1 /* Various declarations for language-independent pretty-print subroutines.
2 Copyright (C) 2003, 2004, 2005, 2007, 2008 Free Software Foundation, Inc.
3 Contributed by Gabriel Dos Reis <gdr@integrable-solutions.net>
5 This file is part of GCC.
7 GCC is free software; you can redistribute it and/or modify it under
8 the terms of the GNU General Public License as published by the Free
9 Software Foundation; either version 3, or (at your option) any later
12 GCC is distributed in the hope that it will be useful, but WITHOUT ANY
13 WARRANTY; without even the implied warranty of MERCHANTABILITY or
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
17 You should have received a copy of the GNU General Public License
18 along with GCC; see the file COPYING3. If not see
19 <http://www.gnu.org/licenses/>. */
22 #undef FLOAT /* This is for hpux. They should change hpux. */
23 #undef FFS /* Some systems define this in param.h. */
25 #include "coretypes.h"
27 #include "pretty-print.h"
35 #define obstack_chunk_alloc xmalloc
36 #define obstack_chunk_free free
38 /* A pointer to the formatted diagnostic message. */
39 #define pp_formatted_text_data(PP) \
40 ((const char *) obstack_base (pp_base (PP)->buffer->obstack))
42 /* Format an integer given by va_arg (ARG, type-specifier T) where
43 type-specifier is a precision modifier as indicated by PREC. F is
44 a string used to construct the appropriate format-specifier. */
45 #define pp_integer_with_precision(PP, ARG, PREC, T, F) \
50 pp_scalar (PP, "%" F, va_arg (ARG, T)); \
54 pp_scalar (PP, "%l" F, va_arg (ARG, long T)); \
58 pp_scalar (PP, "%" HOST_LONG_LONG_FORMAT F, va_arg (ARG, long long T)); \
67 /* Subroutine of pp_set_maximum_length. Set up PRETTY-PRINTER's
68 internal maximum characters per line. */
70 pp_set_real_maximum_length (pretty_printer
*pp
)
72 /* If we're told not to wrap lines then do the obvious thing. In case
73 we'll emit prefix only once per message, it is appropriate
74 not to increase unnecessarily the line-length cut-off. */
75 if (!pp_is_wrapping_line (pp
)
76 || pp_prefixing_rule (pp
) == DIAGNOSTICS_SHOW_PREFIX_ONCE
77 || pp_prefixing_rule (pp
) == DIAGNOSTICS_SHOW_PREFIX_NEVER
)
78 pp
->maximum_length
= pp_line_cutoff (pp
);
81 int prefix_length
= pp
->prefix
? strlen (pp
->prefix
) : 0;
82 /* If the prefix is ridiculously too long, output at least
84 if (pp_line_cutoff (pp
) - prefix_length
< 32)
85 pp
->maximum_length
= pp_line_cutoff (pp
) + 32;
87 pp
->maximum_length
= pp_line_cutoff (pp
);
91 /* Clear PRETTY-PRINTER's output state. */
93 pp_clear_state (pretty_printer
*pp
)
95 pp
->emitted_prefix
= false;
96 pp_indentation (pp
) = 0;
99 /* Flush the formatted text of PRETTY-PRINTER onto the attached stream. */
101 pp_write_text_to_stream (pretty_printer
*pp
)
103 const char *text
= pp_formatted_text (pp
);
104 fputs (text
, pp
->buffer
->stream
);
105 pp_clear_output_area (pp
);
108 /* Wrap a text delimited by START and END into PRETTY-PRINTER. */
110 pp_wrap_text (pretty_printer
*pp
, const char *start
, const char *end
)
112 bool wrapping_line
= pp_is_wrapping_line (pp
);
116 /* Dump anything bordered by whitespaces. */
118 const char *p
= start
;
119 while (p
!= end
&& !ISBLANK (*p
) && *p
!= '\n')
122 && p
- start
>= pp_remaining_character_count_for_line (pp
))
124 pp_append_text (pp
, start
, p
);
128 if (start
!= end
&& ISBLANK (*start
))
133 if (start
!= end
&& *start
== '\n')
141 /* Same as pp_wrap_text but wrap text only when in line-wrapping mode. */
143 pp_maybe_wrap_text (pretty_printer
*pp
, const char *start
, const char *end
)
145 if (pp_is_wrapping_line (pp
))
146 pp_wrap_text (pp
, start
, end
);
148 pp_append_text (pp
, start
, end
);
151 /* Append to the output area of PRETTY-PRINTER a string specified by its
152 STARTing character and LENGTH. */
154 pp_append_r (pretty_printer
*pp
, const char *start
, int length
)
156 obstack_grow (pp
->buffer
->obstack
, start
, length
);
157 pp
->buffer
->line_length
+= length
;
160 /* Insert enough spaces into the output area of PRETTY-PRINTER to bring
161 the column position to the current indentation level, assuming that a
162 newline has just been written to the buffer. */
164 pp_base_indent (pretty_printer
*pp
)
166 int n
= pp_indentation (pp
);
169 for (i
= 0; i
< n
; ++i
)
173 /* The following format specifiers are recognized as being client independent:
174 %d, %i: (signed) integer in base ten.
175 %u: unsigned integer in base ten.
176 %o: unsigned integer in base eight.
177 %x: unsigned integer in base sixteen.
178 %ld, %li, %lo, %lu, %lx: long versions of the above.
179 %lld, %lli, %llo, %llu, %llx: long long versions.
180 %wd, %wi, %wo, %wu, %wx: HOST_WIDE_INT versions.
184 %m: strerror(text->err_no) - does not consume a value from args_ptr.
188 %': apostrophe (should only be used in untranslated messages;
189 translations should use appropriate punctuation directly).
190 %.*s: a substring the length of which is specified by an argument
192 %Ns: likewise, but length specified as constant in the format string.
194 %J: a decl tree, from which DECL_SOURCE_LOCATION will be recorded.
195 %K: a statement, from which EXPR_LOCATION and TREE_BLOCK will be recorded.
196 Flag 'q': quote formatted text (must come immediately after '%').
198 Arguments can be used sequentially, or through %N$ resp. *N$
199 notation Nth argument after the format string. If %N$ / *N$
200 notation is used, it must be used for all arguments, except %m, %%,
201 %<, %> and %', which may not have a number, as they do not consume
202 an argument. When %M$.*N$s is used, M must be N + 1. (This may
203 also be written %M$.*s, provided N is not otherwise used.) The
204 format string must have conversion specifiers with argument numbers
205 1 up to highest argument; each argument may only be used once.
206 A format string can have at most 30 arguments. */
208 /* Formatting phases 1 and 2: render TEXT->format_spec plus
209 TEXT->args_ptr into a series of chunks in PP->buffer->args[].
210 Phase 3 is in pp_base_format_text. */
213 pp_base_format (pretty_printer
*pp
, text_info
*text
)
215 output_buffer
*buffer
= pp
->buffer
;
218 struct chunk_info
*new_chunk_array
;
220 unsigned int curarg
= 0, chunk
= 0, argno
;
221 pp_wrapping_mode_t old_wrapping_mode
;
222 bool any_unnumbered
= false, any_numbered
= false;
223 const char **formatters
[PP_NL_ARGMAX
];
225 /* Allocate a new chunk structure. */
226 new_chunk_array
= XOBNEW (&buffer
->chunk_obstack
, struct chunk_info
);
227 new_chunk_array
->prev
= buffer
->cur_chunk_array
;
228 buffer
->cur_chunk_array
= new_chunk_array
;
229 args
= new_chunk_array
->args
;
231 /* Formatting phase 1: split up TEXT->format_spec into chunks in
232 PP->buffer->args[]. Even-numbered chunks are to be output
233 verbatim, odd-numbered chunks are format specifiers.
234 %m, %%, %<, %>, and %' are replaced with the appropriate text at
237 memset (formatters
, 0, sizeof formatters
);
239 for (p
= text
->format_spec
; *p
; )
241 while (*p
!= '\0' && *p
!= '%')
243 obstack_1grow (&buffer
->chunk_obstack
, *p
);
256 obstack_1grow (&buffer
->chunk_obstack
, '%');
261 obstack_grow (&buffer
->chunk_obstack
,
262 open_quote
, strlen (open_quote
));
268 obstack_grow (&buffer
->chunk_obstack
,
269 close_quote
, strlen (close_quote
));
275 const char *errstr
= xstrerror (text
->err_no
);
276 obstack_grow (&buffer
->chunk_obstack
, errstr
, strlen (errstr
));
282 /* Handled in phase 2. Terminate the plain chunk here. */
283 obstack_1grow (&buffer
->chunk_obstack
, '\0');
284 gcc_assert (chunk
< PP_NL_ARGMAX
* 2);
285 args
[chunk
++] = XOBFINISH (&buffer
->chunk_obstack
, const char *);
292 argno
= strtoul (p
, &end
, 10) - 1;
294 gcc_assert (*p
== '$');
298 gcc_assert (!any_unnumbered
);
303 any_unnumbered
= true;
304 gcc_assert (!any_numbered
);
306 gcc_assert (argno
< PP_NL_ARGMAX
);
307 gcc_assert (!formatters
[argno
]);
308 formatters
[argno
] = &args
[chunk
];
311 obstack_1grow (&buffer
->chunk_obstack
, *p
);
314 while (strchr ("qwl+#", p
[-1]));
318 /* We handle '%.Ns' and '%.*s' or '%M$.*N$s'
319 (where M == N + 1). */
324 obstack_1grow (&buffer
->chunk_obstack
, *p
);
327 while (ISDIGIT (p
[-1]));
328 gcc_assert (p
[-1] == 's');
332 gcc_assert (*p
== '*');
333 obstack_1grow (&buffer
->chunk_obstack
, '*');
339 unsigned int argno2
= strtoul (p
, &end
, 10) - 1;
341 gcc_assert (argno2
== argno
- 1);
342 gcc_assert (!any_unnumbered
);
343 gcc_assert (*p
== '$');
346 formatters
[argno2
] = formatters
[argno
];
350 gcc_assert (!any_numbered
);
351 formatters
[argno
+1] = formatters
[argno
];
354 gcc_assert (*p
== 's');
355 obstack_1grow (&buffer
->chunk_obstack
, 's');
362 obstack_1grow (&buffer
->chunk_obstack
, '\0');
363 gcc_assert (chunk
< PP_NL_ARGMAX
* 2);
364 args
[chunk
++] = XOBFINISH (&buffer
->chunk_obstack
, const char *);
367 obstack_1grow (&buffer
->chunk_obstack
, '\0');
368 gcc_assert (chunk
< PP_NL_ARGMAX
* 2);
369 args
[chunk
++] = XOBFINISH (&buffer
->chunk_obstack
, const char *);
372 /* Set output to the argument obstack, and switch line-wrapping and
374 buffer
->obstack
= &buffer
->chunk_obstack
;
375 old_wrapping_mode
= pp_set_verbatim_wrapping (pp
);
377 /* Second phase. Replace each formatter with the formatted text it
380 for (argno
= 0; formatters
[argno
]; argno
++)
388 /* We do not attempt to enforce any ordering on the modifier
391 for (p
= *formatters
[argno
];; p
++)
416 /* We don't support precision beyond that of "long long". */
417 gcc_assert (precision
< 2);
424 gcc_assert (!wide
|| precision
== 0);
427 pp_string (pp
, open_quote
);
432 pp_character (pp
, va_arg (*text
->args_ptr
, int));
438 pp_wide_integer (pp
, va_arg (*text
->args_ptr
, HOST_WIDE_INT
));
440 pp_integer_with_precision
441 (pp
, *text
->args_ptr
, precision
, int, "d");
446 pp_scalar (pp
, "%" HOST_WIDE_INT_PRINT
"o",
447 va_arg (*text
->args_ptr
, unsigned HOST_WIDE_INT
));
449 pp_integer_with_precision
450 (pp
, *text
->args_ptr
, precision
, unsigned, "o");
454 pp_string (pp
, va_arg (*text
->args_ptr
, const char *));
458 pp_pointer (pp
, va_arg (*text
->args_ptr
, void *));
463 pp_scalar (pp
, HOST_WIDE_INT_PRINT_UNSIGNED
,
464 va_arg (*text
->args_ptr
, unsigned HOST_WIDE_INT
));
466 pp_integer_with_precision
467 (pp
, *text
->args_ptr
, precision
, unsigned, "u");
472 pp_scalar (pp
, HOST_WIDE_INT_PRINT_HEX
,
473 va_arg (*text
->args_ptr
, unsigned HOST_WIDE_INT
));
475 pp_integer_with_precision
476 (pp
, *text
->args_ptr
, precision
, unsigned, "x");
481 location_t
*locus
= va_arg (*text
->args_ptr
, location_t
*);
482 gcc_assert (text
->locus
!= NULL
);
483 *text
->locus
= *locus
;
489 tree t
= va_arg (*text
->args_ptr
, tree
);
490 gcc_assert (text
->locus
!= NULL
);
491 *text
->locus
= DECL_SOURCE_LOCATION (t
);
497 tree t
= va_arg (*text
->args_ptr
, tree
), block
;
498 gcc_assert (text
->locus
!= NULL
);
499 *text
->locus
= EXPR_LOCATION (t
);
500 gcc_assert (text
->abstract_origin
!= NULL
);
501 block
= TREE_BLOCK (t
);
502 *text
->abstract_origin
= NULL
;
504 && TREE_CODE (block
) == BLOCK
505 && BLOCK_ABSTRACT_ORIGIN (block
))
507 tree ao
= BLOCK_ABSTRACT_ORIGIN (block
);
509 while (TREE_CODE (ao
) == BLOCK
510 && BLOCK_ABSTRACT_ORIGIN (ao
)
511 && BLOCK_ABSTRACT_ORIGIN (ao
) != ao
)
512 ao
= BLOCK_ABSTRACT_ORIGIN (ao
);
514 if (TREE_CODE (ao
) == FUNCTION_DECL
)
516 *text
->abstract_origin
= block
;
519 block
= BLOCK_SUPERCONTEXT (block
);
529 /* We handle '%.Ns' and '%.*s' or '%M$.*N$s'
530 (where M == N + 1). The format string should be verified
531 already from the first phase. */
536 n
= strtoul (p
, &end
, 10);
538 gcc_assert (*p
== 's');
542 gcc_assert (*p
== '*');
544 gcc_assert (*p
== 's');
545 n
= va_arg (*text
->args_ptr
, int);
547 /* This consumes a second entry in the formatters array. */
548 gcc_assert (formatters
[argno
] == formatters
[argno
+1]);
552 s
= va_arg (*text
->args_ptr
, const char *);
553 pp_append_text (pp
, s
, s
+ n
);
561 gcc_assert (pp_format_decoder (pp
));
562 ok
= pp_format_decoder (pp
) (pp
, text
, p
,
563 precision
, wide
, plus
, hash
);
569 pp_string (pp
, close_quote
);
571 obstack_1grow (&buffer
->chunk_obstack
, '\0');
572 *formatters
[argno
] = XOBFINISH (&buffer
->chunk_obstack
, const char *);
575 #ifdef ENABLE_CHECKING
576 for (; argno
< PP_NL_ARGMAX
; argno
++)
577 gcc_assert (!formatters
[argno
]);
580 /* Revert to normal obstack and wrapping mode. */
581 buffer
->obstack
= &buffer
->formatted_obstack
;
582 buffer
->line_length
= 0;
583 pp_wrapping_mode (pp
) = old_wrapping_mode
;
587 /* Format of a message pointed to by TEXT. */
589 pp_base_output_formatted_text (pretty_printer
*pp
)
592 output_buffer
*buffer
= pp_buffer (pp
);
593 struct chunk_info
*chunk_array
= buffer
->cur_chunk_array
;
594 const char **args
= chunk_array
->args
;
596 gcc_assert (buffer
->obstack
== &buffer
->formatted_obstack
);
597 gcc_assert (buffer
->line_length
== 0);
599 /* This is a third phase, first 2 phases done in pp_base_format_args.
600 Now we actually print it. */
601 for (chunk
= 0; args
[chunk
]; chunk
++)
602 pp_string (pp
, args
[chunk
]);
604 /* Deallocate the chunk structure and everything after it (i.e. the
605 associated series of formatted strings). */
606 buffer
->cur_chunk_array
= chunk_array
->prev
;
607 obstack_free (&buffer
->chunk_obstack
, chunk_array
);
610 /* Helper subroutine of output_verbatim and verbatim. Do the appropriate
611 settings needed by BUFFER for a verbatim formatting. */
613 pp_base_format_verbatim (pretty_printer
*pp
, text_info
*text
)
615 /* Set verbatim mode. */
616 pp_wrapping_mode_t oldmode
= pp_set_verbatim_wrapping (pp
);
618 /* Do the actual formatting. */
619 pp_format (pp
, text
);
620 pp_output_formatted_text (pp
);
622 /* Restore previous settings. */
623 pp_wrapping_mode (pp
) = oldmode
;
626 /* Flush the content of BUFFER onto the attached stream. */
628 pp_base_flush (pretty_printer
*pp
)
630 pp_write_text_to_stream (pp
);
632 fputc ('\n', pp
->buffer
->stream
);
633 fflush (pp
->buffer
->stream
);
634 pp_needs_newline (pp
) = false;
637 /* Sets the number of maximum characters per line PRETTY-PRINTER can
638 output in line-wrapping mode. A LENGTH value 0 suppresses
641 pp_base_set_line_maximum_length (pretty_printer
*pp
, int length
)
643 pp_line_cutoff (pp
) = length
;
644 pp_set_real_maximum_length (pp
);
647 /* Clear PRETTY-PRINTER output area text info. */
649 pp_base_clear_output_area (pretty_printer
*pp
)
651 obstack_free (pp
->buffer
->obstack
, obstack_base (pp
->buffer
->obstack
));
652 pp
->buffer
->line_length
= 0;
655 /* Set PREFIX for PRETTY-PRINTER. */
657 pp_base_set_prefix (pretty_printer
*pp
, const char *prefix
)
660 pp_set_real_maximum_length (pp
);
661 pp
->emitted_prefix
= false;
662 pp_indentation (pp
) = 0;
665 /* Free PRETTY-PRINTER's prefix, a previously malloc()'d string. */
667 pp_base_destroy_prefix (pretty_printer
*pp
)
669 if (pp
->prefix
!= NULL
)
671 free (CONST_CAST (char *, pp
->prefix
));
676 /* Write out PRETTY-PRINTER's prefix. */
678 pp_base_emit_prefix (pretty_printer
*pp
)
680 if (pp
->prefix
!= NULL
)
682 switch (pp_prefixing_rule (pp
))
685 case DIAGNOSTICS_SHOW_PREFIX_NEVER
:
688 case DIAGNOSTICS_SHOW_PREFIX_ONCE
:
689 if (pp
->emitted_prefix
)
694 pp_indentation (pp
) += 3;
697 case DIAGNOSTICS_SHOW_PREFIX_EVERY_LINE
:
699 int prefix_length
= strlen (pp
->prefix
);
700 pp_append_r (pp
, pp
->prefix
, prefix_length
);
701 pp
->emitted_prefix
= true;
708 /* Construct a PRETTY-PRINTER with PREFIX and of MAXIMUM_LENGTH
709 characters per line. */
711 pp_construct (pretty_printer
*pp
, const char *prefix
, int maximum_length
)
713 memset (pp
, 0, sizeof (pretty_printer
));
714 pp
->buffer
= XCNEW (output_buffer
);
715 obstack_init (&pp
->buffer
->chunk_obstack
);
716 obstack_init (&pp
->buffer
->formatted_obstack
);
717 pp
->buffer
->obstack
= &pp
->buffer
->formatted_obstack
;
718 pp
->buffer
->stream
= stderr
;
719 pp_line_cutoff (pp
) = maximum_length
;
720 pp_prefixing_rule (pp
) = DIAGNOSTICS_SHOW_PREFIX_ONCE
;
721 pp_set_prefix (pp
, prefix
);
722 pp_translate_identifiers (pp
) = true;
725 /* Append a string delimited by START and END to the output area of
726 PRETTY-PRINTER. No line wrapping is done. However, if beginning a
727 new line then emit PRETTY-PRINTER's prefix and skip any leading
728 whitespace if appropriate. The caller must ensure that it is
731 pp_base_append_text (pretty_printer
*pp
, const char *start
, const char *end
)
733 /* Emit prefix and skip whitespace if we're starting a new line. */
734 if (pp
->buffer
->line_length
== 0)
737 if (pp_is_wrapping_line (pp
))
738 while (start
!= end
&& *start
== ' ')
741 pp_append_r (pp
, start
, end
- start
);
744 /* Finishes constructing a NULL-terminated character string representing
745 the PRETTY-PRINTED text. */
747 pp_base_formatted_text (pretty_printer
*pp
)
749 obstack_1grow (pp
->buffer
->obstack
, '\0');
750 return pp_formatted_text_data (pp
);
753 /* Return a pointer to the last character emitted in PRETTY-PRINTER's
754 output area. A NULL pointer means no character available. */
756 pp_base_last_position_in_text (const pretty_printer
*pp
)
758 const char *p
= NULL
;
759 struct obstack
*text
= pp
->buffer
->obstack
;
761 if (obstack_base (text
) != obstack_next_free (text
))
762 p
= ((const char *) obstack_next_free (text
)) - 1;
766 /* Return the amount of characters PRETTY-PRINTER can accept to
767 make a full line. Meaningful only in line-wrapping mode. */
769 pp_base_remaining_character_count_for_line (pretty_printer
*pp
)
771 return pp
->maximum_length
- pp
->buffer
->line_length
;
775 /* Format a message into BUFFER a la printf. */
777 pp_printf (pretty_printer
*pp
, const char *msg
, ...)
785 text
.format_spec
= msg
;
787 pp_format (pp
, &text
);
788 pp_output_formatted_text (pp
);
793 /* Output MESSAGE verbatim into BUFFER. */
795 pp_verbatim (pretty_printer
*pp
, const char *msg
, ...)
803 text
.format_spec
= msg
;
805 pp_format_verbatim (pp
, &text
);
811 /* Have PRETTY-PRINTER start a new line. */
813 pp_base_newline (pretty_printer
*pp
)
815 obstack_1grow (pp
->buffer
->obstack
, '\n');
816 pp
->buffer
->line_length
= 0;
819 /* Have PRETTY-PRINTER add a CHARACTER. */
821 pp_base_character (pretty_printer
*pp
, int c
)
823 if (pp_is_wrapping_line (pp
)
824 && pp_remaining_character_count_for_line (pp
) <= 0)
830 obstack_1grow (pp
->buffer
->obstack
, c
);
831 ++pp
->buffer
->line_length
;
834 /* Append a STRING to the output area of PRETTY-PRINTER; the STRING may
835 be line-wrapped if in appropriate mode. */
837 pp_base_string (pretty_printer
*pp
, const char *str
)
839 pp_maybe_wrap_text (pp
, str
, str
+ (str
? strlen (str
) : 0));
842 /* Maybe print out a whitespace if needed. */
845 pp_base_maybe_space (pretty_printer
*pp
)
847 if (pp_base (pp
)->padding
!= pp_none
)
850 pp_base (pp
)->padding
= pp_none
;
854 /* Print the identifier ID to PRETTY-PRINTER. */
857 pp_base_tree_identifier (pretty_printer
*pp
, tree id
)
859 if (pp_translate_identifiers (pp
))
861 const char *text
= identifier_to_locale (IDENTIFIER_POINTER (id
));
862 pp_append_text (pp
, text
, text
+ strlen (text
));
865 pp_append_text (pp
, IDENTIFIER_POINTER (id
),
866 IDENTIFIER_POINTER (id
) + IDENTIFIER_LENGTH (id
));
869 /* The string starting at P has LEN (at least 1) bytes left; if they
870 start with a valid UTF-8 sequence, return the length of that
871 sequence and set *VALUE to the value of that sequence, and
872 otherwise return 0 and set *VALUE to (unsigned int) -1. */
875 decode_utf8_char (const unsigned char *p
, size_t len
, unsigned int *value
)
886 for (t
= *p
; t
& 0x80; t
<<= 1)
889 if (utf8_len
> len
|| utf8_len
< 2 || utf8_len
> 6)
891 *value
= (unsigned int) -1;
894 ch
= *p
& ((1 << (7 - utf8_len
)) - 1);
895 for (i
= 1; i
< utf8_len
; i
++)
897 unsigned int u
= p
[i
];
898 if ((u
& 0xC0) != 0x80)
900 *value
= (unsigned int) -1;
903 ch
= (ch
<< 6) | (u
& 0x3F);
905 if ( (ch
<= 0x7F && utf8_len
> 1)
906 || (ch
<= 0x7FF && utf8_len
> 2)
907 || (ch
<= 0xFFFF && utf8_len
> 3)
908 || (ch
<= 0x1FFFFF && utf8_len
> 4)
909 || (ch
<= 0x3FFFFFF && utf8_len
> 5)
910 || (ch
>= 0xD800 && ch
<= 0xDFFF))
912 *value
= (unsigned int) -1;
925 /* Given IDENT, an identifier in the internal encoding, return a
926 version of IDENT suitable for diagnostics in the locale character
927 set: either IDENT itself, or a garbage-collected string converted
928 to the locale character set and using escape sequences if not
929 representable in the locale character set or containing control
930 characters or invalid byte sequences. Existing backslashes in
931 IDENT are not doubled, so the result may not uniquely specify the
932 contents of an arbitrary byte sequence identifier. */
935 identifier_to_locale (const char *ident
)
937 const unsigned char *uid
= (const unsigned char *) ident
;
938 size_t idlen
= strlen (ident
);
939 bool valid_printable_utf8
= true;
940 bool all_ascii
= true;
943 for (i
= 0; i
< idlen
;)
946 size_t utf8_len
= decode_utf8_char (&uid
[i
], idlen
- i
, &c
);
947 if (utf8_len
== 0 || c
<= 0x1F || (c
>= 0x7F && c
<= 0x9F))
949 valid_printable_utf8
= false;
957 /* If IDENT contains invalid UTF-8 sequences (which may occur with
958 attributes putting arbitrary byte sequences in identifiers), or
959 control characters, we use octal escape sequences for all bytes
960 outside printable ASCII. */
961 if (!valid_printable_utf8
)
963 char *ret
= GGC_NEWVEC (char, 4 * idlen
+ 1);
965 for (i
= 0; i
< idlen
; i
++)
967 if (uid
[i
] > 0x1F && uid
[i
] < 0x7F)
971 sprintf (p
, "\\%03o", uid
[i
]);
979 /* Otherwise, if it is valid printable ASCII, or printable UTF-8
980 with the locale character set being UTF-8, IDENT is used. */
981 if (all_ascii
|| locale_utf8
)
984 /* Otherwise IDENT is converted to the locale character set if
986 #if defined ENABLE_NLS && defined HAVE_LANGINFO_CODESET && HAVE_ICONV
987 if (locale_encoding
!= NULL
)
989 iconv_t cd
= iconv_open (locale_encoding
, "UTF-8");
990 bool conversion_ok
= true;
992 if (cd
!= (iconv_t
) -1)
994 size_t ret_alloc
= 4 * idlen
+ 1;
997 /* Repeat the whole conversion process as needed with
998 larger buffers so non-reversible transformations can
999 always be detected. */
1000 ICONV_CONST
char *inbuf
= CONST_CAST (char *, ident
);
1002 size_t inbytesleft
= idlen
;
1003 size_t outbytesleft
= ret_alloc
- 1;
1006 ret
= GGC_NEWVEC (char, ret_alloc
);
1009 if (iconv (cd
, 0, 0, 0, 0) == (size_t) -1)
1011 conversion_ok
= false;
1015 iconv_ret
= iconv (cd
, &inbuf
, &inbytesleft
,
1016 &outbuf
, &outbytesleft
);
1017 if (iconv_ret
== (size_t) -1 || inbytesleft
!= 0)
1028 conversion_ok
= false;
1032 else if (iconv_ret
!= 0)
1034 conversion_ok
= false;
1037 /* Return to initial shift state. */
1038 if (iconv (cd
, 0, 0, &outbuf
, &outbytesleft
) == (size_t) -1)
1049 conversion_ok
= false;
1063 /* Otherwise, convert non-ASCII characters in IDENT to UCNs. */
1065 char *ret
= GGC_NEWVEC (char, 10 * idlen
+ 1);
1067 for (i
= 0; i
< idlen
;)
1070 size_t utf8_len
= decode_utf8_char (&uid
[i
], idlen
- i
, &c
);
1075 sprintf (p
, "\\U%08x", c
);