michael@0: /*
michael@0:  * Copyright © 2013  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: /* Hangul shaper */
michael@0: 
michael@0: 
michael@0: /* Same order as the feature array below */
michael@0: enum {
michael@0:   NONE,
michael@0: 
michael@0:   LJMO,
michael@0:   VJMO,
michael@0:   TJMO,
michael@0: 
michael@0:   FIRST_HANGUL_FEATURE = LJMO,
michael@0:   HANGUL_FEATURE_COUNT = TJMO + 1
michael@0: };
michael@0: 
michael@0: static const hb_tag_t hangul_features[HANGUL_FEATURE_COUNT] =
michael@0: {
michael@0:   HB_TAG_NONE,
michael@0:   HB_TAG('l','j','m','o'),
michael@0:   HB_TAG('v','j','m','o'),
michael@0:   HB_TAG('t','j','m','o')
michael@0: };
michael@0: 
michael@0: static void
michael@0: collect_features_hangul (hb_ot_shape_planner_t *plan)
michael@0: {
michael@0:   hb_ot_map_builder_t *map = &plan->map;
michael@0: 
michael@0:   for (unsigned int i = FIRST_HANGUL_FEATURE; i < HANGUL_FEATURE_COUNT; i++)
michael@0:     map->add_feature (hangul_features[i], 1, F_NONE);
michael@0: }
michael@0: 
michael@0: struct hangul_shape_plan_t
michael@0: {
michael@0:   ASSERT_POD ();
michael@0: 
michael@0:   hb_mask_t mask_array[HANGUL_FEATURE_COUNT];
michael@0: };
michael@0: 
michael@0: static void *
michael@0: data_create_hangul (const hb_ot_shape_plan_t *plan)
michael@0: {
michael@0:   hangul_shape_plan_t *hangul_plan = (hangul_shape_plan_t *) calloc (1, sizeof (hangul_shape_plan_t));
michael@0:   if (unlikely (!hangul_plan))
michael@0:     return NULL;
michael@0: 
michael@0:   for (unsigned int i = 0; i < HANGUL_FEATURE_COUNT; i++)
michael@0:     hangul_plan->mask_array[i] = plan->map.get_1_mask (hangul_features[i]);
michael@0: 
michael@0:   return hangul_plan;
michael@0: }
michael@0: 
michael@0: static void
michael@0: data_destroy_hangul (void *data)
michael@0: {
michael@0:   free (data);
michael@0: }
michael@0: 
michael@0: /* Constants for algorithmic hangul syllable [de]composition. */
michael@0: #define LBase 0x1100
michael@0: #define VBase 0x1161
michael@0: #define TBase 0x11A7
michael@0: #define LCount 19
michael@0: #define VCount 21
michael@0: #define TCount 28
michael@0: #define SBase 0xAC00
michael@0: #define NCount (VCount * TCount)
michael@0: #define SCount (LCount * NCount)
michael@0: 
michael@0: #define isCombiningL(u) (hb_in_range<hb_codepoint_t> ((u), LBase, LBase+LCount-1))
michael@0: #define isCombiningV(u) (hb_in_range<hb_codepoint_t> ((u), VBase, VBase+VCount-1))
michael@0: #define isCombiningT(u) (hb_in_range<hb_codepoint_t> ((u), TBase+1, TBase+TCount-1))
michael@0: #define isCombinedS(u) (hb_in_range<hb_codepoint_t> ((u), SBase, SBase+SCount-1))
michael@0: 
michael@0: #define isL(u) (hb_in_ranges<hb_codepoint_t> ((u), 0x1100, 0x115F, 0xA960, 0xA97C))
michael@0: #define isV(u) (hb_in_ranges<hb_codepoint_t> ((u), 0x1160, 0x11A7, 0xD7B0, 0xD7C6))
michael@0: #define isT(u) (hb_in_ranges<hb_codepoint_t> ((u), 0x11A8, 0x11FF, 0xD7CB, 0xD7FB))
michael@0: 
michael@0: #define isHangulTone(u) (hb_in_range<hb_codepoint_t> ((u), 0x302e, 0x302f))
michael@0: 
michael@0: /* buffer var allocations */
michael@0: #define hangul_shaping_feature() complex_var_u8_0() /* hangul jamo shaping feature */
michael@0: 
michael@0: static bool
michael@0: is_zero_width_char (hb_font_t *font,
michael@0: 		    hb_codepoint_t unicode)
michael@0: {
michael@0:   hb_codepoint_t glyph;
michael@0:   return hb_font_get_glyph (font, unicode, 0, &glyph) && hb_font_get_glyph_h_advance (font, glyph) == 0;
michael@0: }
michael@0: 
michael@0: static void
michael@0: preprocess_text_hangul (const hb_ot_shape_plan_t *plan,
michael@0: 			hb_buffer_t              *buffer,
michael@0: 			hb_font_t                *font)
michael@0: {
michael@0:   HB_BUFFER_ALLOCATE_VAR (buffer, hangul_shaping_feature);
michael@0: 
michael@0:   /* Hangul syllables come in two shapes: LV, and LVT.  Of those:
michael@0:    *
michael@0:    *   - LV can be precomposed, or decomposed.  Lets call those
michael@0:    *     <LV> and <L,V>,
michael@0:    *   - LVT can be fully precomposed, partically precomposed, or
michael@0:    *     fully decomposed.  Ie. <LVT>, <LV,T>, or <L,V,T>.
michael@0:    *
michael@0:    * The composition / decomposition is mechanical.  However, not
michael@0:    * all <L,V> sequences compose, and not all <LV,T> sequences
michael@0:    * compose.
michael@0:    *
michael@0:    * Here are the specifics:
michael@0:    *
michael@0:    *   - <L>: U+1100..115F, U+A960..A97F
michael@0:    *   - <V>: U+1160..11A7, U+D7B0..D7C7
michael@0:    *   - <T>: U+11A8..11FF, U+D7CB..D7FB
michael@0:    *
michael@0:    *   - Only the <L,V> sequences for the 11xx ranges combine.
michael@0:    *   - Only <LV,T> sequences for T in U+11A8..11C3 combine.
michael@0:    *
michael@0:    * Here is what we want to accomplish in this shaper:
michael@0:    *
michael@0:    *   - If the whole syllable can be precomposed, do that,
michael@0:    *   - Otherwise, fully decompose and apply ljmo/vjmo/tjmo features.
michael@0:    *   - If a valid syllable is followed by a Hangul tone mark, reorder the tone
michael@0:    *     mark to precede the whole syllable - unless it is a zero-width glyph, in
michael@0:    *     which case we leave it untouched, assuming it's designed to overstrike.
michael@0:    *
michael@0:    * That is, of the different possible syllables:
michael@0:    *
michael@0:    *   <L>
michael@0:    *   <L,V>
michael@0:    *   <L,V,T>
michael@0:    *   <LV>
michael@0:    *   <LVT>
michael@0:    *   <LV, T>
michael@0:    *
michael@0:    * - <L> needs no work.
michael@0:    *
michael@0:    * - <LV> and <LVT> can stay the way they are if the font supports them, otherwise we
michael@0:    *   should fully decompose them if font supports.
michael@0:    *
michael@0:    * - <L,V> and <L,V,T> we should compose if the whole thing can be composed.
michael@0:    *
michael@0:    * - <LV,T> we should compose if the whole thing can be composed, otherwise we should
michael@0:    *   decompose.
michael@0:    */
michael@0: 
michael@0:   buffer->clear_output ();
michael@0:   unsigned int start = 0, end = 0; /* Extent of most recently seen syllable;
michael@0: 				    * valid only if start < end
michael@0: 				    */
michael@0:   unsigned int count = buffer->len;
michael@0: 
michael@0:   for (buffer->idx = 0; buffer->idx < count;)
michael@0:   {
michael@0:     hb_codepoint_t u = buffer->cur().codepoint;
michael@0: 
michael@0:     if (isHangulTone (u))
michael@0:     {
michael@0:       /*
michael@0:        * We could cache the width of the tone marks and the existence of dotted-circle,
michael@0:        * but the use of the Hangul tone mark characters seems to be rare enough that
michael@0:        * I didn't bother for now.
michael@0:        */
michael@0:       if (start < end && end == buffer->out_len)
michael@0:       {
michael@0: 	/* Tone mark follows a valid syllable; move it in front, unless it's zero width. */
michael@0: 	buffer->next_glyph ();
michael@0: 	if (!is_zero_width_char (font, u))
michael@0: 	{
michael@0: 	  hb_glyph_info_t *info = buffer->out_info;
michael@0: 	  hb_glyph_info_t tone = info[end];
michael@0: 	  memmove (&info[start + 1], &info[start], (end - start) * sizeof (hb_glyph_info_t));
michael@0: 	  info[start] = tone;
michael@0: 	}
michael@0: 	/* Merge clusters across the (possibly reordered) syllable+tone.
michael@0: 	 * We want to merge even in the zero-width tone mark case here,
michael@0: 	 * so that clustering behavior isn't dependent on how the tone mark
michael@0: 	 * is handled by the font.
michael@0: 	 */
michael@0: 	buffer->merge_out_clusters (start, end + 1);
michael@0:       }
michael@0:       else
michael@0:       {
michael@0: 	/* No valid syllable as base for tone mark; try to insert dotted circle. */
michael@0: 	if (font->has_glyph (0x25cc))
michael@0: 	{
michael@0: 	  hb_codepoint_t chars[2];
michael@0: 	  if (!is_zero_width_char (font, u)) {
michael@0: 	    chars[0] = u;
michael@0: 	    chars[1] = 0x25cc;
michael@0: 	  } else {
michael@0: 	    chars[0] = 0x25cc;
michael@0: 	    chars[1] = u;
michael@0: 	  }
michael@0: 	  buffer->replace_glyphs (1, 2, chars);
michael@0: 	}
michael@0: 	else
michael@0: 	{
michael@0: 	  /* No dotted circle available in the font; just leave tone mark untouched. */
michael@0: 	  buffer->next_glyph ();
michael@0: 	}
michael@0:       }
michael@0:       start = end = buffer->out_len;
michael@0:       continue;
michael@0:     }
michael@0: 
michael@0:     start = buffer->out_len; /* Remember current position as a potential syllable start;
michael@0: 			      * will only be used if we set end to a later position.
michael@0: 			      */
michael@0: 
michael@0:     if (isL (u) && buffer->idx + 1 < count)
michael@0:     {
michael@0:       hb_codepoint_t l = u;
michael@0:       hb_codepoint_t v = buffer->cur(+1).codepoint;
michael@0:       if (isV (v))
michael@0:       {
michael@0: 	/* Have <L,V> or <L,V,T>. */
michael@0: 	hb_codepoint_t t = 0;
michael@0: 	unsigned int tindex = 0;
michael@0: 	if (buffer->idx + 2 < count)
michael@0: 	{
michael@0: 	  t = buffer->cur(+2).codepoint;
michael@0: 	  if (isT (t))
michael@0: 	    tindex = t - TBase; /* Only used if isCombiningT (t); otherwise invalid. */
michael@0: 	  else
michael@0: 	    t = 0; /* The next character was not a trailing jamo. */
michael@0: 	}
michael@0: 
michael@0: 	/* We've got a syllable <L,V,T?>; see if it can potentially be composed. */
michael@0: 	if (isCombiningL (l) && isCombiningV (v) && (t == 0 || isCombiningT (t)))
michael@0: 	{
michael@0: 	  /* Try to compose; if this succeeds, end is set to start+1. */
michael@0: 	  hb_codepoint_t s = SBase + (l - LBase) * NCount + (v - VBase) * TCount + tindex;
michael@0: 	  if (font->has_glyph (s))
michael@0: 	  {
michael@0: 	    buffer->replace_glyphs (t ? 3 : 2, 1, &s);
michael@0: 	    if (unlikely (buffer->in_error))
michael@0: 	      return;
michael@0: 	    end = start + 1;
michael@0: 	    continue;
michael@0: 	  }
michael@0: 	}
michael@0: 
michael@0: 	/* We didn't compose, either because it's an Old Hangul syllable without a
michael@0: 	 * precomposed character in Unicode, or because the font didn't support the
michael@0: 	 * necessary precomposed glyph.
michael@0: 	 * Set jamo features on the individual glyphs, and advance past them.
michael@0: 	 */
michael@0: 	buffer->cur().hangul_shaping_feature() = LJMO;
michael@0: 	buffer->next_glyph ();
michael@0: 	buffer->cur().hangul_shaping_feature() = VJMO;
michael@0: 	buffer->next_glyph ();
michael@0: 	if (t)
michael@0: 	{
michael@0: 	  buffer->cur().hangul_shaping_feature() = TJMO;
michael@0: 	  buffer->next_glyph ();
michael@0: 	  end = start + 3;
michael@0: 	}
michael@0: 	else
michael@0: 	  end = start + 2;
michael@0: 	buffer->merge_out_clusters (start, end);
michael@0: 	continue;
michael@0:       }
michael@0:     }
michael@0: 
michael@0:     else if (isCombinedS (u))
michael@0:     {
michael@0:       /* Have <LV>, <LVT>, or <LV,T> */
michael@0:       hb_codepoint_t s = u;
michael@0:       bool has_glyph = font->has_glyph (s);
michael@0:       unsigned int lindex = (s - SBase) / NCount;
michael@0:       unsigned int nindex = (s - SBase) % NCount;
michael@0:       unsigned int vindex = nindex / TCount;
michael@0:       unsigned int tindex = nindex % TCount;
michael@0: 
michael@0:       if (!tindex &&
michael@0: 	  buffer->idx + 1 < count &&
michael@0: 	  isCombiningT (buffer->cur(+1).codepoint))
michael@0:       {
michael@0: 	/* <LV,T>, try to combine. */
michael@0: 	unsigned int new_tindex = buffer->cur(+1).codepoint - TBase;
michael@0: 	hb_codepoint_t new_s = s + new_tindex;
michael@0: 	if (font->has_glyph (new_s))
michael@0: 	{
michael@0: 	  buffer->replace_glyphs (2, 1, &new_s);
michael@0: 	  if (unlikely (buffer->in_error))
michael@0: 	    return;
michael@0: 	  end = start + 1;
michael@0: 	  continue;
michael@0: 	}
michael@0:       }
michael@0: 
michael@0:       /* Otherwise, decompose if font doesn't support <LV> or <LVT>,
michael@0:        * or if having non-combining <LV,T>.  Note that we already handled
michael@0:        * combining <LV,T> above. */
michael@0:       if (!has_glyph ||
michael@0: 	  (!tindex &&
michael@0: 	   buffer->idx + 1 < count &&
michael@0: 	   isT (buffer->cur(+1).codepoint)))
michael@0:       {
michael@0: 	hb_codepoint_t decomposed[3] = {LBase + lindex,
michael@0: 					VBase + vindex,
michael@0: 					TBase + tindex};
michael@0: 	if (font->has_glyph (decomposed[0]) &&
michael@0: 	    font->has_glyph (decomposed[1]) &&
michael@0: 	    (!tindex || font->has_glyph (decomposed[2])))
michael@0: 	{
michael@0: 	  unsigned int s_len = tindex ? 3 : 2;
michael@0: 	  buffer->replace_glyphs (1, s_len, decomposed);
michael@0: 	  if (unlikely (buffer->in_error))
michael@0: 	    return;
michael@0: 
michael@0: 	  /* We decomposed S: apply jamo features to the individual glyphs
michael@0: 	   * that are now in buffer->out_info.
michael@0: 	   */
michael@0: 	  hb_glyph_info_t *info = buffer->out_info;
michael@0: 
michael@0: 	  /* If we decomposed an LV because of a non-combining T following,
michael@0: 	   * we want to include this T in the syllable.
michael@0: 	   */
michael@0: 	  if (has_glyph && !tindex)
michael@0: 	  {
michael@0:             buffer->next_glyph ();
michael@0:             s_len++;
michael@0:           }
michael@0:           end = start + s_len;
michael@0: 
michael@0: 	  unsigned int i = start;
michael@0: 	  info[i++].hangul_shaping_feature() = LJMO;
michael@0: 	  info[i++].hangul_shaping_feature() = VJMO;
michael@0: 	  if (i < end)
michael@0: 	    info[i++].hangul_shaping_feature() = TJMO;
michael@0: 	  buffer->merge_out_clusters (start, end);
michael@0: 	  continue;
michael@0: 	}
michael@0:       }
michael@0: 
michael@0:       if (has_glyph)
michael@0:       {
michael@0:         /* We didn't decompose the S, so just advance past it. */
michael@0: 	end = start + 1;
michael@0: 	buffer->next_glyph ();
michael@0: 	continue;
michael@0:       }
michael@0:     }
michael@0: 
michael@0:     /* Didn't find a recognizable syllable, so we leave end <= start;
michael@0:      * this will prevent tone-mark reordering happening.
michael@0:      */
michael@0:     buffer->next_glyph ();
michael@0:   }
michael@0:   buffer->swap_buffers ();
michael@0: }
michael@0: 
michael@0: static void
michael@0: setup_masks_hangul (const hb_ot_shape_plan_t *plan,
michael@0: 		    hb_buffer_t              *buffer,
michael@0: 		    hb_font_t                *font HB_UNUSED)
michael@0: {
michael@0:   const hangul_shape_plan_t *hangul_plan = (const hangul_shape_plan_t *) plan->data;
michael@0: 
michael@0:   if (likely (hangul_plan))
michael@0:   {
michael@0:     unsigned int count = buffer->len;
michael@0:     hb_glyph_info_t *info = buffer->info;
michael@0:     for (unsigned int i = 0; i < count; i++, info++)
michael@0:       info->mask |= hangul_plan->mask_array[info->hangul_shaping_feature()];
michael@0:   }
michael@0: 
michael@0:   HB_BUFFER_DEALLOCATE_VAR (buffer, hangul_shaping_feature);
michael@0: }
michael@0: 
michael@0: 
michael@0: const hb_ot_complex_shaper_t _hb_ot_complex_shaper_hangul =
michael@0: {
michael@0:   "hangul",
michael@0:   collect_features_hangul,
michael@0:   NULL, /* override_features */
michael@0:   data_create_hangul, /* data_create */
michael@0:   data_destroy_hangul, /* data_destroy */
michael@0:   preprocess_text_hangul,
michael@0:   HB_OT_SHAPE_NORMALIZATION_MODE_NONE,
michael@0:   NULL, /* decompose */
michael@0:   NULL, /* compose */
michael@0:   setup_masks_hangul, /* setup_masks */
michael@0:   HB_OT_SHAPE_ZERO_WIDTH_MARKS_NONE,
michael@0:   false, /* fallback_position */
michael@0: };