src/Makefile.am | 3 src/hb-ot-map-private.hh | 8 src/hb-ot-map.cc | 6 src/hb-ot-shape-complex-arabic.cc | 1 src/hb-ot-shape-complex-default.cc | 225 +++++ src/hb-ot-shape-complex-indic.cc | 2 src/hb-ot-shape-complex-misc.cc | 339 -------- src/hb-ot-shape-complex-private.hh | 13 src/hb-ot-shape-complex-thai.cc | 378 ++++++++++ test/shaping/texts/in-tree/shaper-thai/script-thai/misc/MANIFEST | 1 test/shaping/texts/in-tree/shaper-thai/script-thai/misc/pua-shaping.txt | 11 11 files changed, 635 insertions(+), 352 deletions(-)
New commits: commit 1eb3e94fe99a072ce422e60ac4d4d89ef489b08a Author: Behdad Esfahbod <beh...@behdad.org> Date: Wed Nov 14 17:25:03 2012 -0800 [Thai] Implement PUA-based fallback shaping As explained here: http://linux.thai.net/~thep/th-otf/shaping.html Our output now matches Uniscribe for old fonts (eg. XP Tahoma) with no Thai GSUB table. diff --git a/src/hb-ot-shape-complex-thai.cc b/src/hb-ot-shape-complex-thai.cc index 0736eca..87fe521 100644 --- a/src/hb-ot-shape-complex-thai.cc +++ b/src/hb-ot-shape-complex-thai.cc @@ -29,13 +29,230 @@ /* Thai / Lao shaper */ + +/* PUA shaping */ + + +enum thai_consonant_type_t +{ + NC, + AC, + RC, + DC, + NOT_CONSONANT, + NUM_CONSONANT_TYPES = NOT_CONSONANT +}; + +static thai_consonant_type_t +get_consonant_type (hb_codepoint_t u) +{ + if (u == 0x0E1B || u == 0x0E1D || u == 0x0E1F || u == 0x0E2C) + return AC; + if (u == 0x0E0D || u == 0x0E10) + return RC; + if (u == 0x0E0E || u == 0x0E0F) + return DC; + if (hb_in_range<hb_codepoint_t> (u, 0x0E01, 0x0E2E)) + return NC; + return NOT_CONSONANT; +} + + +enum thai_mark_type_t +{ + AV, + BV, + T, + NOT_MARK, + NUM_MARK_TYPES = NOT_MARK +}; + +static thai_mark_type_t +get_mark_type (hb_codepoint_t u) +{ + if (u == 0x0E31 || hb_in_range<hb_codepoint_t> (u, 0x0E34, 0x0E37) || + u == 0x0E47 || hb_in_range<hb_codepoint_t> (u, 0x0E4D, 0x0E4E)) + return AV; + if (hb_in_range<hb_codepoint_t> (u, 0x0E38, 0x0E3A)) + return BV; + if (hb_in_range<hb_codepoint_t> (u, 0x0E48, 0x0E4C)) + return T; + return NOT_MARK; +} + + +enum thai_action_t +{ + NOP, + SD, /* Shift combining-mark down */ + SL, /* Shift combining-mark left */ + SDL, /* Shift combining-mark down-left */ + RD /* Remove descender from base */ +}; + +static hb_codepoint_t +thai_pua_shape (hb_codepoint_t u, thai_action_t action, hb_font_t *font) +{ + struct thai_pua_mapping_t { + hb_codepoint_t u; + hb_codepoint_t win_pua; + hb_codepoint_t mac_pua; + } const *pua_mappings = NULL; + static const thai_pua_mapping_t SD_mappings[] = { + {0x0E48, 0xF70A, 0xF88B}, /* MAI EK */ + {0x0E49, 0xF70B, 0xF88E}, /* MAI THO */ + {0x0E4A, 0xF70C, 0xF891}, /* MAI TRI */ + {0x0E4B, 0xF70D, 0xF894}, /* MAI CHATTAWA */ + {0x0E4C, 0xF70E, 0xF897}, /* THANTHAKHAT */ + {0x0E38, 0xF718, 0xF89B}, /* SARA U */ + {0x0E39, 0xF719, 0xF89C}, /* SARA UU */ + {0x0E3A, 0xF71A, 0xF89D}, /* PHINTHU */ + {0x0000, 0x0000, 0x0000} + }; + static const thai_pua_mapping_t SDL_mappings[] = { + {0x0E48, 0xF705, 0xF88C}, /* MAI EK */ + {0x0E49, 0xF706, 0xF88F}, /* MAI THO */ + {0x0E4A, 0xF707, 0xF892}, /* MAI TRI */ + {0x0E4B, 0xF708, 0xF895}, /* MAI CHATTAWA */ + {0x0E4C, 0xF709, 0xF898}, /* THANTHAKHAT */ + {0x0000, 0x0000, 0x0000} + }; + static const thai_pua_mapping_t SL_mappings[] = { + {0x0E48, 0xF713, 0xF88A}, /* MAI EK */ + {0x0E49, 0xF714, 0xF88D}, /* MAI THO */ + {0x0E4A, 0xF715, 0xF890}, /* MAI TRI */ + {0x0E4B, 0xF716, 0xF893}, /* MAI CHATTAWA */ + {0x0E4C, 0xF717, 0xF896}, /* THANTHAKHAT */ + {0x0E31, 0xF710, 0xF884}, /* MAI HAN-AKAT */ + {0x0E34, 0xF701, 0xF885}, /* SARA I */ + {0x0E35, 0xF702, 0xF886}, /* SARA II */ + {0x0E36, 0xF703, 0xF887}, /* SARA UE */ + {0x0E37, 0xF704, 0xF888}, /* SARA UEE */ + {0x0E47, 0xF712, 0xF889}, /* MAITAIKHU */ + {0x0E4D, 0xF711, 0xF899}, /* NIKHAHIT */ + {0x0000, 0x0000, 0x0000} + }; + static const thai_pua_mapping_t RD_mappings[] = { + {0x0E0D, 0xF70F, 0xF89A}, /* YO YING */ + {0x0E10, 0xF700, 0xF89E}, /* THO THAN */ + {0x0000, 0x0000, 0x0000} + }; + + switch (action) { + default: assert (false); /* Fallthrough */ + case NOP: return u; + case SD: pua_mappings = SD_mappings; break; + case SDL: pua_mappings = SDL_mappings; break; + case SL: pua_mappings = SL_mappings; break; + case RD: pua_mappings = RD_mappings; break; + } + for (; pua_mappings->u; pua_mappings++) + if (pua_mappings->u == u) + { + hb_codepoint_t glyph; + if (hb_font_get_glyph (font, pua_mappings->win_pua, 0, &glyph)) + return pua_mappings->win_pua; + if (hb_font_get_glyph (font, pua_mappings->mac_pua, 0, &glyph)) + return pua_mappings->mac_pua; + break; + } + return u; +} + + +static enum thai_above_state_t +{ /* Cluster above looks like: */ + T0, /* ⣤ */ + T1, /* ⣼ */ + T2, /* ⣾ */ + T3, /* ⣿ */ + NUM_ABOVE_STATES +} thai_above_start_state[NUM_CONSONANT_TYPES + 1/* For NOT_CONSONANT */] = +{ + T0, /* NC */ + T1, /* AC */ + T0, /* RC */ + T0, /* DC */ + T3, /* NOT_CONSONANT */ +}; + +static const struct thai_above_state_machine_edge_t { + thai_action_t action; + thai_above_state_t next_state; +} thai_above_state_machine[NUM_ABOVE_STATES][NUM_MARK_TYPES] = +{ /*AV*/ /*BV*/ /*T*/ +/*T0*/ {{NOP,T3}, {NOP,T0}, {SD, T3}}, +/*T1*/ {{SL, T2}, {NOP,T1}, {SDL,T2}}, +/*T2*/ {{NOP,T3}, {NOP,T2}, {SL, T3}}, +/*T3*/ {{NOP,T3}, {NOP,T3}, {NOP,T3}}, +}; + + +static enum thai_below_state_t +{ + B0, /* No descender */ + B1, /* Removable descender */ + B2, /* Strict descender */ + NUM_BELOW_STATES +} thai_below_start_state[NUM_CONSONANT_TYPES + 1/* For NOT_CONSONANT */] = +{ + B0, /* NC */ + B0, /* AC */ + B1, /* RC */ + B2, /* DC */ + B2, /* NOT_CONSONANT */ +}; + +static const struct thai_below_state_machine_edge_t { + thai_action_t action; + thai_below_state_t next_state; +} thai_below_state_machine[NUM_BELOW_STATES][NUM_MARK_TYPES] = +{ /*AV*/ /*BV*/ /*T*/ +/*B0*/ {{NOP,B0}, {NOP,B2}, {NOP, B0}}, +/*B1*/ {{NOP,B1}, {RD, B2}, {NOP, B1}}, +/*B2*/ {{NOP,B2}, {SD, B2}, {NOP, B2}}, +}; + + static void do_thai_pua_shaping (const hb_ot_shape_plan_t *plan, hb_buffer_t *buffer, hb_font_t *font) { + thai_above_state_t above_state = thai_above_start_state[NOT_CONSONANT]; + thai_below_state_t below_state = thai_below_start_state[NOT_CONSONANT]; + unsigned int base = 0; + + hb_glyph_info_t *info = buffer->info; + unsigned int count = buffer->len; + for (unsigned int i = 0; i < count; i++) + { + thai_mark_type_t mt = get_mark_type (info[i].codepoint); + + if (mt == NOT_MARK) { + thai_consonant_type_t ct = get_consonant_type (info[i].codepoint); + above_state = thai_above_start_state[ct]; + below_state = thai_below_start_state[ct]; + base = i; + continue; + } + + const thai_above_state_machine_edge_t &above_edge = thai_above_state_machine[above_state][mt]; + const thai_below_state_machine_edge_t &below_edge = thai_below_state_machine[below_state][mt]; + above_state = above_edge.next_state; + below_state = below_edge.next_state; + + /* At least one of the above/below actions is NOP. */ + thai_action_t action = above_edge.action != NOP ? above_edge.action : below_edge.action; + + if (action == RD) + info[base].codepoint = thai_pua_shape (info[base].codepoint, action, font); + else + info[i].codepoint = thai_pua_shape (info[i].codepoint, action, font); + } } + static void preprocess_text_thai (const hb_ot_shape_plan_t *plan, hb_buffer_t *buffer, commit 851784f8372004e0a40b698c0cdc2d7db8629aa2 Author: Behdad Esfahbod <beh...@behdad.org> Date: Wed Nov 14 16:24:05 2012 -0800 Improve shaper selection diff --git a/src/hb-ot-map-private.hh b/src/hb-ot-map-private.hh index 11d97e1..c7dbea5 100644 --- a/src/hb-ot-map-private.hh +++ b/src/hb-ot-map-private.hh @@ -115,9 +115,6 @@ struct hb_ot_map_t *lookup_count = end - start; } - inline hb_tag_t get_chosen_script (unsigned int table_index) const - { return chosen_script[table_index]; } - HB_INTERNAL void substitute_closure (const struct hb_ot_shape_plan_t *plan, hb_face_t *face, hb_set_t *glyphs) const; HB_INTERNAL void substitute (const struct hb_ot_shape_plan_t *plan, hb_font_t *font, hb_buffer_t *buffer) const; HB_INTERNAL void position (const struct hb_ot_shape_plan_t *plan, hb_font_t *font, hb_buffer_t *buffer) const; @@ -130,6 +127,9 @@ struct hb_ot_map_t pauses[1].finish (); } + public: + hb_tag_t chosen_script[2]; + bool found_script[2]; private: @@ -140,7 +140,6 @@ struct hb_ot_map_t hb_mask_t global_mask; - hb_tag_t chosen_script[2]; hb_prealloced_array_t<feature_map_t, 8> features; hb_prealloced_array_t<lookup_map_t, 32> lookups[2]; /* GSUB/GPOS */ hb_prealloced_array_t<pause_map_t, 1> pauses[2]; /* GSUB/GPOS */ @@ -200,6 +199,7 @@ struct hb_ot_map_builder_t hb_segment_properties_t props; hb_tag_t chosen_script[2]; + bool found_script[2]; unsigned int script_index[2], language_index[2]; private: diff --git a/src/hb-ot-map.cc b/src/hb-ot-map.cc index 046fa97..024b7df 100644 --- a/src/hb-ot-map.cc +++ b/src/hb-ot-map.cc @@ -79,7 +79,7 @@ hb_ot_map_builder_t::hb_ot_map_builder_t (hb_face_t *face_, for (unsigned int table_index = 0; table_index < 2; table_index++) { hb_tag_t table_tag = table_tags[table_index]; - hb_ot_layout_table_choose_script (face, table_tag, script_tags, &script_index[table_index], &chosen_script[table_index]); + found_script[table_index] = hb_ot_layout_table_choose_script (face, table_tag, script_tags, &script_index[table_index], &chosen_script[table_index]); hb_ot_layout_script_find_language (face, table_tag, script_index[table_index], language_tag, &language_index[table_index]); } } @@ -161,8 +161,10 @@ hb_ot_map_builder_t::compile (hb_ot_map_t &m) { m.global_mask = 1; - for (unsigned int table_index = 0; table_index < 2; table_index++) + for (unsigned int table_index = 0; table_index < 2; table_index++) { m.chosen_script[table_index] = chosen_script[table_index]; + m.found_script[table_index] = found_script[table_index]; + } if (!feature_infos.len) return; diff --git a/src/hb-ot-shape-complex-indic.cc b/src/hb-ot-shape-complex-indic.cc index a7f9b60..324a04b 100644 --- a/src/hb-ot-shape-complex-indic.cc +++ b/src/hb-ot-shape-complex-indic.cc @@ -329,7 +329,7 @@ data_create_indic (const hb_ot_shape_plan_t *plan) break; } - indic_plan->is_old_spec = indic_plan->config->has_old_spec && ((plan->map.get_chosen_script (0) & 0x000000FF) != '2'); + indic_plan->is_old_spec = indic_plan->config->has_old_spec && ((plan->map.chosen_script[0] & 0x000000FF) != '2'); indic_plan->virama_glyph = (hb_codepoint_t) -1; indic_plan->rphf.init (&plan->map, HB_TAG('r','p','h','f')); diff --git a/src/hb-ot-shape-complex-private.hh b/src/hb-ot-shape-complex-private.hh index 043e8e3..9f8cecd 100644 --- a/src/hb-ot-shape-complex-private.hh +++ b/src/hb-ot-shape-complex-private.hh @@ -278,15 +278,18 @@ hb_ot_shape_complex_categorize (const hb_ot_shape_planner_t *planner) case HB_SCRIPT_TAKRI: /* Only use Indic shaper if the font has Indic tables. */ - if (planner->map.chosen_script[0] == HB_OT_TAG_DEFAULT_SCRIPT) - return &_hb_ot_complex_shaper_default; - else + if (planner->map.found_script[0]) return &_hb_ot_complex_shaper_indic; + else + return &_hb_ot_complex_shaper_default; case HB_SCRIPT_KHMER: /* If the font has 'liga', let the generic shaper do it. */ - if (planner->map.chosen_script[0] == HB_OT_TAG_DEFAULT_SCRIPT || - hb_ot_layout_language_find_feature (planner->face, HB_OT_TAG_GSUB, planner->map.script_index[0], planner->map.language_index[0], HB_TAG ('l','i','g','a'), NULL)) + if (!planner->map.found_script[0] || + hb_ot_layout_language_find_feature (planner->face, HB_OT_TAG_GSUB, + planner->map.script_index[0], + planner->map.language_index[0], + HB_TAG ('l','i','g','a'), NULL)) return &_hb_ot_complex_shaper_default; else return &_hb_ot_complex_shaper_indic; diff --git a/src/hb-ot-shape-complex-thai.cc b/src/hb-ot-shape-complex-thai.cc index d6f4669..0736eca 100644 --- a/src/hb-ot-shape-complex-thai.cc +++ b/src/hb-ot-shape-complex-thai.cc @@ -30,10 +30,26 @@ /* Thai / Lao shaper */ static void -preprocess_text_thai (const hb_ot_shape_plan_t *plan HB_UNUSED, +do_thai_pua_shaping (const hb_ot_shape_plan_t *plan, + hb_buffer_t *buffer, + hb_font_t *font) +{ +} + +static void +preprocess_text_thai (const hb_ot_shape_plan_t *plan, hb_buffer_t *buffer, - hb_font_t *font HB_UNUSED) + hb_font_t *font) { + /* This function implements the shaping logic documented here: + * + * http://linux.thai.net/~thep/th-otf/shaping.html + * + * The first shaping rule listed there is needed even if the font has Thai + * OpenType tables. The rest do fallback positioning based on PUA codepoints. + * We implement that only if there exist no Thai GSUB in the font. + */ + /* 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: * @@ -122,6 +138,10 @@ preprocess_text_thai (const hb_ot_shape_plan_t *plan HB_UNUSED, } } buffer->swap_buffers (); + + /* If font has Thai GSUB, we are done. */ + if (plan->props.script == HB_SCRIPT_THAI && !plan->map.found_script[0]) + do_thai_pua_shaping (plan, buffer, font); } const hb_ot_complex_shaper_t _hb_ot_complex_shaper_thai = commit f3584d3a3a627e38dfd7769975a670db340d2a48 Author: Behdad Esfahbod <beh...@behdad.org> Date: Wed Nov 14 15:55:17 2012 -0800 Add test cases for Thai PUA shaping diff --git a/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/MANIFEST b/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/MANIFEST index 6aa865b..911099e 100644 --- a/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/MANIFEST +++ b/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/MANIFEST @@ -1,2 +1,3 @@ phinthu.txt +pua-shaping.txt sara-am.txt diff --git a/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/pua-shaping.txt b/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/pua-shaping.txt new file mode 100644 index 0000000..c17834b --- /dev/null +++ b/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/pua-shaping.txt @@ -0,0 +1,11 @@ +à¸à¸µ +à¸à¹ +à¸à¸µà¹ +à¸à¹à¸µ +à¸à¹ +ภ+à¸à¸¸ +à¸à¸´ +à¸à¸´à¹ +à¸à¹ +à¸à¸¹ commit 43f04a7456419153cb03e610a825056a47824780 Author: Behdad Esfahbod <beh...@behdad.org> Date: Wed Nov 14 15:51:54 2012 -0800 Move Thai shaper into a separate file 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-default.cc b/src/hb-ot-shape-complex-default.cc new file mode 100644 index 0000000..7645e22 --- /dev/null +++ b/src/hb-ot-shape-complex-default.cc @@ -0,0 +1,225 @@ +/* + * 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" + + +/* TODO Add kana, and other small shapers here */ + + +/* The default shaper *only* adds additional per-script features.*/ + +static const hb_tag_t hangul_features[] = +{ + HB_TAG('l','j','m','o'), + HB_TAG('v','j','m','o'), + HB_TAG('t','j','m','o'), + HB_TAG_NONE +}; + +static const hb_tag_t tibetan_features[] = +{ + HB_TAG('a','b','v','s'), + HB_TAG('b','l','w','s'), + HB_TAG('a','b','v','m'), + HB_TAG('b','l','w','m'), + HB_TAG_NONE +}; + +static void +collect_features_default (hb_ot_shape_planner_t *plan) +{ + const hb_tag_t *script_features = NULL; + + switch ((hb_tag_t) plan->props.script) + { + /* Unicode-1.1 additions */ + case HB_SCRIPT_HANGUL: + script_features = hangul_features; + break; + + /* Unicode-2.0 additions */ + case HB_SCRIPT_TIBETAN: + script_features = tibetan_features; + break; + } + + for (; script_features && *script_features; script_features++) + plan->map.add_bool_feature (*script_features); +} + +static hb_ot_shape_normalization_mode_t +normalization_preference_default (const hb_segment_properties_t *props) +{ + switch ((hb_tag_t) props->script) + { + /* Unicode-1.1 additions */ + case HB_SCRIPT_HANGUL: + return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_FULL; + } + return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS; +} + +static hb_bool_t +compose_default (hb_unicode_funcs_t *unicode, + hb_codepoint_t a, + hb_codepoint_t b, + hb_codepoint_t *ab) +{ + /* Hebrew presentation-form shaping. + * https://bugzilla.mozilla.org/show_bug.cgi?id=728866 */ + // Hebrew presentation forms with dagesh, for characters 0x05D0..0x05EA; + // note that some letters do not have a dagesh presForm encoded + static const hb_codepoint_t sDageshForms[0x05EA - 0x05D0 + 1] = { + 0xFB30, // ALEF + 0xFB31, // BET + 0xFB32, // GIMEL + 0xFB33, // DALET + 0xFB34, // HE + 0xFB35, // VAV + 0xFB36, // ZAYIN + 0, // HET + 0xFB38, // TET + 0xFB39, // YOD + 0xFB3A, // FINAL KAF + 0xFB3B, // KAF + 0xFB3C, // LAMED + 0, // FINAL MEM + 0xFB3E, // MEM + 0, // FINAL NUN + 0xFB40, // NUN + 0xFB41, // SAMEKH + 0, // AYIN + 0xFB43, // FINAL PE + 0xFB44, // PE + 0, // FINAL TSADI + 0xFB46, // TSADI + 0xFB47, // QOF + 0xFB48, // RESH + 0xFB49, // SHIN + 0xFB4A // TAV + }; + + hb_bool_t found = unicode->compose (a, b, ab); + + if (!found && (b & ~0x7F) == 0x0580) { + // special-case Hebrew presentation forms that are excluded from + // standard normalization, but wanted for old fonts + switch (b) { + case 0x05B4: // HIRIQ + if (a == 0x05D9) { // YOD + *ab = 0xFB1D; + found = true; + } + break; + case 0x05B7: // patah + if (a == 0x05F2) { // YIDDISH YOD YOD + *ab = 0xFB1F; + found = true; + } else if (a == 0x05D0) { // ALEF + *ab = 0xFB2E; + found = true; + } + break; + case 0x05B8: // QAMATS + if (a == 0x05D0) { // ALEF + *ab = 0xFB2F; + found = true; + } + break; + case 0x05B9: // HOLAM + if (a == 0x05D5) { // VAV + *ab = 0xFB4B; + found = true; + } + break; + case 0x05BC: // DAGESH + if (a >= 0x05D0 && a <= 0x05EA) { + *ab = sDageshForms[a - 0x05D0]; + found = (*ab != 0); + } else if (a == 0xFB2A) { // SHIN WITH SHIN DOT + *ab = 0xFB2C; + found = true; + } else if (a == 0xFB2B) { // SHIN WITH SIN DOT + *ab = 0xFB2D; + found = true; + } + break; + case 0x05BF: // RAFE + switch (a) { + case 0x05D1: // BET + *ab = 0xFB4C; + found = true; + break; + case 0x05DB: // KAF + *ab = 0xFB4D; + found = true; + break; + case 0x05E4: // PE + *ab = 0xFB4E; + found = true; + break; + } + break; + case 0x05C1: // SHIN DOT + if (a == 0x05E9) { // SHIN + *ab = 0xFB2A; + found = true; + } else if (a == 0xFB49) { // SHIN WITH DAGESH + *ab = 0xFB2C; + found = true; + } + break; + case 0x05C2: // SIN DOT + if (a == 0x05E9) { // SHIN + *ab = 0xFB2B; + found = true; + } else if (a == 0xFB49) { // SHIN WITH DAGESH + *ab = 0xFB2D; + found = true; + } + break; + } + } + + return found; +} + +const hb_ot_complex_shaper_t _hb_ot_complex_shaper_default = +{ + "default", + collect_features_default, + NULL, /* override_features */ + NULL, /* data_create */ + NULL, /* data_destroy */ + NULL, /* preprocess_text */ + normalization_preference_default, + NULL, /* decompose */ + compose_default, + NULL, /* setup_masks */ + true, /* zero_width_attached_marks */ + true, /* fallback_position */ +}; diff --git a/src/hb-ot-shape-complex-misc.cc b/src/hb-ot-shape-complex-misc.cc deleted file mode 100644 index afe5f70..0000000 --- a/src/hb-ot-shape-complex-misc.cc +++ /dev/null @@ -1,339 +0,0 @@ -/* - * 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" - - -/* TODO Add kana, and other small shapers here */ - - -/* The default shaper *only* adds additional per-script features.*/ - -static const hb_tag_t hangul_features[] = -{ - HB_TAG('l','j','m','o'), - HB_TAG('v','j','m','o'), - HB_TAG('t','j','m','o'), - HB_TAG_NONE -}; - -static const hb_tag_t tibetan_features[] = -{ - HB_TAG('a','b','v','s'), - HB_TAG('b','l','w','s'), - HB_TAG('a','b','v','m'), - HB_TAG('b','l','w','m'), - HB_TAG_NONE -}; - -static void -collect_features_default (hb_ot_shape_planner_t *plan) -{ - const hb_tag_t *script_features = NULL; - - switch ((hb_tag_t) plan->props.script) - { - /* Unicode-1.1 additions */ - case HB_SCRIPT_HANGUL: - script_features = hangul_features; - break; - - /* Unicode-2.0 additions */ - case HB_SCRIPT_TIBETAN: - script_features = tibetan_features; - break; - } - - for (; script_features && *script_features; script_features++) - plan->map.add_bool_feature (*script_features); -} - -static hb_ot_shape_normalization_mode_t -normalization_preference_default (const hb_segment_properties_t *props) -{ - switch ((hb_tag_t) props->script) - { - /* Unicode-1.1 additions */ - case HB_SCRIPT_HANGUL: - return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_FULL; - } - return HB_OT_SHAPE_NORMALIZATION_MODE_COMPOSED_DIACRITICS; -} - -static hb_bool_t -compose_default (hb_unicode_funcs_t *unicode, - hb_codepoint_t a, - hb_codepoint_t b, - hb_codepoint_t *ab) -{ - /* Hebrew presentation-form shaping. - * https://bugzilla.mozilla.org/show_bug.cgi?id=728866 */ - // Hebrew presentation forms with dagesh, for characters 0x05D0..0x05EA; - // note that some letters do not have a dagesh presForm encoded - static const hb_codepoint_t sDageshForms[0x05EA - 0x05D0 + 1] = { - 0xFB30, // ALEF - 0xFB31, // BET - 0xFB32, // GIMEL - 0xFB33, // DALET - 0xFB34, // HE - 0xFB35, // VAV - 0xFB36, // ZAYIN - 0, // HET - 0xFB38, // TET - 0xFB39, // YOD - 0xFB3A, // FINAL KAF - 0xFB3B, // KAF - 0xFB3C, // LAMED - 0, // FINAL MEM - 0xFB3E, // MEM - 0, // FINAL NUN - 0xFB40, // NUN - 0xFB41, // SAMEKH - 0, // AYIN - 0xFB43, // FINAL PE - 0xFB44, // PE - 0, // FINAL TSADI - 0xFB46, // TSADI - 0xFB47, // QOF - 0xFB48, // RESH - 0xFB49, // SHIN - 0xFB4A // TAV - }; - - hb_bool_t found = unicode->compose (a, b, ab); - - if (!found && (b & ~0x7F) == 0x0580) { - // special-case Hebrew presentation forms that are excluded from - // standard normalization, but wanted for old fonts - switch (b) { - case 0x05B4: // HIRIQ - if (a == 0x05D9) { // YOD - *ab = 0xFB1D; - found = true; - } - break; - case 0x05B7: // patah - if (a == 0x05F2) { // YIDDISH YOD YOD - *ab = 0xFB1F; - found = true; - } else if (a == 0x05D0) { // ALEF - *ab = 0xFB2E; - found = true; - } - break; - case 0x05B8: // QAMATS - if (a == 0x05D0) { // ALEF - *ab = 0xFB2F; - found = true; - } - break; - case 0x05B9: // HOLAM - if (a == 0x05D5) { // VAV - *ab = 0xFB4B; - found = true; - } - break; - case 0x05BC: // DAGESH - if (a >= 0x05D0 && a <= 0x05EA) { - *ab = sDageshForms[a - 0x05D0]; - found = (*ab != 0); - } else if (a == 0xFB2A) { // SHIN WITH SHIN DOT - *ab = 0xFB2C; - found = true; - } else if (a == 0xFB2B) { // SHIN WITH SIN DOT - *ab = 0xFB2D; - found = true; - } - break; - case 0x05BF: // RAFE - switch (a) { - case 0x05D1: // BET - *ab = 0xFB4C; - found = true; - break; - case 0x05DB: // KAF - *ab = 0xFB4D; - found = true; - break; - case 0x05E4: // PE - *ab = 0xFB4E; - found = true; - break; - } - break; - case 0x05C1: // SHIN DOT - if (a == 0x05E9) { // SHIN - *ab = 0xFB2A; - found = true; - } else if (a == 0xFB49) { // SHIN WITH DAGESH - *ab = 0xFB2C; - found = true; - } - break; - case 0x05C2: // SIN DOT - if (a == 0x05E9) { // SHIN - *ab = 0xFB2B; - found = true; - } else if (a == 0xFB49) { // SHIN WITH DAGESH - *ab = 0xFB2D; - found = true; - } - break; - } - } - - return found; -} - -const hb_ot_complex_shaper_t _hb_ot_complex_shaper_default = -{ - "default", - collect_features_default, - NULL, /* override_features */ - NULL, /* data_create */ - NULL, /* data_destroy */ - NULL, /* preprocess_text */ - normalization_preference_default, - NULL, /* decompose */ - compose_default, - NULL, /* setup_masks */ - 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<hb_codepoint_t> ((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<hb_codepoint_t> ((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 */ +}; commit ba82325b7a6311b787ae47f41a56964e2f2cba9f Author: Behdad Esfahbod <beh...@behdad.org> Date: Wed Nov 14 15:36:53 2012 -0800 Add note re 'Phags-pa letter U+A872, which is Joining_Type=L diff --git a/src/hb-ot-shape-complex-arabic.cc b/src/hb-ot-shape-complex-arabic.cc index ea2a68f..e9db005 100644 --- a/src/hb-ot-shape-complex-arabic.cc +++ b/src/hb-ot-shape-complex-arabic.cc @@ -81,6 +81,7 @@ static unsigned int get_joining_type (hb_codepoint_t u, hb_unicode_general_categ if (unlikely (hb_in_range<hb_codepoint_t> (u, 0xA840, 0xA872))) { if (unlikely (u == 0xA872)) + /* XXX Looks like this should be TYPE_L, but we don't support that yet! */ return JOINING_TYPE_R; return JOINING_TYPE_D;
_______________________________________________ HarfBuzz mailing list HarfBuzz@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/harfbuzz