DOCX Output: Implement embedding of fonts

src/calibre/ebooks/conversion/plumber.py

@@ -63,6 +63,7 @@ class CompositeProgressReporter(object):
 ARCHIVE_FMTS = ('zip', 'rar', 'oebzip')
 
 class Plumber(object):
+
     '''
     The `Plumber` manages the conversion pipeline. An UI should call the methods
     :method:`merge_ui_recommendations` and then :method:`run`. The plumber will
@@ -202,7 +203,7 @@ OptionRecommendation(name='embed_font_family',
             'specifies its own fonts, they may override this base font. '
             'You can use the filter style information option to remove fonts from the '
             'input document. Note that font embedding only works '
-            'with some output formats, principally EPUB and AZW3.')
+            'with some output formats, principally EPUB, AZW3 and DOCX.')
         ),
 
 OptionRecommendation(name='embed_all_fonts',
@@ -212,7 +213,7 @@ OptionRecommendation(name='embed_all_fonts',
             'but not already embedded. This will search your system for the '
             'fonts, and if found, they will be embedded. Embedding will only work '
             'if the format you are converting to supports embedded fonts, such as '
-            'EPUB, AZW3 or PDF. Please ensure that you have the proper license for embedding '
+            'EPUB, AZW3, DOCX or PDF. Please ensure that you have the proper license for embedding '
             'the fonts used in this document.'
         )),
 
@@ -1142,8 +1143,7 @@ OptionRecommendation(name='search_replace',
 
         mobi_file_type = getattr(self.opts, 'mobi_file_type', 'old')
         needs_old_markup = (self.output_plugin.file_type == 'lit' or
-                    (self.output_plugin.file_type == 'mobi' and mobi_file_type
+                    (self.output_plugin.file_type == 'mobi' and mobi_file_type == 'old'))
-                     == 'old'))
         flattener = CSSFlattener(fbase=fbase, fkey=fkey,
                 lineh=line_height,
                 untable=needs_old_markup,
@@ -1233,4 +1233,3 @@ def create_oebbook(log, path_or_stream, opts, reader=None,
 
     reader()(oeb, path_or_stream)
     return oeb
-

src/calibre/ebooks/docx/names.py

@@ -18,6 +18,7 @@ APPPROPS  = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships
 STYLES    = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/styles'
 NUMBERING = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/numbering'
 FONTS     = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/fontTable'
+EMBEDDED_FONT = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/font'
 IMAGES    = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/image'
 LINKS     = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink'
 FOOTNOTES = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/footnotes'

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

@@ -91,6 +91,7 @@ class DOCX(object):
         self.font_table = etree.Element('{%s}fonts' % namespaces['w'], nsmap={k:namespaces[k] for k in 'wr'})
         E = ElementMaker(namespace=namespaces['pr'], nsmap={None:namespaces['pr']})
         self.embedded_fonts = E.Relationships()
+        self.fonts = {}
 
     # Boilerplate {{{
     @property
@@ -192,6 +193,8 @@ class DOCX(object):
             zf.writestr('word/_rels/fontTable.xml.rels', xml2str(self.embedded_fonts))
             for fname, data_getter in self.images.iteritems():
                 zf.writestr(fname, data_getter())
+            for fname, data in self.fonts.iteritems():
+                zf.writestr(fname, data)
 
 if __name__ == '__main__':
     d = DOCX(None, None)

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

@@ -6,14 +6,25 @@ from __future__ import (unicode_literals, division, absolute_import,
 __license__ = 'GPL v3'
 __copyright__ = '2015, Kovid Goyal <kovid at kovidgoyal.net>'
 
-from calibre.ebooks.docx.names import makeelement
+from collections import defaultdict
+from uuid import uuid4
+
+from calibre.ebooks.docx.names import makeelement, EMBEDDED_FONT
+from calibre.ebooks.oeb.base import OEB_STYLES
+from calibre.ebooks.oeb.transforms.subset import find_font_face_rules
+
+def obfuscate_font_data(data, key):
+    prefix = bytearray(data[:32])
+    key = bytearray(reversed(key.bytes))
+    prefix = bytes(bytearray(prefix[i]^key[i % len(key)] for i in xrange(len(prefix))))
+    return prefix + data[32:]
 
 class FontsManager(object):
 
-    def __init__(self, oeb):
+    def __init__(self, oeb, opts):
-        self.oeb, self.log = oeb, oeb.log
+        self.oeb, self.log, self.opts = oeb, oeb.log, opts
 
-    def serialize(self, text_styles, fonts, embed_relationships):
+    def serialize(self, text_styles, fonts, embed_relationships, font_data_map):
         font_families, seen = set(), set()
         for ts in text_styles:
             if ts.font_family:
@@ -21,6 +32,44 @@ class FontsManager(object):
                 if lf not in seen:
                     seen.add(lf)
                     font_families.add(ts.font_family)
+        family_map = {}
         for family in sorted(font_families):
-            makeelement(fonts, 'w:font', w_name=family)
+            family_map[family] = makeelement(fonts, 'w:font', w_name=family)
 
+        embedded_fonts = []
+        for item in self.oeb.manifest:
+            if item.media_type in OEB_STYLES and hasattr(item.data, 'cssRules'):
+                embedded_fonts.extend(find_font_face_rules(item, self.oeb))
+
+        num = 0
+        face_map = defaultdict(set)
+        rel_map = {}
+        for ef in embedded_fonts:
+            ff = ef['font-family'][0]
+            if ff not in font_families:
+                continue
+            num += 1
+            bold = ef['weight'] > 400
+            italic = ef['font-style'] != 'normal'
+            tag = 'Regular'
+            if bold or italic:
+                tag = 'Italic'
+                if bold and italic:
+                    tag = 'BoldItalic'
+                elif bold:
+                    tag = 'Bold'
+            if tag in face_map[ff]:
+                continue
+            face_map[ff].add(tag)
+            font = family_map[ff]
+            key = uuid4()
+            item = ef['item']
+            rid = rel_map.get(item)
+            if rid is None:
+                rel_map[item] = rid = 'rId%d' % num
+                fname = 'fonts/font%d.odttf' % num
+                makeelement(embed_relationships, 'Relationship', Id=rid, Type=EMBEDDED_FONT, Target=fname)
+                font_data_map['word/' + fname] = obfuscate_font_data(item.data, key)
+            makeelement(font, 'w:embed' + tag, r_id=rid,
+                        w_fontKey='{%s}' % key.urn.rpartition(':')[-1].upper(),
+                        w_subsetted="true" if self.opts.subset_embedded_fonts else "false")

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

@@ -171,7 +171,7 @@ class Convert(object):
 
         self.styles_manager = StylesManager()
         self.images_manager = ImagesManager(self.oeb, self.docx.document_relationships)
-        self.fonts_manager = FontsManager(self.oeb)
+        self.fonts_manager = FontsManager(self.oeb, self.opts)
 
         for item in self.oeb.spine:
             self.process_item(item)
@@ -298,4 +298,4 @@ class Convert(object):
         self.docx.images = {}
         self.styles_manager.serialize(self.docx.styles)
         self.images_manager.serialize(self.docx.images)
-        self.fonts_manager.serialize(self.styles_manager.text_styles, self.docx.font_table, self.docx.embedded_fonts)
+        self.fonts_manager.serialize(self.styles_manager.text_styles, self.docx.font_table, self.docx.embedded_fonts, self.docx.fonts)