;;; For Malayalam OpenType fonts to draw the reformed Malayalam script.
(font layouter mlym-otf nil
- (font (nil nil unicode-bmp :otf=mlym=haln)))
+ (font (nil nil unicode-bmp :otf=mlym=blwf)))
-;; The first stage is to extract a syllable and re-order characters.
(category
- ;; C: consonant (except for BPR)
- ;; B: consonant that is the subject of "blws" feature
- ;; P: consonant that is the subject of "pstf" feature
+ ;; C: consonant (except for R, B, and P)
;; R: consonant RA
- ;; H: HALANT (VIRAMA)
- ;; m: MATRA (Mpre)
- ;; p: MATRA (Mpost)
- ;; t: MATRA (two-part: Mpre+Mpost)
- ;; A: vowel modifier (post) (ANUSVARA and VISARGA)
+ ;; B: consonant (below)
+ ;; P: consonant (post)
+ ;; H: HALANT
+ ;; m: vowel sign (pre)
+ ;; p: vowel sign (post)
+ ;; A: vowel modifier (post)
;; V: independent vowel
;; N: ZWNJ (ZERO WIDTH NON-JOINER)
;; J: ZWJ (ZERO WIDTH JOINER)
- ;; E: else (all other independnt characters)
- (0x0D00 0x0D7F ?E) ; else
- (0x0D02 0x0D03 ?A) ; ANUSVARA & VISARGA
- (0x0D05 0x0D14 ?V) ; independent vowel
- (0x0D15 0x0D39 ?C) ; consonant
-
- ;; Accorind to www.microsoft.com/typography/otfntdev/indicot/appen.htm,
- ;; these must be classified into 'B' category, but at least, it
- ;; doesn't work with this font
- ;; http://www.supersoftweb.com/THOOLIUC.TTF
- ;;(0x0D1F ?B) ; TTA
- ;;(0x0D23 ?B) ; NNA
- ;;(0x0D26 ?B) ; DA
- (0x0D2F ?P) ; YA
- (0x0D30 ?R) ; RA
- (0x0D32 ?B) ; LA
- (0x0D35 ?P) ; VA
- (0x0D3E 0x0D43 ?p) ; dependent vowel (Mpost)
- (0x0D46 0x0D48 ?m) ; dependent vowel (Mpre)
- (0x0D4A 0x0D4C ?t) ; dependent vowel (two-part)
- (0x0D4D ?H) ; VIRAMA (HALANT)
- (0x0D57 ?p) ; dependent vowel (Mpost)
- (0x0D60 ?V) ; VOCALIC RR
- (0x0D61 ?V) ; VOCALIC LL
+ ;; X: generic
+ ;; Z: internal use
+ (0x0D00 0x0D7F ?X) ; generic
+ (0x0D00 ?Z) ; internal use
+ (0x0D02 0x0D03 ?A) ; SIGN ANUSVARA .. VISARGA
+ (0x0D05 0x0D14 ?V) ; LETTER A .. LETTER AU
+ (0x0D15 0x0D39 ?C) ; LETTER KA .. LETTER HA
+ (0x0D2F ?P) ; LETTER YA
+ (0x0D30 ?R) ; LETTER RA
+ (0x0D32 ?B) ; LETTER LA
+ (0x0D35 ?P) ; LETTER VA
+ (0x0D3E 0x0D43 ?p) ; VOWEL SIGN AA .. VOCALIC R
+ (0x0D46 0x0D48 ?m) ; VOWEL SIGN E .. AI
+ (0x0D4D ?H) ; SIGN VIRAMA
+ (0x0D57 ?p) ; AU LENGTH MARK
+ (0x0D60 0x0D61 ?V) ; LETTER VOCALIC RR .. VOCALIC LL
+ (0x0964 0x0965 ?X) ; DANDA .. DOUBLE DANDA
(0x200C ?N) ; ZWNJ
(0x200D ?J) ; ZWJ
- (0x0D7D ?x) ; marker inserted before base-C
- (0x0D7E ?y) ; marker inserted after base-C
- (0x0D7F ?z) ; marker inserted before last Ra
)
-;; The 1st stage is to extract a syllable while moving a Halant
-;; following Cbase to the tail and partitioning two-part matras into
-;; parts.
+;; Decompose two-part vowels.
(generator
(0
(cond
- ;; If [CR]H is followed by ZWNJ/ZWJ, move ZWNJ/ZWJ) to the head so
- ;; that the later stages find it quickly.
- ("([CBPR]H)([NJ])"
- < | (2 =) (1 = =) | >)
-
- ;; A syllable starting with a consonant and ending with a vowel
- ;; and/or a vowel modifier.
- ("(([CBPR]H)*[CBPR])([pmt]A?|A)"
- < | (1 move-halant) (3 partition = =) | >)
-
- ;; A syllable starting with a consonant and ending without a vowel
- ;; nor a vowel modifier.
- ("(([CBPR]H)*[CBPR])(H)?"
- < | (1 move-halant) (3 =) | >)
-
- ;; A syllable starting with an independent vowel.
+ ((0x0D4A)
+ 0x0D46 0x0D3E)
+ ((0x0D4B)
+ 0x0D47 0x0D3E)
+ ((0x0D4C)
+ 0x0D46 0x0D57)
+ ("." =))
+ *))
+
+;; Syllable identification and reordering.
+;; Put the mark 0x0D00 before the first pre-base consonant and after
+;; the base consonant.
+(generator
+ (0
+ (cond
+ ;; A syllable with a left vowel sign.
+ ;;1 2 3 4
+ ("([CRBP](H[CRBP])*)(m)(p?A?)"
+ < | (3 =) (1 get-rh) 0x0D00 (1 move-h) (4 = *) | >)
+
+ ;; A syllable with a right vowel sign and/or a vowel modifier.
+ ;;1 2 3
+ ("([CRBP](H[CRBP])*)(pA?|A)"
+ < | (1 get-rh) 0x0D00 (1 move-h) (3 = *) | >)
+
+ ;; Explicit Halant / forced chillu form is cut off from the preceders.
+ ("(([CRBP]H)*)([CRBP]H[NJ])"
+ (1 cut-off) (3 < | = * | >))
+
+ ;; A syllable with no vowel signs/modifiers.
+ ("([CRBP](H[CRBP])*H?)"
+ < | (1 get-rh) 0x0D00 (1 move-h) | >)
+
+ ;; Starting with an independent vowel.
("(VA?)"
< | (1 = *) | >)
- ;; Other independent character.
("." =))
*)
- ;; Move Halant to the tail if necessary. Insert 0x0D7D before Cbase,
- ;; 0x0D7E after Cbase, 0x0D7F before the tailing RH (if any). They
- ;; work as markers in the following stages.
- (move-halant
+ ;; Extract RH that will be moved to the left.
+ (get-rh
+ (cond
+ (".+HR$"
+ 0x0D30 0x0D4D)))
+
+ ;; Move the halant after the base consonant to the end.
+ ;; Fill the resulting gap with 0x0D00.
+ ;; Remove final RH if any.
+ (move-h
(cond
- ;; In the first two cases, we don't have to move Halant.
- ("[BPR]$" ; Single consonant.
- 0x0D7D 0x0D7E =)
- ("(([CBPR]H)*)(C)$" ; Cbase is at tail.
- (1 = *) 0x0D7D (3 =) 0x0D7E)
- ;; In the following cases, we must move Halant to the tail.
- ("(([CBPR]H)*)([C])H(([BR]H)*)(R)$" ; Cbase is at head or middle.
- (1 = *) 0x0D7D (3 =) 0x0D7E (4 = *) 0x0D7F 0x0D30 0x0D4D)
- ("(([CBPR]H)*)([C])H(([BR]H)*[BP])$" ; Cbase is at head or middle.
- (1 = *) 0x0D7D (3 =) 0x0D7E (4 = *) 0x0D4D)
- ("(RH)?(.)H(.*)$" ; No Cbase, move the first
- ; Halant to the tail.
- (1 = *) 0x0D7D (2 = ) 0x0D7E (3 = *) 0x0D4D)))
-
- ;; Partition two-part matras into parts.
- (partition
+ ;; There is a C and the syllable ends with an R. Remove final RH.
+ ("([CRBPH]*C)H([RBPH]*)R$"
+ (1 = *) 0x0D00 (2 = *))
+
+ ;; The syllable ends other than R. Move H.
+ ("([CRBPH]*C)(H)([RBPH]+)$"
+ (1 = *) 0x0D00 (3 = *) (2 =))
+
+ ;; No C and end with R. The first consonant will be the base. Remove RH.
+ ("([RBP])H([RBPH]*)R$"
+ (1 =) 0x0D00 (2 = *))
+
+ ;; No C and not end with R. Move H.
+ ("([RBP])(H)([RBPH]+)$"
+ (1 =) 0x0D00 (3 = *) (2 =))
+
+ ;; Otherwise, no need to move H.
+ (".+"
+ = * 0x0D00)))
+
+ ;; If there are some letters make them a separate syllable.
+ (cut-off
(cond
- ((0x0D4A) 0x0D46 0x0D3E)
- ((0x0D4B) 0x0D47 0x0D3E)
- ;; Unicode suggests this partitioning:
- ((0x0D4C) 0x0D46 0x0D57)
- ;; but it is questionable. Perhaps, this substitution is better.
- ;; ((0x0D4C) 0x0D57)
- )))
-
-;; The 2nd stage is to apply GSUB for prebase part.
+ (".+"
+ < | = * | >)))
+ )
+
+;; Apply GSUB features.
+;; According to Microsoft Typography
+;; <http://www.microsoft.com/typography/otfntdev/malayot/features.htm>
+;; 'chillu' are generated with the 'haln' feature. However, some
+;; fonts generate them with 'half'.
(generator
(0
(cond
- ;; Originally [CBPR]H is followed by ZWNJ. Apply no feature (just
- ;; cmap).
- (" N([^ ]*) "
- (1 otf:mlym=+))
-
- ;; Originally [CBPR]H is followed by ZWJ. Apply only GSUB haln
- ;; feature.
- (" J([^ ]*) "
- (1 otf:mlym=haln+))
-
- ;; Syllable starting with a consonant should match this pattern.
- (" ([^ y]*)y([^ z]*)(z..)?([^ ]*) "
- |
- (1 otf:mlym=~blwf,~pstf,~blws,~psts,*+) ; Prebase part.
- 0x0D7E ; Preserve this marker.
- (2 = *) ; Postbase part.
- (3 = *) ; Optional tailing RH.
- (4 = *) ; Optional Mpost.
- |)
-
- ;; A syllable starting with an independent vowel.
- (" ([^ ]*) "
- (1 otf:mlym+))
-
- ;; Any other independent character. Apply no feature (just cmap).
- ("."
- [ otf:mlym=+ ] ))
+ ;; Ordinary syllable. We must avoid applying 'blwf' and 'pstf' to
+ ;; the first consonant. Also 'haln' and 'half' must be avoided
+ ;; because chillu does not appear at the beginning of a word. Thus
+ ;; we apply only 'akhn' here and postpone applying other features
+ ;; to the next stage.
+ ;; 1 2 3 4 5
+ (" (m?(RH)?Z)([CRBPH]+)(Z)([RBPHpA]*) "
+ | (1 = *) (3 otf:mlym=akhn+) (4 =)
+ (5 otf:mlym=blwf,pstf+) |)
+
+ ;; Explicit Halant.
+ (" [CRBP]HN "
+ = *)
+
+ ;; Forced chillu form.
+ (" ([CRBP]H)(J) "
+ | (1 otf:mlym=haln,half+) (2 =) |)
+
+ ;; Sequence that was cut off from a forced form. Again we apply
+ ;; only 'akhn' here.
+ (" ([CRBPH]+) "
+ | (1 otf:mlym=akhn+) |)
+
+ ("." =))
*))
-;; The 3rd stage is to move Mpre (if any) and tailing RH (if any)
-;; forward to just before the base consonant.
+;; Apply postoponed GSUB features to ordinary syllables.
(generator
(0
(cond
- (" ([^x]*)(.)xy([^ zm]*)(z(..))?(m)?([^ ]*) "
- | (1 = *) (4 (5 otf:mlym=pstf+)) (6 =) 0x0D7D (2 =) (3 = *) (7 = *) |)
- (" ([^x]*)x(.)y([^ zm]*)(z(..))?(m)?([^ ]*) "
- | (1 = *) (4 (5 otf:mlym=pstf+)) (6 =) 0x0D7D (2 =) (3 = *) (7 = *) |)
- (" xy([^ m]*)(m)?([^ ]*) "
- | (2 =) 0x0D7D (1 = *) (3 = *) |)
- ("[NJ]")
- ("."
- = ))
+ ;; Ordinary syllable.
+ ;; 1 2 3 4
+ (" (m?(RH)?Z[^Z])([^Z]*)(Z[^ ]*) "
+ | (1 = *) (3 otf:mlym=blwf,half,pstf,haln+) (4 = *) |)
+
+ ;; Explicit Halant / Forced chillu must not change to others.
+ (" (.+)[NJ] "
+ | (1 = *) |)
+
+ ;; Cut off sequence.
+ (" ([^ ])([^ ]*) "
+ | (1 =) (2 otf:mlym=blwf,half,pstf,haln+) |)
+
+ ("." =))
*))
-;; The 4th stage is to apply GSUB to postbase part.
+;; Second reordering. If there are glyphs before the base, move left
+;; vowel sign and RH to the left of the base.
(generator
(0
(cond
- (" ([^x]*)x(.H) "
- | (1 = *) (2 otf:mlym=haln) |)
- (" ([^x]*)x([^ ]*) "
- | (1 = *) (2 otf:mlym=~akhn,~haln,blwf,pstf,vatu,*+) |)
+ ;; 1 2 3 4 5
+ (" (m)?(RH)?Z([^Z]*)([^Z])Z([^ ]*) "
+ | (3 = *) (1 =) (2 otf:mlym=pstf+) (4 =) (5 = *) |)
+
("." =))
*))
-
-
-;; The 5th (last) stage is to apply GPOS features.
+;; Apply substitutions and GPOS features.
(generator
(0
(cond
- (" ([^ ]*) " (1 otf:mlym=))
- ("." =))
+ (" ([^ ]+) "
+ (1 otf:mlym=pres,abvs,blws,psts))
+ ("."
+ [ otf:mlym=+ ]))
*))
;; Local Variables: