DOCX Input: Support horizontal rules

DOCX Input: Add support for horizontal rules created by typing three
hyphens and pressing enter.

src/calibre/ebooks/docx/cleanup.py

@@ -8,6 +8,8 @@
 
 import os
 
+from calibre.ebooks.docx.names import ancestor
+
 def mergeable(previous, current):
     if previous.tail or current.tail:
         return False
@@ -97,6 +99,16 @@ def before_count(root, tag, limit=10):
             return limit
 
 def cleanup_markup(log, root, styles, dest_dir, detect_cover):
+    # Move <hr>s outside paragraphs, if possible.
+    for hr in root.xpath('//span/hr'):
+        p = ancestor(hr, 'p')
+        descendants = tuple(p.iterdescendants())
+        if descendants[-1] is hr:
+            parent = p.getparent()
+            idx = parent.index(p)
+            parent.insert(idx+1, hr)
+            hr.tail = '\n\t'
+
     # Merge consecutive spans that have the same styling
     current_run = []
     for span in root.xpath('//span'):
@@ -165,3 +177,4 @@ def cleanup_markup(log, root, styles, dest_dir, detect_cover):
                     return path
 
 
+

src/calibre/ebooks/docx/images.py

@@ -8,7 +8,7 @@
 
 import os
 
-from lxml.html.builder import IMG
+from lxml.html.builder import IMG, HR
 
 from calibre.ebooks.docx.names import XPath, get, barename
 from calibre.utils.filenames import ascii_filename
@@ -163,6 +163,26 @@ def drawing_to_html(self, drawing, page):
                     yield ans
 
     def pict_to_html(self, pict, page):
+        # First see if we have an <hr>
+        is_hr = len(pict) == 1 and get(pict[0], 'o:hr') in {'t', 'true'}
+        if is_hr:
+            style = {}
+            hr = HR()
+            try:
+                pct = float(get(pict[0], 'o:hrpct'))
+            except (ValueError, TypeError, AttributeError):
+                pass
+            else:
+                if pct > 0:
+                    style['width'] = '%.3g%%' % pct
+            align = get(pict[0], 'o:hralign', 'center')
+            if align in {'left', 'right'}:
+                style['margin-left'] = '0' if align == 'left' else 'auto'
+                style['margin-right'] = 'auto' if align == 'left' else '0'
+            if style:
+                hr.set('style', '; '.join(('%s:%s' % (k, v) for k, v in style.iteritems())))
+            yield hr
+
         for imagedata in XPath('descendant::v:imagedata[@r:id]')(pict):
             rid = get(imagedata, 'r:id')
             if rid in self.rid_map: