feat: add dest argument (#15)

* feat: add dest argument

---------

Signed-off-by: Frost Ming <[email protected]>

.gitignore

@@ -167,4 +167,3 @@ cython_debug/
 *.epub
 *.zip
 .DS_Store
-

epubhv/cli.py

@@ -1,34 +1,36 @@
-from dataclasses import dataclass
+from argparse import ArgumentParser, RawTextHelpFormatter
 from pathlib import Path
-from argparse import ArgumentParser, Namespace, RawTextHelpFormatter
+from typing import cast
 
 from epubhv.epubhv import EPUBHV, list_all_epub_in_dir
 
 
-@dataclass
 class Options:
     epub: str
-    v: bool
-    h: bool
+    method: str
     convert: str
     punctuation: str
     ruby: bool
     cantonese: bool
+    dest: Path
 
 
 def main() -> None:
-    parser: ArgumentParser = ArgumentParser(formatter_class=RawTextHelpFormatter)
+    parser = ArgumentParser(formatter_class=RawTextHelpFormatter)
     parser.add_argument("epub", help="file or dir that contains epub files to change")
     parser.add_argument(
         "--v",
-        dest="v",
-        action="store_true",
+        dest="method",
+        action="store_const",
+        const="to_vertical",
+        default="to_vertical",
         help="change all the epub files to vertical.",
     )
     parser.add_argument(
         "--h",
-        dest="h",
-        action="store_true",
+        dest="method",
+        action="store_const",
+        const="to_horizontal",
         help="change all the epub files to hortical.",
     )
     parser.add_argument(
@@ -43,6 +45,13 @@ def main() -> None:
         action="store_true",
         help="Ruby it for cantonese.",
     )
+    parser.add_argument(
+        "-d",
+        "--dest",
+        help="destination dir to save the epub files, default to current directory",
+        default=".",
+        type=Path,
+    )
 
     parser.add_argument(
         "--punctuation",
@@ -96,29 +105,14 @@ def main() -> None:
         """,
     )
 
-    raw_args: Namespace = parser.parse_args()
-    options: Options = Options(
-        epub=raw_args.epub,
-        v=raw_args.v,
-        h=raw_args.h,
-        convert=raw_args.convert,
-        punctuation=raw_args.punctuation,
-        ruby=raw_args.ruby,
-        cantonese=raw_args.cantonese,
-    )
+    options = cast(Options, parser.parse_args())
     epub_files = Path(options.epub)
-    # default is to to_vertical
-    method: str = "to_vertical"
-    if options.h:
-        method = "to_horizontal"
-    elif options.v:
-        method = "to_vertical"
     if epub_files.exists():
         if epub_files.is_dir():
             files: set[Path] = list_all_epub_in_dir(path=epub_files)
             f: Path
             for f in files:
-                print(f"{str(f)} is {method}")
+                print(f"{str(f)} is {options.method}")
                 try:
                     epubhv: EPUBHV = EPUBHV(
                         file_path=f,
@@ -127,18 +121,18 @@ def main() -> None:
                         need_ruby=options.ruby,
                         need_cantonese=options.cantonese,
                     )
-                    epubhv.run(method=method)
+                    epubhv.run(method=options.method, dest=options.dest)
                 except Exception as e:
-                    print(f"{str(f)} {method} is failed by {str(e)}")
+                    print(f"{str(f)} {options.method} is failed by {str(e)}")
         else:
-            print(f"{str(epub_files)} is {method}")
+            print(f"{str(epub_files)} is {options.method}")
             epubhv: EPUBHV = EPUBHV(
                 file_path=epub_files,
                 convert_to=options.convert,
                 need_ruby=options.ruby,
                 need_cantonese=options.cantonese,
             )
-            epubhv.run(method=method)
+            epubhv.run(method=options.method, dest=options.dest)
     else:
         raise Exception("Please make sure it is a dir contains epub or is a epub file.")
 

epubhv/epubhv.py

@@ -5,7 +5,7 @@
 import os
 import shutil
 import zipfile
-from collections import defaultdict, Counter
+from collections import Counter, defaultdict
 from pathlib import Path
 from typing import Dict, List, Optional
 
@@ -15,11 +15,10 @@
 from bs4 import NavigableString, PageElement, ResultSet, Tag
 from cssutils import CSSParser
 from cssutils.css import CSSStyleSheet
+from langdetect import LangDetectException, detect
 
 from epubhv.punctuation import Punctuation
 from epubhv.yomituki import RubySoup, string_containers  # pyright: ignore
-from langdetect import detect, LangDetectException
-
 
 cssutils.log.setLevel(logging.CRITICAL)  # type: ignore
 
@@ -66,40 +65,31 @@ def load_opf_meta_data(opf_file: Path) -> bs:
 
 
 class EPUBHV:
+    book_path: Path
+    book_name: str
+    opf_file: Path
+
     def __init__(
         self,
         file_path: Path,
         convert_to: Optional[str] = None,
-        convert_punctuation: Optional[str] = "auto",
+        convert_punctuation: str = "auto",
         need_ruby: bool = False,
         need_cantonese: bool = False,
     ) -> None:
         # declare instance fields
-        self.epub_file: Path
+        self.epub_file = file_path
         self.has_css_file: bool = False
         # for language ruby
         self.need_ruby: bool = need_ruby
         self.ruby_language = None
         self.cantonese = need_cantonese
         self.files_dict: Dict[str, List[Path]] = {}
         self.content_files_list: List[Path] = []
-        self.book_path: Path
-        self.book_name: str
-        self.opf_file: Path
-        self.converter: Optional[opencc.OpenCC]
-        self.convert_to: Optional[str]
-        self.convert_punctuation: str = "auto"
-
-        # initialize instance fields
-        self.epub_file = file_path
-        if convert_to is not None:
-            self.converter = opencc.OpenCC(convert_to)
-            self.convert_to = convert_to
-        else:
-            self.converter = None
-            self.convert_to = None
-        if convert_punctuation:
-            self.convert_punctuation = convert_punctuation
+        self.convert_punctuation = convert_punctuation
+        self.convert_to = convert_to
+
+        self.converter = opencc.OpenCC(convert_to) if convert_to is not None else None
 
     def extract_one_epub_to_dir(self) -> None:
         assert self.epub_file.suffix == ".epub", f"{self.epub_file} Must be epub file"
@@ -342,9 +332,11 @@ def convert(self, method: str = "to_vertical") -> None:
                 content: str = f.read()
             soup: bs = bs(content, "html.parser")
             if self.converter:
-                html_element: Optional[Tag | NavigableString] = soup.find("html")
-                assert html_element is not None
-                text_elements: ResultSet[PageElement] = html_element.find_all(string=True)  # type: ignore
+                html_element = soup.find("html")
+                assert isinstance(html_element, Tag)
+                text_elements: ResultSet[PageElement] = html_element.find_all(
+                    string=True
+                )  # type: ignore
 
                 element: Tag
                 for element in text_elements:  # type: ignore
@@ -386,22 +378,25 @@ def convert(self, method: str = "to_vertical") -> None:
                 with open(html_file, "w", encoding="utf-8") as file:
                     file.write(ruby_soup.prettify())
 
-    def pack(self, method: str = "to_vertical") -> None:
-        lang: str = "original"
+    def pack(self, method: str = "to_vertical", dest: Path = Path.cwd()) -> None:
+        lang = "original"
         if self.convert_to is not None:
             lang = self.convert_to
         if self.need_ruby:
             lang = f"{lang}-ruby"
         if method == "to_vertical":
-            book_name: str = f"{self.book_name}-v-{lang}.epub"
+            book_name = f"{self.book_name}-v-{lang}.epub"
         else:
-            book_name: str = f"{self.book_name}-h-{lang}.epub"
+            book_name = f"{self.book_name}-h-{lang}.epub"
+        pack_to = dest / book_name
 
-        shutil.make_archive(base_name=book_name, format="zip", root_dir=self.book_path)
-        os.rename(src=book_name + ".zip", dst=book_name)
+        shutil.make_archive(
+            base_name=str(pack_to), format="zip", root_dir=self.book_path
+        )
+        os.rename(src=f"{pack_to}.zip", dst=pack_to)
         shutil.rmtree(self.book_path)
 
-    def run(self, method: str = "to_vertical") -> None:
+    def run(self, method: str = "to_vertical", dest: Path = Path.cwd()) -> None:
         assert method in [
             "to_horizontal",
             "to_vertical",
@@ -416,4 +411,4 @@ def run(self, method: str = "to_vertical") -> None:
             raise Exception("Only support epub to vertical or horizontal for now")
 
         self.convert(method=method)
-        self.pack(method=method)
+        self.pack(method=method, dest=dest)

pyproject.toml

@@ -33,7 +33,7 @@ requires = ["pdm-backend"]
 build-backend = "pdm.backend"
 
 [tool.pdm.build]
-include = ["epubhv"]
+includes = ["epubhv"]
 
 [tool.pdm.scripts]
 format = "black ."

tests/test_epubhv.py

@@ -1,4 +1,3 @@
-import os
 from pathlib import Path
 from shutil import rmtree
 from typing import Dict, List
@@ -13,19 +12,21 @@
     make_epub_files_dict,
 )
 
+TEST_DIR = Path(__file__).with_name("test_epub")
+
 
 @pytest.fixture
 def epub() -> EPUBHV:
-    return EPUBHV(file_path=Path("tests/test_epub/animal_farm.epub"))
+    return EPUBHV(file_path=TEST_DIR / "animal_farm.epub")
 
 
 def test_find_epub_books() -> None:
-    assert list_all_epub_in_dir(Path("tests/test_epub")) == {
-        Path("tests/test_epub/animal_farm.epub"),
-        Path("tests/test_epub/books/animal.epub"),
-        Path("tests/test_epub/Liber_Esther.epub"),
-        Path("tests/test_epub/books/lemo.epub"),
-        Path("tests/test_epub/sanguo.epub"),
+    assert list_all_epub_in_dir(TEST_DIR) == {
+        TEST_DIR / "animal_farm.epub",
+        TEST_DIR / "books/animal.epub",
+        TEST_DIR / "Liber_Esther.epub",
+        TEST_DIR / "books/lemo.epub",
+        TEST_DIR / "sanguo.epub",
     }
 
 
@@ -47,38 +48,29 @@ def test_make_files_dict(epub: EPUBHV) -> None:
     rmtree(".epub_temp_dir/animal_farm")
 
 
-def test_change_epub_to_vertical():
-    epub_file = Path("animal_farm-v-original.epub")
-    epub_file.unlink(True)
-    b = EPUBHV(Path("tests/test_epub/animal_farm.epub"))
-    b.run()
-    assert b.opf_file == Path(".epub_temp_dir/animal_farm/content.opf")
-    epub_file.unlink(True)
+def test_change_epub_to_vertical(epub: EPUBHV, tmp_path: Path) -> None:
+    epub.run(dest=tmp_path)
+    assert epub.opf_file == Path(".epub_temp_dir/animal_farm/content.opf")
+    assert tmp_path.joinpath("animal_farm-v-original.epub").exists()
 
 
-def test_find_epub_css_files():
-    lemo_output = Path("lemo-h-original.epub")
-    lemo_output.unlink(True)
-    b = EPUBHV(Path("tests/test_epub/animal_farm.epub"))
+def test_find_epub_css_files(tmp_path: Path) -> None:
+    lemo_output = tmp_path / "lemo-h-original.epub"
+    b = EPUBHV(TEST_DIR / "animal_farm.epub")
     b.make_epub_values()
     assert b.has_css_file is False
-    b.run()
-    os.remove("animal_farm-v-original.epub")
-    f: EPUBHV = EPUBHV(Path("tests/test_epub/books/lemo.epub"))
-    f.run("to_horizontal")
+    b.run(dest=tmp_path)
+
+    f = EPUBHV(TEST_DIR / "books/lemo.epub")
+    f.run("to_horizontal", dest=tmp_path)
     assert lemo_output.exists()
-    lemo_output.unlink(True)
-
-
-def test_change_epub_covert() -> None:
-    if os.path.exists("sanguo-v-s2t-v-original.epub"):
-        os.remove("sanguo-v-s2t-v-original.epub")
-    if os.path.exists("sanguo-v-s2t.epub"):
-        os.remove("sanguo-v-s2t.epub")
-    f: EPUBHV = EPUBHV(Path("tests/test_epub/sanguo.epub"), "s2t")
-    f.run("to_vertical")
-    assert os.path.exists("sanguo-v-s2t.epub") is True
-    q: EPUBHV = EPUBHV(Path("sanguo-v-s2t.epub"))
+
+
+def test_change_epub_covert(tmp_path: Path) -> None:
+    f = EPUBHV(TEST_DIR / "sanguo.epub", "s2t")
+    f.run("to_vertical", dest=tmp_path)
+    assert tmp_path.joinpath("sanguo-v-s2t.epub").exists()
+    q = EPUBHV(tmp_path.joinpath("sanguo-v-s2t.epub"))
     q.extract_one_epub_to_dir()
     q.make_epub_values()
     has_t_count: int = 0
@@ -92,31 +84,23 @@ def test_change_epub_covert() -> None:
             if r.find("滾滾長江東逝水") > 0:
                 has_t_count += 1
     assert has_t_count > 0
-    q.run("to_vertical")
-    os.remove("sanguo-v-s2t.epub")
-    os.remove("sanguo-v-s2t-v-original.epub")
+    q.run("to_vertical", dest=tmp_path)
 
 
-def test_ruby() -> None:
-    lemo_output = Path("lemo-h-original-ruby.epub")
-    lemo_output.unlink(True)
-    f: EPUBHV = EPUBHV(Path("tests/test_epub/books/lemo.epub"), need_ruby=True)
-    f.run("to_horizontal")
+def test_ruby(tmp_path: Path) -> None:
+    lemo_output = tmp_path / "lemo-h-original-ruby.epub"
+    f = EPUBHV(TEST_DIR / "books/lemo.epub", need_ruby=True)
+    f.run("to_horizontal", dest=tmp_path)
     assert f.ruby_language == "ja"
     assert lemo_output.exists()
-    lemo_output.unlink(True)
 
 
-def test_cantonese() -> None:
-    animal_output = Path("animal-h-original-ruby.epub")
-    animal_output.unlink(True)
-    f: EPUBHV = EPUBHV(
-        Path("tests/test_epub/books/animal.epub"), need_ruby=True, need_cantonese=True
-    )
-    f.run("to_horizontal")
+def test_cantonese(tmp_path: Path) -> None:
+    animal_output = tmp_path / "animal-h-original-ruby.epub"
+    f = EPUBHV(TEST_DIR / "books/animal.epub", need_ruby=True, need_cantonese=True)
+    f.run("to_horizontal", dest=tmp_path)
     assert f.ruby_language == "cantonese"
     assert animal_output.exists()
-    animal_output.unlink(True)
 
 
 def test_punctuation():