Spaces:
Configuration error
Configuration error
Dien-Hoa
commited on
Commit
·
fa7018d
1
Parent(s):
1826c30
refactor translate 1 and multi languages
Browse files- main.py +2 -2
- translator.py +36 -21
main.py
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
from fasthtml.common import *
|
| 2 |
from fasthtml.xtend import CheckboxX
|
| 3 |
-
from translator import languages as VALID_LANGUAGES,
|
| 4 |
import tempfile
|
| 5 |
from starlette.requests import Request
|
| 6 |
from typing import List
|
|
@@ -72,7 +72,7 @@ def translate_and_process_sbv(content: bytes, filename: str, languages: List[str
|
|
| 72 |
output_dir = os.path.join(temp_dir, "translations")
|
| 73 |
os.makedirs(output_dir, exist_ok=True)
|
| 74 |
|
| 75 |
-
|
| 76 |
|
| 77 |
return create_translation_files(temp_dir, output_dir, filename, selected_languages)
|
| 78 |
except Exception as e:
|
|
|
|
| 1 |
from fasthtml.common import *
|
| 2 |
from fasthtml.xtend import CheckboxX
|
| 3 |
+
from translator import languages as VALID_LANGUAGES, translate_to_languages
|
| 4 |
import tempfile
|
| 5 |
from starlette.requests import Request
|
| 6 |
from typing import List
|
|
|
|
| 72 |
output_dir = os.path.join(temp_dir, "translations")
|
| 73 |
os.makedirs(output_dir, exist_ok=True)
|
| 74 |
|
| 75 |
+
translate_to_languages(input_file, output_dir, selected_languages)
|
| 76 |
|
| 77 |
return create_translation_files(temp_dir, output_dir, filename, selected_languages)
|
| 78 |
except Exception as e:
|
translator.py
CHANGED
|
@@ -90,39 +90,53 @@ def reconstruct_sbv_content(blocks):
|
|
| 90 |
def get_output_file_path(output_dir, base_filename, lang):
|
| 91 |
return os.path.join(output_dir, f"{base_filename}_{lang}.sbv")
|
| 92 |
|
| 93 |
-
def
|
| 94 |
"""
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
This function reads an input SBV file, parses its content, translates it into
|
| 98 |
-
the specified languages, and saves the translated versions as separate files.
|
| 99 |
|
| 100 |
Args:
|
| 101 |
input_file (str): Path to the input SBV file.
|
| 102 |
-
output_dir (str): Path to the directory where translated
|
| 103 |
-
|
| 104 |
|
| 105 |
Returns:
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
Side effects:
|
| 109 |
-
- Creates the output directory if it doesn't exist.
|
| 110 |
-
- Writes translated SBV files to the output directory.
|
| 111 |
-
- Prints progress information to the console.
|
| 112 |
"""
|
| 113 |
lines = read_sbv_file(input_file)
|
| 114 |
base_filename = get_base_filename(input_file)
|
| 115 |
create_output_directory(output_dir)
|
| 116 |
blocks = parse_sbv_file(lines)
|
| 117 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 118 |
for lang in tqdm(selected_languages, desc="Translating languages"):
|
| 119 |
-
|
| 120 |
-
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
|
| 124 |
-
|
| 125 |
-
|
| 126 |
if __name__ == "__main__":
|
| 127 |
parser = argparse.ArgumentParser(description="Translate SBV subtitle files to multiple languages.")
|
| 128 |
parser.add_argument("input_file", help="Path to the input .sbv file")
|
|
@@ -130,4 +144,5 @@ if __name__ == "__main__":
|
|
| 130 |
parser.add_argument("--languages", nargs="+", choices=languages, default=languages, help="Languages to translate to")
|
| 131 |
args = parser.parse_args()
|
| 132 |
|
| 133 |
-
|
|
|
|
|
|
| 90 |
def get_output_file_path(output_dir, base_filename, lang):
|
| 91 |
return os.path.join(output_dir, f"{base_filename}_{lang}.sbv")
|
| 92 |
|
| 93 |
+
def translate_to_language(input_file, output_dir, language):
|
| 94 |
"""
|
| 95 |
+
Translate an SBV subtitle file to a single language.
|
|
|
|
|
|
|
|
|
|
| 96 |
|
| 97 |
Args:
|
| 98 |
input_file (str): Path to the input SBV file.
|
| 99 |
+
output_dir (str): Path to the directory where translated file will be saved.
|
| 100 |
+
language (str): Target language for translation.
|
| 101 |
|
| 102 |
Returns:
|
| 103 |
+
str: Path to the translated output file.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
"""
|
| 105 |
lines = read_sbv_file(input_file)
|
| 106 |
base_filename = get_base_filename(input_file)
|
| 107 |
create_output_directory(output_dir)
|
| 108 |
blocks = parse_sbv_file(lines)
|
| 109 |
|
| 110 |
+
print(f"\nStarting translation for language: {language}")
|
| 111 |
+
translated_blocks = translate_blocks(blocks, language)
|
| 112 |
+
output_content = reconstruct_sbv_content(translated_blocks)
|
| 113 |
+
output_file = get_output_file_path(output_dir, base_filename, language)
|
| 114 |
+
write_sbv_file(output_file, output_content)
|
| 115 |
+
print(f"Translated {language} file saved: {output_file}")
|
| 116 |
+
return output_file
|
| 117 |
+
|
| 118 |
+
def translate_to_languages(input_file, output_dir, selected_languages):
|
| 119 |
+
"""
|
| 120 |
+
Translate an SBV subtitle file to multiple languages.
|
| 121 |
+
|
| 122 |
+
Args:
|
| 123 |
+
input_file (str): Path to the input SBV file.
|
| 124 |
+
output_dir (str): Path to the directory where translated files will be saved.
|
| 125 |
+
selected_languages (list): List of target languages for translation.
|
| 126 |
+
|
| 127 |
+
Returns:
|
| 128 |
+
list: Paths to the translated output files.
|
| 129 |
+
"""
|
| 130 |
+
create_output_directory(output_dir)
|
| 131 |
+
translated_files = []
|
| 132 |
+
|
| 133 |
for lang in tqdm(selected_languages, desc="Translating languages"):
|
| 134 |
+
output_file = translate_to_language(input_file, output_dir, lang)
|
| 135 |
+
translated_files.append(output_file)
|
| 136 |
+
|
| 137 |
+
return translated_files
|
| 138 |
+
|
| 139 |
+
# Update the main block to use the new functions
|
|
|
|
| 140 |
if __name__ == "__main__":
|
| 141 |
parser = argparse.ArgumentParser(description="Translate SBV subtitle files to multiple languages.")
|
| 142 |
parser.add_argument("input_file", help="Path to the input .sbv file")
|
|
|
|
| 144 |
parser.add_argument("--languages", nargs="+", choices=languages, default=languages, help="Languages to translate to")
|
| 145 |
args = parser.parse_args()
|
| 146 |
|
| 147 |
+
translated_files = translate_to_languages(args.input_file, args.output_dir, args.languages)
|
| 148 |
+
print(f"\nTranslation completed. {len(translated_files)} files created.")
|