chore: converting dictionaries

convert

@@ -0,0 +1,195 @@
+#!/usr/bin/env -S deno --allow-read
+
+import { parse, stringify } from "jsr:@std/yaml";
+import miwafu from "npm:fedran-miwafu@0.1.6";
+
+console.log(miwafu);
+
+// Go through all the directories and parse each one.
+let done = false;
+
+for await (const rootEntry of Deno.readDir("src/dictionary")) {
+  const rootPath = "src/dictionary/" + rootEntry.name + "/";
+
+  console.log("-", rootEntry.name);
+
+  for await (const dirEntry of Deno.readDir(rootPath)) {
+    // Load the file and see if we've already converted it.
+    const filePath = rootPath + dirEntry.name;
+
+    console.log("  - path:", filePath);
+
+    const text = await Deno.readTextFile(filePath);
+    const data = parse(text);
+
+    data.version ??= 0;
+
+    console.log("    version:", data.version);
+
+    if (data.version !== 0) {
+      continue;
+    }
+
+    console.log("    converting");
+    console.log(JSON.stringify(data, null, "  "));
+
+    // Convert the object.
+    const newLanguage: NewLanguage = {
+      parts: {},
+    };
+
+    const newEntry: NewEntry = {
+      entry: data.base,
+      languages: { qmw: newLanguage },
+    };
+
+    var newData: NewFile = {
+      version: 1,
+      entries: [newEntry],
+    };
+
+    addGender(newLanguage, data.pos.noun, "noun", data.base);
+    addGender(newLanguage, data.pos.verb, "verb", data.base);
+    addList(newLanguage, data.pos.adv, "adv", data.base);
+    addList(newLanguage, data.pos.adj, "adj", data.base);
+
+    // Write out the results.
+    Deno.writeTextFile(filePath, stringify(newData));
+
+    console.log("    ---");
+    console.log(stringify(newData));
+
+    // We only want to convert one at a time.
+    done = true;
+    break;
+  }
+
+  if (done) break;
+}
+
+function addGender(
+  lang: NewLanguage,
+  gender: OldGender | undefined,
+  part: String,
+  word: String,
+) {
+  if (gender) {
+    addList(lang, gender.masculine, part, miwafu.inflectMasculine(word));
+    addList(lang, gender.feminine, part, miwafu.inflectFeminine(word));
+    addList(lang, gender.neuter, part, miwafu.inflectNeuter(word));
+  }
+}
+
+function addPartList(
+  lang: NewLanguage,
+  list: OldDefinition[] | undefined,
+  part: String,
+  word: String,
+) {
+  if (list) {
+    lang.parts[part] ??= [];
+
+    for (const def of list) {
+      add(lang.parts[part], def, part, word);
+    }
+  }
+}
+
+function addList(
+  lang: NewLanguage,
+  list: OldDefinition[] | undefined,
+  part: String,
+  word: String,
+) {
+  if (list) {
+    lang.parts[part] ??= [];
+
+    for (const def of list) {
+      add(lang.parts[part], def, part, word);
+    }
+  }
+}
+
+function add(dest: NewPart[], def: OldDefinition, part: String, word: String) {
+  console.log("**** def", part, word, def);
+
+  const newPart: NewPart = {
+    word: word,
+    definitions: [{ definition: def.def }],
+  };
+
+  dest.push(newPart);
+}
+
+// Define the old data format.
+interface OldEntry {
+  version?: Number;
+  name: String;
+  pos: [OldPartOfSpeech];
+}
+
+interface OldPartOfSpeech {
+  noun?: OldGender;
+  verb?: OldGender;
+  adv?: OldDefinition[];
+  adj?: OldDefinition[];
+}
+
+interface OldGender {
+  masculine?: OldDefinition[];
+  feminine?: OldDefinition[];
+  neuter?: OldDefinition[];
+}
+
+interface OldDefinition {
+  def?: string;
+  example?: OldExample;
+  oow?: Boolean;
+  reference?: OldReference[];
+}
+
+interface OldExample {
+  miw: String;
+  en: String;
+}
+
+interface OldReference {
+  identifier?: String;
+  title?: String;
+  url?: String;
+  excerpt?: String;
+}
+
+// Define the new data format.
+interface NewFile {
+  version: 1;
+  entries: NewEntry[];
+}
+
+interface NewEntry {
+  entry: String;
+  languages: { [language: String]: NewLanguage };
+}
+
+interface NewLanguage {
+  parts: { [pos: String]: NewPart[] };
+}
+
+interface NewPart {
+  word: String;
+}
+
+/*
+pos:
+  noun:
+    masculine:
+      - def: A long or a full night's sleep.
+    feminine:
+      - def: A nap.
+  verb:
+    masculine:
+      - def: To sleep for a long time.
+    feminine:
+      - def: To take a nap.
+base: zushi
+*/

deno.lock

@@ -0,0 +1,27 @@
+{
+  "version": "4",
+  "specifiers": {
+    "jsr:@std/yaml@*": "1.0.5"
+  },
+  "jsr": {
+    "@std/yaml@1.0.5": {
+      "integrity": "71ba3d334305ee2149391931508b2c293a8490f94a337eef3a09cade1a2a2742"
+    }
+  },
+  "workspace": {
+    "packageJson": {
+      "dependencies": [
+        "npm:@commitlint/cli@^7.2.1",
+        "npm:@commitlint/config-conventional@^7.1.2",
+        "npm:commitizen@^3.0.5",
+        "npm:cz-conventional-changelog@^2.1.0",
+        "npm:fedran-miwafu@~0.1.6",
+        "npm:fs-extra@^7.0.1",
+        "npm:husky@^1.2.1",
+        "npm:js-yaml@^3.13.1",
+        "npm:json-stable-stringify@^1.0.1",
+        "npm:pajv@^1.2.0"
+      ]
+    }
+  }
+}

flake.nix

@@ -25,6 +25,7 @@
             packages = [
               pkgs.nodejs_20
               pkgs.python3
+              pkgs.deno
             ]
             ++ config.packages;
 

package.json

@@ -34,5 +34,7 @@
       "commit-msg": "commitlint -E HUSKY_GIT_PARAMS"
     }
   },
-  "dependencies": {}
+  "dependencies": {
+    "fedran-miwafu": "^0.1.6"
+  }
 }

src/dictionary/ka/kadu.yaml

@@ -1,5 +1,10 @@
-pos:
+version: 1
+entries:
+  - entry: kadu
+    languages:
+      qmw:
+        parts:
           noun:
-    feminine:
-      - def: Hand.
-base: kadu
+            - word: kàdu
+              definitions:
+                - definition: Hand.

src/new-dictionary/_meta.yaml

@@ -1,5 +1,5 @@
 ---
-version: 0
+version: 1
 
 languages:
   qmw:

src/new-dictionary/asanogi.yaml

@@ -1,5 +1,5 @@
 ---
-version: 0
+version: 1
 
 entries:
   - entry: asanogi
@@ -9,11 +9,10 @@ entries:
         ipa: "/a.sa.noː.ɡi/"
 
     languages:
-      - lang: qmw
-
+      qmw:
         parts:
-          - part: noun
-            word: asanōgi
+          noun:
+            - word: asanōgi
 
               definitions:
                 - definition: A drink made from infusing roasted tea leaves in hot water for a long period of time.