Add dictionary parsing

.gitignore

@@ -103,3 +103,4 @@ dist
 # TernJS port file
 .tern-port
 /reports
+.DS_Store

src/index.ts

@@ -1,5 +1,15 @@
-export const example = (): string => 'test string';
+import path from 'path';
+import { DictionaryEntry } from './models';
+import { read } from './reader';
+import { parse } from './parser';
 
-export default {
-  example,
-};
+const COMPRESSED_DICTIONARY_PATH = path.join(`${__dirname}/../resources/dictionary.json.gz`);
+
+export function getDictionary() : DictionaryEntry[] {
+  const content = read(COMPRESSED_DICTIONARY_PATH);
+  const entries = parse(content);
+
+  return entries;
+}
+
+export { DictionaryEntry } from './models';

src/models.ts

@@ -0,0 +1,4 @@
+export interface DictionaryEntry {
+    headword: string,
+    definitions: string[],
+}

src/parser.ts

@@ -0,0 +1,14 @@
+import zlib from 'zlib';
+import { DictionaryEntry } from './models';
+
+export const parse = (content: Buffer): DictionaryEntry[] => {
+  // Decompress the gzipped data
+  const decompressedData = zlib.gunzipSync(content);
+
+  // Parse the JSON content
+  return JSON.parse(decompressedData.toString());
+};
+
+export default {
+  parse,
+};

src/reader.ts

@@ -0,0 +1,7 @@
+import fs from 'fs';
+
+export const read = (location: string) : Buffer => fs.readFileSync(location);
+
+export default {
+  read,
+};

tests/old-danish-dictionary.test.ts

@@ -0,0 +1,38 @@
+import { getDictionary } from '../src';
+
+describe('Old Danish Dictionary tests', () => {
+  test('Dictionary contains expected amount of entries', () => {
+    const result = getDictionary();
+
+    expect(result.length).toBe(45408);
+  });
+
+  test('Dictionary entries are parsed in correct object format', () => {
+    const result = getDictionary();
+
+    result.forEach((entry) => {
+      // Only expected keys.
+      expect(Object.keys(entry)).toEqual(['headword', 'definitions']);
+
+      // TODO: should assert headwords are not empty.
+      // Current dataset contains 8 empties, so cant be done yet.
+    });
+  });
+
+  test('Dictionary contains expected content', () => {
+    const result = getDictionary();
+
+    const skrotte = result.filter((entry) => entry.headword === 'Skrotte')[0];
+    const ulovagtig = result.filter((entry) => entry.headword === 'Ulovagtig')[0];
+    const vandhagl = result.filter((entry) => entry.headword === 'Vandhagl')[0];
+
+    expect(skrotte.headword).toBe('Skrotte');
+    expect(skrotte.definitions[0]).toBe('go. tvivle. M. Hele ordr\u00e6kken synes at hvile p\u00e5 en misforst\u00e5else eller forvanskning.');
+
+    expect(ulovagtig.headword).toBe('Ulovagtig');
+    expect(ulovagtig.definitions[0]).toBe('no. ulovlig,ikke gyldig efter loven; (1597) KhS VIL 495 (ovf. III 235b8); teste depr\u00e6hensi, drefine, wlouactige vidre TV 400. Jf logagtig (ovf. II 78122).');
+
+    expect(vandhagl.headword).toBe('Vandhagl');
+    expect(vandhagl.definitions[0]).toBe('no. et slags hagl; lade skyde med vandhagl 12 urh\u00f8ns (1642). AL. Efter M: mindste h.');
+  });
+});