2780 * subscript (',' subscript)* [',']
2782 static int
2783 validate_subscriptlist(node *tree)
2785 return (validate_repeating_list(tree, subscriptlist,
2786 validate_subscript, "subscriptlist"));
2790 /* subscript:
2792 * '.' '.' '.' | test | [test] ':' [test] [sliceop]
2794 static int
2795 validate_subscript(node *tree)
2797 int offset = 0;
2798 int nch = NCH(tree);
2799 int res = validate_ntype(tree, subscript) && (nch >= 1) && (nch <= 4);
2801 if (!res) {
2802 if (!PyErr_Occurred())
2803 err_string("invalid number of arguments for subscript node");
2804 return (0);
2806 if (TYPE(CHILD(tree, 0)) == DOT)
2807 /* take care of ('.' '.' '.') possibility */
2808 return (validate_numnodes(tree, 3, "subscript")
2809 && validate_dot(CHILD(tree, 0))
2810 && validate_dot(CHILD(tree, 1))
2811 && validate_dot(CHILD(tree, 2)));
2812 if (nch == 1) {
2813 if (TYPE(CHILD(tree, 0)) == test)
2814 res = validate_test(CHILD(tree, 0));
2815 else
2816 res = validate_colon(CHILD(tree, 0));
2817 return (res);
2819 /* Must be [test] ':' [test] [sliceop],
2820 * but at least one of the optional components will
2821 * be present, but we don't know which yet.
2823 if ((TYPE(CHILD(tree, 0)) != COLON) || (nch == 4)) {
2824 res = validate_test(CHILD(tree, 0));
2825 offset = 1;
2827 if (res)
2828 res = validate_colon(CHILD(tree, offset));
2829 if (res) {
2830 int rem = nch - ++offset;
2831 if (rem) {
2832 if (TYPE(CHILD(tree, offset)) == test) {
2833 res = validate_test(CHILD(tree, offset));
2834 ++offset;
2835 --rem;
2837 if (res && rem)
2838 res = validate_sliceop(CHILD(tree, offset));
2841 return (res);
2845 static int
2846 validate_sliceop(node *tree)
2848 int nch = NCH(tree);
2849 int res = ((nch == 1) || validate_numnodes(tree, 2, "sliceop"))
2850 && validate_ntype(tree, sliceop);
2851 if (!res && !PyErr_Occurred()) {
2852 res = validate_numnodes(tree, 1, "sliceop");
2854 if (res)
2855 res = validate_colon(CHILD(tree, 0));
2856 if (res && (nch == 2))
2857 res = validate_test(CHILD(tree, 1));
2859 return (res);
2863 static int
2864 validate_exprlist(node *tree)
2866 return (validate_repeating_list(tree, exprlist,
2867 validate_expr, "exprlist"));
2871 static int
2872 validate_dictmaker(node *tree)
2874 int nch = NCH(tree);
2875 int res = (validate_ntype(tree, dictmaker)
2876 && (nch >= 3)
2877 && validate_test(CHILD(tree, 0))
2878 && validate_colon(CHILD(tree, 1))
2879 && validate_test(CHILD(tree, 2)));
2881 if (res && ((nch % 4) == 0))
2882 res = validate_comma(CHILD(tree, --nch));
2883 else if (res)
2884 res = ((nch % 4) == 3);
2886 if (res && (nch > 3)) {
2887 int pos = 3;
2888 /* ( ',' test ':' test )* */
2889 while (res && (pos < nch)) {
2890 res = (validate_comma(CHILD(tree, pos))
2891 && validate_test(CHILD(tree, pos + 1))
2892 && validate_colon(CHILD(tree, pos + 2))
2893 && validate_test(CHILD(tree, pos + 3)));
2894 pos += 4;
2897 return (res);
2901 static int
2902 validate_eval_input(node *tree)
2904 int pos;
2905 int nch = NCH(tree);
2906 int res = (validate_ntype(tree, eval_input)
2907 && (nch >= 2)
2908 && validate_testlist(CHILD(tree, 0))
2909 && validate_ntype(CHILD(tree, nch - 1), ENDMARKER));
2911 for (pos = 1; res && (pos < (nch - 1)); ++pos)
2912 res = validate_ntype(CHILD(tree, pos), NEWLINE);
2914 return (res);
2918 static int
2919 validate_node(node *tree)
2921 int nch = 0; /* num. children on current node */
2922 int res = 1; /* result value */
2923 node* next = 0; /* node to process after this one */
2925 while (res && (tree != 0)) {
2926 nch = NCH(tree);
2927 next = 0;
2928 switch (TYPE(tree)) {
2930 * Definition nodes.
2932 case funcdef:
2933 res = validate_funcdef(tree);
2934 break;
2935 case classdef:
2936 res = validate_class(tree);
2937 break;
2938 case decorated:
2939 res = validate_decorated(tree);
2940 break;
2942 * "Trivial" parse tree nodes.
2943 * (Why did I call these trivial?)
2945 case stmt:
2946 res = validate_stmt(tree);
2947 break;
2948 case small_stmt:
2950 * expr_stmt | print_stmt | del_stmt | pass_stmt | flow_stmt
2951 * | import_stmt | global_stmt | exec_stmt | assert_stmt
2953 res = validate_small_stmt(tree);
2954 break;
2955 case flow_stmt:
2956 res = (validate_numnodes(tree, 1, "flow_stmt")
2957 && ((TYPE(CHILD(tree, 0)) == break_stmt)
2958 || (TYPE(CHILD(tree, 0)) == continue_stmt)
2959 || (TYPE(CHILD(tree, 0)) == yield_stmt)
2960 || (TYPE(CHILD(tree, 0)) == return_stmt)
2961 || (TYPE(CHILD(tree, 0)) == raise_stmt)));
2962 if (res)
2963 next = CHILD(tree, 0);
2964 else if (nch == 1)
2965 err_string("illegal flow_stmt type");
2966 break;
2967 case yield_stmt:
2968 res = validate_yield_stmt(tree);
2969 break;
2971 * Compound statements.
2973 case simple_stmt:
2974 res = validate_simple_stmt(tree);
2975 break;
2976 case compound_stmt:
2977 res = validate_compound_stmt(tree);
2978 break;
2980 * Fundamental statements.
2982 case expr_stmt:
2983 res = validate_expr_stmt(tree);
2984 break;
2985 case print_stmt:
2986 res = validate_print_stmt(tree);
2987 break;
2988 case del_stmt:
2989 res = validate_del_stmt(tree);
2990 break;
2991 case pass_stmt:
2992 res = (validate_numnodes(tree, 1, "pass")
2993 && validate_name(CHILD(tree, 0), "pass"));
2994 break;
2995 case break_stmt:
2996 res = (validate_numnodes(tree, 1, "break")
2997 && validate_name(CHILD(tree, 0), "break"));
2998 break;
2999 case continue_stmt:
3000 res = (validate_numnodes(tree, 1, "continue")
3001 && validate_name(CHILD(tree, 0), "continue"));
3002 break;
3003 case return_stmt:
3004 res = validate_return_stmt(tree);
3005 break;
3006 case raise_stmt:
3007 res = validate_raise_stmt(tree);
3008 break;
3009 case import_stmt:
3010 res = validate_import_stmt(tree);
3011 break;
3012 case import_name:
3013 res = validate_import_name(tree);
3014 break;
3015 case import_from:
3016 res = validate_import_from(tree);
3017 break;
3018 case global_stmt:
3019 res = validate_global_stmt(tree);
3020 break;
3021 case exec_stmt:
3022 res = validate_exec_stmt(tree);
3023 break;
3024 case assert_stmt:
3025 res = validate_assert_stmt(tree);
3026 break;
3027 case if_stmt:
3028 res = validate_if(tree);
3029 break;
3030 case while_stmt:
3031 res = validate_while(tree);
3032 break;
3033 case for_stmt:
3034 res = validate_for(tree);
3035 break;
3036 case try_stmt:
3037 res = validate_try(tree);
3038 break;
3039 case suite:
3040 res = validate_suite(tree);
3041 break;
3043 * Expression nodes.
3045 case testlist:
3046 res = validate_testlist(tree);
3047 break;
3048 case yield_expr:
3049 res = validate_yield_expr(tree);
3050 break;
3051 case testlist1:
3052 res = validate_testlist1(tree);
3053 break;
3054 case test:
3055 res = validate_test(tree);
3056 break;
3057 case and_test:
3058 res = validate_and_test(tree);
3059 break;
3060 case not_test:
3061 res = validate_not_test(tree);
3062 break;
3063 case comparison:
3064 res = validate_comparison(tree);
3065 break;
3066 case exprlist:
3067 res = validate_exprlist(tree);
3068 break;
3069 case comp_op:
3070 res = validate_comp_op(tree);
3071 break;
3072 case expr:
3073 res = validate_expr(tree);
3074 break;
3075 case xor_expr:
3076 res = validate_xor_expr(tree);
3077 break;
3078 case and_expr:
3079 res = validate_and_expr(tree);
3080 break;
3081 case shift_expr:
3082 res = validate_shift_expr(tree);
3083 break;
3084 case arith_expr:
3085 res = validate_arith_expr(tree);
3086 break;
3087 case term:
3088 res = validate_term(tree);
3089 break;
3090 case factor:
3091 res = validate_factor(tree);
3092 break;
3093 case power:
3094 res = validate_power(tree);
3095 break;
3096 case atom:
3097 res = validate_atom(tree);
3098 break;
3100 default:
3101 /* Hopefully never reached! */
3102 err_string("unrecognized node type");
3103 res = 0;
3104 break;
3106 tree = next;
3108 return (res);
3112 static int
3113 validate_expr_tree(node *tree)
3115 int res = validate_eval_input(tree);
3117 if (!res && !PyErr_Occurred())
3118 err_string("could not validate expression tuple");
3120 return (res);
3124 /* file_input:
3125 * (NEWLINE | stmt)* ENDMARKER
3127 static int
3128 validate_file_input(node *tree)
3130 int j;
3131 int nch = NCH(tree) - 1;
3132 int res = ((nch >= 0)
3133 && validate_ntype(CHILD(tree, nch), ENDMARKER));
3135 for (j = 0; res && (j < nch); ++j) {
3136 if (TYPE(CHILD(tree, j)) == stmt)
3137 res = validate_stmt(CHILD(tree, j));
3138 else
3139 res = validate_newline(CHILD(tree, j));
3141 /* This stays in to prevent any internal failures from getting to the
3142 * user. Hopefully, this won't be needed. If a user reports getting
3143 * this, we have some debugging to do.
3145 if (!res && !PyErr_Occurred())
3146 err_string("VALIDATION FAILURE: report this to the maintainer!");
3148 return (res);
3151 static int
3152 validate_encoding_decl(node *tree)
3154 int nch = NCH(tree);
3155 int res = ((nch == 1)
3156 && validate_file_input(CHILD(tree, 0)));
3158 if (!res && !PyErr_Occurred())
3159 err_string("Error Parsing encoding_decl");
3161 return res;
3164 static PyObject*
3165 pickle_constructor = NULL;
3168 static PyObject*
3169 parser__pickler(PyObject *self, PyObject *args)
3171 NOTE(ARGUNUSED(self))
3172 PyObject *result = NULL;
3173 PyObject *st = NULL;
3174 PyObject *empty_dict = NULL;
3176 if (PyArg_ParseTuple(args, "O!:_pickler", &PyST_Type, &st)) {
3177 PyObject *newargs;
3178 PyObject *tuple;
3180 if ((empty_dict = PyDict_New()) == NULL)
3181 goto finally;
3182 if ((newargs = Py_BuildValue("Oi", st, 1)) == NULL)
3183 goto finally;
3184 tuple = parser_st2tuple((PyST_Object*)NULL, newargs, empty_dict);
3185 if (tuple != NULL) {
3186 result = Py_BuildValue("O(O)", pickle_constructor, tuple);
3187 Py_DECREF(tuple);
3189 Py_DECREF(empty_dict);
3190 Py_DECREF(newargs);
3192 finally:
3193 Py_XDECREF(empty_dict);
3195 return (result);
3199 /* Functions exported by this module. Most of this should probably
3200 * be converted into an ST object with methods, but that is better
3201 * done directly in Python, allowing subclasses to be created directly.
3202 * We'd really have to write a wrapper around it all anyway to allow
3203 * inheritance.
3205 static PyMethodDef parser_functions[] = {
3206 {"ast2tuple", (PyCFunction)parser_st2tuple, PUBLIC_METHOD_TYPE,
3207 PyDoc_STR("Creates a tuple-tree representation of an ST.")},
3208 {"ast2list", (PyCFunction)parser_st2list, PUBLIC_METHOD_TYPE,
3209 PyDoc_STR("Creates a list-tree representation of an ST.")},
3210 {"compileast", (PyCFunction)parser_compilest, PUBLIC_METHOD_TYPE,
3211 PyDoc_STR("Compiles an ST object into a code object.")},
3212 {"compilest", (PyCFunction)parser_compilest, PUBLIC_METHOD_TYPE,
3213 PyDoc_STR("Compiles an ST object into a code object.")},
3214 {"expr", (PyCFunction)parser_expr, PUBLIC_METHOD_TYPE,
3215 PyDoc_STR("Creates an ST object from an expression.")},
3216 {"isexpr", (PyCFunction)parser_isexpr, PUBLIC_METHOD_TYPE,
3217 PyDoc_STR("Determines if an ST object was created from an expression.")},
3218 {"issuite", (PyCFunction)parser_issuite, PUBLIC_METHOD_TYPE,
3219 PyDoc_STR("Determines if an ST object was created from a suite.")},
3220 {"suite", (PyCFunction)parser_suite, PUBLIC_METHOD_TYPE,
3221 PyDoc_STR("Creates an ST object from a suite.")},
3222 {"sequence2ast", (PyCFunction)parser_tuple2st, PUBLIC_METHOD_TYPE,
3223 PyDoc_STR("Creates an ST object from a tree representation.")},
3224 {"sequence2st", (PyCFunction)parser_tuple2st, PUBLIC_METHOD_TYPE,
3225 PyDoc_STR("Creates an ST object from a tree representation.")},
3226 {"st2tuple", (PyCFunction)parser_st2tuple, PUBLIC_METHOD_TYPE,
3227 PyDoc_STR("Creates a tuple-tree representation of an ST.")},
3228 {"st2list", (PyCFunction)parser_st2list, PUBLIC_METHOD_TYPE,
3229 PyDoc_STR("Creates a list-tree representation of an ST.")},
3230 {"tuple2ast", (PyCFunction)parser_tuple2st, PUBLIC_METHOD_TYPE,
3231 PyDoc_STR("Creates an ST object from a tree representation.")},
3232 {"tuple2st", (PyCFunction)parser_tuple2st, PUBLIC_METHOD_TYPE,
3233 PyDoc_STR("Creates an ST object from a tree representation.")},
3235 /* private stuff: support pickle module */
3236 {"_pickler", (PyCFunction)parser__pickler, METH_VARARGS,
3237 PyDoc_STR("Returns the pickle magic to allow ST objects to be pickled.")},
3239 {NULL, NULL, 0, NULL}
3243 PyMODINIT_FUNC initparser(void); /* supply a prototype */
3245 PyMODINIT_FUNC
3246 initparser(void)
3248 PyObject *module, *copyreg;
3250 Py_TYPE(&PyST_Type) = &PyType_Type;
3251 module = Py_InitModule("parser", parser_functions);
3252 if (module == NULL)
3253 return;
3255 if (parser_error == 0)
3256 parser_error = PyErr_NewException("parser.ParserError", NULL, NULL);
3258 if (parser_error == 0)
3259 /* caller will check PyErr_Occurred() */
3260 return;
3261 /* CAUTION: The code next used to skip bumping the refcount on
3262 * parser_error. That's a disaster if initparser() gets called more
3263 * than once. By incref'ing, we ensure that each module dict that
3264 * gets created owns its reference to the shared parser_error object,
3265 * and the file static parser_error vrbl owns a reference too.
3267 Py_INCREF(parser_error);
3268 if (PyModule_AddObject(module, "ParserError", parser_error) != 0)
3269 return;
3271 Py_INCREF(&PyST_Type);
3272 PyModule_AddObject(module, "ASTType", (PyObject*)&PyST_Type);
3273 Py_INCREF(&PyST_Type);
3274 PyModule_AddObject(module, "STType", (PyObject*)&PyST_Type);
3276 PyModule_AddStringConstant(module, "__copyright__",
3277 parser_copyright_string);
3278 PyModule_AddStringConstant(module, "__doc__",
3279 parser_doc_string);
3280 PyModule_AddStringConstant(module, "__version__",
3281 parser_version_string);
3283 /* Register to support pickling.
3284 * If this fails, the import of this module will fail because an
3285 * exception will be raised here; should we clear the exception?
3287 copyreg = PyImport_ImportModuleNoBlock("copy_reg");
3288 if (copyreg != NULL) {
3289 PyObject *func, *pickler;
3291 func = PyObject_GetAttrString(copyreg, "pickle");
3292 pickle_constructor = PyObject_GetAttrString(module, "sequence2st");
3293 pickler = PyObject_GetAttrString(module, "_pickler");
3294 Py_XINCREF(pickle_constructor);
3295 if ((func != NULL) && (pickle_constructor != NULL)
3296 && (pickler != NULL)) {
3297 PyObject *res;
3299 res = PyObject_CallFunctionObjArgs(func, &PyST_Type, pickler,
3300 pickle_constructor, NULL);
3301 Py_XDECREF(res);
3303 Py_XDECREF(func);
3304 Py_XDECREF(pickle_constructor);
3305 Py_XDECREF(pickler);
3306 Py_DECREF(copyreg);