DOCX Output: Insert book cover as full page image at the start of the document

src/calibre/ebooks/conversion/plugins/docx_output.py

@@ -28,6 +28,10 @@ class DOCXOutput(OutputFormatPlugin):
             'EG. `123x321` to specify the width and height (in pts). '
             'This overrides any specified page-size.')),
 
+        OptionRecommendation(name='docx_no_cover', recommended_value=False,
+            help=_('Do not insert the book cover as an image at the start of the document.'
+                   ' If you use this option, the book cover will be discarded.')),
+
         OptionRecommendation(name='extract_to',
             help=_('Extract the contents of the generated %s file to the '
                 'specified directory. The contents of the directory are first '
@@ -55,7 +59,7 @@ class DOCXOutput(OutputFormatPlugin):
         from calibre.ebooks.docx.writer.from_html import Convert
         docx = DOCX(opts, log)
         self.convert_metadata(oeb)
-        Convert(oeb, docx, self.mi)()
+        Convert(oeb, docx, self.mi, not opts.docx_no_cover)()
         docx.write(output_path, self.mi)
         if opts.extract_to:
             from calibre.ebooks.docx.dump import do_dump

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

@@ -27,6 +27,12 @@ def xml2str(root, pretty_print=False, with_tail=False):
                           pretty_print=pretty_print, with_tail=with_tail)
     return ans
 
+def page_size(opts):
+    width, height = PAPER_SIZES[opts.docx_page_size]
+    if opts.docx_custom_page_size is not None:
+        width, height = map(float, opts.docx_custom_page_size.partition('x')[0::2])
+    return width, height
+
 def create_skeleton(opts, namespaces=None):
     namespaces = namespaces or DOCXNamespace().namespaces
     def w(x):
@@ -36,9 +42,7 @@ def create_skeleton(opts, namespaces=None):
     doc = E.document()
     body = E.body()
     doc.append(body)
-    width, height = PAPER_SIZES[opts.docx_page_size]
+    width, height = page_size(opts)
-    if opts.docx_custom_page_size is not None:
-        width, height = map(float, opts.docx_custom_page_size.partition('x')[0::2])
     width, height = int(20 * width), int(20 * height)
     def margin(which):
         return w(which), str(int(getattr(opts, 'margin_'+which) * 20))

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

@@ -9,7 +9,7 @@ __copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
 import re
 from collections import Counter
 
-from calibre.ebooks.docx.writer.container import create_skeleton
+from calibre.ebooks.docx.writer.container import create_skeleton, page_size
 from calibre.ebooks.docx.writer.styles import StylesManager, FloatSpec
 from calibre.ebooks.docx.writer.links import LinksManager
 from calibre.ebooks.docx.writer.images import ImagesManager
@@ -390,10 +390,11 @@ class Convert(object):
     a[href] { text-decoration: underline; color: blue }
     '''
 
-    def __init__(self, oeb, docx, mi):
+    def __init__(self, oeb, docx, mi, add_cover):
-        self.oeb, self.docx = oeb, docx
+        self.oeb, self.docx, self.add_cover = oeb, docx, add_cover
         self.log, self.opts = docx.log, docx.opts
         self.mi = mi
+        self.cover_img = None
 
     def __call__(self):
         from calibre.ebooks.oeb.transforms.rasterize import SVGRasterizer
@@ -411,6 +412,11 @@ class Convert(object):
         for item in self.oeb.spine:
             self.process_item(item)
 
+        if self.add_cover and self.oeb.metadata.cover and unicode(self.oeb.metadata.cover[0]) in self.oeb.manifest.ids:
+            cover_id = unicode(self.oeb.metadata.cover[0])
+            item = self.oeb.manifest.ids[cover_id]
+            self.cover_img = self.images_manager.read_image(item.href)
+
         all_blocks = self.blocks.all_blocks
         remove_blocks = []
         for i, block in enumerate(all_blocks):
@@ -427,6 +433,8 @@ class Convert(object):
         self.blocks.apply_page_break_after()
         self.blocks.resolve_language()
 
+        if self.cover_img is not None:
+            self.cover_img = self.images_manager.create_cover_markup(self.cover_img, *page_size(self.opts))
         self.lists_manager.finalize(all_blocks)
         self.styles_manager.finalize(all_blocks)
         self.write()
@@ -549,6 +557,8 @@ class Convert(object):
         self.docx.document, self.docx.styles, body = create_skeleton(self.opts)
         self.blocks.serialize(body)
         body.append(body[0])  # Move <sectPr> to the end
+        if self.cover_img is not None:
+            self.images_manager.write_cover_block(body, self.cover_img)
         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.docx.fonts)

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

@@ -44,11 +44,7 @@ class ImagesManager(object):
         self.document_relationships = document_relationships
         self.count = 0
 
-    def add_image(self, img, block, stylizer, bookmark=None, as_block=False):
+    def read_image(self, href):
-        src = img.get('src')
-        if not src:
-            return
-        href = self.abshref(src)
         if href not in self.images:
             item = self.oeb.manifest.hrefs.get(href)
             if item is None or not isinstance(item.data, bytes):
@@ -58,9 +54,17 @@ class ImagesManager(object):
             image_rid = self.document_relationships.add_image(image_fname)
             self.images[href] = Image(image_rid, image_fname, width, height, fmt, item)
             item.unload_data_from_memory()
+        return self.images[href]
+
+    def add_image(self, img, block, stylizer, bookmark=None, as_block=False):
+        src = img.get('src')
+        if not src:
+            return
+        href = self.abshref(src)
+        rid = self.read_image(href).rid
         drawing = self.create_image_markup(img, stylizer, href, as_block=as_block)
         block.add_image(drawing, bookmark=bookmark)
-        return self.images[href].rid
+        return rid
 
     def create_image_markup(self, html_img, stylizer, href, as_block=False):
         # TODO: img inside a link (clickable image)
@@ -95,7 +99,7 @@ class ImagesManager(object):
             makeelement(parent, 'wp:simplePos', x='0', y='0')
             makeelement(makeelement(parent, 'wp:positionH', relativeFrom='margin'), 'wp:align').text = floating
             makeelement(makeelement(parent, 'wp:positionV', relativeFrom='line'), 'wp:align').text = 'top'
-        makeelement(parent, 'wp:extent', cx=str(width), cy=str(width))
+        makeelement(parent, 'wp:extent', cx=str(width), cy=str(height))
         if fake_margins:
             # DOCX does not support setting margins for inline images, so we
             # fake it by using effect extents to simulate margins
@@ -108,22 +112,26 @@ class ImagesManager(object):
                 makeelement(parent, 'wp:wrapTopAndBottom')
             else:
                 makeelement(parent, 'wp:wrapSquare', wrapText='bothSides')
-        makeelement(parent, 'wp:docPr', id=str(self.count), name=name, descr=html_img.get('alt') or name)
+        self.create_docx_image_markup(parent, name, html_img.get('alt') or name, img.rid, width, height)
+        return ans
+
+    def create_docx_image_markup(self, parent, name, alt, img_rid, width, height):
+        makeelement, namespaces = self.document_relationships.namespace.makeelement, self.document_relationships.namespace.namespaces
+        makeelement(parent, 'wp:docPr', id=str(self.count), name=name, descr=alt)
         makeelement(makeelement(parent, 'wp:cNvGraphicFramePr'), 'a:graphicFrameLocks', noChangeAspect="1")
         g = makeelement(parent, 'a:graphic')
         gd = makeelement(g, 'a:graphicData', uri=namespaces['pic'])
         pic = makeelement(gd, 'pic:pic')
         nvPicPr = makeelement(pic, 'pic:nvPicPr')
-        makeelement(nvPicPr, 'pic:cNvPr', id='0', name=name, descr=html_img.get('alt') or name)
+        makeelement(nvPicPr, 'pic:cNvPr', id='0', name=name, descr=alt)
         makeelement(nvPicPr, 'pic:cNvPicPr')
         bf = makeelement(pic, 'pic:blipFill')
-        makeelement(bf, 'a:blip', r_embed=img.rid)
+        makeelement(bf, 'a:blip', r_embed=img_rid)
         makeelement(makeelement(bf, 'a:stretch'), 'a:fillRect')
         spPr = makeelement(pic, 'pic:spPr')
         xfrm = makeelement(spPr, 'a:xfrm')
         makeelement(xfrm, 'a:off', x='0', y='0'), makeelement(xfrm, 'a:ext', cx=str(width), cy=str(height))
         makeelement(makeelement(spPr, 'a:prstGeom', prst='rect'), 'a:avLst')
-        return ans
 
     def create_filename(self, href, fmt):
         fname = ascii_filename(urlunquote(posixpath.basename(href)))
@@ -147,3 +155,31 @@ class ImagesManager(object):
             return item.data
         finally:
             item.unload_data_from_memory(False)
+
+    def create_cover_markup(self, img, width, height):
+        self.count += 1
+        makeelement, namespaces = self.document_relationships.namespace.makeelement, self.document_relationships.namespace.namespaces
+
+        root = etree.Element('root', nsmap=namespaces)
+        ans = makeelement(root, 'w:drawing', append=False)
+        parent = makeelement(ans, 'wp:anchor', **{'dist'+edge:'0' for edge in 'LRTB'})
+        parent.set('simplePos', '0'), parent.set('relativeHeight', '1'), parent.set('behindDoc',"0"), parent.set('locked', "0")
+        parent.set('layoutInCell', "1"), parent.set('allowOverlap', '1')
+        makeelement(parent, 'wp:simplePos', x='0', y='0')
+        makeelement(makeelement(parent, 'wp:positionH', relativeFrom='page'), 'wp:align').text = 'center'
+        makeelement(makeelement(parent, 'wp:positionV', relativeFrom='page'), 'wp:align').text = 'center'
+        width, height = map(pt_to_emu, (width, height))
+        makeelement(parent, 'wp:extent', cx=str(width), cy=str(height))
+        makeelement(parent, 'wp:effectExtent', l='0', r='0', t='0', b='0')
+        makeelement(parent, 'wp:wrapTopAndBottom')
+        self.create_docx_image_markup(parent, 'cover.jpg', _('Cover'), img.rid, width, height)
+        return ans
+
+    def write_cover_block(self, body, cover_image):
+        makeelement, namespaces = self.document_relationships.namespace.makeelement, self.document_relationships.namespace.namespaces
+        pbb = body[0].xpath('//*[local-name()="pageBreakBefore"]')[0]
+        pbb.set('{%s}val' % namespaces['w'], 'on')
+        p = makeelement(body, 'w:p', append=False)
+        body.insert(0, p)
+        r = makeelement(p, 'w:r')
+        r.append(cover_image)

src/calibre/gui2/convert/docx_output.py

@@ -19,7 +19,7 @@ class PluginWidget(Widget, Ui_Form):
 
     def __init__(self, parent, get_option, get_help, db=None, book_id=None):
         Widget.__init__(self, parent, [
-            'docx_page_size', 'docx_custom_page_size',
+            'docx_page_size', 'docx_custom_page_size', 'docx_no_cover',
         ])
         for x in get_option('docx_page_size').option.choices:
             self.opt_docx_page_size.addItem(x)

src/calibre/gui2/convert/docx_output.ui

@@ -47,6 +47,13 @@
      </property>
     </widget>
    </item>
+   <item row="2" column="0" colspan="2">
+    <widget class="QCheckBox" name="opt_docx_no_cover">
+     <property name="text">
+      <string>Do not insert &amp;cover as image at start of document</string>
+     </property>
+    </widget>
+   </item>
   </layout>
  </widget>
  <resources/>