DOCX Input: Start work on tables

src/calibre/ebooks/docx/block_styles.py

@@ -65,35 +65,41 @@ LINE_STYLES = {  # {{{
 }  # }}}
 
 # Read from XML {{{
-def read_border(parent, dest):
-    tvals = {'padding_%s':inherit, 'border_%s_width':inherit,
-            'border_%s_style':inherit, 'border_%s_color':inherit}
-    vals = {}
-    for edge in ('left', 'top', 'right', 'bottom'):
-        vals.update({k % edge:v for k, v in tvals.iteritems()})
 
-    for border in XPath('./w:pBdr')(parent):
-        for edge in ('left', 'top', 'right', 'bottom'):
-            for elem in XPath('./w:%s' % edge)(border):
-                color = get(elem, 'w:color')
-                if color is not None:
-                    vals['border_%s_color' % edge] = simple_color(color)
-                style = get(elem, 'w:val')
-                if style is not None:
-                    vals['border_%s_style' % edge] = LINE_STYLES.get(style, 'solid')
-                space = get(elem, 'w:space')
-                if space is not None:
-                    try:
-                        vals['padding_%s' % edge] = float(space)
-                    except (ValueError, TypeError):
-                        pass
-                sz = get(elem, 'w:sz')
-                if sz is not None:
-                    # we dont care about art borders (they are only used for page borders)
-                    try:
-                        vals['border_%s_width' % edge] = min(96, max(2, float(sz))) / 8
-                    except (ValueError, TypeError):
-                        pass
+border_props = ('padding_%s', 'border_%s_width', 'border_%s_style', 'border_%s_color')
+
+def read_single_border(parent, edge):
+    color = style = width = padding = None
+    for elem in XPath('./w:%s' % edge)(parent):
+        c = get(elem, 'w:color')
+        if c is not None:
+            color = simple_color(c)
+        s = get(elem, 'w:val')
+        if s is not None:
+            style = LINE_STYLES.get(s, 'solid')
+        space = get(elem, 'w:space')
+        if space is not None:
+            try:
+                padding = float(space)
+            except (ValueError, TypeError):
+                pass
+        sz = get(elem, 'w:sz')
+        if sz is not None:
+            # we dont care about art borders (they are only used for page borders)
+            try:
+                width = min(96, max(2, float(sz))) / 8
+            except (ValueError, TypeError):
+                pass
+    return {p:v for p, v in zip(border_props, (padding, width, style, color))}
+
+def read_border(parent, dest, border_edges=('left', 'top', 'right', 'bottom'), name='pBdr'):
+    vals = {k % edge:inherit for edge in border_edges for k in border_props}
+
+    for border in XPath('./w:' + name)(parent):
+        for edge in border_edges:
+            for prop, val in read_single_border(border, edge).iteritems():
+                if val is not None:
+                    vals[prop % edge] = val
 
     for key, val in vals.iteritems():
         setattr(dest, key, val)

src/calibre/ebooks/docx/footnotes.py

@@ -17,7 +17,7 @@ class Note(object):
         self.parent = parent
 
     def __iter__(self):
-        for p in descendants(self.parent, 'w:p'):
+        for p in descendants(self.parent, 'w:p', 'w:tbl'):
             yield p
 
 class Footnotes(object):

src/calibre/ebooks/docx/styles.py

@@ -11,6 +11,7 @@ from collections import OrderedDict, Counter
 
 from calibre.ebooks.docx.block_styles import ParagraphStyle, inherit
 from calibre.ebooks.docx.char_styles import RunStyle
+from calibre.ebooks.docx.tables import TableStyle
 from calibre.ebooks.docx.names import XPath, get
 
 class PageProperties(object):
@@ -66,10 +67,17 @@ class Style(object):
             self.based_on = None
         self.is_default = get(elem, 'w:default') in {'1', 'on', 'true'}
 
-        self.paragraph_style = self.character_style = None
+        self.paragraph_style = self.character_style = self.table_style = None
 
-        if self.style_type in {'paragraph', 'character'}:
-            if self.style_type == 'paragraph':
+        if self.style_type in {'paragraph', 'character', 'table'}:
+            if self.style_type == 'table':
+                for tblPr in XPath('./w:tblPr')(elem):
+                    ts = TableStyle(tblPr)
+                    if self.table_style is None:
+                        self.table_style = ts
+                    else:
+                        self.table_style.update(ts)
+            if self.style_type in {'paragraph', 'table'}:
                 for pPr in XPath('./w:pPr')(elem):
                     ps = ParagraphStyle(pPr)
                     if self.paragraph_style is None:
@@ -90,6 +98,10 @@ class Style(object):
                 self.numbering_style_link = get(x, 'w:val')
 
     def resolve_based_on(self, parent):
+        if parent.table_style is not None:
+            if self.table_style is None:
+                self.table_style = TableStyle()
+            self.table_style.resolve_based_on(parent.table_style)
         if parent.paragraph_style is not None:
             if self.paragraph_style is None:
                 self.paragraph_style = ParagraphStyle()

src/calibre/ebooks/docx/tables.py

@@ -0,0 +1,152 @@
+#!/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>'
+
+from collections import OrderedDict
+
+from lxml.html.builder import TABLE, TR, TD
+
+from calibre.ebooks.docx.block_styles import inherit, read_shd, read_border ,border_props  # noqa
+from calibre.ebooks.docx.names import XPath, get
+
+def _read_width(elem):
+    ans = inherit
+    try:
+        w = int(get(elem, 'w:w'))
+    except (TypeError, ValueError):
+        w = 0
+    typ = get(elem, 'w:type', 'auto')
+    if typ == 'nil':
+        ans = '0'
+    elif typ == 'auto':
+        ans = 'auto'
+    elif typ == 'dxa':
+        ans = '%.3gpt' % (w/20)
+    elif typ == 'pct':
+        ans = '%.3g%%' % (w/50)
+    return ans
+
+def read_width(parent, dest):
+    ans = inherit
+    for tblW in XPath('./w:tblW')(parent):
+        ans = _read_width(tblW)
+    setattr(dest, 'width', ans)
+
+def read_padding(parent, dest):
+    name = 'tblCellMar' if parent.tag.endswith('}tblPr') else 'tcMar'
+    left = top = bottom = right = inherit
+    for mar in XPath('./w:%s' % name)(parent):
+        for x in ('left', 'top', 'right', 'bottom'):
+            for edge in XPath('./w:%s' % x)(mar):
+                locals()[x] = _read_width(edge)
+    for x in ('left', 'top', 'right', 'bottom'):
+        setattr(dest, 'cell_padding_%s' % x, locals()[x])
+
+def read_justification(parent, dest):
+    left = right = inherit
+    for jc in XPath('./w:jc[@w:val]')(parent):
+        val = get(jc, 'w:val')
+        if not val:
+            continue
+        if val == 'left':
+            right = 'auto'
+        elif val == 'right':
+            left = 'auto'
+        elif val == 'center':
+            left = right = 'auto'
+    setattr(dest, 'margin_left', left)
+    setattr(dest, 'margin_right', right)
+
+def read_spacing(parent, dest):
+    ans = inherit
+    for cs in XPath('./w:tblCellSpacing')(parent):
+        ans = _read_width(cs)
+    setattr(dest, 'spacing', ans)
+
+def read_indent(parent, dest):
+    ans = inherit
+    for cs in XPath('./w:tblInd')(parent):
+        ans = _read_width(cs)
+    setattr(dest, 'indent', ans)
+
+border_edges = ('left', 'top', 'right', 'bottom', 'insideH', 'insideV')
+
+def read_borders(parent, dest):
+    name = 'tblBorders' if parent.tag.endswith('}tblPr') else 'tcBorders'
+    read_border(parent, dest, border_edges, name)
+
+class TableStyle(object):
+
+    all_properties = (
+        'width', 'cell_padding_left', 'cell_padding_right', 'cell_padding_top',
+        'cell_padding_bottom', 'margin_left', 'margin_right', 'background_color',
+        'spacing', 'indent',
+    ) + tuple(k % edge for edge in border_edges for k in border_props)
+
+    def __init__(self, tblPr=None):
+        if tblPr is None:
+            for p in self.all_properties:
+                setattr(self, p, inherit)
+        else:
+            for x in ('width', 'padding', 'shd', 'justification', 'spacing', 'indent', 'borders'):
+                f = globals()['read_%s' % x]
+                f(tblPr, self)
+
+        self._css = None
+
+    def update(self, other):
+        for prop in self.all_properties:
+            nval = getattr(other, prop)
+            if nval is not inherit:
+                setattr(self, prop, nval)
+
+    def resolve_based_on(self, parent):
+        for p in self.all_properties:
+            val = getattr(self, p)
+            if val is inherit:
+                setattr(self, p, getattr(parent, p))
+
+    @property
+    def css(self):
+        return self._css
+
+class Tables(object):
+
+    def __init__(self):
+        self.tables = OrderedDict()
+
+    def register(self, tbl):
+        self.tables[tbl] = self.current_table = []
+
+    def add(self, p):
+        self.current_table.append(p)
+
+    def apply_markup(self, object_map):
+        rmap = {v:k for k, v in object_map.iteritems()}
+        for tbl, blocks in self.tables.iteritems():
+            if not blocks:
+                continue
+            parent = rmap[blocks[0]].getparent()
+            table = TABLE('\n\t\t')
+            idx = parent.index(rmap[blocks[0]])
+            parent.insert(idx, table)
+            for row in XPath('./w:tr')(tbl):
+                tr = TR('\n\t\t\t')
+                tr.tail = '\n\t\t'
+                table.append(tr)
+                for tc in XPath('./w:tc')(row):
+                    td = TD()
+                    td.tail = '\n\t\t\t'
+                    tr.append(td)
+                    for p in XPath('./w:p')(tc):
+                        block = rmap[p]
+                        td.append(block)
+                if len(tr):
+                    tr[-1].tail = '\n\t\t'
+            if len(table):
+                table[-1].tail = '\n\t'
+

src/calibre/ebooks/docx/to_html.py

@@ -21,6 +21,7 @@ from calibre.ebooks.docx.styles import Styles, inherit, PageProperties
 from calibre.ebooks.docx.numbering import Numbering
 from calibre.ebooks.docx.fonts import Fonts
 from calibre.ebooks.docx.images import Images
+from calibre.ebooks.docx.tables import Tables
 from calibre.ebooks.docx.footnotes import Footnotes
 from calibre.ebooks.metadata.opf2 import OPFCreator
 from calibre.ebooks.metadata.toc import TOC
@@ -47,6 +48,7 @@ class Convert(object):
         self.body = BODY()
         self.styles = Styles()
         self.images = Images()
+        self.tables = Tables()
         self.object_map = OrderedDict()
         self.html = HTML(
             HEAD(
@@ -98,15 +100,26 @@ class Convert(object):
                 dl.append(DT('[', A('←' + text, href='#back_%s' % anchor, title=text), id=anchor))
                 dl[-1][0].tail = ']'
                 dl.append(DD())
+                in_table = False
                 for wp in note:
+                    if wp.tag.endswith('}tbl'):
+                        self.tables.register(wp)
+                        in_table = True
+                        continue
+                    if in_table:
+                        if ancestor(wp, 'w:tbl') is not None:
+                            self.tables.add(wp)
+                        else:
+                            in_table = False
                     p = self.convert_p(wp)
                     dl[-1].append(p)
 
         self.resolve_links(relationships_by_id)
-        # TODO: tables <w:tbl> child of <w:body> (nested tables?)
 
         self.styles.cascade(self.layers)
 
+        self.tables.apply_markup(self.object_map)
+
         numbered = []
         for html_obj, obj in self.object_map.iteritems():
             raw = obj.get('calibre_num_id', None)
@@ -154,7 +167,13 @@ class Convert(object):
         current = []
         self.page_map = OrderedDict()
 
-        for p in descendants(doc, 'w:p'):
+        in_table = False
+
+        for p in descendants(doc, 'w:p', 'w:tbl'):
+            if p.tag.endswith('}tbl'):
+                in_table = True
+                self.tables.register(p)
+                continue
             sect = tuple(descendants(p, 'w:sectPr'))
             if sect:
                 pr = PageProperties(sect)
@@ -163,6 +182,11 @@ class Convert(object):
                 current = []
             else:
                 current.append(p)
+            if in_table:
+                if ancestor(p, 'w:tbl') is not None:
+                    self.tables.add(p)
+                else:
+                    in_table = False
         if current:
             last = XPath('./w:body/w:sectPr')(doc)
             pr = PageProperties(last)