DOCX Input: Cascade the font css

src/calibre/ebooks/docx/block_styles.py

@@ -208,7 +208,7 @@ class ParagraphStyle(object):
 
         # Misc.
         'text_indent', 'text_align', 'line_height', 'direction', 'background_color',
-        'numbering',
+        'numbering', 'font_family', 'font_size',
     )
 
     def __init__(self, pPr=None):
@@ -232,6 +232,8 @@ 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._css = None
 
     def update(self, other):
@@ -274,10 +276,13 @@ 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'):
+            for x in ('text_indent', 'text_align', 'background_color', 'font_family', 'font_size'):
                 val = getattr(self, x)
                 if val is not inherit:
+                    if x == 'font_size':
+                        val = '%.3gpt' % val
                     c[x.replace('_', '-')] = val
+
         return self._css
 
         # TODO: keepNext must be done at markup level

src/calibre/ebooks/docx/char_styles.py

@@ -172,6 +172,18 @@ class RunStyle(object):
             if val is inherit:
                 setattr(self, p, getattr(parent, p))
 
+    def get_border_css(self, ans):
+        for x in ('color', 'style', 'width'):
+            val = getattr(self, 'border_'+x)
+            if x == 'width' and val is not inherit:
+                val = '%.3gpt' % val
+            if val is not inherit:
+                ans['border-%s' % x] = val
+
+    def clear_border_css(self):
+        for x in ('color', 'style', 'width'):
+            setattr(self, 'border_'+x, inherit)
+
     @property
     def css(self):
         if self._css is None:
@@ -196,12 +208,7 @@ class RunStyle(object):
             if self.vanish is True:
                 c['display'] = 'none'
 
-            for x in ('color', 'style', 'width'):
-                val = getattr(self, 'border_'+x)
-                if x == 'width' and val is not inherit:
-                    val = '%.3gpt' % val
-                if val is not inherit:
-                    c['border-%s' % x] = val
+            self.get_border_css(c)
             if self.padding is not inherit:
                 c['padding'] = '%.3gpt' % self.padding
 
@@ -223,6 +230,7 @@ class RunStyle(object):
 
             if self.font_family is not inherit:
                 c['font-family'] = self.font_family
+
         return self._css
 
     def same_border(self, other):

src/calibre/ebooks/docx/styles.py

@@ -258,6 +258,55 @@ class Styles(object):
         if obj.tag.endswith('}r'):
             return self.resolve_run(obj)
 
+    def cascade(self, layers):
+        self.body_font_family = 'serif'
+        self.body_font_size = '10pt'
+
+        for p, runs in layers.iteritems():
+            char_styles = [self.resolve_run(r) for r in runs]
+            block_style = self.resolve_paragraph(p)
+            c = Counter()
+            for s in char_styles:
+                if s.font_family is not inherit:
+                    c[s.font_family] += 1
+            if c:
+                family = c.most_common(1)[0][0]
+                block_style.font_family = family
+                for s in char_styles:
+                    if s.font_family == family:
+                        s.font_family = inherit
+
+            sizes = [s.font_size for s in char_styles if s.font_size is not inherit]
+            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]
+        c = Counter()
+        for s in block_styles:
+            if s.font_family is not inherit:
+                c[s.font_family] += 1
+
+        if c:
+            self.body_font_family = family = c.most_common(1)[0][0]
+            for s in block_styles:
+                if s.font_family == family:
+                    s.font_family = inherit
+
+        c = Counter()
+        for s in block_styles:
+            if s.font_size is not inherit:
+                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
         # must be discarder and pStyle used instead.
@@ -298,12 +347,14 @@ class Styles(object):
         ef = self.fonts.embed_fonts(dest_dir, docx)
         prefix = textwrap.dedent(
             '''\
+            body { font-family: %s; font-size: %s }
+
             p { text-indent: 1.5em }
 
             ul, ol, p { margin: 0; padding: 0 }
-            ''')
+            ''') % (self.body_font_family, self.body_font_size)
         if ef:
-            prefix += '\n' + ef
+            prefix = ef + '\n' + prefix
 
         ans = []
         for (cls, css) in sorted(self.classes.itervalues(), key=lambda x:x[0]):

src/calibre/ebooks/docx/to_html.py

@@ -64,12 +64,15 @@ class Convert(object):
         doc = self.docx.document
         relationships_by_id, relationships_by_type = self.docx.document_relationships
         self.read_styles(relationships_by_type)
+        self.layers = OrderedDict()
         for wp in XPath('//w:p')(doc):
             p = self.convert_p(wp)
             self.body.append(p)
         # TODO: tables <w:tbl> child of <w:body> (nested tables?)
         # TODO: Last section properties <w:sectPr> child of <w:body>
 
+        self.styles.cascade(self.layers)
+
         numbered = []
         for html_obj, obj in self.object_map.iteritems():
             raw = obj.get('calibre_num_id', None)
@@ -156,9 +159,11 @@ class Convert(object):
         dest = P()
         self.object_map[dest] = p
         style = self.styles.resolve_paragraph(p)
+        self.layers[p] = []
         for run in XPath('descendant::w:r')(p):
             span = self.convert_run(run)
             dest.append(span)
+            self.layers[p].append(run)
 
         m = re.match(r'heading\s+(\d+)$', style.style_name or '', re.IGNORECASE)
         if m is not None:
@@ -184,12 +189,9 @@ class Convert(object):
             spans = []
             bs = {}
             for span, style in border_run:
-                c = style.css
+                style.get_border_css(bs)
+                style.clear_border_css()
                 spans.append(span)
-                for x in ('width', 'color', 'style'):
-                    val = c.pop('border-%s' % x, None)
-                    if val is not None:
-                        bs['border-%s' % x] = val
             if bs:
                 cls = self.styles.register(bs, 'text_border')
                 wrapper = self.wrap_elems(spans, SPAN())