[prakriya] Refactor dhatu logic

Our previous rule ordering made certain derivations impossible. This
commit standardizes our rule order as follows:

1. We create a complete sanadi dhatu before adding any other pratyayas.
2. Generally, we apply guna before applying dvitva.

Along the way, we found and corrected a variety of small errors and
substantially increased our rule coverage and unit test coverage:

Rules:      ~1094 --> ~1196
Unit tests:   755 -->   973

vidyut-prakriya/.gitignore

@@ -0,0 +1 @@
+www/static/*

vidyut-prakriya/Makefile

@@ -53,28 +53,28 @@ test_tinantas:
 	cargo build --release
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-basic-kartari.csv \
-	    --hash "72f94a7fa706cb5b2c96b0f74baa17f1a0c263d96cb92edc669f244dfabafcb2"
+	  --hash "ab8ff253c90d44461bd71c9681d748141d77685758282ea04484966fb1014bfa"
 	../target/release/test_tinantas \
-		--test-cases test-files/tinantas-nic-kartari.csv \
-	    --hash "c16ff79b6041e25a3297a614611515537020712341febb340e30ace459a8104b"
+  	--test-cases test-files/tinantas-nic-kartari.csv \
+    --hash "f8a3b4c50eb88eda52db0a6936ff694168d1a80408a015d71561ab88b0e43269"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-san-kartari.csv \
-	    --hash "ba174c9268b8be00262dd1f856161b2cb505b515c4e232feb70cb711eb6ce652"
+		--hash "04bfbbd22493ae53c7d15c753131242d3923c4d4d5e1aee847665cb87bb2d9f1"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-yan-kartari.csv \
-	    --hash "1645ee474a8e370b6cd1a5d487092f2cc84864afd706e2a1ddabea6a8e4c49e3"
+		--hash "2679e81f6a1e3a2957df5a4fdd6b8bd9666d42d1fb79ab5122ad02bd3e8dd7f2"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-basic-karmani.csv \
-	    --hash "a2bbcd572c04973eee057d58a0caf93ea84fb8eb18845633ce4ef271a0abdffd"
+		--hash "cfb9bf65e38289abc3119f839e41615ecd2bb7f548a8ef78a45cd3d24c183d59"
 
 test_krdantas:
 	cargo build --release
 	../target/release/test_krdantas \
 		--test-cases test-files/krdantas-ktvA.csv \
-	    --hash "8b7e855313d13de73b34c55e1c41daad262575365d2a5748c2469de09a3df15a"
+	    --hash "e8be9f718fa56dd602be71166d1bcfd24da054e77dd2cd628dedda1b004bc4f8"
 	../target/release/test_krdantas \
 		--test-cases test-files/krdantas-kta.csv \
-	    --hash "b13cefedbb913354a12b47ff6ae4187c335e2ebe73066b601ac7eeb32c10ea2a"
+	    --hash "223db03e3c5bd4492919cf8fd27d09fd23c0660449df6596bcf6c6dbe9c45caa"
 
 test_subantas:
 	cargo run --bin test_subantas -- \

vidyut-prakriya/README.md

@@ -122,7 +122,7 @@ can take in the comments on the `Rule` type. We suggest using `ashtadhyayi.com`
 to learn more about these rules.
 
 The right column shows the in-progress prakriya. We use an output convention
-that is common on other Ashtadhyayi websites. The encoding foromat for this
+that is common on other Ashtadhyayi websites. The encoding format for this
 text is SLP1, which is the encoding format we use throughout the crate.
 
 [sv]: https://github.com/drdhaval2785/SanskritVerb

vidyut-prakriya/data/dhatupatha.tsv

@@ -1636,7 +1636,7 @@ code	dhatu	artha
 06.0165	mila~^	saNgame
 06.0166	mu\cx~^	mokzaRe
 06.0167	lu\px~^	Cedane
-06.0168	vidx~^	lABe
+06.0168	vi\dx~^	lABe
 06.0169	li\pa~^	upadehe
 06.0170	zi\ca~^	kzaraRe
 06.0171	kftI~	Cedane

vidyut-prakriya/src/ac_sandhi.rs

@@ -22,6 +22,7 @@ lazy_static! {
     static ref EC: Set = s("ec");
     static ref VAL: Set = s("val");
     static ref HAL: Set = s("hal");
+    static ref YAN: Set = s("yaR");
 }
 
 pub fn try_lopo_vyor_vali(p: &mut Prakriya) {
@@ -37,9 +38,11 @@ pub fn try_lopo_vyor_vali(p: &mut Prakriya) {
             let t = get_at(p, i).expect("should be present");
             // Ignore if it starts an upadesha, otherwise roots like "vraj" would by vyartha.
             // Likewise for roots ending with 'v'.
+            // Likewise for pratipadikas.
+            //
             // For now, just check if the term is a dhatu.
             let is_upadesha_adi = t.is_dhatu() && (t.has_adi('v') || t.has_adi('y'));
-            vyor_vali && !is_upadesha_adi
+            vyor_vali && !is_upadesha_adi && !t.is_pratipadika()
         },
         |p, _, i| {
             set_at(p, i, "");
@@ -92,10 +95,21 @@ pub fn apply_general_ac_sandhi(p: &mut Prakriya) {
         },
     );
 
+    // HACK: ignore sandhi between upasarga and dhatu so that we can correctly derive prARinat,
+    // etc.
+    fn is_upasarga_sanadi_dhatu(p: &Prakriya, i: usize) -> bool {
+        get_at(p, i).expect("present").is_upasarga()
+            && p.terms().last().expect("present").is_dhatu()
+    }
+
     char_rule(
         p,
         xy(|x, y| AK.contains(x) && AK.contains(y) && al::savarna(x).contains(y)),
         |p, text, i| {
+            if is_upasarga_sanadi_dhatu(p, i) {
+                return false;
+            }
+
             let x = text.as_bytes()[i] as char;
             set_at(p, i, &al::to_dirgha(x).expect("should be ac").to_string());
             set_at(p, i + 1, "");
@@ -165,14 +179,19 @@ pub fn apply_general_ac_sandhi(p: &mut Prakriya) {
         p,
         xy(|x, y| A.contains(x) && AC.contains(y)),
         |p, text, i| {
+            if is_upasarga_sanadi_dhatu(p, i) {
+                return false;
+            }
+
             let j = i + 1;
             let y = text.as_bytes()[i + 1] as char;
+
             if EC.contains(y) {
-                set_at(p, j, al::to_vrddhi(y).expect("should be set"));
+                set_at(p, j, al::to_vrddhi(y).expect("should have vrddhi"));
                 set_at(p, i, "");
                 p.step("6.1.88");
             } else {
-                set_at(p, j, al::to_guna(y).expect("should be set"));
+                set_at(p, j, al::to_guna(y).expect("should have guna"));
                 set_at(p, i, "");
                 p.step("6.1.87");
             }
@@ -361,7 +380,30 @@ fn hacky_apply_ni_asiddhavat_rules(p: &mut Prakriya) -> Option<()> {
     Some(())
 }
 
-pub fn run_common(p: &mut Prakriya) {
+/// Runs antaranga ac-sandhi rules.
+///
+/// (Example: div -> dyU + sa -> dudyUzati)
+pub fn run_antaranga(p: &mut Prakriya) -> Option<()> {
+    for i in 0..p.terms().len() {
+        let cur = p.get(i)?;
+        if cur.has_upadha(&*IK) && cur.has_antya(&*AC) {
+            let x = cur.upadha()?;
+            let res = match x {
+                'i' | 'I' => "y",
+                'u' | 'U' => "v",
+                'f' | 'F' => "r",
+                'x' | 'X' => "l",
+                _ => panic!("Unexpected res"),
+            };
+            p.op_term("6.1.77", i, |t| t.set_upadha(res));
+        }
+    }
+
+    p.maybe_save_sthanivat();
+    Some(())
+}
+
+pub fn run_common(p: &mut Prakriya) -> Option<()> {
     try_ver_aprktasya(p);
 
     for i in 0..p.terms().len() {
@@ -372,4 +414,6 @@ pub fn run_common(p: &mut Prakriya) {
     hacky_apply_ni_asiddhavat_rules(p);
 
     try_sut_kat_purva(p);
+
+    Some(())
 }