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 __renderChar(self, item):
#if not self.__isLine:
# return
#font = enc(item.font.basefont)
font = enc(item.fontname)
#size = item.get_size()
size = item.size
fontChanged = False
if (self.__font == None or
(self.__font.fullName != font
or self.__font.size != size)) or not self.__hasFont:
# zmienil sie font lub nie jestesmy w obrebie tagu <span> dla fontu
#if self.__font == None:
# print "None",
#else:
# print self.__font.psname,
if self.__hasFont:
if self.__icu != None and self.__inWord:
if not (self.__wordInd - 1) in self.__whites:
self.__outfp.write("</span>")
fontChanged = True
self.__endSpecialTags(self.__font)
self.__outfp.write("</span>")
self.__font = self.__lib.findFont(self.__page,
font).instantiate(size)
#print self.__font.psname
self.__outfp.write("<span style=\"")
name = None
if self.__fontMap != None:
name = self.__fontMap.get(self.__font.name)
if name == None:
name = self.__font.name
self.__outfp.write("font-family: " + self.__font.name)
if self.__font.bold:
self.__outfp.write("; font-weight: bold")
if self.__font.italic:
self.__outfp.write("; font-style: italic")
#self.__outfp.write("; ps-name: " + font)
self.__outfp.write("; font-size: " + str(self.__font.size))
self.__outfp.write("\">")
self.__startSpecialTags(self.__font)
self.__hasFont = True
if self.__icu != None:
if self.__ind in self.__divs or fontChanged: # TODO: I jezeli hasFont bylo False przy wywolaniu renderChar
# i wypisalismy nowy font to wtedy nie moze byc srodek slowa (wiec __ind bedzie w self.__divs bo to 0)
if not self.__wordInd in self.__whites:
self.__outfp.write(
"<span class=\"ocrx_word\" title=\"bbox " + bbox2str(
changeCoords(self.__pagebbox, self.__divbboxes[
self.__wordInd])) + "\">")
self.__wordInd += 1
self.__inWord = True
self.__outfp.write(item.get_text().replace("<", "<").replace(
"&", "&").encode("utf-8"))
if self.__icu != None:
self.__ind += 1
if self.__ind in self.__divs:
if not (self.__wordInd - 1) in self.__whites:
self.__outfp.write("</span>")
self.__inWord = False
def render(item):
if isinstance(item, LTPage):
self.outfp.write('<a id="page_%s" data-bbox="%s" data-rotate="%d"></a>\n' %
(item.pageid, bbox2str(item.bbox), item.rotate))
for child in item:
render(child)
elif isinstance(item, LTLine):
self.outfp.write('<line linewidth="%d" bbox="%s" />\n' %
(item.linewidth, bbox2str(item.bbox)))
elif isinstance(item, LTRect):
self.outfp.write('<rect linewidth="%d" bbox="%s" />\n' %
(item.linewidth, bbox2str(item.bbox)))
elif isinstance(item, LTCurve):
self.outfp.write('<curve linewidth="%d" bbox="%s" pts="%s"/>\n' %
(item.linewidth, bbox2str(item.bbox), item.get_pts()))
elif isinstance(item, LTFigure):
self.outfp.write('<figure name="%s" bbox="%s">\n' %
(item.name, bbox2str(item.bbox)))
for child in item:
render(child)
self.outfp.write('</figure>\n')
elif isinstance(item, LTTextLine):
self.outfp.write('<span data-bbox="%s"/>\n' % bbox2str(item.bbox))
for child in item:
render(child)
# self.outfp.write('</p>\n')
elif isinstance(item, LTTextBox):
wmode = ''
if isinstance(item, LTTextBoxVertical):
wmode = ' wmode="vertical"'
self.outfp.write('<div id="%d" data-bbox="%s" data-wmode="%s"><p>\n' %
(item.index, bbox2str(item.bbox), wmode))
for child in item:
render(child)
self.outfp.write('</p></div>\n')
elif isinstance(item, LTChar):
# self.outfp.write('<text font="%s" bbox="%s" size="%.3f">' %
# (enc(item.fontname), bbox2str(item.bbox), item.size))
self.write_text(item.get_text())
# self.outfp.write('</text>\n')
elif isinstance(item, LTText):
self.outfp.write(item.get_text())
elif isinstance(item, LTImage):
if self.imagewriter is not None:
name = self.imagewriter.export_image(item)
self.outfp.write('<img src="%s" width="%d" height="%d" />\n' %
(enc(name), item.width, item.height))
else:
self.outfp.write('<img width="%d" height="%d" />\n' %
(item.width, item.height))
else:
assert 0, item
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:
char = font.to_unicode(cid)
text += char
except PDFUnicodeNotDefined:
pass
self.outfp.write(enc(text, self.codec))
return
def write_text(self, text, item=None):
if self.ignoring():
self.state.step(u'\n')
return
text = self.CONTROL.sub(u'', text)
if item and text.strip():
self.register_font(item)
if self.has_new_line:
self.handle_new_line()
if self.has_new_chunk:
self.handle_new_chunk()
if text:
self.state.step(unicode(text))
self.handle_style()
self.buffer.write(enc(text, 'utf-8'))
self.last_char = text
return
def write_text(self, text, item=None):
if self.ignoring():
self.state.step(u'\n')
return
text = self.CONTROL.sub(u'', text)
if item and text.strip():
self.register_font(item)
if self.has_new_line:
self.handle_new_line()
if self.has_new_chunk:
self.handle_new_chunk()
if text:
self.state.step(unicode(text))
self.handle_style()
self.buffer.write(enc(text, 'utf-8'))
self.last_char = text
return
def write_header(self):
self.outfp.write('<?xml version="1.0" encoding="%s" ?>\n' % self.codec)
self.outfp.write('<html xmlns="http://www.w3.org/1999/xhtml">\n')
self.outfp.write('<head>\n')
self.outfp.write('<meta charset="%s"/>\n' % self.codec)
if self.document is not None:
contents = self.document.info[0]
if contents is not None:
for name in contents:
if 'itle' in name:
self.outfp.write('<title>%s</title>\n' % contents[name])
else:
self.outfp.write('<meta name="%s" content="%s"/>\n' % (enc(name), contents[name]) )
self.outfp.write('<meta name="Note" content="Converted with PDFminer.py for xhtml format"/>\n')
self.outfp.write('<link rel="stylesheet" type="text/css" href="css/style.css"/>\n')
self.outfp.write('</head>\n')
self.outfp.write('<body>\n')
return
def render(item):
if isinstance(item, LTPage):
self.outfp.write('<page id="%s" bbox="%s" rotate="%d">\n' %
(item.id, item.get_bbox(), item.rotate))
for child in item:
render(child)
self.outfp.write('</page>\n')
elif isinstance(item, LTLine):
self.outfp.write('<line linewidth="%d" direction="%s" bbox="%s" />' % (item.linewidth, item.direction, item.get_bbox()))
elif isinstance(item, LTRect):
self.outfp.write('<rect linewidth="%d" bbox="%s" />' % (item.linewidth, item.get_bbox()))
elif isinstance(item, LTFigure):
self.outfp.write('<figure id="%s">\n' % (item.id))
for child in item:
render(child)
self.outfp.write('</figure>\n')
elif isinstance(item, LTTextLine):
self.outfp.write('<textline bbox="%s">\n' % (item.get_bbox()))
for child in item:
render(child)
self.outfp.write('</textline>\n')
elif isinstance(item, LTTextBox):
self.outfp.write('<textbox id="%s" bbox="%s">\n' % (item.id, item.get_bbox()))
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(),
item.get_bbox(), item.fontsize))
self.write(item.text)
self.outfp.write('</text>\n')
elif isinstance(item, LTText):
self.outfp.write('<text>%s</text>\n' % item.text)
else:
assert 0, item
return
def render(item):
if isinstance(item, LTPage):
metainfo = {
'pid': item.pageid,
'rotate': item.rotate,
'x0': item.bbox[0],
'y0': item.bbox[1],
'x1': item.bbox[2],
'y1': item.bbox[3]
}
self.page = {
'metainfo': metainfo,
'text': [],
'line': [],
'rect': [],
'curve': [],
'figure': [],
'textline': [],
'textbox': [],
'textgroup': [],
'image': []
}
for child in item:
render(child)
if item.groups is not None:
for group in item.groups:
show_group(group)
self.doc.append(self.page)
elif isinstance(item, LTLine):
self.page['line'].append({
'linewidth': item.linewidth,
'x0': item.bbox[0],
'y0': item.bbox[1],
'x1': item.bbox[2],
'y1': item.bbox[3]
})
elif isinstance(item, LTRect):
self.page['rect'].append({
'linewidth': item.linewidth,
'x0': item.bbox[0],
'y0': item.bbox[1],
'x1': item.bbox[2],
'y1': item.bbox[3]
})
elif isinstance(item, LTCurve):
curve = {
'linewidth': item.linewidth,
'pts': item.get_pts(),
'x0': item.bbox[0],
'y0': item.bbox[1],
'x1': item.bbox[2],
'y1': item.bbox[3]
}
self.page['curve'].append(curve)
elif isinstance(item, LTFigure):
self.page['figure'].append({
'name': item.name,
'x0': item.bbox[0],
'y0': item.bbox[1],
'x1': item.bbox[2],
'y1': item.bbox[3]
})
for child in item:
render(child)
elif isinstance(item, LTTextLine):
self.page['textline'].append({
'x0': item.bbox[0],
'y0': item.bbox[1],
'x1': item.bbox[2],
'y1': item.bbox[3]
})
for child in item:
render(child)
elif isinstance(item, LTTextBox):
wmode = 'vertical' if isinstance(
item, LTTextBoxVertical) else 'horizontal'
tb = {
'id': item.index,
'wmode': wmode,
'x0': item.bbox[0],
'y0': item.bbox[1],
'x1': item.bbox[2],
'y1': item.bbox[3]
}
self.page['textbox'].append(tb)
for child in item:
render(child)
elif isinstance(item, LTChar):
# bbox (x0,y0,x1,y1)
# x0: the distance from the left of the page to the left edge of the box.
# y0: the distance from the bottom of the page to the lower edge of the box.
# x1: the distance from the left of the page to the right edge of the box.
# y1: the distance from the bottom of the page to the upper edge of the box.
txt = {
'text': item.get_text(),
'font': enc(item.fontname),
'size': item.size,
'colorspace': item.ncs.name,
'color': json.dumps(item.graphicstate.ncolor),
'x0': item.bbox[0],
'y0': item.bbox[1],
'x1': item.bbox[2],
'y1': item.bbox[3]
}
self.page['text'].append(txt)
elif isinstance(item, LTText):
# LTText is the interface for things that have text.
# LTAnno inherits from LTText.
self.page['text'].append({'text': item.get_text()})
elif isinstance(item, LTImage):
if self.imagewriter is not None:
name = self.imagewriter.export_image(item)
img = {
'src': enc(name),
'width': item.width,
'height': item.height
}
self.page['image'].append(img)
else:
self.page['image'].append({
'width': item.width,
'height': item.height
})
else:
assert False, str(('Unhandled', item))
return
def render(item):
if isinstance(item, LTPage):
s = '<page id="%s" bbox="%s" rotate="%d">\n' % (
item.pageid,
bbox2str(item.bbox),
item.rotate,
)
self.current_page = RpaPdfPage(item.pageid, item.bbox,
item.rotate)
self.write(s)
for child in item:
render(child)
if item.groups is not None:
self.write("<layout>\n")
for group in item.groups:
show_group(group)
self.write("</layout>\n")
self.write("</page>\n")
self.rpa_pdf_document.add_page(self.current_page)
elif isinstance(item, LTLine):
s = '<line linewidth="%d" bbox="%s" />\n' % (
item.linewidth,
bbox2str(item.bbox),
)
self.write(s)
elif isinstance(item, LTRect):
s = '<rect linewidth="%d" bbox="%s" />\n' % (
item.linewidth,
bbox2str(item.bbox),
)
self.write(s)
elif isinstance(item, LTCurve):
s = '<curve linewidth="%d" bbox="%s" pts="%s"/>\n' % (
item.linewidth,
bbox2str(item.bbox),
item.get_pts(),
)
self.write(s)
elif isinstance(item, LTFigure):
s = '<figure name="%s" bbox="%s">\n' % (item.name,
bbox2str(item.bbox))
self.write(s)
self.figure = RpaFigure(item.name, item.bbox)
for child in item:
self.figure.set_item(item)
render(child)
self.write("</figure>\n")
self.current_page.add_content(self.figure)
self.figure = None
elif isinstance(item, LTTextLine):
self.write('<textline bbox="%s">\n' % bbox2str(item.bbox))
for child in item:
render(child)
self.write("</textline>\n")
elif isinstance(item, LTTextBox):
wmode = ""
if isinstance(item, LTTextBoxVertical):
wmode = ' wmode="vertical"'
s = '<textbox id="%d" bbox="%s"%s>\n' % (
item.index,
bbox2str(item.bbox),
wmode,
)
box = RpaTextBox(item.index, item.bbox, wmode)
self.write(s)
box.set_item(item)
self.current_page.add_content(box)
for child in item:
render(child)
self.write("</textbox>\n")
elif isinstance(item, LTChar):
s = ('<text font="%s" bbox="%s" colourspace="%s" '
'ncolour="%s" size="%.3f">' % (
enc(item.fontname),
bbox2str(item.bbox),
item.ncs.name,
item.graphicstate.ncolor,
item.size,
))
self.write(s)
self.write_text(item.get_text())
self.write("</text>\n")
elif isinstance(item, LTText):
self.write("<text>%s</text>\n" % item.get_text())
elif isinstance(item, LTImage):
if self.figure:
self.figure.set_item(item)
if self.imagewriter is not None:
name = self.imagewriter.export_image(item)
self.write('<image src="%s" width="%d" height="%d" />\n' %
(enc(name), item.width, item.height))
else:
self.write('<image width="%d" height="%d" />\n' %
(item.width, item.height))
else:
assert False, str(("Unhandled", item))
def encode_text(self):
for textline in self.taglists['textline']:
textline["text"] = subst_control_chars(enc(textline["text"]))
def write_text(self, text: str):
if self.stripcontrol:
text = self.CONTROL.sub("", text)
self.write(enc(text))
def write_text(self, text):
if self.stripcontrol:
text = self.CONTROL.sub(u'', text)
self.outfp.write(enc(text, self.codec))
return
def end_tag(self):
assert self.tag
self.outfp.write('</%s>' % enc(self.tag.name))
self.tag = None
return
def write(self, text): self.outfp.write(enc(text, self.codec)) return
def render(item):
self.__stack.append(self.__node)
parent = self.__node
if isinstance(item, LTPage):
self.__num += 1
self.__page = PDFMinerNode("page")
self.__page.setPageId(self.__num)
self.__setBbox(self.__page, bbox2str(normalize(item.bbox)))
if self.__lib != None:
self.__lib.addBbox(self.__page.getPageId(),
self.__page.getBbox())
self.__node = self.__page
for child in item:
render(child)
elif isinstance(item, LTLine):
pass
elif isinstance(item, LTRect):
pass
elif isinstance(item, LTCurve):
pass
elif isinstance(item, LTFigure):
pass
elif isinstance(item, LTTextLine):
#print "textline"
self.__node = PDFMinerNode("textline")
self.__node.setPageId(self.__num)
parent.add(self.__node)
self.__setBbox(self.__node, bbox2str(normalize(item.bbox)))
for child in item:
render(child)
elif isinstance(item, LTTextBox):
self.__node = PDFMinerNode("textbox", item.index)
self.__node.setPageId(self.__num)
parent.add(self.__node)
self.__setBbox(self.__node, bbox2str(normalize(item.bbox)))
for child in item:
render(child)
elif isinstance(item, LTChar):
#font = enc(item.font.fontname)
font = enc(item.fontname)
#size = item.get_size()
size = item.size
self.__font = self.__fontDict.get(font + str(size))
if self.__font == None:
self.__font = self.__fontDict.setdefault(
font + str(size),
self.__lib.findFont(self.__pdfminerpage,
font).instantiate(size))
self.__node = PDFMinerNode("text")
self.__node.setPageId(self.__num)
self.__node.setLeaf()
#self.__node.setContentType("Text")
parent.add(self.__node)
#print parent.textOf()
#assert(parent.textOf() == "textline")
self.__setBbox(self.__node, bbox2str(normalize(item.bbox)))
if self.__font != None:
self.__node.add(self.__font)
#assert(self.__node.textOf() == "text")
self.__node.add(item.get_text())
elif isinstance(item, LTText):
pass
# TODO: NOTE ignorujemy tekst pusty (tu byly same spacje)
#self.outfp.write('<text>%s</text>\n' % item.get_text())
elif isinstance(item, LTImage):
pass
elif isinstance(item, LTTextGroup):
self.__node = PDFMinerNode("textgroup")
self.__node.setPageId(self.__num)
parent.add(self.__node)
self.__setBbox(self.__node, bbox2str(normalize(item.bbox)))
for child in item:
render(child)
else:
assert 0, item
self.__node = self.__stack.pop()
return
def write_text(self, text):
if self.stripcontrol:
text = self.CONTROL.sub('', text)
self.write(enc(text))
return
def render(item):
if isinstance(item, LTPage):
self.current_page = Page(item.pageid, item.bbox, item.rotate)
self.write(self.current_page.tag + "\n")
for child in item:
render(child)
if item.groups is not None:
self.write("<layout>\n")
for group in item.groups:
show_group(group)
self.write("</layout>\n")
self.write("</page>\n")
self.active_pdf_document.add_page(self.current_page)
elif isinstance(item, LTLine):
s = '<line linewidth="%d" bbox="%s" />\n' % (
item.linewidth,
bbox2str(item.bbox),
)
self.write(s)
elif isinstance(item, LTRect):
s = '<rect linewidth="%d" bbox="%s" />\n' % (
item.linewidth,
bbox2str(item.bbox),
)
self.write(s)
elif isinstance(item, LTCurve):
s = '<curve linewidth="%d" bbox="%s" pts="%s"/>\n' % (
item.linewidth,
bbox2str(item.bbox),
item.get_pts(),
)
self.write(s)
elif isinstance(item, LTFigure):
figure = Figure(item)
s = '<figure name="%s" bbox="%s">\n' % (
item.name,
bbox2str(item.bbox),
)
self.write(s)
for child in item:
render(child)
self.write("</figure>\n")
self._add_unique_figure(figure)
elif isinstance(item, LTTextLine):
self.write('<textline bbox="%s">\n' % bbox2str(item.bbox))
for child in item:
render(child)
self.write("</textline>\n")
elif isinstance(item, LTTextBox):
wmode = ""
if isinstance(item, LTTextBoxVertical):
wmode = ' wmode="vertical"'
s = '<textbox id="%d" bbox="%s"%s>\n' % (
item.index,
bbox2str(item.bbox),
wmode,
)
box = TextBox(item.index, item=item, trim=self.trim)
self.write(s)
self.current_page.add_content(box)
for child in item:
render(child)
self.write("</textbox>\n")
elif isinstance(item, LTChar):
s = ('<text font="%s" bbox="%s" colourspace="%s" '
'ncolour="%s" size="%.3f">' % (
enc(item.fontname),
bbox2str(item.bbox),
item.ncs.name,
item.graphicstate.ncolor,
item.size,
))
self.write(s)
self.write_text(item.get_text())
self.write("</text>\n")
elif isinstance(item, LTText):
self.write("<text>%s</text>\n" % item.get_text())
elif isinstance(item, LTImage):
figure = Figure(item)
if self.imagewriter is not None:
name = self.imagewriter.export_image(item)
self.write('<image src="%s" width="%d" height="%d" />\n' %
(enc(name), item.width, item.height))
else:
self.write('<image width="%d" height="%d" />\n' %
(item.width, item.height))
self._add_unique_figure(figure)
else:
self._logger.warning("Unknown item: %r", item)
