refactor: Replace html2text with markdownify

frappe/core/utils.py

@@ -1,6 +1,8 @@
 # Copyright (c) 2015, Frappe Technologies Pvt. Ltd. and Contributors
 # License: MIT. See LICENSE
 
+from markdownify import markdownify as md
+
 import frappe
 
 
@@ -86,3 +88,8 @@ def ljust_list(_list, length, fill_word=None):
 		_list.extend([fill_word] * fill_length)
 
 	return _list
+
+
+def html2text(html, strip_links=False, wrap=True):
+	strip = ["a"] if strip_links else None
+	return md(html, heading_style="ATX", strip=strip, wrap=wrap)

frappe/desk/form/document_follow.py

@@ -183,7 +183,7 @@ def get_version(doctype, doc_name, frequency, user):
 
 
 def get_comments(doctype, doc_name, frequency, user):
-	from html2text import html2text
+	from frappe.core.utils import html2text
 
 	timeline = []
 	filters = get_filters("reference_name", doc_name, frequency, user)
@@ -225,7 +225,7 @@ def get_follow_users(doctype, doc_name):
 
 
 def get_row_changed(row_changed, time, doctype, doc_name, v):
-	from html2text import html2text
+	from frappe.core.utils import html2text
 
 	items = []
 	for d in row_changed:
@@ -269,7 +269,7 @@ def get_added_row(added, time, doctype, doc_name, v):
 
 
 def get_field_changed(changed, time, doctype, doc_name, v):
-	from html2text import html2text
+	from frappe.core.utils import html2text
 
 	items = []
 	for d in changed:

frappe/email/doctype/email_queue/email_queue.py

@@ -8,11 +8,11 @@ import traceback
 from email.parser import Parser
 from email.policy import SMTPUTF8
 
-from html2text import html2text
 from rq.timeouts import JobTimeoutException
 
 import frappe
 from frappe import _, safe_encode, task
+from frappe.core.utils import html2text
 from frappe.email.doctype.email_account.email_account import EmailAccount
 from frappe.email.email_body import add_attachment, get_email, get_formatted_html
 from frappe.email.queue import get_unsubcribed_url, get_unsubscribe_message

frappe/integrations/frappe_providers/frappecloud.py

@@ -1,8 +1,8 @@
 import click
 import requests
-from html2text import html2text
 
 import frappe
+from frappe.core.utils import html2text
 
 
 def frappecloud_migrator(local_site):

frappe/utils/data.py

@@ -1911,7 +1911,7 @@ def get_string_between(start: str, string: str, end: str) -> str:
 def to_markdown(html: str) -> str:
 	from html.parser import HTMLParser
 
-	from html2text import html2text
+	from frappe.core.utils import html2text
 
 	try:
 		return html2text(html or "")

frappe/utils/safe_exec.py

@@ -4,7 +4,6 @@ import json
 import mimetypes
 
 import RestrictedPython.Guards
-from html2text import html2text
 from RestrictedPython import compile_restricted, safe_globals
 
 import frappe
@@ -13,6 +12,7 @@ import frappe.integrations.utils
 import frappe.utils
 import frappe.utils.data
 from frappe import _
+from frappe.core.utils import html2text
 from frappe.frappeclient import FrappeClient
 from frappe.handler import execute_cmd
 from frappe.model.delete_doc import delete_doc

frappe/utils/xlsxutils.py

@@ -52,7 +52,7 @@ def make_xlsx(data, sheet_name, wb=None, column_widths=None):
 
 
 def handle_html(data):
-	from html2text import HTML2Text
+	from frappe.core.utils import html2text
 
 	# return if no html tags found
 	data = frappe.as_unicode(data)
@@ -62,12 +62,8 @@ def handle_html(data):
 
 	h = unescape_html(data or "")
 
-	obj = HTML2Text()
-	obj.ignore_links = True
-	obj.body_width = 0
-
 	try:
-		value = obj.handle(h)
+		value = html2text(h, strip_links=True, wrap=False)
 	except Exception:
 		# unable to parse html, send it raw
 		return data

frappe/www/search.py

@@ -1,8 +1,8 @@
-from html2text import html2text
 from jinja2 import utils
 
 import frappe
 from frappe import _
+from frappe.core.utils import html2text
 from frappe.utils import sanitize_html
 from frappe.utils.global_search import web_search
 

pyproject.toml

@@ -35,7 +35,6 @@ dependencies = [
     "git-url-parse~=1.2.2",
     "gitdb~=4.0.7",
     "gunicorn~=20.1.0",
-    "html2text==2020.1.16",
     "html5lib~=1.1",
     "ipython~=8.4.0",
     "ldap3~=2.9",
@@ -73,6 +72,7 @@ dependencies = [
     "urllib3~=1.26.4",
     "xlrd~=2.0.1",
     "zxcvbn-python~=4.4.24",
+    "markdownify~=0.11.2",
 
     # integration dependencies
     "boto3~=1.17.53",