[prakriya] Increase Kaumudi test coverage

Quality: - Substantially increase test coverage in Kaumudi 43 - 52. - Add around 70 new tests. - Model varttikas explicitly and consistently with the `Varttika` variant. - Add varttikas and Dhatupatha gana-sutras to `data/`. - Fix various small typos in `dhatupatha.tsv`. - Increase performance by around 10%. We did this mainly by reducing use of `CompactString` in routines that focus on string manipulation and avoiding an unnecessary allocation in the `it_samjna` rules. - Add broad support for kaNDvAdi-dhAtus. - Add better support for iT-Agama rules with liT. API: - Upgrade wasm API to more richly model the effects of a rule. - Update web frontend to hide empty terms and show varttika text. Development: - Add `create_all_tinantas` script to generate all tinantas and write them to one file.
ambuda-org · Dec 11, 2023 · 2eb4434 · 2eb4434
1 parent b39418f
commit 2eb4434
Show file tree

Hide file tree

Showing 81 changed files with 3,612 additions and 1,148 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/vidyut-lipi/src/lib.rs b/vidyut-lipi/src/lib.rs
@@ -1,17 +1,19 @@
+//! Hacky transliteration functions that other crates might need.
 #![doc = include_str!("../README.md")]
 #![deny(missing_docs)]
 #![deny(clippy::unwrap_used)]
 
-//! Hacky transliteration functions that other crates might need.
 use std::cmp;
 
 /// Defines the available transliteration schemes.
-#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+#[derive(Clone, Copy, Debug, Hash, Eq, PartialEq)]
 pub enum Scheme {
     /// SlP1 transliteration.
     Slp1,
     /// IAST transliteration.
     Iast,
+    /// Devanagari.
+    Devanagari,
 }
 
 fn map_char(cur: &str) -> Option<&'static str> {
@@ -87,13 +89,118 @@ fn iast_to_slp1(input: &str) -> String {
     ret
 }
 
+fn slp1_to_devanagari(text: &str) -> String {
+    const VIRAMA: char = '\u{094D}';
+
+    let mut ret = String::new();
+    for c in text.chars() {
+        let out = match c {
+            'a' => "अ",
+            'A' => "आ",
+            'i' => "इ",
+            'I' => "ई",
+            'u' => "उ",
+            'U' => "ऊ",
+            'f' => "ऋ",
+            'F' => "ॠ",
+            'x' => "ऌ",
+            'X' => "ॡ",
+            'e' => "ए",
+            'E' => "ऐ",
+            'o' => "ओ",
+            'O' => "औ",
+            '~' => "\u{0901}",
+            'M' => "\u{0902}",
+            'H' => "\u{0903}",
+            'k' => "क",
+            'K' => "ख",
+            'g' => "ग",
+            'G' => "घ",
+            'N' => "ङ",
+            'c' => "च",
+            'C' => "छ",
+            'j' => "ज",
+            'J' => "झ",
+            'Y' => "ञ",
+            'w' => "ट",
+            'W' => "ठ",
+            'q' => "ड",
+            'Q' => "ढ",
+            'R' => "ण",
+            't' => "त",
+            'T' => "थ",
+            'd' => "द",
+            'D' => "ध",
+            'n' => "न",
+            'p' => "प",
+            'P' => "फ",
+            'b' => "ब",
+            'B' => "भ",
+            'm' => "म",
+            'y' => "य",
+            'r' => "र",
+            'l' => "ल",
+            'v' => "व",
+            'S' => "श",
+            'z' => "ष",
+            's' => "स",
+            'h' => "ह",
+            'L' => "ळ",
+            other => {
+                ret.push(other);
+                continue;
+            }
+        };
+
+        let vowel_mark = match c {
+            'a' => Some(""),
+            'A' => Some("\u{093E}"),
+            'i' => Some("\u{093F}"),
+            'I' => Some("\u{0940}"),
+            'u' => Some("\u{0941}"),
+            'U' => Some("\u{0942}"),
+            'f' => Some("\u{0943}"),
+            'F' => Some("\u{0944}"),
+            'x' => Some("\u{0962}"),
+            'X' => Some("\u{0963}"),
+            'e' => Some("\u{0947}"),
+            'E' => Some("\u{0948}"),
+            'o' => Some("\u{094B}"),
+            'O' => Some("\u{094C}"),
+            _ => None,
+        };
+
+        if ret.chars().last() == Some(VIRAMA) && vowel_mark.is_some() {
+            // Pop virama and add.
+            ret.pop();
+            ret += vowel_mark.expect("ok");
+        } else {
+            ret += out;
+        }
+
+        let is_consonant = "kKgGNcCjJYwWqQRtTdDnpPbBmyrlvSzshL".contains(c);
+        if is_consonant {
+            ret.push(VIRAMA);
+        }
+    }
+    ret
+}
+
 /// Transliterates the given input text.
 ///
-/// Only the IAST -> SLP1 mapping is defined. All other mappings will panic.
+/// ### Panics
+///
+/// Only the IAST -> SLP1 and SLP1 -> Devanagari mappings are defined. All other mappings will
+/// panic.
 pub fn transliterate(input: &str, from: Scheme, to: Scheme) -> String {
-    assert!(from == Scheme::Iast);
-    assert!(to == Scheme::Slp1);
-    iast_to_slp1(input)
+    use Scheme::*;
+    if from == Iast && to == Slp1 {
+        iast_to_slp1(input)
+    } else if from == Slp1 && to == Devanagari {
+        slp1_to_devanagari(input)
+    } else {
+        panic!("Unsupported scheme combination: {from:?} -> {to:?}")
+    }
 }
 
 #[cfg(test)]

diff --git a/vidyut-prakriya/Cargo.toml b/vidyut-prakriya/Cargo.toml
@@ -26,6 +26,7 @@ console_error_panic_hook = "0.1.7"
 
 [dev-dependencies]
 test_utils = { path = "test_utils" }
+vidyut-lipi = { path = "../vidyut-lipi" }
 
 [lib]
 crate-type = ["cdylib", "rlib"]
diff --git a/vidyut-prakriya/Makefile b/vidyut-prakriya/Makefile
@@ -53,40 +53,40 @@ test_tinantas:
 	cargo build --release
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-basic-kartari.csv \
-	  --hash "ff57ac933f1a44a21285448f1b064d1b4464a7aee650981cb30244c8ee3ed111"
+	  --hash "28293a6aa746f04dff1a5c3b822396c236df269460d691bd137ba01a2c80557f"
 	../target/release/test_tinantas \
   	--test-cases test-files/tinantas-nic-kartari.csv \
-    --hash "263fa8bc08eeb912ffe4daa876452a1bea1a4117620d4e3d86aa876f03795140"
+    --hash "5d508a8b3a52ebe92a1524b0b51bbed90e27301f265669aa8e7a887d55bf1eff"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-san-kartari.csv \
-		--hash "c9c2c2e79eecf4a6bbe105537ceb64f77004ce3ff3b53a07a4ac410f0131de0e"
+		--hash "ef278531cd45ba1df9af60e5b4b8ef582fdbe7a7eb31478fe35b9600c42380fc"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-yan-kartari.csv \
-		--hash "585c83fa6a241d7f31d2b12e7382e65d05b3f7703edaa1609ee3617810aa74d2"
+		--hash "be90fccc6f3907862cf820969094fd958b44b58a447a7a6369edbecde75d8cd4"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-yan-luk-kartari.csv \
-		--hash "9168fb70bce7903708004da407dfe1c14aedc199ab770c2ee524b7f573233cfc"
+		--hash "aec7b4524b159e195e38ce4b0e5e1335940b60cb41606ba31d1b9fbd010f7d11"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-basic-karmani.csv \
-    --hash "63d076bf9d8b88d7a03f224626410ec2be751f2b50ef0f0777c5cddcecad3dfd"
+    --hash "b3621fe5bf2295f2f1d15270aa26b9fc973450126cd4a551e53f91571e4d531c"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-nic-karmani.csv \
-    --hash "2b9aa1158d26ce7a9da28f7ec8685ce61554daa3f8434770aec3e16e000eefe2"
+    --hash "f4760cd6da0496f6db82f1812ba168a18e8cde4e5dffafee815e1e88d19abb36"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-san-karmani.csv \
-    --hash "e60669056aec2124c27c8d60f3cebe19b75e11d193e87522458be06291db221f"
+    --hash "089160604d7a5a40f8f71e1234cccc3667135b3e2654c4ac346677727e5aae80"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-yan-karmani.csv \
-    --hash "7e0ed1e074eff8eb104dc88b8cf7525d1e28b8686bb82dccb9d036a4696caed7"
+    --hash "260a684b78e4e66a68f89366c07850d1680a72424e2934392678359289a1916b"
 
 test_krdantas:
 	cargo build --release
 	../target/release/test_krdantas \
 		--test-cases test-files/krdantas-ktvA.csv \
-	    --hash "6d480b70491153eeac8958d5f66c32e4e3e428f54c6e0fc2d75154889642ef7b"
+	    --hash "59470473bec6cb2d2254217d604a2302c6ee8926571441b3281004a30abc81ce"
 	../target/release/test_krdantas \
 		--test-cases test-files/krdantas-kta.csv \
-	    --hash "bb99612ddcc34a66aff16553ed325f8b1a09d4bfcc920f86d312c959c6b1800b"
+	    --hash "2ce641122f4c6ac243acb348d376f35586c5f960ea338b9b1f7b79f0f72e3c12"
 
 test_subantas:
 	cargo run --bin test_subantas -- \
@@ -102,7 +102,7 @@ check_rule_coverage:
 # Profiles the program's execution time on OSX. This command will probably not
 # work on other operating systems.
 profile-time-osx:
-	cargo instruments -t time --release --bin create_test_file
+	cargo instruments -t time --release --example create_all_tinantas > /dev/null
 
 
 # Other

diff --git a/vidyut-prakriya/data/dhatupatha-ganasutras.tsv b/vidyut-prakriya/data/dhatupatha-ganasutras.tsv
@@ -0,0 +1,18 @@
+01.0933_GawAdayo mitaH
+01.0934_janIjFzkanasuraYjo'mantASca
+01.0935_jvalahvalahmalanamAmanupasargAdvA
+01.0936_glAsnAvanuvamAM ca
+01.0937_na kamyamicamAm
+01.0938_Samo darSane
+01.0939_yamo'parivezaRe
+01.0940_sKadiravapariByAM ca
+02.0076_carkarItaM ca
+04.0162_svAdaya oditaH
+10.0493_jYapAdayo mitaH
+10.0494_nAnye mito'hetO
+10.0495_kusma nAmno vA
+10.0496_A kusmAdAtmanepadinaH
+10.0497_A garvAdAtmanepadinaH
+10.0498_A DfzAdvA
+10.0499_A svadaH sakarmakAt
+10.0500_hantyarTASca
diff --git a/vidyut-prakriya/data/dhatupatha.tsv b/vidyut-prakriya/data/dhatupatha.tsv
@@ -781,7 +781,7 @@ code	dhatu	artha
 01.0780	Iza~	uYCe
 01.0781	kaza~	hiMsAyAm
 01.0782	Kaza~	hiMsAyAm
-01.0783	Siza~	hiMsAyAm
+01.0783	Si\\za~	hiMsAyAm
 01.0784	jaza~	hiMsAyAm
 01.0785	Jaza~	hiMsAyAm
 01.0786	Saza~	hiMsAyAm
@@ -1246,7 +1246,7 @@ code	dhatu	artha
 03.0002	YiBI\	Baye
 03.0003	hrI\	lajjAyAm
 03.0004	pF	pAlanapUraRayoH
-03.0005	pf	pAlanapUraRayoH
+03.0005	pf\	pAlanapUraRayoH
 03.0006	quBf\Y	DAraRapozaRayoH
 03.0007	mA\N	mAne Sabde ca
 03.0008	o~hA\N	gatO
@@ -1785,7 +1785,7 @@ code	dhatu	artha
 10.0034	kuwwa~	CedanaBartsanayoH
 10.0035	puwwa~	alpIBAve
 10.0036	cuwwa~	alpIBAve
-10.0037	awwa~	anAdare
+10.0037	adwa~	anAdare
 10.0038	zuwwa~	anAdare
 10.0039	lunwa~	steye
 10.0040	lunWa~	steye
@@ -1810,7 +1810,7 @@ code	dhatu	artha
 10.0059	Sliza~	SlezaRe
 10.0060	paTi~	gatO
 10.0061	piCa~	kuwwane
-10.0062	Cada~	saMvaraRe
+10.0062	Cadi~	saMvaraRe
 10.0063	SraRa~	dAne
 10.0064	taqa~	AGAte
 10.0065	Kaqa~	Bedane
@@ -2086,7 +2086,7 @@ code	dhatu	artha
 10.0335	ruja~	hiMsAyAm
 10.0336	zvada~	AsvAdane
 10.0337	svAda~	AsvAdane
-10.0338	yuja~	saMyamane
+10.0338	yu\ja~	saMyamane
 10.0339	pfca~	saMyamane
 10.0340	arca~	pUjAyAm
 10.0341	zaha~	marzaRe

diff --git a/vidyut-prakriya/data/varttikas.tsv b/vidyut-prakriya/data/varttikas.tsv
@@ -0,0 +1,93 @@
+1.1.33.1	viBAzAprakaraRe tIyasya NitsUpasaNKyAnam
+1.2.1.1	vyaceH kuwAditvamanasIti vaktavyam
+1.2.6.1	SranTigranTidamBisvaYjInAmiti vaktavyam
+1.3.3.1	ira itsaMjJA vAcyA
+1.3.22.1	ANaH pratijYAyAm
+1.3.29.1	dfSeSceti vaktavyam
+1.3.30.1	upasargAdasyatyUhrorvA vacanam
+1.4.60.3	duraH zatvaRatvayorupasargatvapratizeDo vaktavyaH
+1.4.65.1	antaHSabdasya aNkiviDiRatvezu upasargasaMjYA vaktavyA
+2.1.37.1	BayaBItaBItiBIBiriti vaktavyam
+2.4.45.1	iRvadika iti vaktavyam
+2.4.54.1	varjane pratizeDo vaktavyaH
+2.4.56.2	valAdAvArDaDAtuke vezyate
+3.1.11.1	ojaso'psaraso nityamitarezAM viBAzayA
+3.1.14.1	satrakazwakakzakfcCragahaneByaH kaRvacikIrzAyAm iti vaktavyam
+3.1.15.2	tapasaH parasmEpadaM ca
+3.1.16.1	PenAcceti vaktavyam
+3.1.17.1	sudinadurdinanIhareByaSceti vaktavyam
+3.1.17.2	awAwwASIkAkowApowAsowApruzwApluzwAgrahaRaM kartavyam
+3.1.22.1	sUcisUtramUtryawyartyaSUrRotInAM grahaRaM
+3.1.25.1	arTavedasatyAnAmApugvaktavyaH
+3.1.35.1	kAsyanekAca iti vaktavyam
+3.1.36.1	UrRoteSca pratizeDo vaktavyaH
+3.1.44.1	spfSamfSakfzatfpadfpAM cleH sijvA vaktavyaH
+3.1.48.1	RiSridrusruzu kamerupasaMKyAnaM kartavyam
+3.1.96.1	vasestavyat kartari Ricca
+3.1.96.2	kelimara upasaNKyAnam
+3.1.97.1	takiSasicatiyatijanInAmupasaNKyAnam
+3.1.97.2	hano vA vaDa ca
+3.1.100.1	carerANi cAgurO
+3.2.8.1	pibateH surASIDvoriti vaktavyam
+3.2.9.1	SaktilANgalANkuSayazwitomaraGawaGawIDanuzzu graherupasaNKyAnam
+3.2.9.2	sUtre ca DAryarTe
+3.2.60.1	samAnAnyayoSceti vaktavyam
+3.2.174.1	krukannapi vaktavyaH
+3.2.178.1	kvibvacipracCAyatastukawaprujuSrIRAM dIrGo'saMprasAraRaM ca
+3.3.125.1	KanerqaqarekekavakA vAcyAH
+4.1.6.1	DAtorugitaH pratizeDaH
+4.1.68.1	SvaSurasyokArAkAralopaSca
+4.1.97.1	vyAsavaruqanizAdacaRqAlabimbAnAm iti vaktavyam
+4.1.128.1	cawakAc ceti vaktavyam
+4.2.28.1	CaprakaraRe pENgAkzIputrAdiBya upasaNKyAnam
+4.2.28.2	SatarudrAcCaSca GaSca
+4.2.40.1	gaRikAyASca yaY vaktavyaH
+4.2.42.1	pfzWAdupasaNKyAnam
+4.2.43.1	gajAcceti vaktavyam
+4.2.93.1	avArapArAdvigfhItAdapi viparItAcceti vaktavyam
+4.4.41.1	aDarmAcceti vaktavyam
+4.4.49.1	narAcceti vaktavyam
+5.1.6.1	yatprakaraRe raTAcca
+5.1.33.1	kevalAyASceti vaktavyam
+5.1.38.2	saMnipAtAcceti vaktavyam
+5.1.74.1	kroSaSatayojanaSatayorupasaNKyAnam
+5.1.77.1	AhftaprakaraRe vArijaNgalasTalkAntArapUrvapadAdupasaNKyAnam
+5.1.77.2	ajapaTaSaNkupaTAByAM copasaNKyAnam
+5.1.77.3	maDukamaricayoraR sTalAt
+5.2.29.1	alAbUtilomABaNgAByo rajasyupasaNKyAnam
+5.2.112.1	anyeByo'pi dfSyata iti vaktavyam
+5.4.3.1	caYcabfhayorupasaNKyAnam
+5.4.78.1	palyarAjaByAM ca
+6.1.3.3	Irzyates tftIyasya dve Bavata iti vaktavyam
+6.1.64.1	subDAtuzWivuzvazkatInAM satvapratizeDo vaktavyaH
+6.1.64.2	zWivu ityasya dvitIyasTakArazWakAraSca izyate
+6.3.11.1	antAcca
+6.3.118.1	anutsAhaBrAtfpitFRAm ityeva
+6.4.16.1	gameriNAdeSasyeti vaktavyam
+6.4.24.2	raYjerRO mfgaramaRa upasaNKyAnaM kartavyam
+6.4.24.3	GinuRi ca raYjerupasaNKyAnaM kartavyam
+6.4.34.1	kvO ca SAsa ittvaM BavatIti vaktavyam
+6.4.34.2	kvippratyaye tu tasya api BavatIti vaktavyam
+6.4.84.1	dfnkarapunaH pUrvasya Buvo yaRvaktavyaH
+6.4.114.1	daridrAterArDaDAtuke lopo vaktavyaH
+6.4.114.2	adyatanyAM veti vaktavyam
+6.4.120.1	damBeretvaM vaktavyam
+6.4.122.1	SranTeSceti vaktavyam
+6.4.144.1	nAnatasya wilope sabrahmacAripIWasarpikalApikuTumitEtilijAjalilANgaliSilAliSiKaRqisUkarasadmasuparvaRAmupasaNKyAnaM kartavyam 
+6.4.148.1	yasyetyONaH SyAM pratizeDo vaktavyaH
+6.4.155.1	RAvizWavat prAtipadikasya kAryaM BavatIti vaktavyam
+7.1.26.1	ekatarAtpratizeDo vaktavyaH
+7.1.59.1	Se tfmPAdInAmupasaNKyAnaM kartavyam
+7.1.90.1	oto Riditi vAcyam
+7.2.49.1	tanipatidaridrARAmupasaNKyAnam
+7.2.68.1	dfSeSceti vaktavyam
+7.3.34.1	anAcamikamivamInAm iti vaktavyam
+7.3.37.1	lugAgamastu tasya vaktavyaH
+7.3.37.2	DUYprIYornugvaktavyaH
+7.4.3.1	kARyAdInAM ceti vaktavyam
+7.4.30.1	hanterhisAyAM yaNi GnIBAvo vaktavyaH
+7.4.54.1	sani rADo hiMsAyAmaca is vaktavyaH
+7.4.90.1	rIgftvata iti vaktavyam
+8.2.8.1	NAvuttarapade pratizeDasya pratizeDo vaktavyaH
+8.2.8.3	vA napuMsakAnAm iti vaktavyam
+8.3.118.1	svaYjerupasaNKyAnaM kartavyam