/* parsermodule.c * * Copyright 1995-1996 by Fred L. Drake, Jr. and Virginia Polytechnic * Institute and State University, Blacksburg, Virginia, USA. * Portions copyright 1991-1995 by Stichting Mathematisch Centrum, * Amsterdam, The Netherlands. Copying is permitted under the terms * associated with the main Python distribution, with the additional * restriction that this additional notice be included and maintained * on all distributed copies. * * This module serves to replace the original parser module written * by Guido. The functionality is not matched precisely, but the * original may be implemented on top of this. This is desirable * since the source of the text to be parsed is now divorced from * this interface. * * Unlike the prior interface, the ability to give a parse tree * produced by Python code as a tuple to the compiler is enabled by * this module. See the documentation for more details. * * I've added some annotations that help with the lint code-checking * program, but they're not complete by a long shot. The real errors * that lint detects are gone, but there are still warnings with * Py_[X]DECREF() and Py_[X]INCREF() macros. The lint annotations * look like "NOTE(...)". */ #include "Python.h" /* general Python API */ #include "graminit.h" /* symbols defined in the grammar */ #include "node.h" /* internal parser structure */ #include "token.h" /* token definitions */ /* ISTERMINAL() / ISNONTERMINAL() */ #include "compile.h" /* PyNode_Compile() */ #ifdef lint #include #else #define NOTE(x) #endif #ifdef macintosh char *strdup(char *); #endif /* String constants used to initialize module attributes. * */ static char* parser_copyright_string = "Copyright 1995-1996 by Virginia Polytechnic Institute & State\n\ University, Blacksburg, Virginia, USA, and Fred L. Drake, Jr., Reston,\n\ Virginia, USA. Portions copyright 1991-1995 by Stichting Mathematisch\n\ Centrum, Amsterdam, The Netherlands."; static char* parser_doc_string = "This is an interface to Python's internal parser."; static char* parser_version_string = "0.5"; typedef PyObject* (*SeqMaker) (int length); typedef int (*SeqInserter) (PyObject* sequence, int index, PyObject* element); /* The function below is copyrighted by Stichting Mathematisch Centrum. The * original copyright statement is included below, and continues to apply * in full to the function immediately following. All other material is * original, copyrighted by Fred L. Drake, Jr. and Virginia Polytechnic * Institute and State University. Changes were made to comply with the * new naming conventions. Added arguments to provide support for creating * lists as well as tuples, and optionally including the line numbers. */ static PyObject* node2tuple(node *n, /* node to convert */ SeqMaker mkseq, /* create sequence */ SeqInserter addelem, /* func. to add elem. in seq. */ int lineno) /* include line numbers? */ { if (n == NULL) { Py_INCREF(Py_None); return (Py_None); } if (ISNONTERMINAL(TYPE(n))) { int i; PyObject *v; PyObject *w; v = mkseq(1 + NCH(n)); if (v == NULL) return (v); w = PyInt_FromLong(TYPE(n)); if (w == NULL) { Py_DECREF(v); return ((PyObject*) NULL); } (void) addelem(v, 0, w); for (i = 0; i < NCH(n); i++) { w = node2tuple(CHILD(n, i), mkseq, addelem, lineno); if (w == NULL) { Py_DECREF(v); return ((PyObject*) NULL); } (void) addelem(v, i+1, w); } return (v); } else if (ISTERMINAL(TYPE(n))) { PyObject *result = mkseq(2 + lineno); if (result != NULL) { (void) addelem(result, 0, PyInt_FromLong(TYPE(n))); (void) addelem(result, 1, PyString_FromString(STR(n))); if (lineno == 1) (void) addelem(result, 2, PyInt_FromLong(n->n_lineno)); } return (result); } else { PyErr_SetString(PyExc_SystemError, "unrecognized parse tree node type"); return ((PyObject*) NULL); } } /* * End of material copyrighted by Stichting Mathematisch Centrum. */ /* There are two types of intermediate objects we're interested in: * 'eval' and 'exec' types. These constants can be used in the st_type * field of the object type to identify which any given object represents. * These should probably go in an external header to allow other extensions * to use them, but then, we really should be using C++ too. ;-) */ #define PyST_EXPR 1 #define PyST_SUITE 2 /* These are the internal objects and definitions required to implement the * ST type. Most of the internal names are more reminiscent of the 'old' * naming style, but the code uses the new naming convention. */ static PyObject* parser_error = 0; typedef struct { PyObject_HEAD /* standard object header */ node* st_node; /* the node* returned by the parser */ int st_type; /* EXPR or SUITE ? */ } PyST_Object; staticforward void parser_free(PyST_Object *st); staticforward int parser_compare(PyST_Object *left, PyST_Object *right); staticforward PyObject * parser_getattr(PyObject *self, char *name); static PyTypeObject PyST_Type = { PyObject_HEAD_INIT(NULL) 0, "st", /* tp_name */ (int) sizeof(PyST_Object), /* tp_basicsize */ 0, /* tp_itemsize */ (destructor)parser_free, /* tp_dealloc */ 0, /* tp_print */ parser_getattr, /* tp_getattr */ 0, /* tp_setattr */ (cmpfunc)parser_compare, /* tp_compare */ 0, /* tp_repr */ 0, /* tp_as_number */ 0, /* tp_as_sequence */ 0, /* tp_as_mapping */ 0, /* tp_hash */ 0, /* tp_call */ 0, /* tp_str */ 0, /* tp_getattro */ 0, /* tp_setattro */ /* Functions to access object as input/output buffer */ 0, /* tp_as_buffer */ Py_TPFLAGS_DEFAULT, /* tp_flags */ /* __doc__ */ "Intermediate representation of a Python parse tree." }; /* PyST_Type */ static int parser_compare_nodes(node *left, node *right) { int j; if (TYPE(left) < TYPE(right)) return (-1); if (TYPE(right) < TYPE(left)) return (1); if (ISTERMINAL(TYPE(left))) return (strcmp(STR(left), STR(right))); if (NCH(left) < NCH(right)) return (-1); if (NCH(right) < NCH(left)) return (1); for (j = 0; j < NCH(left); ++j) { int v = parser_compare_nodes(CHILD(left, j), CHILD(right, j)); if (v != 0) return (v); } return (0); } /* int parser_compare(PyST_Object* left, PyST_Object* right) * * Comparison function used by the Python operators ==, !=, <, >, <=, >= * This really just wraps a call to parser_compare_nodes() with some easy * checks and protection code. * */ static int parser_compare(PyST_Object *left, PyST_Object *right) { if (left == right) return (0); if ((left == 0) || (right == 0)) return (-1); return (parser_compare_nodes(left->st_node, right->st_node)); } /* parser_newstobject(node* st) * * Allocates a new Python object representing an ST. This is simply the * 'wrapper' object that holds a node* and allows it to be passed around in * Python code. * */ static PyObject* parser_newstobject(node *st, int type) { PyST_Object* o = PyObject_New(PyST_Object, &PyST_Type); if (o != 0) { o->st_node = st; o->st_type = type; } else { PyNode_Free(st); } return ((PyObject*)o); } /* void parser_free(PyST_Object* st) * * This is called by a del statement that reduces the reference count to 0. * */ static void parser_free(PyST_Object *st) { PyNode_Free(st->st_node); PyObject_Del(st); } /* parser_st2tuple(PyObject* self, PyObject* args, PyObject* kw) * * This provides conversion from a node* to a tuple object that can be * returned to the Python-level caller. The ST object is not modified. * */ static PyObject* parser_st2tuple(PyST_Object *self, PyObject *args, PyObject *kw) { PyObject *line_option = 0; PyObject *res = 0; int ok; static char *keywords[] = {"ast", "line_info", NULL}; if (self == NULL) { ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|O:st2tuple", keywords, &PyST_Type, &self, &line_option); } else ok = PyArg_ParseTupleAndKeywords(args, kw, "|O:totuple", &keywords[1], &line_option); if (ok != 0) { int lineno = 0; if (line_option != NULL) { lineno = (PyObject_IsTrue(line_option) != 0) ? 1 : 0; } /* * Convert ST into a tuple representation. Use Guido's function, * since it's known to work already. */ res = node2tuple(((PyST_Object*)self)->st_node, PyTuple_New, PyTuple_SetItem, lineno); } return (res); } /* parser_st2list(PyObject* self, PyObject* args, PyObject* kw) * * This provides conversion from a node* to a list object that can be * returned to the Python-level caller. The ST object is not modified. * */ static PyObject* parser_st2list(PyST_Object *self, PyObject *args, PyObject *kw) { PyObject *line_option = 0; PyObject *res = 0; int ok; static char *keywords[] = {"ast", "line_info", NULL}; if (self == NULL) ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|O:st2list", keywords, &PyST_Type, &self, &line_option); else ok = PyArg_ParseTupleAndKeywords(args, kw, "|O:tolist", &keywords[1], &line_option); if (ok) { int lineno = 0; if (line_option != 0) { lineno = PyObject_IsTrue(line_option) ? 1 : 0; } /* * Convert ST into a tuple representation. Use Guido's function, * since it's known to work already. */ res = node2tuple(self->st_node, PyList_New, PyList_SetItem, lineno); } return (res); } /* parser_compilest(PyObject* self, PyObject* args) * * This function creates code objects from the parse tree represented by * the passed-in data object. An optional file name is passed in as well. * */ static PyObject* parser_compilest(PyST_Object *self, PyObject *args, PyObject *kw) { PyObject* res = 0; char* str = ""; int ok; static char *keywords[] = {"ast", "filename", NULL}; if (self == NULL) ok = PyArg_ParseTupleAndKeywords(args, kw, "O!|s:compilest", keywords, &PyST_Type, &self, &str); else ok = PyArg_ParseTupleAndKeywords(args, kw, "|s:compile", &keywords[1], &str); if (ok) res = (PyObject *)PyNode_Compile(self->st_node, str); return (res); } /* PyObject* parser_isexpr(PyObject* self, PyObject* args) * PyObject* parser_issuite(PyObject* self, PyObject* args) * * Checks the passed-in ST object to determine if it is an expression or * a statement suite, respectively. The return is a Python truth value. * */ static PyObject* parser_isexpr(PyST_Object *self, PyObject *args, PyObject *kw) { PyObject* res = 0; int ok; static char *keywords[] = {"ast", NULL}; if (self == NULL) ok = PyArg_ParseTupleAndKeywords(args, kw, "O!:isexpr", keywords, &PyST_Type, &self); else ok = PyArg_ParseTupleAndKeywords(args, kw, ":isexpr", &keywords[1]); if (ok) { /* Check to see if the ST represents an expression or not. */ res = (self->st_type == PyST_EXPR) ? Py_True : Py_False; Py_INCREF(res); } return (res); } static PyObject* parser_issuite(PyST_Object *self, PyObject *args, PyObject *kw) { PyObject* res = 0; int ok; static char *keywords[] = {"ast", NULL}; if (self == NULL) ok = PyArg_ParseTupleAndKeywords(args, kw, "O!:issuite", keywords, &PyST_Type, &self); else ok = PyArg_ParseTupleAndKeywords(args, kw, ":issuite", &keywords[1]); if (ok) { /* Check to see if the ST represents an expression or not. */ res = (self->st_type == PyST_EXPR) ? Py_False : Py_True; Py_INCREF(res); } return (res); } #define PUBLIC_METHOD_TYPE (METH_VARARGS|METH_KEYWORDS) static PyMethodDef parser_methods[] = { {"compile", (PyCFunction)parser_compilest, PUBLIC_METHOD_TYPE, "Compile this ST object into a code object."}, {"isexpr", (PyCFunction)parser_isexpr, PUBLIC_METHOD_TYPE, "Determines if this ST object was created from an expression."}, {"issuite", (PyCFunction)parser_issuite, PUBLIC_METHOD_TYPE, "Determines if this ST object was created from a suite."}, {"tolist", (PyCFunction)parser_st2list, PUBLIC_METHOD_TYPE, "Creates a list-tree representation of this ST."}, {"totuple", (PyCFunction)parser_st2tuple, PUBLIC_METHOD_TYPE, "Creates a tuple-tree representation of this ST."}, {NULL, NULL, 0, NULL} }; static PyObject* parser_getattr(PyObject *self, char *name) { return (Py_FindMethod(parser_methods, self, name)); } /* err_string(char* message) * * Sets the error string for an exception of type ParserError. * */ static void err_string(char *message) { PyErr_SetString(parser_error, message); } /* PyObject* parser_do_parse(PyObject* args, int type) * * Internal function to actually execute the parse and return the result if * successful, or set an exception if not. * */ static PyObject* parser_do_parse(PyObject *args, PyObject *kw, char *argspec, int type) { char* string = 0; PyObject* res = 0; static char *keywords[] = {"source", NULL}; if (PyArg_ParseTupleAndKeywords(args, kw, argspec, keywords, &string)) { node* n = PyParser_SimpleParseString(string, (type == PyST_EXPR) ? eval_input : file_input); if (n != 0) res = parser_newstobject(n, type); else err_string("could not parse string"); } return (res); } /* PyObject* parser_expr(PyObject* self, PyObject* args) * PyObject* parser_suite(PyObject* self, PyObject* args) * * External interfaces to the parser itself. Which is called determines if * the parser attempts to recognize an expression ('eval' form) or statement * suite ('exec' form). The real work is done by parser_do_parse() above. * */ static PyObject* parser_expr(PyST_Object *self, PyObject *args, PyObject *kw) { NOTE(ARGUNUSED(self)) return (parser_do_parse(args, kw, "s:expr", PyST_EXPR)); } static PyObject* parser_suite(PyST_Object *self, PyObject *args, PyObject *kw) { NOTE(ARGUNUSED(self)) return (parser_do_parse(args, kw, "s:suite", PyST_SUITE)); } /* This is the messy part of the code. Conversion from a tuple to an ST * object requires that the input tuple be valid without having to rely on * catching an exception from the compiler. This is done to allow the * compiler itself to remain fast, since most of its input will come from * the parser directly, and therefore be known to be syntactically correct. * This validation is done to ensure that we don't core dump the compile * phase, returning an exception instead. * * Two aspects can be broken out in this code: creating a node tree from * the tuple passed in, and verifying that it is indeed valid. It may be * advantageous to expand the number of ST types to include funcdefs and * lambdadefs to take advantage of the optimizer, recognizing those STs * here. They are not necessary, and not quite as useful in a raw form. * For now, let's get expressions and suites working reliably. */ staticforward node* build_node_tree(PyObject *tuple); staticforward int validate_expr_tree(node *tree); staticforward int validate_file_input(node *tree); /* PyObject* parser_tuple2st(PyObject* self, PyObject* args) * * This is the public function, called from the Python code. It receives a * single tuple object from the caller, and creates an ST object if the * tuple can be validated. It does this by checking the first code of the * tuple, and, if acceptable, builds the internal representation. If this * step succeeds, the internal representation is validated as fully as * possible with the various validate_*() routines defined below. * * This function must be changed if support is to be added for PyST_FRAGMENT * ST objects. * */ static PyObject* parser_tuple2st(PyST_Object *self, PyObject *args, PyObject *kw) { NOTE(ARGUNUSED(self)) PyObject *st = 0; PyObject *tuple; node *tree; static char *keywords[] = {"sequence", NULL}; if (!PyArg_ParseTupleAndKeywords(args, kw, "O:sequence2st", keywords, &tuple)) return (0); if (!PySequence_Check(tuple)) { PyErr_SetString(PyExc_ValueError, "sequence2st() requires a single sequence argument"); return (0); } /* * Convert the tree to the internal form before checking it. */ tree = build_node_tree(tuple); if (tree != 0) { int start_sym = TYPE(tree); if (start_sym == eval_input) { /* Might be an eval form. */ if (validate_expr_tree(tree)) st = parser_newstobject(tree, PyST_EXPR); } else if (start_sym == file_input) { /* This looks like an exec form so far. */ if (validate_file_input(tree)) st = parser_newstobject(tree, PyST_SUITE); } else { /* This is a fragment, at best. */ PyNode_Free(tree); err_string("parse tree does not use a valid start symbol"); } } /* Make sure we throw an exception on all errors. We should never * get this, but we'd do well to be sure something is done. */ if (st == NULL && !PyErr_Occurred()) err_string("unspecified ST error occurred"); return st; } /* node* build_node_children() * * Iterate across the children of the current non-terminal node and build * their structures. If successful, return the root of this portion of * the tree, otherwise, 0. Any required exception will be specified already, * and no memory will have been deallocated. * */ static node* build_node_children(PyObject *tuple, node *root, int *line_num) { int len = PyObject_Size(tuple); int i; for (i = 1; i < len; ++i) { /* elem must always be a sequence, however simple */ PyObject* elem = PySequence_GetItem(tuple, i); int ok = elem != NULL; long type = 0; char *strn = 0; if (ok) ok = PySequence_Check(elem); if (ok) { PyObject *temp = PySequence_GetItem(elem, 0); if (temp == NULL) ok = 0; else { ok = PyInt_Check(temp); if (ok) type = PyInt_AS_LONG(temp); Py_DECREF(temp); } } if (!ok) { PyErr_SetObject(parser_error, Py_BuildValue("os", elem, "Illegal node construct.")); Py_XDECREF(elem); return (0); } if (ISTERMINAL(type)) { int len = PyObject_Size(elem); PyObject *temp; if ((len != 2) && (len != 3)) { err_string("terminal nodes must have 2 or 3 entries"); return 0; } temp = PySequence_GetItem(elem, 1); if (temp == NULL) return 0; if (!PyString_Check(temp)) { PyErr_Format(parser_error, "second item in terminal node must be a string," " found %s", ((PyTypeObject*)PyObject_Type(temp))->tp_name); Py_DECREF(temp); return 0; } if (len == 3) { PyObject *o = PySequence_GetItem(elem, 2); if (o != NULL) { if (PyInt_Check(o)) *line_num = PyInt_AS_LONG(o); else { PyErr_Format(parser_error, "third item in terminal node must be an" " integer, found %s", ((PyTypeObject*)PyObject_Type(temp))->tp_name); Py_DECREF(o); Py_DECREF(temp); return 0; } Py_DECREF(o); } } len = PyString_GET_SIZE(temp) + 1; strn = (char *)PyMem_MALLOC(len); if (strn != NULL) (void) memcpy(strn, PyString_AS_STRING(temp), len); Py_DECREF(temp); } else if (!ISNONTERMINAL(type)) { /* * It has to be one or the other; this is an error. * Throw an exception. */ PyErr_SetObject(parser_error, Py_BuildValue("os", elem, "unknown node type.")); Py_XDECREF(elem); return (0); } PyNode_AddChild(root, type, strn, *line_num); if (ISNONTERMINAL(type)) { node* new_child = CHILD(root, i - 1); if (new_child != build_node_children(elem, new_child, line_num)) { Py_XDECREF(elem); return (0); } } else if (type == NEWLINE) { /* It's true: we increment the */ ++(*line_num); /* line number *after* the newline! */ } Py_XDECREF(elem); } return (root); } static node* build_node_tree(PyObject *tuple) { node* res = 0; PyObject *temp = PySequence_GetItem(tuple, 0); long num = -1; if (temp != NULL) num = PyInt_AsLong(temp); Py_XDECREF(temp); if (ISTERMINAL(num)) { /* * The tuple is simple, but it doesn't start with a start symbol. * Throw an exception now and be done with it. */ tuple = Py_BuildValue("os", tuple, "Illegal syntax-tree; cannot start with terminal symbol."); PyErr_SetObject(parser_error, tuple); } else if (ISNONTERMINAL(num)) { /* * Not efficient, but that can be handled later. */ int line_num = 0; res = PyNode_New(num); if (res != build_node_children(tuple, res, &line_num)) { PyNode_Free(res); res = 0; } } else /* The tuple is illegal -- if the number is neither TERMINAL nor * NONTERMINAL, we can't use it. Not sure the implementation * allows this condition, but the API doesn't preclude it. */ PyErr_SetObject(parser_error, Py_BuildValue("os", tuple, "Illegal component tuple.")); return (res); } /* * Validation routines used within the validation section: */ staticforward int validate_terminal(node *terminal, int type, char *string); #define validate_ampersand(ch) validate_terminal(ch, AMPER, "&") #define validate_circumflex(ch) validate_terminal(ch, CIRCUMFLEX, "^") #define validate_colon(ch) validate_terminal(ch, COLON, ":") #define validate_comma(ch) validate_terminal(ch, COMMA, ",") #define validate_dedent(ch) validate_terminal(ch, DEDENT, "") #define validate_equal(ch) validate_terminal(ch, EQUAL, "=") #define validate_indent(ch) validate_terminal(ch, INDENT, (char*)NULL) #define validate_lparen(ch) validate_terminal(ch, LPAR, "(") #define validate_newline(ch) validate_terminal(ch, NEWLINE, (char*)NULL) #define validate_rparen(ch) validate_terminal(ch, RPAR, ")") #define validate_semi(ch) validate_terminal(ch, SEMI, ";") #define validate_star(ch) validate_terminal(ch, STAR, "*") #define validate_vbar(ch) validate_terminal(ch, VBAR, "|") #define validate_doublestar(ch) validate_terminal(ch, DOUBLESTAR, "**") #define validate_dot(ch) validate_terminal(ch, DOT, ".") #define validate_name(ch, str) validate_terminal(ch, NAME, str) #define VALIDATER(n) static int validate_##n(node *tree) VALIDATER(node); VALIDATER(small_stmt); VALIDATER(class); VALIDATER(node); VALIDATER(parameters); VALIDATER(suite); VALIDATER(testlist); VALIDATER(varargslist); VALIDATER(fpdef); VALIDATER(fplist); VALIDATER(stmt); VALIDATER(simple_stmt); VALIDATER(expr_stmt); VALIDATER(power); VALIDATER(print_stmt); VALIDATER(del_stmt); VALIDATER(return_stmt); VALIDATER(list_iter); VALIDATER(raise_stmt); VALIDATER(import_stmt); VALIDATER(global_stmt); VALIDATER(list_if); VALIDATER(assert_stmt); VALIDATER(list_for); VALIDATER(exec_stmt); VALIDATER(compound_stmt); VALIDATER(while); VALIDATER(for); VALIDATER(try); VALIDATER(except_clause); VALIDATER(test); VALIDATER(and_test); VALIDATER(not_test); VALIDATER(comparison); VALIDATER(comp_op); VALIDATER(expr); VALIDATER(xor_expr); VALIDATER(and_expr); VALIDATER(shift_expr); VALIDATER(arith_expr); VALIDATER(term); VALIDATER(factor); VALIDATER(atom); VALIDATER(lambdef); VALIDATER(trailer); VALIDATER(subscript); VALIDATER(subscriptlist); VALIDATER(sliceop); VALIDATER(exprlist); VALIDATER(dictmaker); VALIDATER(arglist); VALIDATER(argument); VALIDATER(listmaker); VALIDATER(yield_stmt); #undef VALIDATER #define is_even(n) (((n) & 1) == 0) #define is_odd(n) (((n) & 1) == 1) static int validate_ntype(node *n, int t) { if (TYPE(n) != t) { PyErr_Format(parser_error, "Expected node type %d, got %d.", t, TYPE(n)); return 0; } return 1; } /* Verifies that the number of child nodes is exactly 'num', raising * an exception if it isn't. The exception message does not indicate * the exact number of nodes, allowing this to be used to raise the * "right" exception when the wrong number of nodes is present in a * specific variant of a statement's syntax. This is commonly used * in that fashion. */ static int validate_numnodes(node *n, int num, const char *const name) { if (NCH(n) != num) { PyErr_Format(parser_error, "Illegal number of children for %s node.", name); return 0; } return 1; } static int validate_terminal(node *terminal, int type, char *string) { int res = (validate_ntype(terminal, type) && ((string == 0) || (strcmp(string, STR(terminal)) == 0))); if (!res && !PyErr_Occurred()) { PyErr_Format(parser_error, "Illegal terminal: expected \"%s\"", string); } return (res); } /* X (',' X) [','] */ static int validate_repeating_list(node *tree, int ntype, int (*vfunc)(node *), const char *const name) { int nch = NCH(tree); int res = (nch && validate_ntype(tree, ntype) && vfunc(CHILD(tree, 0))); if (!res && !PyErr_Occurred()) (void) validate_numnodes(tree, 1, name); else { if (is_even(nch)) res = validate_comma(CHILD(tree, --nch)); if (res && nch > 1) { int pos = 1; for ( ; res && pos < nch; pos += 2) res = (validate_comma(CHILD(tree, pos)) && vfunc(CHILD(tree, pos + 1))); } } return (res); } /* validate_class() * * classdef: * 'class' NAME ['(' testlist ')'] ':' suite */ static int validate_class(node *tree) { int nch = NCH(tree); int res = validate_ntype(tree, classdef) && ((nch == 4) || (nch == 7)); if (res) { res = (validate_name(CHILD(tree, 0), "class") && validate_ntype(CHILD(tree, 1), NAME) && validate_colon(CHILD(tree, nch - 2)) && validate_suite(CHILD(tree, nch - 1))); } else (void) validate_numnodes(tree, 4, "class"); if (res && (nch == 7)) { res = (validate_lparen(CHILD(tree, 2)) && validate_testlist(CHILD(tree, 3)) && validate_rparen(CHILD(tree, 4))); } return (res); } /* if_stmt: * 'if' test ':' suite ('elif' test ':' suite)* ['else' ':' suite] */ static int validate_if(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, if_stmt) && (nch >= 4) && validate_name(CHILD(tree, 0), "if") && validate_test(CHILD(tree, 1)) && validate_colon(CHILD(tree, 2)) && validate_suite(CHILD(tree, 3))); if (res && ((nch % 4) == 3)) { /* ... 'else' ':' suite */ res = (validate_name(CHILD(tree, nch - 3), "else") && validate_colon(CHILD(tree, nch - 2)) && validate_suite(CHILD(tree, nch - 1))); nch -= 3; } else if (!res && !PyErr_Occurred()) (void) validate_numnodes(tree, 4, "if"); if ((nch % 4) != 0) /* Will catch the case for nch < 4 */ res = validate_numnodes(tree, 0, "if"); else if (res && (nch > 4)) { /* ... ('elif' test ':' suite)+ ... */ int j = 4; while ((j < nch) && res) { res = (validate_name(CHILD(tree, j), "elif") && validate_colon(CHILD(tree, j + 2)) && validate_test(CHILD(tree, j + 1)) && validate_suite(CHILD(tree, j + 3))); j += 4; } } return (res); } /* parameters: * '(' [varargslist] ')' * */ static int validate_parameters(node *tree) { int nch = NCH(tree); int res = validate_ntype(tree, parameters) && ((nch == 2) || (nch == 3)); if (res) { res = (validate_lparen(CHILD(tree, 0)) && validate_rparen(CHILD(tree, nch - 1))); if (res && (nch == 3)) res = validate_varargslist(CHILD(tree, 1)); } else { (void) validate_numnodes(tree, 2, "parameters"); } return (res); } /* validate_suite() * * suite: * simple_stmt * | NEWLINE INDENT stmt+ DEDENT */ static int validate_suite(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, suite) && ((nch == 1) || (nch >= 4))); if (res && (nch == 1)) res = validate_simple_stmt(CHILD(tree, 0)); else if (res) { /* NEWLINE INDENT stmt+ DEDENT */ res = (validate_newline(CHILD(tree, 0)) && validate_indent(CHILD(tree, 1)) && validate_stmt(CHILD(tree, 2)) && validate_dedent(CHILD(tree, nch - 1))); if (res && (nch > 4)) { int i = 3; --nch; /* forget the DEDENT */ for ( ; res && (i < nch); ++i) res = validate_stmt(CHILD(tree, i)); } else if (nch < 4) res = validate_numnodes(tree, 4, "suite"); } return (res); } static int validate_testlist(node *tree) { return (validate_repeating_list(tree, testlist, validate_test, "testlist")); } /* '*' NAME [',' '**' NAME] | '**' NAME */ static int validate_varargslist_trailer(node *tree, int start) { int nch = NCH(tree); int res = 0; int sym; if (nch <= start) { err_string("expected variable argument trailer for varargslist"); return 0; } sym = TYPE(CHILD(tree, start)); if (sym == STAR) { /* * ('*' NAME [',' '**' NAME] */ if (nch-start == 2) res = validate_name(CHILD(tree, start+1), NULL); else if (nch-start == 5) res = (validate_name(CHILD(tree, start+1), NULL) && validate_comma(CHILD(tree, start+2)) && validate_doublestar(CHILD(tree, start+3)) && validate_name(CHILD(tree, start+4), NULL)); } else if (sym == DOUBLESTAR) { /* * '**' NAME */ if (nch-start == 2) res = validate_name(CHILD(tree, start+1), NULL); } if (!res) err_string("illegal variable argument trailer for varargslist"); return res; } /* validate_varargslist() * * varargslist: * (fpdef ['=' test] ',')* * ('*' NAME [',' '**' NAME] * | '**' NAME) * | fpdef ['=' test] (',' fpdef ['=' test])* [','] * */ static int validate_varargslist(node *tree) { int nch = NCH(tree); int res = validate_ntype(tree, varargslist) && (nch != 0); int sym; if (!res) return 0; if (nch < 1) { err_string("varargslist missing child nodes"); return 0; } sym = TYPE(CHILD(tree, 0)); if (sym == STAR || sym == DOUBLESTAR) /* whole thing matches: * '*' NAME [',' '**' NAME] | '**' NAME */ res = validate_varargslist_trailer(tree, 0); else if (sym == fpdef) { int i = 0; sym = TYPE(CHILD(tree, nch-1)); if (sym == NAME) { /* * (fpdef ['=' test] ',')+ * ('*' NAME [',' '**' NAME] * | '**' NAME) */ /* skip over (fpdef ['=' test] ',')+ */ while (res && (i+2 <= nch)) { res = validate_fpdef(CHILD(tree, i)); ++i; if (res && TYPE(CHILD(tree, i)) == EQUAL && (i+2 <= nch)) { res = (validate_equal(CHILD(tree, i)) && validate_test(CHILD(tree, i+1))); if (res) i += 2; } if (res && i < nch) { res = validate_comma(CHILD(tree, i)); ++i; if (res && i < nch && (TYPE(CHILD(tree, i)) == DOUBLESTAR || TYPE(CHILD(tree, i)) == STAR)) break; } } /* ... '*' NAME [',' '**' NAME] | '**' NAME * i --^^^ */ if (res) res = validate_varargslist_trailer(tree, i); } else { /* * fpdef ['=' test] (',' fpdef ['=' test])* [','] */ /* strip trailing comma node */ if (sym == COMMA) { res = validate_comma(CHILD(tree, nch-1)); if (!res) return 0; --nch; } /* * fpdef ['=' test] (',' fpdef ['=' test])* */ res = validate_fpdef(CHILD(tree, 0)); ++i; if (res && (i+2 <= nch) && TYPE(CHILD(tree, i)) == EQUAL) { res = (validate_equal(CHILD(tree, i)) && validate_test(CHILD(tree, i+1))); i += 2; } /* * ... (',' fpdef ['=' test])* * i ---^^^ */ while (res && (nch - i) >= 2) { res = (validate_comma(CHILD(tree, i)) && validate_fpdef(CHILD(tree, i+1))); i += 2; if (res && (nch - i) >= 2 && TYPE(CHILD(tree, i)) == EQUAL) { res = (validate_equal(CHILD(tree, i)) && validate_test(CHILD(tree, i+1))); i += 2; } } if (res && nch - i != 0) { res = 0; err_string("illegal formation for varargslist"); } } } return res; } /* list_iter: list_for | list_if */ static int validate_list_iter(node *tree) { int res = (validate_ntype(tree, list_iter) && validate_numnodes(tree, 1, "list_iter")); if (res && TYPE(CHILD(tree, 0)) == list_for) res = validate_list_for(CHILD(tree, 0)); else res = validate_list_if(CHILD(tree, 0)); return res; } /* list_for: 'for' exprlist 'in' testlist [list_iter] */ static int validate_list_for(node *tree) { int nch = NCH(tree); int res; if (nch == 5) res = validate_list_iter(CHILD(tree, 4)); else res = validate_numnodes(tree, 4, "list_for"); if (res) res = (validate_name(CHILD(tree, 0), "for") && validate_exprlist(CHILD(tree, 1)) && validate_name(CHILD(tree, 2), "in") && validate_testlist(CHILD(tree, 3))); return res; } /* list_if: 'if' test [list_iter] */ static int validate_list_if(node *tree) { int nch = NCH(tree); int res; if (nch == 3) res = validate_list_iter(CHILD(tree, 2)); else res = validate_numnodes(tree, 2, "list_if"); if (res) res = (validate_name(CHILD(tree, 0), "if") && validate_test(CHILD(tree, 1))); return res; } /* validate_fpdef() * * fpdef: * NAME * | '(' fplist ')' */ static int validate_fpdef(node *tree) { int nch = NCH(tree); int res = validate_ntype(tree, fpdef); if (res) { if (nch == 1) res = validate_ntype(CHILD(tree, 0), NAME); else if (nch == 3) res = (validate_lparen(CHILD(tree, 0)) && validate_fplist(CHILD(tree, 1)) && validate_rparen(CHILD(tree, 2))); else res = validate_numnodes(tree, 1, "fpdef"); } return (res); } static int validate_fplist(node *tree) { return (validate_repeating_list(tree, fplist, validate_fpdef, "fplist")); } /* simple_stmt | compound_stmt * */ static int validate_stmt(node *tree) { int res = (validate_ntype(tree, stmt) && validate_numnodes(tree, 1, "stmt")); if (res) { tree = CHILD(tree, 0); if (TYPE(tree) == simple_stmt) res = validate_simple_stmt(tree); else res = validate_compound_stmt(tree); } return (res); } /* small_stmt (';' small_stmt)* [';'] NEWLINE * */ static int validate_simple_stmt(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, simple_stmt) && (nch >= 2) && validate_small_stmt(CHILD(tree, 0)) && validate_newline(CHILD(tree, nch - 1))); if (nch < 2) res = validate_numnodes(tree, 2, "simple_stmt"); --nch; /* forget the NEWLINE */ if (res && is_even(nch)) res = validate_semi(CHILD(tree, --nch)); if (res && (nch > 2)) { int i; for (i = 1; res && (i < nch); i += 2) res = (validate_semi(CHILD(tree, i)) && validate_small_stmt(CHILD(tree, i + 1))); } return (res); } static int validate_small_stmt(node *tree) { int nch = NCH(tree); int res = validate_numnodes(tree, 1, "small_stmt"); if (res) { int ntype = TYPE(CHILD(tree, 0)); if ( (ntype == expr_stmt) || (ntype == print_stmt) || (ntype == del_stmt) || (ntype == pass_stmt) || (ntype == flow_stmt) || (ntype == import_stmt) || (ntype == global_stmt) || (ntype == assert_stmt) || (ntype == exec_stmt)) res = validate_node(CHILD(tree, 0)); else { res = 0; err_string("illegal small_stmt child type"); } } else if (nch == 1) { res = 0; PyErr_Format(parser_error, "Unrecognized child node of small_stmt: %d.", TYPE(CHILD(tree, 0))); } return (res); } /* compound_stmt: * if_stmt | while_stmt | for_stmt | try_stmt | funcdef | classdef */ static int validate_compound_stmt(node *tree) { int res = (validate_ntype(tree, compound_stmt) && validate_numnodes(tree, 1, "compound_stmt")); int ntype; if (!res) return (0); tree = CHILD(tree, 0); ntype = TYPE(tree); if ( (ntype == if_stmt) || (ntype == while_stmt) || (ntype == for_stmt) || (ntype == try_stmt) || (ntype == funcdef) || (ntype == classdef)) res = validate_node(tree); else { res = 0; PyErr_Format(parser_error, "Illegal compound statement type: %d.", TYPE(tree)); } return (res); } static int validate_expr_stmt(node *tree) { int j; int nch = NCH(tree); int res = (validate_ntype(tree, expr_stmt) && is_odd(nch) && validate_testlist(CHILD(tree, 0))); if (res && nch == 3 && TYPE(CHILD(tree, 1)) == augassign) { res = (validate_numnodes(CHILD(tree, 1), 1, "augassign") && validate_testlist(CHILD(tree, 2))); if (res) { char *s = STR(CHILD(CHILD(tree, 1), 0)); res = (strcmp(s, "+=") == 0 || strcmp(s, "-=") == 0 || strcmp(s, "*=") == 0 || strcmp(s, "/=") == 0 || strcmp(s, "%=") == 0 || strcmp(s, "&=") == 0 || strcmp(s, "|=") == 0 || strcmp(s, "^=") == 0 || strcmp(s, "<<=") == 0 || strcmp(s, ">>=") == 0 || strcmp(s, "**=") == 0); if (!res) err_string("illegal augmmented assignment operator"); } } else { for (j = 1; res && (j < nch); j += 2) res = (validate_equal(CHILD(tree, j)) && validate_testlist(CHILD(tree, j + 1))); } return (res); } /* print_stmt: * * 'print' ( [ test (',' test)* [','] ] * | '>>' test [ (',' test)+ [','] ] ) */ static int validate_print_stmt(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, print_stmt) && (nch > 0) && validate_name(CHILD(tree, 0), "print")); if (res && nch > 1) { int sym = TYPE(CHILD(tree, 1)); int i = 1; int allow_trailing_comma = 1; if (sym == test) res = validate_test(CHILD(tree, i++)); else { if (nch < 3) res = validate_numnodes(tree, 3, "print_stmt"); else { res = (validate_ntype(CHILD(tree, i), RIGHTSHIFT) && validate_test(CHILD(tree, i+1))); i += 2; allow_trailing_comma = 0; } } if (res) { /* ... (',' test)* [','] */ while (res && i+2 <= nch) { res = (validate_comma(CHILD(tree, i)) && validate_test(CHILD(tree, i+1))); allow_trailing_comma = 1; i += 2; } if (res && !allow_trailing_comma) res = validate_numnodes(tree, i, "print_stmt"); else if (res && i < nch) res = validate_comma(CHILD(tree, i)); } } return (res); } static int validate_del_stmt(node *tree) { return (validate_numnodes(tree, 2, "del_stmt") && validate_name(CHILD(tree, 0), "del") && validate_exprlist(CHILD(tree, 1))); } static int validate_return_stmt(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, return_stmt) && ((nch == 1) || (nch == 2)) && validate_name(CHILD(tree, 0), "return")); if (res && (nch == 2)) res = validate_testlist(CHILD(tree, 1)); return (res); } static int validate_raise_stmt(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, raise_stmt) && ((nch == 1) || (nch == 2) || (nch == 4) || (nch == 6))); if (res) { res = validate_name(CHILD(tree, 0), "raise"); if (res && (nch >= 2)) res = validate_test(CHILD(tree, 1)); if (res && nch > 2) { res = (validate_comma(CHILD(tree, 2)) && validate_test(CHILD(tree, 3))); if (res && (nch > 4)) res = (validate_comma(CHILD(tree, 4)) && validate_test(CHILD(tree, 5))); } } else (void) validate_numnodes(tree, 2, "raise"); if (res && (nch == 4)) res = (validate_comma(CHILD(tree, 2)) && validate_test(CHILD(tree, 3))); return (res); } /* yield_stmt: 'yield' testlist */ static int validate_yield_stmt(node *tree) { return (validate_ntype(tree, yield_stmt) && validate_numnodes(tree, 2, "yield_stmt") && validate_name(CHILD(tree, 0), "yield") && validate_testlist(CHILD(tree, 1))); } static int validate_import_as_name(node *tree) { int nch = NCH(tree); int ok = validate_ntype(tree, import_as_name); if (ok) { if (nch == 1) ok = validate_name(CHILD(tree, 0), NULL); else if (nch == 3) ok = (validate_name(CHILD(tree, 0), NULL) && validate_name(CHILD(tree, 1), "as") && validate_name(CHILD(tree, 2), NULL)); else ok = validate_numnodes(tree, 3, "import_as_name"); } return ok; } /* dotted_name: NAME ("." NAME)* */ static int validate_dotted_name(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, dotted_name) && is_odd(nch) && validate_name(CHILD(tree, 0), NULL)); int i; for (i = 1; res && (i < nch); i += 2) { res = (validate_dot(CHILD(tree, i)) && validate_name(CHILD(tree, i+1), NULL)); } return res; } /* dotted_as_name: dotted_name [NAME NAME] */ static int validate_dotted_as_name(node *tree) { int nch = NCH(tree); int res = validate_ntype(tree, dotted_as_name); if (res) { if (nch == 1) res = validate_dotted_name(CHILD(tree, 0)); else if (nch == 3) res = (validate_dotted_name(CHILD(tree, 0)) && validate_name(CHILD(tree, 1), "as") && validate_name(CHILD(tree, 2), NULL)); else { res = 0; err_string("illegal number of children for dotted_as_name"); } } return res; } /* import_stmt: * * 'import' dotted_as_name (',' dotted_as_name)* * | 'from' dotted_name 'import' ('*' | import_as_name (',' import_as_name)*) */ static int validate_import_stmt(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, import_stmt) && (nch >= 2) && is_even(nch) && validate_ntype(CHILD(tree, 0), NAME)); if (res && (strcmp(STR(CHILD(tree, 0)), "import") == 0)) { int j; res = validate_dotted_as_name(CHILD(tree, 1)); for (j = 2; res && (j < nch); j += 2) res = (validate_comma(CHILD(tree, j)) && validate_dotted_as_name(CHILD(tree, j + 1))); } else if (res && (res = validate_name(CHILD(tree, 0), "from"))) { res = ((nch >= 4) && is_even(nch) && validate_dotted_name(CHILD(tree, 1)) && validate_name(CHILD(tree, 2), "import")); if (nch == 4) { if (TYPE(CHILD(tree, 3)) == import_as_name) res = validate_import_as_name(CHILD(tree, 3)); else res = validate_star(CHILD(tree, 3)); } else { /* 'from' dotted_name 'import' import_as_name * (',' import_as_name)+ */ int j; res = validate_import_as_name(CHILD(tree, 3)); for (j = 4; res && (j < nch); j += 2) res = (validate_comma(CHILD(tree, j)) && validate_import_as_name(CHILD(tree, j + 1))); } } else res = 0; return (res); } static int validate_global_stmt(node *tree) { int j; int nch = NCH(tree); int res = (validate_ntype(tree, global_stmt) && is_even(nch) && (nch >= 2)); if (res) res = (validate_name(CHILD(tree, 0), "global") && validate_ntype(CHILD(tree, 1), NAME)); for (j = 2; res && (j < nch); j += 2) res = (validate_comma(CHILD(tree, j)) && validate_ntype(CHILD(tree, j + 1), NAME)); return (res); } /* exec_stmt: * * 'exec' expr ['in' test [',' test]] */ static int validate_exec_stmt(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, exec_stmt) && ((nch == 2) || (nch == 4) || (nch == 6)) && validate_name(CHILD(tree, 0), "exec") && validate_expr(CHILD(tree, 1))); if (!res && !PyErr_Occurred()) err_string("illegal exec statement"); if (res && (nch > 2)) res = (validate_name(CHILD(tree, 2), "in") && validate_test(CHILD(tree, 3))); if (res && (nch == 6)) res = (validate_comma(CHILD(tree, 4)) && validate_test(CHILD(tree, 5))); return (res); } /* assert_stmt: * * 'assert' test [',' test] */ static int validate_assert_stmt(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, assert_stmt) && ((nch == 2) || (nch == 4)) && (validate_name(CHILD(tree, 0), "__assert__") || validate_name(CHILD(tree, 0), "assert")) && validate_test(CHILD(tree, 1))); if (!res && !PyErr_Occurred()) err_string("illegal assert statement"); if (res && (nch > 2)) res = (validate_comma(CHILD(tree, 2)) && validate_test(CHILD(tree, 3))); return (res); } static int validate_while(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, while_stmt) && ((nch == 4) || (nch == 7)) && validate_name(CHILD(tree, 0), "while") && validate_test(CHILD(tree, 1)) && validate_colon(CHILD(tree, 2)) && validate_suite(CHILD(tree, 3))); if (res && (nch == 7)) res = (validate_name(CHILD(tree, 4), "else") && validate_colon(CHILD(tree, 5)) && validate_suite(CHILD(tree, 6))); return (res); } static int validate_for(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, for_stmt) && ((nch == 6) || (nch == 9)) && validate_name(CHILD(tree, 0), "for") && validate_exprlist(CHILD(tree, 1)) && validate_name(CHILD(tree, 2), "in") && validate_testlist(CHILD(tree, 3)) && validate_colon(CHILD(tree, 4)) && validate_suite(CHILD(tree, 5))); if (res && (nch == 9)) res = (validate_name(CHILD(tree, 6), "else") && validate_colon(CHILD(tree, 7)) && validate_suite(CHILD(tree, 8))); return (res); } /* try_stmt: * 'try' ':' suite (except_clause ':' suite)+ ['else' ':' suite] * | 'try' ':' suite 'finally' ':' suite * */ static int validate_try(node *tree) { int nch = NCH(tree); int pos = 3; int res = (validate_ntype(tree, try_stmt) && (nch >= 6) && ((nch % 3) == 0)); if (res) res = (validate_name(CHILD(tree, 0), "try") && validate_colon(CHILD(tree, 1)) && validate_suite(CHILD(tree, 2)) && validate_colon(CHILD(tree, nch - 2)) && validate_suite(CHILD(tree, nch - 1))); else if (!PyErr_Occurred()) { const char* name = "except"; if (TYPE(CHILD(tree, nch - 3)) != except_clause) name = STR(CHILD(tree, nch - 3)); PyErr_Format(parser_error, "Illegal number of children for try/%s node.", name); } /* Skip past except_clause sections: */ while (res && (TYPE(CHILD(tree, pos)) == except_clause)) { res = (validate_except_clause(CHILD(tree, pos)) && validate_colon(CHILD(tree, pos + 1)) && validate_suite(CHILD(tree, pos + 2))); pos += 3; } if (res && (pos < nch)) { res = validate_ntype(CHILD(tree, pos), NAME); if (res && (strcmp(STR(CHILD(tree, pos)), "finally") == 0)) res = (validate_numnodes(tree, 6, "try/finally") && validate_colon(CHILD(tree, 4)) && validate_suite(CHILD(tree, 5))); else if (res) { if (nch == (pos + 3)) { res = ((strcmp(STR(CHILD(tree, pos)), "except") == 0) || (strcmp(STR(CHILD(tree, pos)), "else") == 0)); if (!res) err_string("illegal trailing triple in try statement"); } else if (nch == (pos + 6)) { res = (validate_name(CHILD(tree, pos), "except") && validate_colon(CHILD(tree, pos + 1)) && validate_suite(CHILD(tree, pos + 2)) && validate_name(CHILD(tree, pos + 3), "else")); } else res = validate_numnodes(tree, pos + 3, "try/except"); } } return (res); } static int validate_except_clause(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, except_clause) && ((nch == 1) || (nch == 2) || (nch == 4)) && validate_name(CHILD(tree, 0), "except")); if (res && (nch > 1)) res = validate_test(CHILD(tree, 1)); if (res && (nch == 4)) res = (validate_comma(CHILD(tree, 2)) && validate_test(CHILD(tree, 3))); return (res); } static int validate_test(node *tree) { int nch = NCH(tree); int res = validate_ntype(tree, test) && is_odd(nch); if (res && (TYPE(CHILD(tree, 0)) == lambdef)) res = ((nch == 1) && validate_lambdef(CHILD(tree, 0))); else if (res) { int pos; res = validate_and_test(CHILD(tree, 0)); for (pos = 1; res && (pos < nch); pos += 2) res = (validate_name(CHILD(tree, pos), "or") && validate_and_test(CHILD(tree, pos + 1))); } return (res); } static int validate_and_test(node *tree) { int pos; int nch = NCH(tree); int res = (validate_ntype(tree, and_test) && is_odd(nch) && validate_not_test(CHILD(tree, 0))); for (pos = 1; res && (pos < nch); pos += 2) res = (validate_name(CHILD(tree, pos), "and") && validate_not_test(CHILD(tree, 0))); return (res); } static int validate_not_test(node *tree) { int nch = NCH(tree); int res = validate_ntype(tree, not_test) && ((nch == 1) || (nch == 2)); if (res) { if (nch == 2) res = (validate_name(CHILD(tree, 0), "not") && validate_not_test(CHILD(tree, 1))); else if (nch == 1) res = validate_comparison(CHILD(tree, 0)); } return (res); } static int validate_comparison(node *tree) { int pos; int nch = NCH(tree); int res = (validate_ntype(tree, comparison) && is_odd(nch) && validate_expr(CHILD(tree, 0))); for (pos = 1; res && (pos < nch); pos += 2) res = (validate_comp_op(CHILD(tree, pos)) && validate_expr(CHILD(tree, pos + 1))); return (res); } static int validate_comp_op(node *tree) { int res = 0; int nch = NCH(tree); if (!validate_ntype(tree, comp_op)) return (0); if (nch == 1) { /* * Only child will be a terminal with a well-defined symbolic name * or a NAME with a string of either 'is' or 'in' */ tree = CHILD(tree, 0); switch (TYPE(tree)) { case LESS: case GREATER: case EQEQUAL: case EQUAL: case LESSEQUAL: case GREATEREQUAL: case NOTEQUAL: res = 1; break; case NAME: res = ((strcmp(STR(tree), "in") == 0) || (strcmp(STR(tree), "is") == 0)); if (!res) { PyErr_Format(parser_error, "illegal operator '%s'", STR(tree)); } break; default: err_string("illegal comparison operator type"); break; } } else if ((res = validate_numnodes(tree, 2, "comp_op")) != 0) { res = (validate_ntype(CHILD(tree, 0), NAME) && validate_ntype(CHILD(tree, 1), NAME) && (((strcmp(STR(CHILD(tree, 0)), "is") == 0) && (strcmp(STR(CHILD(tree, 1)), "not") == 0)) || ((strcmp(STR(CHILD(tree, 0)), "not") == 0) && (strcmp(STR(CHILD(tree, 1)), "in") == 0)))); if (!res && !PyErr_Occurred()) err_string("unknown comparison operator"); } return (res); } static int validate_expr(node *tree) { int j; int nch = NCH(tree); int res = (validate_ntype(tree, expr) && is_odd(nch) && validate_xor_expr(CHILD(tree, 0))); for (j = 2; res && (j < nch); j += 2) res = (validate_xor_expr(CHILD(tree, j)) && validate_vbar(CHILD(tree, j - 1))); return (res); } static int validate_xor_expr(node *tree) { int j; int nch = NCH(tree); int res = (validate_ntype(tree, xor_expr) && is_odd(nch) && validate_and_expr(CHILD(tree, 0))); for (j = 2; res && (j < nch); j += 2) res = (validate_circumflex(CHILD(tree, j - 1)) && validate_and_expr(CHILD(tree, j))); return (res); } static int validate_and_expr(node *tree) { int pos; int nch = NCH(tree); int res = (validate_ntype(tree, and_expr) && is_odd(nch) && validate_shift_expr(CHILD(tree, 0))); for (pos = 1; res && (pos < nch); pos += 2) res = (validate_ampersand(CHILD(tree, pos)) && validate_shift_expr(CHILD(tree, pos + 1))); return (res); } static int validate_chain_two_ops(node *tree, int (*termvalid)(node *), int op1, int op2) { int pos = 1; int nch = NCH(tree); int res = (is_odd(nch) && (*termvalid)(CHILD(tree, 0))); for ( ; res && (pos < nch); pos += 2) { if (TYPE(CHILD(tree, pos)) != op1) res = validate_ntype(CHILD(tree, pos), op2); if (res) res = (*termvalid)(CHILD(tree, pos + 1)); } return (res); } static int validate_shift_expr(node *tree) { return (validate_ntype(tree, shift_expr) && validate_chain_two_ops(tree, validate_arith_expr, LEFTSHIFT, RIGHTSHIFT)); } static int validate_arith_expr(node *tree) { return (validate_ntype(tree, arith_expr) && validate_chain_two_ops(tree, validate_term, PLUS, MINUS)); } static int validate_term(node *tree) { int pos = 1; int nch = NCH(tree); int res = (validate_ntype(tree, term) && is_odd(nch) && validate_factor(CHILD(tree, 0))); for ( ; res && (pos < nch); pos += 2) res = (((TYPE(CHILD(tree, pos)) == STAR) || (TYPE(CHILD(tree, pos)) == SLASH) || (TYPE(CHILD(tree, pos)) == PERCENT)) && validate_factor(CHILD(tree, pos + 1))); return (res); } /* factor: * * factor: ('+'|'-'|'~') factor | power */ static int validate_factor(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, factor) && (((nch == 2) && ((TYPE(CHILD(tree, 0)) == PLUS) || (TYPE(CHILD(tree, 0)) == MINUS) || (TYPE(CHILD(tree, 0)) == TILDE)) && validate_factor(CHILD(tree, 1))) || ((nch == 1) && validate_power(CHILD(tree, 0))))); return (res); } /* power: * * power: atom trailer* ('**' factor)* */ static int validate_power(node *tree) { int pos = 1; int nch = NCH(tree); int res = (validate_ntype(tree, power) && (nch >= 1) && validate_atom(CHILD(tree, 0))); while (res && (pos < nch) && (TYPE(CHILD(tree, pos)) == trailer)) res = validate_trailer(CHILD(tree, pos++)); if (res && (pos < nch)) { if (!is_even(nch - pos)) { err_string("illegal number of nodes for 'power'"); return (0); } for ( ; res && (pos < (nch - 1)); pos += 2) res = (validate_doublestar(CHILD(tree, pos)) && validate_factor(CHILD(tree, pos + 1))); } return (res); } static int validate_atom(node *tree) { int pos; int nch = NCH(tree); int res = validate_ntype(tree, atom); if (res && nch < 1) res = validate_numnodes(tree, nch+1, "atom"); if (res) { switch (TYPE(CHILD(tree, 0))) { case LPAR: res = ((nch <= 3) && (validate_rparen(CHILD(tree, nch - 1)))); if (res && (nch == 3)) res = validate_testlist(CHILD(tree, 1)); break; case LSQB: if (nch == 2) res = validate_ntype(CHILD(tree, 1), RSQB); else if (nch == 3) res = (validate_listmaker(CHILD(tree, 1)) && validate_ntype(CHILD(tree, 2), RSQB)); else { res = 0; err_string("illegal list display atom"); } break; case LBRACE: res = ((nch <= 3) && validate_ntype(CHILD(tree, nch - 1), RBRACE)); if (res && (nch == 3)) res = validate_dictmaker(CHILD(tree, 1)); break; case BACKQUOTE: res = ((nch == 3) && validate_testlist(CHILD(tree, 1)) && validate_ntype(CHILD(tree, 2), BACKQUOTE)); break; case NAME: case NUMBER: res = (nch == 1); break; case STRING: for (pos = 1; res && (pos < nch); ++pos) res = validate_ntype(CHILD(tree, pos), STRING); break; default: res = 0; break; } } return (res); } /* listmaker: * test ( list_for | (',' test)* [','] ) */ static int validate_listmaker(node *tree) { int nch = NCH(tree); int ok = nch; if (nch == 0) err_string("missing child nodes of listmaker"); else ok = validate_test(CHILD(tree, 0)); /* * list_iter | (',' test)* [','] */ if (nch == 2 && TYPE(CHILD(tree, 1)) == list_for) ok = validate_list_for(CHILD(tree, 1)); else { /* (',' test)* [','] */ int i = 1; while (ok && nch - i >= 2) { ok = (validate_comma(CHILD(tree, i)) && validate_test(CHILD(tree, i+1))); i += 2; } if (ok && i == nch-1) ok = validate_comma(CHILD(tree, i)); else if (i != nch) { ok = 0; err_string("illegal trailing nodes for listmaker"); } } return ok; } /* funcdef: * 'def' NAME parameters ':' suite * */ static int validate_funcdef(node *tree) { return (validate_ntype(tree, funcdef) && validate_numnodes(tree, 5, "funcdef") && validate_name(CHILD(tree, 0), "def") && validate_ntype(CHILD(tree, 1), NAME) && validate_colon(CHILD(tree, 3)) && validate_parameters(CHILD(tree, 2)) && validate_suite(CHILD(tree, 4))); } static int validate_lambdef(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, lambdef) && ((nch == 3) || (nch == 4)) && validate_name(CHILD(tree, 0), "lambda") && validate_colon(CHILD(tree, nch - 2)) && validate_test(CHILD(tree, nch - 1))); if (res && (nch == 4)) res = validate_varargslist(CHILD(tree, 1)); else if (!res && !PyErr_Occurred()) (void) validate_numnodes(tree, 3, "lambdef"); return (res); } /* arglist: * * (argument ',')* (argument [','] | '*' test [',' '**' test] | '**' test) */ static int validate_arglist(node *tree) { int nch = NCH(tree); int i = 0; int ok = 1; if (nch <= 0) /* raise the right error from having an invalid number of children */ return validate_numnodes(tree, nch + 1, "arglist"); while (ok && nch-i >= 2) { /* skip leading (argument ',') */ ok = (validate_argument(CHILD(tree, i)) && validate_comma(CHILD(tree, i+1))); if (ok) i += 2; else PyErr_Clear(); } ok = 1; if (nch-i > 0) { /* * argument | '*' test [',' '**' test] | '**' test */ int sym = TYPE(CHILD(tree, i)); if (sym == argument) { ok = validate_argument(CHILD(tree, i)); if (ok && i+1 != nch) { err_string("illegal arglist specification" " (extra stuff on end)"); ok = 0; } } else if (sym == STAR) { ok = validate_star(CHILD(tree, i)); if (ok && (nch-i == 2)) ok = validate_test(CHILD(tree, i+1)); else if (ok && (nch-i == 5)) ok = (validate_test(CHILD(tree, i+1)) && validate_comma(CHILD(tree, i+2)) && validate_doublestar(CHILD(tree, i+3)) && validate_test(CHILD(tree, i+4))); else { err_string("illegal use of '*' in arglist"); ok = 0; } } else if (sym == DOUBLESTAR) { if (nch-i == 2) ok = (validate_doublestar(CHILD(tree, i)) && validate_test(CHILD(tree, i+1))); else { err_string("illegal use of '**' in arglist"); ok = 0; } } else { err_string("illegal arglist specification"); ok = 0; } } return (ok); } /* argument: * * [test '='] test */ static int validate_argument(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, argument) && ((nch == 1) || (nch == 3)) && validate_test(CHILD(tree, 0))); if (res && (nch == 3)) res = (validate_equal(CHILD(tree, 1)) && validate_test(CHILD(tree, 2))); return (res); } /* trailer: * * '(' [arglist] ')' | '[' subscriptlist ']' | '.' NAME */ static int validate_trailer(node *tree) { int nch = NCH(tree); int res = validate_ntype(tree, trailer) && ((nch == 2) || (nch == 3)); if (res) { switch (TYPE(CHILD(tree, 0))) { case LPAR: res = validate_rparen(CHILD(tree, nch - 1)); if (res && (nch == 3)) res = validate_arglist(CHILD(tree, 1)); break; case LSQB: res = (validate_numnodes(tree, 3, "trailer") && validate_subscriptlist(CHILD(tree, 1)) && validate_ntype(CHILD(tree, 2), RSQB)); break; case DOT: res = (validate_numnodes(tree, 2, "trailer") && validate_ntype(CHILD(tree, 1), NAME)); break; default: res = 0; break; } } else { (void) validate_numnodes(tree, 2, "trailer"); } return (res); } /* subscriptlist: * * subscript (',' subscript)* [','] */ static int validate_subscriptlist(node *tree) { return (validate_repeating_list(tree, subscriptlist, validate_subscript, "subscriptlist")); } /* subscript: * * '.' '.' '.' | test | [test] ':' [test] [sliceop] */ static int validate_subscript(node *tree) { int offset = 0; int nch = NCH(tree); int res = validate_ntype(tree, subscript) && (nch >= 1) && (nch <= 4); if (!res) { if (!PyErr_Occurred()) err_string("invalid number of arguments for subscript node"); return (0); } if (TYPE(CHILD(tree, 0)) == DOT) /* take care of ('.' '.' '.') possibility */ return (validate_numnodes(tree, 3, "subscript") && validate_dot(CHILD(tree, 0)) && validate_dot(CHILD(tree, 1)) && validate_dot(CHILD(tree, 2))); if (nch == 1) { if (TYPE(CHILD(tree, 0)) == test) res = validate_test(CHILD(tree, 0)); else res = validate_colon(CHILD(tree, 0)); return (res); } /* Must be [test] ':' [test] [sliceop], * but at least one of the optional components will * be present, but we don't know which yet. */ if ((TYPE(CHILD(tree, 0)) != COLON) || (nch == 4)) { res = validate_test(CHILD(tree, 0)); offset = 1; } if (res) res = validate_colon(CHILD(tree, offset)); if (res) { int rem = nch - ++offset; if (rem) { if (TYPE(CHILD(tree, offset)) == test) { res = validate_test(CHILD(tree, offset)); ++offset; --rem; } if (res && rem) res = validate_sliceop(CHILD(tree, offset)); } } return (res); } static int validate_sliceop(node *tree) { int nch = NCH(tree); int res = ((nch == 1) || validate_numnodes(tree, 2, "sliceop")) && validate_ntype(tree, sliceop); if (!res && !PyErr_Occurred()) { res = validate_numnodes(tree, 1, "sliceop"); } if (res) res = validate_colon(CHILD(tree, 0)); if (res && (nch == 2)) res = validate_test(CHILD(tree, 1)); return (res); } static int validate_exprlist(node *tree) { return (validate_repeating_list(tree, exprlist, validate_expr, "exprlist")); } static int validate_dictmaker(node *tree) { int nch = NCH(tree); int res = (validate_ntype(tree, dictmaker) && (nch >= 3) && validate_test(CHILD(tree, 0)) && validate_colon(CHILD(tree, 1)) && validate_test(CHILD(tree, 2))); if (res && ((nch % 4) == 0)) res = validate_comma(CHILD(tree, --nch)); else if (res) res = ((nch % 4) == 3); if (res && (nch > 3)) { int pos = 3; /* ( ',' test ':' test )* */ while (res && (pos < nch)) { res = (validate_comma(CHILD(tree, pos)) && validate_test(CHILD(tree, pos + 1)) && validate_colon(CHILD(tree, pos + 2)) && validate_test(CHILD(tree, pos + 3))); pos += 4; } } return (res); } static int validate_eval_input(node *tree) { int pos; int nch = NCH(tree); int res = (validate_ntype(tree, eval_input) && (nch >= 2) && validate_testlist(CHILD(tree, 0)) && validate_ntype(CHILD(tree, nch - 1), ENDMARKER)); for (pos = 1; res && (pos < (nch - 1)); ++pos) res = validate_ntype(CHILD(tree, pos), NEWLINE); return (res); } static int validate_node(node *tree) { int nch = 0; /* num. children on current node */ int res = 1; /* result value */ node* next = 0; /* node to process after this one */ while (res && (tree != 0)) { nch = NCH(tree); next = 0; switch (TYPE(tree)) { /* * Definition nodes. */ case funcdef: res = validate_funcdef(tree); break; case classdef: res = validate_class(tree); break; /* * "Trivial" parse tree nodes. * (Why did I call these trivial?) */ case stmt: res = validate_stmt(tree); break; case small_stmt: /* * expr_stmt | print_stmt | del_stmt | pass_stmt | flow_stmt * | import_stmt | global_stmt | exec_stmt | assert_stmt */ res = validate_small_stmt(tree); break; case flow_stmt: res = (validate_numnodes(tree, 1, "flow_stmt") && ((TYPE(CHILD(tree, 0)) == break_stmt) || (TYPE(CHILD(tree, 0)) == continue_stmt) || (TYPE(CHILD(tree, 0)) == yield_stmt) || (TYPE(CHILD(tree, 0)) == return_stmt) || (TYPE(CHILD(tree, 0)) == raise_stmt))); if (res) next = CHILD(tree, 0); else if (nch == 1) err_string("illegal flow_stmt type"); break; case yield_stmt: res = validate_yield_stmt(tree); break; /* * Compound statements. */ case simple_stmt: res = validate_simple_stmt(tree); break; case compound_stmt: res = validate_compound_stmt(tree); break; /* * Fundamental statements. */ case expr_stmt: res = validate_expr_stmt(tree); break; case print_stmt: res = validate_print_stmt(tree); break; case del_stmt: res = validate_del_stmt(tree); break; case pass_stmt: res = (validate_numnodes(tree, 1, "pass") && validate_name(CHILD(tree, 0), "pass")); break; case break_stmt: res = (validate_numnodes(tree, 1, "break") && validate_name(CHILD(tree, 0), "break")); break; case continue_stmt: res = (validate_numnodes(tree, 1, "continue") && validate_name(CHILD(tree, 0), "continue")); break; case return_stmt: res = validate_return_stmt(tree); break; case raise_stmt: res = validate_raise_stmt(tree); break; case import_stmt: res = validate_import_stmt(tree); break; case global_stmt: res = validate_global_stmt(tree); break; case exec_stmt: res = validate_exec_stmt(tree); break; case assert_stmt: res = validate_assert_stmt(tree); break; case if_stmt: res = validate_if(tree); break; case while_stmt: res = validate_while(tree); break; case for_stmt: res = validate_for(tree); break; case try_stmt: res = validate_try(tree); break; case suite: res = validate_suite(tree); break; /* * Expression nodes. */ case testlist: res = validate_testlist(tree); break; case test: res = validate_test(tree); break; case and_test: res = validate_and_test(tree); break; case not_test: res = validate_not_test(tree); break; case comparison: res = validate_comparison(tree); break; case exprlist: res = validate_exprlist(tree); break; case comp_op: res = validate_comp_op(tree); break; case expr: res = validate_expr(tree); break; case xor_expr: res = validate_xor_expr(tree); break; case and_expr: res = validate_and_expr(tree); break; case shift_expr: res = validate_shift_expr(tree); break; case arith_expr: res = validate_arith_expr(tree); break; case term: res = validate_term(tree); break; case factor: res = validate_factor(tree); break; case power: res = validate_power(tree); break; case atom: res = validate_atom(tree); break; default: /* Hopefully never reached! */ err_string("unrecognized node type"); res = 0; break; } tree = next; } return (res); } static int validate_expr_tree(node *tree) { int res = validate_eval_input(tree); if (!res && !PyErr_Occurred()) err_string("could not validate expression tuple"); return (res); } /* file_input: * (NEWLINE | stmt)* ENDMARKER */ static int validate_file_input(node *tree) { int j; int nch = NCH(tree) - 1; int res = ((nch >= 0) && validate_ntype(CHILD(tree, nch), ENDMARKER)); for (j = 0; res && (j < nch); ++j) { if (TYPE(CHILD(tree, j)) == stmt) res = validate_stmt(CHILD(tree, j)); else res = validate_newline(CHILD(tree, j)); } /* This stays in to prevent any internal failures from getting to the * user. Hopefully, this won't be needed. If a user reports getting * this, we have some debugging to do. */ if (!res && !PyErr_Occurred()) err_string("VALIDATION FAILURE: report this to the maintainer!"); return (res); } static PyObject* pickle_constructor = NULL; static PyObject* parser__pickler(PyObject *self, PyObject *args) { NOTE(ARGUNUSED(self)) PyObject *result = NULL; PyObject *st = NULL; PyObject *empty_dict = NULL; if (PyArg_ParseTuple(args, "O!:_pickler", &PyST_Type, &st)) { PyObject *newargs; PyObject *tuple; if ((empty_dict = PyDict_New()) == NULL) goto finally; if ((newargs = Py_BuildValue("Oi", st, 1)) == NULL) goto finally; tuple = parser_st2tuple((PyST_Object*)NULL, newargs, empty_dict); if (tuple != NULL) { result = Py_BuildValue("O(O)", pickle_constructor, tuple); Py_DECREF(tuple); } Py_DECREF(empty_dict); Py_DECREF(newargs); } finally: Py_XDECREF(empty_dict); return (result); } /* Functions exported by this module. Most of this should probably * be converted into an ST object with methods, but that is better * done directly in Python, allowing subclasses to be created directly. * We'd really have to write a wrapper around it all anyway to allow * inheritance. */ static PyMethodDef parser_functions[] = { {"ast2tuple", (PyCFunction)parser_st2tuple, PUBLIC_METHOD_TYPE, "Creates a tuple-tree representation of an ST."}, {"ast2list", (PyCFunction)parser_st2list, PUBLIC_METHOD_TYPE, "Creates a list-tree representation of an ST."}, {"compileast", (PyCFunction)parser_compilest, PUBLIC_METHOD_TYPE, "Compiles an ST object into a code object."}, {"compilest", (PyCFunction)parser_compilest, PUBLIC_METHOD_TYPE, "Compiles an ST object into a code object."}, {"expr", (PyCFunction)parser_expr, PUBLIC_METHOD_TYPE, "Creates an ST object from an expression."}, {"isexpr", (PyCFunction)parser_isexpr, PUBLIC_METHOD_TYPE, "Determines if an ST object was created from an expression."}, {"issuite", (PyCFunction)parser_issuite, PUBLIC_METHOD_TYPE, "Determines if an ST object was created from a suite."}, {"suite", (PyCFunction)parser_suite, PUBLIC_METHOD_TYPE, "Creates an ST object from a suite."}, {"sequence2ast", (PyCFunction)parser_tuple2st, PUBLIC_METHOD_TYPE, "Creates an ST object from a tree representation."}, {"sequence2st", (PyCFunction)parser_tuple2st, PUBLIC_METHOD_TYPE, "Creates an ST object from a tree representation."}, {"st2tuple", (PyCFunction)parser_st2tuple, PUBLIC_METHOD_TYPE, "Creates a tuple-tree representation of an ST."}, {"st2list", (PyCFunction)parser_st2list, PUBLIC_METHOD_TYPE, "Creates a list-tree representation of an ST."}, {"tuple2ast", (PyCFunction)parser_tuple2st, PUBLIC_METHOD_TYPE, "Creates an ST object from a tree representation."}, {"tuple2st", (PyCFunction)parser_tuple2st, PUBLIC_METHOD_TYPE, "Creates an ST object from a tree representation."}, /* private stuff: support pickle module */ {"_pickler", (PyCFunction)parser__pickler, METH_VARARGS, "Returns the pickle magic to allow ST objects to be pickled."}, {NULL, NULL, 0, NULL} }; DL_EXPORT(void) initparser(void); /* supply a prototype */ DL_EXPORT(void) initparser(void) { PyObject* module; PyObject* dict; PyST_Type.ob_type = &PyType_Type; module = Py_InitModule("parser", parser_functions); dict = PyModule_GetDict(module); if (parser_error == 0) parser_error = PyErr_NewException("parser.ParserError", NULL, NULL); if ((parser_error == 0) || (PyDict_SetItemString(dict, "ParserError", parser_error) != 0)) { /* caller will check PyErr_Occurred() */ return; } Py_INCREF(&PyST_Type); PyDict_SetItemString(dict, "ASTType", (PyObject*)&PyST_Type); Py_INCREF(&PyST_Type); PyDict_SetItemString(dict, "STType", (PyObject*)&PyST_Type); PyDict_SetItemString(dict, "__copyright__", PyString_FromString(parser_copyright_string)); PyDict_SetItemString(dict, "__doc__", PyString_FromString(parser_doc_string)); PyDict_SetItemString(dict, "__version__", PyString_FromString(parser_version_string)); /* register to support pickling */ module = PyImport_ImportModule("copy_reg"); if (module != NULL) { PyObject *func, *pickler; func = PyObject_GetAttrString(module, "pickle"); pickle_constructor = PyDict_GetItemString(dict, "sequence2st"); pickler = PyDict_GetItemString(dict, "_pickler"); Py_XINCREF(pickle_constructor); if ((func != NULL) && (pickle_constructor != NULL) && (pickler != NULL)) { PyObject *res; res = PyObject_CallFunction(func, "OOO", &PyST_Type, pickler, pickle_constructor); Py_XDECREF(res); } Py_XDECREF(func); Py_DECREF(module); } }