[prakriya] Add around 250 taddhita tests

This commit massively expands and streamlines the taddhita section.
Although more cleanup work can be done here, there's enough of a
framework that the rest of the rules are easier to fill in.

vidyut-prakriya/.gitignore

@@ -1,2 +1,3 @@
 www/static/data/
 www/static/wasm/
+*.log

vidyut-prakriya/Makefile

@@ -59,7 +59,7 @@ test_tinantas:
     --hash "409c5efa5dbcebba7375732a64867f7212ee43d3b112c8c8c494099158e68a0a"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-san-kartari.csv \
-		--hash "da47a054b8f909c9496f042975b0311d77b106208eb1402d2bb81c7a35da2305"
+		--hash "ec5acedd0e7a5ca478c9ede25ce69daa1f1acaeb7d161c1565e7c505cd843c55"
 	../target/release/test_tinantas \
 		--test-cases test-files/tinantas-yan-kartari.csv \
 		--hash "71cf9817679767971c06bcb806b61f91d29402f27f158837425e02a67a5c0a01"
@@ -80,6 +80,9 @@ test_subantas:
 	cargo run --bin test_subantas -- \
 		--test-cases test-files/subantas.csv
 
+check_rule_coverage:
+	./scripts/check_rule_coverage.py > report.log 
+
 
 # Performance
 # ~~~~~~~~~~~

vidyut-prakriya/scripts/check_rule_coverage.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env python3
+from pathlib import Path
+import re
+import glob
+
+RULE_OK = "✅"
+RULE_UNTESTED = "⚠️ "
+RULE_MISSING = "❌"
+
+def print_legend():
+    print("===== Legend ======")
+    print(f"{RULE_OK}\t\tSutra is tested.")
+    print(f"{RULE_UNTESTED}\t\tSutra is implemented but untested.")
+    print(f"{RULE_MISSING}\t\tSutra is missing.")
+    print()
+    print("These statuses are heuristics. Verify them by checking the underlying code.")
+    print("===================")
+
+base = Path(__file__).parent.parent
+src = base / "src"
+tests = base / "tests"
+
+all_rules = []
+with open(base / "data/sutrapatha.tsv") as f:
+    for line in f:
+        code, text = line.split('\t')
+        all_rules.append(code)
+
+
+implemented_rules = set()
+for path in glob.glob("**/*.rs", root_dir=src, recursive=True):
+    with open(src / path) as f:
+        for line in f:
+            if m := re.search(r'"(\d+\.\d+\.\d+)', line):
+                implemented_rules.add(m.group(1))
+
+tested_rules = set()
+for path in glob.glob("**/*.rs", root_dir=tests, recursive=True):
+    with open(tests / path) as f:
+        for line in f:
+            if m := re.search(r"(\d+_\d+_\d+)", line):
+                tested_rules.add(m.group(1).replace('_', '.'))
+
+print_legend()
+for rule in all_rules:
+    status = None
+    if rule in tested_rules:
+        status = RULE_OK
+    elif rule in implemented_rules:
+        status = RULE_UNTESTED
+    else:
+        status = RULE_MISSING
+    print(f"{status}\t\t{rule}")
+print_legend()

vidyut-prakriya/src/angasya.rs

@@ -38,6 +38,7 @@ use lazy_static::lazy_static;
 
 lazy_static! {
     static ref AC: Set = s("ac");
+    static ref UK: Set = s("uk");
     static ref DANTYA: Set = s("tu~ v");
     static ref OSHTHYA: Set = s("pu~ v");
     static ref II: Set = s("i");
@@ -312,6 +313,7 @@ pub fn try_pratyaya_adesha(p: &mut Prakriya) -> Option<()> {
 
     let i = len - 1;
     let last = p.get_if(i, |t| t.is_pratyaya())?;
+    let prev = p.get(i - 1)?;
 
     if last.has_text_in(&["yu~", "vu~"]) {
         if last.has_text("yu~") {
@@ -322,13 +324,16 @@ pub fn try_pratyaya_adesha(p: &mut Prakriya) -> Option<()> {
     } else if let Some(sub) = replace_pha_dha_and_others(last) {
         p.op_term("7.1.2", i, op::adi(sub));
     } else if last.has_adi('W') {
-        // Run 7.3.50 because it has no clear place otherwise.
-        p.op_term("7.3.50", i, |t| t.set_adi("ik"));
+        // Run 7.3.50 and 7.3.51 because they have no clear place otherwise.
+        if prev.has_suffix_in(&["is", "us", "t"]) || prev.has_antya(&*UK) {
+            p.op_term("7.3.51", i, |t| t.set_adi("k"));
+        } else {
+            p.op_term("7.3.50", i, |t| t.set_adi("ik"));
+        }
 
     // 7.1.34 (daDyA -> daDyO) happens later on after the dhatu's vowel change (DyE -> DyA)
-
-    // -tAt substitution needs to occur early because it conditions samprasarana.
     } else if p.has(i, |t| t.has_tag(T::Tin) && t.has_text_in(&["tu", "hi"])) {
+        // -tAt substitution needs to occur early because it conditions samprasarana.
         // N is to block pit-guNa, not for replacement of the last letter.
         op::optional_adesha("7.1.35", p, i, "tAta~N");
     }
@@ -671,35 +676,36 @@ fn try_sarvadhatuke(p: &mut Prakriya) -> Option<()> {
 }
 
 /// (7.4.21 - 7.4.31)
-fn try_change_dhatu_before_y(p: &mut Prakriya) -> Option<()> {
-    let i = p.find_first(T::Dhatu)?;
+fn try_change_anga_before_y(p: &mut Prakriya) -> Option<()> {
+    p.debug("try_change");
+    let i = p.find_first_where(|t| t.is_dhatu() || t.is_pratipadika())?;
     let i_n = p.find_next_where(i, |t| !t.is_empty())?;
-    let dhatu = p.get(i)?;
+    let anga = p.get(i)?;
     let n = p.view(i_n)?;
 
     let mut temp = CompactString::new("");
-    temp.replace_range(.., &dhatu.text);
+    temp.replace_range(.., &anga.text);
 
     let akrt_sarva = !n.has_tag_in(&[T::Sarvadhatuka, T::Krt]);
     let has_upasarga = i > 0 && p.has(i - 1, |t| t.has_tag(T::Upasarga));
     let yi_kniti = n.has_adi('y') && n.is_knit();
 
-    if dhatu.has_u("SIN") && n.has_tag(T::Sarvadhatuka) {
+    if anga.has_u("SIN") && n.has_tag(T::Sarvadhatuka) {
         p.op_term("7.4.21", i, op::text("Se"));
-    } else if dhatu.has_u("SIN") && yi_kniti {
+    } else if anga.has_u("SIN") && yi_kniti {
         p.op_term("7.4.22", i, op::text("Say"));
         p.set(i, |t| t.force_save_sthanivat());
-    } else if has_upasarga && yi_kniti && dhatu.has_u("Uha~\\") {
+    } else if has_upasarga && yi_kniti && anga.has_u("Uha~\\") {
         // Example: sam[u]hyate
         p.op_term("7.4.23", i, op::adi("u"));
     } else if has_upasarga
         && yi_kniti
-        && dhatu.has_u("i\\R")
+        && anga.has_u("i\\R")
         && p.terms().last()?.has_lakshana("li~N")
     {
         // Example: ud[i]yAt
         p.op_term("7.4.24", i, op::adi("i"));
-    } else if dhatu.has_antya('f') {
+    } else if anga.has_antya('f') {
         let dhatu = p.get(i)?;
         let n = p.view(i_n)?;
         let is_sha_or_yak = n.has_u_in(&["Sa", "yak"]);
@@ -726,10 +732,10 @@ fn try_change_dhatu_before_y(p: &mut Prakriya) -> Option<()> {
             // mantrIyati
             p.op_term("7.4.27", i, op::antya("rI"));
         }
-    } else if dhatu.has_u_in(&["GrA\\", "DmA\\"]) && n.has_u("yaN") {
+    } else if anga.has_u_in(&["GrA\\", "DmA\\"]) && n.has_u("yaN") {
         p.op_term("7.4.31", i, op::antya("I"));
     } else if n.has_adi('y') {
-        let sub = al::to_dirgha(dhatu.antya()?)?;
+        let sub = al::to_dirgha(anga.antya()?)?;
         if n.has_u("cvi") {
             p.op_term("7.4.26", i, op::antya(&sub.to_string()));
         } else if akrt_sarva && n.is_knit() {
@@ -1589,7 +1595,7 @@ pub fn run_before_dvitva(p: &mut Prakriya) -> Option<()> {
     p.debug("==== Guna-vrddhi ====");
     guna_vrddhi::run(p);
 
-    try_change_dhatu_before_y(p);
+    try_change_anga_before_y(p);
     try_cchvoh(p);
 
     // Must precede ft-AdeSa (f -> ir)

vidyut-prakriya/src/angasya/abhyasasya.rs

@@ -117,7 +117,7 @@ fn try_abhyasa_lopa_and_dhatu_change_before_san(p: &mut Prakriya) -> Option<()>
     Some(())
 }
 
-/// `i` is the index of an abhyasa..
+/// `i` is the index of an abhyasa.
 fn run_for_sani_or_cani_at_index(p: &mut Prakriya, i: usize) -> Option<()> {
     const SMR_DR: &[&str] = &[
         "smf",
@@ -161,7 +161,7 @@ fn run_for_sani_or_cani_at_index(p: &mut Prakriya, i: usize) -> Option<()> {
                 p.op_term("7.4.79", i, op::antya("i"));
             } else if abhyasa.has_antya(&*UU) && anga.has_adi(&*PU_YAN_J) && anga.has_at(1, &*AA) {
                 p.op_term("7.4.80", i, op::antya("i"));
-            } else if anga.has_u_in(SRU_ADI) && anga.has_upadha('a') {
+            } else if anga.has_u_in(SRU_ADI) && p.has(i + 2, |t| !t.has_u("san")) {
                 // Example: sru -> sisrAvayizyati
                 // Note that this rule must run after guna for the upadha check to be meaningful.
                 p.op_optional("7.4.81", op::t(i, op::antya("i")));
@@ -260,7 +260,7 @@ fn try_general_rules(p: &mut Prakriya, i: usize) -> Option<()> {
     let abhyasa = p.get(i)?;
     if al::is_dirgha(abhyasa.antya()?) {
         let val = al::to_hrasva(abhyasa.antya()?)?;
-        p.op_term("7.4.60", i, op::antya(&val.to_string()));
+        p.op_term("7.4.59", i, op::antya(&val.to_string()));
     }
 
     if p.has(i, |t| t.has_antya('f')) {

vidyut-prakriya/src/angasya/asiddhavat.rs

@@ -979,6 +979,8 @@ pub fn bhasya(p: &mut Prakriya) -> Option<()> {
                 }
             } else if ani && bha.ends_with("an") {
                 p.step("6.4.167");
+            } else if bha.has_text_in(&["Atman", "aDvan"]) && next.has_u("Ka") {
+                p.step("6.4.169");
             } else if bha.has_text("ahan") {
                 if next.has_u_in(&["wac", "KA"]) {
                     p.op_term("6.4.145", i, op::ti(""));
@@ -1005,6 +1007,43 @@ pub fn bhasya(p: &mut Prakriya) -> Option<()> {
         }
     } else if bha.has_antya('A') && bha.is_dhatu() {
         p.op_term("6.4.140", i, op::antya(""));
+    } else if (bha.has_u("daRqin") && next.has_u("Pak"))
+        || (bha.has_u("hastin") && next.has_u("Pak"))
+        || (bha.has_u("aTarvan") && next.has_u("Wak"))
+        || (bha.has_u("jihmASin") && next.has_u("Qak"))
+        || (bha.has_u("vAsin") && next.has_u("PiY"))
+        || (bha.has_u("BrURahan") && next.has_u("zyaY"))
+        || (bha.has_u("DIvat") && next.has_u("zyaY"))
+        || (bha.has_u("sarayU") && next.has_u("aR"))
+        || (bha.has_u("ikzvAku") && next.has_u_in(&["aY", "aR"]))
+        || (bha.has_u("mitrayu") && next.has_u("QaY"))
+        || (bha.has_u("hiraRya") && next.has_u("mayaw"))
+    {
+        let code = "6.4.174";
+        let sub = if bha.has_u("daRqin") {
+            "daRqinAyana"
+        } else if bha.has_u("hastin") {
+            "hAstinAyana"
+        } else if bha.has_u("aTarvan") {
+            "aTarvaRika"
+        } else if bha.has_u("jihmASin") {
+            "jEhmASineya"
+        } else if bha.has_u("vAsin") {
+            "vAsinAyani"
+        } else if bha.has_u("BrURahan") {
+            "BrORahatya"
+        } else if bha.has_u("DIvat") {
+            "DEvatya"
+        } else if bha.has_u("sarayU") {
+            "sArava"
+        } else if bha.has_u("ikzvAku") {
+            "EkzvAka"
+        } else if bha.has_u("mitrayu") {
+            "mEtreya"
+        } else {
+            "hiraRmaya"
+        };
+        p.op(code, op::nipatana(sub));
     } else if bha.has_antya(&*UU) && taddhita {
         if next.has_tag(T::Qit) && !bha.has_text("kadrU") {
             p.op_term("6.4.147", i, |t| t.set_antya(""));

vidyut-prakriya/src/angasya/guna_vrddhi.rs

@@ -60,43 +60,43 @@ fn try_taddhita_vrddhi(p: &mut Prakriya, i: usize) -> Option<()> {
     let anga = p.get(i)?;
     let n = p.get_if(i + 1, |t| t.is_taddhita())?;
 
-    if anga.has_text_in(&["devikA", "SiMSapA", "dityavAh", "dIrGasatra", "Sreyas"]) {
-        // dAvikA, ...
-        let adi_ac = anga.text.find(al::is_ac)?;
-        p.op_term("7.3.1", i, |t| {
-            t.set_at(adi_ac, "A");
-        });
-
-        return Some(());
-    }
-
     let rule = if n.has_tag_in(&[T::Yit, T::Rit]) {
-        Some("7.2.117")
+        "7.2.117"
     } else if n.has_tag(T::kit) {
-        Some("7.2.118")
+        "7.2.118"
     } else {
-        None
+        return None;
     };
 
-    if let Some(rule) = rule {
-        if anga.has_u_in(DVARA_ADI) {
-            // dvAra -> dOvArika, ...
-            p.op_term("7.3.4", i, |t| {
-                let i_yan = t.text.rfind(|c| c == 'y' || c == 'v').expect("ok");
-                if t.text.get(i_yan..i_yan + 1) == Some("y") {
-                    t.text.insert(i_yan, 'E');
-                } else {
-                    t.text.insert(i_yan, 'O');
-                }
-            });
-        } else {
-            let adi_ac = anga.text.find(al::is_ac)?;
-            let ac = anga.get_at(adi_ac)?;
-            let vrddhi = al::to_vrddhi(ac)?;
-            p.op_term(rule, i, |t| {
-                t.set_at(adi_ac, vrddhi);
-            });
-        }
+    if anga.has_text_in(&["devikA", "SiMSapA", "dityavAh", "dIrGasatra", "Sreyas"]) {
+        // dAvikA, ...
+        let adi_ac = anga.text.find(al::is_ac)?;
+        p.op_term("7.3.1", i, |t| t.set_at(adi_ac, "A"));
+    } else if anga.has_text_in(&["kekaya", "mitrayu", "pralaya"]) {
+        p.op_term("7.3.2", i, |t| t.find_and_replace_text("y", "iy"));
+    } else if anga.text.starts_with("vy") {
+        // HACK: should properly be only with vi-upasarga.
+        // TODO: also apply for sv-, .etc.
+        p.op_term("7.3.3", i, |t| t.text.replace_range(..2, "vEy"));
+    } else if anga.has_u_in(DVARA_ADI) {
+        // dvAra -> dOvArika, ...
+        p.op_term("7.3.4", i, |t| {
+            let i_yan = t.text.rfind(|c| c == 'y' || c == 'v').expect("ok");
+            if t.text.get(i_yan..i_yan + 1) == Some("y") {
+                t.text.insert(i_yan, 'E');
+            } else {
+                t.text.insert(i_yan, 'O');
+            }
+        });
+    } else if anga.has_text("nyagroDa") {
+        p.op_term("7.3.5", i, |t| t.text.replace_range(..2, "nEy"));
+    } else {
+        let adi_ac = anga.text.find(al::is_ac)?;
+        let ac = anga.get_at(adi_ac)?;
+        let vrddhi = al::to_vrddhi(ac)?;
+        p.op_term(rule, i, |t| {
+            t.set_at(adi_ac, vrddhi);
+        });
     }
 
     Some(())