DOCX: Improve style cascading

1) Add cascading for colors
    2) Remove unnecessary text-decoration:none entries
    3) Only cascade font properties if they are the same for all runs in a
    block.
    4) Set sensible defaults for cascaded properties that are never
    specified.

src/calibre/ebooks/docx/block_styles.py

@@ -300,7 +300,7 @@ class ParagraphStyle(object):
 
         # Misc.
         'text_indent', 'text_align', 'line_height', 'direction', 'background_color',
-        'numbering', 'font_family', 'font_size', 'frame',
+        'numbering', 'font_family', 'font_size', 'color', 'frame',
     )
 
     def __init__(self, pPr=None):
@@ -324,7 +324,7 @@ class ParagraphStyle(object):
             for s in XPath('./w:pStyle[@w:val]')(pPr):
                 self.linked_style = get(s, 'w:val')
 
-            self.font_family = self.font_size = inherit
+            self.font_family = self.font_size = self.color = inherit
 
         self._css = None
 
@@ -368,7 +368,7 @@ class ParagraphStyle(object):
             if self.line_height not in {inherit, '1'}:
                 c['line-height'] = self.line_height
 
-            for x in ('text_indent', 'text_align', 'background_color', 'font_family', 'font_size'):
+            for x in ('text_indent', 'text_align', 'background_color', 'font_family', 'font_size', 'color'):
                 val = getattr(self, x)
                 if val is not inherit:
                     if x == 'font_size':

src/calibre/ebooks/docx/styles.py

@@ -317,51 +317,63 @@ class Styles(object):
     def cascade(self, layers):
         self.body_font_family = 'serif'
         self.body_font_size = '10pt'
+        self.body_color = 'black'
+
+        def promote_property(char_styles, block_style, prop):
+            vals = {getattr(s, prop) for s in char_styles}
+            if len(vals) == 1:
+                # All the character styles have the same value
+                for s in char_styles:
+                    setattr(s, prop, inherit)
+                setattr(block_style, prop, next(iter(vals)))
 
         for p, runs in layers.iteritems():
+            has_links = '1' in {r.get('is-link', None) for r in runs}
             char_styles = [self.resolve_run(r) for r in runs]
             block_style = self.resolve_paragraph(p)
-            c = Counter()
+            for prop in ('font_family', 'font_size', 'color'):
+                if has_links and prop == 'color':
+                    # We cannot promote color as browser rendering engines will
+                    # override the link color setting it to blue, unless the
+                    # color is specified on the link element itself
+                    continue
+                promote_property(char_styles, block_style, prop)
             for s in char_styles:
-                if s.font_family is not inherit:
+                if s.text_decoration == 'none':
-                    c[s.font_family] += 1
+                    # The default text decoration is 'none'
+                    s.text_decoration = inherit
+
+        def promote_most_common(block_styles, prop, default):
+            c = Counter()
+            for s in block_styles:
+                val = getattr(s, prop)
+                if val is not inherit:
+                    c[val] += 1
+            val = None
             if c:
-                family = c.most_common(1)[0][0]
+                val = c.most_common(1)[0][0]
-                block_style.font_family = family
+                for s in block_styles:
-                for s in char_styles:
+                    oval = getattr(s, prop)
-                    if s.font_family == family:
+                    if oval is inherit:
-                        s.font_family = inherit
+                        if default != val:
+                            setattr(s, prop, default)
+                    elif oval == val:
+                        setattr(s, prop, inherit)
+            return val
 
-            sizes = [s.font_size for s in char_styles if s.font_size is not inherit]
+        block_styles = tuple(self.resolve_paragraph(p) for p in layers)
-            if sizes:
-                sz = block_style.font_size = sizes[0]
-                for s in char_styles:
-                    if s.font_size == sz:
-                        s.font_size = inherit
 
-        block_styles = [self.resolve_paragraph(p) for p in layers]
+        ff = promote_most_common(block_styles, 'font_family', self.body_font_family)
-        c = Counter()
+        if ff is not None:
-        for s in block_styles:
+            self.body_font_family = ff
-            if s.font_family is not inherit:
-                c[s.font_family] += 1
 
-        if c:
+        fs = promote_most_common(block_styles, 'font_size', int(self.body_font_size[:2]))
-            self.body_font_family = family = c.most_common(1)[0][0]
+        if fs is not None:
-            for s in block_styles:
+            self.body_font_size = '%.3gpt' % fs
-                if s.font_family == family:
-                    s.font_family = inherit
 
-        c = Counter()
+        color = promote_most_common(block_styles, 'color', self.body_color)
-        for s in block_styles:
+        if color is not None:
-            if s.font_size is not inherit:
+            self.body_color = color
-                c[s.font_size] += 1
-
-        if c:
-            sz = c.most_common(1)[0][0]
-            for s in block_styles:
-                if s.font_size == sz:
-                    s.font_size = inherit
-            self.body_font_size = '%.3gpt' % sz
 
     def resolve_numbering(self, numbering):
         # When a numPr element appears inside a paragraph style, the lvl info
@@ -403,7 +415,7 @@ class Styles(object):
         ef = self.fonts.embed_fonts(dest_dir, docx)
         prefix = textwrap.dedent(
             '''\
-            body { font-family: %s; font-size: %s }
+            body { font-family: %s; font-size: %s; color: %s }
 
             ul, ol, p { margin: 0; padding: 0 }
 
@@ -419,7 +431,7 @@ class Styles(object):
 
             dl.notes dd:last-of-type { page-break-after: avoid }
 
-            ''') % (self.body_font_family, self.body_font_size)
+            ''') % (self.body_font_family, self.body_font_size, self.body_color)
         if ef:
             prefix = ef + '\n' + prefix
 
@@ -430,3 +442,4 @@ class Styles(object):
             ans.append('.%s {\n%s\n}\n' % (cls, b.rstrip(';')))
         return prefix + '\n' + '\n'.join(ans)
 
+

src/calibre/ebooks/docx/to_html.py

@@ -325,6 +325,7 @@ class Convert(object):
                     try:
                         hl = hl_xpath(x)[0]
                         self.link_map[hl].append(span)
+                        x.set('is-link', '1')
                     except IndexError:
                         current_hyperlink = None
                 dest.append(span)