Start work on converting character styles

src/calibre/ebooks/docx/writer/from_html.py

@@ -6,12 +6,121 @@ from __future__ import (unicode_literals, division, absolute_import,
 __license__ = 'GPL v3'
 __copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
 
+from lxml import etree
+
+from calibre.ebooks.docx.writer.utils import convert_color, int_or_zero
+from calibre.ebooks.oeb.stylizer import Stylizer
+from calibre.ebooks.oeb.base import XPath, barename
+
+
+class TextStyle(object):
+
+    def __init__(self, css):
+        # for x in ('text-transform', 'text-shadow', 'font-variant', 'letter-spacing', 'vertical-align'):
+        self.font_family = css['font-family']  # TODO: Resolve multiple font families and generic font family names
+        try:
+            self.font_size = int(float(css['font-size']) * 2)  # stylizer normalizes all font sizes into pts
+        except (ValueError, TypeError, AttributeError):
+            self.font_size = None
+
+        fw = self.font_weight = css['font-weight']
+        self.bold = fw in {'bold', 'bolder'} or int_or_zero(fw) >= 700
+        self.font_style = css['font-style']
+        self.italic = self.font_style in {'italic', 'oblique'}
+        self.color = convert_color(css['color'])
+        self.background_color = convert_color(css.backgroundColor)
+        td = set((css.effective_text_decoration or '').split())
+        self.underline = 'underline' in td
+        self.dstrike = 'line-through' in td and 'overline' in td
+        self.strike = not self.dstrike and 'line-through' in td
+
+        # TODO: Borders and padding
+
+    def __hash__(self):
+        return hash(tuple(
+            getattr(self, x) for x in ('font_family', 'font_size', 'bold', 'italic', 'color', 'background_color', 'underline', 'strike', 'dstrike')))
+
+    def __eq__(self, other):
+        return hash(self) == hash(other)
+
+class TextRun(object):
+
+    def __init__(self, style):
+        self.style = style
+
+class Block(object):
+
+    def __init__(self):
+        self.texts = []
+
+    def add_text(self, text, style):
+        pass
+
 class Convert(object):
 
     def __init__(self, oeb, docx):
         self.oeb, self.docx = oeb, docx
         self.log, self.opts = docx.log, docx.opts
 
-    def __call__(self):
+        self.blocks = []
-        pass
+
+    def __call__(self):
+        from calibre.ebooks.oeb.transforms.rasterize import SVGRasterizer
+        SVGRasterizer()(self.oeb, self.opts)
+
+        for item in self.oeb.spine:
+            self.process_item(item)
+
+    def process_item(self, item):
+        stylizer = Stylizer(item.data, item.href, self.oeb, self.opts, self.opts.output_profile)
+
+        for body in XPath('//h:body')(item.data):
+            b = Block()
+            self.blocks.append(b)
+            self.process_block(body, b, stylizer, ignore_tail=True)
+
+    def process_block(self, html_block, docx_block, stylizer, ignore_tail=False):
+        if html_block.text:
+            docx_block.add_text(html_block.text, stylizer.style(html_block))
+
+        for child in html_block.iterchildren(etree.Element):
+            tag = barename(child.tag)
+            style = stylizer.style(child)
+            display = style.get('display', 'inline')
+            if tag == 'img':
+                return  # TODO: Handle images
+            if display == 'block':
+                b = Block()
+                self.blocks.append(b)
+                self.process_block(child, b, stylizer)
+            else:
+                self.process_inline(child, self.blocks[-1], stylizer)
+
+        if ignore_tail is False and html_block.tail:
+            b = docx_block
+            if b is not self.blocks[-1]:
+                b = Block()
+                self.blocks.append(b)
+            b.add_text(html_block.tail, stylizer.style(html_block.getparent()))
+
+    def process_inline(self, html_child, docx_block, stylizer):
+        tag = barename(html_child.tag)
+        if tag == 'img':
+            return  # TODO: Handle images
+        style = stylizer.style(html_child)
+        if html_child.text:
+            docx_block.add_text(html_child.text, style)
+        for child in html_child.iterchildren(etree.Element):
+            style = stylizer.style(child)
+            display = style.get('display', 'inline')
+            if display == 'block':
+                b = Block()
+                self.blocks.append(b)
+                self.process_block(child, b, stylizer)
+            else:
+                self.process_inline(child, self.blocks[-1], stylizer)
+
+        if html_child.tail:
+            docx_block.add_text(html_child.tail, stylizer.style(html_child.getparent()))
+
 

src/calibre/ebooks/docx/writer/utils.py

@@ -0,0 +1,64 @@
+#!/usr/bin/env python
+# vim:fileencoding=utf-8
+from __future__ import (unicode_literals, division, absolute_import,
+                        print_function)
+
+__license__ = 'GPL v3'
+__copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
+
+import re
+from cssutils.css.colors import COLORS
+
+def int_or_zero(raw):
+    try:
+        return int(raw)
+    except (ValueError, TypeError, AttributeError):
+        return 0
+
+# convert_color() {{{
+hex_pat = re.compile(r'#([0-9a-f]{6})')
+hex3_pat = re.compile(r'#([0-9a-f]{3})')
+rgb_pat = re.compile(r'rgba?\s*\((\d+)\s*,\s*(\d+)\s*,\s*(\d+)')
+
+def convert_color(c):
+    if not c:
+        return None
+    c = c.lower().strip()
+    if c == 'transparent':
+        return None
+    try:
+        cval = COLORS[c]
+    except KeyError:
+        m = hex_pat.match(c)
+        if m is not None:
+            return c.upper()
+        m = hex3_pat.match(c)
+        if m is not None:
+            return '#' + (c[1]*2) + (c[2]*2) + (c[3]*2)
+        m = rgb_pat.match(c)
+        if m is not None:
+            return '#' + ''.join('%02X' % int(m.group(i)) for i in (1, 2, 3))
+    else:
+        return '#' + ''.join('%02X' % int(x) for x in cval[:3])
+    return None
+
+def test_convert_color():
+    import unittest
+    class TestColors(unittest.TestCase):
+
+        def test_color_conversion(self):
+            ae = self.assertEqual
+            cc = convert_color
+            ae(None, cc(None))
+            ae(None, cc('transparent'))
+            ae(None, cc('none'))
+            ae(None, cc('#12j456'))
+            ae('#F0F8FF', cc('AliceBlue'))
+            ae('#000000', cc('black'))
+            ae(cc('#001'), '#000011')
+            ae('#12345D', cc('#12345d'))
+            ae('#FFFFFF', cc('rgb(255, 255, 255)'))
+            ae('#FF0000', cc('rgba(255, 0, 0, 23)'))
+    tests = unittest.defaultTestLoader.loadTestsFromTestCase(TestColors)
+    unittest.TextTestRunner(verbosity=4).run(tests)
+# }}}

src/calibre/ebooks/oeb/stylizer.py

@@ -494,6 +494,9 @@ class Style(object):
             result = DEFAULTS[name]
         return result
 
+    def get(self, name, default=None):
+        return self._style.get(name, default)
+
     def _unit_convert(self, value, base=None, font=None):
         'Return value in pts'
         if base is None: