txt2ebook 0.1.140__py3-none-any.whl → 0.1.142__py3-none-any.whl

txt2ebook/parser.py

@@ -19,6 +19,8 @@ import argparse
 import logging
 from dataclasses import dataclass
 from importlib import import_module
+from importlib import import_module
+from types import ModuleType
 from typing import List, Tuple, Union
 
 import regex as re
@@ -36,14 +38,15 @@ class Parser:
 
     raw_content: str
     config: argparse.Namespace
+    langconf: ModuleType
 
-    def __init__(self, raw_content: str, config: argparse.Namespace) -> None:
+    def __init__(
+        self, raw_content: str, config: argparse.Namespace, langconf: ModuleType
+    ) -> None:
         """Set the constructor for the Parser."""
         self.raw_content = raw_content
         self.config = config
-
-        config_lang = config.language.replace("-", "_")
-        self.langconf = import_module(f"txt2ebook.languages.{config_lang}")
+        self.langconf = langconf
 
     def parse(self) -> Book:
         """Parse the content into volumes (optional) and chapters.

txt2ebook/subcommands/parse.py

@@ -19,10 +19,14 @@ import argparse
 import logging
 import sys
 
+import logging
+import sys
+from importlib import import_module
+
 import jieba.analyse
 from bs4 import UnicodeDammit
-from langdetect import detect
 
+from txt2ebook import detect_and_expect_language
 from txt2ebook.exceptions import EmptyFileError
 from txt2ebook.models import Book
 from txt2ebook.parser import Parser
@@ -73,26 +77,20 @@ def run(args: argparse.Namespace) -> Book:
     logger.info("Detect encoding : %s", unicode.original_encoding)
 
     content = unicode.unicode_markup
+
     if not content:
         raise EmptyFileError(f"Empty file content in {args.input_file.name}")
 
-    args_language = args.language
-    detect_language = detect(content)
-    args.language = args_language or detect_language
-    logger.info("args language: %s", args_language)
-    logger.info("Detect language: %s", detect_language)
+    logger.info("Detect encoding : %s", unicode.original_encoding)
 
-    if args_language and args_language != detect_language:
-        logger.warning(
-            "args (%s) and detect (%s) language mismatch",
-            args_language,
-            detect_language,
-        )
+    args.language = detect_and_expect_language(content, args.language)
+    config_lang = args.language.replace("-", "_")
+    langconf = import_module(f"txt2ebook.languages.{config_lang}")
 
     tags = jieba.analyse.extract_tags(content, topK=100)
     logger.info("tags: %s", " ".join(tags))
 
-    parser = Parser(content, args)
+    parser = Parser(content, args, langconf)
     book = parser.parse()
 
     if args.debug:

{txt2ebook-0.1.140.dist-info → txt2ebook-0.1.142.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: txt2ebook
-Version: 0.1.140
+Version: 0.1.142
 Summary: CLI tool to convert txt file to ebook format
 Project-URL: Homepage, https://github.com/kianmeng/txt2ebook
 Project-URL: Repository, https://github.com/kianmeng/txt2ebook
@@ -108,12 +108,12 @@ positional arguments:
     typ
       generate ebook in Typst format
 
-options:
-  -of, --output-folder OUTPUT_FOLDER
+optional arguments:
+  -of OUTPUT_FOLDER, --output-folder OUTPUT_FOLDER
       set default output folder (default: 'output')
   -p, --purge
       remove converted ebooks specified by --output-folder option (default: 'False')
-  -l, --language LANGUAGE
+  -l LANGUAGE, --language LANGUAGE
       language of the ebook (default: 'None')
   -rw, --raise-on-warning
       raise exception and stop parsing upon warning

{txt2ebook-0.1.140.dist-info → txt2ebook-0.1.142.dist-info}/RECORD

@@ -2,7 +2,7 @@ txt2ebook/__init__.py,sha256=Oq0Yor9IB6LPfAsVVTl-wbh-EFVy8T309BR1UVMC0kw,3055
 txt2ebook/__main__.py,sha256=L29rlfPSx9XMnVaHBYP2dyYgDmutJvONR3yUejjYwRY,855
 txt2ebook/cli.py,sha256=i8NrYJyC9ckMC5opCGkIcs42p4AFzhE0lTGKSU-S8Zw,4418
 txt2ebook/exceptions.py,sha256=PT3m85PE5QopHHUfRwEQzp0kJ4AA9yjLO6V6lYC8WhQ,858
-txt2ebook/parser.py,sha256=rf_iS73yW4FFVzHFupCUPJbypmWrqN6yuYwjLTZ08KQ,8989
+txt2ebook/parser.py,sha256=foAO-ezeb_cSPW2U3Ul83xsf4xdOZsBJaqJmcy1_rio,9015
 txt2ebook/tokenizer.py,sha256=UGyOBGxlKOXJtvP2UFp38EgFym8-PAU3A7Jl9RF3w6Y,10299
 txt2ebook/zh_utils.py,sha256=0Yq9r-JL4HntW68vFR6TBP9yQim1a07mfsh_sp-XmaE,4887
 txt2ebook/formats/__init__.py,sha256=CBZSA9zbLL4-4VYH7Xp76HK4kHTyISoNs7gMs7lBIzY,1646
@@ -41,12 +41,12 @@ txt2ebook/subcommands/epub.py,sha256=_obM1_fvVBPHOBXBOCYK8nyJadBX3_gOn9kaXA5HipA
 txt2ebook/subcommands/gmi.py,sha256=ANnPg-RFsylTo44fUzFOSHN1fC3Ce82gBzrv-sBv5fU,3318
 txt2ebook/subcommands/massage.py,sha256=EuC-C03NMJk9V1_PEUOa-n4SmQCRpj1TJ_GwSJE8_Ss,11809
 txt2ebook/subcommands/md.py,sha256=PmIqrqrnzLywvN4qTkle0V9N3FTIJGRWpC0Xbk76B5o,3329
-txt2ebook/subcommands/parse.py,sha256=FaYTWa2yqkowwPAmHWJC7iCii2Rnus3SUHG10GjjJp4,3022
+txt2ebook/subcommands/parse.py,sha256=3LP3GGgX5amfde3fpvobZf6Ks1_nA9PqFh2hjYslmaA,2929
 txt2ebook/subcommands/pdf.py,sha256=1JQtpugzAIaho6G3CK1rGYk74hotAexXZxPH9PHpRps,2980
 txt2ebook/subcommands/tex.py,sha256=ToYdFXnFLwsXxTsZzCRsURo7TCeOIFJtp5sFJDt0R-E,3131
 txt2ebook/subcommands/typ.py,sha256=qXpHMmtu_1nAMs264oKUSolWAUBjZpTziTSBcTe2JgA,3681
-txt2ebook-0.1.140.dist-info/METADATA,sha256=_nJDfV9pCfmUWs5OGwLB3LUvLpxf7CUE0tvF3aZj5Q0,4867
-txt2ebook-0.1.140.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-txt2ebook-0.1.140.dist-info/entry_points.txt,sha256=3jm5vpUsDRgoM6S3CQVMMiP7tJQqfq1vfV0sh_KaK9s,74
-txt2ebook-0.1.140.dist-info/licenses/LICENSE.md,sha256=tGtFDwxWTjuR9syrJoSv1Hiffd2u8Tu8cYClfrXS_YU,31956
-txt2ebook-0.1.140.dist-info/RECORD,,
+txt2ebook-0.1.142.dist-info/METADATA,sha256=0-WMp4gWS4JzWBe0UDwU6LFtGgHr-WF4G6JieMtZ78Y,4901
+txt2ebook-0.1.142.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+txt2ebook-0.1.142.dist-info/entry_points.txt,sha256=3jm5vpUsDRgoM6S3CQVMMiP7tJQqfq1vfV0sh_KaK9s,74
+txt2ebook-0.1.142.dist-info/licenses/LICENSE.md,sha256=tGtFDwxWTjuR9syrJoSv1Hiffd2u8Tu8cYClfrXS_YU,31956
+txt2ebook-0.1.142.dist-info/RECORD,,