Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-08 02:34:06 -04:00
Code Issues Packages Projects Releases Wiki Activity

...

Browse Source
This commit is contained in:
Kovid Goyal 2010-02-14 11:34:48 -07:00
parent d21f3ffb52
commit bd7796bcd2
1 changed files with 33 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

55
src/calibre/ebooks/pdf/reflow.py
View File

@ -143,6 +143,14 @@ class Column(object):
def add(self, elem): def add(self, elem):
if elem in self.elements: return if elem in self.elements: return
self.elements.append(elem) self.elements.append(elem)
self._post_add()
def prepend(self, elem):
if elem in self.elements: return
self.elements.insert(0, elem)
self._post_add()
def _post_add(self):
self.elements.sort(cmp=lambda x,y:cmp(x.bottom,y.bottom)) self.elements.sort(cmp=lambda x,y:cmp(x.bottom,y.bottom))
self.top = self.elements[0].top self.top = self.elements[0].top
self.bottom = self.elements[-1].bottom self.bottom = self.elements[-1].bottom
@ -307,16 +315,14 @@ class Region(object):
self.absorb_region(region, at) self.absorb_region(region, at)
def absorb_region(self, region, at): def absorb_region(self, region, at):
src_iter = lambda x:x if at == 'bottom' else reversed
if len(region.columns) <= len(self.columns): if len(region.columns) <= len(self.columns):
for i in range(len(region.columns)): for i in range(len(region.columns)):
src, dest = region.columns[i], self.columns[i] src, dest = region.columns[i], self.columns[i]
for elem in src_iter(src): if at != 'bottom':
if at == 'bottom': src = reversed(list(iter(src)))
dest.append(elem) for elem in src:
else: func = dest.add if at == 'bottom' else dest.prepend
dest.insert(0, elem) func(elem)
else: else:
col_map = {} col_map = {}
@ -331,15 +337,16 @@ class Region(object):
max_overlap_index = j max_overlap_index = j
col_map[i] = max_overlap_index col_map[i] = max_overlap_index
lines = max(map(len, region.columns)) lines = max(map(len, region.columns))
for i in range(src_iter(lines)): if at == 'bottom':
lines = range(lines)
else:
lines = range(lines-1, -1, -1)
for i in lines:
for j, src in enumerate(region.columns): for j, src in enumerate(region.columns):
dest = self.columns[col_map[j]] dest = self.columns[col_map[j]]
if i < len(src): if i < len(src):
if at == 'bottom': func = dest.add if at == 'bottom' else dest.prepend
dest.append(src[i]) func(src.elements[i])
else:
dest.insert(0, src[i])
def linearize(self): def linearize(self):
self.elements = [] self.elements = []
@ -480,6 +487,8 @@ class Page(object):
while found: while found:
found = False found = False
for i, region in enumerate(self.regions): for i, region in enumerate(self.regions):
if region in absorbed:
continue
if region.is_small: if region.is_small:
found = True found = True
regions = [region] regions = [region]
@ -504,23 +513,24 @@ class Page(object):
absorb_into = None absorb_into = None
else: else:
absorb_into = prev_region absorb_into = prev_region
if next_region.line_count >= prev_region.line_count: if self.regions[next_region].line_count >= \
self.regions[prev_region].line_count:
avg_column_count = sum([len(r.columns) for r in avg_column_count = sum([len(r.columns) for r in
regions])/float(len(regions)) regions])/float(len(regions))
if next_region.line_count > prev_region.line_count \ if self.regions[next_region].line_count > \
or abs(avg_column_count - len(prev_region.columns)) \ self.regions[prev_region].line_count \
> abs(avg_column_count - len(next_region.columns)): or abs(avg_column_count -
len(self.regions[prev_region].columns)) \
> abs(avg_column_count -
len(self.regions[next_region].columns)):
absorb_into = next_region absorb_into = next_region
absorb_at = 'top' absorb_at = 'top'
if absorb_into is not None: if absorb_into is not None:
absorb_into.absorb_regions(regions, absorb_at) self.regions[absorb_into].absorb_regions(regions, absorb_at)
absorbed.update(regions) absorbed.update(regions)
i = j
for region in absorbed: for region in absorbed:
self.regions.remove(region) self.regions.remove(region)
def sort_into_columns(self, elem, neighbors): def sort_into_columns(self, elem, neighbors):
neighbors.add(elem) neighbors.add(elem)
neighbors = sorted(neighbors, cmp=lambda x,y:cmp(x.left, y.left)) neighbors = sorted(neighbors, cmp=lambda x,y:cmp(x.left, y.left))
@ -639,8 +649,9 @@ class PDFDocument(object):
for elem in self.elements: for elem in self.elements:
html.extend(elem.to_html()) html.extend(elem.to_html())
html += ['</body>', '</html>'] html += ['</body>', '</html>']
raw = (u'\n'.join(html)).replace('</strong><strong>', '')
with open('index.html', 'wb') as f: with open('index.html', 'wb') as f:
f.write((u'\n'.join(html)).encode('utf-8')) f.write(raw.encode('utf-8'))