summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authortoasted-nutbread <toasted-nutbread@users.noreply.github.com>2023-11-27 16:09:55 -0500
committertoasted-nutbread <toasted-nutbread@users.noreply.github.com>2023-11-27 16:09:55 -0500
commitd897c3b069f80fd170b7f1d0c0362fec4037d714 (patch)
tree76448c197b510c394463d3ba6b0eb31b01bcd7f0
parentac562ca36417f2d9fdb860d1f8a879fdccde438d (diff)
Update dictionary importer
-rw-r--r--ext/js/language/dictionary-importer.js130
-rw-r--r--jsconfig.json5
-rw-r--r--package-lock.json13
-rw-r--r--package.json1
-rw-r--r--types/ext/dictionary-importer.d.ts27
5 files changed, 123 insertions, 53 deletions
diff --git a/ext/js/language/dictionary-importer.js b/ext/js/language/dictionary-importer.js
index 2a2f4063..115e0726 100644
--- a/ext/js/language/dictionary-importer.js
+++ b/ext/js/language/dictionary-importer.js
@@ -16,10 +16,23 @@
* along with this program. If not, see <https://www.gnu.org/licenses/>.
*/
-import * as ajvSchemas from '../../lib/validate-schemas.js';
-import {BlobWriter, TextWriter, Uint8ArrayReader, ZipReader, configure} from '../../lib/zip.js';
+import * as ajvSchemas0 from '../../lib/validate-schemas.js';
+import {
+ BlobWriter as BlobWriter0,
+ TextWriter as TextWriter0,
+ Uint8ArrayReader as Uint8ArrayReader0,
+ ZipReader as ZipReader0,
+ configure
+} from '../../lib/zip.js';
import {stringReverse} from '../core.js';
import {MediaUtil} from '../media/media-util.js';
+import {ExtensionError} from '../core/extension-error.js';
+
+const ajvSchemas = /** @type {import('dictionary-importer').CompiledSchemaValidators} */ (/** @type {unknown} */ (ajvSchemas0));
+const BlobWriter = /** @type {typeof import('@zip.js/zip.js').BlobWriter} */ (/** @type {unknown} */ (BlobWriter0));
+const TextWriter = /** @type {typeof import('@zip.js/zip.js').TextWriter} */ (/** @type {unknown} */ (TextWriter0));
+const Uint8ArrayReader = /** @type {typeof import('@zip.js/zip.js').Uint8ArrayReader} */ (/** @type {unknown} */ (Uint8ArrayReader0));
+const ZipReader = /** @type {typeof import('@zip.js/zip.js').ZipReader} */ (/** @type {unknown} */ (ZipReader0));
export class DictionaryImporter {
/**
@@ -62,21 +75,21 @@ export class DictionaryImporter {
const zipFileReader = new Uint8ArrayReader(new Uint8Array(archiveContent));
const zipReader = new ZipReader(zipFileReader);
const zipEntries = await zipReader.getEntries();
- const zipEntriesObject = {};
+ /** @type {import('dictionary-importer').ArchiveFileMap} */
+ const fileMap = new Map();
for (const entry of zipEntries) {
- zipEntriesObject[entry.filename] = entry;
+ fileMap.set(entry.filename, entry);
}
// Read and validate index
const indexFileName = 'index.json';
- const indexFile = zipEntriesObject[indexFileName];
- if (!indexFile) {
+ const indexFile = fileMap.get(indexFileName);
+ if (typeof indexFile === 'undefined') {
throw new Error('No dictionary index found in archive');
}
+ const indexFile2 = /** @type {import('@zip.js/zip.js').Entry} */ (indexFile);
- const indexContent = await indexFile.getData(
- new TextWriter()
- );
- const index = JSON.parse(indexContent);
+ const indexContent = await this._getData(indexFile2, new TextWriter());
+ const index = /** @type {import('dictionary-data').Index} */ (JSON.parse(indexContent));
if (!ajvSchemas.dictionaryIndex(index)) {
throw this._formatAjvSchemaError(ajvSchemas.dictionaryIndex, indexFileName);
@@ -99,11 +112,11 @@ export class DictionaryImporter {
const dataBankSchemas = this._getDataBankSchemas(version);
// Files
- const termFiles = this._getArchiveFiles(zipEntriesObject, 'term_bank_?.json');
- const termMetaFiles = this._getArchiveFiles(zipEntriesObject, 'term_meta_bank_?.json');
- const kanjiFiles = this._getArchiveFiles(zipEntriesObject, 'kanji_bank_?.json');
- const kanjiMetaFiles = this._getArchiveFiles(zipEntriesObject, 'kanji_meta_bank_?.json');
- const tagFiles = this._getArchiveFiles(zipEntriesObject, 'tag_bank_?.json');
+ const termFiles = this._getArchiveFiles(fileMap, 'term_bank_?.json');
+ const termMetaFiles = this._getArchiveFiles(fileMap, 'term_meta_bank_?.json');
+ const kanjiFiles = this._getArchiveFiles(fileMap, 'kanji_bank_?.json');
+ const kanjiMetaFiles = this._getArchiveFiles(fileMap, 'kanji_meta_bank_?.json');
+ const tagFiles = this._getArchiveFiles(fileMap, 'tag_bank_?.json');
// Load data
this._progressNextStep(termFiles.length + termMetaFiles.length + kanjiFiles.length + kanjiMetaFiles.length + tagFiles.length);
@@ -153,7 +166,7 @@ export class DictionaryImporter {
// Async requirements
this._progressNextStep(requirements.length);
- const {media} = await this._resolveAsyncRequirements(requirements, zipEntriesObject);
+ const {media} = await this._resolveAsyncRequirements(requirements, fileMap);
// Add dictionary descriptor
this._progressNextStep(termList.length + termMetaList.length + kanjiList.length + kanjiMetaList.length + tagList.length + media.length);
@@ -274,20 +287,20 @@ export class DictionaryImporter {
}
/**
- *
- * @param schema
- * @param fileName
+ * @param {import('ajv').ValidateFunction} schema
+ * @param {string} fileName
+ * @returns {ExtensionError}
*/
_formatAjvSchemaError(schema, fileName) {
- const e2 = new Error(`Dictionary has invalid data in '${fileName}'`);
+ const e2 = new ExtensionError(`Dictionary has invalid data in '${fileName}'`);
e2.data = schema.errors;
return e2;
}
/**
- *
- * @param version
+ * @param {number} version
+ * @returns {import('dictionary-importer').CompiledSchemaNameArray}
*/
_getDataBankSchemas(version) {
const termBank = (
@@ -402,13 +415,15 @@ export class DictionaryImporter {
}
/**
- *
- * @param requirements
- * @param zipEntriesObject
+ * @param {import('dictionary-importer').ImportRequirement[]} requirements
+ * @param {import('dictionary-importer').ArchiveFileMap} fileMap
+ * @returns {Promise<{media: import('dictionary-database').MediaDataArrayBufferContent[]}>}
*/
- async _resolveAsyncRequirements(requirements, zipEntriesObject) {
+ async _resolveAsyncRequirements(requirements, fileMap) {
+ /** @type {Map<string, import('dictionary-database').MediaDataArrayBufferContent>} */
const media = new Map();
- const context = {zipEntriesObject, media};
+ /** @type {import('dictionary-importer').ImportRequirementContext} */
+ const context = {fileMap, media};
for (const requirement of requirements) {
await this._resolveAsyncRequirement(context, requirement);
@@ -537,15 +552,13 @@ export class DictionaryImporter {
}
// Find file in archive
- const file = context.zipEntriesObject[path];
- if (file === null) {
+ const file = context.fileMap.get(path);
+ if (typeof file === 'undefined') {
throw createError('Could not find image');
}
// Load file content
- let content = await (await file.getData(
- new BlobWriter()
- )).arrayBuffer();
+ let content = await (await this._getData(file, new BlobWriter())).arrayBuffer();
const mediaType = MediaUtil.getImageMediaTypeFromFileName(path);
if (mediaType === null) {
@@ -683,46 +696,48 @@ export class DictionaryImporter {
}
/**
- *
- * @param zipEntriesObject
- * @param fileNameFormat
+ * @param {import('dictionary-importer').ArchiveFileMap} fileMap
+ * @param {string} fileNameFormat
+ * @returns {import('@zip.js/zip.js').Entry[]}
*/
- _getArchiveFiles(zipEntriesObject, fileNameFormat) {
+ _getArchiveFiles(fileMap, fileNameFormat) {
const indexPosition = fileNameFormat.indexOf('?');
const prefix = fileNameFormat.substring(0, indexPosition);
const suffix = fileNameFormat.substring(indexPosition + 1);
+ /** @type {import('@zip.js/zip.js').Entry[]} */
const results = [];
- for (const f of Object.keys(zipEntriesObject)) {
- if (f.startsWith(prefix) && f.endsWith(suffix)) {
- results.push(zipEntriesObject[f]);
+ for (const [name, value] of fileMap.entries()) {
+ if (name.startsWith(prefix) && name.endsWith(suffix)) {
+ results.push(value);
}
}
return results;
}
/**
- *
- * @param files
- * @param convertEntry
- * @param schemaName
- * @param dictionaryTitle
+ * @template [TEntry=unknown]
+ * @template [TResult=unknown]
+ * @param {import('@zip.js/zip.js').Entry[]} files
+ * @param {(entry: TEntry, dictionaryTitle: string) => TResult} convertEntry
+ * @param {import('dictionary-importer').CompiledSchemaName} schemaName
+ * @param {string} dictionaryTitle
+ * @returns {Promise<TResult[]>}
*/
async _readFileSequence(files, convertEntry, schemaName, dictionaryTitle) {
const progressData = this._progressData;
let startIndex = 0;
const results = [];
- for (const fileName of Object.keys(files)) {
- const content = await files[fileName].getData(
- new TextWriter()
- );
- const entries = JSON.parse(content);
+ for (const file of files) {
+ const content = await this._getData(file, new TextWriter());
+ const entries = /** @type {unknown} */ (JSON.parse(content));
startIndex = progressData.index;
this._progress();
- if (!ajvSchemas[schemaName](entries)) {
- throw this._formatAjvSchemaError(ajvSchemas[schemaName], fileName);
+ const schema = ajvSchemas[schemaName];
+ if (!schema(entries)) {
+ throw this._formatAjvSchemaError(schema, file.filename);
}
progressData.index = startIndex + 1;
@@ -771,4 +786,17 @@ export class DictionaryImporter {
// - '\ufa67'.normalize('NFC') => '\u9038' (逸 => 逸)
return text;
}
+
+ /**
+ * @template [T=unknown]
+ * @param {import('@zip.js/zip.js').Entry} entry
+ * @param {import('@zip.js/zip.js').Writer<T>|import('@zip.js/zip.js').WritableWriter} writer
+ * @returns {Promise<T>}
+ */
+ async _getData(entry, writer) {
+ if (typeof entry.getData === 'undefined') {
+ throw new Error(`Cannot read ${entry.filename}`);
+ }
+ return await entry.getData(writer);
+ }
}
diff --git a/jsconfig.json b/jsconfig.json
index 7d23472f..ace0c2aa 100644
--- a/jsconfig.json
+++ b/jsconfig.json
@@ -20,7 +20,10 @@
"handlebars",
"jszip",
"parse5",
- "wanakana"
+ "wanakana",
+ "zip.js",
+ "dexie",
+ "ajv"
]
},
"include": [
diff --git a/package-lock.json b/package-lock.json
index 3fb951f7..8980b096 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -29,6 +29,7 @@
"@types/jsdom": "^21.1.6",
"@types/node": "^20.10.0",
"@types/wanakana": "^4.0.6",
+ "@types/zip.js": "^2.0.32",
"@typescript-eslint/eslint-plugin": "^6.12.0",
"@typescript-eslint/parser": "^6.12.0",
"@vitest/coverage-v8": "^0.34.6",
@@ -949,6 +950,12 @@
"integrity": "sha512-al8hJELQI+RDcexy6JLV/BqghQ/nP0B9d62m0F3jEvPyxAq9RXFH9xDoGa73oT9/keCUKRxWCA6l37wv4TCfQw==",
"dev": true
},
+ "node_modules/@types/zip.js": {
+ "version": "2.0.32",
+ "resolved": "https://registry.npmjs.org/@types/zip.js/-/zip.js-2.0.32.tgz",
+ "integrity": "sha512-+/r1iYLsLUCTNsDiGcrqK7LQ9ui11GVC98Dj3x0GtpuvzKM2PK8k/gXeu2RyZWTiVR3k6pxodHnAiBMBVsNebw==",
+ "dev": true
+ },
"node_modules/@typescript-eslint/eslint-plugin": {
"version": "6.12.0",
"resolved": "https://registry.npmjs.org/@typescript-eslint/eslint-plugin/-/eslint-plugin-6.12.0.tgz",
@@ -6477,6 +6484,12 @@
"integrity": "sha512-al8hJELQI+RDcexy6JLV/BqghQ/nP0B9d62m0F3jEvPyxAq9RXFH9xDoGa73oT9/keCUKRxWCA6l37wv4TCfQw==",
"dev": true
},
+ "@types/zip.js": {
+ "version": "2.0.32",
+ "resolved": "https://registry.npmjs.org/@types/zip.js/-/zip.js-2.0.32.tgz",
+ "integrity": "sha512-+/r1iYLsLUCTNsDiGcrqK7LQ9ui11GVC98Dj3x0GtpuvzKM2PK8k/gXeu2RyZWTiVR3k6pxodHnAiBMBVsNebw==",
+ "dev": true
+ },
"@typescript-eslint/eslint-plugin": {
"version": "6.12.0",
"resolved": "https://registry.npmjs.org/@typescript-eslint/eslint-plugin/-/eslint-plugin-6.12.0.tgz",
diff --git a/package.json b/package.json
index bc2aebd7..0d9796bb 100644
--- a/package.json
+++ b/package.json
@@ -52,6 +52,7 @@
"@types/jsdom": "^21.1.6",
"@types/node": "^20.10.0",
"@types/wanakana": "^4.0.6",
+ "@types/zip.js": "^2.0.32",
"@typescript-eslint/eslint-plugin": "^6.12.0",
"@typescript-eslint/parser": "^6.12.0",
"@vitest/coverage-v8": "^0.34.6",
diff --git a/types/ext/dictionary-importer.d.ts b/types/ext/dictionary-importer.d.ts
index 16ce66ce..de85d04a 100644
--- a/types/ext/dictionary-importer.d.ts
+++ b/types/ext/dictionary-importer.d.ts
@@ -15,6 +15,8 @@
* along with this program. If not, see <https://www.gnu.org/licenses/>.
*/
+import type * as Ajv from 'ajv';
+import type * as ZipJS from '@zip.js/zip.js';
import type * as DictionaryData from './dictionary-data';
import type * as DictionaryDatabase from './dictionary-database';
import type * as StructuredContent from './structured-content';
@@ -90,6 +92,29 @@ export type StructuredContentImageImportRequirement = {
};
export type ImportRequirementContext = {
- archive: import('jszip');
+ fileMap: ArchiveFileMap;
media: Map<string, DictionaryDatabase.MediaDataArrayBufferContent>;
};
+
+export type ArchiveFileMap = Map<string, ZipJS.Entry>;
+
+export type CompiledSchemaNameArray = [
+ termBank: CompiledSchemaName,
+ termMetaBank: CompiledSchemaName,
+ kanjiBank: CompiledSchemaName,
+ kanjiMetaBank: CompiledSchemaName,
+ tagBank: CompiledSchemaName,
+];
+
+export type CompiledSchemaValidators = {
+ dictionaryIndex: Ajv.ValidateFunction<unknown>;
+ dictionaryTermBankV1: Ajv.ValidateFunction<unknown>;
+ dictionaryTermBankV3: Ajv.ValidateFunction<unknown>;
+ dictionaryTermMetaBankV3: Ajv.ValidateFunction<unknown>;
+ dictionaryKanjiBankV1: Ajv.ValidateFunction<unknown>;
+ dictionaryKanjiBankV3: Ajv.ValidateFunction<unknown>;
+ dictionaryKanjiMetaBankV3: Ajv.ValidateFunction<unknown>;
+ dictionaryTagBankV3: Ajv.ValidateFunction<unknown>;
+};
+
+export type CompiledSchemaName = keyof CompiledSchemaValidators;