WIP

wydengyre · Mar 2, 2024 · 1a23884 · 1a23884
1 parent 2f751dd
commit 1a23884
Show file tree

Hide file tree

Showing 42 changed files with 1,232,067 additions and 9 deletions.
diff --git a/package-lock.json b/package-lock.json
diff --git a/packages/cli/main.ts b/packages/cli/main.ts
@@ -109,13 +109,15 @@ async function goWithLanguagesAndText(
  const hunalignWasm = await readFile(HUNALIGN_WASM_PATH);
  const hunalignLib = await HunalignLib.Hunalign.create(hunalignWasm);
 
+ const dictName = `${targetLangAbbr}-${sourceLangAbbr}.dic`;
  const hunalignDictData = await readFile(
  fileURLToPath(
- import.meta.resolve(
- `@bitextual/hunalign/dictionaries/${targetLangAbbr}-${sourceLangAbbr}.dic`,
- ),
+ import.meta.resolve(`@bitextual/wiktionary/dictionaries/${dictName}`),
  ),
  );
+ console.error(
+ `using dictionary: ${dictName} with size ${hunalignDictData.length} bytes`,
+ );
 
  const sourceLang = languageCodes.get(sourceLangAbbr);
  if (sourceLang === undefined) {

diff --git a/packages/cli/package.json b/packages/cli/package.json
@@ -20,6 +20,7 @@
  },
  "dependencies": {
  "@bitextual/core": "*",
- "@bitextual/hunalign": "*"
+ "@bitextual/hunalign": "*",
+ "@bitextual/wiktionary": "*"
  }
 }
diff --git a/packages/hunalign/dictionaries/en-fr.dic b/packages/hunalign/dictionaries/en-fr.dic
@@ -1,4 +1,3 @@
-
 -- @ --
 --- @ ---
 ---- @ ----

diff --git a/packages/test/bovary.aligned.html b/packages/test/bovary.aligned.html
diff --git a/packages/test/marianela.aligned.html b/packages/test/marianela.aligned.html
diff --git a/packages/wiktionary/.gitignore b/packages/wiktionary/.gitignore
@@ -0,0 +1 @@
+raw-wiktextract-data.json
diff --git a/packages/wiktionary/README.md b/packages/wiktionary/README.md
@@ -0,0 +1 @@
+Data from https://kaikki.org/dictionary/raw-wiktextract-data.json
diff --git a/packages/wiktionary/dictionaries/README.md b/packages/wiktionary/dictionaries/README.md
@@ -0,0 +1 @@
+Dictionary files generated from Wiktionary data go in this directory