X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/c547693afdacf4b865027d3bb8c53f601eadb5ae..671850a1c3e1aaac93a89fabc6d810be6a4a77fe:/src/scan-code.l diff --git a/src/scan-code.l b/src/scan-code.l index b8c0c1b0..308d1d0d 100644 --- a/src/scan-code.l +++ b/src/scan-code.l @@ -1,6 +1,6 @@ /* Bison Action Scanner -*- C -*- - Copyright (C) 2006-2012 Free Software Foundation, Inc. + Copyright (C) 2006-2015 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -47,6 +47,9 @@ YY_DECL; #define YY_USER_ACTION location_compute (loc, &loc->end, yytext, yyleng); +static char *fetch_type_name (char *cp, char const **type_name, + location dollar_loc); + static void handle_action_dollar (symbol_list *rule, char *cp, location dollar_loc); static void handle_action_at (symbol_list *rule, char *cp, location at_loc); @@ -75,8 +78,9 @@ static bool untyped_var_seen; /* POSIX says that a tag must be both an id and a C union member, but historically almost any character is allowed in a tag. We disallow - NUL and newline, as this simplifies our implementation. */ -tag [^\0\n>]+ + NUL and newline, as this simplifies our implementation. We allow + "->" as a means to dereference a pointer. */ +tag ([^\0\n>]|->)+ /* Zero or more instances of backslash-newline. Following GCC, allow white space between the backslash and the newline. */ @@ -92,22 +96,6 @@ ref -?[0-9]+|{id}|"["{id}"]"|"$" %% %{ - /* Nesting level of the current code in braces. */ - int braces_level = 0; - - /* Whether a semicolon is probably needed. - The heuristic is that a semicolon is not needed after '{', '}', ';', - or a C preprocessor directive, and that whitespaces and comments - do not affect this flag. - Note that '{' does not need a semicolon because of '{}'. - A semicolon may be needed before a cpp direcive, but don't bother. */ - bool need_semicolon = false; - - /* Whether in a C preprocessor directive. Don't use a start condition - for this because, at the end of strings and comments, we still need - to know whether we're in a directive. */ - bool in_cpp = false; - /* This scanner is special: it is invoked only once, henceforth is expected to return only once. This initialization is therefore done once per action to translate. */ @@ -158,114 +146,43 @@ ref -?[0-9]+|{id}|"["{id}"]"|"$" } -{ - "'" { - STRING_GROW; - BEGIN SC_CHARACTER; - need_semicolon = true; - } - "\"" { - STRING_GROW; - BEGIN SC_STRING; - need_semicolon = true; - } - "/"{splice}"*" { - STRING_GROW; - BEGIN SC_COMMENT; - } - "/"{splice}"/" { - STRING_GROW; - BEGIN SC_LINE_COMMENT; + +{ + "'" STRING_GROW; BEGIN SC_CHARACTER; + "\"" STRING_GROW; BEGIN SC_STRING; + "/"{splice}"*" STRING_GROW; BEGIN SC_COMMENT; + "/"{splice}"/" STRING_GROW; BEGIN SC_LINE_COMMENT; + + [$@] { + complain (loc, Wother, _("stray '%s'"), yytext); + obstack_escape (&obstack_for_string, yytext); } } { "$"("<"{tag}">")?{ref} { - ref_tail_fields = 0; + ref_tail_fields = NULL; handle_action_dollar (self->rule, yytext, *loc); - if (ref_tail_fields) { + if (ref_tail_fields) obstack_sgrow (&obstack_for_string, ref_tail_fields); - } - need_semicolon = true; } "@"{ref} { - ref_tail_fields = 0; + ref_tail_fields = NULL; handle_action_at (self->rule, yytext, *loc); - if (ref_tail_fields) { + if (ref_tail_fields) obstack_sgrow (&obstack_for_string, ref_tail_fields); - } - need_semicolon = true; - } - "$" { - warn_at (*loc, _("stray '$'")); - obstack_sgrow (&obstack_for_string, "$]["); - need_semicolon = true; - } - "@" { - warn_at (*loc, _("stray '@'")); - obstack_sgrow (&obstack_for_string, "@@"); - need_semicolon = true; - } - "[" { - obstack_sgrow (&obstack_for_string, "@{"); - need_semicolon = true; } - "]" { - obstack_sgrow (&obstack_for_string, "@}"); - need_semicolon = true; - } - - ";" STRING_GROW; need_semicolon = false; - "{" STRING_GROW; ++braces_level; need_semicolon = false; - "}" { - bool outer_brace = --braces_level == 0; - - /* As an undocumented Bison extension, append ';' before the last - brace in braced code, so that the user code can omit trailing - ';'. But do not append ';' if emulating Yacc, since Yacc does - not append one. */ - if (outer_brace && !yacc_flag && language_prio == default_prio - && skeleton_prio == default_prio && need_semicolon && ! in_cpp) - { - warn_at (*loc, _("a ';' might be needed at the end of action code")); - warn_at (*loc, _("future versions of Bison will not add the ';'")); - obstack_1grow (&obstack_for_string, ';'); - } - - STRING_GROW; - need_semicolon = false; - } - - /* Preprocessing directives should only be recognized at the beginning - of lines, allowing whitespace including comments, but in C/C++, - '#' can only be the start of preprocessor directives or within - '#define' directives anyway, so don't bother with begin of line. */ - "#" STRING_GROW; in_cpp = true; - - {splice} STRING_GROW; - [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; - [ \t\f] STRING_GROW; - - /* YYFAIL is undocumented and was formally deprecated in Bison - 2.4.2. */ - YYFAIL { - STRING_GROW; need_semicolon = true; - warn_at (*loc, _("use of YYFAIL, which is deprecated and will be" - " removed")); - } - - /* The sole purpose of this is to make sure identifiers that merely - contain YYFAIL don't produce the above warning. */ - [A-Za-z_][0-9A-Za-z_]* STRING_GROW; need_semicolon = true; - - . STRING_GROW; need_semicolon = true; } { - "$$" { - obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar["); + "$"("<"{tag}">")?"$" { + const char *type_name = NULL; + fetch_type_name (yytext + 1, &type_name, *loc)[-1] = 0; + obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar("); + obstack_quote (&obstack_for_string, type_name); + obstack_sgrow (&obstack_for_string, ")["); self->is_value_used = true; } "@$" { @@ -275,29 +192,17 @@ ref -?[0-9]+|{id}|"["{id}"]"|"$" } - /*-----------------------------------------. - | Escape M4 quoting characters in C code. | - `-----------------------------------------*/ - <*> { - \$ obstack_sgrow (&obstack_for_string, "$]["); - \@ obstack_sgrow (&obstack_for_string, "@@"); - \[ obstack_sgrow (&obstack_for_string, "@{"); - \] obstack_sgrow (&obstack_for_string, "@}"); -} - - /*-----------------------------------------------------. - | By default, grow the string obstack with the input. | - `-----------------------------------------------------*/ + /* Escape M4 quoting characters in C code. */ + [$@\[\]] obstack_escape (&obstack_for_string, yytext); -<*>.|\n STRING_GROW; + /* By default, grow the string obstack with the input. */ + .|\n STRING_GROW; /* End of processing. */ -<*><> { - STRING_FINISH; - return last_string; - } + <> STRING_FINISH; return last_string; +} %% @@ -346,7 +251,7 @@ typedef struct not visible from current midrule. */ #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2) -static variant *variant_table = 0; +static variant *variant_table = NULL; static unsigned variant_table_size = 0; static unsigned variant_count = 0; @@ -368,7 +273,7 @@ static void variant_table_free (void) { free (variant_table); - variant_table = 0; + variant_table = NULL; variant_table_size = variant_count = 0; } @@ -422,77 +327,83 @@ get_at_spec(unsigned symbol_index) } static void -show_sub_messages (const char* cp, bool explicit_bracketing, - int midrule_rhs_index, char dollar_or_at, - bool is_warning, unsigned indent) +show_sub_message (warnings warning, + const char* cp, bool explicit_bracketing, + int midrule_rhs_index, char dollar_or_at, + unsigned indent, const variant *var) { - unsigned i; + const char *at_spec = get_at_spec (var->symbol_index); - for (i = 0; i < variant_count; ++i) + if (var->err == 0) + complain_indent (&var->loc, warning, &indent, + _("refers to: %c%s at %s"), dollar_or_at, + var->id, at_spec); + else { - const variant *var = &variant_table[i]; - const char *at_spec = get_at_spec (var->symbol_index); + static struct obstack msg_buf; + const char *tail = explicit_bracketing ? "" : cp + strlen (var->id); + const char *id; + location id_loc; - if (var->err == 0) + if (var->hidden_by) { - if (is_warning) - warn_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), - dollar_or_at, var->id, at_spec); - else - complain_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), - dollar_or_at, var->id, at_spec); + id = var->hidden_by->id; + id_loc = var->hidden_by->loc; } else { - static struct obstack msg_buf; - const char *tail = explicit_bracketing ? "" : - cp + strlen (var->id); - const char *id = var->hidden_by ? var->hidden_by->id : - var->id; - location id_loc = var->hidden_by ? var->hidden_by->loc : - var->loc; - - /* Create the explanation message. */ - obstack_init (&msg_buf); - - obstack_fgrow1 (&msg_buf, _("possibly meant: %c"), dollar_or_at); - if (contains_dot_or_dash (id)) - obstack_fgrow1 (&msg_buf, "[%s]", id); - else - obstack_sgrow (&msg_buf, id); - obstack_sgrow (&msg_buf, tail); - - if (var->err & VARIANT_HIDDEN) - { - obstack_fgrow1 (&msg_buf, _(", hiding %c"), dollar_or_at); - if (contains_dot_or_dash (var->id)) - obstack_fgrow1 (&msg_buf, "[%s]", var->id); - else - obstack_sgrow (&msg_buf, var->id); - obstack_sgrow (&msg_buf, tail); - } + id = var->id; + id_loc = var->loc; + } - obstack_fgrow1 (&msg_buf, _(" at %s"), at_spec); + /* Create the explanation message. */ + obstack_init (&msg_buf); - if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE) - { - const char *format = - _(", cannot be accessed from mid-rule action at $%d"); - obstack_fgrow1 (&msg_buf, format, midrule_rhs_index); - } + obstack_printf (&msg_buf, _("possibly meant: %c"), dollar_or_at); + if (contains_dot_or_dash (id)) + obstack_printf (&msg_buf, "[%s]", id); + else + obstack_sgrow (&msg_buf, id); + obstack_sgrow (&msg_buf, tail); - obstack_1grow (&msg_buf, '\0'); - if (is_warning) - warn_at_indent (id_loc, &indent, "%s", - (char *) obstack_finish (&msg_buf)); + if (var->err & VARIANT_HIDDEN) + { + obstack_printf (&msg_buf, _(", hiding %c"), dollar_or_at); + if (contains_dot_or_dash (var->id)) + obstack_printf (&msg_buf, "[%s]", var->id); else - complain_at_indent (id_loc, &indent, "%s", - (char *) obstack_finish (&msg_buf)); - obstack_free (&msg_buf, 0); + obstack_sgrow (&msg_buf, var->id); + obstack_sgrow (&msg_buf, tail); } + + obstack_printf (&msg_buf, _(" at %s"), at_spec); + + if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE) + obstack_printf (&msg_buf, + _(", cannot be accessed from mid-rule action at $%d"), + midrule_rhs_index); + + complain_indent (&id_loc, warning, &indent, "%s", + obstack_finish0 (&msg_buf)); + obstack_free (&msg_buf, 0); } } +static void +show_sub_messages (warnings warning, + const char* cp, bool explicit_bracketing, + int midrule_rhs_index, char dollar_or_at, + unsigned indent) +{ + unsigned i; + + for (i = 0; i < variant_count; ++i) + show_sub_message (warning | silent, + cp, explicit_bracketing, + midrule_rhs_index, dollar_or_at, + indent, &variant_table[i]); +} + /* Returned from "parse_ref" when the reference is inappropriate. */ #define INVALID_REF (INT_MIN) @@ -501,9 +412,6 @@ show_sub_messages (const char* cp, bool explicit_bracketing, points to LHS ($$) of the current rule or midrule. */ #define LHS_REF (INT_MIN + 1) -/* Sub-messages indent. */ -#define SUB_INDENT (4) - /* Parse named or positional reference. In case of positional references, can return negative values for $-n "deep" stack accesses. */ @@ -529,8 +437,8 @@ parse_ref (char *cp, symbol_list *rule, int rule_length, return num; else { - complain_at (text_loc, _("integer out of range: %s"), - quote (text)); + complain (&text_loc, complaint, _("integer out of range: %s"), + quote (text)); return INVALID_REF; } } @@ -618,37 +526,33 @@ parse_ref (char *cp, symbol_list *rule, int rule_length, cp_end - cp : ref_tail_fields - cp; unsigned indent = 0; - complain_at_indent (text_loc, &indent, _("invalid reference: %s"), - quote (text)); + complain_indent (&text_loc, complaint, &indent, + _("invalid reference: %s"), quote (text)); indent += SUB_INDENT; if (len == 0) { location sym_loc = text_loc; sym_loc.start.column += 1; sym_loc.end = sym_loc.start; - const char *format = - _("syntax error after '%c', expecting integer, letter," - " '_', '[', or '$'"); - complain_at_indent (sym_loc, &indent, format, dollar_or_at); + complain_indent (&sym_loc, complaint, &indent, + _("syntax error after '%c', expecting integer, " + "letter, '_', '[', or '$'"), + dollar_or_at); } else if (midrule_rhs_index) - { - const char *format = - _("symbol not found in production before $%d: %.*s"); - complain_at_indent (rule->location, &indent, format, - midrule_rhs_index, len, cp); - } + complain_indent (&rule->location, complaint, &indent, + _("symbol not found in production before $%d: " + "%.*s"), + midrule_rhs_index, len, cp); else - { - const char *format = - _("symbol not found in production: %.*s"); - complain_at_indent (rule->location, &indent, format, - len, cp); - } + complain_indent (&rule->location, complaint, &indent, + _("symbol not found in production: %.*s"), + len, cp); if (variant_count > 0) - show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, - dollar_or_at, false, indent); + show_sub_messages (complaint, + cp, explicit_bracketing, midrule_rhs_index, + dollar_or_at, indent); return INVALID_REF; } case 1: @@ -656,10 +560,11 @@ parse_ref (char *cp, symbol_list *rule, int rule_length, unsigned indent = 0; if (variant_count > 1) { - warn_at_indent (text_loc, &indent, _("misleading reference: %s"), - quote (text)); - show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, - dollar_or_at, true, indent + SUB_INDENT); + complain_indent (&text_loc, Wother, &indent, + _("misleading reference: %s"), quote (text)); + show_sub_messages (Wother, + cp, explicit_bracketing, midrule_rhs_index, + dollar_or_at, indent + SUB_INDENT); } { unsigned symbol_index = @@ -671,16 +576,14 @@ parse_ref (char *cp, symbol_list *rule, int rule_length, default: { unsigned indent = 0; - complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"), - quote (text)); - show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, - dollar_or_at, false, indent + SUB_INDENT); + complain_indent (&text_loc, complaint, &indent, + _("ambiguous reference: %s"), quote (text)); + show_sub_messages (complaint, + cp, explicit_bracketing, midrule_rhs_index, + dollar_or_at, indent + SUB_INDENT); return INVALID_REF; } } - - /* Not reachable. */ - return INVALID_REF; } /* Keeps track of the maximum number of semantic values to the left of @@ -689,6 +592,33 @@ parse_ref (char *cp, symbol_list *rule, int rule_length, int max_left_semantic_context = 0; +/* If CP points to a typename (i.e., <.*?>), set TYPE_NAME to its + beginning (i.e., after the opening "<", and return the pointer + immediately after it. */ + +static +char * +fetch_type_name (char *cp, char const **type_name, + location dollar_loc) +{ + if (*cp == '<') + { + *type_name = ++cp; + /* Series of non-'>' or "->". */ + while (*cp != '>' || cp[-1] == '-') + ++cp; + + /* The '>' symbol will be later replaced by '\0'. Original + 'text' is needed for error messages. */ + ++cp; + if (untyped_var_seen) + complain (&dollar_loc, complaint, + _("explicit type given in untyped grammar")); + tag_seen = true; + } + return cp; +} + /*------------------------------------------------------------------. | TEXT is pointing to a wannabee semantic value (i.e., a '$'). | | | @@ -702,7 +632,6 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) { char const *type_name = NULL; char *cp = text + 1; - char *gt_ptr = 0; symbol_list *effective_rule; int effective_rule_length; int n; @@ -719,26 +648,14 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) } /* Get the type name if explicit. */ - if (*cp == '<') - { - type_name = ++cp; - while (*cp != '>') - ++cp; - - /* The '>' symbol will be later replaced by '\0'. Original - 'text' is needed for error messages. */ - gt_ptr = cp; - ++cp; - if (untyped_var_seen) - complain_at (dollar_loc, _("explicit type given in untyped grammar")); - tag_seen = true; - } + cp = fetch_type_name (cp, &type_name, dollar_loc); n = parse_ref (cp, effective_rule, effective_rule_length, rule->midrule_parent_rhs_index, text, dollar_loc, '$'); - if (gt_ptr) - *gt_ptr = '\0'; + /* End type_name. */ + if (type_name) + cp[-1] = '\0'; switch (n) { @@ -754,22 +671,23 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) if (union_seen | tag_seen) { if (rule->midrule_parent_rule) - complain_at (dollar_loc, + complain (&dollar_loc, complaint, _("$$ for the midrule at $%d of %s" " has no declared type"), rule->midrule_parent_rhs_index, quote (effective_rule->content.sym->tag)); else - complain_at (dollar_loc, _("$$ of %s has no declared type"), - quote (rule->content.sym->tag)); + complain (&dollar_loc, complaint, + _("$$ of %s has no declared type"), + quote (rule->content.sym->tag)); } else untyped_var_seen = true; - type_name = ""; } - obstack_fgrow1 (&obstack_for_string, - "]b4_lhs_value([%s])[", type_name); + obstack_sgrow (&obstack_for_string, "]b4_lhs_value("); + obstack_quote (&obstack_for_string, type_name); + obstack_sgrow (&obstack_for_string, ")["); rule->action_props.is_value_used = true; break; @@ -782,17 +700,18 @@ handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) if (!type_name) { if (union_seen | tag_seen) - complain_at (dollar_loc, _("$%s of %s has no declared type"), - cp, quote (effective_rule->content.sym->tag)); + complain (&dollar_loc, complaint, + _("$%s of %s has no declared type"), cp, + quote (effective_rule->content.sym->tag)); else untyped_var_seen = true; - type_name = ""; } - obstack_fgrow3 (&obstack_for_string, - "]b4_rhs_value(%d, %d, [%s])[", - effective_rule_length, n, type_name); - if (n > 0) + obstack_printf (&obstack_for_string, + "]b4_rhs_value(%d, %d, ", effective_rule_length, n); + obstack_quote (&obstack_for_string, type_name); + obstack_sgrow (&obstack_for_string, ")["); + if (0 < n) symbol_list_n_get (effective_rule, n)->action_props.is_value_used = true; break; @@ -838,7 +757,7 @@ handle_action_at (symbol_list *rule, char *text, location at_loc) break; default: - obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location(%d, %d)[", + obstack_printf (&obstack_for_string, "]b4_rhs_location(%d, %d)[", effective_rule_length, n); break; } @@ -882,30 +801,26 @@ code_props_none_init (code_props *self) *self = code_props_none; } -code_props const code_props_none = CODE_PROPS_NONE_INIT; +code_props code_props_none = CODE_PROPS_NONE_INIT; void code_props_plain_init (code_props *self, char const *code, location code_loc) { + code_props_none_init (self); self->kind = CODE_PROPS_PLAIN; self->code = code; self->location = code_loc; - self->is_value_used = false; - self->rule = NULL; - self->named_ref = NULL; } void code_props_symbol_action_init (code_props *self, char const *code, location code_loc) { + code_props_none_init (self); self->kind = CODE_PROPS_SYMBOL_ACTION; self->code = code; self->location = code_loc; - self->is_value_used = false; - self->rule = NULL; - self->named_ref = NULL; } void @@ -913,10 +828,10 @@ code_props_rule_action_init (code_props *self, char const *code, location code_loc, symbol_list *rule, named_ref *name, bool is_predicate) { + code_props_none_init (self); self->kind = CODE_PROPS_RULE_ACTION; self->code = code; self->location = code_loc; - self->is_value_used = false; self->rule = rule; self->named_ref = name; self->is_predicate = is_predicate;