src/hb-ot-shape-complex-misc.cc | 49 ++++++---- src/hb-private.hh | 6 + src/hb-unicode.cc | 6 + test/shaping/texts/in-tree/shaper-thai/MANIFEST | 1 test/shaping/texts/in-tree/shaper-thai/script-lao/MANIFEST | 1 test/shaping/texts/in-tree/shaper-thai/script-lao/misc/MANIFEST | 1 test/shaping/texts/in-tree/shaper-thai/script-lao/misc/sara-am.txt | 20 ++++ test/shaping/texts/in-tree/shaper-thai/script-thai/misc/MANIFEST | 1 test/shaping/texts/in-tree/shaper-thai/script-thai/misc/phinthu.txt | 16 +++ test/shaping/texts/in-tree/shaper-thai/script-thai/misc/sara-am.txt | 18 +++ 10 files changed, 102 insertions(+), 17 deletions(-)
New commits: commit 42848453bf260b456b46a07f066e31b8c3aac2f1 Author: Behdad Esfahbod <beh...@behdad.org> Date: Mon Jul 23 13:52:07 2012 -0400 [Thai] Reorder U+0E3A THAI VOWEL SIGN PHINTHU Uniscribe reorders U+0E3A to be after U+0E38 and U+0E39. We do that by modifying the ccc for U+0E3A. Fixes the two remaining Thai failures (see previous commit). diff --git a/src/hb-ot-shape-complex-misc.cc b/src/hb-ot-shape-complex-misc.cc index 17e2625..6852d47 100644 --- a/src/hb-ot-shape-complex-misc.cc +++ b/src/hb-ot-shape-complex-misc.cc @@ -132,6 +132,13 @@ _hb_ot_shape_complex_setup_masks_thai (hb_ot_map_t *map HB_UNUSED, * 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 _hb_unicode_modified_combining_class (). Lao does NOT have a U+0E3A + * equivalent. */ diff --git a/src/hb-unicode.cc b/src/hb-unicode.cc index 140f382..3569b20 100644 --- a/src/hb-unicode.cc +++ b/src/hb-unicode.cc @@ -363,6 +363,12 @@ _hb_unicode_modified_combining_class (hb_unicode_funcs_t *ufuncs, }; c = permuted_hebrew_classes[c - 10]; } + else if (unlikely (unicode == 0x0E3A)) /* THAI VOWEL SIGN PHINTHU */ + { + /* Assign 104, so it reorders after the THAI ccc=103 marks. + * Uniscribe does this. */ + c = 104; + } return c; } 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 ffd16f1..6aa865b 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 +1,2 @@ +phinthu.txt sara-am.txt diff --git a/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/phinthu.txt b/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/phinthu.txt new file mode 100644 index 0000000..e304777 --- /dev/null +++ b/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/phinthu.txt @@ -0,0 +1,16 @@ +ภ+à¸à¸º +à¸à¸¸ +à¸à¸¹ +à¸à¸¸à¸¹ +à¸à¸¹à¸¸ +à¸à¸ºà¸¸ +à¸à¸¸à¸º +à¸à¸ºà¸¹ +à¸à¸¹à¸º +à¸à¸ºà¸¸à¸¹ +à¸à¸¸à¸ºà¸¹ +à¸à¸¸à¸¹à¸º +à¸à¸ºà¸¹à¸¸ +à¸à¸¹à¸ºà¸¸ +à¸à¸¹à¸¸à¸º commit 4a7f4f3e56f8f7640ae7337aa1b3324f31e0d4ab Author: Behdad Esfahbod <beh...@behdad.org> Date: Mon Jul 23 13:15:33 2012 -0400 [Thai] Adjust SARA AM reordering to match Uniscribe Adjust the list of marks before SARA AM that get the reordering treatment. Also adjust cluster formation to match Uniscribe. With Wikipedia test data, now I see: - For Thai, with the Angsana New font from Win7, I see 54 failures out of over 4M tests (0.00129107%). Of the 54, two are legitimate reordering issues (fix coming soon), and the other 52 are simply Uniscribe using a zero-width space char instead of an unknown character for missing glyphs. No idea why. The missing-glyph sequences include one that is a Thai character followed by an Arabic Sokun. Someone confused it with Nikhahit I assume! - For Lao, with the Dokchampa font from Win7, 33 tests fail out of 54k (0.0615167%). All seem to be insignificant mark positioning with two marks on a base. Have to investigate. diff --git a/src/hb-ot-shape-complex-misc.cc b/src/hb-ot-shape-complex-misc.cc index 7a11876..17e2625 100644 --- a/src/hb-ot-shape-complex-misc.cc +++ b/src/hb-ot-shape-complex-misc.cc @@ -121,19 +121,20 @@ _hb_ot_shape_complex_setup_masks_thai (hb_ot_map_t *map 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 a, *and* mode the - * nihka hit backwards over any *tone* mark (0E48-0E4B). + * 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 + * 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 uâªser wanted, but the rendering is nevertheless nikhahit above + * not what a user wanted, but the rendering is nevertheless nikhahit above * chattawa. * * Same for Lao. */ + /* * Here are the characters of significance: * @@ -142,9 +143,9 @@ _hb_ot_shape_complex_setup_masks_thai (hb_ot_map_t *map HB_UNUSED, * SARA AA: U+0E32 U+0EB2 * Nikhahit: U+0E4D U+0ECD * - * Tone marks: - * Thai: <0E48..0E4B> CCC=107 - * Lao: <0EC8..0ECB> CCC=122 + * Testing shows that Uniscribe reorder the following marks: + * Thai: <0E31..0E37,0E47..0E4E> + * Lao: <0EB1..0EB7,0EC7..0ECE> * * Note how the Lao versions are the same as Thai + 0x80. */ @@ -154,7 +155,7 @@ _hb_ot_shape_complex_setup_masks_thai (hb_ot_map_t *map HB_UNUSED, #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) (((x) & ~0x0083) == 0x0E48) +#define IS_TONE_MARK(x) (hb_in_ranges<hb_codepoint_t> ((x) & ~0x0080, 0x0E31, 0x0E37, 0x0E47, 0x0E4E)) buffer->clear_output (); unsigned int count = buffer->len; @@ -179,14 +180,23 @@ _hb_ot_shape_complex_setup_masks_thai (hb_ot_map_t *map HB_UNUSED, while (start > 0 && IS_TONE_MARK (buffer->out_info[start - 1].codepoint)) start--; - /* Move Nikhahit (end-2) to the beginning */ - 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; - - buffer->merge_out_clusters (start, end); + 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 (); } diff --git a/src/hb-private.hh b/src/hb-private.hh index 3f710ed..0b9c4ef 100644 --- a/src/hb-private.hh +++ b/src/hb-private.hh @@ -729,6 +729,12 @@ hb_in_range (T u, T lo, T hi) return lo <= u && u <= hi; } +template <typename T> static inline bool +hb_in_ranges (T u, T lo1, T hi1, T lo2, T hi2) +{ + return hb_in_range (u, lo1, hi1) || hb_in_range (u, lo2, hi2); +} + /* Useful for set-operations on small enums. * For example, for testing "x â {x1, x2, x3}" use: diff --git a/test/shaping/texts/in-tree/shaper-thai/MANIFEST b/test/shaping/texts/in-tree/shaper-thai/MANIFEST index 22bc0ed..32b5476 100644 --- a/test/shaping/texts/in-tree/shaper-thai/MANIFEST +++ b/test/shaping/texts/in-tree/shaper-thai/MANIFEST @@ -1 +1,2 @@ +script-lao script-thai diff --git a/test/shaping/texts/in-tree/shaper-thai/script-lao/MANIFEST b/test/shaping/texts/in-tree/shaper-thai/script-lao/MANIFEST new file mode 100644 index 0000000..b8752e7 --- /dev/null +++ b/test/shaping/texts/in-tree/shaper-thai/script-lao/MANIFEST @@ -0,0 +1 @@ +misc diff --git a/test/shaping/texts/in-tree/shaper-thai/script-lao/misc/MANIFEST b/test/shaping/texts/in-tree/shaper-thai/script-lao/misc/MANIFEST new file mode 100644 index 0000000..ffd16f1 --- /dev/null +++ b/test/shaping/texts/in-tree/shaper-thai/script-lao/misc/MANIFEST @@ -0,0 +1 @@ +sara-am.txt diff --git a/test/shaping/texts/in-tree/shaper-thai/script-lao/misc/sara-am.txt b/test/shaping/texts/in-tree/shaper-thai/script-lao/misc/sara-am.txt new file mode 100644 index 0000000..234d8c0 --- /dev/null +++ b/test/shaping/texts/in-tree/shaper-thai/script-lao/misc/sara-am.txt @@ -0,0 +1,20 @@ +àºàº³ +àºà»àº³ +àºà»à»àº² +àºà»à»àº² +ມັຳ +ມິຳ +ມີຳ +ມຶຳ +ມືຳ +ມຸຳ +ມູຳ +ມ຺ຳ +ມà»àº³ +ມà»àº³ +ມà»àº³ +ມà»àº³ +ມà»àº³ +ມà»àº³ +ມà»àº³ +ມà»àº³ diff --git a/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/sara-am.txt b/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/sara-am.txt index 6d385ef..9f044ce 100644 --- a/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/sara-am.txt +++ b/test/shaping/texts/in-tree/shaper-thai/script-thai/misc/sara-am.txt @@ -1,4 +1,20 @@ -ำ à¸à¸³ à¸à¹à¸³ à¸à¹à¹à¸² +à¸à¹à¹à¸² +มัำ +มิำ +มีำ +มึำ +มืำ +มุำ +มูำ +มฺำ +มà¹à¸³ +มà¹à¸³ +มà¹à¸³ +มà¹à¸³ +มà¹à¸³ +มà¹à¸³ +มà¹à¸³ +มà¹à¸³
_______________________________________________ HarfBuzz mailing list HarfBuzz@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/harfbuzz