DOCX Input: Add support for DOCX files created by Word 2013 in "Strict" mode

src/calibre/ebooks/docx/block_styles.py

@@ -28,7 +28,15 @@ def simple_float(val, mult=1.0):
     try:
         return float(val) * mult
     except (ValueError, TypeError, AttributeError, KeyError):
-        return None
+        pass
+
+def twips(val, mult=0.05):
+    ''' Parse val as either a pure number representing twentieths of a point or a number followed by the suffix pt, representing pts.'''
+    try:
+        return float(val) * mult
+    except (ValueError, TypeError, AttributeError, KeyError):
+        if val and val.endswith('pt') and mult == 0.05:
+            return twips(val[:-2], mult=1.0)
 
 
 LINE_STYLES = {  # {{{
@@ -160,6 +168,7 @@ def read_spacing(parent, dest, XPath, get):
         l, lr = get(s, 'w:line'), get(s, 'w:lineRule', 'auto')
         if l is not None:
             lh = simple_float(l, 0.05) if lr in {'exact', 'atLeast'} else simple_float(l, 1/240.0)
+            if lh is not None:
                 line_height = '%.3g%s' % (lh, 'pt' if lr in {'exact', 'atLeast'} else '')
 
     setattr(dest, 'margin_top', padding_top)

src/calibre/ebooks/docx/names.py

@@ -30,6 +30,11 @@ TRANSITIONAL_NAMES = {
     'WEB_SETTINGS' : 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/webSettings',
 }
 
+STRICT_NAMES = {
+    k:v.replace('http://schemas.openxmlformats.org/officeDocument/2006',  'http://purl.oclc.org/ooxml/officeDocument')
+    for k, v in TRANSITIONAL_NAMES.iteritems()
+}
+
 TRANSITIONAL_NAMESPACES = {
     'mo': 'http://schemas.microsoft.com/office/mac/office/2008/main',
     'o': 'urn:schemas-microsoft-com:office:office',
@@ -60,6 +65,14 @@ TRANSITIONAL_NAMESPACES = {
     'dcmitype': 'http://purl.org/dc/dcmitype/',
     'dcterms': 'http://purl.org/dc/terms/'
 }
+
+STRICT_NAMESPACES = {
+    k:v.replace(
+        'http://schemas.openxmlformats.org/officeDocument/2006', 'http://purl.oclc.org/ooxml/officeDocument').replace(
+        'http://schemas.openxmlformats.org/wordprocessingml/2006', 'http://purl.oclc.org/ooxml/wordprocessingml').replace(
+        'http://schemas.openxmlformats.org/drawingml/2006', 'http://purl.oclc.org/ooxml/drawingml')
+    for k, v in TRANSITIONAL_NAMESPACES.iteritems()
+}
 # }}}
 
 def barename(x):
@@ -83,6 +96,9 @@ class DOCXNamespace(object):
         if transitional:
             self.namespaces = TRANSITIONAL_NAMESPACES.copy()
             self.names = TRANSITIONAL_NAMES.copy()
+        else:
+            self.namespaces = STRICT_NAMESPACES.copy()
+            self.names = STRICT_NAMES.copy()
 
     def XPath(self, expr):
         ans = self.xpath_cache.get(expr, None)

src/calibre/ebooks/docx/styles.py

@@ -9,7 +9,7 @@ __copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
 import textwrap
 from collections import OrderedDict, Counter
 
-from calibre.ebooks.docx.block_styles import ParagraphStyle, inherit
+from calibre.ebooks.docx.block_styles import ParagraphStyle, inherit, twips
 from calibre.ebooks.docx.char_styles import RunStyle
 from calibre.ebooks.docx.tables import TableStyle
 
@@ -21,29 +21,21 @@ class PageProperties(object):
     '''
 
     def __init__(self, namespace, elems=()):
-        self.width = self.height = 595.28, 841.89  # pts, A4
+        self.width, self.height = 595.28, 841.89  # pts, A4
         self.margin_left = self.margin_right = 72  # pts
+
+        def setval(attr, val):
+            val = twips(val)
+            if val is not None:
+                setattr(self, attr, val)
+
         for sectPr in elems:
             for pgSz in namespace.XPath('./w:pgSz')(sectPr):
                 w, h = namespace.get(pgSz, 'w:w'), namespace.get(pgSz, 'w:h')
-                try:
+                setval('width', w), setval('height', h)
-                    self.width = int(w)/20
-                except (ValueError, TypeError):
-                    pass
-                try:
-                    self.height = int(h)/20
-                except (ValueError, TypeError):
-                    pass
             for pgMar in namespace.XPath('./w:pgMar')(sectPr):
                 l, r = namespace.get(pgMar, 'w:left'), namespace.get(pgMar, 'w:right')
-                try:
+                setval('margin_left', l), setval('margin_right', r)
-                    self.margin_left = int(l)/20
-                except (ValueError, TypeError):
-                    pass
-                try:
-                    self.margin_right = int(r)/20
-                except (ValueError, TypeError):
-                    pass
 
 
 class Style(object):