print 'legend: (for abbyy format)'

print ' ta - align'

print ' li - leftIndent'

print ' ri - rightIndent'

print ' si - startIndent'

print ' ls - lineSpacing'

print

print 'Colors:'

for sty_name in styles:

import optparse

parser = optparse.OptionParser()

parser = optparse.OptionParser(usage='usage: %prog [options]',

version='%prog 0.1',

description='A visualizer for '

'coordinate-annotated OCR data.')

def legend_callback(option, opt_str, value, parser):

legend()

sys.exit(0)

parser.add_option('--legend', '-l',

action='callback',

callback=legend_callback,

help='Display legend information - for generated images')

parser.add_option('--reduce',

action='store',

type='int',

metavar='n',

default=2,

help='For jp2 input images, reduce jp2 resolution '

'by 2 ^ n when reading '

'original image, for speed. This also reduces the '

'output scale by 2 ^ n, unless otherwise specified '

'with --scale.')

parser.add_option('--scale',

action='store',

type='int',

default=0,

help='Scale result images down from original scan '

'resolution.')

parser.add_option('--last',

action='store',

type='int',

metavar='leaf',

default=0,

help='Stop generating output leaves '

'after the specified leaf')

parser.add_option('--first',

action='store',

type='int',

metavar='leaf',

default=0,

help='Don\'t generate output leaves until the '

'specified leaf')

parser.add_option('--leaf',

action='store',

type='int',

metavar='leaf',

default=0,

help='Only generate output for the specified leaf')

parser.add_option('--text',

action='store_true',

default=False,

help='Generate output characters for OCRed '

'text in input files')

parser.add_option('--outdir',

help='Output directory. Default is source_type + \'_viz\'')

parser.add_option('--source',

choices=['abbyy', 'pdftoxml', 'djvu'],

default='abbyy',

help='Which source to use for OCR data/coordinates.')

parser.add_option('--show-opts',

action='store_true',

# help=optparse.SUPPRESS_HELP

help='Display parsed options/defaults and exit')

global opts

opts, args = parser.parse_args(argv)

if opts.reduce < 0 or opts.reduce > 4:

parser.error('--reduce must be between 0 and 4')

if opts.scale == 0:

opts.scale = 2 ** opts.reduce

if opts.leaf != 0:

if opts.first > 0 or opts.last > 0:

parser.error('can\'t specify --last or --first with --leaf')

opts.last = opts.first = opts.leaf

if opts.source == 'djvu':

parser.error('--source=djvu not supported at the moment')

if opts.outdir is None:

opts.outdir = opts.source + '_viz'

if opts.show_opts:

print 'Options: ' + str(opts)

print 'Args: ' + str(args)

sys.exit(0)

parser.destroy()

if not os.path.isdir('./' + opts.outdir + '/'):

os.mkdir('./' + opts.outdir + '/')

id = iarchive.infer_book_id()

iabook = iarchive.Book(id, '', '.')

# scandata = objectify.parse(iabook.get_scandata_path()).getroot()

scandata = iabook.get_scandata()

if opts.source == 'abbyy':

context = etree.iterparse(iabook.get_abbyy(), tag=abbyyns+'page')

elif opts.source == 'pdfxml':

context = etree.iterparse(iabook.get_pdfxml_xml(), tag='PAGE')

elif opts.source == 'djvu':

context = etree.iterparse(iabook.get_djvu_xml(), tag='OBJECT')

if opts.source == 'abbyy':

return (el.get('t') is not None)

elif opts.source == 'pdfxml':

return (el.get('y') is not None)

elif opts.source == 'djvu':

def iget(ltrb):

return int(math.floor(float(el.get(ltrb))))

if opts.source == 'pdfxml':

if ltrb == 't':

return iget('y')

elif ltrb == 'l':

return iget('x')

elif ltrb == 'r':

return iget('x') + iget('width')

elif ltrb == 'b':

return iget('y') + iget('height')

elif opts.source == 'djvu':

raise 'djvuness not yet implemented here'

else: # abbyy

col = color.color[sty['col']]

if sty['width'] == 0:

return

lt, rb = use_coords if use_coords is not None else tag_coords(el, opts.scale)

x1, y1 = lt

x2, y2 = rb

enclosing = None

if use_coords is None:

enclosing = enclosing_tag(el)

if enclosing is not None and enclosing.get('t') is not None:

elt, erb = tag_coords(enclosing, s)

ex1, ey1 = elt

ex2, ey2 = erb

margin = sty['margin']

near = 3

x1 = x1 + margin if x1 - ex1 < near else x1

x2 = x2 - margin if ex2 - x2 < near else x2

y1 = y1 + margin if y1 - ey1 < near else y1

y2 = y2 - margin if ey2 - y2 < near else y2

# x1 += sty['offset']

# y1 += sty['offset']

# x2 += sty['offset']

# y2 += sty['offset']

draw.line([(x1, y1), (x2, y1), (x2, y2), (x1, y2), (x1, y1)],

t = int(tag.get('t'))/s

b = int(tag.get('b'))/s

l = int(tag.get('l'))/s

r = int(tag.get('r'))/s

l = int(tag.get('l'))/s

t = int(tag.get('t'))/s

r = int(tag.get('r'))/s

b = int(tag.get('b'))/s

tag_with_coords = tag

while tag_with_coords is not None:

tag_with_coords = tag.getparent()

if tag_with_coords is not None and tag.get('t') is not None:

break

'block_text' : { 'col':'yellow', 'width':1, 'offset':0, 'margin':10 },

'block_picture' : { 'col':'red', 'width':1, 'offset':7, 'margin':10 },

'block_table' : { 'col':'purple', 'width':10, 'offset':7, 'margin':10 },

'rect' : { 'col':'orange', 'width':1, 'offset':-4, 'margin':10 },

'par' : { 'col':'green', 'width':2, 'offset':0, 'margin':10 },

'line' : { 'col':'blue', 'width':1, 'offset':0, 'margin':10 },

if len(par) > 0:

(o_l, o_t), (o_r, o_b) = tag_coords(par[0], opts.scale)

for line in par:

(l, t), (r, b) = tag_coords(line, opts.scale)

o_l = l if (l - o_l < 0) else o_l

o_t = t if (t - o_t < 0) else o_t

o_r = r if (o_r - r < 0) else o_r

o_b = b if (o_b - b < 0) else o_b

return (o_l, o_t), (o_r, o_b)

else:

book_id = iabook.get_book_id()

scandata_pages = scandata.pageData.page

scandata_ns = iabook.get_scandata_ns()

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:

if opts.first > 0:

if i < opts.first:

i += 1

continue

orig_width = int(page.get('width'))

orig_height = int(page.get('height'))

orig_size = (orig_width, orig_height)

requested_size = (orig_width / opts.scale, orig_height / opts.scale)

image = Image.new('RGB', requested_size)

image_str = iabook.get_page_image(i, requested_size,

out_img_type='ppm',

kdu_reduce=opts.reduce)

page_image = None

if image_str is not None:

page_image = Image.open(StringIO.StringIO(image_str))

if requested_size != page_image.size:

page_image = page_image.resize(requested_size)

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, opts.scale))

image.paste(cropped, four_coords(block, opts.scale))

for block in page:

blocktype = ''

if block.get('blockType') == 'Text':

blocktype = 'Text'

render(draw, block, 'block_text')

if block.get('blockType') == 'Picture':

blocktype = 'Picture'

render(draw, block, 'block_picture')

if block.get('blockType') == 'Table':

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':

for rect in el:

if rect.tag != abyns+'rect':

raise 'found non-rect in region: ' + rect.tag

render(draw, rect, 'rect')

elif el.tag == abyns+'row':

for cell in el:

for text in cell:

draw_text_el(draw, dpi, text)

elif el.tag == abyns+'text':

draw_text_el(draw, dpi, el)

elif (el.tag == abyns+'row'):

raise 'unexpected row el'

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)

page_scandata = iabook.get_page_scandata(i)

if page_scandata is not None:

t = page_scandata.pageType.text

# pageno_string = page_scandata.pageNumber.text

pageno = page_scandata.find(scandata_ns + 'pageNumber')

if pageno:

pageno_string = pageno.text

t += ' ' + pageno_string

handside = page_scandata.find(scandata_ns + 'handSide')

if handside:

t += ' ' + handside.text

f = font.get_font("Courier", dpi / opts.scale, 12)

page_w, page_h = image.size

draw.text((.02 * dpi,

.02 * dpi),

t.encode('utf-8'),

font=f,

fill=color.green)

image.save(opts.outdir + '/img' + scandata_pages[i].get('leafNum').zfill(3) + '.png')

print 'leaf index: ' + str(i)

page.clear()

i += 1

if opts.last > 0:

if i > opts.last:

sys.exit(0)

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,

if page is None:

return False

add = page.find('addToAccessFormats')

if add is None:

add = page.addToAccessFormats

if add is not None and add.text == 'true':

return True

else: