Send non-stroke color to XML conversion
Inspired by https://github.com/euske/pdfminer/pull/158 from @andruo11 and https://github.com/euske/pdfminer/pull/197 from @staccatosound.pull/129/head
parent
b6c63bedc6
commit
2231f0892e
|
@ -112,7 +112,7 @@ class PDFLayoutAnalyzer(PDFTextDevice):
|
||||||
evenodd, gstate.scolor, gstate.ncolor))
|
evenodd, gstate.scolor, gstate.ncolor))
|
||||||
return
|
return
|
||||||
|
|
||||||
def render_char(self, matrix, font, fontsize, scaling, rise, cid):
|
def render_char(self, matrix, font, fontsize, scaling, rise, cid, ncs, graphicstate):
|
||||||
try:
|
try:
|
||||||
text = font.to_unichr(cid)
|
text = font.to_unichr(cid)
|
||||||
assert isinstance(text, six.text_type), str(type(text))
|
assert isinstance(text, six.text_type), str(type(text))
|
||||||
|
@ -120,7 +120,7 @@ class PDFLayoutAnalyzer(PDFTextDevice):
|
||||||
text = self.handle_undefined_char(font, cid)
|
text = self.handle_undefined_char(font, cid)
|
||||||
textwidth = font.char_width(cid)
|
textwidth = font.char_width(cid)
|
||||||
textdisp = font.char_disp(cid)
|
textdisp = font.char_disp(cid)
|
||||||
item = LTChar(matrix, font, fontsize, scaling, rise, text, textwidth, textdisp)
|
item = LTChar(matrix, font, fontsize, scaling, rise, text, textwidth, textdisp, ncs, graphicstate)
|
||||||
self.cur_item.add(item)
|
self.cur_item.add(item)
|
||||||
return item.adv
|
return item.adv
|
||||||
|
|
||||||
|
@ -520,8 +520,9 @@ class XMLConverter(PDFConverter):
|
||||||
render(child)
|
render(child)
|
||||||
self.write('</textbox>\n')
|
self.write('</textbox>\n')
|
||||||
elif isinstance(item, LTChar):
|
elif isinstance(item, LTChar):
|
||||||
self.write('<text font="%s" bbox="%s" size="%.3f">' %
|
self.write('<text font="%s" bbox="%s" colourspace="%s" ncolour=%s" size="%.3f">' %
|
||||||
(enc(item.fontname, None), bbox2str(item.bbox), item.size))
|
(enc(item.fontname, None), bbox2str(item.bbox),
|
||||||
|
item.ncs.name, item.graphicstate.ncolor, item.size))
|
||||||
self.write_text(item.get_text())
|
self.write_text(item.get_text())
|
||||||
self.write('</text>\n')
|
self.write('</text>\n')
|
||||||
elif isinstance(item, LTText):
|
elif isinstance(item, LTText):
|
||||||
|
|
|
@ -228,11 +228,13 @@ class LTAnno(LTItem, LTText):
|
||||||
class LTChar(LTComponent, LTText):
|
class LTChar(LTComponent, LTText):
|
||||||
|
|
||||||
def __init__(self, matrix, font, fontsize, scaling, rise,
|
def __init__(self, matrix, font, fontsize, scaling, rise,
|
||||||
text, textwidth, textdisp):
|
text, textwidth, textdisp, ncs, graphicstate):
|
||||||
LTText.__init__(self)
|
LTText.__init__(self)
|
||||||
self._text = text
|
self._text = text
|
||||||
self.matrix = matrix
|
self.matrix = matrix
|
||||||
self.fontname = font.fontname
|
self.fontname = font.fontname
|
||||||
|
self.ncs = ncs
|
||||||
|
self.graphicstate = graphicstate
|
||||||
self.adv = textwidth * fontsize * scaling
|
self.adv = textwidth * fontsize * scaling
|
||||||
# compute the boundary rectangle.
|
# compute the boundary rectangle.
|
||||||
if font.is_vertical():
|
if font.is_vertical():
|
||||||
|
|
|
@ -63,7 +63,7 @@ class PDFDevice(object):
|
||||||
##
|
##
|
||||||
class PDFTextDevice(PDFDevice):
|
class PDFTextDevice(PDFDevice):
|
||||||
|
|
||||||
def render_string(self, textstate, seq):
|
def render_string(self, textstate, seq, ncs, graphicstate):
|
||||||
matrix = utils.mult_matrix(textstate.matrix, self.ctm)
|
matrix = utils.mult_matrix(textstate.matrix, self.ctm)
|
||||||
font = textstate.font
|
font = textstate.font
|
||||||
fontsize = textstate.fontsize
|
fontsize = textstate.fontsize
|
||||||
|
@ -77,15 +77,16 @@ class PDFTextDevice(PDFDevice):
|
||||||
if font.is_vertical():
|
if font.is_vertical():
|
||||||
textstate.linematrix = self.render_string_vertical(
|
textstate.linematrix = self.render_string_vertical(
|
||||||
seq, matrix, textstate.linematrix, font, fontsize,
|
seq, matrix, textstate.linematrix, font, fontsize,
|
||||||
scaling, charspace, wordspace, rise, dxscale)
|
scaling, charspace, wordspace, rise, dxscale, ncs, graphicstate)
|
||||||
else:
|
else:
|
||||||
textstate.linematrix = self.render_string_horizontal(
|
textstate.linematrix = self.render_string_horizontal(
|
||||||
seq, matrix, textstate.linematrix, font, fontsize,
|
seq, matrix, textstate.linematrix, font, fontsize,
|
||||||
scaling, charspace, wordspace, rise, dxscale)
|
scaling, charspace, wordspace, rise, dxscale, ncs, graphicstate)
|
||||||
return
|
return
|
||||||
|
|
||||||
def render_string_horizontal(self, seq, matrix, pos,
|
def render_string_horizontal(self, seq, matrix, pos,
|
||||||
font, fontsize, scaling, charspace, wordspace, rise, dxscale):
|
font, fontsize, scaling, charspace, wordspace,
|
||||||
|
rise, dxscale, ncs, graphicstate):
|
||||||
(x, y) = pos
|
(x, y) = pos
|
||||||
needcharspace = False
|
needcharspace = False
|
||||||
for obj in seq:
|
for obj in seq:
|
||||||
|
@ -97,14 +98,16 @@ class PDFTextDevice(PDFDevice):
|
||||||
if needcharspace:
|
if needcharspace:
|
||||||
x += charspace
|
x += charspace
|
||||||
x += self.render_char(utils.translate_matrix(matrix, (x, y)),
|
x += self.render_char(utils.translate_matrix(matrix, (x, y)),
|
||||||
font, fontsize, scaling, rise, cid)
|
font, fontsize, scaling, rise, cid,
|
||||||
|
ncs, graphicstate)
|
||||||
if cid == 32 and wordspace:
|
if cid == 32 and wordspace:
|
||||||
x += wordspace
|
x += wordspace
|
||||||
needcharspace = True
|
needcharspace = True
|
||||||
return (x, y)
|
return (x, y)
|
||||||
|
|
||||||
def render_string_vertical(self, seq, matrix, pos,
|
def render_string_vertical(self, seq, matrix, pos,
|
||||||
font, fontsize, scaling, charspace, wordspace, rise, dxscale):
|
font, fontsize, scaling, charspace, wordspace,
|
||||||
|
rise, dxscale, ncs, graphicstate):
|
||||||
(x, y) = pos
|
(x, y) = pos
|
||||||
needcharspace = False
|
needcharspace = False
|
||||||
for obj in seq:
|
for obj in seq:
|
||||||
|
@ -116,13 +119,14 @@ class PDFTextDevice(PDFDevice):
|
||||||
if needcharspace:
|
if needcharspace:
|
||||||
y += charspace
|
y += charspace
|
||||||
y += self.render_char(utils.translate_matrix(matrix, (x, y)),
|
y += self.render_char(utils.translate_matrix(matrix, (x, y)),
|
||||||
font, fontsize, scaling, rise, cid)
|
font, fontsize, scaling, rise, cid,
|
||||||
|
ncs, graphicstate)
|
||||||
if cid == 32 and wordspace:
|
if cid == 32 and wordspace:
|
||||||
y += wordspace
|
y += wordspace
|
||||||
needcharspace = True
|
needcharspace = True
|
||||||
return (x, y)
|
return (x, y)
|
||||||
|
|
||||||
def render_char(self, matrix, font, fontsize, scaling, rise, cid):
|
def render_char(self, matrix, font, fontsize, scaling, rise, cid, ncs, graphicstate):
|
||||||
return 0
|
return 0
|
||||||
|
|
||||||
|
|
||||||
|
|
|
@ -586,13 +586,13 @@ class PDFPageInterpreter(object):
|
||||||
|
|
||||||
# setgray-stroking
|
# setgray-stroking
|
||||||
def do_G(self, gray):
|
def do_G(self, gray):
|
||||||
self.graphicstate.color = gray
|
self.graphicstate.scolor = gray
|
||||||
#self.do_CS(LITERAL_DEVICE_GRAY)
|
#self.do_CS(LITERAL_DEVICE_GRAY)
|
||||||
return
|
return
|
||||||
|
|
||||||
# setgray-non-stroking
|
# setgray-non-stroking
|
||||||
def do_g(self, gray):
|
def do_g(self, gray):
|
||||||
self.graphicstate.color = gray
|
self.graphicstate.ncolor = gray
|
||||||
#self.do_cs(LITERAL_DEVICE_GRAY)
|
#self.do_cs(LITERAL_DEVICE_GRAY)
|
||||||
return
|
return
|
||||||
|
|
||||||
|
@ -769,7 +769,7 @@ class PDFPageInterpreter(object):
|
||||||
if settings.STRICT:
|
if settings.STRICT:
|
||||||
raise PDFInterpreterError('No font specified!')
|
raise PDFInterpreterError('No font specified!')
|
||||||
return
|
return
|
||||||
self.device.render_string(self.textstate, seq)
|
self.device.render_string(self.textstate, seq, self.ncs, self.graphicstate.copy())
|
||||||
return
|
return
|
||||||
|
|
||||||
# show
|
# show
|
||||||
|
|
Loading…
Reference in New Issue