This commit was manufactured by cvs2svn to create tag 'r201'.
[python/dscho.git] / Modules / parsermodule.c
blob27d193c48597d1a3d2c7b5fd8341a8fa2fecf32d
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 "token.h" /* token definitions */
32 /* ISTERMINAL() / ISNONTERMINAL() */
33 #include "compile.h" /* PyNode_Compile() */
35 #ifdef lint
36 #include <note.h>
37 #else
38 #define NOTE(x)
39 #endif
41 #ifdef macintosh
42 char *strdup(char *);
43 #endif
45 /* String constants used to initialize module attributes.
48 static char*
49 parser_copyright_string
50 = "Copyright 1995-1996 by Virginia Polytechnic Institute & State\n\
51 University, Blacksburg, Virginia, USA, and Fred L. Drake, Jr., Reston,\n\
52 Virginia, USA. Portions copyright 1991-1995 by Stichting Mathematisch\n\
53 Centrum, Amsterdam, The Netherlands.";
56 static char*
57 parser_doc_string
58 = "This is an interface to Python's internal parser.";
60 static char*
61 parser_version_string = "0.5";
64 typedef PyObject* (*SeqMaker) (int length);
65 typedef int (*SeqInserter) (PyObject* sequence,
66 int index,
67 PyObject* element);
69 /* The function below is copyrighted by Stichting Mathematisch Centrum. The
70 * original copyright statement is included below, and continues to apply
71 * in full to the function immediately following. All other material is
72 * original, copyrighted by Fred L. Drake, Jr. and Virginia Polytechnic
73 * Institute and State University. Changes were made to comply with the
74 * new naming conventions. Added arguments to provide support for creating
75 * lists as well as tuples, and optionally including the line numbers.
79 static PyObject*
80 node2tuple(node *n, /* node to convert */
81 SeqMaker mkseq, /* create sequence */
82 SeqInserter addelem, /* func. to add elem. in seq. */
83 int lineno) /* include line numbers? */
85 if (n == NULL) {
86 Py_INCREF(Py_None);
87 return (Py_None);
89 if (ISNONTERMINAL(TYPE(n))) {
90 int i;
91 PyObject *v;
92 PyObject *w;
94 v = mkseq(1 + NCH(n));
95 if (v == NULL)
96 return (v);
97 w = PyInt_FromLong(TYPE(n));
98 if (w == NULL) {
99 Py_DECREF(v);
100 return ((PyObject*) NULL);
102 (void) addelem(v, 0, w);
103 for (i = 0; i < NCH(n); i++) {
104 w = node2tuple(CHILD(n, i), mkseq, addelem, lineno);
105 if (w == NULL) {
106 Py_DECREF(v);
107 return ((PyObject*) NULL);
109 (void) addelem(v, i+1, w);
111 return (v);
113 else if (ISTERMINAL(TYPE(n))) {
114 PyObject *result = mkseq(2 + lineno);
115 if (result != NULL) {
116 (void) addelem(result, 0, PyInt_FromLong(TYPE(n)));
117 (void) addelem(result, 1, PyString_FromString(STR(n)));
118 if (lineno == 1)
119 (void) addelem(result, 2, PyInt_FromLong(n->n_lineno));
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 ast_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. ;-)
141 * The PyAST_FRAGMENT type is not currently supported. Maybe not useful?
142 * Haven't decided yet.
145 #define PyAST_EXPR 1
146 #define PyAST_SUITE 2
147 #define PyAST_FRAGMENT 3
150 /* These are the internal objects and definitions required to implement the
151 * AST type. Most of the internal names are more reminiscent of the 'old'
152 * naming style, but the code uses the new naming convention.
155 static PyObject*
156 parser_error = 0;
159 typedef struct _PyAST_Object {
160 PyObject_HEAD /* standard object header */
161 node* ast_node; /* the node* returned by the parser */
162 int ast_type; /* EXPR or SUITE ? */
163 } PyAST_Object;
166 staticforward void
167 parser_free(PyAST_Object *ast);
169 staticforward int
170 parser_compare(PyAST_Object *left, PyAST_Object *right);
172 staticforward PyObject *
173 parser_getattr(PyObject *self, char *name);
176 static
177 PyTypeObject PyAST_Type = {
178 PyObject_HEAD_INIT(NULL)
180 "ast", /* tp_name */
181 (int) sizeof(PyAST_Object), /* tp_basicsize */
182 0, /* tp_itemsize */
183 (destructor)parser_free, /* tp_dealloc */
184 0, /* tp_print */
185 parser_getattr, /* tp_getattr */
186 0, /* tp_setattr */
187 (cmpfunc)parser_compare, /* tp_compare */
188 0, /* tp_repr */
189 0, /* tp_as_number */
190 0, /* tp_as_sequence */
191 0, /* tp_as_mapping */
192 0, /* tp_hash */
193 0, /* tp_call */
194 0, /* tp_str */
195 0, /* tp_getattro */
196 0, /* tp_setattro */
198 /* Functions to access object as input/output buffer */
199 0, /* tp_as_buffer */
201 Py_TPFLAGS_DEFAULT, /* tp_flags */
203 /* __doc__ */
204 "Intermediate representation of a Python parse tree."
205 }; /* PyAST_Type */
208 static int
209 parser_compare_nodes(node *left, node *right)
211 int j;
213 if (TYPE(left) < TYPE(right))
214 return (-1);
216 if (TYPE(right) < TYPE(left))
217 return (1);
219 if (ISTERMINAL(TYPE(left)))
220 return (strcmp(STR(left), STR(right)));
222 if (NCH(left) < NCH(right))
223 return (-1);
225 if (NCH(right) < NCH(left))
226 return (1);
228 for (j = 0; j < NCH(left); ++j) {
229 int v = parser_compare_nodes(CHILD(left, j), CHILD(right, j));
231 if (v != 0)
232 return (v);
234 return (0);
238 /* int parser_compare(PyAST_Object* left, PyAST_Object* right)
240 * Comparison function used by the Python operators ==, !=, <, >, <=, >=
241 * This really just wraps a call to parser_compare_nodes() with some easy
242 * checks and protection code.
245 static int
246 parser_compare(PyAST_Object *left, PyAST_Object *right)
248 if (left == right)
249 return (0);
251 if ((left == 0) || (right == 0))
252 return (-1);
254 return (parser_compare_nodes(left->ast_node, right->ast_node));
258 /* parser_newastobject(node* ast)
260 * Allocates a new Python object representing an AST. This is simply the
261 * 'wrapper' object that holds a node* and allows it to be passed around in
262 * Python code.
265 static PyObject*
266 parser_newastobject(node *ast, int type)
268 PyAST_Object* o = PyObject_New(PyAST_Object, &PyAST_Type);
270 if (o != 0) {
271 o->ast_node = ast;
272 o->ast_type = type;
274 else {
275 PyNode_Free(ast);
277 return ((PyObject*)o);
281 /* void parser_free(PyAST_Object* ast)
283 * This is called by a del statement that reduces the reference count to 0.
286 static void
287 parser_free(PyAST_Object *ast)
289 PyNode_Free(ast->ast_node);
290 PyObject_Del(ast);
294 /* parser_ast2tuple(PyObject* self, PyObject* args, PyObject* kw)
296 * This provides conversion from a node* to a tuple object that can be
297 * returned to the Python-level caller. The AST object is not modified.
300 static PyObject*
301 parser_ast2tuple(PyAST_Object *self, PyObject *args, PyObject *kw)
303 PyObject *line_option = 0;
304 PyObject *res = 0;
305 int ok;
307 static char *keywords[] = {"ast", "line_info", NULL};
309 if (self == NULL) {
310 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|O:ast2tuple", keywords,
311 &PyAST_Type, &self, &line_option);
313 else
314 ok = PyArg_ParseTupleAndKeywords(args, kw, "|O:totuple", &keywords[1],
315 &line_option);
316 if (ok != 0) {
317 int lineno = 0;
318 if (line_option != NULL) {
319 lineno = (PyObject_IsTrue(line_option) != 0) ? 1 : 0;
322 * Convert AST into a tuple representation. Use Guido's function,
323 * since it's known to work already.
325 res = node2tuple(((PyAST_Object*)self)->ast_node,
326 PyTuple_New, PyTuple_SetItem, lineno);
328 return (res);
332 /* parser_ast2list(PyObject* self, PyObject* args, PyObject* kw)
334 * This provides conversion from a node* to a list object that can be
335 * returned to the Python-level caller. The AST object is not modified.
338 static PyObject*
339 parser_ast2list(PyAST_Object *self, PyObject *args, PyObject *kw)
341 PyObject *line_option = 0;
342 PyObject *res = 0;
343 int ok;
345 static char *keywords[] = {"ast", "line_info", NULL};
347 if (self == NULL)
348 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|O:ast2list", keywords,
349 &PyAST_Type, &self, &line_option);
350 else
351 ok = PyArg_ParseTupleAndKeywords(args, kw, "|O:tolist", &keywords[1],
352 &line_option);
353 if (ok) {
354 int lineno = 0;
355 if (line_option != 0) {
356 lineno = PyObject_IsTrue(line_option) ? 1 : 0;
359 * Convert AST into a tuple representation. Use Guido's function,
360 * since it's known to work already.
362 res = node2tuple(self->ast_node,
363 PyList_New, PyList_SetItem, lineno);
365 return (res);
369 /* parser_compileast(PyObject* self, PyObject* args)
371 * This function creates code objects from the parse tree represented by
372 * the passed-in data object. An optional file name is passed in as well.
375 static PyObject*
376 parser_compileast(PyAST_Object *self, PyObject *args, PyObject *kw)
378 PyObject* res = 0;
379 char* str = "<ast>";
380 int ok;
382 static char *keywords[] = {"ast", "filename", NULL};
384 if (self == NULL)
385 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|s:compileast", keywords,
386 &PyAST_Type, &self, &str);
387 else
388 ok = PyArg_ParseTupleAndKeywords(args, kw, "|s:compile", &keywords[1],
389 &str);
391 if (ok)
392 res = (PyObject *)PyNode_Compile(self->ast_node, str);
394 return (res);
398 /* PyObject* parser_isexpr(PyObject* self, PyObject* args)
399 * PyObject* parser_issuite(PyObject* self, PyObject* args)
401 * Checks the passed-in AST object to determine if it is an expression or
402 * a statement suite, respectively. The return is a Python truth value.
405 static PyObject*
406 parser_isexpr(PyAST_Object *self, PyObject *args, PyObject *kw)
408 PyObject* res = 0;
409 int ok;
411 static char *keywords[] = {"ast", NULL};
413 if (self == NULL)
414 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!:isexpr", keywords,
415 &PyAST_Type, &self);
416 else
417 ok = PyArg_ParseTupleAndKeywords(args, kw, ":isexpr", &keywords[1]);
419 if (ok) {
420 /* Check to see if the AST represents an expression or not. */
421 res = (self->ast_type == PyAST_EXPR) ? Py_True : Py_False;
422 Py_INCREF(res);
424 return (res);
428 static PyObject*
429 parser_issuite(PyAST_Object *self, PyObject *args, PyObject *kw)
431 PyObject* res = 0;
432 int ok;
434 static char *keywords[] = {"ast", NULL};
436 if (self == NULL)
437 ok = PyArg_ParseTupleAndKeywords(args, kw, "O!:issuite", keywords,
438 &PyAST_Type, &self);
439 else
440 ok = PyArg_ParseTupleAndKeywords(args, kw, ":issuite", &keywords[1]);
442 if (ok) {
443 /* Check to see if the AST represents an expression or not. */
444 res = (self->ast_type == PyAST_EXPR) ? Py_False : Py_True;
445 Py_INCREF(res);
447 return (res);
451 #define PUBLIC_METHOD_TYPE (METH_VARARGS|METH_KEYWORDS)
453 static PyMethodDef
454 parser_methods[] = {
455 {"compile", (PyCFunction)parser_compileast, PUBLIC_METHOD_TYPE,
456 "Compile this AST object into a code object."},
457 {"isexpr", (PyCFunction)parser_isexpr, PUBLIC_METHOD_TYPE,
458 "Determines if this AST object was created from an expression."},
459 {"issuite", (PyCFunction)parser_issuite, PUBLIC_METHOD_TYPE,
460 "Determines if this AST object was created from a suite."},
461 {"tolist", (PyCFunction)parser_ast2list, PUBLIC_METHOD_TYPE,
462 "Creates a list-tree representation of this AST."},
463 {"totuple", (PyCFunction)parser_ast2tuple, PUBLIC_METHOD_TYPE,
464 "Creates a tuple-tree representation of this AST."},
466 {NULL, NULL, 0, NULL}
470 static PyObject*
471 parser_getattr(PyObject *self, char *name)
473 return (Py_FindMethod(parser_methods, self, name));
477 /* err_string(char* message)
479 * Sets the error string for an exception of type ParserError.
482 static void
483 err_string(char *message)
485 PyErr_SetString(parser_error, message);
489 /* PyObject* parser_do_parse(PyObject* args, int type)
491 * Internal function to actually execute the parse and return the result if
492 * successful, or set an exception if not.
495 static PyObject*
496 parser_do_parse(PyObject *args, PyObject *kw, char *argspec, int type)
498 char* string = 0;
499 PyObject* res = 0;
501 static char *keywords[] = {"source", NULL};
503 if (PyArg_ParseTupleAndKeywords(args, kw, argspec, keywords, &string)) {
504 node* n = PyParser_SimpleParseString(string,
505 (type == PyAST_EXPR)
506 ? eval_input : file_input);
508 if (n != 0)
509 res = parser_newastobject(n, type);
510 else
511 err_string("Could not parse string.");
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(PyAST_Object *self, PyObject *args, PyObject *kw)
528 NOTE(ARGUNUSED(self))
529 return (parser_do_parse(args, kw, "s:expr", PyAST_EXPR));
533 static PyObject*
534 parser_suite(PyAST_Object *self, PyObject *args, PyObject *kw)
536 NOTE(ARGUNUSED(self))
537 return (parser_do_parse(args, kw, "s:suite", PyAST_SUITE));
542 /* This is the messy part of the code. Conversion from a tuple to an AST
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 AST types to include funcdefs and
553 * lambdadefs to take advantage of the optimizer, recognizing those ASTs
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 staticforward node* build_node_tree(PyObject *tuple);
560 staticforward int validate_expr_tree(node *tree);
561 staticforward int validate_file_input(node *tree);
564 /* PyObject* parser_tuple2ast(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 AST 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 PyAST_FRAGMENT
574 * AST objects.
577 static PyObject*
578 parser_tuple2ast(PyAST_Object *self, PyObject *args, PyObject *kw)
580 NOTE(ARGUNUSED(self))
581 PyObject *ast = 0;
582 PyObject *tuple;
583 node *tree;
585 static char *keywords[] = {"sequence", NULL};
587 if (!PyArg_ParseTupleAndKeywords(args, kw, "O:sequence2ast", keywords,
588 &tuple))
589 return (0);
590 if (!PySequence_Check(tuple)) {
591 PyErr_SetString(PyExc_ValueError,
592 "sequence2ast() 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 ast = parser_newastobject(tree, PyAST_EXPR);
606 else if (start_sym == file_input) {
607 /* This looks like an exec form so far. */
608 if (validate_file_input(tree))
609 ast = parser_newastobject(tree, PyAST_SUITE);
611 else {
612 /* This is a fragment, at best. */
613 PyNode_Free(tree);
614 err_string("Parse tree does not use a valid start symbol.");
617 /* Make sure we throw an exception on all errors. We should never
618 * get this, but we'd do well to be sure something is done.
620 if ((ast == 0) && !PyErr_Occurred())
621 err_string("Unspecified ast error occurred.");
623 return (ast);
627 /* node* build_node_children()
629 * Iterate across the children of the current non-terminal node and build
630 * their structures. If successful, return the root of this portion of
631 * the tree, otherwise, 0. Any required exception will be specified already,
632 * and no memory will have been deallocated.
635 static node*
636 build_node_children(PyObject *tuple, node *root, int *line_num)
638 int len = PyObject_Size(tuple);
639 int i;
641 for (i = 1; i < len; ++i) {
642 /* elem must always be a sequence, however simple */
643 PyObject* elem = PySequence_GetItem(tuple, i);
644 int ok = elem != NULL;
645 long type = 0;
646 char *strn = 0;
648 if (ok)
649 ok = PySequence_Check(elem);
650 if (ok) {
651 PyObject *temp = PySequence_GetItem(elem, 0);
652 if (temp == NULL)
653 ok = 0;
654 else {
655 ok = PyInt_Check(temp);
656 if (ok)
657 type = PyInt_AS_LONG(temp);
658 Py_DECREF(temp);
661 if (!ok) {
662 PyErr_SetObject(parser_error,
663 Py_BuildValue("os", elem,
664 "Illegal node construct."));
665 Py_XDECREF(elem);
666 return (0);
668 if (ISTERMINAL(type)) {
669 int len = PyObject_Size(elem);
670 PyObject *temp;
672 if ((len != 2) && (len != 3)) {
673 err_string("Terminal nodes must have 2 or 3 entries.");
674 return 0;
676 temp = PySequence_GetItem(elem, 1);
677 if (temp == NULL)
678 return 0;
679 if (!PyString_Check(temp)) {
680 PyErr_Format(parser_error,
681 "Second item in terminal node must be a string,"
682 " found %s.",
683 ((PyTypeObject*)PyObject_Type(temp))->tp_name);
684 Py_DECREF(temp);
685 return 0;
687 if (len == 3) {
688 PyObject *o = PySequence_GetItem(elem, 2);
689 if (o != NULL) {
690 if (PyInt_Check(o))
691 *line_num = PyInt_AS_LONG(o);
692 else {
693 PyErr_Format(parser_error,
694 "Third item in terminal node must be an"
695 " integer, found %s.",
696 ((PyTypeObject*)PyObject_Type(temp))->tp_name);
697 Py_DECREF(o);
698 Py_DECREF(temp);
699 return 0;
701 Py_DECREF(o);
704 len = PyString_GET_SIZE(temp) + 1;
705 strn = (char *)PyMem_MALLOC(len);
706 if (strn != NULL)
707 (void) memcpy(strn, PyString_AS_STRING(temp), len);
708 Py_DECREF(temp);
710 else if (!ISNONTERMINAL(type)) {
712 * It has to be one or the other; this is an error.
713 * Throw an exception.
715 PyErr_SetObject(parser_error,
716 Py_BuildValue("os", elem, "Unknown node type."));
717 Py_XDECREF(elem);
718 return (0);
720 PyNode_AddChild(root, type, strn, *line_num);
722 if (ISNONTERMINAL(type)) {
723 node* new_child = CHILD(root, i - 1);
725 if (new_child != build_node_children(elem, new_child, line_num)) {
726 Py_XDECREF(elem);
727 return (0);
730 else if (type == NEWLINE) { /* It's true: we increment the */
731 ++(*line_num); /* line number *after* the newline! */
733 Py_XDECREF(elem);
735 return (root);
739 static node*
740 build_node_tree(PyObject *tuple)
742 node* res = 0;
743 PyObject *temp = PySequence_GetItem(tuple, 0);
744 long num = -1;
746 if (temp != NULL)
747 num = PyInt_AsLong(temp);
748 Py_XDECREF(temp);
749 if (ISTERMINAL(num)) {
751 * The tuple is simple, but it doesn't start with a start symbol.
752 * Throw an exception now and be done with it.
754 tuple = Py_BuildValue("os", tuple,
755 "Illegal ast tuple; cannot start with terminal symbol.");
756 PyErr_SetObject(parser_error, tuple);
758 else if (ISNONTERMINAL(num)) {
760 * Not efficient, but that can be handled later.
762 int line_num = 0;
764 res = PyNode_New(num);
765 if (res != build_node_children(tuple, res, &line_num)) {
766 PyNode_Free(res);
767 res = 0;
770 else
771 /* The tuple is illegal -- if the number is neither TERMINAL nor
772 * NONTERMINAL, we can't use it. Not sure the implementation
773 * allows this condition, but the API doesn't preclude it.
775 PyErr_SetObject(parser_error,
776 Py_BuildValue("os", tuple,
777 "Illegal component tuple."));
779 return (res);
784 * Validation routines used within the validation section:
786 staticforward int validate_terminal(node *terminal, int type, char *string);
788 #define validate_ampersand(ch) validate_terminal(ch, AMPER, "&")
789 #define validate_circumflex(ch) validate_terminal(ch, CIRCUMFLEX, "^")
790 #define validate_colon(ch) validate_terminal(ch, COLON, ":")
791 #define validate_comma(ch) validate_terminal(ch, COMMA, ",")
792 #define validate_dedent(ch) validate_terminal(ch, DEDENT, "")
793 #define validate_equal(ch) validate_terminal(ch, EQUAL, "=")
794 #define validate_indent(ch) validate_terminal(ch, INDENT, (char*)NULL)
795 #define validate_lparen(ch) validate_terminal(ch, LPAR, "(")
796 #define validate_newline(ch) validate_terminal(ch, NEWLINE, (char*)NULL)
797 #define validate_rparen(ch) validate_terminal(ch, RPAR, ")")
798 #define validate_semi(ch) validate_terminal(ch, SEMI, ";")
799 #define validate_star(ch) validate_terminal(ch, STAR, "*")
800 #define validate_vbar(ch) validate_terminal(ch, VBAR, "|")
801 #define validate_doublestar(ch) validate_terminal(ch, DOUBLESTAR, "**")
802 #define validate_dot(ch) validate_terminal(ch, DOT, ".")
803 #define validate_name(ch, str) validate_terminal(ch, NAME, str)
805 #define VALIDATER(n) static int validate_##n(node *tree)
807 VALIDATER(node); VALIDATER(small_stmt);
808 VALIDATER(class); VALIDATER(node);
809 VALIDATER(parameters); VALIDATER(suite);
810 VALIDATER(testlist); VALIDATER(varargslist);
811 VALIDATER(fpdef); VALIDATER(fplist);
812 VALIDATER(stmt); VALIDATER(simple_stmt);
813 VALIDATER(expr_stmt); VALIDATER(power);
814 VALIDATER(print_stmt); VALIDATER(del_stmt);
815 VALIDATER(return_stmt); VALIDATER(list_iter);
816 VALIDATER(raise_stmt); VALIDATER(import_stmt);
817 VALIDATER(global_stmt); VALIDATER(list_if);
818 VALIDATER(assert_stmt); VALIDATER(list_for);
819 VALIDATER(exec_stmt); VALIDATER(compound_stmt);
820 VALIDATER(while); VALIDATER(for);
821 VALIDATER(try); VALIDATER(except_clause);
822 VALIDATER(test); VALIDATER(and_test);
823 VALIDATER(not_test); VALIDATER(comparison);
824 VALIDATER(comp_op); VALIDATER(expr);
825 VALIDATER(xor_expr); VALIDATER(and_expr);
826 VALIDATER(shift_expr); VALIDATER(arith_expr);
827 VALIDATER(term); VALIDATER(factor);
828 VALIDATER(atom); VALIDATER(lambdef);
829 VALIDATER(trailer); VALIDATER(subscript);
830 VALIDATER(subscriptlist); VALIDATER(sliceop);
831 VALIDATER(exprlist); VALIDATER(dictmaker);
832 VALIDATER(arglist); VALIDATER(argument);
833 VALIDATER(listmaker);
835 #undef VALIDATER
837 #define is_even(n) (((n) & 1) == 0)
838 #define is_odd(n) (((n) & 1) == 1)
841 static int
842 validate_ntype(node *n, int t)
844 if (TYPE(n) != t) {
845 PyErr_Format(parser_error, "Expected node type %d, got %d.",
846 t, TYPE(n));
847 return 0;
849 return 1;
853 /* Verifies that the number of child nodes is exactly 'num', raising
854 * an exception if it isn't. The exception message does not indicate
855 * the exact number of nodes, allowing this to be used to raise the
856 * "right" exception when the wrong number of nodes is present in a
857 * specific variant of a statement's syntax. This is commonly used
858 * in that fashion.
860 static int
861 validate_numnodes(node *n, int num, const char *const name)
863 if (NCH(n) != num) {
864 PyErr_Format(parser_error,
865 "Illegal number of children for %s node.", name);
866 return 0;
868 return 1;
872 static int
873 validate_terminal(node *terminal, int type, char *string)
875 int res = (validate_ntype(terminal, type)
876 && ((string == 0) || (strcmp(string, STR(terminal)) == 0)));
878 if (!res && !PyErr_Occurred()) {
879 PyErr_Format(parser_error,
880 "Illegal terminal: expected \"%s\"", string);
882 return (res);
886 /* X (',' X) [',']
888 static int
889 validate_repeating_list(node *tree, int ntype, int (*vfunc)(node *),
890 const char *const name)
892 int nch = NCH(tree);
893 int res = (nch && validate_ntype(tree, ntype)
894 && vfunc(CHILD(tree, 0)));
896 if (!res && !PyErr_Occurred())
897 (void) validate_numnodes(tree, 1, name);
898 else {
899 if (is_even(nch))
900 res = validate_comma(CHILD(tree, --nch));
901 if (res && nch > 1) {
902 int pos = 1;
903 for ( ; res && pos < nch; pos += 2)
904 res = (validate_comma(CHILD(tree, pos))
905 && vfunc(CHILD(tree, pos + 1)));
908 return (res);
912 /* validate_class()
914 * classdef:
915 * 'class' NAME ['(' testlist ')'] ':' suite
917 static int
918 validate_class(node *tree)
920 int nch = NCH(tree);
921 int res = validate_ntype(tree, classdef) && ((nch == 4) || (nch == 7));
923 if (res) {
924 res = (validate_name(CHILD(tree, 0), "class")
925 && validate_ntype(CHILD(tree, 1), NAME)
926 && validate_colon(CHILD(tree, nch - 2))
927 && validate_suite(CHILD(tree, nch - 1)));
929 else
930 (void) validate_numnodes(tree, 4, "class");
931 if (res && (nch == 7)) {
932 res = (validate_lparen(CHILD(tree, 2))
933 && validate_testlist(CHILD(tree, 3))
934 && validate_rparen(CHILD(tree, 4)));
936 return (res);
940 /* if_stmt:
941 * 'if' test ':' suite ('elif' test ':' suite)* ['else' ':' suite]
943 static int
944 validate_if(node *tree)
946 int nch = NCH(tree);
947 int res = (validate_ntype(tree, if_stmt)
948 && (nch >= 4)
949 && validate_name(CHILD(tree, 0), "if")
950 && validate_test(CHILD(tree, 1))
951 && validate_colon(CHILD(tree, 2))
952 && validate_suite(CHILD(tree, 3)));
954 if (res && ((nch % 4) == 3)) {
955 /* ... 'else' ':' suite */
956 res = (validate_name(CHILD(tree, nch - 3), "else")
957 && validate_colon(CHILD(tree, nch - 2))
958 && validate_suite(CHILD(tree, nch - 1)));
959 nch -= 3;
961 else if (!res && !PyErr_Occurred())
962 (void) validate_numnodes(tree, 4, "if");
963 if ((nch % 4) != 0)
964 /* Will catch the case for nch < 4 */
965 res = validate_numnodes(tree, 0, "if");
966 else if (res && (nch > 4)) {
967 /* ... ('elif' test ':' suite)+ ... */
968 int j = 4;
969 while ((j < nch) && res) {
970 res = (validate_name(CHILD(tree, j), "elif")
971 && validate_colon(CHILD(tree, j + 2))
972 && validate_test(CHILD(tree, j + 1))
973 && validate_suite(CHILD(tree, j + 3)));
974 j += 4;
977 return (res);
981 /* parameters:
982 * '(' [varargslist] ')'
985 static int
986 validate_parameters(node *tree)
988 int nch = NCH(tree);
989 int res = validate_ntype(tree, parameters) && ((nch == 2) || (nch == 3));
991 if (res) {
992 res = (validate_lparen(CHILD(tree, 0))
993 && validate_rparen(CHILD(tree, nch - 1)));
994 if (res && (nch == 3))
995 res = validate_varargslist(CHILD(tree, 1));
997 else {
998 (void) validate_numnodes(tree, 2, "parameters");
1000 return (res);
1004 /* validate_suite()
1006 * suite:
1007 * simple_stmt
1008 * | NEWLINE INDENT stmt+ DEDENT
1010 static int
1011 validate_suite(node *tree)
1013 int nch = NCH(tree);
1014 int res = (validate_ntype(tree, suite) && ((nch == 1) || (nch >= 4)));
1016 if (res && (nch == 1))
1017 res = validate_simple_stmt(CHILD(tree, 0));
1018 else if (res) {
1019 /* NEWLINE INDENT stmt+ DEDENT */
1020 res = (validate_newline(CHILD(tree, 0))
1021 && validate_indent(CHILD(tree, 1))
1022 && validate_stmt(CHILD(tree, 2))
1023 && validate_dedent(CHILD(tree, nch - 1)));
1025 if (res && (nch > 4)) {
1026 int i = 3;
1027 --nch; /* forget the DEDENT */
1028 for ( ; res && (i < nch); ++i)
1029 res = validate_stmt(CHILD(tree, i));
1031 else if (nch < 4)
1032 res = validate_numnodes(tree, 4, "suite");
1034 return (res);
1038 static int
1039 validate_testlist(node *tree)
1041 return (validate_repeating_list(tree, testlist,
1042 validate_test, "testlist"));
1046 /* '*' NAME [',' '**' NAME] | '**' NAME
1048 static int
1049 validate_varargslist_trailer(node *tree, int start)
1051 int nch = NCH(tree);
1052 int res = 0;
1053 int sym;
1055 if (nch <= start) {
1056 err_string("expected variable argument trailer for varargslist");
1057 return 0;
1059 sym = TYPE(CHILD(tree, start));
1060 if (sym == STAR) {
1062 * ('*' NAME [',' '**' NAME]
1064 if (nch-start == 2)
1065 res = validate_name(CHILD(tree, start+1), NULL);
1066 else if (nch-start == 5)
1067 res = (validate_name(CHILD(tree, start+1), NULL)
1068 && validate_comma(CHILD(tree, start+2))
1069 && validate_doublestar(CHILD(tree, start+3))
1070 && validate_name(CHILD(tree, start+4), NULL));
1072 else if (sym == DOUBLESTAR) {
1074 * '**' NAME
1076 if (nch-start == 2)
1077 res = validate_name(CHILD(tree, start+1), NULL);
1079 if (!res)
1080 err_string("illegal variable argument trailer for varargslist");
1081 return res;
1085 /* validate_varargslist()
1087 * varargslist:
1088 * (fpdef ['=' test] ',')*
1089 * ('*' NAME [',' '**' NAME]
1090 * | '**' NAME)
1091 * | fpdef ['=' test] (',' fpdef ['=' test])* [',']
1094 static int
1095 validate_varargslist(node *tree)
1097 int nch = NCH(tree);
1098 int res = validate_ntype(tree, varargslist) && (nch != 0);
1099 int sym;
1101 if (!res)
1102 return 0;
1103 if (nch < 1) {
1104 err_string("varargslist missing child nodes");
1105 return 0;
1107 sym = TYPE(CHILD(tree, 0));
1108 if (sym == STAR || sym == DOUBLESTAR)
1109 /* whole thing matches:
1110 * '*' NAME [',' '**' NAME] | '**' NAME
1112 res = validate_varargslist_trailer(tree, 0);
1113 else if (sym == fpdef) {
1114 int i = 0;
1116 sym = TYPE(CHILD(tree, nch-1));
1117 if (sym == NAME) {
1119 * (fpdef ['=' test] ',')+
1120 * ('*' NAME [',' '**' NAME]
1121 * | '**' NAME)
1123 /* skip over (fpdef ['=' test] ',')+ */
1124 while (res && (i+2 <= nch)) {
1125 res = validate_fpdef(CHILD(tree, i));
1126 ++i;
1127 if (res && TYPE(CHILD(tree, i)) == EQUAL && (i+2 <= nch)) {
1128 res = (validate_equal(CHILD(tree, i))
1129 && validate_test(CHILD(tree, i+1)));
1130 if (res)
1131 i += 2;
1133 if (res && i < nch) {
1134 res = validate_comma(CHILD(tree, i));
1135 ++i;
1136 if (res && i < nch
1137 && (TYPE(CHILD(tree, i)) == DOUBLESTAR
1138 || TYPE(CHILD(tree, i)) == STAR))
1139 break;
1142 /* ... '*' NAME [',' '**' NAME] | '**' NAME
1143 * i --^^^
1145 if (res)
1146 res = validate_varargslist_trailer(tree, i);
1148 else {
1150 * fpdef ['=' test] (',' fpdef ['=' test])* [',']
1152 /* strip trailing comma node */
1153 if (sym == COMMA) {
1154 res = validate_comma(CHILD(tree, nch-1));
1155 if (!res)
1156 return 0;
1157 --nch;
1160 * fpdef ['=' test] (',' fpdef ['=' test])*
1162 res = validate_fpdef(CHILD(tree, 0));
1163 ++i;
1164 if (res && (i+2 <= nch) && TYPE(CHILD(tree, i)) == EQUAL) {
1165 res = (validate_equal(CHILD(tree, i))
1166 && validate_test(CHILD(tree, i+1)));
1167 i += 2;
1170 * ... (',' fpdef ['=' test])*
1171 * i ---^^^
1173 while (res && (nch - i) >= 2) {
1174 res = (validate_comma(CHILD(tree, i))
1175 && validate_fpdef(CHILD(tree, i+1)));
1176 i += 2;
1177 if (res && (nch - i) >= 2 && TYPE(CHILD(tree, i)) == EQUAL) {
1178 res = (validate_equal(CHILD(tree, i))
1179 && validate_test(CHILD(tree, i+1)));
1180 i += 2;
1183 if (res && nch - i != 0) {
1184 res = 0;
1185 err_string("illegal formation for varargslist");
1189 return res;
1193 /* list_iter: list_for | list_if
1195 static int
1196 validate_list_iter(node *tree)
1198 int res = (validate_ntype(tree, list_iter)
1199 && validate_numnodes(tree, 1, "list_iter"));
1200 if (res && TYPE(CHILD(tree, 0)) == list_for)
1201 res = validate_list_for(CHILD(tree, 0));
1202 else
1203 res = validate_list_if(CHILD(tree, 0));
1205 return res;
1208 /* list_for: 'for' exprlist 'in' testlist [list_iter]
1210 static int
1211 validate_list_for(node *tree)
1213 int nch = NCH(tree);
1214 int res;
1216 if (nch == 5)
1217 res = validate_list_iter(CHILD(tree, 4));
1218 else
1219 res = validate_numnodes(tree, 4, "list_for");
1221 if (res)
1222 res = (validate_name(CHILD(tree, 0), "for")
1223 && validate_exprlist(CHILD(tree, 1))
1224 && validate_name(CHILD(tree, 2), "in")
1225 && validate_testlist(CHILD(tree, 3)));
1227 return res;
1230 /* list_if: 'if' test [list_iter]
1232 static int
1233 validate_list_if(node *tree)
1235 int nch = NCH(tree);
1236 int res;
1238 if (nch == 3)
1239 res = validate_list_iter(CHILD(tree, 2));
1240 else
1241 res = validate_numnodes(tree, 2, "list_if");
1243 if (res)
1244 res = (validate_name(CHILD(tree, 0), "if")
1245 && validate_test(CHILD(tree, 1)));
1247 return res;
1251 /* validate_fpdef()
1253 * fpdef:
1254 * NAME
1255 * | '(' fplist ')'
1257 static int
1258 validate_fpdef(node *tree)
1260 int nch = NCH(tree);
1261 int res = validate_ntype(tree, fpdef);
1263 if (res) {
1264 if (nch == 1)
1265 res = validate_ntype(CHILD(tree, 0), NAME);
1266 else if (nch == 3)
1267 res = (validate_lparen(CHILD(tree, 0))
1268 && validate_fplist(CHILD(tree, 1))
1269 && validate_rparen(CHILD(tree, 2)));
1270 else
1271 res = validate_numnodes(tree, 1, "fpdef");
1273 return (res);
1277 static int
1278 validate_fplist(node *tree)
1280 return (validate_repeating_list(tree, fplist,
1281 validate_fpdef, "fplist"));
1285 /* simple_stmt | compound_stmt
1288 static int
1289 validate_stmt(node *tree)
1291 int res = (validate_ntype(tree, stmt)
1292 && validate_numnodes(tree, 1, "stmt"));
1294 if (res) {
1295 tree = CHILD(tree, 0);
1297 if (TYPE(tree) == simple_stmt)
1298 res = validate_simple_stmt(tree);
1299 else
1300 res = validate_compound_stmt(tree);
1302 return (res);
1306 /* small_stmt (';' small_stmt)* [';'] NEWLINE
1309 static int
1310 validate_simple_stmt(node *tree)
1312 int nch = NCH(tree);
1313 int res = (validate_ntype(tree, simple_stmt)
1314 && (nch >= 2)
1315 && validate_small_stmt(CHILD(tree, 0))
1316 && validate_newline(CHILD(tree, nch - 1)));
1318 if (nch < 2)
1319 res = validate_numnodes(tree, 2, "simple_stmt");
1320 --nch; /* forget the NEWLINE */
1321 if (res && is_even(nch))
1322 res = validate_semi(CHILD(tree, --nch));
1323 if (res && (nch > 2)) {
1324 int i;
1326 for (i = 1; res && (i < nch); i += 2)
1327 res = (validate_semi(CHILD(tree, i))
1328 && validate_small_stmt(CHILD(tree, i + 1)));
1330 return (res);
1334 static int
1335 validate_small_stmt(node *tree)
1337 int nch = NCH(tree);
1338 int res = validate_numnodes(tree, 1, "small_stmt");
1340 if (res) {
1341 int ntype = TYPE(CHILD(tree, 0));
1343 if ( (ntype == expr_stmt)
1344 || (ntype == print_stmt)
1345 || (ntype == del_stmt)
1346 || (ntype == pass_stmt)
1347 || (ntype == flow_stmt)
1348 || (ntype == import_stmt)
1349 || (ntype == global_stmt)
1350 || (ntype == assert_stmt)
1351 || (ntype == exec_stmt))
1352 res = validate_node(CHILD(tree, 0));
1353 else {
1354 res = 0;
1355 err_string("illegal small_stmt child type");
1358 else if (nch == 1) {
1359 res = 0;
1360 PyErr_Format(parser_error,
1361 "Unrecognized child node of small_stmt: %d.",
1362 TYPE(CHILD(tree, 0)));
1364 return (res);
1368 /* compound_stmt:
1369 * if_stmt | while_stmt | for_stmt | try_stmt | funcdef | classdef
1371 static int
1372 validate_compound_stmt(node *tree)
1374 int res = (validate_ntype(tree, compound_stmt)
1375 && validate_numnodes(tree, 1, "compound_stmt"));
1376 int ntype;
1378 if (!res)
1379 return (0);
1381 tree = CHILD(tree, 0);
1382 ntype = TYPE(tree);
1383 if ( (ntype == if_stmt)
1384 || (ntype == while_stmt)
1385 || (ntype == for_stmt)
1386 || (ntype == try_stmt)
1387 || (ntype == funcdef)
1388 || (ntype == classdef))
1389 res = validate_node(tree);
1390 else {
1391 res = 0;
1392 PyErr_Format(parser_error,
1393 "Illegal compound statement type: %d.", TYPE(tree));
1395 return (res);
1399 static int
1400 validate_expr_stmt(node *tree)
1402 int j;
1403 int nch = NCH(tree);
1404 int res = (validate_ntype(tree, expr_stmt)
1405 && is_odd(nch)
1406 && validate_testlist(CHILD(tree, 0)));
1408 if (res && nch == 3
1409 && TYPE(CHILD(tree, 1)) == augassign) {
1410 res = (validate_numnodes(CHILD(tree, 1), 1, "augassign")
1411 && validate_testlist(CHILD(tree, 2)));
1413 if (res) {
1414 char *s = STR(CHILD(CHILD(tree, 1), 0));
1416 res = (strcmp(s, "+=") == 0
1417 || strcmp(s, "-=") == 0
1418 || strcmp(s, "*=") == 0
1419 || strcmp(s, "/=") == 0
1420 || strcmp(s, "%=") == 0
1421 || strcmp(s, "&=") == 0
1422 || strcmp(s, "|=") == 0
1423 || strcmp(s, "^=") == 0
1424 || strcmp(s, "<<=") == 0
1425 || strcmp(s, ">>=") == 0
1426 || strcmp(s, "**=") == 0);
1427 if (!res)
1428 err_string("illegal augmmented assignment operator");
1431 else {
1432 for (j = 1; res && (j < nch); j += 2)
1433 res = (validate_equal(CHILD(tree, j))
1434 && validate_testlist(CHILD(tree, j + 1)));
1436 return (res);
1440 /* print_stmt:
1442 * 'print' ( [ test (',' test)* [','] ]
1443 * | '>>' test [ (',' test)+ [','] ] )
1445 static int
1446 validate_print_stmt(node *tree)
1448 int nch = NCH(tree);
1449 int res = (validate_ntype(tree, print_stmt)
1450 && (nch > 0)
1451 && validate_name(CHILD(tree, 0), "print"));
1453 if (res && nch > 1) {
1454 int sym = TYPE(CHILD(tree, 1));
1455 int i = 1;
1456 int allow_trailing_comma = 1;
1458 if (sym == test)
1459 res = validate_test(CHILD(tree, i++));
1460 else {
1461 if (nch < 3)
1462 res = validate_numnodes(tree, 3, "print_stmt");
1463 else {
1464 res = (validate_ntype(CHILD(tree, i), RIGHTSHIFT)
1465 && validate_test(CHILD(tree, i+1)));
1466 i += 2;
1467 allow_trailing_comma = 0;
1470 if (res) {
1471 /* ... (',' test)* [','] */
1472 while (res && i+2 <= nch) {
1473 res = (validate_comma(CHILD(tree, i))
1474 && validate_test(CHILD(tree, i+1)));
1475 allow_trailing_comma = 1;
1476 i += 2;
1478 if (res && !allow_trailing_comma)
1479 res = validate_numnodes(tree, i, "print_stmt");
1480 else if (res && i < nch)
1481 res = validate_comma(CHILD(tree, i));
1484 return (res);
1488 static int
1489 validate_del_stmt(node *tree)
1491 return (validate_numnodes(tree, 2, "del_stmt")
1492 && validate_name(CHILD(tree, 0), "del")
1493 && validate_exprlist(CHILD(tree, 1)));
1497 static int
1498 validate_return_stmt(node *tree)
1500 int nch = NCH(tree);
1501 int res = (validate_ntype(tree, return_stmt)
1502 && ((nch == 1) || (nch == 2))
1503 && validate_name(CHILD(tree, 0), "return"));
1505 if (res && (nch == 2))
1506 res = validate_testlist(CHILD(tree, 1));
1508 return (res);
1512 static int
1513 validate_raise_stmt(node *tree)
1515 int nch = NCH(tree);
1516 int res = (validate_ntype(tree, raise_stmt)
1517 && ((nch == 1) || (nch == 2) || (nch == 4) || (nch == 6)));
1519 if (res) {
1520 res = validate_name(CHILD(tree, 0), "raise");
1521 if (res && (nch >= 2))
1522 res = validate_test(CHILD(tree, 1));
1523 if (res && nch > 2) {
1524 res = (validate_comma(CHILD(tree, 2))
1525 && validate_test(CHILD(tree, 3)));
1526 if (res && (nch > 4))
1527 res = (validate_comma(CHILD(tree, 4))
1528 && validate_test(CHILD(tree, 5)));
1531 else
1532 (void) validate_numnodes(tree, 2, "raise");
1533 if (res && (nch == 4))
1534 res = (validate_comma(CHILD(tree, 2))
1535 && validate_test(CHILD(tree, 3)));
1537 return (res);
1541 static int
1542 validate_import_as_name(node *tree)
1544 int nch = NCH(tree);
1545 int ok = validate_ntype(tree, import_as_name);
1547 if (ok) {
1548 if (nch == 1)
1549 ok = validate_name(CHILD(tree, 0), NULL);
1550 else if (nch == 3)
1551 ok = (validate_name(CHILD(tree, 0), NULL)
1552 && validate_name(CHILD(tree, 1), "as")
1553 && validate_name(CHILD(tree, 2), NULL));
1554 else
1555 ok = validate_numnodes(tree, 3, "import_as_name");
1557 return ok;
1561 /* dotted_as_name: dotted_name [NAME NAME]
1563 static int
1564 validate_dotted_as_name(node *tree)
1566 int nch = NCH(tree);
1567 int res = validate_ntype(tree, dotted_as_name);
1569 if (res) {
1570 if (nch == 1)
1571 res = validate_ntype(CHILD(tree, 0), dotted_name);
1572 else if (nch == 3)
1573 res = (validate_ntype(CHILD(tree, 0), dotted_name)
1574 && validate_name(CHILD(tree, 1), "as")
1575 && validate_name(CHILD(tree, 2), NULL));
1576 else {
1577 res = 0;
1578 err_string("Illegal number of children for dotted_as_name.");
1581 return res;
1585 /* import_stmt:
1587 * 'import' dotted_as_name (',' dotted_as_name)*
1588 * | 'from' dotted_name 'import' ('*' | import_as_name (',' import_as_name)*)
1590 static int
1591 validate_import_stmt(node *tree)
1593 int nch = NCH(tree);
1594 int res = (validate_ntype(tree, import_stmt)
1595 && (nch >= 2) && is_even(nch)
1596 && validate_ntype(CHILD(tree, 0), NAME));
1598 if (res && (strcmp(STR(CHILD(tree, 0)), "import") == 0)) {
1599 int j;
1601 res = validate_dotted_as_name(CHILD(tree, 1));
1602 for (j = 2; res && (j < nch); j += 2)
1603 res = (validate_comma(CHILD(tree, j))
1604 && validate_ntype(CHILD(tree, j + 1), dotted_name));
1606 else if (res && (res = validate_name(CHILD(tree, 0), "from"))) {
1607 res = ((nch >= 4) && is_even(nch)
1608 && validate_name(CHILD(tree, 2), "import")
1609 && validate_dotted_as_name(CHILD(tree, 1)));
1610 if (nch == 4) {
1611 if (TYPE(CHILD(tree, 3)) == import_as_name)
1612 res = validate_import_as_name(CHILD(tree, 3));
1613 else
1614 res = validate_star(CHILD(tree, 3));
1616 else {
1617 /* 'from' dotted_name 'import' import_as_name
1618 * (',' import_as_name)+
1620 int j;
1621 res = validate_import_as_name(CHILD(tree, 3));
1622 for (j = 4; res && (j < nch); j += 2)
1623 res = (validate_comma(CHILD(tree, j))
1624 && validate_import_as_name(CHILD(tree, j + 1)));
1627 else
1628 res = 0;
1630 return (res);
1634 static int
1635 validate_global_stmt(node *tree)
1637 int j;
1638 int nch = NCH(tree);
1639 int res = (validate_ntype(tree, global_stmt)
1640 && is_even(nch) && (nch >= 2));
1642 if (res)
1643 res = (validate_name(CHILD(tree, 0), "global")
1644 && validate_ntype(CHILD(tree, 1), NAME));
1645 for (j = 2; res && (j < nch); j += 2)
1646 res = (validate_comma(CHILD(tree, j))
1647 && validate_ntype(CHILD(tree, j + 1), NAME));
1649 return (res);
1653 /* exec_stmt:
1655 * 'exec' expr ['in' test [',' test]]
1657 static int
1658 validate_exec_stmt(node *tree)
1660 int nch = NCH(tree);
1661 int res = (validate_ntype(tree, exec_stmt)
1662 && ((nch == 2) || (nch == 4) || (nch == 6))
1663 && validate_name(CHILD(tree, 0), "exec")
1664 && validate_expr(CHILD(tree, 1)));
1666 if (!res && !PyErr_Occurred())
1667 err_string("Illegal exec statement.");
1668 if (res && (nch > 2))
1669 res = (validate_name(CHILD(tree, 2), "in")
1670 && validate_test(CHILD(tree, 3)));
1671 if (res && (nch == 6))
1672 res = (validate_comma(CHILD(tree, 4))
1673 && validate_test(CHILD(tree, 5)));
1675 return (res);
1679 /* assert_stmt:
1681 * 'assert' test [',' test]
1683 static int
1684 validate_assert_stmt(node *tree)
1686 int nch = NCH(tree);
1687 int res = (validate_ntype(tree, assert_stmt)
1688 && ((nch == 2) || (nch == 4))
1689 && (validate_name(CHILD(tree, 0), "__assert__") ||
1690 validate_name(CHILD(tree, 0), "assert"))
1691 && validate_test(CHILD(tree, 1)));
1693 if (!res && !PyErr_Occurred())
1694 err_string("Illegal assert statement.");
1695 if (res && (nch > 2))
1696 res = (validate_comma(CHILD(tree, 2))
1697 && validate_test(CHILD(tree, 3)));
1699 return (res);
1703 static int
1704 validate_while(node *tree)
1706 int nch = NCH(tree);
1707 int res = (validate_ntype(tree, while_stmt)
1708 && ((nch == 4) || (nch == 7))
1709 && validate_name(CHILD(tree, 0), "while")
1710 && validate_test(CHILD(tree, 1))
1711 && validate_colon(CHILD(tree, 2))
1712 && validate_suite(CHILD(tree, 3)));
1714 if (res && (nch == 7))
1715 res = (validate_name(CHILD(tree, 4), "else")
1716 && validate_colon(CHILD(tree, 5))
1717 && validate_suite(CHILD(tree, 6)));
1719 return (res);
1723 static int
1724 validate_for(node *tree)
1726 int nch = NCH(tree);
1727 int res = (validate_ntype(tree, for_stmt)
1728 && ((nch == 6) || (nch == 9))
1729 && validate_name(CHILD(tree, 0), "for")
1730 && validate_exprlist(CHILD(tree, 1))
1731 && validate_name(CHILD(tree, 2), "in")
1732 && validate_testlist(CHILD(tree, 3))
1733 && validate_colon(CHILD(tree, 4))
1734 && validate_suite(CHILD(tree, 5)));
1736 if (res && (nch == 9))
1737 res = (validate_name(CHILD(tree, 6), "else")
1738 && validate_colon(CHILD(tree, 7))
1739 && validate_suite(CHILD(tree, 8)));
1741 return (res);
1745 /* try_stmt:
1746 * 'try' ':' suite (except_clause ':' suite)+ ['else' ':' suite]
1747 * | 'try' ':' suite 'finally' ':' suite
1750 static int
1751 validate_try(node *tree)
1753 int nch = NCH(tree);
1754 int pos = 3;
1755 int res = (validate_ntype(tree, try_stmt)
1756 && (nch >= 6) && ((nch % 3) == 0));
1758 if (res)
1759 res = (validate_name(CHILD(tree, 0), "try")
1760 && validate_colon(CHILD(tree, 1))
1761 && validate_suite(CHILD(tree, 2))
1762 && validate_colon(CHILD(tree, nch - 2))
1763 && validate_suite(CHILD(tree, nch - 1)));
1764 else if (!PyErr_Occurred()) {
1765 const char* name = "except";
1766 if (TYPE(CHILD(tree, nch - 3)) != except_clause)
1767 name = STR(CHILD(tree, nch - 3));
1769 PyErr_Format(parser_error,
1770 "Illegal number of children for try/%s node.", name);
1772 /* Skip past except_clause sections: */
1773 while (res && (TYPE(CHILD(tree, pos)) == except_clause)) {
1774 res = (validate_except_clause(CHILD(tree, pos))
1775 && validate_colon(CHILD(tree, pos + 1))
1776 && validate_suite(CHILD(tree, pos + 2)));
1777 pos += 3;
1779 if (res && (pos < nch)) {
1780 res = validate_ntype(CHILD(tree, pos), NAME);
1781 if (res && (strcmp(STR(CHILD(tree, pos)), "finally") == 0))
1782 res = (validate_numnodes(tree, 6, "try/finally")
1783 && validate_colon(CHILD(tree, 4))
1784 && validate_suite(CHILD(tree, 5)));
1785 else if (res) {
1786 if (nch == (pos + 3)) {
1787 res = ((strcmp(STR(CHILD(tree, pos)), "except") == 0)
1788 || (strcmp(STR(CHILD(tree, pos)), "else") == 0));
1789 if (!res)
1790 err_string("Illegal trailing triple in try statement.");
1792 else if (nch == (pos + 6)) {
1793 res = (validate_name(CHILD(tree, pos), "except")
1794 && validate_colon(CHILD(tree, pos + 1))
1795 && validate_suite(CHILD(tree, pos + 2))
1796 && validate_name(CHILD(tree, pos + 3), "else"));
1798 else
1799 res = validate_numnodes(tree, pos + 3, "try/except");
1802 return (res);
1806 static int
1807 validate_except_clause(node *tree)
1809 int nch = NCH(tree);
1810 int res = (validate_ntype(tree, except_clause)
1811 && ((nch == 1) || (nch == 2) || (nch == 4))
1812 && validate_name(CHILD(tree, 0), "except"));
1814 if (res && (nch > 1))
1815 res = validate_test(CHILD(tree, 1));
1816 if (res && (nch == 4))
1817 res = (validate_comma(CHILD(tree, 2))
1818 && validate_test(CHILD(tree, 3)));
1820 return (res);
1824 static int
1825 validate_test(node *tree)
1827 int nch = NCH(tree);
1828 int res = validate_ntype(tree, test) && is_odd(nch);
1830 if (res && (TYPE(CHILD(tree, 0)) == lambdef))
1831 res = ((nch == 1)
1832 && validate_lambdef(CHILD(tree, 0)));
1833 else if (res) {
1834 int pos;
1835 res = validate_and_test(CHILD(tree, 0));
1836 for (pos = 1; res && (pos < nch); pos += 2)
1837 res = (validate_name(CHILD(tree, pos), "or")
1838 && validate_and_test(CHILD(tree, pos + 1)));
1840 return (res);
1844 static int
1845 validate_and_test(node *tree)
1847 int pos;
1848 int nch = NCH(tree);
1849 int res = (validate_ntype(tree, and_test)
1850 && is_odd(nch)
1851 && validate_not_test(CHILD(tree, 0)));
1853 for (pos = 1; res && (pos < nch); pos += 2)
1854 res = (validate_name(CHILD(tree, pos), "and")
1855 && validate_not_test(CHILD(tree, 0)));
1857 return (res);
1861 static int
1862 validate_not_test(node *tree)
1864 int nch = NCH(tree);
1865 int res = validate_ntype(tree, not_test) && ((nch == 1) || (nch == 2));
1867 if (res) {
1868 if (nch == 2)
1869 res = (validate_name(CHILD(tree, 0), "not")
1870 && validate_not_test(CHILD(tree, 1)));
1871 else if (nch == 1)
1872 res = validate_comparison(CHILD(tree, 0));
1874 return (res);
1878 static int
1879 validate_comparison(node *tree)
1881 int pos;
1882 int nch = NCH(tree);
1883 int res = (validate_ntype(tree, comparison)
1884 && is_odd(nch)
1885 && validate_expr(CHILD(tree, 0)));
1887 for (pos = 1; res && (pos < nch); pos += 2)
1888 res = (validate_comp_op(CHILD(tree, pos))
1889 && validate_expr(CHILD(tree, pos + 1)));
1891 return (res);
1895 static int
1896 validate_comp_op(node *tree)
1898 int res = 0;
1899 int nch = NCH(tree);
1901 if (!validate_ntype(tree, comp_op))
1902 return (0);
1903 if (nch == 1) {
1905 * Only child will be a terminal with a well-defined symbolic name
1906 * or a NAME with a string of either 'is' or 'in'
1908 tree = CHILD(tree, 0);
1909 switch (TYPE(tree)) {
1910 case LESS:
1911 case GREATER:
1912 case EQEQUAL:
1913 case EQUAL:
1914 case LESSEQUAL:
1915 case GREATEREQUAL:
1916 case NOTEQUAL:
1917 res = 1;
1918 break;
1919 case NAME:
1920 res = ((strcmp(STR(tree), "in") == 0)
1921 || (strcmp(STR(tree), "is") == 0));
1922 if (!res) {
1923 PyErr_Format(parser_error,
1924 "Illegal operator: '%s'.", STR(tree));
1926 break;
1927 default:
1928 err_string("Illegal comparison operator type.");
1929 break;
1932 else if ((res = validate_numnodes(tree, 2, "comp_op")) != 0) {
1933 res = (validate_ntype(CHILD(tree, 0), NAME)
1934 && validate_ntype(CHILD(tree, 1), NAME)
1935 && (((strcmp(STR(CHILD(tree, 0)), "is") == 0)
1936 && (strcmp(STR(CHILD(tree, 1)), "not") == 0))
1937 || ((strcmp(STR(CHILD(tree, 0)), "not") == 0)
1938 && (strcmp(STR(CHILD(tree, 1)), "in") == 0))));
1939 if (!res && !PyErr_Occurred())
1940 err_string("Unknown comparison operator.");
1942 return (res);
1946 static int
1947 validate_expr(node *tree)
1949 int j;
1950 int nch = NCH(tree);
1951 int res = (validate_ntype(tree, expr)
1952 && is_odd(nch)
1953 && validate_xor_expr(CHILD(tree, 0)));
1955 for (j = 2; res && (j < nch); j += 2)
1956 res = (validate_xor_expr(CHILD(tree, j))
1957 && validate_vbar(CHILD(tree, j - 1)));
1959 return (res);
1963 static int
1964 validate_xor_expr(node *tree)
1966 int j;
1967 int nch = NCH(tree);
1968 int res = (validate_ntype(tree, xor_expr)
1969 && is_odd(nch)
1970 && validate_and_expr(CHILD(tree, 0)));
1972 for (j = 2; res && (j < nch); j += 2)
1973 res = (validate_circumflex(CHILD(tree, j - 1))
1974 && validate_and_expr(CHILD(tree, j)));
1976 return (res);
1980 static int
1981 validate_and_expr(node *tree)
1983 int pos;
1984 int nch = NCH(tree);
1985 int res = (validate_ntype(tree, and_expr)
1986 && is_odd(nch)
1987 && validate_shift_expr(CHILD(tree, 0)));
1989 for (pos = 1; res && (pos < nch); pos += 2)
1990 res = (validate_ampersand(CHILD(tree, pos))
1991 && validate_shift_expr(CHILD(tree, pos + 1)));
1993 return (res);
1997 static int
1998 validate_chain_two_ops(node *tree, int (*termvalid)(node *), int op1, int op2)
2000 int pos = 1;
2001 int nch = NCH(tree);
2002 int res = (is_odd(nch)
2003 && (*termvalid)(CHILD(tree, 0)));
2005 for ( ; res && (pos < nch); pos += 2) {
2006 if (TYPE(CHILD(tree, pos)) != op1)
2007 res = validate_ntype(CHILD(tree, pos), op2);
2008 if (res)
2009 res = (*termvalid)(CHILD(tree, pos + 1));
2011 return (res);
2015 static int
2016 validate_shift_expr(node *tree)
2018 return (validate_ntype(tree, shift_expr)
2019 && validate_chain_two_ops(tree, validate_arith_expr,
2020 LEFTSHIFT, RIGHTSHIFT));
2024 static int
2025 validate_arith_expr(node *tree)
2027 return (validate_ntype(tree, arith_expr)
2028 && validate_chain_two_ops(tree, validate_term, PLUS, MINUS));
2032 static int
2033 validate_term(node *tree)
2035 int pos = 1;
2036 int nch = NCH(tree);
2037 int res = (validate_ntype(tree, term)
2038 && is_odd(nch)
2039 && validate_factor(CHILD(tree, 0)));
2041 for ( ; res && (pos < nch); pos += 2)
2042 res = (((TYPE(CHILD(tree, pos)) == STAR)
2043 || (TYPE(CHILD(tree, pos)) == SLASH)
2044 || (TYPE(CHILD(tree, pos)) == PERCENT))
2045 && validate_factor(CHILD(tree, pos + 1)));
2047 return (res);
2051 /* factor:
2053 * factor: ('+'|'-'|'~') factor | power
2055 static int
2056 validate_factor(node *tree)
2058 int nch = NCH(tree);
2059 int res = (validate_ntype(tree, factor)
2060 && (((nch == 2)
2061 && ((TYPE(CHILD(tree, 0)) == PLUS)
2062 || (TYPE(CHILD(tree, 0)) == MINUS)
2063 || (TYPE(CHILD(tree, 0)) == TILDE))
2064 && validate_factor(CHILD(tree, 1)))
2065 || ((nch == 1)
2066 && validate_power(CHILD(tree, 0)))));
2067 return (res);
2071 /* power:
2073 * power: atom trailer* ('**' factor)*
2075 static int
2076 validate_power(node *tree)
2078 int pos = 1;
2079 int nch = NCH(tree);
2080 int res = (validate_ntype(tree, power) && (nch >= 1)
2081 && validate_atom(CHILD(tree, 0)));
2083 while (res && (pos < nch) && (TYPE(CHILD(tree, pos)) == trailer))
2084 res = validate_trailer(CHILD(tree, pos++));
2085 if (res && (pos < nch)) {
2086 if (!is_even(nch - pos)) {
2087 err_string("Illegal number of nodes for 'power'.");
2088 return (0);
2090 for ( ; res && (pos < (nch - 1)); pos += 2)
2091 res = (validate_doublestar(CHILD(tree, pos))
2092 && validate_factor(CHILD(tree, pos + 1)));
2094 return (res);
2098 static int
2099 validate_atom(node *tree)
2101 int pos;
2102 int nch = NCH(tree);
2103 int res = validate_ntype(tree, atom);
2105 if (res && nch < 1)
2106 res = validate_numnodes(tree, nch+1, "atom");
2107 if (res) {
2108 switch (TYPE(CHILD(tree, 0))) {
2109 case LPAR:
2110 res = ((nch <= 3)
2111 && (validate_rparen(CHILD(tree, nch - 1))));
2113 if (res && (nch == 3))
2114 res = validate_testlist(CHILD(tree, 1));
2115 break;
2116 case LSQB:
2117 if (nch == 2)
2118 res = validate_ntype(CHILD(tree, 1), RSQB);
2119 else if (nch == 3)
2120 res = (validate_listmaker(CHILD(tree, 1))
2121 && validate_ntype(CHILD(tree, 2), RSQB));
2122 else {
2123 res = 0;
2124 err_string("illegal list display atom");
2126 break;
2127 case LBRACE:
2128 res = ((nch <= 3)
2129 && validate_ntype(CHILD(tree, nch - 1), RBRACE));
2131 if (res && (nch == 3))
2132 res = validate_dictmaker(CHILD(tree, 1));
2133 break;
2134 case BACKQUOTE:
2135 res = ((nch == 3)
2136 && validate_testlist(CHILD(tree, 1))
2137 && validate_ntype(CHILD(tree, 2), BACKQUOTE));
2138 break;
2139 case NAME:
2140 case NUMBER:
2141 res = (nch == 1);
2142 break;
2143 case STRING:
2144 for (pos = 1; res && (pos < nch); ++pos)
2145 res = validate_ntype(CHILD(tree, pos), STRING);
2146 break;
2147 default:
2148 res = 0;
2149 break;
2152 return (res);
2156 /* listmaker:
2157 * test ( list_for | (',' test)* [','] )
2159 static int
2160 validate_listmaker(node *tree)
2162 int nch = NCH(tree);
2163 int ok = nch;
2165 if (nch == 0)
2166 err_string("missing child nodes of listmaker");
2167 else
2168 ok = validate_test(CHILD(tree, 0));
2171 * list_iter | (',' test)* [',']
2173 if (nch == 2 && TYPE(CHILD(tree, 1)) == list_for)
2174 ok = validate_list_for(CHILD(tree, 1));
2175 else {
2176 /* (',' test)* [','] */
2177 int i = 1;
2178 while (ok && nch - i >= 2) {
2179 ok = (validate_comma(CHILD(tree, i))
2180 && validate_test(CHILD(tree, i+1)));
2181 i += 2;
2183 if (ok && i == nch-1)
2184 ok = validate_comma(CHILD(tree, i));
2185 else if (i != nch) {
2186 ok = 0;
2187 err_string("illegal trailing nodes for listmaker");
2190 return ok;
2194 /* funcdef:
2195 * 'def' NAME parameters ':' suite
2198 static int
2199 validate_funcdef(node *tree)
2201 return (validate_ntype(tree, funcdef)
2202 && validate_numnodes(tree, 5, "funcdef")
2203 && validate_name(CHILD(tree, 0), "def")
2204 && validate_ntype(CHILD(tree, 1), NAME)
2205 && validate_colon(CHILD(tree, 3))
2206 && validate_parameters(CHILD(tree, 2))
2207 && validate_suite(CHILD(tree, 4)));
2211 static int
2212 validate_lambdef(node *tree)
2214 int nch = NCH(tree);
2215 int res = (validate_ntype(tree, lambdef)
2216 && ((nch == 3) || (nch == 4))
2217 && validate_name(CHILD(tree, 0), "lambda")
2218 && validate_colon(CHILD(tree, nch - 2))
2219 && validate_test(CHILD(tree, nch - 1)));
2221 if (res && (nch == 4))
2222 res = validate_varargslist(CHILD(tree, 1));
2223 else if (!res && !PyErr_Occurred())
2224 (void) validate_numnodes(tree, 3, "lambdef");
2226 return (res);
2230 /* arglist:
2232 * (argument ',')* (argument [','] | '*' test [',' '**' test] | '**' test)
2234 static int
2235 validate_arglist(node *tree)
2237 int nch = NCH(tree);
2238 int i = 0;
2239 int ok = 1;
2241 if (nch <= 0)
2242 /* raise the right error from having an invalid number of children */
2243 return validate_numnodes(tree, nch + 1, "arglist");
2245 while (ok && nch-i >= 2) {
2246 /* skip leading (argument ',') */
2247 ok = (validate_argument(CHILD(tree, i))
2248 && validate_comma(CHILD(tree, i+1)));
2249 if (ok)
2250 i += 2;
2251 else
2252 PyErr_Clear();
2254 ok = 1;
2255 if (nch-i > 0) {
2257 * argument | '*' test [',' '**' test] | '**' test
2259 int sym = TYPE(CHILD(tree, i));
2261 if (sym == argument) {
2262 ok = validate_argument(CHILD(tree, i));
2263 if (ok && i+1 != nch) {
2264 err_string("illegal arglist specification"
2265 " (extra stuff on end)");
2266 ok = 0;
2269 else if (sym == STAR) {
2270 ok = validate_star(CHILD(tree, i));
2271 if (ok && (nch-i == 2))
2272 ok = validate_test(CHILD(tree, i+1));
2273 else if (ok && (nch-i == 5))
2274 ok = (validate_test(CHILD(tree, i+1))
2275 && validate_comma(CHILD(tree, i+2))
2276 && validate_doublestar(CHILD(tree, i+3))
2277 && validate_test(CHILD(tree, i+4)));
2278 else {
2279 err_string("illegal use of '*' in arglist");
2280 ok = 0;
2283 else if (sym == DOUBLESTAR) {
2284 if (nch-i == 2)
2285 ok = (validate_doublestar(CHILD(tree, i))
2286 && validate_test(CHILD(tree, i+1)));
2287 else {
2288 err_string("illegal use of '**' in arglist");
2289 ok = 0;
2292 else {
2293 err_string("illegal arglist specification");
2294 ok = 0;
2297 return (ok);
2302 /* argument:
2304 * [test '='] test
2306 static int
2307 validate_argument(node *tree)
2309 int nch = NCH(tree);
2310 int res = (validate_ntype(tree, argument)
2311 && ((nch == 1) || (nch == 3))
2312 && validate_test(CHILD(tree, 0)));
2314 if (res && (nch == 3))
2315 res = (validate_equal(CHILD(tree, 1))
2316 && validate_test(CHILD(tree, 2)));
2318 return (res);
2323 /* trailer:
2325 * '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME
2327 static int
2328 validate_trailer(node *tree)
2330 int nch = NCH(tree);
2331 int res = validate_ntype(tree, trailer) && ((nch == 2) || (nch == 3));
2333 if (res) {
2334 switch (TYPE(CHILD(tree, 0))) {
2335 case LPAR:
2336 res = validate_rparen(CHILD(tree, nch - 1));
2337 if (res && (nch == 3))
2338 res = validate_arglist(CHILD(tree, 1));
2339 break;
2340 case LSQB:
2341 res = (validate_numnodes(tree, 3, "trailer")
2342 && validate_subscriptlist(CHILD(tree, 1))
2343 && validate_ntype(CHILD(tree, 2), RSQB));
2344 break;
2345 case DOT:
2346 res = (validate_numnodes(tree, 2, "trailer")
2347 && validate_ntype(CHILD(tree, 1), NAME));
2348 break;
2349 default:
2350 res = 0;
2351 break;
2354 else {
2355 (void) validate_numnodes(tree, 2, "trailer");
2357 return (res);
2361 /* subscriptlist:
2363 * subscript (',' subscript)* [',']
2365 static int
2366 validate_subscriptlist(node *tree)
2368 return (validate_repeating_list(tree, subscriptlist,
2369 validate_subscript, "subscriptlist"));
2373 /* subscript:
2375 * '.' '.' '.' | test | [test] ':' [test] [sliceop]
2377 static int
2378 validate_subscript(node *tree)
2380 int offset = 0;
2381 int nch = NCH(tree);
2382 int res = validate_ntype(tree, subscript) && (nch >= 1) && (nch <= 4);
2384 if (!res) {
2385 if (!PyErr_Occurred())
2386 err_string("invalid number of arguments for subscript node");
2387 return (0);
2389 if (TYPE(CHILD(tree, 0)) == DOT)
2390 /* take care of ('.' '.' '.') possibility */
2391 return (validate_numnodes(tree, 3, "subscript")
2392 && validate_dot(CHILD(tree, 0))
2393 && validate_dot(CHILD(tree, 1))
2394 && validate_dot(CHILD(tree, 2)));
2395 if (nch == 1) {
2396 if (TYPE(CHILD(tree, 0)) == test)
2397 res = validate_test(CHILD(tree, 0));
2398 else
2399 res = validate_colon(CHILD(tree, 0));
2400 return (res);
2402 /* Must be [test] ':' [test] [sliceop],
2403 * but at least one of the optional components will
2404 * be present, but we don't know which yet.
2406 if ((TYPE(CHILD(tree, 0)) != COLON) || (nch == 4)) {
2407 res = validate_test(CHILD(tree, 0));
2408 offset = 1;
2410 if (res)
2411 res = validate_colon(CHILD(tree, offset));
2412 if (res) {
2413 int rem = nch - ++offset;
2414 if (rem) {
2415 if (TYPE(CHILD(tree, offset)) == test) {
2416 res = validate_test(CHILD(tree, offset));
2417 ++offset;
2418 --rem;
2420 if (res && rem)
2421 res = validate_sliceop(CHILD(tree, offset));
2424 return (res);
2428 static int
2429 validate_sliceop(node *tree)
2431 int nch = NCH(tree);
2432 int res = ((nch == 1) || validate_numnodes(tree, 2, "sliceop"))
2433 && validate_ntype(tree, sliceop);
2434 if (!res && !PyErr_Occurred()) {
2435 res = validate_numnodes(tree, 1, "sliceop");
2437 if (res)
2438 res = validate_colon(CHILD(tree, 0));
2439 if (res && (nch == 2))
2440 res = validate_test(CHILD(tree, 1));
2442 return (res);
2446 static int
2447 validate_exprlist(node *tree)
2449 return (validate_repeating_list(tree, exprlist,
2450 validate_expr, "exprlist"));
2454 static int
2455 validate_dictmaker(node *tree)
2457 int nch = NCH(tree);
2458 int res = (validate_ntype(tree, dictmaker)
2459 && (nch >= 3)
2460 && validate_test(CHILD(tree, 0))
2461 && validate_colon(CHILD(tree, 1))
2462 && validate_test(CHILD(tree, 2)));
2464 if (res && ((nch % 4) == 0))
2465 res = validate_comma(CHILD(tree, --nch));
2466 else if (res)
2467 res = ((nch % 4) == 3);
2469 if (res && (nch > 3)) {
2470 int pos = 3;
2471 /* ( ',' test ':' test )* */
2472 while (res && (pos < nch)) {
2473 res = (validate_comma(CHILD(tree, pos))
2474 && validate_test(CHILD(tree, pos + 1))
2475 && validate_colon(CHILD(tree, pos + 2))
2476 && validate_test(CHILD(tree, pos + 3)));
2477 pos += 4;
2480 return (res);
2484 static int
2485 validate_eval_input(node *tree)
2487 int pos;
2488 int nch = NCH(tree);
2489 int res = (validate_ntype(tree, eval_input)
2490 && (nch >= 2)
2491 && validate_testlist(CHILD(tree, 0))
2492 && validate_ntype(CHILD(tree, nch - 1), ENDMARKER));
2494 for (pos = 1; res && (pos < (nch - 1)); ++pos)
2495 res = validate_ntype(CHILD(tree, pos), NEWLINE);
2497 return (res);
2501 static int
2502 validate_node(node *tree)
2504 int nch = 0; /* num. children on current node */
2505 int res = 1; /* result value */
2506 node* next = 0; /* node to process after this one */
2508 while (res & (tree != 0)) {
2509 nch = NCH(tree);
2510 next = 0;
2511 switch (TYPE(tree)) {
2513 * Definition nodes.
2515 case funcdef:
2516 res = validate_funcdef(tree);
2517 break;
2518 case classdef:
2519 res = validate_class(tree);
2520 break;
2522 * "Trivial" parse tree nodes.
2523 * (Why did I call these trivial?)
2525 case stmt:
2526 res = validate_stmt(tree);
2527 break;
2528 case small_stmt:
2530 * expr_stmt | print_stmt | del_stmt | pass_stmt | flow_stmt
2531 * | import_stmt | global_stmt | exec_stmt | assert_stmt
2533 res = validate_small_stmt(tree);
2534 break;
2535 case flow_stmt:
2536 res = (validate_numnodes(tree, 1, "flow_stmt")
2537 && ((TYPE(CHILD(tree, 0)) == break_stmt)
2538 || (TYPE(CHILD(tree, 0)) == continue_stmt)
2539 || (TYPE(CHILD(tree, 0)) == return_stmt)
2540 || (TYPE(CHILD(tree, 0)) == raise_stmt)));
2541 if (res)
2542 next = CHILD(tree, 0);
2543 else if (nch == 1)
2544 err_string("Illegal flow_stmt type.");
2545 break;
2547 * Compound statements.
2549 case simple_stmt:
2550 res = validate_simple_stmt(tree);
2551 break;
2552 case compound_stmt:
2553 res = validate_compound_stmt(tree);
2554 break;
2556 * Fundamental statements.
2558 case expr_stmt:
2559 res = validate_expr_stmt(tree);
2560 break;
2561 case print_stmt:
2562 res = validate_print_stmt(tree);
2563 break;
2564 case del_stmt:
2565 res = validate_del_stmt(tree);
2566 break;
2567 case pass_stmt:
2568 res = (validate_numnodes(tree, 1, "pass")
2569 && validate_name(CHILD(tree, 0), "pass"));
2570 break;
2571 case break_stmt:
2572 res = (validate_numnodes(tree, 1, "break")
2573 && validate_name(CHILD(tree, 0), "break"));
2574 break;
2575 case continue_stmt:
2576 res = (validate_numnodes(tree, 1, "continue")
2577 && validate_name(CHILD(tree, 0), "continue"));
2578 break;
2579 case return_stmt:
2580 res = validate_return_stmt(tree);
2581 break;
2582 case raise_stmt:
2583 res = validate_raise_stmt(tree);
2584 break;
2585 case import_stmt:
2586 res = validate_import_stmt(tree);
2587 break;
2588 case global_stmt:
2589 res = validate_global_stmt(tree);
2590 break;
2591 case exec_stmt:
2592 res = validate_exec_stmt(tree);
2593 break;
2594 case assert_stmt:
2595 res = validate_assert_stmt(tree);
2596 break;
2597 case if_stmt:
2598 res = validate_if(tree);
2599 break;
2600 case while_stmt:
2601 res = validate_while(tree);
2602 break;
2603 case for_stmt:
2604 res = validate_for(tree);
2605 break;
2606 case try_stmt:
2607 res = validate_try(tree);
2608 break;
2609 case suite:
2610 res = validate_suite(tree);
2611 break;
2613 * Expression nodes.
2615 case testlist:
2616 res = validate_testlist(tree);
2617 break;
2618 case test:
2619 res = validate_test(tree);
2620 break;
2621 case and_test:
2622 res = validate_and_test(tree);
2623 break;
2624 case not_test:
2625 res = validate_not_test(tree);
2626 break;
2627 case comparison:
2628 res = validate_comparison(tree);
2629 break;
2630 case exprlist:
2631 res = validate_exprlist(tree);
2632 break;
2633 case comp_op:
2634 res = validate_comp_op(tree);
2635 break;
2636 case expr:
2637 res = validate_expr(tree);
2638 break;
2639 case xor_expr:
2640 res = validate_xor_expr(tree);
2641 break;
2642 case and_expr:
2643 res = validate_and_expr(tree);
2644 break;
2645 case shift_expr:
2646 res = validate_shift_expr(tree);
2647 break;
2648 case arith_expr:
2649 res = validate_arith_expr(tree);
2650 break;
2651 case term:
2652 res = validate_term(tree);
2653 break;
2654 case factor:
2655 res = validate_factor(tree);
2656 break;
2657 case power:
2658 res = validate_power(tree);
2659 break;
2660 case atom:
2661 res = validate_atom(tree);
2662 break;
2664 default:
2665 /* Hopefully never reached! */
2666 err_string("Unrecogniged node type.");
2667 res = 0;
2668 break;
2670 tree = next;
2672 return (res);
2676 static int
2677 validate_expr_tree(node *tree)
2679 int res = validate_eval_input(tree);
2681 if (!res && !PyErr_Occurred())
2682 err_string("Could not validate expression tuple.");
2684 return (res);
2688 /* file_input:
2689 * (NEWLINE | stmt)* ENDMARKER
2691 static int
2692 validate_file_input(node *tree)
2694 int j = 0;
2695 int nch = NCH(tree) - 1;
2696 int res = ((nch >= 0)
2697 && validate_ntype(CHILD(tree, nch), ENDMARKER));
2699 for ( ; res && (j < nch); ++j) {
2700 if (TYPE(CHILD(tree, j)) == stmt)
2701 res = validate_stmt(CHILD(tree, j));
2702 else
2703 res = validate_newline(CHILD(tree, j));
2705 /* This stays in to prevent any internal failures from getting to the
2706 * user. Hopefully, this won't be needed. If a user reports getting
2707 * this, we have some debugging to do.
2709 if (!res && !PyErr_Occurred())
2710 err_string("VALIDATION FAILURE: report this to the maintainer!.");
2712 return (res);
2716 static PyObject*
2717 pickle_constructor = NULL;
2720 static PyObject*
2721 parser__pickler(PyObject *self, PyObject *args)
2723 NOTE(ARGUNUSED(self))
2724 PyObject *result = NULL;
2725 PyObject *ast = NULL;
2726 PyObject *empty_dict = NULL;
2728 if (PyArg_ParseTuple(args, "O!:_pickler", &PyAST_Type, &ast)) {
2729 PyObject *newargs;
2730 PyObject *tuple;
2732 if ((empty_dict = PyDict_New()) == NULL)
2733 goto finally;
2734 if ((newargs = Py_BuildValue("Oi", ast, 1)) == NULL)
2735 goto finally;
2736 tuple = parser_ast2tuple((PyAST_Object*)NULL, newargs, empty_dict);
2737 if (tuple != NULL) {
2738 result = Py_BuildValue("O(O)", pickle_constructor, tuple);
2739 Py_DECREF(tuple);
2741 Py_DECREF(empty_dict);
2742 Py_DECREF(newargs);
2744 finally:
2745 Py_XDECREF(empty_dict);
2747 return (result);
2751 /* Functions exported by this module. Most of this should probably
2752 * be converted into an AST object with methods, but that is better
2753 * done directly in Python, allowing subclasses to be created directly.
2754 * We'd really have to write a wrapper around it all anyway to allow
2755 * inheritance.
2757 static PyMethodDef parser_functions[] = {
2758 {"ast2tuple", (PyCFunction)parser_ast2tuple, PUBLIC_METHOD_TYPE,
2759 "Creates a tuple-tree representation of an AST."},
2760 {"ast2list", (PyCFunction)parser_ast2list, PUBLIC_METHOD_TYPE,
2761 "Creates a list-tree representation of an AST."},
2762 {"compileast", (PyCFunction)parser_compileast, PUBLIC_METHOD_TYPE,
2763 "Compiles an AST object into a code object."},
2764 {"expr", (PyCFunction)parser_expr, PUBLIC_METHOD_TYPE,
2765 "Creates an AST object from an expression."},
2766 {"isexpr", (PyCFunction)parser_isexpr, PUBLIC_METHOD_TYPE,
2767 "Determines if an AST object was created from an expression."},
2768 {"issuite", (PyCFunction)parser_issuite, PUBLIC_METHOD_TYPE,
2769 "Determines if an AST object was created from a suite."},
2770 {"suite", (PyCFunction)parser_suite, PUBLIC_METHOD_TYPE,
2771 "Creates an AST object from a suite."},
2772 {"sequence2ast", (PyCFunction)parser_tuple2ast, PUBLIC_METHOD_TYPE,
2773 "Creates an AST object from a tree representation."},
2774 {"tuple2ast", (PyCFunction)parser_tuple2ast, PUBLIC_METHOD_TYPE,
2775 "Creates an AST object from a tree representation."},
2777 /* private stuff: support pickle module */
2778 {"_pickler", (PyCFunction)parser__pickler, METH_VARARGS,
2779 "Returns the pickle magic to allow ast objects to be pickled."},
2781 {NULL, NULL, 0, NULL}
2785 DL_EXPORT(void) initparser(void); /* supply a prototype */
2787 DL_EXPORT(void)
2788 initparser(void)
2790 PyObject* module;
2791 PyObject* dict;
2793 PyAST_Type.ob_type = &PyType_Type;
2794 module = Py_InitModule("parser", parser_functions);
2795 dict = PyModule_GetDict(module);
2797 if (parser_error == 0)
2798 parser_error = PyErr_NewException("parser.ParserError", NULL, NULL);
2800 if ((parser_error == 0)
2801 || (PyDict_SetItemString(dict, "ParserError", parser_error) != 0))
2803 /* caller will check PyErr_Occurred() */
2804 return;
2807 * Nice to have, but don't cry if we fail.
2809 Py_INCREF(&PyAST_Type);
2810 PyDict_SetItemString(dict, "ASTType", (PyObject*)&PyAST_Type);
2812 PyDict_SetItemString(dict, "__copyright__",
2813 PyString_FromString(parser_copyright_string));
2814 PyDict_SetItemString(dict, "__doc__",
2815 PyString_FromString(parser_doc_string));
2816 PyDict_SetItemString(dict, "__version__",
2817 PyString_FromString(parser_version_string));
2819 /* register to support pickling */
2820 module = PyImport_ImportModule("copy_reg");
2821 if (module != NULL) {
2822 PyObject *func, *pickler;
2824 func = PyObject_GetAttrString(module, "pickle");
2825 pickle_constructor = PyDict_GetItemString(dict, "sequence2ast");
2826 pickler = PyDict_GetItemString(dict, "_pickler");
2827 Py_XINCREF(pickle_constructor);
2828 if ((func != NULL) && (pickle_constructor != NULL)
2829 && (pickler != NULL)) {
2830 PyObject *res;
2832 res = PyObject_CallFunction(
2833 func, "OOO", &PyAST_Type, pickler, pickle_constructor);
2834 Py_XDECREF(res);
2836 Py_XDECREF(func);
2837 Py_DECREF(module);