2 * Copyright © 2011 Google, Inc.
4 * This is part of HarfBuzz, a text shaping library.
6 * Permission is hereby granted, without written agreement and without
7 * license or royalty fees, to use, copy, modify, and distribute this
8 * software and its documentation for any purpose, provided that the
9 * above copyright notice and the following two paragraphs appear in
10 * all copies of this software.
12 * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
13 * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
14 * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
15 * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
18 * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
19 * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
20 * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
21 * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
22 * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
24 * Google Author(s): Behdad Esfahbod
27 #include "hb-ot-shape-private.hh"
28 #include "hb-ot-shape-complex-private.hh"
35 * This file exports one main function: _hb_ot_shape_normalize().
37 * This function closely reflects the Unicode Normalization Algorithm,
38 * yet it's different. The shaper an either prefer decomposed (NFD) or
41 * In general what happens is that: each grapheme is decomposed in a chain
42 * of 1:2 decompositions, marks reordered, and then recomposed if desires,
43 * so far it's like Unicode Normalization. However, the decomposition and
44 * recomposition only happens if the font supports the resulting characters.
48 * - Try to render all canonically equivalent strings similarly. To really
49 * achieve this we have to always do the full decomposition and then
50 * selectively recompose from there. It's kinda too expensive though, so
51 * we skip some cases. For example, if composed is desired, we simply
52 * don't touch 1-character clusters that are supported by the font, even
53 * though their NFC may be different.
55 * - When a font has a precomposed character for a sequence but the 'ccmp'
56 * feature in the font is not adequate, form use the precomposed character
57 * which typically has better mark positioning.
59 * - When a font does not support a character but supports its decomposition,
60 * well, use the decomposition.
62 * - The Indic shaper requests decomposed output. This will handle splitting
63 * matra for the Indic shaper.
68 decompose (hb_ot_shape_context_t *c,
72 hb_codepoint_t a, b, glyph;
73 bool has_this = hb_font_get_glyph (c->font, ab, 0, &glyph);
75 /* If recomposing and the single char is supported by the font, we're good. */
76 if (recompose && has_this)
79 if (hb_unicode_decompose (c->buffer->unicode, ab, &a, &b) &&
80 hb_font_get_glyph (c->font, b, 0, &glyph) &&
81 decompose (c, recompose, a))
83 /* Successfully decomposed. */
86 /* Try composing b with base if not blocked */
97 decompose_single_char_cluster (hb_ot_shape_context_t *c,
101 // c->buffer->copy ();
102 // bool ret = decompose (c, recompose, c->buffer->info[i].codepoint);
103 // c->buffer->skip ();
109 handle_single_char_cluster (hb_ot_shape_context_t *c,
114 decompose_single_char_cluster (c, recompose, i);
118 handle_multi_char_cluster (hb_ot_shape_context_t *c,
123 /* TODO Currently if there's a variation-selector we give-up, it's just too hard. */
124 for (unsigned int i = start; i < end; i++)
125 if (unlikely (is_variation_selector (c->buffer->info[i].codepoint)))
131 _hb_ot_shape_normalize (hb_ot_shape_context_t *c)
133 hb_buffer_t *buffer = c->buffer;
134 bool recompose = !hb_ot_shape_complex_prefer_decomposed (c->plan->shaper);
136 buffer->clear_output ();
138 unsigned int count = buffer->len;
139 for (buffer->idx = 0; buffer->idx < count;)
143 for (end = buffer->idx + 1; end < count; end++)
144 if (buffer->info[buffer->idx].cluster != buffer->info[end].cluster)
147 if (buffer->idx + 1 == end)
148 handle_single_char_cluster (c, recompose, buffer->idx);
150 handle_multi_char_cluster (c, recompose, buffer->idx, end);
151 while (buffer->idx < end)
152 c->buffer->next_glyph ();
155 buffer->swap_buffers ();