indickit 0.6.0 → 0.9.0

package/dist/types/js/phonetic.d.ts

@@ -10,6 +10,13 @@
  *
  * @module
  */
+/** @internal For ./phonetic-search: the engine compiler, the normalizer and the
+ * search key's engine (keys of a normalized word). Not part of the API. */
+export declare const _internal: {
+    compile: (r: unknown) => (word: string) => string[];
+    normalize: (w: string) => string;
+    engine: (word: string) => string[];
+};
 /** The rules every key from this module was made with. */
 export declare const RULES_VERSION: string;
 /** NameKeys stops here: a long Tamil name can have many key combinations. */
@@ -23,6 +30,13 @@ export declare function words(name: string): string[];
  * separated by a space. Two names match when these share an element, so
  * these are what to index. At most MAX_NAME_KEYS. */
 export declare function nameKeys(name: string): string[];
+/** Joined keys shorter than this (in classes) are not returned: they find too many names. */
+export declare const JOINED_MIN_CLASSES: number;
+/** The keys of a name written as one word, its words joined: "Ram Nath" and
+ * இராம்நாத் share no nameKeys, but can share a joined key. Only keys of at
+ * least JOINED_MIN_CLASSES classes. A search looks them up only when
+ * nameKeys find nobody; match does not use them. */
+export declare function joinedKeys(name: string): string[];
 /** Whether two names have the same number of words and every pair of words,
  * in order, shares a key. */
 export declare function match(a: string, b: string): boolean;

package/dist/types/js/stem.d.ts

@@ -6,6 +6,7 @@
  * least two code points; in languages with the vowel step it then drops one
  * final vowel sign (the Dravidian enunciative u). A word with no listed
  * ending is its own stem, and so is every word of a language with no table.
+ * Kannada, whose endings stack, runs all of this twice.
  *
  * A stem is a key, not text. Apply `stem` once to the words you index and
  * once to the query, and compare. stem(stem(w)) may cut again, as Snowball

package/dist/types/js/unidata.d.ts

@@ -0,0 +1,36 @@
+/**
+ * The Unicode data indickit reads inside the blocks it has rules for (Arabic,
+ * Devanagari to Malayalam, Ol Chiki, Vedic, Devanagari Extended, Meetei
+ * Mayek): one pinned table, internal/unidata/unicode15.json, which the Go
+ * package internal/unidata also reads. So Go and TypeScript give one answer
+ * whatever Unicode version the engine ships, and a move to a new Unicode
+ * version is a change to the table: a rules change, with a conformance run
+ * (MAINTAINING.md, "Upkeep").
+ *
+ * Outside the blocks the engine answers (`\p{..}`, `String.prototype.normalize`).
+ * The Latin blocks indickit also reads are closed (every code point is
+ * assigned), and Unicode's stability policy freezes the decomposition and
+ * combining class of an assigned code point, so the engine cannot drift there.
+ *
+ * @module
+ */
+/** The Unicode version of the table. */
+export declare const UNICODE_VERSION: string;
+/** Whether the table answers for this code point. */
+export declare const inBlocks: (cp: number) => boolean;
+/** General_Category Mn. */
+export declare function isMn(cp: number): boolean;
+/** General_Category Mn, Mc or Me. */
+export declare function isMark(cp: number): boolean;
+/** General_Category L. */
+export declare function isLetter(cp: number): boolean;
+/** The full canonical decomposition of one code point. */
+export declare function nfd(cp: number): number[];
+/**
+ * Unicode NFC. The string is cut before each stable code point of the
+ * blocks. A piece made only of code points of the blocks is composed from
+ * the table; a piece holding any other code point goes to the engine, whose
+ * data for the blocks' assigned code points equals the table's (stability
+ * policy), and which alone knows the combining classes outside the blocks.
+ */
+export declare function nfc(s: string): string;

package/dist/types/js/wellformed.d.ts

@@ -0,0 +1,31 @@
+/**
+ * wellformed: can a font draw this text whole? A text renderer (HarfBuzz)
+ * draws a dotted circle (◌) where a character cannot join the syllable
+ * before it: a vowel sign with no letter before it (ि at the start of a
+ * word, as PDF text layers give it), a vowel sign after a virama, a third
+ * anusvara. isWellFormed("िहन्दी") is false; brokenAt gives the index
+ * of the first such character. Use it to reject or flag text before you
+ * store or index it.
+ *
+ * The rules are HarfBuzz's own verdicts, swept over every character after
+ * every character of each script, and kept in rules.json; this code only
+ * walks them. A text is cut into runs of one script (with the invisible
+ * characters and the placeholders NBSP and ◌ between them); a space, a
+ * digit or a letter of another script ends a run. In a run, a character is
+ * broken when it starts the run and its class is in the start set, or the
+ * widest window that holds its class and up to three classes before it
+ * says so (quads, then triples, then pairs), or it ends the run and its
+ * class is pending (Malayalam dot reph ൎ, which waits for a consonant).
+ *
+ * Arabic and Ol Chiki runs are always whole: HarfBuzz draws no dotted
+ * circle inside them. Only HarfBuzz was tested; DirectWrite and CoreText
+ * have their own grammars.
+ *
+ * @module
+ */
+/** True when a font draws text with no dotted circle. */
+export declare function isWellFormed(text: string): boolean;
+/** The UTF-16 index of the first character a font draws with a dotted circle, or -1. */
+export declare function brokenAt(text: string): number;
+/** The rules version; a stored verdict goes stale when it changes. */
+export declare const RULES_VERSION: string;

package/dist/wellformed.js

@@ -0,0 +1 @@
+var o={version:"2026-10-07",invisible:"‌‍​⁠\uFEFF­",placeholder:" ◌",scripts:{Devanagari:{names:["Avagraha","Bindu","Bindu.1","Cantillation_Mark","Cantillation_Mark.1","Consonant","Consonant.1","Consonant.2","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Nukta","Number","Other","Other.1","Other.2","Placeholder","Virama","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Dependent.3","Vowel_Dependent.4","Vowel_Dependent.5","Vowel_Dependent.6","Vowel_Independent","Vowel_Independent.1","Vowel_Independent.2","Vowel_Independent.3","Vowel_Independent.4","Vowel_Independent.5"],blocks:{"0900":"111ltuvwtxtttttyttttt5555555555555555555555555556555555555mne0noopoooqqrsnnnnkong34hhonn77777777ttooggffffffffffggtttttt55555555",A8E0:"33333333333333333322iiiiggggggto"},shared:"bacd89jj",start:"134ehklmnopqrs",pending:"",pairs:{"0":"kmnopqrs","1":"ekmnopqrs","2":"kmnopqrs","3":"1hlmnopqrs","4":"1hlmnopqrs","8":"134ehklmnopqrs","9":"134ehklmnopqrs",a:"134ehklmnopqrs",b:"134ehklmnopqrs",c:"134ehklmnopqrs",d:"134ehklmnopqrs",g:"134ehklmnopqrs",h:"4mnopqrs",i:"kmnopqrs",k:"kmnopqrs",l:"ekmnopqrs",u:"mnq",v:"mqrs",x:"p",y:"qr"},triples_broken:{"03":"k","04":"k","0e":"ekmnopqrs","0h":"k","11":"1hl","13":"ek","14":"ek","1h":"1ehkl","1l":"1hl","23":"k","24":"k","2e":"ekmnopqrs","2h":"k","3e":"1hlmnopqrs","3k":"1hl","4e":"1hlmnopqrs","4k":"1hl","6k":"w","7e":"e",ee:"e",h1:"1hl",h3:"4",he:"4mnopqrs",hh:"1hl",hk:"4",hl:"1hl",i3:"k",i4:"k",ie:"ekmnopqrs",ih:"k",k3:"k",k4:"k",ke:"kmnopqrs",kh:"k",l1:"1hl",l3:"ek",l4:"ek",lh:"1ehkl",ll:"1hl",me:"e",ne:"e",oe:"e",pe:"e",qe:"e",re:"e",se:"e"},triples_whole:{"0a":"1hl","0b":"1hl","1b":"34","2a":"1hl","2b":"1hl","5a":"134ehklmnopqrs","5b":"1hklmnopqrs","6a":"134ehklmnopqrs","6b":"1hklmnopqrs","7a":"1hklmnopqrs","7b":"1hklmnopqrs",ea:"1hklmnopqrs",eb:"1hklmnopqrs",fa:"1hklmnopqrs",fb:"1hklmnopqrs",hb:"34",ia:"1hl",ib:"1hl",ja:"1hklmnopqrs",jb:"1hklmnopqrs",ka:"134ehl",kb:"134hl",lb:"34",ma:"1hlmnopqrs",mb:"1hlmnopqrs",mk:"mnopqrs",na:"1hlmnopqrs",nb:"1hlmnopqrs",nk:"mnopqrs",oa:"1hlmnopqrs",ob:"1hlmnopqrs",ok:"mnopqrs",pa:"1hlmnopqrs",pb:"1hlmnopqrs",pk:"mnopqrs",qa:"1hlmnopqrs",qb:"1hlmnopqrs",qk:"mnopqrs",ra:"1hlmnopqrs",rb:"1hlmnopqrs",rk:"mnopqrs",sa:"1hlmnopqrs",sb:"1hlmnopqrs",sk:"mnopqrs",ta:"1hklmnopqrs",tb:"1hklmnopqrs",ua:"1hklmnopqrs",ub:"1hklmnopqrs",va:"1hklmnopqrs",vb:"1hklmnopqrs",wa:"1hklmnopqrs",wb:"1hklmnopqrs",xa:"1hklmnopqrs",xb:"1hklmnopqrs",ya:"1hklmnopqrs",yb:"1hklmnopqrs"},quads_broken:{},quads_whole:{"5aa":"1hklmnopqrs","5ab":"1hklmnopqrs","5ba":"mnopqrs","5bb":"mnopqrs","6aa":"1hklmnopqrs","6ab":"1hklmnopqrs","6ba":"mnopqrs","6bb":"mnopqrs","7aa":"mnopqrs","7ab":"mnopqrs","7ba":"mnopqrs","7bb":"mnopqrs",eaa:"mnopqrs",eab:"mnopqrs",eba:"mnopqrs",ebb:"mnopqrs",faa:"mnopqrs",fab:"mnopqrs",fba:"mnopqrs",fbb:"mnopqrs",jaa:"mnopqrs",jab:"mnopqrs",jba:"mnopqrs",jbb:"mnopqrs",kaa:"1hl",kab:"1hl",kba:"1hl",kbb:"1hl",maa:"mnopqrs",mab:"mnopqrs",mba:"mnopqrs",mbb:"mnopqrs",naa:"mnopqrs",nab:"mnopqrs",nba:"mnopqrs",nbb:"mnopqrs",oaa:"mnopqrs",oab:"mnopqrs",oba:"mnopqrs",obb:"mnopqrs",paa:"mnopqrs",pab:"mnopqrs",pba:"mnopqrs",pbb:"mnopqrs",qaa:"mnopqrs",qab:"mnopqrs",qba:"mnopqrs",qbb:"mnopqrs",raa:"mnopqrs",rab:"mnopqrs",rba:"mnopqrs",rbb:"mnopqrs",saa:"mnopqrs",sab:"mnopqrs",sba:"mnopqrs",sbb:"mnopqrs",taa:"mnopqrs",tab:"mnopqrs",tba:"mnopqrs",tbb:"mnopqrs",uaa:"mnopqrs",uab:"mnopqrs",uba:"mnopqrs",ubb:"mnopqrs",vaa:"mnopqrs",vab:"mnopqrs",vba:"mnopqrs",vbb:"mnopqrs",waa:"mnopqrs",wab:"mnopqrs",wba:"mnopqrs",wbb:"mnopqrs",xaa:"mnopqrs",xab:"mnopqrs",xba:"mnopqrs",xbb:"mnopqrs",yaa:"mnopqrs",yab:"mnopqrs",yba:"mnopqrs",ybb:"mnopqrs"}},Bengali:{names:["Avagraha","Bindu","Bindu.1","Consonant","Consonant_Dead","Consonant_Placeholder","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Nukta","Number","Other","Placeholder","Syllable_Modifier","Virama","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Dependent.3","Vowel_Independent","Vowel_Independent.1","Vowel_Independent.2","Vowel_Independent.3"],blocks:{"0980":"511i.nooooopq..oo..oo33333333333333333333.3333333.3...3333..c0jkkkklk..kk..kkh4........k....33.3oomk..dddddddddd33eeeeeeeeee2eg"},shared:"98ab67ff",start:"1cghijklm",pending:"",pairs:{"0":"hjklm","1":"chjklm","6":"1cghijklm","7":"1cghijklm","8":"1cghijklm","9":"1cghijklm",a:"1cghijklm",b:"1cghijklm",e:"1cghijklm",g:"jklm",h:"hjklm",i:"chjklm",n:"j",p:"l",q:"m"},triples_broken:{"0c":"chjklm","0g":"h","11":"1gi","1g":"1cghi","1i":"1gi",cc:"c",g1:"1gi",gc:"jklm",gg:"1gi",gi:"1gi",hc:"hjklm",hg:"h",i1:"1gi",ig:"1cghi",ii:"1gi",jc:"c",kc:"c",lc:"c",mc:"c"},triples_whole:{"08":"1gi","09":"1gi","28":"1ghijklm","29":"1ghijklm","38":"1cghijklm","39":"1ghijklm","48":"1cghijklm","49":"1ghijklm","58":"1ghijklm","59":"1ghijklm",c8:"1ghijklm",c9:"1ghijklm",d8:"1ghijklm",d9:"1ghijklm",f8:"1ghijklm",f9:"1ghijklm",h8:"1cgi",h9:"1gi",j8:"1gijklm",j9:"1gijklm",jh:"jklm",k8:"1gijklm",k9:"1gijklm",kh:"jklm",l8:"1gijklm",l9:"1gijklm",lh:"jklm",m8:"1gijklm",m9:"1gijklm",mh:"jklm",n8:"1ghijklm",n9:"1ghijklm",o8:"1ghijklm",o9:"1ghijklm",p8:"1ghijklm",p9:"1ghijklm",q8:"1ghijklm",q9:"1ghijklm"},quads_broken:{},quads_whole:{"288":"jklm","289":"jklm","298":"jklm","299":"jklm","388":"1ghijklm","389":"1ghijklm","398":"jklm","399":"jklm","488":"1ghijklm","489":"1ghijklm","498":"jklm","499":"jklm","588":"jklm","589":"jklm","598":"jklm","599":"jklm",c88:"jklm",c89:"jklm",c98:"jklm",c99:"jklm",d88:"jklm",d89:"jklm",d98:"jklm",d99:"jklm",f88:"jklm",f89:"jklm",f98:"jklm",f99:"jklm",h88:"1gi",h89:"1gi",h98:"1gi",h99:"1gi",j88:"jklm",j89:"jklm",j98:"jklm",j99:"jklm",k88:"jklm",k89:"jklm",k98:"jklm",k99:"jklm",l88:"jklm",l89:"jklm",l98:"jklm",l99:"jklm",m88:"jklm",m89:"jklm",m98:"jklm",m99:"jklm",n88:"jklm",n89:"jklm",n98:"jklm",n99:"jklm",o88:"jklm",o89:"jklm",o98:"jklm",o99:"jklm",p88:"jklm",p89:"jklm",p98:"jklm",p99:"jklm",q88:"jklm",q89:"jklm",q98:"jklm",q99:"jklm"}},Gurmukhi:{names:["Bindu","Cantillation_Mark","Consonant","Consonant.1","Consonant_Medial","Consonant_Placeholder","Consonant_Placeholder.1","Gemination_Mark","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Nukta","Number","Other","Placeholder","Virama","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Dependent.3","Vowel_Independent","Vowel_Independent.1"],blocks:{"0A00":".00j.oppppp....pp..pp22222222222222222222.2222222.23.23.22..e.klmnn....lk..nki...1.......3332.3.......ffffffffff0756g4g"},shared:"bacd89hh",start:"0147eijklmn",pending:"",pairs:{"0":"14eikln","1":"4","4":"4e","5":"lm","6":"n","7":"14eikln","8":"0147eijklmn","9":"0147eijklmn",a:"0147eijklmn",b:"0147eijklmn",c:"0147eijklmn",d:"0147eijklmn",g:"0147eijklmn",i:"14iklmn",j:"14eikln",k:"4",l:"4",m:"4",n:"4",o:"k"},triples_broken:{"00":"07jm","07":"07jm","0j":"07jm","1e":"4e","3e":"e","4i":"e","70":"07jm","77":"07jm","7j":"07jm",ee:"e",i0:"m",i7:"m",ie:"14iklmn",ij:"m",j0:"07jm",j7:"07jm",jj:"07jm",ke:"4e",le:"4e",me:"4e",ne:"4e"},triples_whole:{"1a":"017jklmn","1b":"017jklmn","1i":"1klmn","2a":"0147eijklmn","2b":"017ijklmn","3a":"017ijklmn","3b":"017ijklmn","4a":"017ijklmn","4b":"017ijklmn","5a":"0147eijklmn","5b":"017ijklmn","6a":"0147eijklmn","6b":"017ijklmn",ea:"017ijklmn",eb:"017ijklmn",fa:"017ijklmn",fb:"017ijklmn",ha:"017ijklmn",hb:"017ijklmn",ia:"07ej",ib:"07j",ka:"017jklmn",kb:"017jklmn",ki:"1klmn",la:"017jklmn",lb:"017jklmn",li:"1klmn",ma:"017jklmn",mb:"017jklmn",mi:"1klmn",na:"017jklmn",nb:"017jklmn",ni:"1klmn",oa:"017ijklmn",ob:"017ijklmn",pa:"017ijklmn",pb:"017ijklmn"},quads_broken:{},quads_whole:{"1aa":"1klmn","1ab":"1klmn","1ba":"1klmn","1bb":"1klmn","2aa":"017ijklmn","2ab":"017ijklmn","2ba":"1klmn","2bb":"1klmn","3aa":"1klmn","3ab":"1klmn","3ba":"1klmn","3bb":"1klmn","4aa":"1klmn","4ab":"1klmn","4ba":"1klmn","4bb":"1klmn","5aa":"017ijklmn","5ab":"017ijklmn","5ba":"1klmn","5bb":"1klmn","6aa":"017ijklmn","6ab":"017ijklmn","6ba":"1klmn","6bb":"1klmn",eaa:"1klmn",eab:"1klmn",eba:"1klmn",ebb:"1klmn",faa:"1klmn",fab:"1klmn",fba:"1klmn",fbb:"1klmn",haa:"1klmn",hab:"1klmn",hba:"1klmn",hbb:"1klmn",iaa:"07j",iab:"07j",iba:"07j",ibb:"07j",kaa:"1klmn",kab:"1klmn",kba:"1klmn",kbb:"1klmn",laa:"1klmn",lab:"1klmn",lba:"1klmn",lbb:"1klmn",maa:"1klmn",mab:"1klmn",mba:"1klmn",mbb:"1klmn",naa:"1klmn",nab:"1klmn",nba:"1klmn",nbb:"1klmn",oaa:"1klmn",oab:"1klmn",oba:"1klmn",obb:"1klmn",paa:"1klmn",pab:"1klmn",pba:"1klmn",pbb:"1klmn"}},Gujarati:{names:["Avagraha","Bindu","Cantillation_Mark","Consonant","Gemination_Mark","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Nukta","Nukta.1","Number","Other","Placeholder","Virama","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Dependent.3","Vowel_Independent","Vowel_Independent.1"],blocks:{"0A80":".11h.mnnnnnnnn.nnn.nn33333333333333333333.3333333.33.33333..b0ijjjjjjk.lll.llg..e...............nnjj..ddddddddddee.......3242ccc"},shared:"879a56ff",start:"124bcghijkl",pending:"",pairs:{"0":"gijkl","1":"4bcgijkl","2":"14bcghijkl","5":"124bcghijkl","6":"124bcghijkl","7":"124bcghijkl","8":"124bcghijkl","9":"124bcghijkl",a:"124bcghijkl",e:"124bcghijkl",g:"4cgijkl",h:"4bcgijkl",k:"i",m:"ikl"},triples_broken:{"04":"4bcgijkl","0b":"4bcgijkl","0c":"4bcgijkl","11":"1h","1h":"1h","44":"4bc","4b":"4bc","4c":"4bc",b4:"4bc",bb:"4bc",bc:"4bc",c4:"4bc",cb:"4bc",cc:"4bc",gb:"4cgijkl",h1:"1h",hh:"1h",i4:"4bc",ib:"4bc",ic:"4bc",j4:"4bc",jb:"4bc",jc:"4bc",k4:"4bc",kb:"4bc",kc:"4bc",l4:"4bc",lb:"4bc",lc:"4bc"},triples_whole:{"07":"1h","08":"1h","18":"2","37":"124bcghijkl","38":"1ghijkl","47":"1ghijkl","48":"1ghijkl",b7:"1ghijkl",b8:"1ghijkl",c7:"1ghijkl",c8:"1ghijkl",d7:"1ghijkl",d8:"1ghijkl",f7:"1ghijkl",f8:"1ghijkl",g7:"124bch",g8:"12h",h8:"2",i7:"1hijkl",i8:"1hijkl",ig:"ijkl",j7:"1hijkl",j8:"1hijkl",jg:"ijkl",k7:"1hijkl",k8:"1hijkl",kg:"ijkl",l7:"1hijkl",l8:"1hijkl",lg:"ijkl",m7:"1ghijkl",m8:"1ghijkl",n7:"1ghijkl",n8:"1ghijkl"},quads_broken:{},quads_whole:{"377":"1ghijkl","378":"1ghijkl","387":"ijkl","388":"ijkl","477":"ijkl","478":"ijkl","487":"ijkl","488":"ijkl",b77:"ijkl",b78:"ijkl",b87:"ijkl",b88:"ijkl",c77:"ijkl",c78:"ijkl",c87:"ijkl",c88:"ijkl",d77:"ijkl",d78:"ijkl",d87:"ijkl",d88:"ijkl",f77:"ijkl",f78:"ijkl",f87:"ijkl",f88:"ijkl",g77:"1h",g78:"1h",g87:"1h",g88:"1h",i77:"ijkl",i78:"ijkl",i87:"ijkl",i88:"ijkl",j77:"ijkl",j78:"ijkl",j87:"ijkl",j88:"ijkl",k77:"ijkl",k78:"ijkl",k87:"ijkl",k88:"ijkl",l77:"ijkl",l78:"ijkl",l87:"ijkl",l88:"ijkl",m77:"ijkl",m78:"ijkl",m87:"ijkl",m88:"ijkl",n77:"ijkl",n78:"ijkl",n87:"ijkl",n88:"ijkl"}},Oriya:{names:["Avagraha","Bindu","Consonant","Consonant.1","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Nukta","Number","Other","Placeholder","Virama","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Dependent.3","Vowel_Independent","Vowel_Independent.1","Vowel_Independent.2"],blocks:{"0B00":".11f.klllllll..ml..ml22222222222222222222.2222222.22.22222..a0ghhhhhh..hh..hhe.......ihj....33.2llhh..bbbbbbbbbbc2cccccc"},shared:"768945dd",start:"1aefghij",pending:"",pairs:{"0":"eghj","1":"aeghij","4":"1aefghij","5":"1aefghij","6":"1aefghij","7":"1aefghij","8":"1aefghij","9":"1aefghij",c:"1aefghij",e:"eghij",f:"aeghij",k:"g",m:"j"},triples_broken:{"0a":"aeghij","0i":"aeghij","11":"1f","1f":"1f","3a":"ai","3i":"ai",aa:"ai",ai:"ai",ea:"eghij",f1:"1f",ff:"1f",ga:"ai",gi:"ai",ha:"ai",hi:"ai",ia:"ai",ii:"ai",ja:"ai",ji:"ai"},triples_whole:{"06":"1f","07":"1f","26":"1aefghij","27":"1efghj","36":"1efghj","37":"1efghj",a6:"1efghj",a7:"1efghj",b6:"1efghj",b7:"1efghj",d6:"1efghj",d7:"1efghj",e6:"1afi",e7:"1f",g6:"1fghj",g7:"1fghj",ge:"ghj",h6:"1fghj",h7:"1fghj",he:"ghj",i6:"1efghj",i7:"1efghj",j6:"1fghj",j7:"1fghj",je:"ghj",k6:"1efghj",k7:"1efghj",l6:"1efghj",l7:"1efghj",m6:"1efghj",m7:"1efghj"},quads_broken:{},quads_whole:{"266":"1efghj","267":"1efghj","276":"ghj","277":"ghj","366":"ghj","367":"ghj","376":"ghj","377":"ghj",a66:"ghj",a67:"ghj",a76:"ghj",a77:"ghj",b66:"ghj",b67:"ghj",b76:"ghj",b77:"ghj",d66:"ghj",d67:"ghj",d76:"ghj",d77:"ghj",e66:"1f",e67:"1f",e76:"1f",e77:"1f",g66:"ghj",g67:"ghj",g76:"ghj",g77:"ghj",h66:"ghj",h67:"ghj",h76:"ghj",h77:"ghj",i66:"ghj",i67:"ghj",i76:"ghj",i77:"ghj",j66:"ghj",j67:"ghj",j76:"ghj",j77:"ghj",k66:"ghj",k67:"ghj",k76:"ghj",k77:"ghj",l66:"ghj",l67:"ghj",l76:"ghj",l77:"ghj",m66:"ghj",m67:"ghj",m76:"ghj",m77:"ghj"}},Tamil:{names:["Bindu","Consonant","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Modifying_Letter","Number","Other","Placeholder","Virama","Vowel_Dependent","Vowel_Dependent.1","Vowel_Independent","Vowel_Independent.1"],blocks:{"0B80":"..08.fggggg...ggg.ggg1...11.1.11...11...111...111111111111....dddde...ddd.dddc..a......d..............9999999999aaaaaaaaaaa"},shared:"546723bb",start:"0cde",pending:"",pairs:{"0":"cde","2":"0cde","3":"0cde","4":"0cde","5":"0cde","6":"0cde","7":"0cde","8":"0cde",a:"0cde",c:"cde",f:"e"},triples_broken:{"00":"0"},triples_whole:{"14":"0cde","15":"0cde","94":"0cde","95":"0cde",b4:"0cde",b5:"0cde",c4:"0",c5:"0",d4:"0de",d5:"0de",dc:"de",e4:"0de",e5:"0de",ec:"de",f4:"0cde",f5:"0cde",g4:"0cde",g5:"0cde"},quads_broken:{},quads_whole:{"144":"0cde","145":"0cde","154":"de","155":"de","944":"de","945":"de","954":"de","955":"de",b44:"de",b45:"de",b54:"de",b55:"de",c44:"0",c45:"0",c54:"0",c55:"0",d44:"de",d45:"de",d54:"de",d55:"de",e44:"de",e45:"de",e54:"de",e55:"de",f44:"de",f45:"de",f54:"de",f55:"de",g44:"de",g45:"de",g54:"de",g55:"de"}},Telugu:{names:["Avagraha","Bindu","Consonant","Consonant_Dead","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Nukta","Number","Other","Placeholder","Virama","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Dependent.3","Vowel_Dependent.4","Vowel_Independent","Vowel_Independent.1"],blocks:{"0C00":"111f1llllllll.lll.mll22222222222222222222.2222222222222222..a0ghggggg.hgg.hgie.......jk.222..3..llgg..bbbbbbbbbb.......ccccccccc"},shared:"768945dd",start:"1aefghijk",pending:"",pairs:{"0":"eghijk","1":"aeghijk","4":"1aefghijk","5":"1aefghijk","6":"1aefghijk","7":"1aefghijk","8":"1aefghijk","9":"1aefghijk",c:"1aefghijk",e:"eghi",f:"aeghijk",h:"j",m:"ij"},triples_broken:{"0a":"aeghijk","11":"1f","1f":"1f",aa:"ajk",aj:"a",ak:"a",ea:"eghi",ej:"e",ek:"e",f1:"1f",ff:"1f",ga:"a",ha:"a",ia:"a",ja:"a",ka:"a"},triples_whole:{"06":"1f","07":"1f","26":"1aefghijk","27":"1efghijk","36":"1aefghijk","37":"1efghijk",a6:"1efghijk",a7:"1efghijk",b6:"1efghijk",b7:"1efghijk",d6:"1efghijk",d7:"1efghijk",e6:"1af",e7:"1f",g6:"1fghijk",g7:"1fghijk",ge:"ghi",h6:"1fghijk",h7:"1fghijk",he:"ghi",i6:"1fghijk",i7:"1fghijk",ie:"ghi",j6:"1fghijk",j7:"1fghijk",je:"ghi",k6:"1fghijk",k7:"1fghijk",ke:"ghi",l6:"1efghijk",l7:"1efghijk",m6:"1efghijk",m7:"1efghijk"},quads_broken:{},quads_whole:{"266":"1efghijk","267":"1efghijk","276":"ghijk","277":"ghijk","366":"1efghijk","367":"1efghijk","376":"ghijk","377":"ghijk",a66:"ghijk",a67:"ghijk",a76:"ghijk",a77:"ghijk",b66:"ghijk",b67:"ghijk",b76:"ghijk",b77:"ghijk",d66:"ghijk",d67:"ghijk",d76:"ghijk",d77:"ghijk",e66:"1f",e67:"1f",e76:"1f",e77:"1f",g66:"ghijk",g67:"ghijk",g76:"ghijk",g77:"ghijk",h66:"ghijk",h67:"ghijk",h76:"ghijk",h77:"ghijk",i66:"ghijk",i67:"ghijk",i76:"ghijk",i77:"ghijk",j66:"ghijk",j67:"ghijk",j76:"ghijk",j77:"ghijk",k66:"ghijk",k67:"ghijk",k76:"ghijk",k77:"ghijk",l66:"ghijk",l67:"ghijk",l76:"ghijk",l77:"ghijk",m66:"ghijk",m67:"ghijk",m76:"ghijk",m77:"ghijk"}},Kannada:{names:["Avagraha","Bindu","Bindu.1","Consonant","Consonant_Dead","Consonant_With_Stacker","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Nukta","Number","Other","Placeholder","Virama","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Independent","Vowel_Independent.1","Vowel_Independent.2"],blocks:{"0C80":"122hellllmlml.lll.nll33333333333333333333.3333333333.33333..c0ijjjjjj.jjj.jjkg.......jj......43.lljj..dddddddddd.552"},shared:"98ab67ff",start:"2cghijk",pending:"",pairs:{"0":"gijk","2":"cgijk","5":"2cghijk","6":"2cghijk","7":"2cghijk","8":"2cghijk","9":"2cghijk",a:"2cghijk",b:"2cghijk",e:"2cghijk",g:"gijk",h:"cgijk",m:"i",n:"k"},triples_broken:{"0c":"cgijk","22":"2h","2h":"2h",cc:"c",gc:"gijk",h2:"2h",hh:"2h",ic:"c",jc:"c",kc:"c"},triples_whole:{"08":"2h","09":"2h","18":"2ghijk","19":"2ghijk","38":"2cghijk","39":"2ghijk","48":"2cghijk","49":"2ghijk",c8:"2ghijk",c9:"2ghijk",d8:"2ghijk",d9:"2ghijk",f8:"2ghijk",f9:"2ghijk",g8:"2ch",g9:"2h",i8:"2hijk",i9:"2hijk",ig:"ijk",j8:"2hijk",j9:"2hijk",jg:"ijk",k8:"2hijk",k9:"2hijk",kg:"ijk",l8:"2ghijk",l9:"2ghijk",m8:"2ghijk",m9:"2ghijk",n8:"2ghijk",n9:"2ghijk"},quads_broken:{},quads_whole:{"188":"ijk","189":"ijk","198":"ijk","199":"ijk","388":"2ghijk","389":"2ghijk","398":"ijk","399":"ijk","488":"2ghijk","489":"2ghijk","498":"ijk","499":"ijk",c88:"ijk",c89:"ijk",c98:"ijk",c99:"ijk",d88:"ijk",d89:"ijk",d98:"ijk",d99:"ijk",f88:"ijk",f89:"ijk",f98:"ijk",f99:"ijk",g88:"2h",g89:"2h",g98:"2h",g99:"2h",i88:"ijk",i89:"ijk",i98:"ijk",i99:"ijk",j88:"ijk",j89:"ijk",j98:"ijk",j99:"ijk",k88:"ijk",k89:"ijk",k98:"ijk",k99:"ijk",l88:"ijk",l89:"ijk",l98:"ijk",l99:"ijk",m88:"ijk",m89:"ijk",m98:"ijk",m99:"ijk",n88:"ijk",n89:"ijk",n98:"ijk",n99:"ijk"}},Malayalam:{names:["Avagraha","Bindu","Bindu.1","Consonant","Consonant_Dead","Consonant_Preceding_Repha","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Number","Other","Placeholder","Pure_Killer","Virama","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Dependent.3","Vowel_Independent","Vowel_Independent.1","Vowel_Independent.2","Vowel_Independent.3"],blocks:{"0D00":"111h2mmnmnmmm.omm.pmm33333333333333333333333333333333333333ff0ijjjjjj.kjj.jjjg5d....444ldddddddmmmjj..ccccccccccdddddddddd444444"},shared:"98ab67ee",start:"1fghijkl",pending:"5",pairs:{"0":"fgijkl","1":"fgijkl","5":"016789abdfghijkl","6":"1fghijkl","7":"1fghijkl","8":"1fghijkl","9":"1fghijkl",a:"1fghijkl",b:"1fghijkl",d:"1fghijkl",g:"fgijkl",h:"fgijkl",n:"l",o:"k",p:"il"},triples_broken:{"11":"1h","1h":"1h",h1:"1h",hh:"1h"},triples_whole:{"08":"1h","09":"1h","28":"1fghijkl","29":"1fghijkl","38":"1fghijkl","39":"1fghijkl","48":"1fghijkl","49":"1fghijkl",c8:"1fghijkl",c9:"1fghijkl",e8:"1fghijkl",e9:"1fghijkl",f8:"1fhijkl",f9:"1fhijkl",fg:"fijkl",g8:"1h",g9:"1h",i8:"1fhijkl",i9:"1fhijkl",ig:"fijkl",j8:"1fhijkl",j9:"1fhijkl",jg:"fijkl",k8:"1fhijkl",k9:"1fhijkl",kg:"fijkl",l8:"1fhijkl",l9:"1fhijkl",lg:"fijkl",m8:"1fghijkl",m9:"1fghijkl",n8:"1fghijkl",n9:"1fghijkl",o8:"1fghijkl",o9:"1fghijkl",p8:"1fghijkl",p9:"1fghijkl"},quads_broken:{},quads_whole:{"288":"fijkl","289":"fijkl","298":"fijkl","299":"fijkl","388":"1fghijkl","389":"1fghijkl","398":"fijkl","399":"fijkl","488":"1fghijkl","489":"1fghijkl","498":"fijkl","499":"fijkl",c88:"fijkl",c89:"fijkl",c98:"fijkl",c99:"fijkl",e88:"fijkl",e89:"fijkl",e98:"fijkl",e99:"fijkl",f88:"fijkl",f89:"fijkl",f98:"fijkl",f99:"fijkl",g88:"1h",g89:"1h",g98:"1h",g99:"1h",i88:"fijkl",i89:"fijkl",i98:"fijkl",i99:"fijkl",j88:"fijkl",j89:"fijkl",j98:"fijkl",j99:"fijkl",k88:"fijkl",k89:"fijkl",k98:"fijkl",k99:"fijkl",l88:"fijkl",l89:"fijkl",l98:"fijkl",l99:"fijkl",m88:"fijkl",m89:"fijkl",m98:"fijkl",m99:"fijkl",n88:"fijkl",n89:"fijkl",n98:"fijkl",n99:"fijkl",o88:"fijkl",o89:"fijkl",o98:"fijkl",o99:"fijkl",p88:"fijkl",p89:"fijkl",p98:"fijkl",p99:"fijkl"}},Arabic:{names:["Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Other","Placeholder"],blocks:{"0600":"6666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666666","0750":"666666666666666666666666666666666666666666666666"},shared:"32450177",start:"",pending:"",pairs:{},triples_broken:{},triples_whole:{},quads_broken:{},quads_whole:{}},OlChiki:{names:["Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Other","Placeholder"],blocks:{"1C50":"666666666666666666666666666666666666666666666666"},shared:"32450177",start:"",pending:"",pairs:{},triples_broken:{},triples_whole:{},quads_broken:{},quads_whole:{}},MeeteiMayek:{names:["Consonant","Consonant_Final","Inv-f","Inv-h","Inv-j","Inv-n","Inv-s","Inv-w","Invisible_Stacker","Number","Other","Placeholder","Pure_Killer","Tone_Mark","Visarga","Vowel_Dependent","Vowel_Dependent.1","Vowel_Dependent.2","Vowel_Dependent.3","Vowel_Independent"],blocks:{AAE0:"jj000000000fghfiaaaaae8",ABC0:"00000000000000jj0j00000000011111111iihiigiiadc..9999999999"},shared:"546723bb",start:"58cdefghi",pending:"",pairs:{"2":"58cdefghi","3":"58cdefghi","4":"58cdefghi","5":"5","6":"58cdefghi","7":"58cdefghi","8":"8cdefghi",c:"8fh",d:"8cfghi",e:"8cfghi",f:"8",g:"8fh",h:"8f",i:"8cfgh"},triples_broken:{"85":"8cdefghi",c5:"8fh",d5:"8cfghi",e5:"8cfghi",f5:"8",g5:"8fh",h5:"8f",i5:"8cfgh"},triples_whole:{"04":"58cdefghi","14":"58cdefghi","54":"8cdefghi","84":"5","94":"58cdefghi",a4:"58cdefghi",b4:"58cdefghi",c4:"5cdegi",d4:"5de",e4:"5de",f4:"5cdefghi",g4:"5cdegi",h4:"5cdeghi",i4:"5dei",j4:"58cdefghi"},quads_broken:{"845":"8cdefghi","854":"8cdefghi",c45:"8fh",c54:"8fh",d45:"8cfghi",d54:"8cfghi",e45:"8cfghi",e54:"8cfghi",f45:"8",f54:"8",g45:"8fh",g54:"8fh",h45:"8f",h54:"8f",i45:"8cfgh",i54:"8cfgh"},quads_whole:{"044":"58cdefghi","144":"58cdefghi","544":"8cdefghi","844":"5","944":"58cdefghi",a44:"58cdefghi",b44:"58cdefghi",c44:"5cdegi",d44:"5de",e44:"5de",f44:"5cdefghi",g44:"5cdegi",h44:"5cdeghi",i44:"5dei",j44:"58cdefghi"}}}};function w(g){let f=new Map,b=Array.from(g.invisible+g.placeholder,(l)=>l.codePointAt(0)),q=new Set(b),m=(l)=>new Set(Object.entries(l).flatMap(([n,k])=>[...k].map((j)=>n+j)));for(let l of Object.values(g.scripts)){let n={cls:new Map,start:new Set(l.start),pending:new Set(l.pending),pairs:m(l.pairs),tb:m(l.triples_broken),tw:m(l.triples_whole),qb:m(l.quads_broken),qw:m(l.quads_whole)};for(let[k,j]of Object.entries(l.blocks)){let i=parseInt(k,16);for(let e=0;e<j.length;e++)if(j[e]!==".")n.cls.set(i+e,j[e]),f.set(i+e,n)}b.forEach((k,j)=>n.cls.set(k,l.shared[j]))}function t(l,n){let k="",j=0;for(let i=0;i<n.length;i++){let e=l.cls.get(n[i]),h;if(j===0)h=l.start.has(e);else if(j>=3&&l.qb.has(k.slice(-3)+e))h=!0;else if(j>=3&&l.qw.has(k.slice(-3)+e))h=!1;else if(j>=2&&l.tb.has(k.slice(-2)+e))h=!0;else if(j>=2&&l.tw.has(k.slice(-2)+e))h=!1;else h=l.pairs.has(k.slice(-1)+e);if(h)return i;if(k+=e,j+=e.length,k.length>3)k=k.slice(-3)}if(j>0&&l.pending.has(k[k.length-1]))return n.length-1;return-1}return function(n){let k,j=[],i=0,e=0,h=()=>k&&j.length?t(k,j):-1;for(let r of n){let d=r.codePointAt(0),a=f.get(d);if(a===void 0&&q.has(d)||a!==void 0&&(k===void 0||a===k)){if(j.length===0)i=e;if(j.push(d),a!==void 0)k=a}else{let c=h();if(c>=0)return i+c;j=a?[d]:[],k=a,i=e}e+=r.length}let p=h();return p>=0?i+p:-1}}var s=w(o);function V(g){return s(g)<0}function D(g){return s(g)}var u=o.version;export{V as isWellFormed,D as brokenAt,u as RULES_VERSION};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "indickit",
-  "version": "0.6.0",
+  "version": "0.9.0",
   "description": "Small, exact utilities for text in Indian languages: a phonetic key that matches a name across 21 Indian scripts and English, a normalizer that gives text that looks the same one encoding, a segmenter for the letters a reader sees, and a stemmer for search.",
   "license": "MIT",
   "packageManager": "bun@1.3.14",
@@ -29,6 +29,18 @@
     "./romanize": {
       "types": "./dist/types/js/romanize.d.ts",
       "default": "./dist/romanize.js"
+    },
+    "./phonetic-search": {
+      "types": "./dist/types/js/phonetic-search.d.ts",
+      "default": "./dist/phonetic-search.js"
+    },
+    "./deromanize": {
+      "types": "./dist/types/js/deromanize.d.ts",
+      "default": "./dist/deromanize.js"
+    },
+    "./wellformed": {
+      "types": "./dist/types/js/wellformed.d.ts",
+      "default": "./dist/wellformed.js"
     }
   },
   "files": [
@@ -36,7 +48,7 @@
   ],
   "sideEffects": false,
   "scripts": {
-    "build": "bun build js/phonetic.ts --outfile dist/phonetic.js --target browser --format esm && bun build js/normalize.ts --outfile dist/normalize.js --target browser --format esm && bun build js/segment.ts --outfile dist/segment.js --target browser --format esm && bun build js/stem.ts --outfile dist/stem.js --target browser --format esm && bun build js/romanize.ts --outfile dist/romanize.js --target browser --format esm && tsc -p tsconfig.json",
+    "build": "bun build js/phonetic.ts --outfile dist/phonetic.js --target browser --format esm --minify && bun build js/normalize.ts --outfile dist/normalize.js --target browser --format esm --minify && bun build js/segment.ts --outfile dist/segment.js --target browser --format esm --minify && bun build js/stem.ts --outfile dist/stem.js --target browser --format esm --minify && bun build js/romanize.ts --outfile dist/romanize.js --target browser --format esm --minify && bun build js/phonetic-search.ts --outfile dist/phonetic-search.js --target browser --format esm --minify && bun build js/deromanize.ts --outfile dist/deromanize.js --target browser --format esm --minify && bun build js/wellformed.ts --outfile dist/wellformed.js --target browser --format esm --minify && tsc -p tsconfig.json",
     "test": "bun test",
     "check-dist": "node js/check-dist.mjs"
   },