X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fexpressions%2Fparse.c;h=272090b17db88983700ae9c1fe952d75dc80abb1;hb=115c7ddfdd16e742de67b00844d0d09e11c88d5f;hp=43aa553d894456227bcb7eb7eb328627a15de455;hpb=3a61659a8fc11c51ad5af02b20f5613dcde50382;p=pspp-builds.git diff --git a/src/language/expressions/parse.c b/src/language/expressions/parse.c index 43aa553d..272090b1 100644 --- a/src/language/expressions/parse.c +++ b/src/language/expressions/parse.c @@ -1,6 +1,5 @@ /* PSPP - computes sample statistics. Copyright (C) 1997-9, 2000, 2006 Free Software Foundation, Inc. - Written by Ben Pfaff . This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as @@ -45,7 +44,7 @@ /* Declarations. */ /* Recursive descent parser in order of increasing precedence. */ -typedef union any_node *parse_recursively_func (struct expression *); +typedef union any_node *parse_recursively_func (struct lexer *, struct expression *); static parse_recursively_func parse_or, parse_and, parse_not; static parse_recursively_func parse_rel, parse_add, parse_mul; static parse_recursively_func parse_neg, parse_exp; @@ -73,7 +72,7 @@ static union any_node *allocate_unary_variable (struct expression *, Returns the new expression if successful or a null pointer otherwise. */ struct expression * -expr_parse (struct dataset *ds, enum expr_type type) +expr_parse (struct lexer *lexer, struct dataset *ds, enum expr_type type) { union any_node *n; struct expression *e; @@ -81,7 +80,7 @@ expr_parse (struct dataset *ds, enum expr_type type) assert (type == EXPR_NUMBER || type == EXPR_STRING || type == EXPR_BOOLEAN); e = expr_create (ds); - n = parse_or (e); + n = parse_or (lexer, e); if (n != NULL && type_check (e, &n, type)) return finish_expression (expr_optimize (n, e), e); else @@ -95,11 +94,12 @@ expr_parse (struct dataset *ds, enum expr_type type) expr_parse(), and sets up so that destroying POOL will free the expression as well. */ struct expression * -expr_parse_pool (struct pool *pool, +expr_parse_pool (struct lexer *lexer, + struct pool *pool, struct dataset *ds, enum expr_type type) { - struct expression *e = expr_parse (ds, type); + struct expression *e = expr_parse (lexer, ds, type); if (e != NULL) pool_add_subpool (pool, e->expr_pool); return e; @@ -114,13 +114,13 @@ expr_free (struct expression *e) } struct expression * -expr_parse_any (struct dataset *ds, bool optimize) +expr_parse_any (struct lexer *lexer, struct dataset *ds, bool optimize) { union any_node *n; struct expression *e; e = expr_create (ds); - n = parse_or (e); + n = parse_or (lexer, e); if (n == NULL) { expr_free (e); @@ -357,7 +357,7 @@ type_coercion_core (struct expression *e, msg_disable (); if ((*node)->type == OP_format && fmt_check_input (&(*node)->format.f) - && fmt_check_type_compat (&(*node)->format.f, NUMERIC)) + && fmt_check_type_compat (&(*node)->format.f, VAR_NUMERIC)) { msg_enable (); if (do_coercion) @@ -371,7 +371,7 @@ type_coercion_core (struct expression *e, msg_disable (); if ((*node)->type == OP_format && fmt_check_output (&(*node)->format.f) - && fmt_check_type_compat (&(*node)->format.f, NUMERIC)) + && fmt_check_type_compat (&(*node)->format.f, VAR_NUMERIC)) { msg_enable (); if (do_coercion) @@ -399,6 +399,15 @@ type_coercion_core (struct expression *e, } break; + case OP_var: + if ((*node)->type == OP_NUM_VAR || (*node)->type == OP_STR_VAR) + { + if (do_coercion) + *node = (*node)->composite.args[0]; + return true; + } + break; + case OP_pos_int: if ((*node)->type == OP_number && floor ((*node)->number.n) == (*node)->number.n @@ -455,6 +464,16 @@ is_coercible (atom_type required_type, union any_node *const *node) (union any_node **) node, NULL, false); } +/* Returns true if ACTUAL_TYPE is a kind of REQUIRED_TYPE, false + otherwise. */ +static bool +is_compatible (atom_type required_type, atom_type actual_type) +{ + return (required_type == actual_type + || (required_type == OP_var + && (actual_type == OP_num_var || actual_type == OP_str_var))); +} + /* How to parse an operator. */ struct operator { @@ -469,7 +488,7 @@ struct operator On failure, returns false and, if OPERATOR is non-null, sets *OPERATOR to a null pointer. */ static bool -match_operator (const struct operator ops[], size_t op_cnt, +match_operator (struct lexer *lexer, const struct operator ops[], size_t op_cnt, const struct operator **operator) { const struct operator *op; @@ -477,8 +496,8 @@ match_operator (const struct operator ops[], size_t op_cnt, for (op = ops; op < ops + op_cnt; op++) { if (op->token == '-') - lex_negative_to_dash (); - if (lex_match (op->token)) + lex_negative_to_dash (lexer); + if (lex_match (lexer, op->token)) { if (operator != NULL) *operator = op; @@ -501,7 +520,7 @@ check_operator (const struct operator *op, int arg_cnt, atom_type arg_type) assert (o->arg_cnt == arg_cnt); assert ((o->flags & OPF_ARRAY_OPERAND) == 0); for (i = 0; i < arg_cnt; i++) - assert (o->args[i] == arg_type); + assert (is_compatible (arg_type, o->args[i])); return true; } @@ -529,7 +548,7 @@ get_operand_type (const struct operator *op) is non-null, then it will be issued as a warning if more than one operator/operand pair is parsed. */ static union any_node * -parse_binary_operators (struct expression *e, union any_node *node, +parse_binary_operators (struct lexer *lexer, struct expression *e, union any_node *node, const struct operator ops[], size_t op_cnt, parse_recursively_func *parse_next_level, const char *chain_warning) @@ -542,7 +561,7 @@ parse_binary_operators (struct expression *e, union any_node *node, if (node == NULL) return node; - for (op_count = 0; match_operator (ops, op_cnt, &operator); op_count++) + for (op_count = 0; match_operator (lexer, ops, op_cnt, &operator); op_count++) { union any_node *rhs; @@ -552,7 +571,7 @@ parse_binary_operators (struct expression *e, union any_node *node, /* Parse the right-hand side and coerce to type OPERAND_TYPE. */ - rhs = parse_next_level (e); + rhs = parse_next_level (lexer, e); if (!type_coercion (e, operand_type, &rhs, operator->name)) return NULL; node = expr_allocate_binary (e, operator->type, node, rhs); @@ -565,7 +584,7 @@ parse_binary_operators (struct expression *e, union any_node *node, } static union any_node * -parse_inverting_unary_operator (struct expression *e, +parse_inverting_unary_operator (struct lexer *lexer, struct expression *e, const struct operator *op, parse_recursively_func *parse_next_level) { @@ -575,10 +594,10 @@ parse_inverting_unary_operator (struct expression *e, check_operator (op, 1, get_operand_type (op)); op_count = 0; - while (match_operator (op, 1, NULL)) + while (match_operator (lexer, op, 1, NULL)) op_count++; - node = parse_next_level (e); + node = parse_next_level (lexer, e); if (op_count > 0 && type_coercion (e, get_operand_type (op), &node, op->name) && op_count % 2 != 0) @@ -589,36 +608,37 @@ parse_inverting_unary_operator (struct expression *e, /* Parses the OR level. */ static union any_node * -parse_or (struct expression *e) +parse_or (struct lexer *lexer, struct expression *e) { static const struct operator op = { T_OR, OP_OR, "logical disjunction (\"OR\")" }; - return parse_binary_operators (e, parse_and (e), &op, 1, parse_and, NULL); + return parse_binary_operators (lexer, e, parse_and (lexer, e), &op, 1, parse_and, NULL); } /* Parses the AND level. */ static union any_node * -parse_and (struct expression *e) +parse_and (struct lexer *lexer, struct expression *e) { static const struct operator op = { T_AND, OP_AND, "logical conjunction (\"AND\")" }; - return parse_binary_operators (e, parse_not (e), &op, 1, parse_not, NULL); + return parse_binary_operators (lexer, e, parse_not (lexer, e), + &op, 1, parse_not, NULL); } /* Parses the NOT level. */ static union any_node * -parse_not (struct expression *e) +parse_not (struct lexer *lexer, struct expression *e) { static const struct operator op = { T_NOT, OP_NOT, "logical negation (\"NOT\")" }; - return parse_inverting_unary_operator (e, &op, parse_rel); + return parse_inverting_unary_operator (lexer, e, &op, parse_rel); } /* Parse relational operators. */ static union any_node * -parse_rel (struct expression *e) +parse_rel (struct lexer *lexer, struct expression *e) { const char *chain_warning = _("Chaining relational operators (e.g. \"a < b < c\") will " @@ -628,7 +648,7 @@ parse_rel (struct expression *e) "If chaining is really intended, parentheses will disable " "this warning (e.g. \"(a < b) < c\".)"); - union any_node *node = parse_add (e); + union any_node *node = parse_add (lexer, e); if (node == NULL) return NULL; @@ -649,7 +669,8 @@ parse_rel (struct expression *e) { T_NE, OP_NE, "numeric inequality (\"<>\")" }, }; - return parse_binary_operators (e, node, ops, sizeof ops / sizeof *ops, + return parse_binary_operators (lexer, e, node, ops, + sizeof ops / sizeof *ops, parse_add, chain_warning); } @@ -666,7 +687,8 @@ parse_rel (struct expression *e) { T_NE, OP_NE_STRING, "string inequality (\"<>\")" }, }; - return parse_binary_operators (e, node, ops, sizeof ops / sizeof *ops, + return parse_binary_operators (lexer, e, node, ops, + sizeof ops / sizeof *ops, parse_add, chain_warning); } @@ -677,7 +699,7 @@ parse_rel (struct expression *e) /* Parses the addition and subtraction level. */ static union any_node * -parse_add (struct expression *e) +parse_add (struct lexer *lexer, struct expression *e) { static const struct operator ops[] = { @@ -685,14 +707,14 @@ parse_add (struct expression *e) { '-', OP_SUB, "subtraction (\"-\")" }, }; - return parse_binary_operators (e, parse_mul (e), + return parse_binary_operators (lexer, e, parse_mul (lexer, e), ops, sizeof ops / sizeof *ops, parse_mul, NULL); } /* Parses the multiplication and division level. */ static union any_node * -parse_mul (struct expression *e) +parse_mul (struct lexer *lexer, struct expression *e) { static const struct operator ops[] = { @@ -700,21 +722,21 @@ parse_mul (struct expression *e) { '/', OP_DIV, "division (\"/\")" }, }; - return parse_binary_operators (e, parse_neg (e), + return parse_binary_operators (lexer, e, parse_neg (lexer, e), ops, sizeof ops / sizeof *ops, parse_neg, NULL); } /* Parses the unary minus level. */ static union any_node * -parse_neg (struct expression *e) +parse_neg (struct lexer *lexer, struct expression *e) { static const struct operator op = { '-', OP_NEG, "negation (\"-\")" }; - return parse_inverting_unary_operator (e, &op, parse_exp); + return parse_inverting_unary_operator (lexer, e, &op, parse_exp); } static union any_node * -parse_exp (struct expression *e) +parse_exp (struct lexer *lexer, struct expression *e) { static const struct operator op = { T_EXP, OP_POW, "exponentiation (\"**\")" }; @@ -725,17 +747,17 @@ parse_exp (struct expression *e) "That is, \"a**b**c\" equals \"(a**b)**c\", not as \"a**(b**c)\". " "To disable this warning, insert parentheses."); - return parse_binary_operators (e, parse_primary (e), &op, 1, + return parse_binary_operators (lexer, e, parse_primary (lexer, e), &op, 1, parse_primary, chain_warning); } /* Parses system variables. */ static union any_node * -parse_sysvar (struct expression *e) +parse_sysvar (struct lexer *lexer, struct expression *e) { - if (lex_match_id ("$CASENUM")) + if (lex_match_id (lexer, "$CASENUM")) return expr_allocate_nullary (e, OP_CASENUM); - else if (lex_match_id ("$DATE")) + else if (lex_match_id (lexer, "$DATE")) { static const char *months[12] = { @@ -753,13 +775,13 @@ parse_sysvar (struct expression *e) return expr_allocate_string_buffer (e, temp_buf, strlen (temp_buf)); } - else if (lex_match_id ("$TRUE")) + else if (lex_match_id (lexer, "$TRUE")) return expr_allocate_boolean (e, 1.0); - else if (lex_match_id ("$FALSE")) + else if (lex_match_id (lexer, "$FALSE")) return expr_allocate_boolean (e, 0.0); - else if (lex_match_id ("$SYSMIS")) + else if (lex_match_id (lexer, "$SYSMIS")) return expr_allocate_number (e, SYSMIS); - else if (lex_match_id ("$JDATE")) + else if (lex_match_id (lexer, "$JDATE")) { time_t time = time_of_last_procedure (e->ds); struct tm *tm = localtime (&time); @@ -767,7 +789,7 @@ parse_sysvar (struct expression *e) tm->tm_mon + 1, tm->tm_mday)); } - else if (lex_match_id ("$TIME")) + else if (lex_match_id (lexer, "$TIME")) { time_t time = time_of_last_procedure (e->ds); struct tm *tm = localtime (&time); @@ -779,45 +801,45 @@ parse_sysvar (struct expression *e) + tm->tm_min * 60. + tm->tm_sec); } - else if (lex_match_id ("$LENGTH")) + else if (lex_match_id (lexer, "$LENGTH")) return expr_allocate_number (e, get_viewlength ()); - else if (lex_match_id ("$WIDTH")) + else if (lex_match_id (lexer, "$WIDTH")) return expr_allocate_number (e, get_viewwidth ()); else { - msg (SE, _("Unknown system variable %s."), tokid); + msg (SE, _("Unknown system variable %s."), lex_tokid (lexer)); return NULL; } } /* Parses numbers, varnames, etc. */ static union any_node * -parse_primary (struct expression *e) +parse_primary (struct lexer *lexer, struct expression *e) { - switch (token) + switch (lex_token (lexer)) { case T_ID: - if (lex_look_ahead () == '(') + if (lex_look_ahead (lexer) == '(') { /* An identifier followed by a left parenthesis may be a vector element reference. If not, it's a function call. */ - if (e->ds != NULL && dict_lookup_vector (dataset_dict (e->ds), tokid) != NULL) - return parse_vector_element (e); + if (e->ds != NULL && dict_lookup_vector (dataset_dict (e->ds), lex_tokid (lexer)) != NULL) + return parse_vector_element (lexer, e); else - return parse_function (e); + return parse_function (lexer, e); } - else if (tokid[0] == '$') + else if (lex_tokid (lexer)[0] == '$') { /* $ at the beginning indicates a system variable. */ - return parse_sysvar (e); + return parse_sysvar (lexer, e); } - else if (e->ds != NULL && dict_lookup_var (dataset_dict (e->ds), tokid)) + else if (e->ds != NULL && dict_lookup_var (dataset_dict (e->ds), lex_tokid (lexer))) { /* It looks like a user variable. (It could be a format specifier, but we'll assume it's a variable unless proven otherwise. */ - return allocate_unary_variable (e, parse_variable (dataset_dict (e->ds))); + return allocate_unary_variable (e, parse_variable (lexer, dataset_dict (e->ds))); } else { @@ -826,14 +848,14 @@ parse_primary (struct expression *e) bool ok; msg_disable (); - ok = parse_format_specifier (&fmt); + ok = parse_format_specifier (lexer, &fmt); msg_enable (); if (ok) return expr_allocate_format (e, &fmt); /* All attempts failed. */ - msg (SE, _("Unknown identifier %s."), tokid); + msg (SE, _("Unknown identifier %s."), lex_tokid (lexer)); return NULL; } break; @@ -841,40 +863,40 @@ parse_primary (struct expression *e) case T_POS_NUM: case T_NEG_NUM: { - union any_node *node = expr_allocate_number (e, tokval); - lex_get (); + union any_node *node = expr_allocate_number (e, lex_tokval (lexer) ); + lex_get (lexer); return node; } case T_STRING: { union any_node *node = expr_allocate_string_buffer ( - e, ds_cstr (&tokstr), ds_length (&tokstr)); - lex_get (); + e, ds_cstr (lex_tokstr (lexer) ), ds_length (lex_tokstr (lexer) )); + lex_get (lexer); return node; } case '(': { union any_node *node; - lex_get (); - node = parse_or (e); - if (node != NULL && !lex_match (')')) + lex_get (lexer); + node = parse_or (lexer, e); + if (node != NULL && !lex_match (lexer, ')')) { - lex_error (_("expecting `)'")); + lex_error (lexer, _("expecting `)'")); return NULL; } return node; } default: - lex_error (_("in expression")); + lex_error (lexer, _("in expression")); return NULL; } } static union any_node * -parse_vector_element (struct expression *e) +parse_vector_element (struct lexer *lexer, struct expression *e) { const struct vector *vector; union any_node *element; @@ -882,22 +904,22 @@ parse_vector_element (struct expression *e) /* Find vector, skip token. The caller must already have verified that the current token is the name of a vector. */ - vector = dict_lookup_vector (dataset_dict (e->ds), tokid); + vector = dict_lookup_vector (dataset_dict (e->ds), lex_tokid (lexer)); assert (vector != NULL); - lex_get (); + lex_get (lexer); /* Skip left parenthesis token. The caller must have verified that the lookahead is a left parenthesis. */ - assert (token == '('); - lex_get (); + assert (lex_token (lexer) == '('); + lex_get (lexer); - element = parse_or (e); + element = parse_or (lexer, e); if (!type_coercion (e, OP_number, &element, "vector indexing") - || !lex_match (')')) + || !lex_match (lexer, ')')) return NULL; - return expr_allocate_binary (e, (vector->var[0]->type == NUMERIC + return expr_allocate_binary (e, (vector_get_type (vector) == VAR_NUMERIC ? OP_VEC_ELEM_NUM : OP_VEC_ELEM_STR), element, expr_allocate_vector (e, vector)); } @@ -940,8 +962,11 @@ word_matches (const char **test, const char **name) } static int -compare_names (const char *test, const char *name) +compare_names (const char *test, const char *name, bool abbrev_ok) { + if (!abbrev_ok) + return true; + for (;;) { if (!word_matches (&test, &name)) @@ -952,14 +977,15 @@ compare_names (const char *test, const char *name) } static int -compare_strings (const char *test, const char *name) +compare_strings (const char *test, const char *name, bool abbrev_ok UNUSED) { return strcasecmp (test, name); } static bool lookup_function_helper (const char *name, - int (*compare) (const char *test, const char *name), + int (*compare) (const char *test, const char *name, + bool abbrev_ok), const struct operation **first, const struct operation **last) { @@ -967,11 +993,12 @@ lookup_function_helper (const char *name, for (f = operations + OP_function_first; f <= operations + OP_function_last; f++) - if (!compare (name, f->name)) + if (!compare (name, f->name, !(f->flags & OPF_NO_ABBREV))) { *first = f; - while (f <= operations + OP_function_last && !compare (name, f->name)) + while (f <= operations + OP_function_last + && !compare (name, f->name, !(f->flags & OPF_NO_ABBREV))) f++; *last = f; @@ -1155,7 +1182,7 @@ no_match (const char *func_name, } static union any_node * -parse_function (struct expression *e) +parse_function (struct lexer *lexer, struct expression *e) { int min_valid; const struct operation *f, *first, *last; @@ -1168,17 +1195,17 @@ parse_function (struct expression *e) union any_node *n; - ds_init_string (&func_name, &tokstr); - min_valid = extract_min_valid (ds_cstr (&tokstr)); - if (!lookup_function (ds_cstr (&tokstr), &first, &last)) + ds_init_string (&func_name, lex_tokstr (lexer)); + min_valid = extract_min_valid (ds_cstr (lex_tokstr (lexer))); + if (!lookup_function (ds_cstr (lex_tokstr (lexer)), &first, &last)) { - msg (SE, _("No function or vector named %s."), ds_cstr (&tokstr)); + msg (SE, _("No function or vector named %s."), ds_cstr (lex_tokstr (lexer))); ds_destroy (&func_name); return NULL; } - lex_get (); - if (!lex_force_match ('(')) + lex_get (lexer); + if (!lex_force_match (lexer, '(')) { ds_destroy (&func_name); return NULL; @@ -1186,16 +1213,16 @@ parse_function (struct expression *e) args = NULL; arg_cnt = arg_cap = 0; - if (token != ')') + if (lex_token (lexer) != ')') for (;;) { - if (token == T_ID && lex_look_ahead () == 'T') + if (lex_token (lexer) == T_ID && lex_look_ahead (lexer) == 'T') { struct variable **vars; size_t var_cnt; size_t i; - if (!parse_variables (dataset_dict (e->ds), &vars, &var_cnt, PV_SINGLE)) + if (!parse_variables (lexer, dataset_dict (e->ds), &vars, &var_cnt, PV_SINGLE)) goto fail; for (i = 0; i < var_cnt; i++) add_arg (&args, &arg_cnt, &arg_cap, @@ -1204,17 +1231,17 @@ parse_function (struct expression *e) } else { - union any_node *arg = parse_or (e); + union any_node *arg = parse_or (lexer, e); if (arg == NULL) goto fail; add_arg (&args, &arg_cnt, &arg_cap, arg); } - if (lex_match (')')) + if (lex_match (lexer, ')')) break; - else if (!lex_match (',')) + else if (!lex_match (lexer, ',')) { - lex_error (_("expecting `,' or `)' invoking %s function"), + lex_error (lexer, _("expecting `,' or `)' invoking %s function"), first->name); goto fail; } @@ -1351,13 +1378,13 @@ is_valid_node (union any_node *n) assert (is_composite (n->type)); assert (c->arg_cnt >= op->arg_cnt); for (i = 0; i < op->arg_cnt; i++) - assert (expr_node_returns (c->args[i]) == op->args[i]); + assert (is_compatible (op->args[i], expr_node_returns (c->args[i]))); if (c->arg_cnt > op->arg_cnt && !is_operator (n->type)) { assert (op->flags & OPF_ARRAY_OPERAND); for (i = 0; i < c->arg_cnt; i++) - assert (operations[c->args[i]->type].returns - == op->args[op->arg_cnt - 1]); + assert (is_compatible (op->args[op->arg_cnt - 1], + expr_node_returns (c->args[i]))); } } @@ -1460,7 +1487,7 @@ union any_node * expr_allocate_variable (struct expression *e, struct variable *v) { union any_node *n = pool_alloc (e->expr_pool, sizeof n->variable); - n->type = v->type == NUMERIC ? OP_num_var : OP_str_var; + n->type = var_is_numeric (v) ? OP_num_var : OP_str_var; n->variable.v = v; return n; } @@ -1480,6 +1507,6 @@ static union any_node * allocate_unary_variable (struct expression *e, struct variable *v) { assert (v != NULL); - return expr_allocate_unary (e, v->type == NUMERIC ? OP_NUM_VAR : OP_STR_VAR, + return expr_allocate_unary (e, var_is_numeric (v) ? OP_NUM_VAR : OP_STR_VAR, expr_allocate_variable (e, v)); }