michael@0: /*
michael@0:  * Copyright © 2010,2012  Google, Inc.
michael@0:  *
michael@0:  *  This is part of HarfBuzz, a text shaping library.
michael@0:  *
michael@0:  * Permission is hereby granted, without written agreement and without
michael@0:  * license or royalty fees, to use, copy, modify, and distribute this
michael@0:  * software and its documentation for any purpose, provided that the
michael@0:  * above copyright notice and the following two paragraphs appear in
michael@0:  * all copies of this software.
michael@0:  *
michael@0:  * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
michael@0:  * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
michael@0:  * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
michael@0:  * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
michael@0:  * DAMAGE.
michael@0:  *
michael@0:  * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
michael@0:  * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
michael@0:  * FITNESS FOR A PARTICULAR PURPOSE.  THE SOFTWARE PROVIDED HEREUNDER IS
michael@0:  * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
michael@0:  * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
michael@0:  *
michael@0:  * Google Author(s): Behdad Esfahbod
michael@0:  */
michael@0: 
michael@0: #include "hb-ot-shape-complex-private.hh"
michael@0: 
michael@0: 
michael@0: /* Thai / Lao shaper */
michael@0: 
michael@0: 
michael@0: /* PUA shaping */
michael@0: 
michael@0: 
michael@0: enum thai_consonant_type_t
michael@0: {
michael@0:   NC,
michael@0:   AC,
michael@0:   RC,
michael@0:   DC,
michael@0:   NOT_CONSONANT,
michael@0:   NUM_CONSONANT_TYPES = NOT_CONSONANT
michael@0: };
michael@0: 
michael@0: static thai_consonant_type_t
michael@0: get_consonant_type (hb_codepoint_t u)
michael@0: {
michael@0:   if (u == 0x0E1B || u == 0x0E1D || u == 0x0E1F/* || u == 0x0E2C*/)
michael@0:     return AC;
michael@0:   if (u == 0x0E0D || u == 0x0E10)
michael@0:     return RC;
michael@0:   if (u == 0x0E0E || u == 0x0E0F)
michael@0:     return DC;
michael@0:   if (hb_in_range<hb_codepoint_t> (u, 0x0E01, 0x0E2E))
michael@0:     return NC;
michael@0:   return NOT_CONSONANT;
michael@0: }
michael@0: 
michael@0: 
michael@0: enum thai_mark_type_t
michael@0: {
michael@0:   AV,
michael@0:   BV,
michael@0:   T,
michael@0:   NOT_MARK,
michael@0:   NUM_MARK_TYPES = NOT_MARK
michael@0: };
michael@0: 
michael@0: static thai_mark_type_t
michael@0: get_mark_type (hb_codepoint_t u)
michael@0: {
michael@0:   if (u == 0x0E31 || hb_in_range<hb_codepoint_t> (u, 0x0E34, 0x0E37) ||
michael@0:       u == 0x0E47 || hb_in_range<hb_codepoint_t> (u, 0x0E4D, 0x0E4E))
michael@0:     return AV;
michael@0:   if (hb_in_range<hb_codepoint_t> (u, 0x0E38, 0x0E3A))
michael@0:     return BV;
michael@0:   if (hb_in_range<hb_codepoint_t> (u, 0x0E48, 0x0E4C))
michael@0:     return T;
michael@0:   return NOT_MARK;
michael@0: }
michael@0: 
michael@0: 
michael@0: enum thai_action_t
michael@0: {
michael@0:   NOP,
michael@0:   SD,  /* Shift combining-mark down */
michael@0:   SL,  /* Shift combining-mark left */
michael@0:   SDL, /* Shift combining-mark down-left */
michael@0:   RD   /* Remove descender from base */
michael@0: };
michael@0: 
michael@0: static hb_codepoint_t
michael@0: thai_pua_shape (hb_codepoint_t u, thai_action_t action, hb_font_t *font)
michael@0: {
michael@0:   struct thai_pua_mapping_t {
michael@0:     hb_codepoint_t u;
michael@0:     hb_codepoint_t win_pua;
michael@0:     hb_codepoint_t mac_pua;
michael@0:   } const *pua_mappings = NULL;
michael@0:   static const thai_pua_mapping_t SD_mappings[] = {
michael@0:     {0x0E48, 0xF70A, 0xF88B}, /* MAI EK */
michael@0:     {0x0E49, 0xF70B, 0xF88E}, /* MAI THO */
michael@0:     {0x0E4A, 0xF70C, 0xF891}, /* MAI TRI */
michael@0:     {0x0E4B, 0xF70D, 0xF894}, /* MAI CHATTAWA */
michael@0:     {0x0E4C, 0xF70E, 0xF897}, /* THANTHAKHAT */
michael@0:     {0x0E38, 0xF718, 0xF89B}, /* SARA U */
michael@0:     {0x0E39, 0xF719, 0xF89C}, /* SARA UU */
michael@0:     {0x0E3A, 0xF71A, 0xF89D}, /* PHINTHU */
michael@0:     {0x0000, 0x0000, 0x0000}
michael@0:   };
michael@0:   static const thai_pua_mapping_t SDL_mappings[] = {
michael@0:     {0x0E48, 0xF705, 0xF88C}, /* MAI EK */
michael@0:     {0x0E49, 0xF706, 0xF88F}, /* MAI THO */
michael@0:     {0x0E4A, 0xF707, 0xF892}, /* MAI TRI */
michael@0:     {0x0E4B, 0xF708, 0xF895}, /* MAI CHATTAWA */
michael@0:     {0x0E4C, 0xF709, 0xF898}, /* THANTHAKHAT */
michael@0:     {0x0000, 0x0000, 0x0000}
michael@0:   };
michael@0:   static const thai_pua_mapping_t SL_mappings[] = {
michael@0:     {0x0E48, 0xF713, 0xF88A}, /* MAI EK */
michael@0:     {0x0E49, 0xF714, 0xF88D}, /* MAI THO */
michael@0:     {0x0E4A, 0xF715, 0xF890}, /* MAI TRI */
michael@0:     {0x0E4B, 0xF716, 0xF893}, /* MAI CHATTAWA */
michael@0:     {0x0E4C, 0xF717, 0xF896}, /* THANTHAKHAT */
michael@0:     {0x0E31, 0xF710, 0xF884}, /* MAI HAN-AKAT */
michael@0:     {0x0E34, 0xF701, 0xF885}, /* SARA I */
michael@0:     {0x0E35, 0xF702, 0xF886}, /* SARA II */
michael@0:     {0x0E36, 0xF703, 0xF887}, /* SARA UE */
michael@0:     {0x0E37, 0xF704, 0xF888}, /* SARA UEE */
michael@0:     {0x0E47, 0xF712, 0xF889}, /* MAITAIKHU */
michael@0:     {0x0E4D, 0xF711, 0xF899}, /* NIKHAHIT */
michael@0:     {0x0000, 0x0000, 0x0000}
michael@0:   };
michael@0:   static const thai_pua_mapping_t RD_mappings[] = {
michael@0:     {0x0E0D, 0xF70F, 0xF89A}, /* YO YING */
michael@0:     {0x0E10, 0xF700, 0xF89E}, /* THO THAN */
michael@0:     {0x0000, 0x0000, 0x0000}
michael@0:   };
michael@0: 
michael@0:   switch (action) {
michael@0:     default: assert (false); /* Fallthrough */
michael@0:     case NOP: return u;
michael@0:     case SD:  pua_mappings = SD_mappings; break;
michael@0:     case SDL: pua_mappings = SDL_mappings; break;
michael@0:     case SL:  pua_mappings = SL_mappings; break;
michael@0:     case RD:  pua_mappings = RD_mappings; break;
michael@0:   }
michael@0:   for (; pua_mappings->u; pua_mappings++)
michael@0:     if (pua_mappings->u == u)
michael@0:     {
michael@0:       hb_codepoint_t glyph;
michael@0:       if (hb_font_get_glyph (font, pua_mappings->win_pua, 0, &glyph))
michael@0: 	return pua_mappings->win_pua;
michael@0:       if (hb_font_get_glyph (font, pua_mappings->mac_pua, 0, &glyph))
michael@0: 	return pua_mappings->mac_pua;
michael@0:       break;
michael@0:     }
michael@0:   return u;
michael@0: }
michael@0: 
michael@0: 
michael@0: static enum thai_above_state_t
michael@0: {     /* Cluster above looks like: */
michael@0:   T0, /*  ⣤                      */
michael@0:   T1, /*     ⣼                   */
michael@0:   T2, /*        ⣾                */
michael@0:   T3, /*           ⣿             */
michael@0:   NUM_ABOVE_STATES
michael@0: } thai_above_start_state[NUM_CONSONANT_TYPES + 1/* For NOT_CONSONANT */] =
michael@0: {
michael@0:   T0, /* NC */
michael@0:   T1, /* AC */
michael@0:   T0, /* RC */
michael@0:   T0, /* DC */
michael@0:   T3, /* NOT_CONSONANT */
michael@0: };
michael@0: 
michael@0: static const struct thai_above_state_machine_edge_t {
michael@0:   thai_action_t action;
michael@0:   thai_above_state_t next_state;
michael@0: } thai_above_state_machine[NUM_ABOVE_STATES][NUM_MARK_TYPES] =
michael@0: {        /*AV*/    /*BV*/    /*T*/
michael@0: /*T0*/ {{NOP,T3}, {NOP,T0}, {SD, T3}},
michael@0: /*T1*/ {{SL, T2}, {NOP,T1}, {SDL,T2}},
michael@0: /*T2*/ {{NOP,T3}, {NOP,T2}, {SL, T3}},
michael@0: /*T3*/ {{NOP,T3}, {NOP,T3}, {NOP,T3}},
michael@0: };
michael@0: 
michael@0: 
michael@0: static enum thai_below_state_t
michael@0: {
michael@0:   B0, /* No descender */
michael@0:   B1, /* Removable descender */
michael@0:   B2, /* Strict descender */
michael@0:   NUM_BELOW_STATES
michael@0: } thai_below_start_state[NUM_CONSONANT_TYPES + 1/* For NOT_CONSONANT */] =
michael@0: {
michael@0:   B0, /* NC */
michael@0:   B0, /* AC */
michael@0:   B1, /* RC */
michael@0:   B2, /* DC */
michael@0:   B2, /* NOT_CONSONANT */
michael@0: };
michael@0: 
michael@0: static const struct thai_below_state_machine_edge_t {
michael@0:   thai_action_t action;
michael@0:   thai_below_state_t next_state;
michael@0: } thai_below_state_machine[NUM_BELOW_STATES][NUM_MARK_TYPES] =
michael@0: {        /*AV*/    /*BV*/    /*T*/
michael@0: /*B0*/ {{NOP,B0}, {NOP,B2}, {NOP, B0}},
michael@0: /*B1*/ {{NOP,B1}, {RD, B2}, {NOP, B1}},
michael@0: /*B2*/ {{NOP,B2}, {SD, B2}, {NOP, B2}},
michael@0: };
michael@0: 
michael@0: 
michael@0: static void
michael@0: do_thai_pua_shaping (const hb_ot_shape_plan_t *plan HB_UNUSED,
michael@0: 		     hb_buffer_t              *buffer,
michael@0: 		     hb_font_t                *font)
michael@0: {
michael@0:   thai_above_state_t above_state = thai_above_start_state[NOT_CONSONANT];
michael@0:   thai_below_state_t below_state = thai_below_start_state[NOT_CONSONANT];
michael@0:   unsigned int base = 0;
michael@0: 
michael@0:   hb_glyph_info_t *info = buffer->info;
michael@0:   unsigned int count = buffer->len;
michael@0:   for (unsigned int i = 0; i < count; i++)
michael@0:   {
michael@0:     thai_mark_type_t mt = get_mark_type (info[i].codepoint);
michael@0: 
michael@0:     if (mt == NOT_MARK) {
michael@0:       thai_consonant_type_t ct = get_consonant_type (info[i].codepoint);
michael@0:       above_state = thai_above_start_state[ct];
michael@0:       below_state = thai_below_start_state[ct];
michael@0:       base = i;
michael@0:       continue;
michael@0:     }
michael@0: 
michael@0:     const thai_above_state_machine_edge_t &above_edge = thai_above_state_machine[above_state][mt];
michael@0:     const thai_below_state_machine_edge_t &below_edge = thai_below_state_machine[below_state][mt];
michael@0:     above_state = above_edge.next_state;
michael@0:     below_state = below_edge.next_state;
michael@0: 
michael@0:     /* At least one of the above/below actions is NOP. */
michael@0:     thai_action_t action = above_edge.action != NOP ? above_edge.action : below_edge.action;
michael@0: 
michael@0:     if (action == RD)
michael@0:       info[base].codepoint = thai_pua_shape (info[base].codepoint, action, font);
michael@0:     else
michael@0:       info[i].codepoint = thai_pua_shape (info[i].codepoint, action, font);
michael@0:   }
michael@0: }
michael@0: 
michael@0: 
michael@0: static void
michael@0: preprocess_text_thai (const hb_ot_shape_plan_t *plan,
michael@0: 		      hb_buffer_t              *buffer,
michael@0: 		      hb_font_t                *font)
michael@0: {
michael@0:   /* This function implements the shaping logic documented here:
michael@0:    *
michael@0:    *   http://linux.thai.net/~thep/th-otf/shaping.html
michael@0:    *
michael@0:    * The first shaping rule listed there is needed even if the font has Thai
michael@0:    * OpenType tables.  The rest do fallback positioning based on PUA codepoints.
michael@0:    * We implement that only if there exist no Thai GSUB in the font.
michael@0:    */
michael@0: 
michael@0:   /* The following is NOT specified in the MS OT Thai spec, however, it seems
michael@0:    * to be what Uniscribe and other engines implement.  According to Eric Muller:
michael@0:    *
michael@0:    * When you have a SARA AM, decompose it in NIKHAHIT + SARA AA, *and* move the
michael@0:    * NIKHAHIT backwards over any tone mark (0E48-0E4B).
michael@0:    *
michael@0:    * <0E14, 0E4B, 0E33> -> <0E14, 0E4D, 0E4B, 0E32>
michael@0:    *
michael@0:    * This reordering is legit only when the NIKHAHIT comes from a SARA AM, not
michael@0:    * when it's there to start with. The string <0E14, 0E4B, 0E4D> is probably
michael@0:    * not what a user wanted, but the rendering is nevertheless nikhahit above
michael@0:    * chattawa.
michael@0:    *
michael@0:    * Same for Lao.
michael@0:    *
michael@0:    * Note:
michael@0:    *
michael@0:    * Uniscribe also does some below-marks reordering.  Namely, it positions U+0E3A
michael@0:    * after U+0E38 and U+0E39.  We do that by modifying the ccc for U+0E3A.
michael@0:    * See unicode->modified_combining_class ().  Lao does NOT have a U+0E3A
michael@0:    * equivalent.
michael@0:    */
michael@0: 
michael@0: 
michael@0:   /*
michael@0:    * Here are the characters of significance:
michael@0:    *
michael@0:    *			Thai	Lao
michael@0:    * SARA AM:		U+0E33	U+0EB3
michael@0:    * SARA AA:		U+0E32	U+0EB2
michael@0:    * Nikhahit:		U+0E4D	U+0ECD
michael@0:    *
michael@0:    * Testing shows that Uniscribe reorder the following marks:
michael@0:    * Thai:	<0E31,0E34..0E37,0E47..0E4E>
michael@0:    * Lao:	<0EB1,0EB4..0EB7,0EC7..0ECE>
michael@0:    *
michael@0:    * Note how the Lao versions are the same as Thai + 0x80.
michael@0:    */
michael@0: 
michael@0:   /* We only get one script at a time, so a script-agnostic implementation
michael@0:    * is adequate here. */
michael@0: #define IS_SARA_AM(x) (((x) & ~0x0080) == 0x0E33)
michael@0: #define NIKHAHIT_FROM_SARA_AM(x) ((x) - 0xE33 + 0xE4D)
michael@0: #define SARA_AA_FROM_SARA_AM(x) ((x) - 1)
michael@0: #define IS_TONE_MARK(x) (hb_in_ranges<hb_codepoint_t> ((x) & ~0x0080, 0x0E34, 0x0E37, 0x0E47, 0x0E4E, 0x0E31, 0x0E31))
michael@0: 
michael@0:   buffer->clear_output ();
michael@0:   unsigned int count = buffer->len;
michael@0:   for (buffer->idx = 0; buffer->idx < count;)
michael@0:   {
michael@0:     hb_codepoint_t u = buffer->cur().codepoint;
michael@0:     if (likely (!IS_SARA_AM (u))) {
michael@0:       buffer->next_glyph ();
michael@0:       continue;
michael@0:     }
michael@0: 
michael@0:     /* Is SARA AM. Decompose and reorder. */
michael@0:     hb_codepoint_t decomposed[2] = {hb_codepoint_t (NIKHAHIT_FROM_SARA_AM (u)),
michael@0: 				    hb_codepoint_t (SARA_AA_FROM_SARA_AM (u))};
michael@0:     buffer->replace_glyphs (1, 2, decomposed);
michael@0:     if (unlikely (buffer->in_error))
michael@0:       return;
michael@0: 
michael@0:     /* Ok, let's see... */
michael@0:     unsigned int end = buffer->out_len;
michael@0:     unsigned int start = end - 2;
michael@0:     while (start > 0 && IS_TONE_MARK (buffer->out_info[start - 1].codepoint))
michael@0:       start--;
michael@0: 
michael@0:     if (start + 2 < end)
michael@0:     {
michael@0:       /* Move Nikhahit (end-2) to the beginning */
michael@0:       buffer->merge_out_clusters (start, end);
michael@0:       hb_glyph_info_t t = buffer->out_info[end - 2];
michael@0:       memmove (buffer->out_info + start + 1,
michael@0: 	       buffer->out_info + start,
michael@0: 	       sizeof (buffer->out_info[0]) * (end - start - 2));
michael@0:       buffer->out_info[start] = t;
michael@0:     }
michael@0:     else
michael@0:     {
michael@0:       /* Since we decomposed, and NIKHAHIT is combining, merge clusters with the
michael@0:        * previous cluster. */
michael@0:       if (start)
michael@0: 	buffer->merge_out_clusters (start - 1, end);
michael@0:     }
michael@0:   }
michael@0:   buffer->swap_buffers ();
michael@0: 
michael@0:   /* If font has Thai GSUB, we are done. */
michael@0:   if (plan->props.script == HB_SCRIPT_THAI && !plan->map.found_script[0])
michael@0:     do_thai_pua_shaping (plan, buffer, font);
michael@0: }
michael@0: 
michael@0: const hb_ot_complex_shaper_t _hb_ot_complex_shaper_thai =
michael@0: {
michael@0:   "thai",
michael@0:   NULL, /* collect_features */
michael@0:   NULL, /* override_features */
michael@0:   NULL, /* data_create */
michael@0:   NULL, /* data_destroy */
michael@0:   preprocess_text_thai,
michael@0:   HB_OT_SHAPE_NORMALIZATION_MODE_DEFAULT,
michael@0:   NULL, /* decompose */
michael@0:   NULL, /* compose */
michael@0:   NULL, /* setup_masks */
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_DEFAULT,
michael@0:   false,/* fallback_position */
michael@0: };