Edit

kc3-lang/harfbuzz/src/hb-ot-shape-fallback.cc

Branch :

  • Show log

    Commit

  • Author : Behdad Esfahbod
    Date : 2022-01-29 10:24:38
    Hash : b47b3b99
    Message : [fallback-kern] Move buffer message to correct position

  • src/hb-ot-shape-fallback.cc
  • /*
     * Copyright © 2011,2012  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.hh"
    
    #ifndef HB_NO_OT_SHAPE
    
    #include "hb-ot-shape-fallback.hh"
    #include "hb-kern.hh"
    
    static unsigned int
    recategorize_combining_class (hb_codepoint_t u,
    			      unsigned int klass)
    {
      if (klass >= 200)
        return klass;
    
      /* Thai / Lao need some per-character work. */
      if ((u & ~0xFF) == 0x0E00u)
      {
        if (unlikely (klass == 0))
        {
          switch (u)
          {
    	case 0x0E31u:
    	case 0x0E34u:
    	case 0x0E35u:
    	case 0x0E36u:
    	case 0x0E37u:
    	case 0x0E47u:
    	case 0x0E4Cu:
    	case 0x0E4Du:
    	case 0x0E4Eu:
    	  klass = HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT;
    	  break;
    
    	case 0x0EB1u:
    	case 0x0EB4u:
    	case 0x0EB5u:
    	case 0x0EB6u:
    	case 0x0EB7u:
    	case 0x0EBBu:
    	case 0x0ECCu:
    	case 0x0ECDu:
    	  klass = HB_UNICODE_COMBINING_CLASS_ABOVE;
    	  break;
    
    	case 0x0EBCu:
    	  klass = HB_UNICODE_COMBINING_CLASS_BELOW;
    	  break;
          }
        } else {
          /* Thai virama is below-right */
          if (u == 0x0E3Au)
    	klass = HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT;
        }
      }
    
      switch (klass)
      {
    
        /* Hebrew */
    
        case HB_MODIFIED_COMBINING_CLASS_CCC10: /* sheva */
        case HB_MODIFIED_COMBINING_CLASS_CCC11: /* hataf segol */
        case HB_MODIFIED_COMBINING_CLASS_CCC12: /* hataf patah */
        case HB_MODIFIED_COMBINING_CLASS_CCC13: /* hataf qamats */
        case HB_MODIFIED_COMBINING_CLASS_CCC14: /* hiriq */
        case HB_MODIFIED_COMBINING_CLASS_CCC15: /* tsere */
        case HB_MODIFIED_COMBINING_CLASS_CCC16: /* segol */
        case HB_MODIFIED_COMBINING_CLASS_CCC17: /* patah */
        case HB_MODIFIED_COMBINING_CLASS_CCC18: /* qamats & qamats qatan */
        case HB_MODIFIED_COMBINING_CLASS_CCC20: /* qubuts */
        case HB_MODIFIED_COMBINING_CLASS_CCC22: /* meteg */
          return HB_UNICODE_COMBINING_CLASS_BELOW;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC23: /* rafe */
          return HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC24: /* shin dot */
          return HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC25: /* sin dot */
        case HB_MODIFIED_COMBINING_CLASS_CCC19: /* holam & holam haser for vav */
          return HB_UNICODE_COMBINING_CLASS_ABOVE_LEFT;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC26: /* point varika */
          return HB_UNICODE_COMBINING_CLASS_ABOVE;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC21: /* dagesh */
          break;
    
    
        /* Arabic and Syriac */
    
        case HB_MODIFIED_COMBINING_CLASS_CCC27: /* fathatan */
        case HB_MODIFIED_COMBINING_CLASS_CCC28: /* dammatan */
        case HB_MODIFIED_COMBINING_CLASS_CCC30: /* fatha */
        case HB_MODIFIED_COMBINING_CLASS_CCC31: /* damma */
        case HB_MODIFIED_COMBINING_CLASS_CCC33: /* shadda */
        case HB_MODIFIED_COMBINING_CLASS_CCC34: /* sukun */
        case HB_MODIFIED_COMBINING_CLASS_CCC35: /* superscript alef */
        case HB_MODIFIED_COMBINING_CLASS_CCC36: /* superscript alaph */
          return HB_UNICODE_COMBINING_CLASS_ABOVE;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC29: /* kasratan */
        case HB_MODIFIED_COMBINING_CLASS_CCC32: /* kasra */
          return HB_UNICODE_COMBINING_CLASS_BELOW;
    
    
        /* Thai */
    
        case HB_MODIFIED_COMBINING_CLASS_CCC103: /* sara u / sara uu */
          return HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC107: /* mai */
          return HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT;
    
    
        /* Lao */
    
        case HB_MODIFIED_COMBINING_CLASS_CCC118: /* sign u / sign uu */
          return HB_UNICODE_COMBINING_CLASS_BELOW;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC122: /* mai */
          return HB_UNICODE_COMBINING_CLASS_ABOVE;
    
    
        /* Tibetan */
    
        case HB_MODIFIED_COMBINING_CLASS_CCC129: /* sign aa */
          return HB_UNICODE_COMBINING_CLASS_BELOW;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC130: /* sign i*/
          return HB_UNICODE_COMBINING_CLASS_ABOVE;
    
        case HB_MODIFIED_COMBINING_CLASS_CCC132: /* sign u */
          return HB_UNICODE_COMBINING_CLASS_BELOW;
    
      }
    
      return klass;
    }
    
    void
    _hb_ot_shape_fallback_mark_position_recategorize_marks (const hb_ot_shape_plan_t *plan HB_UNUSED,
    							hb_font_t *font HB_UNUSED,
    							hb_buffer_t  *buffer)
    {
    #ifdef HB_NO_OT_SHAPE_FALLBACK
      return;
    #endif
    
      unsigned int count = buffer->len;
      hb_glyph_info_t *info = buffer->info;
      for (unsigned int i = 0; i < count; i++)
        if (_hb_glyph_info_get_general_category (&info[i]) == HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK) {
          unsigned int combining_class = _hb_glyph_info_get_modified_combining_class (&info[i]);
          combining_class = recategorize_combining_class (info[i].codepoint, combining_class);
          _hb_glyph_info_set_modified_combining_class (&info[i], combining_class);
        }
    }
    
    
    static void
    zero_mark_advances (hb_buffer_t *buffer,
    		    unsigned int start,
    		    unsigned int end,
    		    bool adjust_offsets_when_zeroing)
    {
      hb_glyph_info_t *info = buffer->info;
      for (unsigned int i = start; i < end; i++)
        if (_hb_glyph_info_get_general_category (&info[i]) == HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)
        {
          if (adjust_offsets_when_zeroing)
          {
    	buffer->pos[i].x_offset -= buffer->pos[i].x_advance;
    	buffer->pos[i].y_offset -= buffer->pos[i].y_advance;
          }
          buffer->pos[i].x_advance = 0;
          buffer->pos[i].y_advance = 0;
        }
    }
    
    static inline void
    position_mark (const hb_ot_shape_plan_t *plan HB_UNUSED,
    	       hb_font_t *font,
    	       hb_buffer_t  *buffer,
    	       hb_glyph_extents_t &base_extents,
    	       unsigned int i,
    	       unsigned int combining_class)
    {
      hb_glyph_extents_t mark_extents;
      if (!font->get_glyph_extents (buffer->info[i].codepoint, &mark_extents))
        return;
    
      hb_position_t y_gap = font->y_scale / 16;
    
      hb_glyph_position_t &pos = buffer->pos[i];
      pos.x_offset = pos.y_offset = 0;
    
    
      /* We don't position LEFT and RIGHT marks. */
    
      /* X positioning */
      switch (combining_class)
      {
        case HB_UNICODE_COMBINING_CLASS_DOUBLE_BELOW:
        case HB_UNICODE_COMBINING_CLASS_DOUBLE_ABOVE:
          if (buffer->props.direction == HB_DIRECTION_LTR) {
    	pos.x_offset += base_extents.x_bearing + base_extents.width - mark_extents.width / 2 - mark_extents.x_bearing;
    	break;
          } else if (buffer->props.direction == HB_DIRECTION_RTL) {
    	pos.x_offset += base_extents.x_bearing - mark_extents.width / 2 - mark_extents.x_bearing;
    	break;
          }
          HB_FALLTHROUGH;
    
        default:
        case HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW:
        case HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE:
        case HB_UNICODE_COMBINING_CLASS_BELOW:
        case HB_UNICODE_COMBINING_CLASS_ABOVE:
          /* Center align. */
          pos.x_offset += base_extents.x_bearing + (base_extents.width - mark_extents.width) / 2 - mark_extents.x_bearing;
          break;
    
        case HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW_LEFT:
        case HB_UNICODE_COMBINING_CLASS_BELOW_LEFT:
        case HB_UNICODE_COMBINING_CLASS_ABOVE_LEFT:
          /* Left align. */
          pos.x_offset += base_extents.x_bearing - mark_extents.x_bearing;
          break;
    
        case HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE_RIGHT:
        case HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT:
        case HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT:
          /* Right align. */
          pos.x_offset += base_extents.x_bearing + base_extents.width - mark_extents.width - mark_extents.x_bearing;
          break;
      }
    
      /* Y positioning */
      switch (combining_class)
      {
        case HB_UNICODE_COMBINING_CLASS_DOUBLE_BELOW:
        case HB_UNICODE_COMBINING_CLASS_BELOW_LEFT:
        case HB_UNICODE_COMBINING_CLASS_BELOW:
        case HB_UNICODE_COMBINING_CLASS_BELOW_RIGHT:
          /* Add gap, fall-through. */
          base_extents.height -= y_gap;
          HB_FALLTHROUGH;
    
        case HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW_LEFT:
        case HB_UNICODE_COMBINING_CLASS_ATTACHED_BELOW:
          pos.y_offset = base_extents.y_bearing + base_extents.height - mark_extents.y_bearing;
          /* Never shift up "below" marks. */
          if ((y_gap > 0) == (pos.y_offset > 0))
          {
    	base_extents.height -= pos.y_offset;
    	pos.y_offset = 0;
          }
          base_extents.height += mark_extents.height;
          break;
    
        case HB_UNICODE_COMBINING_CLASS_DOUBLE_ABOVE:
        case HB_UNICODE_COMBINING_CLASS_ABOVE_LEFT:
        case HB_UNICODE_COMBINING_CLASS_ABOVE:
        case HB_UNICODE_COMBINING_CLASS_ABOVE_RIGHT:
          /* Add gap, fall-through. */
          base_extents.y_bearing += y_gap;
          base_extents.height -= y_gap;
          HB_FALLTHROUGH;
    
        case HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE:
        case HB_UNICODE_COMBINING_CLASS_ATTACHED_ABOVE_RIGHT:
          pos.y_offset = base_extents.y_bearing - (mark_extents.y_bearing + mark_extents.height);
          /* Don't shift down "above" marks too much. */
          if ((y_gap > 0) != (pos.y_offset > 0))
          {
    	int correction = -pos.y_offset / 2;
    	base_extents.y_bearing += correction;
    	base_extents.height -= correction;
    	pos.y_offset += correction;
          }
          base_extents.y_bearing -= mark_extents.height;
          base_extents.height += mark_extents.height;
          break;
      }
    }
    
    static inline void
    position_around_base (const hb_ot_shape_plan_t *plan,
    		      hb_font_t *font,
    		      hb_buffer_t  *buffer,
    		      unsigned int base,
    		      unsigned int end,
    		      bool adjust_offsets_when_zeroing)
    {
      hb_direction_t horiz_dir = HB_DIRECTION_INVALID;
    
      buffer->unsafe_to_break (base, end);
    
      hb_glyph_extents_t base_extents;
      if (!font->get_glyph_extents (buffer->info[base].codepoint,
    				&base_extents))
      {
        /* If extents don't work, zero marks and go home. */
        zero_mark_advances (buffer, base + 1, end, adjust_offsets_when_zeroing);
        return;
      }
      base_extents.y_bearing += buffer->pos[base].y_offset;
      /* Use horizontal advance for horizontal positioning.
       * Generally a better idea.  Also works for zero-ink glyphs.  See:
       * https://github.com/harfbuzz/harfbuzz/issues/1532 */
      base_extents.x_bearing = 0;
      base_extents.width = font->get_glyph_h_advance (buffer->info[base].codepoint);
    
      unsigned int lig_id = _hb_glyph_info_get_lig_id (&buffer->info[base]);
      /* Use integer for num_lig_components such that it doesn't convert to unsigned
       * when we divide or multiply by it. */
      int num_lig_components = _hb_glyph_info_get_lig_num_comps (&buffer->info[base]);
    
      hb_position_t x_offset = 0, y_offset = 0;
      if (HB_DIRECTION_IS_FORWARD (buffer->props.direction)) {
        x_offset -= buffer->pos[base].x_advance;
        y_offset -= buffer->pos[base].y_advance;
      }
    
      hb_glyph_extents_t component_extents = base_extents;
      int last_lig_component = -1;
      unsigned int last_combining_class = 255;
      hb_glyph_extents_t cluster_extents = base_extents; /* Initialization is just to shut gcc up. */
      hb_glyph_info_t *info = buffer->info;
      for (unsigned int i = base + 1; i < end; i++)
        if (_hb_glyph_info_get_modified_combining_class (&info[i]))
        {
          if (num_lig_components > 1) {
    	unsigned int this_lig_id = _hb_glyph_info_get_lig_id (&info[i]);
    	int this_lig_component = _hb_glyph_info_get_lig_comp (&info[i]) - 1;
    	/* Conditions for attaching to the last component. */
    	if (!lig_id || lig_id != this_lig_id || this_lig_component >= num_lig_components)
    	  this_lig_component = num_lig_components - 1;
    	if (last_lig_component != this_lig_component)
    	{
    	  last_lig_component = this_lig_component;
    	  last_combining_class = 255;
    	  component_extents = base_extents;
    	  if (unlikely (horiz_dir == HB_DIRECTION_INVALID)) {
    	    if (HB_DIRECTION_IS_HORIZONTAL (plan->props.direction))
    	      horiz_dir = plan->props.direction;
    	    else
    	      horiz_dir = hb_script_get_horizontal_direction (plan->props.script);
    	  }
    	  if (horiz_dir == HB_DIRECTION_LTR)
    	    component_extents.x_bearing += (this_lig_component * component_extents.width) / num_lig_components;
    	  else
    	    component_extents.x_bearing += ((num_lig_components - 1 - this_lig_component) * component_extents.width) / num_lig_components;
    	  component_extents.width /= num_lig_components;
    	}
          }
    
          unsigned int this_combining_class = _hb_glyph_info_get_modified_combining_class (&info[i]);
          if (last_combining_class != this_combining_class)
          {
    	last_combining_class = this_combining_class;
    	cluster_extents = component_extents;
          }
    
          position_mark (plan, font, buffer, cluster_extents, i, this_combining_class);
    
          buffer->pos[i].x_advance = 0;
          buffer->pos[i].y_advance = 0;
          buffer->pos[i].x_offset += x_offset;
          buffer->pos[i].y_offset += y_offset;
    
        } else {
          if (HB_DIRECTION_IS_FORWARD (buffer->props.direction)) {
    	x_offset -= buffer->pos[i].x_advance;
    	y_offset -= buffer->pos[i].y_advance;
          } else {
    	x_offset += buffer->pos[i].x_advance;
    	y_offset += buffer->pos[i].y_advance;
          }
        }
    }
    
    static inline void
    position_cluster (const hb_ot_shape_plan_t *plan,
    		  hb_font_t *font,
    		  hb_buffer_t  *buffer,
    		  unsigned int start,
    		  unsigned int end,
    		  bool adjust_offsets_when_zeroing)
    {
      if (end - start < 2)
        return;
    
      /* Find the base glyph */
      hb_glyph_info_t *info = buffer->info;
      for (unsigned int i = start; i < end; i++)
        if (!_hb_glyph_info_is_unicode_mark (&info[i]))
        {
          /* Find mark glyphs */
          unsigned int j;
          for (j = i + 1; j < end; j++)
    	if (!_hb_glyph_info_is_unicode_mark (&info[j]))
    	  break;
    
          position_around_base (plan, font, buffer, i, j, adjust_offsets_when_zeroing);
    
          i = j - 1;
        }
    }
    
    void
    _hb_ot_shape_fallback_mark_position (const hb_ot_shape_plan_t *plan,
    				     hb_font_t *font,
    				     hb_buffer_t  *buffer,
    				     bool adjust_offsets_when_zeroing)
    {
    #ifdef HB_NO_OT_SHAPE_FALLBACK
      return;
    #endif
    
      if (!buffer->message (font, "start fallback mark"))
        return;
    
      _hb_buffer_assert_gsubgpos_vars (buffer);
    
      unsigned int start = 0;
      unsigned int count = buffer->len;
      hb_glyph_info_t *info = buffer->info;
      for (unsigned int i = 1; i < count; i++)
        if (likely (!_hb_glyph_info_is_unicode_mark (&info[i]))) {
          position_cluster (plan, font, buffer, start, i, adjust_offsets_when_zeroing);
          start = i;
        }
      position_cluster (plan, font, buffer, start, count, adjust_offsets_when_zeroing);
    
      (void) buffer->message (font, "end fallback mark");
    }
    
    
    #ifndef HB_DISABLE_DEPRECATED
    struct hb_ot_shape_fallback_kern_driver_t
    {
      hb_ot_shape_fallback_kern_driver_t (hb_font_t   *font_,
    				      hb_buffer_t *buffer) :
        font (font_), direction (buffer->props.direction) {}
    
      hb_position_t get_kerning (hb_codepoint_t first, hb_codepoint_t second) const
      {
        hb_position_t kern = 0;
        font->get_glyph_kerning_for_direction (first, second,
    					   direction,
    					   &kern, &kern);
        return kern;
      }
    
      hb_font_t *font;
      hb_direction_t direction;
    };
    #endif
    
    /* Performs font-assisted kerning. */
    void
    _hb_ot_shape_fallback_kern (const hb_ot_shape_plan_t *plan,
    			    hb_font_t *font,
    			    hb_buffer_t *buffer)
    {
    #ifdef HB_NO_OT_SHAPE_FALLBACK
      return;
    #endif
    
    #ifndef HB_DISABLE_DEPRECATED
      if (HB_DIRECTION_IS_HORIZONTAL (buffer->props.direction) ?
          !font->has_glyph_h_kerning_func () :
          !font->has_glyph_v_kerning_func ())
        return;
    
      if (!buffer->message (font, "start fallback kern"))
        return;
    
      bool reverse = HB_DIRECTION_IS_BACKWARD (buffer->props.direction);
    
      if (reverse)
        buffer->reverse ();
    
      hb_ot_shape_fallback_kern_driver_t driver (font, buffer);
      OT::hb_kern_machine_t<hb_ot_shape_fallback_kern_driver_t> machine (driver);
      machine.kern (font, buffer, plan->kern_mask, false);
    
      if (reverse)
        buffer->reverse ();
    
      (void) buffer->message (font, "end fallback kern");
    #endif
    }
    
    
    /* Adjusts width of various spaces. */
    void
    _hb_ot_shape_fallback_spaces (const hb_ot_shape_plan_t *plan HB_UNUSED,
    			      hb_font_t *font,
    			      hb_buffer_t  *buffer)
    {
      hb_glyph_info_t *info = buffer->info;
      hb_glyph_position_t *pos = buffer->pos;
      bool horizontal = HB_DIRECTION_IS_HORIZONTAL (buffer->props.direction);
      unsigned int count = buffer->len;
      for (unsigned int i = 0; i < count; i++)
        if (_hb_glyph_info_is_unicode_space (&info[i]) && !_hb_glyph_info_ligated (&info[i]))
        {
          /* If font had no ASCII space and we used the invisible glyph, give it a 1/4 EM default advance. */
          if (buffer->invisible && info[i].codepoint == buffer->invisible)
          {
            if (horizontal)
    	  pos[i].x_advance = +font->x_scale / 4;
            else
    	  pos[i].y_advance = -font->y_scale / 4;
          }
    
          hb_unicode_funcs_t::space_t space_type = _hb_glyph_info_get_unicode_space_fallback_type (&info[i]);
          hb_codepoint_t glyph;
          typedef hb_unicode_funcs_t t;
          switch (space_type)
          {
    	case t::NOT_SPACE: /* Shouldn't happen. */
    	case t::SPACE:
    	  break;
    
    	case t::SPACE_EM:
    	case t::SPACE_EM_2:
    	case t::SPACE_EM_3:
    	case t::SPACE_EM_4:
    	case t::SPACE_EM_5:
    	case t::SPACE_EM_6:
    	case t::SPACE_EM_16:
    	  if (horizontal)
    	    pos[i].x_advance = +(font->x_scale + ((int) space_type)/2) / (int) space_type;
    	  else
    	    pos[i].y_advance = -(font->y_scale + ((int) space_type)/2) / (int) space_type;
    	  break;
    
    	case t::SPACE_4_EM_18:
    	  if (horizontal)
    	    pos[i].x_advance = (int64_t) +font->x_scale * 4 / 18;
    	  else
    	    pos[i].y_advance = (int64_t) -font->y_scale * 4 / 18;
    	  break;
    
    	case t::SPACE_FIGURE:
    	  for (char u = '0'; u <= '9'; u++)
    	    if (font->get_nominal_glyph (u, &glyph))
    	    {
    	      if (horizontal)
    		pos[i].x_advance = font->get_glyph_h_advance (glyph);
    	      else
    		pos[i].y_advance = font->get_glyph_v_advance (glyph);
    	      break;
    	    }
    	  break;
    
    	case t::SPACE_PUNCTUATION:
    	  if (font->get_nominal_glyph ('.', &glyph) ||
    	      font->get_nominal_glyph (',', &glyph))
    	  {
    	    if (horizontal)
    	      pos[i].x_advance = font->get_glyph_h_advance (glyph);
    	    else
    	      pos[i].y_advance = font->get_glyph_v_advance (glyph);
    	  }
    	  break;
    
    	case t::SPACE_NARROW:
    	  /* Half-space?
    	   * Unicode doc https://unicode.org/charts/PDF/U2000.pdf says ~1/4 or 1/5 of EM.
    	   * However, in my testing, many fonts have their regular space being about that
    	   * size.  To me, a percentage of the space width makes more sense.  Half is as
    	   * good as any. */
    	  if (horizontal)
    	    pos[i].x_advance /= 2;
    	  else
    	    pos[i].y_advance /= 2;
    	  break;
          }
        }
    }
    
    
    #endif