def add_el(self, el, el_len=100):
if self.el_len_total > self.max_el_len_total:
self.flush_els()
if len(self.el_stack) == 0:
assert_d(self.current_part == None)
self.current_part, body_el = self.make_xhtml()
self.el_stack.append(body_el)
self.el_stack[-1].append(el)
self.el_len_total += el_len
return 'part' + str(self.part_number).zfill(4) + '.html'
def draw_text_el(draw, dpi, el):
for par in el:
par_coords = box_from_par(par)
if par_coords is not None:
render(draw, par, 'par', par_coords)
tl, rb = par_coords
t = ''
for att, nick in [ ('align', 'ta'),
('leftIndent', 'li'),
('rightIndent', 'ri'),
('startIndent', 'si'),
('lineSpacing', 'ls') ]:
att_txt = par.get(att)
if att_txt is not None:
t += nick + ':' + att_txt + ' '
if len(t) > 0:
f = font.get_font("Courier", dpi / opts.scale, 12)
draw.text(tl, t, font=f, fill=color.green)
for line in par:
render(draw, line, 'line');
for fmt in line:
assert_d(fmt.tag == abyns+'formatting')
font_name = fmt.get('ff')
font_size = fmt.get('fs')
font_size = int(re.sub('\.', '', font_size))
font_ital = (fmt.get('italic') == 'true')
f = font.get_font(font_name, dpi / opts.scale,
font_size, font_ital)
for cp in fmt:
assert_d(cp.tag == abyns+'charParams')
if opts.text:
draw.text((int(cp.get('l')) / opts.scale,
int(cp.get('b')) / opts.scale),
cp.text.encode('utf-8'),
font=f,
fill=color.yellow)
def assert_tag(el, expected):
shorttag = nons(el.tag)
if (shorttag != 'block'):
assert_d(shorttag == expected)
def scan_pages(context, scandata, iabook):
book_id = iabook.get_book_id()
scandata_pages = scandata.pageData.page
try:
# dpi isn't always there
dpi = int(scandata.bookData.dpi.text)
except AttributeError:
dpi = 300
i = 0
f = ImageFont.load_default()
# f = ImageFont.load('/Users/mccabe/s/archive/epub/Times-18.bdf')
for event, page in context:
orig_width = int(page.get('width'))
orig_height = int(page.get('height'))
width = orig_width / s
height = orig_height / s
image = Image.new('RGB', (width, height))
page_image = None
image_str = iabook.get_page_image(i, width, height, out_img_type='ppm')
if image_str is not None:
page_image = Image.open(StringIO.StringIO(image_str))
(nw, nh) = page_image.size
if nw != width or nh != height:
page_image = page_image.resize((width, height))
# image.paste(page_image, None)
try:
image = Image.blend(image, page_image, .2)
except ValueError:
print 'blending - images didn\'t match'
debug()
pass
draw = ImageDraw.Draw(image)
for block in page:
if block.get('blockType') == 'Picture' and page_image is not None:
cropped = page_image.crop(four_coords(block, scale))
image.paste(cropped, four_coords(block, scale))
for block in page:
if block.get('blockType') == 'Text':
render(draw, block, 'block_text')
if block.get('blockType') == 'Picture':
render(draw, block, 'block_picture')
if block.get('blockType') == 'Table':
render(draw, block, 'block_table')
# else:
# render(draw, block, 'block_picture')
# if i > 0:
# cropped = page_image.crop(four_coords(block))
# image.paste(cropped, four_coords(block))
for el in block:
if el.tag == abyns+'region':
pass
elif el.tag == abyns+'row':
for cell in el:
for text in cell:
for par in text:
par_coords = box_from_par(par)
if par_coords is not None:
render(draw, par, 'par', par_coords)
for line in par:
render(draw, line, 'line');
for fmt in line:
assert_d(fmt.tag == abyns+'formatting')
font_name = fmt.get('ff')
font_size = fmt.get('fs')
font_size = int(re.sub('\.', '', font_size))
font_ital = (fmt.get('italic') == 'true')
f = font.get_font(font_name, dpi / scale, font_size, font_ital)
for cp in fmt:
assert_d(cp.tag == abyns+'charParams')
draw.text((int(cp.get('l'))/s,
int(cp.get('b'))/s),
cp.text.encode('utf-8'),
font=f,
fill=color.yellow)
elif el.tag == abyns+'text':
for par in el:
par_coords = box_from_par(par)
if par_coords is not None:
render(draw, par, 'par', par_coords)
for line in par:
render(draw, line, 'line');
for fmt in line:
assert_d(fmt.tag == abyns+'formatting')
font_name = fmt.get('ff')
font_size = fmt.get('fs')
font_size = int(re.sub('\.', '', font_size))
font_ital = (fmt.get('italic') == 'true')
f = font.get_font(font_name, dpi / scale, font_size, font_ital)
for cp in fmt:
assert_d(cp.tag == abyns+'charParams')
draw.text((int(cp.get('l'))/s,
int(cp.get('b'))/s),
cp.text.encode('utf-8'),
font=f,
fill=color.yellow)
elif (el.tag == abyns+'row'):
pass
else:
print('unexpected tag type' + el.tag)
sys.exit(-1)
if not include_page(scandata_pages[i]):
draw.line([(0, 0), image.size], width=50, fill=color.red)
image.save(outdir + '/img' + scandata_pages[i].get('leafNum') + '.png')
print 'page index: ' + str(i)
page.clear()
i += 1
return None
