X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Flexer%2Flexer.c;h=47c019ec26dfe4208984bc23264aba69f9623aa6;hb=398309051afbe46d2407c9e93f34eaa602316d63;hp=329003406bf0281b0d01f99db81372bd4b5649b0;hpb=510cc9dd9baf3108ba55cfb7893384517c9930b4;p=pspp diff --git a/src/language/lexer/lexer.c b/src/language/lexer/lexer.c index 329003406b..47c019ec26 100644 --- a/src/language/lexer/lexer.c +++ b/src/language/lexer/lexer.c @@ -282,8 +282,8 @@ struct lexer }; static struct lex_source *lex_source__ (const struct lexer *); -static char *lex_source_get_syntax__ (const struct lex_source *, - int n0, int n1); +static char *lex_source_syntax__ (const struct lex_source *, + int ofs0, int ofs1); static const struct lex_token *lex_next__ (const struct lexer *, int n); static void lex_source_push_endcmd__ (struct lex_source *); static void lex_source_push_parse (struct lex_source *, struct lex_token *); @@ -530,6 +530,12 @@ lex_error_expecting_array (struct lexer *lexer, const char **options, size_t n) options[5], options[6], options[7]); break; + case 9: + lex_error (lexer, _("expecting %s, %s, %s, %s, %s, %s, %s, %s, or %s"), + options[0], options[1], options[2], options[3], options[4], + options[5], options[6], options[7], options[8]); + break; + default: lex_error (lexer, NULL); } @@ -966,6 +972,236 @@ lex_force_num (struct lexer *lexer) return false; } +/* If the current token is an number in the closed range [MIN,MAX], does + nothing and returns true. Otherwise, reports an error and returns false. + If NAME is nonnull, then it is used in the error message. */ +bool +lex_force_num_range_closed (struct lexer *lexer, const char *name, + double min, double max) +{ + bool is_number = lex_is_number (lexer); + bool too_small = is_number && lex_number (lexer) < min; + bool too_big = is_number && lex_number (lexer) > max; + if (is_number && !too_small && !too_big) + return true; + + if (min > max) + { + /* Weird, maybe a bug in the caller. Just report that we needed an + number. */ + if (name) + lex_error (lexer, _("Number expected for %s."), name); + else + lex_error (lexer, _("Number expected.")); + } + else if (min == max) + { + if (name) + lex_error (lexer, _("Expected %g for %s."), min, name); + else + lex_error (lexer, _("Expected %g."), min); + } + else + { + bool report_lower_bound = min > -DBL_MAX || too_small; + bool report_upper_bound = max < DBL_MAX || too_big; + + if (report_lower_bound && report_upper_bound) + { + if (name) + lex_error (lexer, + _("Expected number between %g and %g for %s."), + min, max, name); + else + lex_error (lexer, _("Expected number between %g and %g."), + min, max); + } + else if (report_lower_bound) + { + if (min == 0) + { + if (name) + lex_error (lexer, _("Expected non-negative number for %s."), + name); + else + lex_error (lexer, _("Expected non-negative number.")); + } + else + { + if (name) + lex_error (lexer, _("Expected number %g or greater for %s."), + min, name); + else + lex_error (lexer, _("Expected number %g or greater."), min); + } + } + else if (report_upper_bound) + { + if (name) + lex_error (lexer, + _("Expected number less than or equal to %g for %s."), + max, name); + else + lex_error (lexer, _("Expected number less than or equal to %g."), + max); + } + else + { + if (name) + lex_error (lexer, _("Number expected for %s."), name); + else + lex_error (lexer, _("Number expected.")); + } + } + return false; +} + +/* If the current token is an number in the half-open range [MIN,MAX), does + nothing and returns true. Otherwise, reports an error and returns false. + If NAME is nonnull, then it is used in the error message. */ +bool +lex_force_num_range_halfopen (struct lexer *lexer, const char *name, + double min, double max) +{ + bool is_number = lex_is_number (lexer); + bool too_small = is_number && lex_number (lexer) < min; + bool too_big = is_number && lex_number (lexer) >= max; + if (is_number && !too_small && !too_big) + return true; + + if (min >= max) + { + /* Weird, maybe a bug in the caller. Just report that we needed an + number. */ + if (name) + lex_error (lexer, _("Number expected for %s."), name); + else + lex_error (lexer, _("Number expected.")); + } + else + { + bool report_lower_bound = min > -DBL_MAX || too_small; + bool report_upper_bound = max < DBL_MAX || too_big; + + if (report_lower_bound && report_upper_bound) + { + if (name) + lex_error (lexer, _("Expected number in [%g,%g) for %s."), + min, max, name); + else + lex_error (lexer, _("Expected number in [%g,%g)."), + min, max); + } + else if (report_lower_bound) + { + if (min == 0) + { + if (name) + lex_error (lexer, _("Expected non-negative number for %s."), + name); + else + lex_error (lexer, _("Expected non-negative number.")); + } + else + { + if (name) + lex_error (lexer, _("Expected number %g or greater for %s."), + min, name); + else + lex_error (lexer, _("Expected number %g or greater."), min); + } + } + else if (report_upper_bound) + { + if (name) + lex_error (lexer, + _("Expected number less than %g for %s."), max, name); + else + lex_error (lexer, _("Expected number less than %g."), max); + } + else + { + if (name) + lex_error (lexer, _("Number expected for %s."), name); + else + lex_error (lexer, _("Number expected.")); + } + } + return false; +} + +/* If the current token is an number in the open range (MIN,MAX], does + nothing and returns true. Otherwise, reports an error and returns false. + If NAME is nonnull, then it is used in the error message. */ +bool +lex_force_num_range_open (struct lexer *lexer, const char *name, + double min, double max) +{ + bool is_number = lex_is_number (lexer); + bool too_small = is_number && lex_number (lexer) <= min; + bool too_big = is_number && lex_number (lexer) >= max; + if (is_number && !too_small && !too_big) + return true; + + if (min >= max) + { + /* Weird, maybe a bug in the caller. Just report that we needed an + number. */ + if (name) + lex_error (lexer, _("Number expected for %s."), name); + else + lex_error (lexer, _("Number expected.")); + } + else + { + bool report_lower_bound = min > -DBL_MAX || too_small; + bool report_upper_bound = max < DBL_MAX || too_big; + + if (report_lower_bound && report_upper_bound) + { + if (name) + lex_error (lexer, _("Expected number in (%g,%g) for %s."), + min, max, name); + else + lex_error (lexer, _("Expected number in (%g,%g)."), min, max); + } + else if (report_lower_bound) + { + if (min == 0) + { + if (name) + lex_error (lexer, _("Expected positive number for %s."), name); + else + lex_error (lexer, _("Expected positive number.")); + } + else + { + if (name) + lex_error (lexer, _("Expected number greater than %g for %s."), + min, name); + else + lex_error (lexer, _("Expected number greater than %g."), min); + } + } + else if (report_upper_bound) + { + if (name) + lex_error (lexer, _("Expected number less than %g for %s."), + max, name); + else + lex_error (lexer, _("Expected number less than %g."), max); + } + else + { + if (name) + lex_error (lexer, _("Number expected for %s."), name); + else + lex_error (lexer, _("Number expected.")); + } + } + return false; +} + /* If the current token is an identifier, does nothing and returns true. Otherwise, reports an error and returns false. */ bool @@ -1221,15 +1457,37 @@ lex_ofs_end_point (const struct lexer *lexer, int ofs) /* Returns the text of the syntax in tokens N0 ahead of the current one, through N1 ahead of the current one, inclusive. (For example, if N0 and N1 - are both zero, this requests the syntax for the current token.) The caller - must eventually free the returned string (with free()). The syntax is - encoded in UTF-8 and in the original form supplied to the lexer so that, for - example, it may include comments, spaces, and new-lines if it spans multiple - tokens. Macro expansion, however, has already been performed. */ + are both zero, this requests the syntax for the current token.) + + The caller must eventually free the returned string (with free()). The + syntax is encoded in UTF-8 and in the original form supplied to the lexer so + that, for example, it may include comments, spaces, and new-lines if it + spans multiple tokens. Macro expansion, however, has already been + performed. */ char * lex_next_representation (const struct lexer *lexer, int n0, int n1) { - return lex_source_get_syntax__ (lex_source__ (lexer), n0, n1); + const struct lex_source *src = lex_source__ (lexer); + return (src + ? lex_source_syntax__ (src, n0 + src->parse_ofs, n1 + src->parse_ofs) + : xstrdup ("")); +} + + +/* Returns the text of the syntax in tokens with offsets OFS0 to OFS1, + inclusive. (For example, if OFS0 and OFS1 are both zero, this requests the + syntax for the first token in the current command.) + + The caller must eventually free the returned string (with free()). The + syntax is encoded in UTF-8 and in the original form supplied to the lexer so + that, for example, it may include comments, spaces, and new-lines if it + spans multiple tokens. Macro expansion, however, has already been + performed. */ +char * +lex_ofs_representation (const struct lexer *lexer, int ofs0, int ofs1) +{ + const struct lex_source *src = lex_source__ (lexer); + return src ? lex_source_syntax__ (src, ofs0, ofs1) : xstrdup (""); } /* Returns true if the token N ahead of the current one was produced by macro @@ -1557,32 +1815,33 @@ lex_source__ (const struct lexer *lexer) : ll_data (ll_head (&lexer->sources), struct lex_source, ll)); } -/* Returns the text of the syntax in SRC for tokens N0 ahead of the current - one, through N1 ahead of the current one, inclusive. (For example, if N0 - and N1 are both zero, this requests the syntax for the current token.) The - caller must eventually free the returned string (with free()). The syntax - is encoded in UTF-8 and in the original form supplied to the lexer so that, - for example, it may include comments, spaces, and new-lines if it spans - multiple tokens. Macro expansion, however, has already been performed. */ +/* Returns the text of the syntax in SRC for tokens with offsets OFS0 through + OFS1 in the current command, inclusive. (For example, if OFS0 and OFS1 are + both zero, this requests the syntax for the first token in the current + command.) The caller must eventually free the returned string (with + free()). The syntax is encoded in UTF-8 and in the original form supplied + to the lexer so that, for example, it may include comments, spaces, and + new-lines if it spans multiple tokens. Macro expansion, however, has + already been performed. */ static char * -lex_source_get_syntax__ (const struct lex_source *src, int n0, int n1) +lex_source_syntax__ (const struct lex_source *src, int ofs0, int ofs1) { struct string s = DS_EMPTY_INITIALIZER; - for (size_t i = n0; i <= n1; ) + for (size_t i = ofs0; i <= ofs1; ) { /* Find [I,J) as the longest sequence of tokens not produced by macro expansion, or otherwise the longest sequence expanded from a single macro call. */ - const struct lex_token *first = lex_source_next__ (src, i); + const struct lex_token *first = lex_source_ofs__ (src, i); size_t j; - for (j = i + 1; j <= n1; j++) + for (j = i + 1; j <= ofs1; j++) { - const struct lex_token *cur = lex_source_next__ (src, j); + const struct lex_token *cur = lex_source_ofs__ (src, j); if ((first->macro_rep != NULL) != (cur->macro_rep != NULL) || first->macro_rep != cur->macro_rep) break; } - const struct lex_token *last = lex_source_next__ (src, j - 1); + const struct lex_token *last = lex_source_ofs__ (src, j - 1); /* Now add the syntax for this sequence of tokens to SRC. */ if (!ds_is_empty (&s)) @@ -1653,7 +1912,8 @@ lex_source_error_valist (struct lex_source *src, int n0, int n1, else { /* Get the syntax that caused the error. */ - char *raw_syntax = lex_source_get_syntax__ (src, n0, n1); + char *raw_syntax = lex_source_syntax__ (src, n0 + src->parse_ofs, + n1 + src->parse_ofs); char syntax[64]; str_ellipsize (ss_cstr (raw_syntax), syntax, sizeof syntax); free (raw_syntax);