From cc09af522e371caeff4a414647ef1a1f2db36ce9 Mon Sep 17 00:00:00 2001
From: Charles Low <charles@larches.net>
Date: Mon, 4 Nov 2024 01:14:19 -0800
Subject: [PATCH] add pagebreak support

---
 htmldocx/h2d.py | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/htmldocx/h2d.py b/htmldocx/h2d.py
index 6a0e113..b7e7569 100644
--- a/htmldocx/h2d.py
+++ b/htmldocx/h2d.py
@@ -226,6 +226,7 @@ def add_styles_to_paragraph(self, style):
                 self.paragraph.paragraph_format.left_indent = Inches(min(margin // 10 * INDENT, MAX_INDENT))
             # TODO handle non px units
 
+
     def add_styles_to_run(self, style):
         if 'color' in style:
             if 'rgb' in style['color']:
@@ -399,6 +400,9 @@ def handle_link(self, href, text):
         # Add hyperlink to run
         self.paragraph._p.append(hyperlink)
 
+    def handle_pagebreak(self):
+        self.doc.add_page_break()
+
     def handle_starttag(self, tag, attrs):
         if self.skip:
             return
@@ -468,6 +472,13 @@ def handle_starttag(self, tag, attrs):
         elif tag == 'table':
             self.handle_table()
             return
+        # handle pagebreak using following html code
+        # <div style="page-break-after: always;"></div>
+        elif tag == 'div':
+            style = self.parse_dict_string(current_attrs['style'])
+            if(style.get('page-break-after', '') == 'always'):
+                self.handle_pagebreak()
+
 
         # set new run reference point in case of leading line breaks
         if tag in ['p', 'li', 'pre']: