2007-12-31 03:41:45 +00:00
|
|
|
#!/usr/bin/env python
|
2010-01-30 07:30:01 +00:00
|
|
|
import sys, os.path
|
2009-11-03 13:39:34 +00:00
|
|
|
from pdfdevice import PDFDevice, PDFTextDevice
|
2009-11-03 01:27:30 +00:00
|
|
|
from pdffont import PDFUnicodeNotDefined
|
2010-01-30 07:30:01 +00:00
|
|
|
from pdftypes import LITERALS_DCT_DECODE
|
2009-11-03 01:27:30 +00:00
|
|
|
from layout import LayoutContainer
|
2009-12-20 02:38:01 +00:00
|
|
|
from layout import LTPage, LTText, LTLine, LTRect, LTPolygon
|
2010-01-30 07:30:01 +00:00
|
|
|
from layout import LTFigure, LTImage, LTTextItem, LTTextBox, LTTextLine
|
2009-11-03 13:39:34 +00:00
|
|
|
from utils import apply_matrix_pt, mult_matrix
|
2010-01-31 02:09:28 +00:00
|
|
|
from utils import enc, strbbox
|
2009-07-23 14:03:58 +00:00
|
|
|
|
|
|
|
|
|
|
|
## TagExtractor
|
|
|
|
##
|
|
|
|
class TagExtractor(PDFDevice):
|
|
|
|
|
2009-10-24 04:41:59 +00:00
|
|
|
def __init__(self, rsrc, outfp, codec='utf-8'):
|
|
|
|
PDFDevice.__init__(self, rsrc)
|
|
|
|
self.outfp = outfp
|
|
|
|
self.codec = codec
|
|
|
|
self.pageno = 0
|
|
|
|
self.tag = None
|
|
|
|
return
|
|
|
|
|
|
|
|
def render_string(self, textstate, seq):
|
|
|
|
font = textstate.font
|
|
|
|
text = ''
|
|
|
|
for obj in seq:
|
|
|
|
if not isinstance(obj, str): continue
|
|
|
|
chars = font.decode(obj)
|
|
|
|
for cid in chars:
|
|
|
|
try:
|
2009-12-19 14:17:00 +00:00
|
|
|
char = font.to_unichr(cid)
|
2009-10-24 04:41:59 +00:00
|
|
|
text += char
|
|
|
|
except PDFUnicodeNotDefined:
|
|
|
|
pass
|
|
|
|
self.outfp.write(enc(text, self.codec))
|
|
|
|
return
|
|
|
|
|
|
|
|
def begin_page(self, page, ctm):
|
|
|
|
self.outfp.write('<page id="%s" bbox="%s" rotate="%d">' %
|
2010-01-31 02:09:28 +00:00
|
|
|
(self.pageno, strbbox(page.mediabox), page.rotate))
|
2009-10-24 04:41:59 +00:00
|
|
|
return
|
|
|
|
|
|
|
|
def end_page(self, page):
|
|
|
|
self.outfp.write('</page>\n')
|
|
|
|
self.pageno += 1
|
|
|
|
return
|
|
|
|
|
|
|
|
def begin_tag(self, tag, props=None):
|
|
|
|
s = ''
|
|
|
|
if props:
|
|
|
|
s = ''.join( ' %s="%s"' % (enc(k), enc(str(v))) for (k,v)
|
|
|
|
in sorted(props.iteritems()) )
|
|
|
|
self.outfp.write('<%s%s>' % (enc(tag.name), s))
|
|
|
|
self.tag = tag
|
|
|
|
return
|
|
|
|
|
|
|
|
def end_tag(self):
|
|
|
|
assert self.tag
|
|
|
|
self.outfp.write('</%s>' % enc(self.tag.name))
|
|
|
|
self.tag = None
|
|
|
|
return
|
|
|
|
|
|
|
|
def do_tag(self, tag, props=None):
|
|
|
|
self.begin_tag(tag, props)
|
|
|
|
self.tag = None
|
|
|
|
return
|
2008-04-26 06:47:56 +00:00
|
|
|
|
2009-05-15 14:25:32 +00:00
|
|
|
|
|
|
|
## PDFPageAggregator
|
|
|
|
##
|
2009-07-23 14:03:58 +00:00
|
|
|
class PDFPageAggregator(PDFTextDevice):
|
2009-05-15 14:25:32 +00:00
|
|
|
|
2009-10-24 04:41:59 +00:00
|
|
|
def __init__(self, rsrc, pageno=1, laparams=None):
|
|
|
|
PDFTextDevice.__init__(self, rsrc)
|
|
|
|
self.laparams = laparams
|
|
|
|
self.pageno = pageno
|
|
|
|
self.stack = []
|
|
|
|
return
|
|
|
|
|
|
|
|
def begin_page(self, page, ctm):
|
|
|
|
(x0,y0,x1,y1) = page.mediabox
|
|
|
|
(x0,y0) = apply_matrix_pt(ctm, (x0,y0))
|
|
|
|
(x1,y1) = apply_matrix_pt(ctm, (x1,y1))
|
|
|
|
mediabox = (0, 0, abs(x0-x1), abs(y0-y1))
|
|
|
|
self.cur_item = LTPage(self.pageno, mediabox)
|
|
|
|
return
|
|
|
|
|
|
|
|
def end_page(self, _):
|
|
|
|
assert not self.stack
|
|
|
|
assert isinstance(self.cur_item, LTPage)
|
|
|
|
self.cur_item.fixate()
|
|
|
|
if self.laparams:
|
|
|
|
self.cur_item.analyze_layout(self.laparams)
|
|
|
|
self.pageno += 1
|
|
|
|
return self.cur_item
|
|
|
|
|
|
|
|
def begin_figure(self, name, bbox, matrix):
|
|
|
|
self.stack.append(self.cur_item)
|
|
|
|
self.cur_item = LTFigure(name, bbox, mult_matrix(matrix, self.ctm))
|
|
|
|
return
|
|
|
|
|
|
|
|
def end_figure(self, _):
|
|
|
|
fig = self.cur_item
|
|
|
|
self.cur_item.fixate()
|
|
|
|
self.cur_item = self.stack.pop()
|
|
|
|
self.cur_item.add(fig)
|
|
|
|
return
|
|
|
|
|
2010-01-30 07:30:01 +00:00
|
|
|
def render_image(self, name, stream):
|
|
|
|
assert isinstance(self.cur_item, LTFigure)
|
|
|
|
item = LTImage(name, stream['Filter'],
|
|
|
|
(stream['Width'], stream['Height']),
|
|
|
|
(self.cur_item.x0, self.cur_item.y0,
|
|
|
|
self.cur_item.x1, self.cur_item.y1),
|
|
|
|
stream.get_rawdata())
|
|
|
|
self.cur_item.add(item)
|
|
|
|
return
|
|
|
|
|
2009-10-24 04:41:59 +00:00
|
|
|
def paint_path(self, gstate, stroke, fill, evenodd, path):
|
|
|
|
shape = ''.join(x[0] for x in path)
|
2010-01-04 12:50:59 +00:00
|
|
|
if shape == 'ml':
|
|
|
|
# horizontal/vertical line
|
2009-10-24 04:41:59 +00:00
|
|
|
(_,x0,y0) = path[0]
|
|
|
|
(_,x1,y1) = path[1]
|
|
|
|
(x0,y0) = apply_matrix_pt(self.ctm, (x0,y0))
|
|
|
|
(x1,y1) = apply_matrix_pt(self.ctm, (x1,y1))
|
2009-12-20 02:38:01 +00:00
|
|
|
self.cur_item.add(LTLine(gstate.linewidth, (x0,y0), (x1,y1)))
|
2009-10-24 04:41:59 +00:00
|
|
|
elif shape == 'mlllh':
|
|
|
|
# rectangle
|
|
|
|
(_,x0,y0) = path[0]
|
|
|
|
(_,x1,y1) = path[1]
|
|
|
|
(_,x2,y2) = path[2]
|
|
|
|
(_,x3,y3) = path[3]
|
|
|
|
(x0,y0) = apply_matrix_pt(self.ctm, (x0,y0))
|
|
|
|
(x1,y1) = apply_matrix_pt(self.ctm, (x1,y1))
|
|
|
|
(x2,y2) = apply_matrix_pt(self.ctm, (x2,y2))
|
|
|
|
(x3,y3) = apply_matrix_pt(self.ctm, (x3,y3))
|
|
|
|
if ((x0 == x1 and y1 == y2 and x2 == x3 and y3 == y0) or
|
|
|
|
(y0 == y1 and x1 == x2 and y2 == y3 and x3 == x0)):
|
|
|
|
self.cur_item.add(LTRect(gstate.linewidth, (x0,y0,x2,y2)))
|
2009-12-20 02:38:01 +00:00
|
|
|
else:
|
|
|
|
# other polygon
|
|
|
|
pts = []
|
|
|
|
for p in path:
|
|
|
|
for i in xrange(1, len(p), 2):
|
|
|
|
pts.append(apply_matrix_pt(self.ctm, (p[i], p[i+1])))
|
|
|
|
self.cur_item.add(LTPolygon(gstate.linewidth, pts))
|
2009-10-24 04:41:59 +00:00
|
|
|
return
|
|
|
|
|
|
|
|
def render_chars(self, matrix, font, fontsize, charspace, scaling, chars):
|
|
|
|
if not chars: return (0, 0)
|
|
|
|
item = LTTextItem(matrix, font, fontsize, charspace, scaling, chars)
|
|
|
|
self.cur_item.add(item)
|
|
|
|
return item.adv
|
2009-05-15 14:25:32 +00:00
|
|
|
|
2008-06-23 13:22:27 +00:00
|
|
|
|
2009-03-28 17:23:53 +00:00
|
|
|
## PDFConverter
|
2009-05-05 12:26:29 +00:00
|
|
|
##
|
2009-03-28 17:23:53 +00:00
|
|
|
class PDFConverter(PDFPageAggregator):
|
2009-10-24 04:41:59 +00:00
|
|
|
|
|
|
|
def __init__(self, rsrc, outfp, codec='utf-8', pageno=1, laparams=None):
|
|
|
|
PDFPageAggregator.__init__(self, rsrc, pageno=pageno, laparams=laparams)
|
|
|
|
self.outfp = outfp
|
|
|
|
self.codec = codec
|
|
|
|
return
|
|
|
|
|
|
|
|
def write(self, text):
|
|
|
|
self.outfp.write(enc(text, self.codec))
|
|
|
|
return
|
|
|
|
|
|
|
|
|
2009-10-31 03:04:56 +00:00
|
|
|
## XMLConverter
|
2008-07-27 04:30:37 +00:00
|
|
|
##
|
2009-10-31 03:04:56 +00:00
|
|
|
class XMLConverter(PDFConverter):
|
2008-07-27 04:30:37 +00:00
|
|
|
|
2010-01-31 02:09:28 +00:00
|
|
|
def __init__(self, rsrc, outfp, codec='utf-8', pageno=1, laparams=None, outdir=None):
|
2009-10-31 03:04:56 +00:00
|
|
|
PDFConverter.__init__(self, rsrc, outfp, codec=codec, pageno=pageno, laparams=laparams)
|
2010-01-31 02:09:28 +00:00
|
|
|
self.outdir = outdir
|
2009-10-31 03:04:56 +00:00
|
|
|
self.outfp.write('<?xml version="1.0" encoding="%s" ?>\n' % codec)
|
|
|
|
self.outfp.write('<pages>\n')
|
|
|
|
return
|
2010-01-30 07:30:01 +00:00
|
|
|
|
|
|
|
def write_image(self, image):
|
|
|
|
if image.type in LITERALS_DCT_DECODE:
|
|
|
|
ext = '.jpg'
|
|
|
|
else:
|
|
|
|
return None
|
|
|
|
name = image.name+ext
|
2010-01-31 02:09:28 +00:00
|
|
|
path = os.path.join(self.outdir, name)
|
2010-01-30 07:30:01 +00:00
|
|
|
fp = file(path, 'wb')
|
|
|
|
fp.write(image.data)
|
|
|
|
fp.close()
|
|
|
|
return name
|
2009-10-31 03:04:56 +00:00
|
|
|
|
2009-10-24 04:41:59 +00:00
|
|
|
def end_page(self, page):
|
|
|
|
def render(item):
|
|
|
|
if isinstance(item, LTPage):
|
|
|
|
self.outfp.write('<page id="%s" bbox="%s" rotate="%d">\n' %
|
2010-01-31 02:09:28 +00:00
|
|
|
(item.id, strbbox(item.bbox), item.rotate))
|
2009-10-24 04:41:59 +00:00
|
|
|
for child in item:
|
|
|
|
render(child)
|
|
|
|
self.outfp.write('</page>\n')
|
2009-12-20 02:38:01 +00:00
|
|
|
elif isinstance(item, LTLine) and item.direction:
|
2010-01-31 02:09:28 +00:00
|
|
|
self.outfp.write('<line linewidth="%d" direction="%s" bbox="%s" />\n' % (item.linewidth, item.direction, strbbox(item.bbox)))
|
2009-10-24 04:41:59 +00:00
|
|
|
elif isinstance(item, LTRect):
|
2010-01-31 02:09:28 +00:00
|
|
|
self.outfp.write('<rect linewidth="%d" bbox="%s" />\n' % (item.linewidth, strbbox(item.bbox)))
|
2009-12-20 02:38:01 +00:00
|
|
|
elif isinstance(item, LTPolygon):
|
2010-01-31 02:09:28 +00:00
|
|
|
self.outfp.write('<polygon linewidth="%d" bbox="%s" pts="%s"/>\n' % (item.linewidth, strbbox(item.bbox), item.get_pts()))
|
2009-10-24 04:41:59 +00:00
|
|
|
elif isinstance(item, LTFigure):
|
2010-01-31 02:09:28 +00:00
|
|
|
self.outfp.write('<figure id="%s" bbox="%s">\n' % (item.id, strbbox(item.bbox)))
|
2009-10-24 04:41:59 +00:00
|
|
|
for child in item:
|
|
|
|
render(child)
|
|
|
|
self.outfp.write('</figure>\n')
|
|
|
|
elif isinstance(item, LTTextLine):
|
2010-01-31 02:09:28 +00:00
|
|
|
self.outfp.write('<textline bbox="%s">\n' % strbbox(item.bbox))
|
2009-10-24 04:41:59 +00:00
|
|
|
for child in item:
|
|
|
|
render(child)
|
|
|
|
self.outfp.write('</textline>\n')
|
|
|
|
elif isinstance(item, LTTextBox):
|
2010-01-31 02:09:28 +00:00
|
|
|
self.outfp.write('<textbox id="%s" bbox="%s">\n' % (item.id, strbbox(item.bbox)))
|
2009-10-24 04:41:59 +00:00
|
|
|
for child in item:
|
|
|
|
render(child)
|
|
|
|
self.outfp.write('</textbox>\n')
|
|
|
|
elif isinstance(item, LTTextItem):
|
|
|
|
self.outfp.write('<text font="%s" vertical="%s" bbox="%s" fontsize="%.3f">' %
|
|
|
|
(enc(item.font.fontname), item.is_vertical(),
|
2010-01-31 02:09:28 +00:00
|
|
|
strbbox(item.bbox), item.fontsize))
|
2009-10-24 04:41:59 +00:00
|
|
|
self.write(item.text)
|
|
|
|
self.outfp.write('</text>\n')
|
|
|
|
elif isinstance(item, LTText):
|
|
|
|
self.outfp.write('<text>%s</text>\n' % item.text)
|
2010-01-30 07:30:01 +00:00
|
|
|
elif isinstance(item, LTImage):
|
|
|
|
x = ''
|
2010-01-31 02:09:28 +00:00
|
|
|
if self.outdir:
|
2010-01-30 07:30:01 +00:00
|
|
|
name = self.write_image(item)
|
|
|
|
if name:
|
|
|
|
x = 'name="%s" ' % enc(name)
|
|
|
|
self.outfp.write('<image %stype="%s" width="%d" height="%d" />\n' % (x, item.type, item.width, item.height))
|
2009-10-24 04:41:59 +00:00
|
|
|
else:
|
|
|
|
assert 0, item
|
|
|
|
return
|
|
|
|
page = PDFConverter.end_page(self, page)
|
|
|
|
render(page)
|
|
|
|
return
|
2008-06-23 13:22:27 +00:00
|
|
|
|
2009-10-31 03:04:56 +00:00
|
|
|
def close(self):
|
|
|
|
self.outfp.write('</pages>\n')
|
|
|
|
return
|
|
|
|
|
2008-07-27 04:30:37 +00:00
|
|
|
|
|
|
|
## HTMLConverter
|
|
|
|
##
|
2009-03-28 17:23:53 +00:00
|
|
|
class HTMLConverter(PDFConverter):
|
2008-07-27 04:30:37 +00:00
|
|
|
|
2009-10-24 04:41:59 +00:00
|
|
|
def __init__(self, rsrc, outfp, codec='utf-8', pageno=1, laparams=None,
|
2010-01-31 02:09:28 +00:00
|
|
|
scale=1, showpageno=True, pagepad=50, outdir=None):
|
2009-10-24 04:41:59 +00:00
|
|
|
PDFConverter.__init__(self, rsrc, outfp, codec=codec, pageno=pageno, laparams=laparams)
|
|
|
|
self.showpageno = showpageno
|
|
|
|
self.pagepad = pagepad
|
2010-01-31 02:09:28 +00:00
|
|
|
self.outdir = outdir
|
2009-10-24 04:41:59 +00:00
|
|
|
self.scale = scale
|
|
|
|
self.outfp.write('<html><head>\n')
|
|
|
|
self.outfp.write('<meta http-equiv="Content-Type" content="text/html; charset=%s">\n' %
|
|
|
|
self.codec)
|
|
|
|
self.outfp.write('</head><body>\n')
|
|
|
|
self.yoffset = self.pagepad
|
|
|
|
return
|
|
|
|
|
|
|
|
def write_rect(self, color, width, x, y, w, h):
|
|
|
|
self.outfp.write('<span style="position:absolute; border: %s %dpx solid; '
|
|
|
|
'left:%dpx; top:%dpx; width:%dpx; height:%dpx;"></span>\n' %
|
|
|
|
(color, width, x*self.scale, y*self.scale, w*self.scale, h*self.scale))
|
|
|
|
return
|
|
|
|
|
2010-01-30 07:30:01 +00:00
|
|
|
def write_image(self, image):
|
|
|
|
if image.type in LITERALS_DCT_DECODE:
|
|
|
|
ext = '.jpg'
|
|
|
|
else:
|
|
|
|
return
|
|
|
|
name = image.name+ext
|
2010-01-31 02:09:28 +00:00
|
|
|
path = os.path.join(self.outdir, name)
|
2010-01-30 07:30:01 +00:00
|
|
|
fp = file(path, 'wb')
|
|
|
|
fp.write(image.data)
|
|
|
|
fp.close()
|
|
|
|
self.outfp.write('<img src="%s" style="position:absolute; left:%dpx; top:%dpx;" '
|
|
|
|
'width="%d" height="%d" />\n' %
|
|
|
|
(enc(name),
|
2010-01-31 02:09:28 +00:00
|
|
|
image.x0*self.scale, (self.yoffset-image.y1)*self.scale,
|
|
|
|
image.width*self.scale, image.height*self.scale))
|
2010-01-30 07:30:01 +00:00
|
|
|
return
|
|
|
|
|
2009-10-24 04:41:59 +00:00
|
|
|
def end_page(self, page):
|
|
|
|
def render(item):
|
|
|
|
if isinstance(item, LTPage):
|
|
|
|
self.yoffset += item.y1
|
|
|
|
self.write_rect('gray', 1, item.x0, self.yoffset-item.y1, item.width, item.height)
|
|
|
|
if self.showpageno:
|
|
|
|
self.outfp.write('<div style="position:absolute; top:%dpx;">' %
|
|
|
|
((self.yoffset-item.y1)*self.scale))
|
|
|
|
self.outfp.write('<a name="%s">Page %s</a></div>\n' % (page.id, page.id))
|
|
|
|
for child in item:
|
|
|
|
render(child)
|
|
|
|
elif isinstance(item, LTTextItem):
|
|
|
|
if item.vertical:
|
|
|
|
wmode = 'tb-rl'
|
|
|
|
else:
|
|
|
|
wmode = 'lr-tb'
|
|
|
|
self.outfp.write('<span style="position:absolute; writing-mode:%s;'
|
|
|
|
' left:%dpx; top:%dpx; font-size:%dpx;">' %
|
|
|
|
(wmode, item.x0*self.scale, (self.yoffset-item.y1)*self.scale,
|
|
|
|
item.fontsize*self.scale))
|
|
|
|
self.write(item.text)
|
|
|
|
self.outfp.write('</span>\n')
|
|
|
|
if self.debug:
|
|
|
|
self.write_rect('red', 1, item.x0, self.yoffset-item.y1, item.width, item.height)
|
2009-12-20 02:38:01 +00:00
|
|
|
elif isinstance(item, LTPolygon):
|
2009-10-24 04:41:59 +00:00
|
|
|
self.write_rect('black', 1, item.x0, self.yoffset-item.y1, item.width, item.height)
|
|
|
|
elif isinstance(item, LTTextLine):
|
|
|
|
for child in item:
|
|
|
|
render(child)
|
|
|
|
elif isinstance(item, LTTextBox):
|
|
|
|
self.write_rect('blue', 1, item.x0, self.yoffset-item.y1, item.width, item.height)
|
|
|
|
for child in item:
|
|
|
|
render(child)
|
|
|
|
elif isinstance(item, LTFigure):
|
|
|
|
self.write_rect('green', 1, item.x0, self.yoffset-item.y1, item.width, item.height)
|
|
|
|
for child in item:
|
|
|
|
render(child)
|
2010-01-30 07:30:01 +00:00
|
|
|
elif isinstance(item, LTImage):
|
2010-01-31 02:09:28 +00:00
|
|
|
if self.outdir:
|
2010-01-30 07:30:01 +00:00
|
|
|
self.write_image(item)
|
2009-10-24 04:41:59 +00:00
|
|
|
return
|
|
|
|
page = PDFConverter.end_page(self, page)
|
|
|
|
render(page)
|
|
|
|
self.yoffset += self.pagepad
|
|
|
|
return
|
|
|
|
|
|
|
|
def close(self):
|
|
|
|
self.outfp.write('<div style="position:absolute; top:0px;">Page: %s</div>\n' %
|
|
|
|
', '.join('<a href="#%s">%s</a>' % (i,i) for i in xrange(1,self.pageno)))
|
|
|
|
self.outfp.write('</body></html>\n')
|
|
|
|
return
|
2008-07-27 04:30:37 +00:00
|
|
|
|
|
|
|
|
2009-03-28 17:23:53 +00:00
|
|
|
## TextConverter
|
|
|
|
##
|
|
|
|
class TextConverter(PDFConverter):
|
|
|
|
|
2009-10-24 04:41:59 +00:00
|
|
|
def __init__(self, rsrc, outfp, codec='utf-8', pageno=1, laparams=None,
|
|
|
|
showpageno=False):
|
|
|
|
PDFConverter.__init__(self, rsrc, outfp, codec=codec, pageno=pageno, laparams=laparams)
|
|
|
|
self.showpageno = showpageno
|
|
|
|
return
|
|
|
|
|
|
|
|
def write(self, text):
|
|
|
|
self.outfp.write(text.encode(self.codec, 'ignore'))
|
|
|
|
return
|
|
|
|
|
|
|
|
def end_page(self, page):
|
|
|
|
def render(item):
|
|
|
|
if isinstance(item, LTText):
|
|
|
|
self.write(item.text)
|
|
|
|
elif isinstance(item, LayoutContainer):
|
|
|
|
for child in item:
|
|
|
|
render(child)
|
|
|
|
if isinstance(item, LTTextBox):
|
|
|
|
self.write('\n')
|
|
|
|
page = PDFConverter.end_page(self, page)
|
|
|
|
if self.showpageno:
|
|
|
|
self.write('Page %d\n' % page.id)
|
|
|
|
render(page)
|
|
|
|
self.write('\f')
|
|
|
|
return
|