Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-09 03:04:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

DOCX Output: Use standard names for styles corresponding to HTML heading tags

Browse Source
This commit is contained in:
Kovid Goyal 2015-05-11 09:29:35 +05:30
parent 51b3df2c29
commit 89c0028fe8
2 changed files with 22 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
src/calibre/ebooks/docx/writer/from_html.py
View File

@ -124,6 +124,7 @@ class Block(object):
self.numbering_id = None self.numbering_id = None
self.parent_items = None self.parent_items = None
self.html_block = html_block self.html_block = html_block
self.html_tag = barename(html_block.tag)
self.float_spec = float_spec self.float_spec = float_spec
if float_spec is not None: if float_spec is not None:
float_spec.blocks.append(self) float_spec.blocks.append(self)

22
src/calibre/ebooks/docx/writer/styles.py
View File

@ -49,6 +49,7 @@ class CombinedStyle(object):
self.bs, self.rs, self.blocks = bs, rs, blocks self.bs, self.rs, self.blocks = bs, rs, blocks
self.namespace = namespace self.namespace = namespace
self.id = self.name = self.seq = None self.id = self.name = self.seq = None
self.outline_level = None
def apply(self): def apply(self):
for block in self.blocks: for block in self.blocks:
@ -68,6 +69,8 @@ class CombinedStyle(object):
block.set(w('default'), '1') block.set(w('default'), '1')
pPr = makeelement(block, 'w:pPr') pPr = makeelement(block, 'w:pPr')
self.bs.serialize_properties(pPr, normal_style.bs) self.bs.serialize_properties(pPr, normal_style.bs)
if self.outline_level is not None:
makeelement(pPr, 'w:outlineLvl', w_val=str(self.outline_level))
rPr = makeelement(block, 'w:rPr') rPr = makeelement(block, 'w:rPr')
self.rs.serialize_properties(rPr, normal_style.rs) self.rs.serialize_properties(rPr, normal_style.rs)
@ -526,6 +529,8 @@ class StylesManager(object):
block_counts, run_counts = Counter(), Counter() block_counts, run_counts = Counter(), Counter()
block_rmap, run_rmap = defaultdict(list), defaultdict(list) block_rmap, run_rmap = defaultdict(list), defaultdict(list)
used_pairs = defaultdict(list) used_pairs = defaultdict(list)
heading_styles = defaultdict(list)
headings = frozenset('h1 h2 h3 h4 h5 h6'.split())
for block in blocks: for block in blocks:
bs = block.style bs = block.style
@ -540,6 +545,8 @@ class StylesManager(object):
if local_run_counts: if local_run_counts:
rs = local_run_counts.most_common(1)[0][0] rs = local_run_counts.most_common(1)[0][0]
used_pairs[(bs, rs)].append(block) used_pairs[(bs, rs)].append(block)
if block.html_tag in headings:
heading_styles[block.html_tag].append((bs, rs))
rnum = len(str(max(1, len(run_counts) - 1))) rnum = len(str(max(1, len(run_counts) - 1)))
for i, (text_style, count) in enumerate(run_counts.most_common()): for i, (text_style, count) in enumerate(run_counts.most_common()):
@ -553,16 +560,29 @@ class StylesManager(object):
self.text_styles.pop(s) self.text_styles.pop(s)
counts = Counter() counts = Counter()
smap = {}
for (bs, rs), blocks in used_pairs.iteritems(): for (bs, rs), blocks in used_pairs.iteritems():
s = CombinedStyle(bs, rs, blocks, self.namespace) s = CombinedStyle(bs, rs, blocks, self.namespace)
smap[(bs, rs)] = s
counts[s] += sum(1 for b in blocks if not b.is_empty()) counts[s] += sum(1 for b in blocks if not b.is_empty())
for i, heading_tag in enumerate(sorted(heading_styles)):
styles = sorted((smap[k] for k in heading_styles[heading_tag]), key=counts.__getitem__)
styles = filter(lambda s:s.outline_level is None, styles)
if styles:
heading_style = styles[-1]
heading_style.outline_level = i
snum = len(str(max(1, len(counts) - 1))) snum = len(str(max(1, len(counts) - 1)))
for i, (style, count) in enumerate(counts.most_common()): for i, (style, count) in enumerate(counts.most_common()):
if i == 0: if i == 0:
self.normal_style = style self.normal_style = style
style.id = style.name = 'Normal' style.id = style.name = 'Normal'
else: else:
style.id = style.name = 'Para %0{}d'.format(snum) % i if style.outline_level is None:
val = 'Para %0{}d'.format(snum) % i
else:
val = 'Heading %d' % (style.outline_level + 1)
style.id = style.name = val
style.seq = i style.seq = i
self.combined_styles = sorted(counts.iterkeys(), key=attrgetter('seq')) self.combined_styles = sorted(counts.iterkeys(), key=attrgetter('seq'))
[ls.apply() for ls in self.combined_styles] [ls.apply() for ls in self.combined_styles]