X-Git-Url: https://code.delx.au/gnu-emacs/blobdiff_plain/4d0108a132788e0c3903eb4d5875321ed6e8eef1..551562513e304a38a0caa1b94d87cc39a1af7c18:/src/bidi.c diff --git a/src/bidi.c b/src/bidi.c index cbc1820c2a..e3b1f54968 100644 --- a/src/bidi.c +++ b/src/bidi.c @@ -1313,13 +1313,13 @@ bidi_fetch_char (ptrdiff_t charpos, ptrdiff_t bytepos, ptrdiff_t *disp_pos, /* `(space ...)' display specs are handled as paragraph separators for the purposes of the reordering; see UAX#9 section 3 and clause HL1 in section 4.3 there. */ - ch = 0x2029; + ch = PARAGRAPH_SEPARATOR; } else { /* All other display specs are handled as the Unicode Object Replacement Character. */ - ch = 0xFFFC; + ch = OBJECT_REPLACEMENT_CHARACTER; } disp_end_pos = compute_display_string_end (*disp_pos, string); if (disp_end_pos < 0) @@ -1799,6 +1799,11 @@ bidi_explicit_dir_char (int ch) if (!bidi_initialized) emacs_abort (); + if (ch < 0) + { + eassert (ch == BIDI_EOB); + return false; + } ch_type = (bidi_type_t) XINT (CHAR_TABLE_REF (bidi_type_table, ch)); return (ch_type == LRE || ch_type == LRO || ch_type == RLE || ch_type == RLO @@ -2447,7 +2452,7 @@ typedef struct bpa_stack_entry { #define MAX_BPA_STACK ((int)max (MAX_ALLOCA / sizeof (bpa_stack_entry), 1)) /* UAX#9 says to match opening brackets with the matching closing - brackets or their canonical equivalents. As of Unicode 7.0, there + brackets or their canonical equivalents. As of Unicode 8.0, there are only 2 bracket characters that have canonical equivalence decompositions: u+2329 and u+232A. So instead of accessing the table in uni-decomposition.el, we just handle these 2 characters @@ -2477,8 +2482,8 @@ typedef struct bpa_stack_entry { #define CANONICAL_EQU(c) \ ( ASCII_CHAR_P (c) ? c \ - : (c) == 0x2329 ? 0x3008 \ - : (c) == 0x232a ? 0x3009 \ + : (c) == LEFT_POINTING_ANGLE_BRACKET ? LEFT_ANGLE_BRACKET \ + : (c) == RIGHT_POINTING_ANGLE_BRACKET ? RIGHT_ANGLE_BRACKET \ : c ) #ifdef ENABLE_CHECKING @@ -2943,7 +2948,7 @@ bidi_resolve_neutral (struct bidi_it *bidi_it) we are already at paragraph end. */ && (is_neutral || bidi_isolate_fmt_char (type))) /* N1-N2/Retaining */ - || (type == WEAK_BN && bidi_explicit_dir_char (bidi_it->ch))) + || type == WEAK_BN) { if (bidi_it->next_for_neutral.type != UNKNOWN_BT) { @@ -2973,6 +2978,7 @@ bidi_resolve_neutral (struct bidi_it *bidi_it) entering the expensive loop in the "else" clause. */ else if (current_level == 0 && bidi_it->prev_for_neutral.type == STRONG_L + && type != WEAK_BN && !bidi_explicit_dir_char (bidi_it->ch) && !bidi_isolate_fmt_char (type)) type = bidi_resolve_neutral_1 (bidi_it->prev_for_neutral.type, @@ -2986,6 +2992,7 @@ bidi_resolve_neutral (struct bidi_it *bidi_it) && (bidi_it->prev_for_neutral.type == STRONG_R || bidi_it->prev_for_neutral.type == WEAK_EN || bidi_it->prev_for_neutral.type == WEAK_AN) + && type != WEAK_BN && !bidi_explicit_dir_char (bidi_it->ch) && !bidi_isolate_fmt_char (type)) type = bidi_resolve_neutral_1 (bidi_it->prev_for_neutral.type, @@ -3211,6 +3218,7 @@ bidi_level_of_next_char (struct bidi_it *bidi_it) it belongs to a sequence of WS characters preceding a newline or a TAB or a paragraph separator. */ if ((bidi_it->orig_type == NEUTRAL_WS + || bidi_it->orig_type == WEAK_BN || bidi_isolate_fmt_char (bidi_it->orig_type)) && bidi_it->next_for_ws.charpos < bidi_it->charpos) { @@ -3244,11 +3252,14 @@ bidi_level_of_next_char (struct bidi_it *bidi_it) /* Resolve implicit levels. */ if (bidi_it->orig_type == NEUTRAL_B /* L1 */ - || bidi_it->orig_type == NEUTRAL_S - || bidi_it->ch == '\n' || bidi_it->ch == BIDI_EOB - || (bidi_it->orig_type == NEUTRAL_WS - && (bidi_it->next_for_ws.type == NEUTRAL_B - || bidi_it->next_for_ws.type == NEUTRAL_S))) + || bidi_it->orig_type == NEUTRAL_S + || bidi_it->ch == '\n' || bidi_it->ch == BIDI_EOB + || ((bidi_it->orig_type == NEUTRAL_WS + || bidi_it->orig_type == WEAK_BN + || bidi_isolate_fmt_char (bidi_it->orig_type) + || bidi_explicit_dir_char (bidi_it->ch)) + && (bidi_it->next_for_ws.type == NEUTRAL_B + || bidi_it->next_for_ws.type == NEUTRAL_S))) level = bidi_it->level_stack[0].level; else if ((level & 1) == 0) /* I1 */ { @@ -3343,7 +3354,6 @@ bidi_move_to_visually_next (struct bidi_it *bidi_it) { int old_level, new_level, next_level; struct bidi_it sentinel; - struct gcpro gcpro1; if (bidi_it->charpos < 0 || bidi_it->bytepos < 0) emacs_abort (); @@ -3353,11 +3363,6 @@ bidi_move_to_visually_next (struct bidi_it *bidi_it) bidi_it->scan_dir = 1; /* default to logical order */ } - /* The code below can call eval, and thus cause GC. If we are - iterating a Lisp string, make sure it won't be GCed. */ - if (STRINGP (bidi_it->string.lstring)) - GCPRO1 (bidi_it->string.lstring); - /* If we just passed a newline, initialize for the next line. */ if (!bidi_it->first_elt && (bidi_it->ch == '\n' || bidi_it->ch == BIDI_EOB)) @@ -3503,9 +3508,6 @@ bidi_move_to_visually_next (struct bidi_it *bidi_it) eassert (bidi_it->resolved_level >= 0 && bidi_it->resolved_level <= BIDI_MAXDEPTH + 2); - - if (STRINGP (bidi_it->string.lstring)) - UNGCPRO; } /* Utility function for looking for strong directional characters