kjqwdw/pixiv
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

修复注册扫描识别失败问题

Browse Source
This commit is contained in:
kjqwer
2025-10-09 09:01:29 +08:00
parent f32feafd17
commit 1483f93b99
4 changed files with 245 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/services/download-registry.js
+50 -10
View File
@@ -1,6 +1,8 @@
const path = require('path');
const fs = require('fs-extra');
const { defaultLogger } = require('../utils/logger');
const ConfigManager = require('../config/config-manager');
const artworkUtils = require('../utils/artwork-utils');
// 创建logger实例
const logger = defaultLogger.child('DownloadRegistry');
@@ -12,14 +14,14 @@ const logger = defaultLogger.child('DownloadRegistry');
class DownloadRegistry {
constructor(dataPath) {
this.dataPath = dataPath;
this.registryPath = path.join(dataPath, 'download_registry.json');
this.registryPath = path.join(dataPath, 'download-registry.json');
this.registry = {
version: '1.0.5',
created_at: new Date().toISOString(),
updated_at: new Date().toISOString(),
artists: {} // 格式: { artistName: { artworks: [artworkId1, artworkId2, ...] } }
artists: {},
lastUpdated: null
};
this.loaded = false;
this.configManager = new ConfigManager();
}
/**
@@ -185,6 +187,44 @@ class DownloadRegistry {
return false;
}
/**
* 从作品目录名中提取作品ID
* @param {string} artworkDir - 作品目录名
* @returns {number|null} 作品ID，如果无法提取则返回null
*/
async extractArtworkIdFromDir(artworkDir) {
return await artworkUtils.extractArtworkIdFromDir(artworkDir);
}
/**
* 检查作品是否已在注册表中注册
* @param {string} artistName - 作者名称
* @param {string} artworkDir - 作品目录名
* @returns {boolean} 是否已注册
*/
async isArtworkRegistered(artistName, artworkDir) {
if (!this.loaded) {
await this.loadRegistry();
}
// 使用动态方法从作品目录名中提取作品ID
const artworkId = await this.extractArtworkIdFromDir(artworkDir);
if (!artworkId) {
logger.warn(`无法从作品目录名中提取作品ID: ${artworkDir}`);
return false;
}
const normalizedArtistName = this.normalizeArtistName(artistName);
// 检查艺术家是否存在
if (!this.registry.artists[normalizedArtistName]) {
return false;
}
// 检查作品是否在艺术家的作品列表中
return this.registry.artists[normalizedArtistName].artworks.includes(artworkId);
}
/**
* 获取已下载的作品ID列表
* @returns {number[]} 作品ID数组
@@ -387,10 +427,10 @@ class DownloadRegistry {
const isRegistered = await this.isArtworkRegistered(artistDir, artworkDir);
if (!isRegistered) {
// 获取作品信息并添加到注册表
const artworkInfo = await fileManager.getArtworkInfo(artistDir, artworkDir);
if (artworkInfo) {
await this.addArtwork(artistDir, artworkDir, artworkInfo);
// 从作品目录名中提取作品ID并添加到注册表
const artworkId = await this.extractArtworkIdFromDir(artworkDir);
if (artworkId) {
await this.addArtwork(artistDir, artworkId);
stats.addedArtworks++;
logger.debug(`添加作品到注册表: ${artistDir}/${artworkDir}`);
}
@@ -451,8 +491,8 @@ class DownloadRegistry {
const artworkEntries = await fileManager.listDirectory(artistPath);
for (const entry of artworkEntries) {
const match = entry.match(/^(\d+)_(.+)$/);
if (match && parseInt(match[1]) === artworkId) {
const extractedArtworkId = await artworkUtils.extractArtworkIdFromDir(entry);
if (extractedArtworkId && extractedArtworkId === artworkId) {
const artworkPath = path.join(artistPath, entry);
const infoPath = path.join(artworkPath, 'artwork_info.json');
backend/services/download.js
+5 -6
View File
@@ -10,6 +10,7 @@ const DownloadRegistry = require('./download-registry');
const CacheConfigManager = require('../config/cache-config');
const fs = require('fs-extra'); // Added for fs-extra
const { defaultLogger } = require('../utils/logger');
const artworkUtils = require('../utils/artwork-utils');
// 创建logger实例
const logger = defaultLogger.child('DownloadService');
@@ -784,11 +785,9 @@ class DownloadService {
const artworks = await this.fileManager.listDirectory(artistPath);
for (const artwork of artworks) {
// 检查是否是作品目录（包含数字ID）
const artworkMatch = artwork.match(/^(\d+)_(.+)$/);
if (artworkMatch) {
const artworkId = artworkMatch[1];
// 使用工具函数检查是否是作品目录并提取ID
const artworkId = await artworkUtils.extractArtworkIdFromDir(artwork);
if (artworkId) {
// 检查作品目录是否包含图片文件
const artworkPath = path.join(artistPath, artwork);
const artworkStat = await this.fileManager.getFileInfo(artworkPath);
@@ -797,7 +796,7 @@ class DownloadService {
const files = await this.fileManager.listDirectory(artworkPath);
const imageFiles = files.filter(file => /\.(jpg|jpeg|png|gif|webp)$/i.test(file));
if (imageFiles.length > 0) {
downloadedIds.add(parseInt(artworkId));
downloadedIds.add(artworkId);
}
}
}
backend/services/repository.js
+12 -10
View File
@@ -5,6 +5,7 @@ const { exec } = require('child_process')
const ConfigManager = require('../config/config-manager')
const execAsync = promisify(exec)
const { defaultLogger } = require('../utils/logger');
const artworkUtils = require('../utils/artwork-utils');
// 创建logger实例
const logger = defaultLogger.child('RepositoryService');
@@ -465,9 +466,9 @@ class RepositoryService {
for (const artworkEntry of artworkEntries) {
if (!artworkEntry.isDirectory()) continue
// 检查是否是目标作品目录（包含数字ID）
const artworkMatch = artworkEntry.name.match(/^(\d+)_(.+)$/)
if (artworkMatch && artworkMatch[1] === artworkId.toString()) {
// 使用工具函数检查是否是目标作品目录
const extractedArtworkId = await artworkUtils.extractArtworkIdFromDir(artworkEntry.name);
if (extractedArtworkId && extractedArtworkId === parseInt(artworkId)) {
const artworkPath = path.join(artistPath, artworkEntry.name)
// 检查作品信息文件 - 这是最可靠的判断标准
@@ -769,9 +770,10 @@ class RepositoryService {
const artistDir = artwork.artistPath
try {
const artistEntries = await fs.readdir(artistDir, { withFileTypes: true })
const hasArtworks = artistEntries.some(entry =>
entry.isDirectory() && entry.name.match(/^\d+_/)
)
const hasArtworks = artistEntries.some(async (entry) => {
if (!entry.isDirectory()) return false;
return await artworkUtils.isArtworkDirectory(entry.name);
});
if (!hasArtworks) {
await fs.rmdir(artistDir)
@@ -816,11 +818,11 @@ class RepositoryService {
for (const artworkEntry of artworkEntries) {
if (!artworkEntry.isDirectory()) continue
// 检查是否是目标作品目录（包含数字ID）
const artworkMatch = artworkEntry.name.match(/^(\d+)_(.+)$/)
if (artworkMatch && artworkMatch[1] === artworkId.toString()) {
// 使用工具函数检查是否是目标作品目录
const extractedArtworkId = await artworkUtils.extractArtworkIdFromDir(artworkEntry.name);
if (extractedArtworkId && extractedArtworkId === parseInt(artworkId)) {
const artworkPath = path.join(artistPath, artworkEntry.name)
const title = artworkMatch[2]
const title = await artworkUtils.extractTitleFromDir(artworkEntry.name) || 'Unknown Title';
// 找到目标作品，返回基本信息（不需要扫描文件详情）
return {
backend/utils/artwork-utils.js
+178
View File
@@ -0,0 +1,178 @@
const ConfigManager = require('../config/config-manager');
const { defaultLogger } = require('./logger');
const logger = defaultLogger.child('ArtworkUtils');
class ArtworkUtils {
constructor() {
this.configManager = new ConfigManager();
this._cachedConfig = null;
this._configCacheTime = 0;
this.CACHE_DURATION = 30000; // 30秒缓存
}
/**
* 获取配置（带缓存）
* @returns {Promise<Object>} 配置对象
*/
async getConfig() {
const now = Date.now();
if (this._cachedConfig && (now - this._configCacheTime) < this.CACHE_DURATION) {
return this._cachedConfig;
}
this._cachedConfig = await this.configManager.readConfig();
this._configCacheTime = now;
return this._cachedConfig;
}
/**
* 从作品目录名中提取作品ID
* @param {string} artworkDir - 作品目录名
* @returns {Promise<number|null>} 作品ID，如果无法提取则返回null
*/
async extractArtworkIdFromDir(artworkDir) {
try {
// 获取配置中的命名模式
const config = await this.getConfig();
const namingPattern = config.namingPattern || "{artist_name}/{artwork_id}_{title}";
// 从命名模式中提取作品目录部分（去掉 {artist_name}/ 前缀）
let artworkPattern = namingPattern;
if (artworkPattern.includes('/')) {
artworkPattern = artworkPattern.split('/').pop(); // 取最后一部分
}
// 将命名模式转换为正则表达式
// 替换占位符为对应的正则表达式组
let regexPattern = artworkPattern
.replace(/\{artwork_id\}/g, '(\\d+)') // artwork_id 匹配数字
.replace(/\{title\}/g, '(.+)') // title 匹配任意字符
.replace(/\{artist_name\}/g, '(.+)') // artist_name 匹配任意字符（虽然在这里不应该出现）
.replace(/\{[^}]+\}/g, '(.+)'); // 其他占位符匹配任意字符
// 转义特殊字符
regexPattern = regexPattern.replace(/[.*+?^${}()|[\]\\]/g, (match) => {
// 不转义我们添加的正则表达式组
if (match === '(' || match === ')' || match === '\\') {
return match;
}
return '\\' + match;
});
// 创建正则表达式
const regex = new RegExp(`^${regexPattern}$`);
const match = artworkDir.match(regex);
if (match) {
// 找到 artwork_id 在模式中的位置
const placeholders = artworkPattern.match(/\{[^}]+\}/g) || [];
const artworkIdIndex = placeholders.findIndex(placeholder => placeholder === '{artwork_id}');
if (artworkIdIndex !== -1 && match[artworkIdIndex + 1]) {
return parseInt(match[artworkIdIndex + 1]);
}
}
// 如果动态解析失败，回退到默认格式 {artwork_id}_{title}
const fallbackMatch = artworkDir.match(/^(\d+)_(.+)$/);
if (fallbackMatch) {
logger.debug(`使用回退模式解析作品目录: ${artworkDir}`);
return parseInt(fallbackMatch[1]);
}
return null;
} catch (error) {
logger.warn(`解析作品目录名失败: ${artworkDir}`, error.message);
// 发生错误时回退到默认格式
const fallbackMatch = artworkDir.match(/^(\d+)_(.+)$/);
if (fallbackMatch) {
return parseInt(fallbackMatch[1]);
}
return null;
}
}
/**
* 检查目录名是否匹配作品目录格式
* @param {string} dirName - 目录名
* @returns {Promise<boolean>} 是否匹配作品目录格式
*/
async isArtworkDirectory(dirName) {
const artworkId = await this.extractArtworkIdFromDir(dirName);
return artworkId !== null;
}
/**
* 从作品目录名中提取标题
* @param {string} artworkDir - 作品目录名
* @returns {Promise<string|null>} 作品标题，如果无法提取则返回null
*/
async extractTitleFromDir(artworkDir) {
try {
// 获取配置中的命名模式
const config = await this.getConfig();
const namingPattern = config.namingPattern || "{artist_name}/{artwork_id}_{title}";
// 从命名模式中提取作品目录部分（去掉 {artist_name}/ 前缀）
let artworkPattern = namingPattern;
if (artworkPattern.includes('/')) {
artworkPattern = artworkPattern.split('/').pop(); // 取最后一部分
}
// 将命名模式转换为正则表达式
let regexPattern = artworkPattern
.replace(/\{artwork_id\}/g, '(\\d+)') // artwork_id 匹配数字
.replace(/\{title\}/g, '(.+)') // title 匹配任意字符
.replace(/\{artist_name\}/g, '(.+)') // artist_name 匹配任意字符
.replace(/\{[^}]+\}/g, '(.+)'); // 其他占位符匹配任意字符
// 转义特殊字符
regexPattern = regexPattern.replace(/[.*+?^${}()|[\]\\]/g, (match) => {
if (match === '(' || match === ')' || match === '\\') {
return match;
}
return '\\' + match;
});
// 创建正则表达式
const regex = new RegExp(`^${regexPattern}$`);
const match = artworkDir.match(regex);
if (match) {
// 找到 title 在模式中的位置
const placeholders = artworkPattern.match(/\{[^}]+\}/g) || [];
const titleIndex = placeholders.findIndex(placeholder => placeholder === '{title}');
if (titleIndex !== -1 && match[titleIndex + 1]) {
return match[titleIndex + 1];
}
}
// 如果动态解析失败，回退到默认格式 {artwork_id}_{title}
const fallbackMatch = artworkDir.match(/^(\d+)_(.+)$/);
if (fallbackMatch) {
return fallbackMatch[2];
}
return null;
} catch (error) {
logger.warn(`解析作品标题失败: ${artworkDir}`, error.message);
// 发生错误时回退到默认格式
const fallbackMatch = artworkDir.match(/^(\d+)_(.+)$/);
if (fallbackMatch) {
return fallbackMatch[2];
}
return null;
}
}
}
// 创建单例实例
const artworkUtils = new ArtworkUtils();
module.exports = artworkUtils;