From 43f04a7456419153cb03e610a825056a47824780 Mon Sep 17 00:00:00 2001 From: Behdad Esfahbod Date: Wed, 14 Nov 2012 15:51:54 -0800 Subject: [PATCH] Move Thai shaper into a separate file --- src/Makefile.am | 3 +- ...plex-misc.cc => hb-ot-shape-complex-default.cc} | 114 ----------------- src/hb-ot-shape-complex-thai.cc | 141 +++++++++++++++++++++ 3 files changed, 143 insertions(+), 115 deletions(-) rename src/{hb-ot-shape-complex-misc.cc => hb-ot-shape-complex-default.cc} (61%) create mode 100644 src/hb-ot-shape-complex-thai.cc diff --git a/src/Makefile.am b/src/Makefile.am index 372c10f..aca8abf 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -81,11 +81,12 @@ HBSOURCES += \ hb-ot-shape-complex-arabic.cc \ hb-ot-shape-complex-arabic-fallback.hh \ hb-ot-shape-complex-arabic-table.hh \ + hb-ot-shape-complex-default.cc \ hb-ot-shape-complex-indic.cc \ hb-ot-shape-complex-indic-machine.hh \ hb-ot-shape-complex-indic-private.hh \ hb-ot-shape-complex-indic-table.hh \ - hb-ot-shape-complex-misc.cc \ + hb-ot-shape-complex-thai.cc \ hb-ot-shape-complex-private.hh \ hb-ot-shape-normalize-private.hh \ hb-ot-shape-normalize.cc \ diff --git a/src/hb-ot-shape-complex-misc.cc b/src/hb-ot-shape-complex-default.cc similarity index 61% rename from src/hb-ot-shape-complex-misc.cc rename to src/hb-ot-shape-complex-default.cc index afe5f70..7645e22 100644 --- a/src/hb-ot-shape-complex-misc.cc +++ b/src/hb-ot-shape-complex-default.cc @@ -223,117 +223,3 @@ const hb_ot_complex_shaper_t _hb_ot_complex_shaper_default = true, /* zero_width_attached_marks */ true, /* fallback_position */ }; - - -/* Thai / Lao shaper */ - -static void -preprocess_text_thai (const hb_ot_shape_plan_t *plan HB_UNUSED, - hb_buffer_t *buffer, - hb_font_t *font HB_UNUSED) -{ - /* The following is NOT specified in the MS OT Thai spec, however, it seems - * to be what Uniscribe and other engines implement. According to Eric Muller: - * - * When you have a SARA AM, decompose it in NIKHAHIT + SARA AA, *and* move the - * NIKHAHIT backwards over any tone mark (0E48-0E4B). - * - * <0E14, 0E4B, 0E33> -> <0E14, 0E4D, 0E4B, 0E32> - * - * This reordering is legit only when the NIKHAHIT comes from a SARA AM, not - * when it's there to start with. The string <0E14, 0E4B, 0E4D> is probably - * not what a user wanted, but the rendering is nevertheless nikhahit above - * chattawa. - * - * Same for Lao. - * - * Note: - * - * Uniscribe also does so below-marks reordering. Namely, it positions U+0E3A - * after U+0E38 and U+0E39. We do that by modifying the ccc for U+0E3A. - * See unicode->modified_combining_class (). Lao does NOT have a U+0E3A - * equivalent. - */ - - - /* - * Here are the characters of significance: - * - * Thai Lao - * SARA AM: U+0E33 U+0EB3 - * SARA AA: U+0E32 U+0EB2 - * Nikhahit: U+0E4D U+0ECD - * - * Testing shows that Uniscribe reorder the following marks: - * Thai: <0E31,0E34..0E37,0E47..0E4E> - * Lao: <0EB1,0EB4..0EB7,0EC7..0ECE> - * - * Note how the Lao versions are the same as Thai + 0x80. - */ - - /* We only get one script at a time, so a script-agnostic implementation - * is adequate here. */ -#define IS_SARA_AM(x) (((x) & ~0x0080) == 0x0E33) -#define NIKHAHIT_FROM_SARA_AM(x) ((x) - 0xE33 + 0xE4D) -#define SARA_AA_FROM_SARA_AM(x) ((x) - 1) -#define IS_TONE_MARK(x) (hb_in_ranges ((x) & ~0x0080, 0x0E34, 0x0E37, 0x0E47, 0x0E4E, 0x0E31, 0x0E31)) - - buffer->clear_output (); - unsigned int count = buffer->len; - for (buffer->idx = 0; buffer->idx < count;) - { - hb_codepoint_t u = buffer->cur().codepoint; - if (likely (!IS_SARA_AM (u))) { - buffer->next_glyph (); - continue; - } - - /* Is SARA AM. Decompose and reorder. */ - hb_codepoint_t decomposed[2] = {hb_codepoint_t (NIKHAHIT_FROM_SARA_AM (u)), - hb_codepoint_t (SARA_AA_FROM_SARA_AM (u))}; - buffer->replace_glyphs (1, 2, decomposed); - if (unlikely (buffer->in_error)) - return; - - /* Ok, let's see... */ - unsigned int end = buffer->out_len; - unsigned int start = end - 2; - while (start > 0 && IS_TONE_MARK (buffer->out_info[start - 1].codepoint)) - start--; - - if (start + 2 < end) - { - /* Move Nikhahit (end-2) to the beginning */ - buffer->merge_out_clusters (start, end); - hb_glyph_info_t t = buffer->out_info[end - 2]; - memmove (buffer->out_info + start + 1, - buffer->out_info + start, - sizeof (buffer->out_info[0]) * (end - start - 2)); - buffer->out_info[start] = t; - } - else - { - /* Since we decomposed, and NIKHAHIT is combining, merge clusters with the - * previous cluster. */ - if (start) - buffer->merge_out_clusters (start - 1, end); - } - } - buffer->swap_buffers (); -} - -const hb_ot_complex_shaper_t _hb_ot_complex_shaper_thai = -{ - "thai", - NULL, /* collect_features */ - NULL, /* override_features */ - NULL, /* data_create */ - NULL, /* data_destroy */ - preprocess_text_thai, - NULL, /* normalization_preference */ - NULL, /* decompose */ - NULL, /* compose */ - NULL, /* setup_masks */ - true, /* zero_width_attached_marks */ - false,/* fallback_position */ -}; diff --git a/src/hb-ot-shape-complex-thai.cc b/src/hb-ot-shape-complex-thai.cc new file mode 100644 index 0000000..d6f4669 --- /dev/null +++ b/src/hb-ot-shape-complex-thai.cc @@ -0,0 +1,141 @@ +/* + * Copyright © 2010,2012 Google, Inc. + * + * This is part of HarfBuzz, a text shaping library. + * + * Permission is hereby granted, without written agreement and without + * license or royalty fees, to use, copy, modify, and distribute this + * software and its documentation for any purpose, provided that the + * above copyright notice and the following two paragraphs appear in + * all copies of this software. + * + * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR + * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES + * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN + * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH + * DAMAGE. + * + * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, + * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS + * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO + * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS. + * + * Google Author(s): Behdad Esfahbod + */ + +#include "hb-ot-shape-complex-private.hh" + + +/* Thai / Lao shaper */ + +static void +preprocess_text_thai (const hb_ot_shape_plan_t *plan HB_UNUSED, + hb_buffer_t *buffer, + hb_font_t *font HB_UNUSED) +{ + /* The following is NOT specified in the MS OT Thai spec, however, it seems + * to be what Uniscribe and other engines implement. According to Eric Muller: + * + * When you have a SARA AM, decompose it in NIKHAHIT + SARA AA, *and* move the + * NIKHAHIT backwards over any tone mark (0E48-0E4B). + * + * <0E14, 0E4B, 0E33> -> <0E14, 0E4D, 0E4B, 0E32> + * + * This reordering is legit only when the NIKHAHIT comes from a SARA AM, not + * when it's there to start with. The string <0E14, 0E4B, 0E4D> is probably + * not what a user wanted, but the rendering is nevertheless nikhahit above + * chattawa. + * + * Same for Lao. + * + * Note: + * + * Uniscribe also does some below-marks reordering. Namely, it positions U+0E3A + * after U+0E38 and U+0E39. We do that by modifying the ccc for U+0E3A. + * See unicode->modified_combining_class (). Lao does NOT have a U+0E3A + * equivalent. + */ + + + /* + * Here are the characters of significance: + * + * Thai Lao + * SARA AM: U+0E33 U+0EB3 + * SARA AA: U+0E32 U+0EB2 + * Nikhahit: U+0E4D U+0ECD + * + * Testing shows that Uniscribe reorder the following marks: + * Thai: <0E31,0E34..0E37,0E47..0E4E> + * Lao: <0EB1,0EB4..0EB7,0EC7..0ECE> + * + * Note how the Lao versions are the same as Thai + 0x80. + */ + + /* We only get one script at a time, so a script-agnostic implementation + * is adequate here. */ +#define IS_SARA_AM(x) (((x) & ~0x0080) == 0x0E33) +#define NIKHAHIT_FROM_SARA_AM(x) ((x) - 0xE33 + 0xE4D) +#define SARA_AA_FROM_SARA_AM(x) ((x) - 1) +#define IS_TONE_MARK(x) (hb_in_ranges ((x) & ~0x0080, 0x0E34, 0x0E37, 0x0E47, 0x0E4E, 0x0E31, 0x0E31)) + + buffer->clear_output (); + unsigned int count = buffer->len; + for (buffer->idx = 0; buffer->idx < count;) + { + hb_codepoint_t u = buffer->cur().codepoint; + if (likely (!IS_SARA_AM (u))) { + buffer->next_glyph (); + continue; + } + + /* Is SARA AM. Decompose and reorder. */ + hb_codepoint_t decomposed[2] = {hb_codepoint_t (NIKHAHIT_FROM_SARA_AM (u)), + hb_codepoint_t (SARA_AA_FROM_SARA_AM (u))}; + buffer->replace_glyphs (1, 2, decomposed); + if (unlikely (buffer->in_error)) + return; + + /* Ok, let's see... */ + unsigned int end = buffer->out_len; + unsigned int start = end - 2; + while (start > 0 && IS_TONE_MARK (buffer->out_info[start - 1].codepoint)) + start--; + + if (start + 2 < end) + { + /* Move Nikhahit (end-2) to the beginning */ + buffer->merge_out_clusters (start, end); + hb_glyph_info_t t = buffer->out_info[end - 2]; + memmove (buffer->out_info + start + 1, + buffer->out_info + start, + sizeof (buffer->out_info[0]) * (end - start - 2)); + buffer->out_info[start] = t; + } + else + { + /* Since we decomposed, and NIKHAHIT is combining, merge clusters with the + * previous cluster. */ + if (start) + buffer->merge_out_clusters (start - 1, end); + } + } + buffer->swap_buffers (); +} + +const hb_ot_complex_shaper_t _hb_ot_complex_shaper_thai = +{ + "thai", + NULL, /* collect_features */ + NULL, /* override_features */ + NULL, /* data_create */ + NULL, /* data_destroy */ + preprocess_text_thai, + NULL, /* normalization_preference */ + NULL, /* decompose */ + NULL, /* compose */ + NULL, /* setup_masks */ + true, /* zero_width_attached_marks */ + false,/* fallback_position */ +}; -- 2.7.4