finishing parsing noun words

lingdocs · Aug 23, 2024 · 1abc83e · 1abc83e
1 parent 32f7618
commit 1abc83e
Show file tree

Hide file tree

Showing 4 changed files with 132 additions and 9 deletions.
diff --git a/src/lib/src/parsing/parse-fem-noun.ts b/src/lib/src/parsing/parse-fem-noun.ts
@@ -67,7 +67,7 @@ function parsePattern1(
     dictionary
       .queryP(p)
       .filter(andSuccTp(tp.isFemNounEntry, tp.isPattern1Entry));
-  const plain = first.s.endsWith("ه")
+  const plain = ["ه", "ع"].some((v) => first.s.endsWith(v))
     ? p1Lookup(first.s).map<FemNounBaseParse>((entry) => ({
         inflection: [0],
         gender: ["fem"],
@@ -84,12 +84,21 @@ function parsePattern1(
       }))
     : [];
   const inflected = first.s.endsWith("ې")
-    ? p1Lookup(first.s.slice(0, -1) + "ه").map<FemNounBaseParse>((entry) => ({
-        inflection: [1],
-        gender: ["fem"],
-        entry,
-        given: first.s,
-      }))
+    ? (() => {
+        const base = first.s.slice(0, -1);
+        const lookups = [
+          ...p1Lookup(base + "ه"),
+          ...(["ح", "ع"].some((v) => first.s.at(-2) === v)
+            ? p1Lookup(base)
+            : []),
+        ];
+        return lookups.map<FemNounBaseParse>((entry) => ({
+          inflection: [1],
+          gender: ["fem"],
+          entry,
+          given: first.s,
+        }));
+      })()
     : [];
   const doubleInflected = first.s.endsWith("و")
     ? p1Lookup(first.s.slice(0, -1) + "ه").map<FemNounBaseParse>((entry) => ({

diff --git a/src/lib/src/parsing/parse-irregular-plural.ts b/src/lib/src/parsing/parse-irregular-plural.ts
@@ -1,5 +1,6 @@
 import * as T from "../../../types";
 import { DictionaryAPI } from "../dictionary/dictionary";
+import { endsInConsonant } from "../p-text-helpers";
 import * as tp from "../type-predicates";
 import { returnParseResults } from "./utils";
 
@@ -16,7 +17,10 @@ export function parseIrregularPlural(
     .filter(tp.isNounEntry)
     .map<T.ParsedNounWord<T.NounEntry>>((entry) => ({
       entry,
-      gender: tp.isFemNounEntry(entry) ? "fem" : "masc",
+      gender:
+        tp.isFemNounEntry(entry) && !hasMasculineArabicPlural(entry)
+          ? "fem"
+          : "masc",
       inflected: false,
       given: first.s,
       plural: true,
@@ -75,3 +79,8 @@ export function parseIrregularPlural(
     ...inflectedAfterLongSep,
   ];
 }
+
+function hasMasculineArabicPlural(e: T.FemNounEntry): boolean {
+  if (!e.app || !e.apf) return false;
+  return endsInConsonant({ p: e.app, f: e.apf });
+}
diff --git a/src/lib/src/parsing/parse-noun-new.test.ts b/src/lib/src/parsing/parse-noun-new.test.ts
@@ -44,10 +44,12 @@ const zooy = testDictionary.nounLookup("زوی")[0];
 const loor = testDictionary.nounLookup("لور")[0];
 const nabee = testDictionary.nounLookup("نبي")[0];
 const lafz = testDictionary.nounLookup("لفظ")[0];
+const fatha = testDictionary.nounLookup("فتح")[0];
+const nafa = testDictionary.nounLookup("نفع")[0];
+const tajraba = testDictionary.nounLookup("تجربه")[0];
 
 // TODO: test for adjective errors etc
 // bundled plural
-// TODO: منابع
 
 const tests: {
   category: string;
@@ -216,6 +218,93 @@ const tests: {
       },
     ],
   },
+  {
+    category: "pattern 1 fem nouns with ح ه etc",
+    cases: [
+      {
+        input: "فتح",
+        output: [
+          {
+            inflected: false,
+            selection: {
+              ...makeNounSelection(fatha, undefined),
+            },
+          },
+        ],
+      },
+      {
+        input: "فتحې",
+        output: [
+          {
+            inflected: true,
+            selection: {
+              ...makeNounSelection(fatha, undefined),
+            },
+          },
+          {
+            inflected: false,
+            selection: {
+              ...makeNounSelection(fatha, undefined),
+              number: "plural",
+            },
+          },
+        ],
+      },
+      {
+        input: "فتحو",
+        output: [
+          {
+            inflected: true,
+            selection: {
+              ...makeNounSelection(fatha, undefined),
+              number: "plural",
+            },
+          },
+        ],
+      },
+      {
+        input: "نفع",
+        output: [
+          {
+            inflected: false,
+            selection: {
+              ...makeNounSelection(nafa, undefined),
+            },
+          },
+        ],
+      },
+      {
+        input: "نفعې",
+        output: [
+          {
+            inflected: true,
+            selection: {
+              ...makeNounSelection(nafa, undefined),
+            },
+          },
+          {
+            inflected: false,
+            selection: {
+              ...makeNounSelection(nafa, undefined),
+              number: "plural",
+            },
+          },
+        ],
+      },
+      {
+        input: "نفعو",
+        output: [
+          {
+            inflected: true,
+            selection: {
+              ...makeNounSelection(nafa, undefined),
+              number: "plural",
+            },
+          },
+        ],
+      },
+    ],
+  },
   {
     category: "group/plural nouns",
     // to do fem plural تسبیج, هټکرۍ, مایع
@@ -1625,6 +1714,19 @@ const tests: {
           },
         ],
       },
+      {
+        input: "تجارب",
+        output: [
+          {
+            inflected: false,
+            selection: {
+              ...makeNounSelection(tajraba, undefined),
+              number: "plural",
+              gender: "masc",
+            },
+          },
+        ],
+      },
     ],
   },
 ];

diff --git a/vocab/mini-dict-tss.ts b/vocab/mini-dict-tss.ts
@@ -47,4 +47,7 @@ export const entries: T.DictionaryEntry["ts"][] = [
   1527812582, // دعا
   1591803598624, // هتکړۍ
   1527812861, // لور - loor
+  1589023873660, // فتح - fatha
+  1527814342, // نفع - nafa
+  1527815329, // تجربه
 ];