-/* Low-level bidirectional buffer-scanning functions for GNU Emacs.
- Copyright (C) 2000, 2001, 2004, 2005, 2009 Free Software Foundation, Inc.
+/* Low-level bidirectional buffer/string-scanning functions for GNU Emacs.
+ Copyright (C) 2000-2001, 2004-2005, 2009-2011
+ Free Software Foundation, Inc.
This file is part of GNU Emacs.
-GNU Emacs is free software; you can redistribute it and/or modify
+GNU Emacs is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
-the Free Software Foundation; either version 2, or (at your option)
-any later version.
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
GNU Emacs is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
-
You should have received a copy of the GNU General Public License
-along with GNU Emacs; see the file COPYING. If not, write to
-the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
-Boston, MA 02110-1301, USA. */
+along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */
/* Written by Eli Zaretskii <eliz@gnu.org>.
A sequential implementation of the Unicode Bidirectional algorithm,
- as per UAX#9, a part of the Unicode Standard.
+ (UBA) as per UAX#9, a part of the Unicode Standard.
Unlike the reference and most other implementations, this one is
- designed to be called once for every character in the buffer.
+ designed to be called once for every character in the buffer or
+ string.
- The main entry point is bidi_get_next_char_visually. Each time it
+ The main entry point is bidi_move_to_visually_next. Each time it
is called, it finds the next character in the visual order, and
returns its information in a special structure. The caller is then
expected to process this character for display or any other
- purposes, and call bidi_get_next_char_visually for the next
- character. See the comments in bidi_get_next_char_visually for
- more details about its algorithm that finds the next visual-order
+ purposes, and call bidi_move_to_visually_next for the next
+ character. See the comments in bidi_move_to_visually_next for more
+ details about its algorithm that finds the next visual-order
character by resolving their levels on the fly.
+ Two other entry points are bidi_paragraph_init and
+ bidi_mirror_char. The first determines the base direction of a
+ paragraph, while the second returns the mirrored version of its
+ argument character.
+
+ A few auxiliary entry points are used to initialize the bidi
+ iterator for iterating an object (buffer or string), push and pop
+ the bidi iterator state, and save and restore the state of the bidi
+ cache.
+
If you want to understand the code, you will have to read it
together with the relevant portions of UAX#9. The comments include
references to UAX#9 rules, for that very reason.
rule X9 and to its modifications decribed in the "Implementation
Notes" section of UAX#9, under "Retaining Format Codes". */
-#ifdef HAVE_CONFIG_H
#include <config.h>
-#endif
-
#include <stdio.h>
-
-#ifdef HAVE_STRING_H
-#include <string.h>
-#endif
-
#include <setjmp.h>
#include "lisp.h"
static int bidi_initialized = 0;
-static Lisp_Object bidi_type_table;
+static Lisp_Object bidi_type_table, bidi_mirror_table;
-/* FIXME: Remove these when bidi_explicit_dir_char uses a lookup table. */
#define LRM_CHAR 0x200E
#define RLM_CHAR 0x200F
-#define LRE_CHAR 0x202A
-#define RLE_CHAR 0x202B
-#define PDF_CHAR 0x202C
-#define LRO_CHAR 0x202D
-#define RLO_CHAR 0x202E
-
#define BIDI_EOB -1
-#define BIDI_BOB -2 /* FIXME: Is this needed? */
-
-/* Local data structures. (Look in dispextern.h for the rest.) */
-
-/* What we need to know about the current paragraph. */
-struct bidi_paragraph_info {
- int start_bytepos; /* byte position where it begins */
- int end_bytepos; /* byte position where it ends */
- int embedding_level; /* its basic embedding level */
- bidi_dir_t base_dir; /* its base direction */
-};
/* Data type for describing the bidirectional character categories. */
typedef enum {
STRONG
} bidi_category_t;
+extern int bidi_ignore_explicit_marks_for_paragraph_level EXTERNALLY_VISIBLE;
int bidi_ignore_explicit_marks_for_paragraph_level = 1;
-static Lisp_Object fallback_paragraph_start_re, fallback_paragraph_separate_re;
+static Lisp_Object paragraph_start_re, paragraph_separate_re;
static Lisp_Object Qparagraph_start, Qparagraph_separate;
-static void
-bidi_initialize ()
-{
- /* FIXME: This should come from the Unicode Database. */
- struct {
- int from, to;
- bidi_type_t type;
- } bidi_type[] =
- { { 0x0000, 0x0008, WEAK_BN },
- { 0x0009, 0x0000, NEUTRAL_S },
- { 0x000A, 0x0000, NEUTRAL_B },
- { 0x000B, 0x0000, NEUTRAL_S },
- { 0x000C, 0x0000, NEUTRAL_WS },
- { 0x000D, 0x0000, NEUTRAL_B },
- { 0x000E, 0x001B, WEAK_BN },
- { 0x001C, 0x001E, NEUTRAL_B },
- { 0x001F, 0x0000, NEUTRAL_S },
- { 0x0020, 0x0000, NEUTRAL_WS },
- { 0x0021, 0x0022, NEUTRAL_ON },
- { 0x0023, 0x0025, WEAK_ET },
- { 0x0026, 0x002A, NEUTRAL_ON },
- { 0x002B, 0x0000, WEAK_ES },
- { 0x002C, 0x0000, WEAK_CS },
- { 0x002D, 0x0000, WEAK_ES },
- { 0x002E, 0x002F, WEAK_CS },
- { 0x0030, 0x0039, WEAK_EN },
- { 0x003A, 0x0000, WEAK_CS },
- { 0x003B, 0x0040, NEUTRAL_ON },
- { 0x005B, 0x0060, NEUTRAL_ON },
- { 0x007B, 0x007E, NEUTRAL_ON },
- { 0x007F, 0x0084, WEAK_BN },
- { 0x0085, 0x0000, NEUTRAL_B },
- { 0x0086, 0x009F, WEAK_BN },
- { 0x00A0, 0x0000, WEAK_CS },
- { 0x00A1, 0x0000, NEUTRAL_ON },
- { 0x00A2, 0x00A5, WEAK_ET },
- { 0x00A6, 0x00A9, NEUTRAL_ON },
- { 0x00AB, 0x00AC, NEUTRAL_ON },
- { 0x00AD, 0x0000, WEAK_BN },
- { 0x00AE, 0x00Af, NEUTRAL_ON },
- { 0x00B0, 0x00B1, WEAK_ET },
- { 0x00B2, 0x00B3, WEAK_EN },
- { 0x00B4, 0x0000, NEUTRAL_ON },
- { 0x00B6, 0x00B8, NEUTRAL_ON },
- { 0x00B9, 0x0000, WEAK_EN },
- { 0x00BB, 0x00BF, NEUTRAL_ON },
- { 0x00D7, 0x0000, NEUTRAL_ON },
- { 0x00F7, 0x0000, NEUTRAL_ON },
- { 0x02B9, 0x02BA, NEUTRAL_ON },
- { 0x02C2, 0x02CF, NEUTRAL_ON },
- { 0x02D2, 0x02DF, NEUTRAL_ON },
- { 0x02E5, 0x02ED, NEUTRAL_ON },
- { 0x0300, 0x036F, WEAK_NSM },
- { 0x0374, 0x0375, NEUTRAL_ON },
- { 0x037E, 0x0385, NEUTRAL_ON },
- { 0x0387, 0x0000, NEUTRAL_ON },
- { 0x03F6, 0x0000, NEUTRAL_ON },
- { 0x0483, 0x0489, WEAK_NSM },
- { 0x058A, 0x0000, NEUTRAL_ON },
- { 0x0591, 0x05BD, WEAK_NSM },
- { 0x05BE, 0x0000, STRONG_R },
- { 0x05BF, 0x0000, WEAK_NSM },
- { 0x05C0, 0x0000, STRONG_R },
- { 0x05C1, 0x05C2, WEAK_NSM },
- { 0x05C3, 0x0000, STRONG_R },
- { 0x05C4, 0x05C5, WEAK_NSM },
- { 0x05C6, 0x0000, STRONG_R },
- { 0x05C7, 0x0000, WEAK_NSM },
- { 0x05D0, 0x05F4, STRONG_R },
- { 0x060C, 0x0000, WEAK_CS },
- { 0x061B, 0x064A, STRONG_AL },
- { 0x064B, 0x0655, WEAK_NSM },
- { 0x0660, 0x0669, WEAK_AN },
- { 0x066A, 0x0000, WEAK_ET },
- { 0x066B, 0x066C, WEAK_AN },
- { 0x066D, 0x066F, STRONG_AL },
- { 0x0670, 0x0000, WEAK_NSM },
- { 0x0671, 0x06D5, STRONG_AL },
- { 0x06D6, 0x06DC, WEAK_NSM },
- { 0x06DD, 0x0000, STRONG_AL },
- { 0x06DE, 0x06E4, WEAK_NSM },
- { 0x06E5, 0x06E6, STRONG_AL },
- { 0x06E7, 0x06E8, WEAK_NSM },
- { 0x06E9, 0x0000, NEUTRAL_ON },
- { 0x06EA, 0x06ED, WEAK_NSM },
- { 0x06F0, 0x06F9, WEAK_EN },
- { 0x06FA, 0x070D, STRONG_AL },
- { 0x070F, 0x0000, WEAK_BN },
- { 0x0710, 0x0000, STRONG_AL },
- { 0x0711, 0x0000, WEAK_NSM },
- { 0x0712, 0x072C, STRONG_AL },
- { 0x0730, 0x074A, WEAK_NSM },
- { 0x0780, 0x07A5, STRONG_AL },
- { 0x07A6, 0x07B0, WEAK_NSM },
- { 0x07B1, 0x0000, STRONG_AL },
- { 0x0901, 0x0902, WEAK_NSM },
- { 0x093C, 0x0000, WEAK_NSM },
- { 0x0941, 0x0948, WEAK_NSM },
- { 0x094D, 0x0000, WEAK_NSM },
- { 0x0951, 0x0954, WEAK_NSM },
- { 0x0962, 0x0963, WEAK_NSM },
- { 0x0981, 0x0000, WEAK_NSM },
- { 0x09BC, 0x0000, WEAK_NSM },
- { 0x09C1, 0x09C4, WEAK_NSM },
- { 0x09CD, 0x0000, WEAK_NSM },
- { 0x09E2, 0x09E3, WEAK_NSM },
- { 0x09F2, 0x09F3, WEAK_ET },
- { 0x0A02, 0x0000, WEAK_NSM },
- { 0x0A3C, 0x0000, WEAK_NSM },
- { 0x0A41, 0x0A4D, WEAK_NSM },
- { 0x0A70, 0x0A71, WEAK_NSM },
- { 0x0A81, 0x0A82, WEAK_NSM },
- { 0x0ABC, 0x0000, WEAK_NSM },
- { 0x0AC1, 0x0AC8, WEAK_NSM },
- { 0x0ACD, 0x0000, WEAK_NSM },
- { 0x0B01, 0x0000, WEAK_NSM },
- { 0x0B3C, 0x0000, WEAK_NSM },
- { 0x0B3F, 0x0000, WEAK_NSM },
- { 0x0B41, 0x0B43, WEAK_NSM },
- { 0x0B4D, 0x0B56, WEAK_NSM },
- { 0x0B82, 0x0000, WEAK_NSM },
- { 0x0BC0, 0x0000, WEAK_NSM },
- { 0x0BCD, 0x0000, WEAK_NSM },
- { 0x0C3E, 0x0C40, WEAK_NSM },
- { 0x0C46, 0x0C56, WEAK_NSM },
- { 0x0CBF, 0x0000, WEAK_NSM },
- { 0x0CC6, 0x0000, WEAK_NSM },
- { 0x0CCC, 0x0CCD, WEAK_NSM },
- { 0x0D41, 0x0D43, WEAK_NSM },
- { 0x0D4D, 0x0000, WEAK_NSM },
- { 0x0DCA, 0x0000, WEAK_NSM },
- { 0x0DD2, 0x0DD6, WEAK_NSM },
- { 0x0E31, 0x0000, WEAK_NSM },
- { 0x0E34, 0x0E3A, WEAK_NSM },
- { 0x0E3F, 0x0000, WEAK_ET },
- { 0x0E47, 0x0E4E, WEAK_NSM },
- { 0x0EB1, 0x0000, WEAK_NSM },
- { 0x0EB4, 0x0EBC, WEAK_NSM },
- { 0x0EC8, 0x0ECD, WEAK_NSM },
- { 0x0F18, 0x0F19, WEAK_NSM },
- { 0x0F35, 0x0000, WEAK_NSM },
- { 0x0F37, 0x0000, WEAK_NSM },
- { 0x0F39, 0x0000, WEAK_NSM },
- { 0x0F3A, 0x0F3D, NEUTRAL_ON },
- { 0x0F71, 0x0F7E, WEAK_NSM },
- { 0x0F80, 0x0F84, WEAK_NSM },
- { 0x0F86, 0x0F87, WEAK_NSM },
- { 0x0F90, 0x0FBC, WEAK_NSM },
- { 0x0FC6, 0x0000, WEAK_NSM },
- { 0x102D, 0x1030, WEAK_NSM },
- { 0x1032, 0x1037, WEAK_NSM },
- { 0x1039, 0x0000, WEAK_NSM },
- { 0x1058, 0x1059, WEAK_NSM },
- { 0x1680, 0x0000, NEUTRAL_WS },
- { 0x169B, 0x169C, NEUTRAL_ON },
- { 0x1712, 0x1714, WEAK_NSM },
- { 0x1732, 0x1734, WEAK_NSM },
- { 0x1752, 0x1753, WEAK_NSM },
- { 0x1772, 0x1773, WEAK_NSM },
- { 0x17B7, 0x17BD, WEAK_NSM },
- { 0x17C6, 0x0000, WEAK_NSM },
- { 0x17C9, 0x17D3, WEAK_NSM },
- { 0x17DB, 0x0000, WEAK_ET },
- { 0x1800, 0x180A, NEUTRAL_ON },
- { 0x180B, 0x180D, WEAK_NSM },
- { 0x180E, 0x0000, WEAK_BN },
- { 0x18A9, 0x0000, WEAK_NSM },
- { 0x1FBD, 0x0000, NEUTRAL_ON },
- { 0x1FBF, 0x1FC1, NEUTRAL_ON },
- { 0x1FCD, 0x1FCF, NEUTRAL_ON },
- { 0x1FDD, 0x1FDF, NEUTRAL_ON },
- { 0x1FED, 0x1FEF, NEUTRAL_ON },
- { 0x1FFD, 0x1FFE, NEUTRAL_ON },
- { 0x2000, 0x200A, NEUTRAL_WS },
- { 0x200B, 0x200D, WEAK_BN },
- { 0x200F, 0x0000, STRONG_R },
- { 0x2010, 0x2027, NEUTRAL_ON },
- { 0x2028, 0x0000, NEUTRAL_WS },
- { 0x2029, 0x0000, NEUTRAL_B },
- { 0x202A, 0x0000, LRE },
- { 0x202B, 0x0000, RLE },
- { 0x202C, 0x0000, PDF },
- { 0x202D, 0x0000, LRO },
- { 0x202E, 0x0000, RLO },
- { 0x202F, 0x0000, NEUTRAL_WS },
- { 0x2030, 0x2034, WEAK_ET },
- { 0x2035, 0x2057, NEUTRAL_ON },
- { 0x205F, 0x0000, NEUTRAL_WS },
- { 0x2060, 0x206F, WEAK_BN },
- { 0x2070, 0x0000, WEAK_EN },
- { 0x2074, 0x2079, WEAK_EN },
- { 0x207A, 0x207B, WEAK_ET },
- { 0x207C, 0x207E, NEUTRAL_ON },
- { 0x2080, 0x2089, WEAK_EN },
- { 0x208A, 0x208B, WEAK_ET },
- { 0x208C, 0x208E, NEUTRAL_ON },
- { 0x20A0, 0x20B1, WEAK_ET },
- { 0x20D0, 0x20EA, WEAK_NSM },
- { 0x2100, 0x2101, NEUTRAL_ON },
- { 0x2103, 0x2106, NEUTRAL_ON },
- { 0x2108, 0x2109, NEUTRAL_ON },
- { 0x2114, 0x0000, NEUTRAL_ON },
- { 0x2116, 0x2118, NEUTRAL_ON },
- { 0x211E, 0x2123, NEUTRAL_ON },
- { 0x2125, 0x0000, NEUTRAL_ON },
- { 0x2127, 0x0000, NEUTRAL_ON },
- { 0x2129, 0x0000, NEUTRAL_ON },
- { 0x212E, 0x0000, WEAK_ET },
- { 0x2132, 0x0000, NEUTRAL_ON },
- { 0x213A, 0x0000, NEUTRAL_ON },
- { 0x2140, 0x2144, NEUTRAL_ON },
- { 0x214A, 0x215F, NEUTRAL_ON },
- { 0x2190, 0x2211, NEUTRAL_ON },
- { 0x2212, 0x2213, WEAK_ET },
- { 0x2214, 0x2335, NEUTRAL_ON },
- { 0x237B, 0x2394, NEUTRAL_ON },
- { 0x2396, 0x244A, NEUTRAL_ON },
- { 0x2460, 0x249B, WEAK_EN },
- { 0x24EA, 0x0000, WEAK_EN },
- { 0x24EB, 0x2FFB, NEUTRAL_ON },
- { 0x3000, 0x0000, NEUTRAL_WS },
- { 0x3001, 0x3004, NEUTRAL_ON },
- { 0x3008, 0x3020, NEUTRAL_ON },
- { 0x302A, 0x302F, WEAK_NSM },
- { 0x3030, 0x0000, NEUTRAL_ON },
- { 0x3036, 0x3037, NEUTRAL_ON },
- { 0x303D, 0x303F, NEUTRAL_ON },
- { 0x3099, 0x309A, WEAK_NSM },
- { 0x309B, 0x309C, NEUTRAL_ON },
- { 0x30A0, 0x0000, NEUTRAL_ON },
- { 0x30FB, 0x0000, NEUTRAL_ON },
- { 0x3251, 0x325F, NEUTRAL_ON },
- { 0x32B1, 0x32BF, NEUTRAL_ON },
- { 0xA490, 0xA4C6, NEUTRAL_ON },
- { 0xFB1D, 0x0000, STRONG_R },
- { 0xFB1E, 0x0000, WEAK_NSM },
- { 0xFB1F, 0xFB28, STRONG_R },
- { 0xFB29, 0x0000, WEAK_ET },
- { 0xFB2A, 0xFB4F, STRONG_R },
- { 0xFB50, 0xFD3D, STRONG_AL },
- { 0xFD3E, 0xFD3F, NEUTRAL_ON },
- { 0xFD50, 0xFDFC, STRONG_AL },
- { 0xFE00, 0xFE23, WEAK_NSM },
- { 0xFE30, 0xFE4F, NEUTRAL_ON },
- { 0xFE50, 0x0000, WEAK_CS },
- { 0xFE51, 0x0000, NEUTRAL_ON },
- { 0xFE52, 0x0000, WEAK_CS },
- { 0xFE54, 0x0000, NEUTRAL_ON },
- { 0xFE55, 0x0000, WEAK_CS },
- { 0xFE56, 0xFE5E, NEUTRAL_ON },
- { 0xFE5F, 0x0000, WEAK_ET },
- { 0xFE60, 0xFE61, NEUTRAL_ON },
- { 0xFE62, 0xFE63, WEAK_ET },
- { 0xFE64, 0xFE68, NEUTRAL_ON },
- { 0xFE69, 0xFE6A, WEAK_ET },
- { 0xFE6B, 0x0000, NEUTRAL_ON },
- { 0xFE70, 0xFEFC, STRONG_AL },
- { 0xFEFF, 0x0000, WEAK_BN },
- { 0xFF01, 0xFF02, NEUTRAL_ON },
- { 0xFF03, 0xFF05, WEAK_ET },
- { 0xFF06, 0xFF0A, NEUTRAL_ON },
- { 0xFF0B, 0x0000, WEAK_ET },
- { 0xFF0C, 0x0000, WEAK_CS },
- { 0xFF0D, 0x0000, WEAK_ET },
- { 0xFF0E, 0x0000, WEAK_CS },
- { 0xFF0F, 0x0000, WEAK_ES },
- { 0xFF10, 0xFF19, WEAK_EN },
- { 0xFF1A, 0x0000, WEAK_CS },
- { 0xFF1B, 0xFF20, NEUTRAL_ON },
- { 0xFF3B, 0xFF40, NEUTRAL_ON },
- { 0xFF5B, 0xFF65, NEUTRAL_ON },
- { 0xFFE0, 0xFFE1, WEAK_ET },
- { 0xFFE2, 0xFFE4, NEUTRAL_ON },
- { 0xFFE5, 0xFFE6, WEAK_ET },
- { 0xFFE8, 0xFFEE, NEUTRAL_ON },
- { 0xFFF9, 0xFFFB, WEAK_BN },
- { 0xFFFC, 0xFFFD, NEUTRAL_ON },
- { 0x1D167, 0x1D169, WEAK_NSM },
- { 0x1D173, 0x1D17A, WEAK_BN },
- { 0x1D17B, 0x1D182, WEAK_NSM },
- { 0x1D185, 0x1D18B, WEAK_NSM },
- { 0x1D1AA, 0x1D1AD, WEAK_NSM },
- { 0x1D7CE, 0x1D7FF, WEAK_EN },
- { 0xE0001, 0xE007F, WEAK_BN } };
- int i;
-
- bidi_type_table = Fmake_char_table (Qnil, make_number (STRONG_L));
- staticpro (&bidi_type_table);
-
- for (i = 0; i < sizeof bidi_type / sizeof bidi_type[0]; i++)
- char_table_set_range (bidi_type_table, bidi_type[i].from,
- bidi_type[i].to ? bidi_type[i].to : bidi_type[i].from,
- make_number (bidi_type[i].type));
-
- fallback_paragraph_start_re =
- XSYMBOL (Fintern_soft (build_string ("paragraph-start"), Qnil))->value;
- if (!STRINGP (fallback_paragraph_start_re))
- fallback_paragraph_start_re = build_string ("\f\\|[ \t]*$");
- staticpro (&fallback_paragraph_start_re);
- Qparagraph_start = intern ("paragraph-start");
- staticpro (&Qparagraph_start);
- fallback_paragraph_separate_re =
- XSYMBOL (Fintern_soft (build_string ("paragraph-separate"), Qnil))->value;
- if (!STRINGP (fallback_paragraph_separate_re))
- fallback_paragraph_separate_re = build_string ("[ \t\f]*$");
- staticpro (&fallback_paragraph_separate_re);
- Qparagraph_separate = intern ("paragraph-separate");
- staticpro (&Qparagraph_separate);
- bidi_initialized = 1;
-}
+\f
+/***********************************************************************
+ Utilities
+ ***********************************************************************/
/* Return the bidi type of a character CH, subject to the current
directional OVERRIDE. */
-bidi_type_t
+static inline bidi_type_t
bidi_get_type (int ch, bidi_dir_t override)
{
bidi_type_t default_type;
}
}
-void
+static void
bidi_check_type (bidi_type_t type)
{
if (type < UNKNOWN_BT || type > NEUTRAL_ON)
}
/* Given a bidi TYPE of a character, return its category. */
-bidi_category_t
+static inline bidi_category_t
bidi_get_category (bidi_type_t type)
{
switch (type)
}
}
-/* Return the mirrored character of C, if any.
-
- Note: The conditions in UAX#9 clause L4 must be tested by the
- caller. */
-/* FIXME: exceedingly temporary! Should consult the Unicode database
- of character properties. */
+/* Return the mirrored character of C, if it has one. If C has no
+ mirrored counterpart, return C.
+ Note: The conditions in UAX#9 clause L4 regarding the surrounding
+ context must be tested by the caller. */
int
bidi_mirror_char (int c)
{
- static const char mirrored_pairs[] = "()<>[]{}";
- const char *p = c > 0 && c < 128 ? strchr (mirrored_pairs, c) : NULL;
+ Lisp_Object val;
+
+ if (c == BIDI_EOB)
+ return c;
+ if (c < 0 || c > MAX_CHAR)
+ abort ();
- if (p)
+ val = CHAR_TABLE_REF (bidi_mirror_table, c);
+ if (INTEGERP (val))
{
- size_t i = p - mirrored_pairs;
+ int v = XINT (val);
- return mirrored_pairs [(i ^ 1)];
+ if (v < 0 || v > MAX_CHAR)
+ abort ();
+
+ return v;
}
+
return c;
}
+/* Determine the start-of-run (sor) directional type given the two
+ embedding levels on either side of the run boundary. Also, update
+ the saved info about previously seen characters, since that info is
+ generally valid for a single level run. */
+static inline void
+bidi_set_sor_type (struct bidi_it *bidi_it, int level_before, int level_after)
+{
+ int higher_level = level_before > level_after ? level_before : level_after;
+
+ /* The prev_was_pdf gork is required for when we have several PDFs
+ in a row. In that case, we want to compute the sor type for the
+ next level run only once: when we see the first PDF. That's
+ because the sor type depends only on the higher of the two levels
+ that we find on the two sides of the level boundary (see UAX#9,
+ clause X10), and so we don't need to know the final embedding
+ level to which we descend after processing all the PDFs. */
+ if (!bidi_it->prev_was_pdf || level_before < level_after)
+ /* FIXME: should the default sor direction be user selectable? */
+ bidi_it->sor = (higher_level & 1) != 0 ? R2L : L2R;
+ if (level_before > level_after)
+ bidi_it->prev_was_pdf = 1;
+
+ bidi_it->prev.type = UNKNOWN_BT;
+ bidi_it->last_strong.type = bidi_it->last_strong.type_after_w1 =
+ bidi_it->last_strong.orig_type = UNKNOWN_BT;
+ bidi_it->prev_for_neutral.type = bidi_it->sor == R2L ? STRONG_R : STRONG_L;
+ bidi_it->prev_for_neutral.charpos = bidi_it->charpos;
+ bidi_it->prev_for_neutral.bytepos = bidi_it->bytepos;
+ bidi_it->next_for_neutral.type = bidi_it->next_for_neutral.type_after_w1 =
+ bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
+ bidi_it->ignore_bn_limit = -1; /* meaning it's unknown */
+}
+
+/* Push the current embedding level and override status; reset the
+ current level to LEVEL and the current override status to OVERRIDE. */
+static inline void
+bidi_push_embedding_level (struct bidi_it *bidi_it,
+ int level, bidi_dir_t override)
+{
+ bidi_it->stack_idx++;
+ xassert (bidi_it->stack_idx < BIDI_MAXLEVEL);
+ bidi_it->level_stack[bidi_it->stack_idx].level = level;
+ bidi_it->level_stack[bidi_it->stack_idx].override = override;
+}
+
+/* Pop the embedding level and directional override status from the
+ stack, and return the new level. */
+static inline int
+bidi_pop_embedding_level (struct bidi_it *bidi_it)
+{
+ /* UAX#9 says to ignore invalid PDFs. */
+ if (bidi_it->stack_idx > 0)
+ bidi_it->stack_idx--;
+ return bidi_it->level_stack[bidi_it->stack_idx].level;
+}
+
+/* Record in SAVED_INFO the information about the current character. */
+static inline void
+bidi_remember_char (struct bidi_saved_info *saved_info,
+ struct bidi_it *bidi_it)
+{
+ saved_info->charpos = bidi_it->charpos;
+ saved_info->bytepos = bidi_it->bytepos;
+ saved_info->type = bidi_it->type;
+ bidi_check_type (bidi_it->type);
+ saved_info->type_after_w1 = bidi_it->type_after_w1;
+ bidi_check_type (bidi_it->type_after_w1);
+ saved_info->orig_type = bidi_it->orig_type;
+ bidi_check_type (bidi_it->orig_type);
+}
+
/* Copy the bidi iterator from FROM to TO. To save cycles, this only
copies the part of the level stack that is actually in use. */
static inline void
int i;
/* Copy everything except the level stack and beyond. */
- memcpy (to, from, ((size_t)&((struct bidi_it *)0)->level_stack[0]));
+ memcpy (to, from, offsetof (struct bidi_it, level_stack[0]));
/* Copy the active part of the level stack. */
to->level_stack[0] = from->level_stack[0]; /* level zero is always in use */
to->level_stack[i] = from->level_stack[i];
}
-/* Caching the bidi iterator states. */
-
-static struct bidi_it bidi_cache[1000]; /* FIXME: make this dynamically allocated! */
-static int bidi_cache_idx;
-static int bidi_cache_last_idx;
-
+\f
+/***********************************************************************
+ Caching the bidi iterator states
+ ***********************************************************************/
+
+#define BIDI_CACHE_CHUNK 200
+static struct bidi_it *bidi_cache;
+static EMACS_INT bidi_cache_size = 0;
+enum { elsz = sizeof (struct bidi_it) };
+static EMACS_INT bidi_cache_idx; /* next unused cache slot */
+static EMACS_INT bidi_cache_last_idx; /* slot of last cache hit */
+static EMACS_INT bidi_cache_start = 0; /* start of cache for this
+ "stack" level */
+
+/* Reset the cache state to the empty state. We only reset the part
+ of the cache relevant to iteration of the current object. Previous
+ objects, which are pushed on the display iterator's stack, are left
+ intact. This is called when the cached information is no more
+ useful for the current iteration, e.g. when we were reseated to a
+ new position on the same object. */
static inline void
bidi_cache_reset (void)
{
- bidi_cache_idx = 0;
+ bidi_cache_idx = bidi_cache_start;
bidi_cache_last_idx = -1;
}
+/* Shrink the cache to its minimal size. Called when we init the bidi
+ iterator for reordering a buffer or a string that does not come
+ from display properties, because that means all the previously
+ cached info is of no further use. */
static inline void
-bidi_cache_fetch_state (int idx, struct bidi_it *bidi_it)
+bidi_cache_shrink (void)
+{
+ if (bidi_cache_size > BIDI_CACHE_CHUNK)
+ {
+ bidi_cache_size = BIDI_CACHE_CHUNK;
+ bidi_cache =
+ (struct bidi_it *) xrealloc (bidi_cache, bidi_cache_size * elsz);
+ }
+ bidi_cache_reset ();
+}
+
+static inline void
+bidi_cache_fetch_state (EMACS_INT idx, struct bidi_it *bidi_it)
{
int current_scan_dir = bidi_it->scan_dir;
- if (idx < 0 || idx >= bidi_cache_idx)
+ if (idx < bidi_cache_start || idx >= bidi_cache_idx)
abort ();
bidi_copy_it (bidi_it, &bidi_cache[idx]);
level less or equal to LEVEL. if LEVEL is -1, disregard the
resolved levels in cached states. DIR, if non-zero, means search
in that direction from the last cache hit. */
-static inline int
-bidi_cache_search (int charpos, int level, int dir)
+static inline EMACS_INT
+bidi_cache_search (EMACS_INT charpos, int level, int dir)
{
- int i, i_start;
+ EMACS_INT i, i_start;
- if (bidi_cache_idx)
+ if (bidi_cache_idx > bidi_cache_start)
{
+ if (bidi_cache_last_idx == -1)
+ bidi_cache_last_idx = bidi_cache_idx - 1;
if (charpos < bidi_cache[bidi_cache_last_idx].charpos)
- dir = -1;
- else if (charpos > bidi_cache[bidi_cache_last_idx].charpos)
- dir = 1;
- if (dir)
+ {
+ dir = -1;
+ i_start = bidi_cache_last_idx - 1;
+ }
+ else if (charpos > (bidi_cache[bidi_cache_last_idx].charpos
+ + bidi_cache[bidi_cache_last_idx].nchars - 1))
+ {
+ dir = 1;
+ i_start = bidi_cache_last_idx + 1;
+ }
+ else if (dir)
i_start = bidi_cache_last_idx;
else
{
if (dir < 0)
{
/* Linear search for now; FIXME! */
- for (i = i_start; i >= 0; i--)
- if (bidi_cache[i].charpos == charpos
+ for (i = i_start; i >= bidi_cache_start; i--)
+ if (bidi_cache[i].charpos <= charpos
+ && charpos < bidi_cache[i].charpos + bidi_cache[i].nchars
&& (level == -1 || bidi_cache[i].resolved_level <= level))
return i;
}
else
{
for (i = i_start; i < bidi_cache_idx; i++)
- if (bidi_cache[i].charpos == charpos
+ if (bidi_cache[i].charpos <= charpos
+ && charpos < bidi_cache[i].charpos + bidi_cache[i].nchars
&& (level == -1 || bidi_cache[i].resolved_level <= level))
return i;
}
/* Find a cached state where the resolved level changes to a value
that is lower than LEVEL, and return its cache slot index. DIR is
the direction to search, starting with the last used cache slot.
- BEFORE, if non-zero, means return the index of the slot that is
- ``before'' the level change in the search direction. That is,
+ If DIR is zero, we search backwards from the last occupied cache
+ slot. BEFORE, if non-zero, means return the index of the slot that
+ is ``before'' the level change in the search direction. That is,
given the cached levels like this:
1122333442211
C, searching backwards (DIR = -1) for LEVEL = 2 will return the
index of slot B or A, depending whether BEFORE is, respectively,
non-zero or zero. */
-static int
+static EMACS_INT
bidi_cache_find_level_change (int level, int dir, int before)
{
if (bidi_cache_idx)
{
- int i = dir ? bidi_cache_last_idx : bidi_cache_idx - 1;
+ EMACS_INT i = dir ? bidi_cache_last_idx : bidi_cache_idx - 1;
int incr = before ? 1 : 0;
+ xassert (!dir || bidi_cache_last_idx >= 0);
+
if (!dir)
dir = -1;
else if (!incr)
if (dir < 0)
{
- while (i >= incr)
+ while (i >= bidi_cache_start + incr)
{
if (bidi_cache[i - incr].resolved_level >= 0
&& bidi_cache[i - incr].resolved_level < level)
return -1;
}
+static inline void
+bidi_cache_ensure_space (EMACS_INT idx)
+{
+ /* Enlarge the cache as needed. */
+ if (idx >= bidi_cache_size)
+ {
+ while (idx >= bidi_cache_size)
+ bidi_cache_size += BIDI_CACHE_CHUNK;
+ bidi_cache =
+ (struct bidi_it *) xrealloc (bidi_cache, bidi_cache_size * elsz);
+ }
+}
+
static inline void
bidi_cache_iterator_state (struct bidi_it *bidi_it, int resolved)
{
- int idx;
+ EMACS_INT idx;
/* We should never cache on backward scans. */
if (bidi_it->scan_dir == -1)
if (idx < 0)
{
idx = bidi_cache_idx;
- /* Don't overrun the cache limit. */
- if (idx > sizeof (bidi_cache) / sizeof (bidi_cache[0]) - 1)
- abort ();
- /* Don't violate cache integrity: character positions should
- correspond to cache positions 1:1. */
- if (idx > 0 && bidi_it->charpos != bidi_cache[idx - 1].charpos + 1)
+ bidi_cache_ensure_space (idx);
+ /* Character positions should correspond to cache positions 1:1.
+ If we are outside the range of cached positions, the cache is
+ useless and must be reset. */
+ if (idx > bidi_cache_start &&
+ (bidi_it->charpos > (bidi_cache[idx - 1].charpos
+ + bidi_cache[idx - 1].nchars)
+ || bidi_it->charpos < bidi_cache[bidi_cache_start].charpos))
+ {
+ bidi_cache_reset ();
+ idx = bidi_cache_start;
+ }
+ if (bidi_it->nchars <= 0)
abort ();
bidi_copy_it (&bidi_cache[idx], bidi_it);
if (!resolved)
bidi_cache[idx].resolved_level = -1;
- bidi_cache[idx].new_paragraph = 0;
}
else
{
}
static inline bidi_type_t
-bidi_cache_find (int charpos, int level, struct bidi_it *bidi_it)
+bidi_cache_find (EMACS_INT charpos, int level, struct bidi_it *bidi_it)
{
- int i = bidi_cache_search (charpos, level, bidi_it->scan_dir);
+ EMACS_INT i = bidi_cache_search (charpos, level, bidi_it->scan_dir);
- if (i >= 0)
+ if (i >= bidi_cache_start)
{
bidi_dir_t current_scan_dir = bidi_it->scan_dir;
static inline int
bidi_peek_at_next_level (struct bidi_it *bidi_it)
{
- if (bidi_cache_idx == 0 || bidi_cache_last_idx == -1)
+ if (bidi_cache_idx == bidi_cache_start || bidi_cache_last_idx == -1)
abort ();
return bidi_cache[bidi_cache_last_idx + bidi_it->scan_dir].resolved_level;
}
-/* Check if buffer position CHARPOS/BYTEPOS is the end of a paragraph.
- Value is the non-negative length of the paragraph separator
- following the buffer position, -1 if position is at the beginning
- of a new paragraph, or -2 if position is neither at beginning nor
- at end of a paragraph. */
-EMACS_INT
-bidi_at_paragraph_end (EMACS_INT charpos, EMACS_INT bytepos)
+\f
+/***********************************************************************
+ Pushing and popping the bidi iterator state
+ ***********************************************************************/
+/* 5-slot stack for saving the start of the previous level of the
+ cache. xdisp.c maintains a 5-slot stack for its iterator state,
+ and we need the same size of our stack. */
+static EMACS_INT bidi_cache_start_stack[IT_STACK_SIZE];
+static int bidi_cache_sp;
+
+/* Push the bidi iterator state in preparation for reordering a
+ different object, e.g. display string found at certain buffer
+ position. Pushing the bidi iterator boils down to saving its
+ entire state on the cache and starting a new cache "stacked" on top
+ of the current cache. */
+void
+bidi_push_it (struct bidi_it *bidi_it)
{
- Lisp_Object sep_re = Fbuffer_local_value (Qparagraph_separate,
- Fcurrent_buffer ());
- Lisp_Object start_re = Fbuffer_local_value (Qparagraph_start,
- Fcurrent_buffer ());
- EMACS_INT val;
+ /* Save the current iterator state in its entirety after the last
+ used cache slot. */
+ bidi_cache_ensure_space (bidi_cache_idx);
+ memcpy (&bidi_cache[bidi_cache_idx++], bidi_it, sizeof (struct bidi_it));
- if (!STRINGP (sep_re))
- sep_re = fallback_paragraph_separate_re;
- if (!STRINGP (start_re))
- start_re = fallback_paragraph_start_re;
+ /* Push the current cache start onto the stack. */
+ xassert (bidi_cache_sp < IT_STACK_SIZE);
+ bidi_cache_start_stack[bidi_cache_sp++] = bidi_cache_start;
- val = fast_looking_at (sep_re, charpos, bytepos, ZV, ZV_BYTE, Qnil);
- if (val < 0)
+ /* Start a new level of cache, and make it empty. */
+ bidi_cache_start = bidi_cache_idx;
+ bidi_cache_last_idx = -1;
+}
+
+/* Restore the iterator state saved by bidi_push_it and return the
+ cache to the corresponding state. */
+void
+bidi_pop_it (struct bidi_it *bidi_it)
+{
+ if (bidi_cache_start <= 0)
+ abort ();
+
+ /* Reset the next free cache slot index to what it was before the
+ call to bidi_push_it. */
+ bidi_cache_idx = bidi_cache_start - 1;
+
+ /* Restore the bidi iterator state saved in the cache. */
+ memcpy (bidi_it, &bidi_cache[bidi_cache_idx], sizeof (struct bidi_it));
+
+ /* Pop the previous cache start from the stack. */
+ if (bidi_cache_sp <= 0)
+ abort ();
+ bidi_cache_start = bidi_cache_start_stack[--bidi_cache_sp];
+
+ /* Invalidate the last-used cache slot data. */
+ bidi_cache_last_idx = -1;
+}
+
+/* Stash away a copy of the cache and its control variables. */
+void *
+bidi_shelve_cache (void)
+{
+ unsigned char *databuf;
+
+ if (bidi_cache_idx == 0)
+ return NULL;
+
+ databuf = xmalloc (sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it)
+ + sizeof (bidi_cache_start_stack)
+ + sizeof (bidi_cache_sp) + sizeof (bidi_cache_start)
+ + sizeof (bidi_cache_last_idx));
+ memcpy (databuf, &bidi_cache_idx, sizeof (bidi_cache_idx));
+ memcpy (databuf + sizeof (bidi_cache_idx),
+ bidi_cache, bidi_cache_idx * sizeof (struct bidi_it));
+ memcpy (databuf + sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it),
+ bidi_cache_start_stack, sizeof (bidi_cache_start_stack));
+ memcpy (databuf + sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it)
+ + sizeof (bidi_cache_start_stack),
+ &bidi_cache_sp, sizeof (bidi_cache_sp));
+ memcpy (databuf + sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it)
+ + sizeof (bidi_cache_start_stack) + sizeof (bidi_cache_sp),
+ &bidi_cache_start, sizeof (bidi_cache_start));
+ memcpy (databuf + sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it)
+ + sizeof (bidi_cache_start_stack) + sizeof (bidi_cache_sp)
+ + sizeof (bidi_cache_start),
+ &bidi_cache_last_idx, sizeof (bidi_cache_last_idx));
+
+ return databuf;
+}
+
+/* Restore the cache state from a copy stashed away by bidi_shelve_cache. */
+void
+bidi_unshelve_cache (void *databuf)
+{
+ unsigned char *p = databuf;
+
+ if (!p)
{
- if (fast_looking_at (start_re, charpos, bytepos, ZV, ZV_BYTE, Qnil) >= 0)
- val = -1;
- else
- val = -2;
+ /* A NULL pointer means an empty cache. */
+ bidi_cache_start = 0;
+ bidi_cache_sp = 0;
+ bidi_cache_reset ();
}
+ else
+ {
+ memcpy (&bidi_cache_idx, p, sizeof (bidi_cache_idx));
+ bidi_cache_ensure_space (bidi_cache_idx);
+ memcpy (bidi_cache, p + sizeof (bidi_cache_idx),
+ bidi_cache_idx * sizeof (struct bidi_it));
+ memcpy (bidi_cache_start_stack,
+ p + sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it),
+ sizeof (bidi_cache_start_stack));
+ memcpy (&bidi_cache_sp,
+ p + sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it)
+ + sizeof (bidi_cache_start_stack),
+ sizeof (bidi_cache_sp));
+ memcpy (&bidi_cache_start,
+ p + sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it)
+ + sizeof (bidi_cache_start_stack) + sizeof (bidi_cache_sp),
+ sizeof (bidi_cache_start));
+ memcpy (&bidi_cache_last_idx,
+ p + sizeof (bidi_cache_idx)
+ + bidi_cache_idx * sizeof (struct bidi_it)
+ + sizeof (bidi_cache_start_stack) + sizeof (bidi_cache_sp)
+ + sizeof (bidi_cache_start),
+ sizeof (bidi_cache_last_idx));
+
+ xfree (p);
+ }
+}
- return val;
+\f
+/***********************************************************************
+ Initialization
+ ***********************************************************************/
+static void
+bidi_initialize (void)
+{
+
+#include "biditype.h"
+#include "bidimirror.h"
+
+ int i;
+
+ bidi_type_table = Fmake_char_table (Qnil, make_number (STRONG_L));
+ staticpro (&bidi_type_table);
+
+ for (i = 0; i < sizeof bidi_type / sizeof bidi_type[0]; i++)
+ char_table_set_range (bidi_type_table, bidi_type[i].from, bidi_type[i].to,
+ make_number (bidi_type[i].type));
+
+ bidi_mirror_table = Fmake_char_table (Qnil, Qnil);
+ staticpro (&bidi_mirror_table);
+
+ for (i = 0; i < sizeof bidi_mirror / sizeof bidi_mirror[0]; i++)
+ char_table_set (bidi_mirror_table, bidi_mirror[i].from,
+ make_number (bidi_mirror[i].to));
+
+ Qparagraph_start = intern ("paragraph-start");
+ staticpro (&Qparagraph_start);
+ paragraph_start_re = Fsymbol_value (Qparagraph_start);
+ if (!STRINGP (paragraph_start_re))
+ paragraph_start_re = build_string ("\f\\|[ \t]*$");
+ staticpro (¶graph_start_re);
+ Qparagraph_separate = intern ("paragraph-separate");
+ staticpro (&Qparagraph_separate);
+ paragraph_separate_re = Fsymbol_value (Qparagraph_separate);
+ if (!STRINGP (paragraph_separate_re))
+ paragraph_separate_re = build_string ("[ \t\f]*$");
+ staticpro (¶graph_separate_re);
+
+ bidi_cache_sp = 0;
+
+ bidi_initialized = 1;
}
-/* Determine the start-of-run (sor) directional type given the two
- embedding levels on either side of the run boundary. Also, update
- the saved info about previously seen characters, since that info is
- generally valid for a single level run. */
+/* Do whatever UAX#9 clause X8 says should be done at paragraph's
+ end. */
static inline void
-bidi_set_sor_type (struct bidi_it *bidi_it, int level_before, int level_after)
+bidi_set_paragraph_end (struct bidi_it *bidi_it)
{
- int higher_level = level_before > level_after ? level_before : level_after;
-
- /* The prev_was_pdf gork is required for when we have several PDFs
- in a row. In that case, we want to compute the sor type for the
- next level run only once: when we see the first PDF. That's
- because the sor type depends only on the higher of the two levels
- that we find on the two sides of the level boundary (see UAX#9,
- clause X10), and so we don't need to know the final embedding
- level to which we descend after processing all the PDFs. */
- if (!bidi_it->prev_was_pdf || level_before < level_after)
- /* FIXME: should the default sor direction be user selectable? */
- bidi_it->sor = (higher_level & 1) != 0 ? R2L : L2R;
- if (level_before > level_after)
- bidi_it->prev_was_pdf = 1;
+ bidi_it->invalid_levels = 0;
+ bidi_it->invalid_rl_levels = -1;
+ bidi_it->stack_idx = 0;
+ bidi_it->resolved_level = bidi_it->level_stack[0].level;
+}
- bidi_it->prev.type = UNKNOWN_BT;
+/* Initialize the bidi iterator from buffer/string position CHARPOS. */
+void
+bidi_init_it (EMACS_INT charpos, EMACS_INT bytepos, int frame_window_p,
+ struct bidi_it *bidi_it)
+{
+ if (! bidi_initialized)
+ bidi_initialize ();
+ if (charpos >= 0)
+ bidi_it->charpos = charpos;
+ if (bytepos >= 0)
+ bidi_it->bytepos = bytepos;
+ bidi_it->frame_window_p = frame_window_p;
+ bidi_it->nchars = -1; /* to be computed in bidi_resolve_explicit_1 */
+ bidi_it->first_elt = 1;
+ bidi_set_paragraph_end (bidi_it);
+ bidi_it->new_paragraph = 1;
+ bidi_it->separator_limit = -1;
+ bidi_it->type = NEUTRAL_B;
+ bidi_it->type_after_w1 = NEUTRAL_B;
+ bidi_it->orig_type = NEUTRAL_B;
+ bidi_it->prev_was_pdf = 0;
+ bidi_it->prev.type = bidi_it->prev.type_after_w1 =
+ bidi_it->prev.orig_type = UNKNOWN_BT;
bidi_it->last_strong.type = bidi_it->last_strong.type_after_w1 =
bidi_it->last_strong.orig_type = UNKNOWN_BT;
- bidi_it->prev_for_neutral.type = bidi_it->sor == R2L ? STRONG_R : STRONG_L;
- bidi_it->prev_for_neutral.charpos = bidi_it->charpos;
- bidi_it->prev_for_neutral.bytepos = bidi_it->bytepos;
- bidi_it->next_for_neutral.type = bidi_it->next_for_neutral.type_after_w1 =
+ bidi_it->next_for_neutral.charpos = -1;
+ bidi_it->next_for_neutral.type =
+ bidi_it->next_for_neutral.type_after_w1 =
bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
- bidi_it->ignore_bn_limit = 0; /* meaning it's unknown */
+ bidi_it->prev_for_neutral.charpos = -1;
+ bidi_it->prev_for_neutral.type =
+ bidi_it->prev_for_neutral.type_after_w1 =
+ bidi_it->prev_for_neutral.orig_type = UNKNOWN_BT;
+ bidi_it->sor = L2R; /* FIXME: should it be user-selectable? */
+ bidi_it->disp_pos = -1; /* invalid/unknown */
+ /* We can only shrink the cache if we are at the bottom level of its
+ "stack". */
+ if (bidi_cache_start == 0)
+ bidi_cache_shrink ();
+ else
+ bidi_cache_reset ();
}
+/* Perform initializations for reordering a new line of bidi text. */
static void
bidi_line_init (struct bidi_it *bidi_it)
{
bidi_cache_reset ();
}
+\f
+/***********************************************************************
+ Fetching characters
+ ***********************************************************************/
+
+/* Count bytes in string S between BEG/BEGBYTE and END. BEG and END
+ are zero-based character positions in S, BEGBYTE is byte position
+ corresponding to BEG. UNIBYTE, if non-zero, means S is a unibyte
+ string. */
+static inline EMACS_INT
+bidi_count_bytes (const unsigned char *s, const EMACS_INT beg,
+ const EMACS_INT begbyte, const EMACS_INT end, int unibyte)
+{
+ EMACS_INT pos = beg;
+ const unsigned char *p = s + begbyte, *start = p;
+
+ if (unibyte)
+ p = s + end;
+ else
+ {
+ if (!CHAR_HEAD_P (*p))
+ abort ();
+
+ while (pos < end)
+ {
+ p += BYTES_BY_CHAR_HEAD (*p);
+ pos++;
+ }
+ }
+
+ return p - start;
+}
+
+/* Fetch and returns the character at byte position BYTEPOS. If S is
+ non-NULL, fetch the character from string S; otherwise fetch the
+ character from the current buffer. UNIBYTE non-zero means S is a
+ unibyte string. */
+static inline int
+bidi_char_at_pos (EMACS_INT bytepos, const unsigned char *s, int unibyte)
+{
+ if (s)
+ {
+ if (unibyte)
+ return s[bytepos];
+ else
+ return STRING_CHAR (s + bytepos);
+ }
+ else
+ return FETCH_MULTIBYTE_CHAR (bytepos);
+}
+
+/* Fetch and return the character at BYTEPOS/CHARPOS. If that
+ character is covered by a display string, treat the entire run of
+ covered characters as a single character u+FFFC, and return their
+ combined length in CH_LEN and NCHARS. DISP_POS specifies the
+ character position of the next display string, or -1 if not yet
+ computed. When the next character is at or beyond that position,
+ the function updates DISP_POS with the position of the next display
+ string. STRING->s is the C string to iterate, or NULL if iterating
+ over a buffer or a Lisp string; in the latter case, STRING->lstring
+ is the Lisp string. */
+static inline int
+bidi_fetch_char (EMACS_INT bytepos, EMACS_INT charpos, EMACS_INT *disp_pos,
+ struct bidi_string_data *string,
+ int frame_window_p, EMACS_INT *ch_len, EMACS_INT *nchars)
+{
+ int ch;
+ EMACS_INT endpos =
+ (string->s || STRINGP (string->lstring)) ? string->schars : ZV;
+ struct text_pos pos;
+
+ /* If we got past the last known position of display string, compute
+ the position of the next one. That position could be at CHARPOS. */
+ if (charpos < endpos && charpos > *disp_pos)
+ {
+ SET_TEXT_POS (pos, charpos, bytepos);
+ *disp_pos = compute_display_string_pos (&pos, string, frame_window_p);
+ }
+
+ /* Fetch the character at BYTEPOS. */
+ if (charpos >= endpos)
+ {
+ ch = BIDI_EOB;
+ *ch_len = 1;
+ *nchars = 1;
+ *disp_pos = endpos;
+ }
+ else if (charpos >= *disp_pos)
+ {
+ EMACS_INT disp_end_pos;
+
+ /* We don't expect to find ourselves in the middle of a display
+ property. Hopefully, it will never be needed. */
+ if (charpos > *disp_pos)
+ abort ();
+ /* Return the Unicode Object Replacement Character to represent
+ the entire run of characters covered by the display string. */
+ ch = 0xFFFC;
+ disp_end_pos = compute_display_string_end (*disp_pos, string);
+ *nchars = disp_end_pos - *disp_pos;
+ if (*nchars <= 0)
+ abort ();
+ if (string->s)
+ *ch_len = bidi_count_bytes (string->s, *disp_pos, bytepos,
+ disp_end_pos, string->unibyte);
+ else if (STRINGP (string->lstring))
+ *ch_len = bidi_count_bytes (SDATA (string->lstring), *disp_pos,
+ bytepos, disp_end_pos, string->unibyte);
+ else
+ *ch_len = CHAR_TO_BYTE (disp_end_pos) - bytepos;
+ }
+ else
+ {
+ if (string->s)
+ {
+ int len;
+
+ if (!string->unibyte)
+ {
+ ch = STRING_CHAR_AND_LENGTH (string->s + bytepos, len);
+ *ch_len = len;
+ }
+ else
+ {
+ ch = UNIBYTE_TO_CHAR (string->s[bytepos]);
+ *ch_len = 1;
+ }
+ }
+ else if (STRINGP (string->lstring))
+ {
+ int len;
+
+ if (!string->unibyte)
+ {
+ ch = STRING_CHAR_AND_LENGTH (SDATA (string->lstring) + bytepos,
+ len);
+ *ch_len = len;
+ }
+ else
+ {
+ ch = UNIBYTE_TO_CHAR (SREF (string->lstring, bytepos));
+ *ch_len = 1;
+ }
+ }
+ else
+ {
+ ch = FETCH_MULTIBYTE_CHAR (bytepos);
+ *ch_len = CHAR_BYTES (ch);
+ }
+ *nchars = 1;
+ }
+
+ /* If we just entered a run of characters covered by a display
+ string, compute the position of the next display string. */
+ if (charpos + *nchars <= endpos && charpos + *nchars > *disp_pos)
+ {
+ SET_TEXT_POS (pos, charpos + *nchars, bytepos + *ch_len);
+ *disp_pos = compute_display_string_pos (&pos, string, frame_window_p);
+ }
+
+ return ch;
+}
+
+\f
+/***********************************************************************
+ Determining paragraph direction
+ ***********************************************************************/
+
+/* Check if buffer position CHARPOS/BYTEPOS is the end of a paragraph.
+ Value is the non-negative length of the paragraph separator
+ following the buffer position, -1 if position is at the beginning
+ of a new paragraph, or -2 if position is neither at beginning nor
+ at end of a paragraph. */
+static EMACS_INT
+bidi_at_paragraph_end (EMACS_INT charpos, EMACS_INT bytepos)
+{
+ Lisp_Object sep_re;
+ Lisp_Object start_re;
+ EMACS_INT val;
+
+ sep_re = paragraph_separate_re;
+ start_re = paragraph_start_re;
+
+ val = fast_looking_at (sep_re, charpos, bytepos, ZV, ZV_BYTE, Qnil);
+ if (val < 0)
+ {
+ if (fast_looking_at (start_re, charpos, bytepos, ZV, ZV_BYTE, Qnil) >= 0)
+ val = -1;
+ else
+ val = -2;
+ }
+
+ return val;
+}
+
/* Find the beginning of this paragraph by looking back in the buffer.
Value is the byte position of the paragraph's beginning. */
static EMACS_INT
bidi_find_paragraph_start (EMACS_INT pos, EMACS_INT pos_byte)
{
- Lisp_Object re = Fbuffer_local_value (Qparagraph_start, Fcurrent_buffer ());
+ Lisp_Object re = paragraph_start_re;
EMACS_INT limit = ZV, limit_byte = ZV_BYTE;
- if (!STRINGP (re))
- re = fallback_paragraph_start_re;
while (pos_byte > BEGV_BYTE
&& fast_looking_at (re, pos, pos_byte, limit, limit_byte, Qnil) < 0)
{
+ /* FIXME: What if the paragraph beginning is covered by a
+ display string? And what if a display string covering some
+ of the text over which we scan back includes
+ paragraph_start_re? */
pos = find_next_newline_no_quit (pos - 1, -1);
pos_byte = CHAR_TO_BYTE (pos);
}
return pos_byte;
}
-/* Determine the direction, a.k.a. base embedding level, of the
+/* Determine the base direction, a.k.a. base embedding level, of the
paragraph we are about to iterate through. If DIR is either L2R or
R2L, just use that. Otherwise, determine the paragraph direction
- from the first strong character of the paragraph.
-
- Note that this gives the paragraph separator the same direction as
- the preceding paragraph, even though Emacs generally views the
- separartor as not belonging to any paragraph. */
+ from the first strong directional character of the paragraph.
+
+ NO_DEFAULT_P non-zero means don't default to L2R if the paragraph
+ has no strong directional characters and both DIR and
+ bidi_it->paragraph_dir are NEUTRAL_DIR. In that case, search back
+ in the buffer until a paragraph is found with a strong character,
+ or until hitting BEGV. In the latter case, fall back to L2R. This
+ flag is used in current-bidi-paragraph-direction.
+
+ Note that this function gives the paragraph separator the same
+ direction as the preceding paragraph, even though Emacs generally
+ views the separartor as not belonging to any paragraph. */
void
-bidi_paragraph_init (bidi_dir_t dir, struct bidi_it *bidi_it)
+bidi_paragraph_init (bidi_dir_t dir, struct bidi_it *bidi_it, int no_default_p)
{
EMACS_INT bytepos = bidi_it->bytepos;
+ int string_p = bidi_it->string.s != NULL || STRINGP (bidi_it->string.lstring);
+ EMACS_INT pstartbyte;
+ /* Note that begbyte is a byte position, while end is a character
+ position. Yes, this is ugly, but we are trying to avoid costly
+ calls to BYTE_TO_CHAR and its ilk. */
+ EMACS_INT begbyte = string_p ? 0 : BEGV_BYTE;
+ EMACS_INT end = string_p ? bidi_it->string.schars : ZV;
/* Special case for an empty buffer. */
- if (bytepos == BEGV_BYTE && bytepos == ZV_BYTE)
+ if (bytepos == begbyte && bidi_it->charpos == end)
dir = L2R;
/* We should never be called at EOB or before BEGV. */
- else if (bytepos >= ZV_BYTE || bytepos < BEGV_BYTE)
+ else if (bidi_it->charpos >= end || bytepos < begbyte)
abort ();
if (dir == L2R)
}
else if (dir == NEUTRAL_DIR) /* P2 */
{
- int ch, ch_len;
- EMACS_INT pos;
+ int ch;
+ EMACS_INT ch_len, nchars;
+ EMACS_INT pos, disp_pos = -1;
bidi_type_t type;
- EMACS_INT sep_len;
+ const unsigned char *s;
+
+ if (!bidi_initialized)
+ bidi_initialize ();
/* If we are inside a paragraph separator, we are just waiting
for the separator to be exhausted; use the previous paragraph
we are potentially in a new paragraph that doesn't yet
exist. */
pos = bidi_it->charpos;
- if (bytepos > BEGV_BYTE && FETCH_CHAR (bytepos) == '\n')
+ s = STRINGP (bidi_it->string.lstring) ?
+ SDATA (bidi_it->string.lstring) : bidi_it->string.s;
+ if (bytepos > begbyte
+ && bidi_char_at_pos (bytepos, s, bidi_it->string.unibyte) == '\n')
{
bytepos++;
pos++;
/* We are either at the beginning of a paragraph or in the
middle of it. Find where this paragraph starts. */
- bytepos = bidi_find_paragraph_start (pos, bytepos);
-
- /* We should always be at the beginning of a new line at this
- point. */
- if (!(bytepos == BEGV_BYTE || FETCH_CHAR (bytepos - 1) == '\n'))
- abort ();
-
+ if (string_p)
+ {
+ /* We don't support changes of paragraph direction inside a
+ string. It is treated as a single paragraph. */
+ pstartbyte = 0;
+ }
+ else
+ pstartbyte = bidi_find_paragraph_start (pos, bytepos);
bidi_it->separator_limit = -1;
bidi_it->new_paragraph = 0;
- ch = FETCH_CHAR (bytepos);
- ch_len = CHAR_BYTES (ch);
- pos = BYTE_TO_CHAR (bytepos);
- type = bidi_get_type (ch, NEUTRAL_DIR);
-
- for (pos++, bytepos += ch_len;
- /* NOTE: UAX#9 says to search only for L, AL, or R types of
- characters, and ignore RLE, RLO, LRE, and LRO. However,
- I'm not sure it makes sense to omit those 4; should try
- with and without that to see the effect. */
- (bidi_get_category (type) != STRONG)
- || (bidi_ignore_explicit_marks_for_paragraph_level
- && (type == RLE || type == RLO
- || type == LRE || type == LRO));
- type = bidi_get_type (ch, NEUTRAL_DIR))
- {
- if (type == NEUTRAL_B && bidi_at_paragraph_end (pos, bytepos) >= -1)
- break;
- if (bytepos >= ZV_BYTE)
- {
- /* Pretend there's a paragraph separator at end of buffer. */
- type = NEUTRAL_B;
+
+ /* The following loop is run more than once only if NO_DEFAULT_P
+ is non-zero, and only if we are iterating on a buffer. */
+ do {
+ bytepos = pstartbyte;
+ if (!string_p)
+ pos = BYTE_TO_CHAR (bytepos);
+ ch = bidi_fetch_char (bytepos, pos, &disp_pos, &bidi_it->string,
+ bidi_it->frame_window_p, &ch_len, &nchars);
+ type = bidi_get_type (ch, NEUTRAL_DIR);
+
+ for (pos += nchars, bytepos += ch_len;
+ /* NOTE: UAX#9 says to search only for L, AL, or R types
+ of characters, and ignore RLE, RLO, LRE, and LRO.
+ However, I'm not sure it makes sense to omit those 4;
+ should try with and without that to see the effect. */
+ (bidi_get_category (type) != STRONG)
+ || (bidi_ignore_explicit_marks_for_paragraph_level
+ && (type == RLE || type == RLO
+ || type == LRE || type == LRO));
+ type = bidi_get_type (ch, NEUTRAL_DIR))
+ {
+ if (pos >= end)
+ {
+ /* Pretend there's a paragraph separator at end of
+ buffer/string. */
+ type = NEUTRAL_B;
+ break;
+ }
+ if (!string_p
+ && type == NEUTRAL_B
+ && bidi_at_paragraph_end (pos, bytepos) >= -1)
break;
- }
- FETCH_CHAR_ADVANCE (ch, pos, bytepos);
- }
- if (type == STRONG_R || type == STRONG_AL) /* P3 */
- bidi_it->paragraph_dir = R2L;
- else if (type == STRONG_L)
- bidi_it->paragraph_dir = L2R;
+ /* Fetch next character and advance to get past it. */
+ ch = bidi_fetch_char (bytepos, pos, &disp_pos, &bidi_it->string,
+ bidi_it->frame_window_p, &ch_len, &nchars);
+ pos += nchars;
+ bytepos += ch_len;
+ }
+ if (type == STRONG_R || type == STRONG_AL) /* P3 */
+ bidi_it->paragraph_dir = R2L;
+ else if (type == STRONG_L)
+ bidi_it->paragraph_dir = L2R;
+ if (!string_p
+ && no_default_p && bidi_it->paragraph_dir == NEUTRAL_DIR)
+ {
+ /* If this paragraph is at BEGV, default to L2R. */
+ if (pstartbyte == BEGV_BYTE)
+ bidi_it->paragraph_dir = L2R; /* P3 and HL1 */
+ else
+ {
+ EMACS_INT prevpbyte = pstartbyte;
+ EMACS_INT p = BYTE_TO_CHAR (pstartbyte), pbyte = pstartbyte;
+
+ /* Find the beginning of the previous paragraph, if any. */
+ while (pbyte > BEGV_BYTE && prevpbyte >= pstartbyte)
+ {
+ /* FXIME: What if p is covered by a display
+ string? See also a FIXME inside
+ bidi_find_paragraph_start. */
+ p--;
+ pbyte = CHAR_TO_BYTE (p);
+ prevpbyte = bidi_find_paragraph_start (p, pbyte);
+ }
+ pstartbyte = prevpbyte;
+ }
+ }
+ } while (!string_p
+ && no_default_p && bidi_it->paragraph_dir == NEUTRAL_DIR);
}
else
abort ();
/* Contrary to UAX#9 clause P3, we only default the paragraph
direction to L2R if we have no previous usable paragraph
- direction. */
- if (bidi_it->paragraph_dir == NEUTRAL_DIR)
- bidi_it->paragraph_dir = L2R; /* P3 and ``higher protocols'' */
+ direction. This is allowed by the HL1 clause. */
+ if (bidi_it->paragraph_dir != L2R && bidi_it->paragraph_dir != R2L)
+ bidi_it->paragraph_dir = L2R; /* P3 and HL1 ``higher-level protocols'' */
if (bidi_it->paragraph_dir == R2L)
bidi_it->level_stack[0].level = 1;
else
bidi_line_init (bidi_it);
}
-/* Do whatever UAX#9 clause X8 says should be done at paragraph's
- end. */
-static inline void
-bidi_set_paragraph_end (struct bidi_it *bidi_it)
-{
- bidi_it->invalid_levels = 0;
- bidi_it->invalid_rl_levels = -1;
- bidi_it->stack_idx = 0;
- bidi_it->resolved_level = bidi_it->level_stack[0].level;
-}
+\f
+/***********************************************************************
+ Resolving explicit and implicit levels.
+ The rest of this file constitutes the core of the UBA implementation.
+ ***********************************************************************/
-/* Initialize the bidi iterator from buffer position CHARPOS. */
-void
-bidi_init_it (EMACS_INT charpos, EMACS_INT bytepos, struct bidi_it *bidi_it)
-{
- if (! bidi_initialized)
- bidi_initialize ();
- bidi_it->charpos = charpos;
- bidi_it->bytepos = bytepos;
- bidi_it->first_elt = 1;
- bidi_set_paragraph_end (bidi_it);
- bidi_it->new_paragraph = 1;
- bidi_it->separator_limit = -1;
- bidi_it->type = NEUTRAL_B;
- bidi_it->type_after_w1 = UNKNOWN_BT;
- bidi_it->orig_type = UNKNOWN_BT;
- bidi_it->prev_was_pdf = 0;
- bidi_it->prev.type = bidi_it->prev.type_after_w1 = UNKNOWN_BT;
- bidi_it->last_strong.type = bidi_it->last_strong.type_after_w1 =
- bidi_it->last_strong.orig_type = UNKNOWN_BT;
- bidi_it->next_for_neutral.charpos = -1;
- bidi_it->next_for_neutral.type =
- bidi_it->next_for_neutral.type_after_w1 =
- bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
- bidi_it->prev_for_neutral.charpos = -1;
- bidi_it->prev_for_neutral.type =
- bidi_it->prev_for_neutral.type_after_w1 =
- bidi_it->prev_for_neutral.orig_type = UNKNOWN_BT;
- bidi_it->sor = L2R; /* FIXME: should it be user-selectable? */
-}
-
-/* Push the current embedding level and override status; reset the
- current level to LEVEL and the current override status to OVERRIDE. */
-static inline void
-bidi_push_embedding_level (struct bidi_it *bidi_it,
- int level, bidi_dir_t override)
-{
- bidi_it->stack_idx++;
- if (bidi_it->stack_idx >= BIDI_MAXLEVEL)
- abort ();
- bidi_it->level_stack[bidi_it->stack_idx].level = level;
- bidi_it->level_stack[bidi_it->stack_idx].override = override;
-}
-
-/* Pop the embedding level and directional override status from the
- stack, and return the new level. */
static inline int
-bidi_pop_embedding_level (struct bidi_it *bidi_it)
-{
- /* UAX#9 says to ignore invalid PDFs. */
- if (bidi_it->stack_idx > 0)
- bidi_it->stack_idx--;
- return bidi_it->level_stack[bidi_it->stack_idx].level;
-}
-
-/* Record in SAVED_INFO the information about the current character. */
-static inline void
-bidi_remember_char (struct bidi_saved_info *saved_info,
- struct bidi_it *bidi_it)
+bidi_explicit_dir_char (int ch)
{
- saved_info->charpos = bidi_it->charpos;
- saved_info->bytepos = bidi_it->bytepos;
- saved_info->type = bidi_it->type;
- bidi_check_type (bidi_it->type);
- saved_info->type_after_w1 = bidi_it->type_after_w1;
- bidi_check_type (bidi_it->type_after_w1);
- saved_info->orig_type = bidi_it->orig_type;
- bidi_check_type (bidi_it->orig_type);
-}
-
-/* Resolve the type of a neutral character according to the type of
- surrounding strong text and the current embedding level. */
-static inline bidi_type_t
-bidi_resolve_neutral_1 (bidi_type_t prev_type, bidi_type_t next_type, int lev)
-{
- /* N1: European and Arabic numbers are treated as though they were R. */
- if (next_type == WEAK_EN || next_type == WEAK_AN)
- next_type = STRONG_R;
- if (prev_type == WEAK_EN || prev_type == WEAK_AN)
- prev_type = STRONG_R;
-
- if (next_type == prev_type) /* N1 */
- return next_type;
- else if ((lev & 1) == 0) /* N2 */
- return STRONG_L;
- else
- return STRONG_R;
-}
+ bidi_type_t ch_type;
-static inline int
-bidi_explicit_dir_char (int c)
-{
- /* FIXME: this should be replaced with a lookup table with suitable
- bits set, like standard C ctype macros do. */
- return (c == LRE_CHAR || c == LRO_CHAR
- || c == RLE_CHAR || c == RLO_CHAR || c == PDF_CHAR);
+ if (!bidi_initialized)
+ abort ();
+ 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
+ || ch_type == PDF);
}
/* A helper function for bidi_resolve_explicit. It advances to the
int current_level;
int new_level;
bidi_dir_t override;
+ int string_p = bidi_it->string.s != NULL || STRINGP (bidi_it->string.lstring);
- if (bidi_it->bytepos < BEGV_BYTE /* after reseat to BEGV? */
+ /* If reseat()'ed, don't advance, so as to start iteration from the
+ position where we were reseated. bidi_it->bytepos can be less
+ than BEGV_BYTE after reseat to BEGV. */
+ if (bidi_it->bytepos < (string_p ? 0 : BEGV_BYTE)
|| bidi_it->first_elt)
{
bidi_it->first_elt = 0;
- if (bidi_it->charpos < BEGV)
- bidi_it->charpos = BEGV;
- bidi_it->bytepos = CHAR_TO_BYTE (bidi_it->charpos);
+ if (string_p)
+ {
+ const unsigned char *p =
+ STRINGP (bidi_it->string.lstring)
+ ? SDATA (bidi_it->string.lstring) : bidi_it->string.s;
+
+ if (bidi_it->charpos < 0)
+ bidi_it->charpos = 0;
+ bidi_it->bytepos = bidi_count_bytes (p, 0, 0, bidi_it->charpos,
+ bidi_it->string.unibyte);
+ }
+ else
+ {
+ if (bidi_it->charpos < BEGV)
+ bidi_it->charpos = BEGV;
+ bidi_it->bytepos = CHAR_TO_BYTE (bidi_it->charpos);
+ }
}
- else if (bidi_it->bytepos < ZV_BYTE) /* don't move at ZV */
+ /* Don't move at end of buffer/string. */
+ else if (bidi_it->charpos < (string_p ? bidi_it->string.schars : ZV))
{
- bidi_it->charpos++;
+ /* Advance to the next character, skipping characters covered by
+ display strings (nchars > 1). */
+ if (bidi_it->nchars <= 0)
+ abort ();
+ bidi_it->charpos += bidi_it->nchars;
if (bidi_it->ch_len == 0)
abort ();
bidi_it->bytepos += bidi_it->ch_len;
override = bidi_it->level_stack[bidi_it->stack_idx].override;
new_level = current_level;
- /* in case it is a unibyte character (not yet implemented) */
- /* _fetch_multibyte_char_len = 1; */
- if (bidi_it->bytepos >= ZV_BYTE)
+ if (bidi_it->charpos >= (string_p ? bidi_it->string.schars : ZV))
{
curchar = BIDI_EOB;
bidi_it->ch_len = 1;
+ bidi_it->nchars = 1;
+ bidi_it->disp_pos = (string_p ? bidi_it->string.schars : ZV);
}
else
{
- curchar = FETCH_CHAR (bidi_it->bytepos);
- bidi_it->ch_len = CHAR_BYTES (curchar);
+ /* Fetch the character at BYTEPOS. If it is covered by a
+ display string, treat the entire run of covered characters as
+ a single character u+FFFC. */
+ curchar = bidi_fetch_char (bidi_it->bytepos, bidi_it->charpos,
+ &bidi_it->disp_pos, &bidi_it->string,
+ bidi_it->frame_window_p,
+ &bidi_it->ch_len, &bidi_it->nchars);
}
bidi_it->ch = curchar;
bidi_it->type_after_w1 = type;
bidi_check_type (bidi_it->type_after_w1);
type = WEAK_BN; /* X9/Retaining */
- if (bidi_it->ignore_bn_limit <= 0)
+ if (bidi_it->ignore_bn_limit <= -1)
{
if (current_level <= BIDI_MAXLEVEL - 4)
{
bidi_it->type_after_w1 = type;
bidi_check_type (bidi_it->type_after_w1);
type = WEAK_BN; /* X9/Retaining */
- if (bidi_it->ignore_bn_limit <= 0)
+ if (bidi_it->ignore_bn_limit <= -1)
{
if (current_level <= BIDI_MAXLEVEL - 5)
{
bidi_it->type_after_w1 = type;
bidi_check_type (bidi_it->type_after_w1);
type = WEAK_BN; /* X9/Retaining */
- if (bidi_it->ignore_bn_limit <= 0)
+ if (bidi_it->ignore_bn_limit <= -1)
{
if (!bidi_it->invalid_rl_levels)
{
}
/* Given an iterator state in BIDI_IT, advance one character position
- in the buffer to the next character (in the logical order), resolve
- any explicit embeddings and directional overrides, and return the
- embedding level of the character after resolving explicit
- directives and ignoring empty embeddings. */
+ in the buffer/string to the next character (in the logical order),
+ resolve any explicit embeddings and directional overrides, and
+ return the embedding level of the character after resolving
+ explicit directives and ignoring empty embeddings. */
static int
bidi_resolve_explicit (struct bidi_it *bidi_it)
{
int prev_level = bidi_it->level_stack[bidi_it->stack_idx].level;
int new_level = bidi_resolve_explicit_1 (bidi_it);
+ EMACS_INT eob = bidi_it->string.s ? bidi_it->string.schars : ZV;
+ const unsigned char *s = STRINGP (bidi_it->string.lstring)
+ ? SDATA (bidi_it->string.lstring) : bidi_it->string.s;
if (prev_level < new_level
&& bidi_it->type == WEAK_BN
- && bidi_it->ignore_bn_limit == 0 /* only if not already known */
- && bidi_it->ch != BIDI_EOB /* not already at EOB */
- && bidi_explicit_dir_char (FETCH_CHAR (bidi_it->bytepos
- + bidi_it->ch_len)))
+ && bidi_it->ignore_bn_limit == -1 /* only if not already known */
+ && bidi_it->charpos < eob /* not already at EOB */
+ && bidi_explicit_dir_char (bidi_char_at_pos (bidi_it->bytepos
+ + bidi_it->ch_len, s,
+ bidi_it->string.unibyte)))
{
/* Avoid pushing and popping embedding levels if the level run
is empty, as this breaks level runs where it shouldn't.
bidi_copy_it (&saved_it, bidi_it);
- while (bidi_explicit_dir_char (FETCH_CHAR (bidi_it->bytepos
- + bidi_it->ch_len)))
+ while (bidi_explicit_dir_char (bidi_char_at_pos (bidi_it->bytepos
+ + bidi_it->ch_len, s,
+ bidi_it->string.unibyte)))
{
+ /* This advances to the next character, skipping any
+ characters covered by display strings. */
level = bidi_resolve_explicit_1 (bidi_it);
+ /* If string.lstring was relocated inside bidi_resolve_explicit_1,
+ a pointer to its data is no longer valid. */
+ if (STRINGP (bidi_it->string.lstring))
+ s = SDATA (bidi_it->string.lstring);
}
+ if (bidi_it->nchars <= 0)
+ abort ();
if (level == prev_level) /* empty embedding */
- saved_it.ignore_bn_limit = bidi_it->charpos + 1;
+ saved_it.ignore_bn_limit = bidi_it->charpos + bidi_it->nchars;
else /* this embedding is non-empty */
- saved_it.ignore_bn_limit = -1;
+ saved_it.ignore_bn_limit = -2;
bidi_copy_it (bidi_it, &saved_it);
- if (bidi_it->ignore_bn_limit > 0)
+ if (bidi_it->ignore_bn_limit > -1)
{
/* We pushed a level, but we shouldn't have. Undo that. */
if (!bidi_it->invalid_rl_levels)
return new_level;
}
-/* Advance in the buffer, resolve weak types and return the type of
- the next character after weak type resolution. */
-bidi_type_t
+/* Advance in the buffer/string, resolve weak types and return the
+ type of the next character after weak type resolution. */
+static bidi_type_t
bidi_resolve_weak (struct bidi_it *bidi_it)
{
bidi_type_t type;
int next_char;
bidi_type_t type_of_next;
struct bidi_it saved_it;
+ EMACS_INT eob =
+ (STRINGP (bidi_it->string.lstring) || bidi_it->string.s)
+ ? bidi_it->string.schars : ZV;
type = bidi_it->type;
override = bidi_it->level_stack[bidi_it->stack_idx].override;
if (type == WEAK_NSM) /* W1 */
{
/* Note that we don't need to consider the case where the
- prev character has its type overridden by an RLO or LRO:
- such characters are outside the current level run, and
- thus not relevant to this NSM. Thus, NSM gets the
- orig_type of the previous character. */
- if (bidi_it->prev.type != UNKNOWN_BT)
- type = bidi_it->prev.orig_type;
+ prev character has its type overridden by an RLO or LRO,
+ because then either the type of this NSM would have been
+ also overridden, or the previous character is outside the
+ current level run, and thus not relevant to this NSM.
+ This is why NSM gets the type_after_w1 of the previous
+ character. */
+ if (bidi_it->prev.type_after_w1 != UNKNOWN_BT
+ /* if type_after_w1 is NEUTRAL_B, this NSM is at sor */
+ && bidi_it->prev.type_after_w1 != NEUTRAL_B)
+ type = bidi_it->prev.type_after_w1;
else if (bidi_it->sor == R2L)
type = STRONG_R;
else if (bidi_it->sor == L2R)
&& bidi_it->prev.orig_type == WEAK_EN)
|| bidi_it->prev.type_after_w1 == WEAK_AN)))
{
+ const unsigned char *s =
+ STRINGP (bidi_it->string.lstring)
+ ? SDATA (bidi_it->string.lstring) : bidi_it->string.s;
+
next_char =
- bidi_it->bytepos + bidi_it->ch_len >= ZV_BYTE
- ? BIDI_EOB : FETCH_CHAR (bidi_it->bytepos + bidi_it->ch_len);
+ bidi_it->charpos + bidi_it->nchars >= eob
+ ? BIDI_EOB
+ : bidi_char_at_pos (bidi_it->bytepos + bidi_it->ch_len, s,
+ bidi_it->string.unibyte);
type_of_next = bidi_get_type (next_char, override);
if (type_of_next == WEAK_BN
type = WEAK_EN;
else /* W5: ET/BN with EN after it. */
{
- EMACS_INT en_pos = bidi_it->charpos + 1;
+ EMACS_INT en_pos = bidi_it->charpos + bidi_it->nchars;
+ const unsigned char *s =
+ STRINGP (bidi_it->string.lstring)
+ ? SDATA (bidi_it->string.lstring) : bidi_it->string.s;
+ if (bidi_it->nchars <= 0)
+ abort ();
next_char =
- bidi_it->bytepos + bidi_it->ch_len >= ZV_BYTE
- ? BIDI_EOB : FETCH_CHAR (bidi_it->bytepos + bidi_it->ch_len);
+ bidi_it->charpos + bidi_it->nchars >= eob
+ ? BIDI_EOB
+ : bidi_char_at_pos (bidi_it->bytepos + bidi_it->ch_len, s,
+ bidi_it->string.unibyte);
type_of_next = bidi_get_type (next_char, override);
if (type_of_next == WEAK_ET
return type;
}
-bidi_type_t
+/* Resolve the type of a neutral character according to the type of
+ surrounding strong text and the current embedding level. */
+static inline bidi_type_t
+bidi_resolve_neutral_1 (bidi_type_t prev_type, bidi_type_t next_type, int lev)
+{
+ /* N1: European and Arabic numbers are treated as though they were R. */
+ if (next_type == WEAK_EN || next_type == WEAK_AN)
+ next_type = STRONG_R;
+ if (prev_type == WEAK_EN || prev_type == WEAK_AN)
+ prev_type = STRONG_R;
+
+ if (next_type == prev_type) /* N1 */
+ return next_type;
+ else if ((lev & 1) == 0) /* N2 */
+ return STRONG_L;
+ else
+ return STRONG_R;
+}
+
+static bidi_type_t
bidi_resolve_neutral (struct bidi_it *bidi_it)
{
int prev_level = bidi_it->level_stack[bidi_it->stack_idx].level;
/* Arrrgh!! The UAX#9 algorithm is too deeply entrenched in
the assumption of batch-style processing; see clauses W4,
W5, and especially N1, which require to look far forward
- (as well as back) in the buffer. May the fleas of a
- thousand camels infest the armpits of those who design
+ (as well as back) in the buffer/string. May the fleas of
+ a thousand camels infest the armpits of those who design
supposedly general-purpose algorithms by looking at their
own implementations, and fail to consider other possible
implementations! */
}
/* Given an iterator state in BIDI_IT, advance one character position
- in the buffer to the next character (in the logical order), resolve
- the bidi type of that next character, and return that type. */
-bidi_type_t
+ in the buffer/string to the next character (in the logical order),
+ resolve the bidi type of that next character, and return that
+ type. */
+static bidi_type_t
bidi_type_of_next_char (struct bidi_it *bidi_it)
{
bidi_type_t type;
/* Reset the limit until which to ignore BNs if we step out of the
area where we found only empty levels. */
- if ((bidi_it->ignore_bn_limit > 0
+ if ((bidi_it->ignore_bn_limit > -1
&& bidi_it->ignore_bn_limit <= bidi_it->charpos)
- || (bidi_it->ignore_bn_limit == -1
+ || (bidi_it->ignore_bn_limit == -2
&& !bidi_explicit_dir_char (bidi_it->ch)))
- bidi_it->ignore_bn_limit = 0;
+ bidi_it->ignore_bn_limit = -1;
type = bidi_resolve_neutral (bidi_it);
}
/* Given an iterator state BIDI_IT, advance one character position in
- the buffer to the next character (in the logical order), resolve
- the embedding and implicit levels of that next character, and
- return the resulting level. */
-int
+ the buffer/string to the next character (in the current scan
+ direction), resolve the embedding and implicit levels of that next
+ character, and return the resulting level. */
+static int
bidi_level_of_next_char (struct bidi_it *bidi_it)
{
bidi_type_t type;
int level, prev_level = -1;
struct bidi_saved_info next_for_neutral;
+ EMACS_INT next_char_pos = -2;
if (bidi_it->scan_dir == 1)
{
+ EMACS_INT eob =
+ (bidi_it->string.s || STRINGP (bidi_it->string.lstring))
+ ? bidi_it->string.schars : ZV;
+
/* There's no sense in trying to advance if we hit end of text. */
- if (bidi_it->ch == BIDI_EOB)
+ if (bidi_it->charpos >= eob)
return bidi_it->resolved_level;
/* Record the info about the previous character. */
}
next_for_neutral = bidi_it->next_for_neutral;
- /* Perhaps it is already cached. */
- type = bidi_cache_find (bidi_it->charpos + bidi_it->scan_dir, -1, bidi_it);
+ /* Perhaps the character we want is already cached. If it is, the
+ call to bidi_cache_find below will return a type other than
+ UNKNOWN_BT. */
+ if (bidi_cache_idx > bidi_cache_start && !bidi_it->first_elt)
+ {
+ int bob =
+ (bidi_it->string.s || STRINGP (bidi_it->string.lstring)) ? 0 : 1;
+
+ if (bidi_it->scan_dir > 0)
+ {
+ if (bidi_it->nchars <= 0)
+ abort ();
+ next_char_pos = bidi_it->charpos + bidi_it->nchars;
+ }
+ else if (bidi_it->charpos >= bob)
+ /* Implementation note: we allow next_char_pos to be as low as
+ 0 for buffers or -1 for strings, and that is okay because
+ that's the "position" of the sentinel iterator state we
+ cached at the beginning of the iteration. */
+ next_char_pos = bidi_it->charpos - 1;
+ if (next_char_pos >= bob - 1)
+ type = bidi_cache_find (next_char_pos, -1, bidi_it);
+ else
+ type = UNKNOWN_BT;
+ }
+ else
+ type = UNKNOWN_BT;
if (type != UNKNOWN_BT)
{
/* Don't lose the information for resolving neutrals! The
bidi_check_type (bidi_it->type);
/* For L1 below, we need to know, for each WS character, whether
- it belongs to a sequence of WS characters preceeding a newline
+ 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->next_for_ws.type == UNKNOWN_BT)
{
int ch;
- int clen = bidi_it->ch_len;
+ EMACS_INT clen = bidi_it->ch_len;
EMACS_INT bpos = bidi_it->bytepos;
EMACS_INT cpos = bidi_it->charpos;
+ EMACS_INT disp_pos = bidi_it->disp_pos;
+ EMACS_INT nc = bidi_it->nchars;
+ struct bidi_string_data bs = bidi_it->string;
bidi_type_t chtype;
+ int fwp = bidi_it->frame_window_p;
+ if (bidi_it->nchars <= 0)
+ abort ();
do {
- /*_fetch_multibyte_char_len = 1;*/
- ch = bpos + clen >= ZV_BYTE ? BIDI_EOB : FETCH_CHAR (bpos + clen);
- bpos += clen;
- cpos++;
- clen = (ch == BIDI_EOB ? 1 : CHAR_BYTES (ch));
+ ch = bidi_fetch_char (bpos += clen, cpos += nc, &disp_pos, &bs, fwp,
+ &clen, &nc);
if (ch == '\n' || ch == BIDI_EOB /* || ch == LINESEP_CHAR */)
chtype = NEUTRAL_B;
else
we want it to be displayed as
- {RLO}STet{PDF}
+ {PDF}STet{RLO}
not as
If this level's other edge is cached, we simply jump to it, filling
the iterator structure with the iterator state on the other edge.
- Otherwise, we walk the buffer until we come back to the same level
- as LEVEL.
+ Otherwise, we walk the buffer or string until we come back to the
+ same level as LEVEL.
Note: we are not talking here about a ``level run'' in the UAX#9
sense of the term, but rather about a ``level'' which includes
bidi_find_other_level_edge (struct bidi_it *bidi_it, int level, int end_flag)
{
int dir = end_flag ? -bidi_it->scan_dir : bidi_it->scan_dir;
- int idx;
+ EMACS_INT idx;
/* Try the cache first. */
- if ((idx = bidi_cache_find_level_change (level, dir, end_flag)) >= 0)
+ if ((idx = bidi_cache_find_level_change (level, dir, end_flag))
+ >= bidi_cache_start)
bidi_cache_fetch_state (idx, bidi_it);
else
{
}
void
-bidi_get_next_char_visually (struct bidi_it *bidi_it)
+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)
+ abort ();
if (bidi_it->scan_dir == 0)
{
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->orig_type == NEUTRAL_B)
bidi_line_init (bidi_it);
- /* Prepare the sentinel iterator state. */
- if (bidi_cache_idx == 0)
+ /* Prepare the sentinel iterator state, and cache it. When we bump
+ into it, scanning backwards, we'll know that the last non-base
+ level is exhausted. */
+ if (bidi_cache_idx == bidi_cache_start)
{
bidi_copy_it (&sentinel, bidi_it);
if (bidi_it->first_elt)
sentinel.bytepos--;
sentinel.ch = '\n'; /* doesn't matter, but why not? */
sentinel.ch_len = 1;
+ sentinel.nchars = 1;
}
+ bidi_cache_iterator_state (&sentinel, 1);
}
old_level = bidi_it->resolved_level;
/* Reordering of resolved levels (clause L2) is implemented by
jumping to the other edge of the level and flipping direction of
- scanning the buffer whenever we find a level change. */
+ scanning the text whenever we find a level change. */
if (new_level != old_level)
{
int ascending = new_level > old_level;
int incr = ascending ? 1 : -1;
int expected_next_level = old_level + incr;
- /* If we don't have anything cached yet, we need to cache the
- sentinel state, since we'll need it to record where to jump
- when the last non-base level is exhausted. */
- if (bidi_cache_idx == 0)
- bidi_cache_iterator_state (&sentinel, 1);
/* Jump (or walk) to the other edge of this level. */
bidi_find_other_level_edge (bidi_it, level_to_search, !ascending);
/* Switch scan direction and peek at the next character in the
paragraph direction, if needed. We do this at the newline before
the paragraph separator, because the next character might not be
the first character of the next paragraph, due to the bidi
- reordering. */
- if (bidi_it->scan_dir == 1
- && bidi_it->orig_type == NEUTRAL_B
- && bidi_it->bytepos < ZV_BYTE)
+ reordering, whereas we _must_ know the paragraph base direction
+ _before_ we process the paragraph's text, since the base
+ direction affects the reordering. */
+ if (bidi_it->scan_dir == 1 && bidi_it->orig_type == NEUTRAL_B)
{
- EMACS_INT sep_len =
- bidi_at_paragraph_end (bidi_it->charpos + 1,
- bidi_it->bytepos + bidi_it->ch_len);
- if (sep_len >= 0)
+ /* The paragraph direction of the entire string, once
+ determined, is in effect for the entire string. Setting the
+ separator limit to the end of the string prevents
+ bidi_paragraph_init from being called automatically on this
+ string. */
+ if (bidi_it->string.s || STRINGP (bidi_it->string.lstring))
+ bidi_it->separator_limit = bidi_it->string.schars;
+ else if (bidi_it->bytepos < ZV_BYTE)
{
- bidi_it->new_paragraph = 1;
- /* Record the buffer position of the last character of the
- paragraph separator. */
- bidi_it->separator_limit = bidi_it->charpos + 1 + sep_len;
+ EMACS_INT sep_len =
+ bidi_at_paragraph_end (bidi_it->charpos + bidi_it->nchars,
+ bidi_it->bytepos + bidi_it->ch_len);
+ if (bidi_it->nchars <= 0)
+ abort ();
+ if (sep_len >= 0)
+ {
+ bidi_it->new_paragraph = 1;
+ /* Record the buffer position of the last character of the
+ paragraph separator. */
+ bidi_it->separator_limit =
+ bidi_it->charpos + bidi_it->nchars + sep_len;
+ }
}
}
- if (bidi_it->scan_dir == 1 && bidi_cache_idx)
+ if (bidi_it->scan_dir == 1 && bidi_cache_idx > bidi_cache_start)
{
/* If we are at paragraph's base embedding level and beyond the
last cached position, the cache's job is done and we can
discard it. */
if (bidi_it->resolved_level == bidi_it->level_stack[0].level
- && bidi_it->charpos > bidi_cache[bidi_cache_idx - 1].charpos)
+ && bidi_it->charpos > (bidi_cache[bidi_cache_idx - 1].charpos
+ + bidi_cache[bidi_cache_idx - 1].nchars - 1))
bidi_cache_reset ();
/* But as long as we are caching during forward scan, we must
cache each state, or else the cache integrity will be
else
bidi_cache_iterator_state (bidi_it, 1);
}
+
+ if (STRINGP (bidi_it->string.lstring))
+ UNGCPRO;
}
/* This is meant to be called from within the debugger, whenever you
wish to examine the cache contents. */
+void bidi_dump_cached_states (void) EXTERNALLY_VISIBLE;
void
bidi_dump_cached_states (void)
{
fprintf (stderr, "The cache is empty.\n");
return;
}
- fprintf (stderr, "Total of %d state%s in cache:\n",
+ fprintf (stderr, "Total of %"pD"d state%s in cache:\n",
bidi_cache_idx, bidi_cache_idx == 1 ? "" : "s");
for (i = bidi_cache[bidi_cache_idx - 1].charpos; i > 0; i /= 10)
fputs ("\n", stderr);
fputs ("pos ", stderr);
for (i = 0; i < bidi_cache_idx; i++)
- fprintf (stderr, "%*d", ndigits, bidi_cache[i].charpos);
+ fprintf (stderr, "%*"pI"d", ndigits, bidi_cache[i].charpos);
fputs ("\n", stderr);
}
HCoop / bpt / emacs.git / blobdiff