fix: prevent punctuaction to be added in above/below links (141a6058) · Commits · CIM - Context Information Management / NGSI-LD API

md_to_docx_converter/src/to_html/postprocessing.py

+28 −2

Original line number	Diff line number	Diff line
		@@ -496,6 +496,21 @@ def fix_custom_tags(soup: BeautifulSoup):
		image_extensions = [".png", ".jpg", ".jpeg", ".svg"]
		return not any(href.endswith(ext) for ext in image_extensions)

		def remove_trailing_punctuation(a: Tag, needle: str) -> str:
		if not a["href"].endswith(needle):
		# find index of last occurrence of needle
		index = a["href"].rfind(needle)
		if index != -1 and index + len(needle) < len(a["href"]):
		remaining_text = a["href"][index + len(needle):]
		new_href = a["href"][:index + len(needle)]
		# a["href"] = href
		string_index = a.string.rfind(needle)
		new_a_text = a.string[:string_index + len(needle)]
		# a.string = a.string[:string_index + len(needle)]
		a.insert_after(NavigableString(remaining_text))
		return new_href, new_a_text
		return None, None

		# Example: Change <custom-tag> to <div class="custom-tag">
		h1_tag = soup.find("h1", id=True)

		@@ -503,7 +518,13 @@ def fix_custom_tags(soup: BeautifulSoup):

		for a in a_tags:
		href = a.get("href", "")
		if href.endswith("below"):
		if href.find("+++below") != -1:
		new_href, new_a_text = remove_trailing_punctuation(a, "+++below")
		if new_href:
		href = new_href
		if new_a_text:
		a.string = new_a_text

		count_below = href.count("+++below")
		is_table = "Table" in href
		class_name = "TH" if is_table else "FL"
		@@ -528,7 +549,12 @@ def fix_custom_tags(soup: BeautifulSoup):
		)
		)
		os._exit(1)
		elif href.endswith("above"):
		elif href.find("+++above") != -1:
		new_href, new_a_text = remove_trailing_punctuation(a, "+++above")
		if new_href:
		href = new_href
		if new_a_text:
		a.string = new_a_text
		count_above = href.count("+++above")
		is_table = "Table" in href
		class_name = "TH" if is_table else "FL"