novel-downloader 1.1.1__py3-none-any.whl

novel_downloader/resources/config/rules.toml

@@ -0,0 +1,196 @@
+[qidian.profile] # 起点
+book_info_url = "https://book.qidian.com/info/{book_id}/"
+chapter_url = "https://www.qidian.com/chapter/{book_id}/{chapter_id}/"
+
+[qidian.book_info.book_name]
+steps = [
+    { type = "select_one", selector = "em#bookName" },
+    { type = "text" },
+    { type = "strip" }
+]
+
+[qidian.book_info.author]
+steps = [
+    { type = "select_one", selector = "a.writer" },
+    { type = "text" },
+    { type = "strip" }
+]
+
+[qidian.book_info.cover_url]
+steps = [
+    { type = "select_one", selector = "div.book-img img" },
+    { type = "attr", attr = "src" },
+    { type = "strip" }
+]
+
+[qidian.book_info.update_time]
+steps = [
+    { type = "select_one", selector = "span.book-update-time" },
+    { type = "text" },
+    { type = "replace", old = "更新时间", new = "" },
+    { type = "strip" }
+]
+
+[qidian.book_info.serial_status]
+steps = [
+    { type = "select_one", selector = "span.blue" },
+    { type = "text" },
+    { type = "strip" }
+]
+
+[qidian.book_info.word_count]
+steps = [
+    { type = "regex", pattern = "<em>([\\d.]+)</em>\\s*<cite>(.*?)字</cite>", template = "$1$2字" }
+]
+
+[qidian.book_info.summary]
+steps = [
+    { type = "select_one", selector = "div.book-intro p" },
+    { type = "text" },
+    { type = "strip" }
+]
+
+[qidian.book_info.volumes]
+has_volume = true
+volume_selector = "div.volume-wrap div.volume"
+chapter_selector = "li"
+
+volume_name_steps = [
+    { type = "select_one", selector = "h3" },
+    { type = "text" },
+    { type = "split", sep = "·", index = 0 },
+    { type = "replace", old = "订阅本卷", new = "" },
+    { type = "strip" }
+]
+
+chapter_steps = [
+    { key = "title", steps = [
+        { type = "select_one", selector = "a" },
+        { type = "text" }
+    ] },
+
+    { key = "url", steps = [
+        { type = "select_one", selector = "a" },
+        { type = "attr", attr = "href" }
+    ] },
+
+    { key = "chapterId", steps = [
+        { type = "select_one", selector = "a" },
+        { type = "attr", attr = "href" },
+        { type = "regex", pattern = "/(\\d+)", group = 1 }
+    ] }
+]
+
+[qidian.chapter.title] # 章节标题
+steps = [
+    { type = "select_one", selector = "div#app div.print h1.title" },
+    { type = "exclude", selector = "span.review" },
+    { type = "text" },
+    { type = "strip" }
+]
+
+[qidian.chapter.content] # 章节正文
+steps = [
+    { type = "select_one", selector = "div#app div#reader-content main" },
+    { type = "exclude", selector = "span.review" },
+    { type = "find_all", name = "p" },
+    { type = "text" },
+    { type = "strip" },
+    { type = "join", sep = "\n\n" }
+]
+
+[biquge.profile] # 笔趣阁
+book_info_url = "http://www.b520.cc/{book_id}/"
+chapter_url = "http://www.b520.cc/{book_id}/{chapter_id}.html"
+
+[biquge.book_info.book_name]
+steps = [
+    { type = "select_one", selector = "div#info > h1" },
+    { type = "text" },
+    { type = "strip" }
+]
+
+[biquge.book_info.author]
+steps = [
+    { type = "select", selector = "div#info > p", index = 0 },
+    { type = "text" },
+    { type = "replace", old = "\u00a0", new = "" },
+    { type = "regex", pattern = "作\\s*者[:：]?\\s*(\\S+)", group = 1 },
+    { type = "strip" }
+]
+
+[biquge.book_info.cover_url]
+steps = [
+    { type = "select_one", selector = "div#fmimg > img" },
+    { type = "attr", attr = "src" },
+    { type = "strip" }
+]
+
+[biquge.book_info.update_time]
+steps = [
+    { type = "select", selector = "div#info > p", index = 2 },
+    { type = "text" },
+    { type = "regex", pattern = "最后更新[:：]\\s*(\\S+)", group = 1 },
+    { type = "strip" }
+]
+
+# [biquge.book_info.serial_status] # 没有
+
+# [biquge.book_info.word_count] # 没有
+
+[biquge.book_info.summary]
+steps = [
+    { type = "select_one", selector = "div#intro" },
+    { type = "text" },
+    { type = "strip" }
+]
+
+[biquge.book_info.volumes]
+has_volume = true
+volume_mode = "mixed"
+list_selector = "div#list"
+volume_selector = "dt"
+chapter_selector = "dd"
+
+volume_name_steps = [
+    { type = "select_one", selector = "dt" },
+    { type = "text" },
+    { type = "strip" },
+    { type = "split", sep = "\n", index = 0 },
+    { type = "strip" }
+]
+
+chapter_steps = [
+    { key = "title", steps = [
+        { type = "select_one", selector = "a" },
+        { type = "text" }
+    ] },
+
+    { key = "url", steps = [
+        { type = "select_one", selector = "a" },
+        { type = "attr", attr = "href" }
+    ] },
+
+    { key = "chapterId", steps = [
+        { type = "select_one", selector = "a" },
+        { type = "attr", attr = "href" },
+        { type = "replace", old = ".html", new = "" },
+        { type = "regex", pattern = ".*/(\\d+)$", group = 1 }
+    ] }
+]
+
+[biquge.chapter.title] # 章节标题
+steps = [
+    { type = "select_one", selector = "div.bookname > h1" },
+    { type = "text" },
+    { type = "strip" }
+]
+
+[biquge.chapter.content] # 章节正文
+steps = [
+    { type = "select_one", selector = "div#content" },
+    { type = "find_all", name = "p" },
+    { type = "text" },
+    { type = "strip" },
+    { type = "join", sep = "\n\n" }
+]

novel_downloader/resources/config/settings.yaml

@@ -0,0 +1,70 @@
+# 网络请求层设置
+requests:
+  wait_time: 5                      # 每次请求等待时间 (秒)
+  retry_times: 3                    # 请求失败重试次数
+  retry_interval: 5
+  timeout: 30                       # 页面加载超时时间 (秒)
+  # DrissionPage 专用设置
+  headless: false                   # 是否以无头模式启动浏览器
+  user_data_folder: ""              # 浏览器用户数据目录: 为空则使用默认目录
+  profile_name: ""                  # 使用的用户配置名称: 为空则使用默认配置
+  auto_close: true                  # 页面抓取完是否自动关闭浏览器
+  disable_images: false             # 是否禁用图片加载 (加速)
+  mute_audio: true                  # 是否静音
+
+# 全局通用设置
+general:
+  request_interval: 5               # 同一本书各章节请求间隔 (秒)
+  raw_data_dir: "./raw_data"        # 原始章节 HTML/JSON 存放目录
+  output_dir: "./downloads"         # 最终输出文件存放目录
+  cache_dir: "./novel_cache"        # 本地缓存目录 (字体 / 图片等)
+  max_threads: 4                    # 最大并发下载线程数 (未实现)
+  skip_existing: true               # 是否跳过已存在章节
+  debug:
+    save_html: false                # 是否将抓取到的原始 HTML 保留到磁盘
+    log_level: "INFO"               # 日志级别: DEBUG, INFO, WARNING, ERROR
+
+# 各站点的特定配置
+sites:
+  qidian:
+    # 小说 ID 列表
+    # 例如: 访问 https://www.qidian.com/book/1010868264/
+    # 该小说的 ID 就是 1010868264
+    book_ids:
+      - "0000000000"
+      - "0000000000"
+    mode: "browser"                 # browser / session
+    login_required: true            # 是否需要登录才能访问
+    decode_font: false              # 是否尝试本地解码混淆字体
+    use_freq: false                 # 是否使用频率分析
+    ocr_version: "v2.0"             # "v1.0" / "v2.0"
+    use_ocr: true                   # 是否使用 OCR 辅助识别文本
+    use_vec: false                  # 是否使用 Vector 辅助识别文本
+    save_font_debug: false          # 是否保存字体解码调试数据
+    batch_size: 32
+    ocr_weight: 0.6
+    vec_weight: 0.4
+    #
+  sample_site:
+    book_ids:
+      - "ABCDEF"
+    api_key: "your_api_key"
+    special_flag: 42
+  common:
+    mode: "session"                 # async / session
+    login_required: false           # 是否需要登录才能访问
+
+# 输出文件格式及相关选项
+output:
+  clean_text: true                  # 是否对章节文本做清理
+  formats:
+    make_txt: true                  # 是否生成完整 TXT 文件
+    make_epub: false                # 是否生成 EPUB
+    make_md: false                  # 是否生成 Markdown (未实现)
+    make_pdf: false                 # 可能支持 PDF 输出 (未实现)
+  naming:
+    append_timestamp: false         # 在文件名中追加时间戳
+    filename_template: "{title}_{author}" # 文件命名规则
+  epub:
+    include_cover: true             # 是否在 EPUB 中包含封面
+    include_toc: false              # 是否自动生成目录

novel_downloader/resources/css_styles/main.css

@@ -0,0 +1,104 @@
+html, body {
+    height: 95%;
+    padding: 0.5em;
+}
+
+.images img {
+    width:100%; /*自适应*/
+}
+.color .images {
+    margin:0 0 1.5em;
+}
+.article .images {
+    margin:1.0em 0;
+    border:1px solid #000
+}
+
+.new-page-after {
+    page-break-after:always;
+}
+
+.cover {
+    width:100%;
+    padding:0px;
+}
+.center {
+    text-align: center;
+    margin-left: 0%;
+    margin-right: 0%;
+}
+.left {
+    text-align: center;
+    margin-left: 0%;
+    margin-right: 0%;
+}
+.right {
+    text-align: right;
+    margin-left: 0%;
+    margin-right: 0%;
+}
+
+hr {
+    margin: 0;
+    border-style: dotted;
+    border-width: 1px 0 0 0;
+    margin-top: 0.5em;
+    margin-bottom: 0.5em;
+}
+
+h1 {
+    text-indent: 0;
+    duokan-text-indent: 0;
+    line-height: 110%;
+    text-align: center;
+    font-size: 1.2em;
+    margin-top: 0.4em;
+    margin-bottom: 0.4em;
+}
+
+h2, h3, h4, h5, h6 {
+    text-indent: 0;
+    duokan-text-indent: 0;
+    line-height: 110%;
+    font-size: 1.1em;
+    margin-top: 0.4em;
+    margin-bottom: 0.4em;
+}
+
+p {
+    text-align: justify;  /* 让文本两端对齐 */
+    text-indent: 2em;
+    duokan-text-indent: 2em;
+    line-height: 100%;
+    margin-top: 0.4em;
+    margin-bottom: 0.4em;
+}
+
+div.list {
+    text-align: left;
+    font-size: 0.85em;
+    line-height: 130%;
+}
+
+span {
+    text-indent: 0;
+    duokan-text-indent: 0;
+    text-align: left;
+    font-size: 1em;
+}
+
+ol {
+    text-align: left;
+    margin-top: 1.2em;
+    margin-bottom: 1.2em;
+    margin-left: 0;
+    list-style-type: none;
+}
+
+li {
+    text-align: left;
+    margin-top: 0.4em;
+    margin-bottom: 0.4em;
+    margin-left: 0;
+    list-style-type: none;
+}

novel_downloader/resources/css_styles/volume-intro.css

@@ -0,0 +1,56 @@
+/*标题页*/
+
+h1.volume-title-line1 {
+    font-size: 1.25rem;              /* 第一行较大 */
+    font-weight: bold;
+    text-align: center;
+    text-indent: 0em;
+    duokan-text-indent: 0em;
+    color: #6e471c;
+    text-shadow: 1px 1px 2px rgba(255, 255, 255, 0.8);
+    margin: 0;
+    display: block;
+}
+
+p.volume-title-line2 {
+    font-size: 1rem;            /* 第二行稍小 */
+    font-weight: bold;
+    text-align: center;
+    text-indent: 0em;
+    duokan-text-indent: 0em;
+    color: #6e471c;
+    text-shadow: 1px 1px 2px rgba(255, 255, 255, 0.8);
+    margin: 0;
+    display: block;
+}
+
+div.border1 {
+    margin: 35% auto 0 auto;
+    text-align: center;
+    text-indent: 0em;
+    duokan-text-indent: 0em;
+}
+
+div.border2 {
+    margin: 0 auto 0 auto;
+    text-align: center;
+    text-indent: 0em;
+    duokan-text-indent: 0em;
+    transform: rotate(180deg);
+    -ms-transform: rotate(180deg);
+    /* IE 9 */
+    -moz-transform: rotate(180deg);
+    /* Firefox */
+    -webkit-transform: rotate(180deg);
+    /* Safari 和 Chrome */
+    -o-transform: rotate(180deg);
+    /* Opera */;
+}
+
+img.border1 {
+    width: 100%;
+}
+
+img.border2 {
+    width: 80%;
+}

novel_downloader/resources/js_scripts/qidian_decrypt_node.js

@@ -0,0 +1,82 @@
+window = global;
+null_fun = function(){console.log(arguments);}
+window.outerHeight = 1000
+window.innerHeight = 100
+globalThis = window
+self = window
+window.location = {}
+location.protocol = "https:"
+location.hostname = "vipreader.qidian.com"
+setTimeout = null_fun
+setInterval = null_fun
+document = {createElement: null_fun, documentElement: {}, createEvent: null_fun, currentScript: {src: "https://qdfepccdn.qidian.com/www.qidian.com/fock/116594983210.js"}, domain: 'qidian.com'}
+navigator = {userAgent: 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/135.0.0.0 Safari/537.36'}
+performance = {}
+performance.navigation = {type: 1}
+
+// ---- DECRYPT FUNCTION ----
+const Fock = require('./4819793b.qeooxh.js');
+
+async function decrypt(enContent, cuChapterId, fkp, fuid) {
+  Fock.initialize();
+  Fock.setupUserKey(fuid);
+  eval(atob(fkp));
+  isFockInit = true;
+
+  return new Promise((resolve, reject) => {
+    Fock.unlock(enContent, cuChapterId, (code, decrypted) => {
+      if (code === 0) {
+        resolve(decrypted);
+      } else {
+        reject(new Error(`Fock.unlock failed, code=${code}`));
+      }
+    });
+  });
+}
+
+// ---- MAIN ----
+const fs = require('fs');
+
+(async () => {
+  const [inputPath, outputPath] = process.argv.slice(2);
+  if (!inputPath || !outputPath) {
+    console.error(
+      "Usage: node script.js <input.json> <output.txt>"
+    );
+    process.exit(1);
+  }
+
+  try {
+    const inputData = fs.readFileSync(inputPath, "utf-8");
+    const [
+      raw_enContent,
+      raw_cuChapterId,
+      raw_fkp,
+      raw_fuid
+    ] = JSON.parse(inputData);
+
+    const decryptPromise = decrypt(
+      String(raw_enContent),
+      String(raw_cuChapterId),
+      String(raw_fkp),
+      String(raw_fuid)
+    );
+
+    const timeoutMs = 5000;
+    const timerPromise = new Promise((_, reject) => {
+      setTimeout(() => {
+        reject(new Error(`decrypt timeout after ${timeoutMs}ms`));
+      }, timeoutMs);
+    });
+
+    const result = await Promise.race([
+      decryptPromise,
+      timerPromise
+    ]);
+
+    fs.writeFileSync(outputPath, result, "utf-8");
+  } catch (err) {
+    console.error("Failed to decrypt:", err);
+    process.exit(1);
+  }
+})();

novel_downloader/resources/json/replace_word_map.json

@@ -0,0 +1,4 @@
+{
+  "source_1": "target_1",
+  "source_2": "target_2"
+}

novel_downloader/resources/text/blacklist.txt

@@ -0,0 +1,22 @@
+求票
+月票
+投票
+首订
+订阅
+追订
+追读
+加更
+推荐
+收藏
+新书
+成绩
+建议
+太监
+烂尾
+完本
+更新
+支持
+感谢
+推书
+盟主
+提示

novel_downloader/utils/cache.py

@@ -0,0 +1,24 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+novel_downloader.utils.cache
+----------------------------
+
+Provides decorators for caching function results,
+specifically optimized for configuration loading functions.
+"""
+
+from functools import lru_cache, wraps
+from typing import Any, Callable, TypeVar, cast
+
+T = TypeVar("T", bound=Callable[..., Any])
+
+
+def cached_load_config(func: T) -> T:
+    """
+    A decorator to cache the result of a config-loading function.
+    Uses LRU cache with maxsize=1.
+    """
+    cached = lru_cache(maxsize=1)(func)
+    wrapped = wraps(func)(cached)
+    return cast(T, wrapped)