novel-downloader 2.0.1__tar.gz → 2.0.2__tar.gz

{novel_downloader-2.0.1/src/novel_downloader.egg-info → novel_downloader-2.0.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: novel-downloader
-Version: 2.0.1
+Version: 2.0.2
 Summary: A command-line tool for downloading Chinese web novels from Qidian and similar platforms.
 Author-email: Saudade Z <saudadez217@gmail.com>
 License: MIT License
@@ -123,6 +123,7 @@ novel-cli download 123456
 
 * 支持站点见: [支持站点列表](https://github.com/saudadez21/novel-downloader/blob/main/docs/4-supported-sites.md)
 * 更多示例见: [CLI 使用示例](https://github.com/saudadez21/novel-downloader/blob/main/docs/5-cli-usage-examples.md)
+* 运行中可使用 `CTRL+C` 取消任务
 
 ### 3. 图形界面 (GUI / Web)
 
@@ -134,6 +135,8 @@ novel-web
 # novel-web --listen public
 ```
 
+* 运行中可使用 `CTRL+C` 停止服务
+
 ---
 
 ## 从源码安装 (开发版)

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/README.md

@@ -56,6 +56,7 @@ novel-cli download 123456
 
 * 支持站点见: [支持站点列表](https://github.com/saudadez21/novel-downloader/blob/main/docs/4-supported-sites.md)
 * 更多示例见: [CLI 使用示例](https://github.com/saudadez21/novel-downloader/blob/main/docs/5-cli-usage-examples.md)
+* 运行中可使用 `CTRL+C` 取消任务
 
 ### 3. 图形界面 (GUI / Web)
 
@@ -67,6 +68,8 @@ novel-web
 # novel-web --listen public
 ```
 
+* 运行中可使用 `CTRL+C` 停止服务
+
 ---
 
 ## 从源码安装 (开发版)

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/pyproject.toml

@@ -111,7 +111,7 @@ testpaths = ["tests"]
 
 [tool.commitizen]
 name = "cz_conventional_commits"
-version = "2.0.1"
+version = "2.0.2"
 version_files = [
     "src/novel_downloader/__init__.py"
 ]

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/src/novel_downloader/__init__.py

@@ -6,7 +6,7 @@ novel_downloader
 Core package for the Novel Downloader project.
 """
 
-__version__ = "2.0.1"
+__version__ = "2.0.2"
 
 __author__ = "Saudade Z"
 __email__ = "saudadez217@gmail.com"

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/src/novel_downloader/cli/download.py

@@ -178,7 +178,14 @@ async def _download(
 
         for book in valid_books:
             ui.info(t("download_downloading", book_id=book["book_id"], site=site))
-            await downloader.download(book, progress_hook=_print_progress)
+
+            hook, close = ui.create_progress_hook(
+                prefix=t("download_progress_prefix"), unit="chapters"
+            )
+            try:
+                await downloader.download(book, progress_hook=hook)
+            finally:
+                close()
 
             if not no_export and exporter is not None:
                 await asyncio.to_thread(exporter.export, book["book_id"])
@@ -186,6 +193,9 @@ async def _download(
         if downloader_cfg.login_required and fetcher.is_logged_in:
             await fetcher.save_state()
 
+    if exporter is not None:
+        exporter.close()
+
 
 async def _prompt_login_fields(
     fields: list[LoginField],
@@ -235,10 +245,3 @@ async def _prompt_login_fields(
         result[field.name] = value
 
     return result
-
-
-async def _print_progress(done: int, total: int) -> None:
-    """Progress hook passed into the downloader."""
-    ui.print_progress(
-        done, total, prefix=t("download_progress_prefix"), unit="chapters"
-    )

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/src/novel_downloader/cli/export.py

@@ -55,22 +55,22 @@ def handle_export(args: Namespace) -> None:
     adapter = ConfigAdapter(config=config_data, site=site)
     exporter_cfg = adapter.get_exporter_config()
     log_level = adapter.get_log_level()
-    exporter = get_exporter(site, exporter_cfg)
     setup_logging(console_level=log_level)
 
-    for book_id in book_ids:
-        ui.info(t("export_processing", book_id=book_id, format=export_format))
-
-        if export_format in {"txt", "all"}:
-            try:
-                exporter.export_as_txt(book_id)
-                ui.success(t("export_success_txt", book_id=book_id))
-            except Exception as e:
-                ui.error(t("export_failed_txt", book_id=book_id, err=str(e)))
-
-        if export_format in {"epub", "all"}:
-            try:
-                exporter.export_as_epub(book_id)
-                ui.success(t("export_success_epub", book_id=book_id))
-            except Exception as e:
-                ui.error(t("export_failed_epub", book_id=book_id, err=str(e)))
+    with get_exporter(site, exporter_cfg) as exporter:
+        for book_id in book_ids:
+            ui.info(t("export_processing", book_id=book_id, format=export_format))
+
+            if export_format in {"txt", "all"}:
+                try:
+                    exporter.export_as_txt(book_id)
+                    ui.success(t("export_success_txt", book_id=book_id))
+                except Exception as e:
+                    ui.error(t("export_failed_txt", book_id=book_id, err=str(e)))
+
+            if export_format in {"epub", "all"}:
+                try:
+                    exporter.export_as_epub(book_id)
+                    ui.success(t("export_success_epub", book_id=book_id))
+                except Exception as e:
+                    ui.error(t("export_failed_epub", book_id=book_id, err=str(e)))

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/src/novel_downloader/cli/ui.py

@@ -17,9 +17,10 @@ Public API:
 
 from __future__ import annotations
 
-from collections.abc import Iterable, Sequence
+from collections.abc import Awaitable, Callable, Iterable, Sequence
 
 from rich.console import Console
+from rich.progress import Progress, TaskID
 from rich.prompt import Confirm, Prompt
 from rich.table import Table
 
@@ -154,3 +155,29 @@ def print_progress(
     total = max(1, total)
     pct = done / total * 100.0
     _CONSOLE.print(f"[dim]{prefix}[/] {done}/{total} {unit} ({pct:.2f}%)")
+
+
+def create_progress_hook(
+    prefix: str = "Progress",
+    unit: str = "item",
+) -> tuple[Callable[[int, int], Awaitable[None]], Callable[[], None]]:
+    progress = Progress(console=_CONSOLE)
+    task_id: TaskID | None = None
+
+    async def hook(done: int, total: int) -> None:
+        nonlocal task_id
+        if task_id is None:
+            progress.start()
+            task_id = progress.add_task(f"[cyan]{prefix}[/]", total=max(1, total))
+
+        progress.update(
+            task_id,
+            completed=done,
+            total=max(1, total),
+            description=f"{prefix} ({done}/{total} {unit})",
+        )
+
+    def close() -> None:
+        progress.stop()
+
+    return hook, close

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/src/novel_downloader/config/adapter.py

@@ -17,6 +17,7 @@ from novel_downloader.models import (
     DownloaderConfig,
     ExporterConfig,
     FetcherConfig,
+    FontOCRConfig,
     ParserConfig,
     TextCleanerConfig,
 )
@@ -97,13 +98,16 @@ class ConfigAdapter:
         """
         g = self._gen_cfg
         s = self._site_cfg
-        font_ocr = g.get("font_ocr") or {}
+        g_font = g.get("font_ocr") or {}
+        s_font = s.get("font_ocr") or {}
+        font_ocr: dict[str, Any] = {**g_font, **s_font}
         return ParserConfig(
             cache_dir=g.get("cache_dir", "./novel_cache"),
             use_truncation=bool(s.get("use_truncation", True)),
             decode_font=bool(font_ocr.get("decode_font", False)),
             save_font_debug=bool(font_ocr.get("save_font_debug", False)),
             batch_size=int(font_ocr.get("batch_size", 32)),
+            fontocr_cfg=self._dict_to_fontocr_cfg(font_ocr),
         )
 
     def get_exporter_config(self) -> ExporterConfig:
@@ -283,6 +287,28 @@ class ConfigAdapter:
                 out["ignore_ids"] = [str(x) for x in data["ignore_ids"]]
         return out
 
+    @staticmethod
+    def _dict_to_fontocr_cfg(data: dict[str, Any]) -> FontOCRConfig:
+        """
+        Convert a raw ``font_ocr`` dict into a :class:`FontOCRConfig`.
+        """
+        if not isinstance(data, dict):
+            return FontOCRConfig()
+
+        ishape = data.get("input_shape")
+        if isinstance(ishape, list):
+            ishape = tuple(ishape)  # [C, H, W] -> (C, H, W)
+
+        return FontOCRConfig(
+            model_name=data.get("model_name"),
+            model_dir=data.get("model_dir"),
+            input_shape=ishape,
+            device=data.get("device"),
+            precision=data.get("precision", "fp32"),
+            cpu_threads=data.get("cpu_threads", 10),
+            enable_hpi=data.get("enable_hpi", False),
+        )
+
     @classmethod
     def _dict_to_cleaner_cfg(cls, cfg: dict[str, Any]) -> TextCleanerConfig:
         """

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/src/novel_downloader/core/archived/deqixs/fetcher.py

@@ -8,8 +8,6 @@ novel_downloader.core.archived.deqixs.fetcher
 from typing import Any
 
 from novel_downloader.core.fetchers.base import BaseSession
-from novel_downloader.models import FetcherConfig
-from novel_downloader.utils import async_jitter_sleep
 
 # from novel_downloader.core.fetchers.registry import register_fetcher
 
@@ -26,25 +24,11 @@ class DeqixsSession(BaseSession):
     BOOK_INFO_URL = "https://www.deqixs.com/xiaoshuo/{book_id}/"
     CHAPTER_URL = "https://www.deqixs.com/xiaoshuo/{book_id}/{chapter_id}.html"
 
-    def __init__(
-        self,
-        config: FetcherConfig,
-        cookies: dict[str, str] | None = None,
-        **kwargs: Any,
-    ) -> None:
-        super().__init__("deqixs", config, cookies, **kwargs)
-
     async def get_book_info(
         self,
         book_id: str,
         **kwargs: Any,
     ) -> list[str]:
-        """
-        Fetch the raw HTML of the book info page asynchronously.
-
-        :param book_id: The book identifier.
-        :return: The page content as a string.
-        """
         url = self.book_info_url(book_id=book_id)
         return [await self.fetch(url, **kwargs)]
 
@@ -54,13 +38,6 @@ class DeqixsSession(BaseSession):
         chapter_id: str,
         **kwargs: Any,
     ) -> list[str]:
-        """
-        Fetch the raw HTML of a single chapter asynchronously.
-
-        :param book_id: The book identifier.
-        :param chapter_id: The chapter identifier.
-        :return: The chapter content as a string.
-        """
         html_pages: list[str] = []
         idx = 1
 
@@ -85,11 +62,7 @@ class DeqixsSession(BaseSession):
 
             html_pages.append(html)
             idx += 1
-            await async_jitter_sleep(
-                self.request_interval,
-                mul_spread=1.1,
-                max_sleep=self.request_interval + 2,
-            )
+            await self._sleep()
 
         return html_pages
 

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/src/novel_downloader/core/downloaders/__init__.py

@@ -11,9 +11,11 @@ __all__ = [
     "CommonDownloader",
     "QianbiDownloader",
     "QidianDownloader",
+    "QqbookDownloader",
 ]
 
 from .common import CommonDownloader
 from .qianbi import QianbiDownloader
 from .qidian import QidianDownloader
+from .qqbook import QqbookDownloader
 from .registry import get_downloader

{novel_downloader-2.0.1 → novel_downloader-2.0.2}/src/novel_downloader/core/downloaders/base.py

@@ -10,15 +10,11 @@ import abc
 import asyncio
 import json
 import logging
-from collections.abc import AsyncIterator, Awaitable, Callable, Sequence
+from collections.abc import Awaitable, Callable, Sequence
 from pathlib import Path
-from typing import Any, cast
+from typing import Any, ClassVar, cast
 
-from novel_downloader.core.interfaces import (
-    DownloaderProtocol,
-    FetcherProtocol,
-    ParserProtocol,
-)
+from novel_downloader.core.interfaces import FetcherProtocol, ParserProtocol
 from novel_downloader.models import (
     BookConfig,
     BookInfoDict,
@@ -28,7 +24,7 @@ from novel_downloader.models import (
 from novel_downloader.utils import time_diff
 
 
-class BaseDownloader(DownloaderProtocol, abc.ABC):
+class BaseDownloader(abc.ABC):
     """
     Abstract base class for novel downloaders.
 
@@ -39,8 +35,8 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
     a single book, using the provided fetcher and parser components.
     """
 
-    DEFAULT_SOURCE_ID = 0
-    PRIORITIES_MAP = {
+    DEFAULT_SOURCE_ID: ClassVar[int] = 0
+    PRIORITIES_MAP: ClassVar[dict[int, int]] = {
         DEFAULT_SOURCE_ID: 0,
     }
 
@@ -61,15 +57,23 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
         """
         self._fetcher = fetcher
         self._parser = parser
-        self._config = config
         self._site = site
 
+        self._save_html = config.save_html
+        self._skip_existing = config.skip_existing
+        self._login_required = config.login_required
+        self._request_interval = config.request_interval
+        self._retry_times = config.retry_times
+        self._backoff_factor = config.backoff_factor
+        self._workers = config.workers
+        self._storage_batch_size = max(1, config.storage_batch_size)
+
         self._raw_data_dir = Path(config.raw_data_dir) / site
         self._raw_data_dir.mkdir(parents=True, exist_ok=True)
         self._debug_dir = Path.cwd() / "debug" / site
         self._debug_dir.mkdir(parents=True, exist_ok=True)
 
-        self.logger = logging.getLogger(f"{self.__class__.__name__}")
+        self.logger = logging.getLogger(f"{__name__}.{self.__class__.__name__}")
 
     async def download_many(
         self,
@@ -87,7 +91,7 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
                                 args: completed_count, total_count.
         :param cancel_event: Optional asyncio.Event to allow cancellation.
         """
-        if not await self._ensure_ready():
+        if not self._check_login():
             book_ids = [b["book_id"] for b in books]
             self.logger.warning(
                 "[%s] login failed, skipping download of books: %s",
@@ -116,8 +120,6 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
             except Exception as e:
                 self._handle_download_exception(book, e)
 
-        await self._finalize()
-
     async def download(
         self,
         book: BookConfig,
@@ -134,7 +136,7 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
                                 args: completed_count, total_count.
         :param cancel_event: Optional asyncio.Event to allow cancellation.
         """
-        if not await self._ensure_ready():
+        if not self._check_login():
             self.logger.warning(
                 "[%s] login failed, skipping download of book: %s (%s-%s)",
                 self._site,
@@ -142,6 +144,7 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
                 book.get("start_id", "-"),
                 book.get("end_id", "-"),
             )
+            return
 
         # if already cancelled before starting
         if cancel_event and cancel_event.is_set():
@@ -162,8 +165,6 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
         except Exception as e:
             self._handle_download_exception(book, e)
 
-        await self._finalize()
-
     async def load_book_info(
         self,
         book_id: str,
@@ -200,23 +201,6 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
         """
         ...
 
-    async def _prepare(self) -> None:
-        """
-        Optional hook called before downloading.
-
-        Subclasses can override this method to perform pre-download setup.
-        """
-        return
-
-    async def _finalize(self) -> None:
-        """
-        Optional hook called after downloading is complete.
-
-        Subclasses can override this method to perform post-download tasks,
-        such as saving state or releasing resources.
-        """
-        return
-
     def _load_book_info(
         self,
         book_id: str,
@@ -283,25 +267,22 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
         :param filename: used as filename prefix
         :param html_list: list of HTML strings to save
         """
-        if not self.save_html:
+        if not self._save_html:
             return
-
         html_dir.mkdir(parents=True, exist_ok=True)
         for i, html in enumerate(html_list):
-            file_path = html_dir / f"{filename}_{i}.html"
-            file_path.write_text(html, encoding="utf-8")
+            (html_dir / f"{filename}_{i}.html").write_text(html, encoding="utf-8")
 
     @staticmethod
-    async def _chapter_ids(
-        volumes: list[VolumeInfoDict],
+    def _planned_chapter_ids(
+        vols: list[VolumeInfoDict],
         start_id: str | None,
         end_id: str | None,
-    ) -> AsyncIterator[str]:
-        """
-        Yield each chapterId in order, respecting start/end bounds.
-        """
+        ignore: set[str],
+    ) -> list[str]:
         seen_start = start_id is None
-        for vol in volumes:
+        out: list[str] = []
+        for vol in vols:
             for chap in vol["chapters"]:
                 cid = chap.get("chapterId")
                 if not cid:
@@ -311,9 +292,11 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
                         seen_start = True
                     else:
                         continue
-                yield cid
+                if cid not in ignore:
+                    out.append(cid)
                 if end_id is not None and cid == end_id:
-                    return
+                    return out
+        return out
 
     @property
     def fetcher(self) -> FetcherProtocol:
@@ -323,38 +306,6 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
     def parser(self) -> ParserProtocol:
         return self._parser
 
-    @property
-    def save_html(self) -> bool:
-        return self._config.save_html
-
-    @property
-    def skip_existing(self) -> bool:
-        return self._config.skip_existing
-
-    @property
-    def login_required(self) -> bool:
-        return self._config.login_required
-
-    @property
-    def request_interval(self) -> float:
-        return self._config.request_interval
-
-    @property
-    def retry_times(self) -> int:
-        return self._config.retry_times
-
-    @property
-    def backoff_factor(self) -> float:
-        return self._config.backoff_factor
-
-    @property
-    def workers(self) -> int:
-        return self._config.workers
-
-    @property
-    def storage_batch_size(self) -> int:
-        return max(1, self._config.storage_batch_size)
-
     def _handle_download_exception(self, book: BookConfig, error: Exception) -> None:
         """
         Handle download errors in a consistent way.
@@ -373,10 +324,8 @@ class BaseDownloader(DownloaderProtocol, abc.ABC):
             error,
         )
 
-    async def _ensure_ready(self) -> bool:
+    def _check_login(self) -> bool:
         """
-        Run pre-download preparation and check login if needed.
+        Check login if needed.
         """
-        await self._prepare()
-
-        return self.fetcher.is_logged_in if self.login_required else True
+        return self.fetcher.is_logged_in if self._login_required else True