support import translation from TXT/markdown #639 (a4c50344) · Commits · git-mirror / BallonsTranslator

ui/config_proj.py

+67 −0

Original line number	Diff line number	Diff line
		@@ -26,6 +26,39 @@ def read_jpg_metadata(imgpath: str):
		bubdict = json.loads(user_comment)
		return bubdict

		page_start_pattern = re.compile(r'^###\s+', re.MULTILINE)
		text_blkid_start_pattern = re.compile(r'^\d+\.', re.MULTILINE)

		def parse_txt_translation(file_path: str):
		with open(file_path, 'r', encoding='utf8') as f:
		content = f.read()
		page_start = None
		page_list = []
		for matched in page_start_pattern.finditer(content):
		start, end = matched.span()
		if page_start is not None:
		page_list.append({'page_content': content[page_start: start]})
		page_start = start
		if page_start is not None:
		page_list.append({'page_content': content[page_start:]})

		for page_dict in page_list:
		page_content = page_dict['page_content']
		page_dict['page_name'] = page_start_pattern.sub('', page_content.split('\n')[0]).strip()
		blkid_start = blkid_end = None
		blk_list = []
		for matched in text_blkid_start_pattern.finditer(page_content):
		start, end = matched.span()
		if blkid_start is not None:
		blk_list.append(page_content[blkid_end: start].strip())
		blkid_start = start
		blkid_end = end
		if blkid_start is not None:
		blk_list.append(page_content[blkid_end:])
		page_dict['blk_list'] = blk_list

		return page_list


		class TextBlkEncoder(NumpyEncoder):
		def default(self, obj):
		@@ -139,6 +172,40 @@ class ProjImgTrans:
		if len(self.pages) > 0:
		self.set_current_img_byidx(0)

		def load_translation_from_txt(self, file_path: str):
		page_list = parse_txt_translation(file_path)
		missing_pages = []
		unmatched_pages = []
		unexpected_pages = []
		matched_pages = []
		for page_dict in page_list:
		page_name = page_dict['page_name']
		if page_name in self.pages:
		matched_pages.append(page_name)
		else:
		unexpected_pages.append(page_name)
		continue
		blklist = self.pages[page_name]
		n_blk = len(blklist)
		src_blk_list = page_dict['blk_list']
		n_src_blk = len(src_blk_list)
		if n_src_blk != n_blk:
		LOGGER.warning(f'Unmatched text blocks in {page_name}, number of text blocks in this page vs source file: {n_blk}-{n_src_blk}')
		unmatched_pages.append(page_name)
		for blkid in range(min(n_blk, n_src_blk)):
		blk = blklist[blkid]
		blk.rich_text = ''
		blk.translation = src_blk_list[blkid]

		matched_pages = set(matched_pages)
		if len(matched_pages) != self.num_pages:
		for page_name in self.pages:
		if page_name not in matched_pages:
		missing_pages.append(page_name)

		all_matched = len(missing_pages) == 0 and len(unmatched_pages) == 0 and len(unexpected_pages) == 0
		return all_matched, {'missing_pages': missing_pages, 'unmatched_pages': unmatched_pages, 'unexpected_pages': unexpected_pages, 'matched_pages': matched_pages}

		def load_from_json(self, json_path: str):
		old_dir = self.directory
		directory = osp.dirname(json_path)

ui/mainwindow.py

+37 −3

Original line number	Diff line number	Diff line
		@@ -14,7 +14,7 @@ from utils.logger import logger as LOGGER
		from utils.text_processing import is_cjk, full_len, half_len
		from utils.textblock import TextBlock, TextAlignment
		from utils import shared
		from utils import create_error_dialog
		from utils import create_error_dialog, create_info_dialog
		from modules.translators.trans_chatgpt import GPTTranslator
		from .misc import parse_stylesheet, set_html_family, QKEY
		from utils.config import ProgramConfig, pcfg, save_config, text_styles, save_text_styles, load_textstyle_from, FontFormat
		@@ -139,6 +139,7 @@ class MainWindow(mainwindow_cls):
		self.leftBar.export_trans_txt.connect(lambda : self.on_export_txt(dump_target='translation'))
		self.leftBar.export_src_md.connect(lambda : self.on_export_txt(dump_target='source', suffix='.md'))
		self.leftBar.export_trans_md.connect(lambda : self.on_export_txt(dump_target='translation', suffix='.md'))
		self.leftBar.import_trans_txt.connect(self.on_import_trans_txt)

		self.pageList = PageListView()
		self.pageList.reveal_file.connect(self.on_reveal_file)
		@@ -476,7 +477,6 @@ class MainWindow(mainwindow_cls):
		self.imgtrans_proj.set_current_img(item.text())
		self.canvas.clear_undostack(update_saved_step=True)
		self.canvas.updateCanvas()
		self.st_manager.hovering_transwidget = None
		self.st_manager.updateSceneTextitems()
		self.titleBar.setTitleContent(page_name=self.imgtrans_proj.current_img)
		self.module_manager.handle_page_changed()
		@@ -1186,7 +1186,41 @@ class MainWindow(mainwindow_cls):
		msg.setText(self.tr('Text file exported to ') + self.imgtrans_proj.dump_txt_path(dump_target, suffix))
		msg.exec_()
		except Exception as e:
		create_error_dialog(e, self.tr('failed to export as TEXT file'))
		create_error_dialog(e, self.tr('Failed to export as TEXT file'))

		def on_import_trans_txt(self):
		try:
		selected_file = ''
		dialog = QFileDialog()
		selected_file = str(dialog.getOpenFileUrl(self.parent(), self.tr('Import .md/.txt'), filter=".txt .md .TXT .MD")[0].toLocalFile())
		if not osp.exists(selected_file):
		return

		all_matched, match_rst = self.imgtrans_proj.load_translation_from_txt(selected_file)
		matched_pages = match_rst['matched_pages']

		if self.imgtrans_proj.current_img in matched_pages:
		self.canvas.clear_undostack(update_saved_step=True)
		self.st_manager.updateSceneTextitems()

		if all_matched:
		msg = self.tr('Translation imported and matched successfully.')
		else:
		msg = self.tr('Imported txt file not fully matched with current project, please make sure source txt file structured like results from \"export TXT/markdown\"')
		if len(match_rst['missing_pages']) > 0:
		msg += '\n' + self.tr('Missing pages: ') + '\n'
		msg += '\n'.join(match_rst['missing_pages'])
		if len(match_rst['unexpected_pages']) > 0:
		msg += '\n' + self.tr('Unexpected pages: ') + '\n'
		msg += '\n'.join(match_rst['unexpected_pages'])
		if len(match_rst['unmatched_pages']) > 0:
		msg += '\n' + self.tr('Unmatched pages: ') + '\n'
		msg += '\n'.join(match_rst['unmatched_pages'])
		msg = msg.strip()
		create_info_dialog(msg)

		except Exception as e:
		create_error_dialog(e, self.tr('Failed to import translation from ') + selected_file)

		def on_reveal_file(self):
		current_img_path = self.imgtrans_proj.current_img_path()

ui/mainwindowbars.py

+4 −0

Original line number	Diff line number	Diff line
		@@ -154,6 +154,9 @@ class LeftBar(Widget):
		actionExportTranslationMD = QAction(self.tr("Export translation as markdown"), self)
		self.export_trans_md = actionExportTranslationMD.triggered

		actionImportTranslationTxt = QAction(self.tr("Import translation from TXT/markdown"), self)
		self.import_trans_txt = actionImportTranslationTxt.triggered

		self.recentMenu = QMenu(self.tr("Open Recent"), self)

		openMenu = QMenu(self)
		@@ -168,6 +171,7 @@ class LeftBar(Widget):
		actionExportTranslationTxt,
		actionExportSrcMD,
		actionExportTranslationMD,
		actionImportTranslationTxt,
		])
		self.openBtn = OpenBtn()
		self.openBtn.setFixedSize(LEFTBTN_WIDTH, LEFTBTN_WIDTH)

ui/scenetext_manager.py

+1 −0

Original line number	Diff line number	Diff line
		@@ -440,6 +440,7 @@ class SceneTextManager(QObject):
		self.pairwidget_list.clear()

		def updateSceneTextitems(self):
		self.hovering_transwidget = None
		self.txtblkShapeControl.setBlkItem(None)
		self.clearSceneTextitems()
		for textblock in self.imgtrans_proj.current_block_list():