def _cache_xobj(contents, resources, mbox, bbox, rotation):
''' Return a cached Form XObject, or create a new one and cache it.
Adds private members x, y, w, h
'''
cachedict = contents.xobj_cachedict
if cachedict is None:
cachedict = contents.private.xobj_cachedict = {}
cachekey = mbox, bbox, rotation
result = cachedict.get(cachekey)
if result is None:
func = (_get_fullpage, _get_subpage)[mbox != bbox]
result = PdfDict(
func(contents, resources, mbox, bbox, rotation),
Type=PdfName.XObject,
Subtype=PdfName.Form,
FormType=1,
BBox=PdfArray(bbox),
)
rect = bbox
if rotation:
matrix = rotate_point((1, 0), rotation) + \
rotate_point((0, 1), rotation)
result.Matrix = PdfArray(matrix + (0, 0))
rect = rotate_rect(rect, rotation)
result.private.x = rect[0]
result.private.y = rect[1]
result.private.w = rect[2] - rect[0]
result.private.h = rect[3] - rect[1]
cachedict[cachekey] = result
return result
def wrap_object(obj, width, margin):
''' Wrap an xobj in its own page object.
'''
fmt = 'q %s 0 0 %s %s %s cm /MyImage Do Q'
contents = PdfDict(indirect=True)
subtype = obj.Subtype
if subtype == PdfName.Form:
contents._stream = obj.stream
contents.Length = obj.Length
contents.Filter = obj.Filter
contents.DecodeParms = obj.DecodeParms
resources = obj.Resources
mbox = obj.BBox
elif subtype == PdfName.Image: # Image
xoffset = margin[0]
yoffset = margin[1]
cw = width - margin[0] - margin[2]
iw, ih = float(obj.Width), float(obj.Height)
ch = 1.0 * cw / iw * ih
height = ch + margin[1] + margin[3]
p = tuple(('%.9f' % x).rstrip('0').rstrip('.') for x in (cw, ch, xoffset, yoffset))
contents.stream = fmt % p
resources = PdfDict(XObject=PdfDict(MyImage=obj))
mbox = PdfArray((0, 0, width, height))
else:
raise TypeError("Expected Form or Image XObject")
return PdfDict(
indirect=True,
Type=PdfName.Page,
MediaBox=mbox,
Resources=resources,
Contents=contents,
)
def parse_xref_stream(self,
source,
int=int,
range=range,
enumerate=enumerate,
islice=itertools.islice,
defaultdict=collections.defaultdict,
hexlify=binascii.hexlify):
''' Parse (one of) the cross-reference file section(s)
'''
def readint(s, lengths):
offset = 0
for length in itertools.cycle(lengths):
next = offset + length
yield int(hexlify(s[offset:next]), 16) if length else None
offset = next
setdefault = source.obj_offsets.setdefault
next = source.next
# check for xref stream object
objid = source.multiple(3)
ok = len(objid) == 3
ok = ok and objid[0].isdigit()
ok = ok and objid[1] == 'obj'
ok = ok and objid[2] == '<<'
if not ok:
source.exception('Expected xref stream start')
obj = self.readdict(source)
if obj.Type != PdfName.XRef:
source.exception('Expected dict type of /XRef')
tok = next()
self.readstream(obj, self.findstream(obj, tok, source), source, True)
old_strm = obj.stream
if not uncompress([obj], True):
source.exception('Could not decompress Xref stream')
stream = obj.stream
# Fix for issue #76 -- goofy compressed xref stream
# that is NOT ACTUALLY COMPRESSED
stream = stream if stream is not old_strm else convert_store(old_strm)
num_pairs = obj.Index or PdfArray(['0', obj.Size])
num_pairs = [int(x) for x in num_pairs]
num_pairs = zip(num_pairs[0::2], num_pairs[1::2])
entry_sizes = [int(x) for x in obj.W]
if len(entry_sizes) != 3:
source.exception('Invalid entry size')
object_streams = defaultdict(list)
get = readint(stream, entry_sizes)
for objnum, size in num_pairs:
for cnt in range(size):
xtype, p1, p2 = islice(get, 3)
if xtype in (1, None):
if p1:
setdefault((objnum, p2 or 0), p1)
elif xtype == 2:
object_streams[p1].append((objnum, p2))
objnum += 1
obj.private.object_streams = object_streams
return obj
def readarray(self, source, PdfArray=PdfArray):
''' Found a [ token. Parse the tokens after that.
'''
specialget = self.special.get
result = []
pop = result.pop
append = result.append
for value in source:
if value in ']R':
if value == ']':
break
generation = pop()
value = self.findindirect(pop(), generation)
else:
func = specialget(value)
if func is not None:
value = func(source)
append(value)
return PdfArray(result)
def parsexref(self, source, int=int, range=range):
''' Parse (one of) the cross-reference file section(s)
'''
def _pairs(array):
i = 0
while 1:
yield int(array[i]), int(array[i + 1])
i += 2
if (i + 1) >= len(array):
break
def convert_to_int(d, size):
if size > 8:
source.exception('Invalid size in convert_to_int')
d = '\x00\x00\x00\x00\x00\x00\x00\x00' + d
d = d[-8:]
return struct.unpack('>q', d)[0]
def read_trailer():
tok = next()
if tok != '<<':
source.exception('Expected "<<" starting catalog')
return self.readdict(source)
setdefault = source.obj_offsets.setdefault
add_offset = source.all_offsets.append
next = source.next
tok = next()
if tok.isdigit():
# check for xref stream object
objid = source.multiple(2)
ok = len(objid) == 2
ok = ok and objid[0].isdigit()
ok = ok and objid[1] == 'obj'
if ok:
next() # start of dict
obj = self.readdict(source)
assert obj.Type == '/XRef'
tok = next()
end = source.floc + int(obj.Length)
self.readstream(obj, self.findstream(obj, tok, source), source)
uncompress([obj])
num_pairs = obj.Index or PdfArray(['0', obj.Size])
entry_sizes = [int(x) for x in obj.W]
object_streams = {}
for num, size in _pairs(num_pairs):
cnt = 0
stream_offset = 0
while cnt < size:
for i in range(len(entry_sizes)):
d = obj.stream[stream_offset:stream_offset +
entry_sizes[i]]
stream_offset += entry_sizes[i]
di = convert_to_int(d, entry_sizes[i])
if i == 0:
xref_type = di
if xref_type == 0 and entry_sizes[0] == 0:
xref_type = 1
elif i == 1:
if xref_type == 1:
offset = di
elif xref_type == 2:
objnum = di
elif i == 2:
if xref_type == 1:
generation = di
elif xref_type == 2:
obstr_idx = di
if xref_type == 1 and offset != 0:
setdefault((num, generation), offset)
add_offset(offset)
elif xref_type == 2:
if not objnum in object_streams:
object_streams[objnum] = []
object_streams[objnum].append(obstr_idx)
cnt += 1
num += 1
self.load_stream_objects(object_streams)
source.floc = end
endit = source.multiple(2)
if endit != ['endstream', 'endobj']:
source.exception('Expected endstream endobj')
return obj
else:
source.exception('Expected xref stream')
elif tok == 'xref':
# plain xref table
start = source.floc
try:
while 1:
tok = next()
if tok == 'trailer':
return read_trailer()
startobj = int(tok)
for objnum in range(startobj, startobj + int(next())):
offset = int(next())
generation = int(next())
inuse = next()
if inuse == 'n':
if offset != 0:
setdefault((objnum, generation), offset)
add_offset(offset)
elif inuse != 'f':
raise ValueError
except:
pass
try:
# Table formatted incorrectly.
# See if we can figure it out anyway.
end = source.fdata.rindex('trailer', start)
table = source.fdata[start:end].splitlines()
for line in table:
tokens = line.split()
if len(tokens) == 2:
objnum = int(tokens[0])
elif len(tokens) == 3:
offset, generation, inuse = \
int(tokens[0]), int(tokens[1]), tokens[2]
if offset != 0 and inuse == 'n':
setdefault((objnum, generation), offset)
add_offset(offset)
objnum += 1
elif tokens:
log.error('Invalid line in xref table: %s' %
repr(line))
raise ValueError
log.warning('Badly formatted xref table')
source.floc = end
next()
except:
source.floc = start
source.exception('Invalid table format')
return read_trailer()
else:
source.exception('Expected "xref" keyword or xref stream object')
def __init__(self, version='1.3', compress=False):
self.pagearray = PdfArray()
self.compress = compress
self.version = version
self.killobj = {}
class PdfWriter(object):
_trailer = None
def __init__(self, version='1.3', compress=False):
self.pagearray = PdfArray()
self.compress = compress
self.version = version
self.killobj = {}
def addpage(self, page):
self._trailer = None
if page.Type != PdfName.Page:
raise PdfOutputError('Bad /Type: Expected %s, found %s'
% (PdfName.Page, page.Type))
inheritable = page.inheritable # searches for resources
self.pagearray.append(
IndirectPdfDict(
page,
Resources=inheritable.Resources,
MediaBox=inheritable.MediaBox,
CropBox=inheritable.CropBox,
Rotate=inheritable.Rotate,
)
)
# Add parents in the hierarchy to objects we
# don't want to output
killobj = self.killobj
obj = page.Parent
while obj is not None:
objid = id(obj)
if objid in killobj:
break
killobj[objid] = obj
obj = obj.Parent
return self
addPage = addpage # for compatibility with pyPdf
def addpages(self, pagelist):
for page in pagelist:
self.addpage(page)
return self
def _get_trailer(self):
trailer = self._trailer
if trailer is not None:
return trailer
# Create the basic object structure of the PDF file
trailer = PdfDict(
Root=IndirectPdfDict(
Type=PdfName.Catalog,
Pages=IndirectPdfDict(
Type=PdfName.Pages,
Count=PdfObject(len(self.pagearray)),
Kids=self.pagearray
)
)
)
# Make all the pages point back to the page dictionary and
# ensure they are indirect references
pagedict = trailer.Root.Pages
for page in pagedict.Kids:
page.Parent = pagedict
self._trailer = trailer
return trailer
def _set_trailer(self, trailer):
self._trailer = trailer
trailer = property(_get_trailer, _set_trailer)
def write(self, fname, trailer=None):
trailer = trailer or self.trailer
# Dump the data. We either have a filename or a preexisting
# file object.
preexisting = hasattr(fname, 'write')
f = preexisting and fname or open(fname, 'wb')
FormatObjects(f, trailer, self.version, self.compress, self.killobj)
if not preexisting:
f.close()
# Dump the trace.
if trailer.active_trace != None:
fname_trace = fname + '.trace'
f = open(fname_trace, 'wb')
pickle.dump(trailer.active_trace, f)
f.close()
def __init__(self, version='1.3', compress=False):
self.pagearray = PdfArray()
self.compress = compress
self.version = version
self.killobj = {}
class PdfWriter(object):
_trailer = None
def __init__(self, version='1.3', compress=False):
self.pagearray = PdfArray()
self.compress = compress
self.version = version
self.killobj = {}
def addpage(self, page):
self._trailer = None
if page.Type != PdfName.Page:
raise PdfOutputError('Bad /Type: Expected %s, found %s' %
(PdfName.Page, page.Type))
inheritable = page.inheritable # searches for resources
self.pagearray.append(
IndirectPdfDict(
page,
Resources=inheritable.Resources,
MediaBox=inheritable.MediaBox,
CropBox=inheritable.CropBox,
Rotate=inheritable.Rotate,
))
# Add parents in the hierarchy to objects we
# don't want to output
killobj = self.killobj
obj = page.Parent
while obj is not None:
objid = id(obj)
if objid in killobj:
break
killobj[objid] = obj
obj = obj.Parent
return self
addPage = addpage # for compatibility with pyPdf
def addpages(self, pagelist):
for page in pagelist:
self.addpage(page)
return self
def _get_trailer(self):
trailer = self._trailer
if trailer is not None:
return trailer
# Create the basic object structure of the PDF file
trailer = PdfDict(Root=IndirectPdfDict(
Type=PdfName.Catalog,
Pages=IndirectPdfDict(Type=PdfName.Pages,
Count=PdfObject(len(self.pagearray)),
Kids=self.pagearray)))
# Make all the pages point back to the page dictionary and
# ensure they are indirect references
pagedict = trailer.Root.Pages
for page in pagedict.Kids:
page.Parent = pagedict
self._trailer = trailer
return trailer
def _set_trailer(self, trailer):
self._trailer = trailer
trailer = property(_get_trailer, _set_trailer)
def write(self, fname, trailer=None):
trailer = trailer or self.trailer
# Dump the data. We either have a filename or a preexisting
# file object.
preexisting = hasattr(fname, 'write')
f = preexisting and fname or open(fname, 'wb')
FormatObjects(f, trailer, self.version, self.compress, self.killobj)
if not preexisting:
f.close()
# Dump the trace.
if trailer.active_trace != None:
fname_trace = fname + '.trace'
f = open(fname_trace, 'wb')
pickle.dump(trailer.active_trace, f)
f.close()
def make_page(page, pdf_page, psem):
# Prepare everything in parallel
@asyncio.coroutine
def get_pdf_thumbnail(psem):
if page.thumbnail is None:
return None
return (yield from page.thumbnail.pdf_thumbnail(psem))
@asyncio.coroutine
def get_pdf_background(psem):
if page.background is None:
return None
return (yield from page.background.pdf_image(psem))
@asyncio.coroutine
def get_pdf_mask(foreground, psem):
if foreground.color is not None:
return None
return (yield from foreground.pdf_mask(psem))
pdf_thumbnail, pdf_background, pdf_foregrounds, pdf_masks = (
yield from asyncio.gather(
get_pdf_thumbnail(psem), get_pdf_background(psem),
asyncio.gather(
*[fg.pdf_image(psem) for fg in page.foreground]),
asyncio.gather(
*[get_pdf_mask(fg, psem) for fg in page.foreground])))
pdf_page.MediaBox = PdfArray(
[0, 0, PdfNumber(page.width),
PdfNumber(page.height)])
pdf_page.Group = pdf_group
pdf_resources = PdfDict()
pdf_colorspace = PdfDict()
pdf_colorspace.DefaultRGB = default_rgb_colorspace
pdf_resources.ColorSpace = pdf_colorspace
pdf_xobject = PdfDict()
if pdf_thumbnail is not None:
pdf_page.Thumb = pdf_thumbnail
im_index = 0
# Save graphics state and scale unity rectangle to page size
matrix = TransformationMatrix()
matrix.scale(page.width, page.height)
before_graphics = ("q\n" + "%s cm\n" % matrix.to_pdf())
after_graphics = "\nQ\n"
contents = ""
graphics = ""
current_color = None
if page.color != self._factory.WHITE:
if current_color != page.color:
current_color = page.color
graphics += page.color.to_pdf() + " rg "
graphics += ("0 0 1 1 re " + "f\n")
if pdf_background is not None:
pdf_xobject[PdfName("Im%d" % im_index)] = pdf_background
graphics += "/Im%d Do\n" % im_index
im_index += 1
for foreground, pdf_foreground, pdf_mask in zip(
page.foreground, pdf_foregrounds, pdf_masks):
if pdf_mask is not None:
pdf_xobject[PdfName("Im%d" % im_index)] = pdf_mask
im_index += 1
pdf_xobject[PdfName("Im%d" % im_index)] = pdf_foreground
if (foreground.color is not None
and current_color != foreground.color):
current_color = foreground.color
graphics += foreground.color.to_pdf() + " rg "
graphics += "/Im%d Do\n" % im_index
im_index += 1
if graphics:
contents += (before_graphics + graphics.rstrip(" \n") +
after_graphics)
current_color = None
before_text = ("BT\n" + "/F1 1 Tf 3 Tr\n")
after_text = "\nET\n"
text = ""
pdf_annots = []
for t in page.text:
if t.text:
matrix = TransformationMatrix()
# Glyph size is 0.5 x 1
matrix.scale(2 / len(t.text), 1)
matrix.translate(-0.5, -0.5)
if t.direction == "ltr":
pass
elif t.direction == "rtl":
matrix.translate(0, -1)
elif t.direction == "ttb":
matrix.rotate(90)
matrix.rotate(-t.rotation)
matrix.translate(0.5, 0.5)
matrix.scale(t.width, t.height)
matrix.translate(t.x, t.y)
text += "%s Tm %s Tj\n" % (
matrix.to_pdf(), PdfString().from_bytes(
t.text.encode("utf-16-be"), bytes_encoding="hex"))
if t.external_link is not None or t.internal_link is not None:
pdf_annot = PdfDict()
pdf_annots.append(pdf_annot)
pdf_annot.Type = PdfName.Annot
pdf_annot.Subtype = PdfName.Link
pdf_annot.Border = [0, 0, 0]
pdf_annot.Rect = [
PdfNumber(t.x),
PdfNumber(t.y),
PdfNumber(t.x + t.width),
PdfNumber(t.y + t.height)
]
if t.external_link is not None:
pdf_a = PdfDict()
pdf_annot.A = pdf_a
pdf_a.Type = PdfName.Action
pdf_a.S = PdfName.URI
pdf_a.URI = t.external_link.decode("latin-1")
if t.internal_link is not None:
pdf_target_page = pdf_pages[t.internal_link[0]]
target_x, target_y = t.internal_link[1]
pdf_annot.Dest = [
pdf_target_page, PdfName.XYZ,
PdfNumber(target_x),
PdfNumber(target_y), 0
]
text = text.rstrip(" \n")
if text:
pdf_resources.Font = pdf_font_mapping
contents += (before_text + text + after_text)
contents = contents.rstrip(" \n")
if contents:
pdf_contents = PdfDict()
pdf_contents.indirect = True
pdf_page.Contents = pdf_contents
if COMPRESS_PAGE_CONTENTS:
pdf_contents.Filter = [PdfName.FlateDecode]
pdf_contents.stream = zlib.compress(
contents.encode("latin-1"), 9).decode("latin-1")
else:
pdf_contents.stream = contents
if pdf_annots:
pdf_page.Annots = pdf_annots
if pdf_xobject:
pdf_resources.XObject = pdf_xobject
if pdf_resources:
pdf_page.Resources = pdf_resources
# Report progress
nonlocal finished_pages
finished_pages += 1
if progress_cb:
progress_cb(finished_pages / len(self._pages))
def write_async(self, outfile, process_semaphore, progress_cb=None):
pdf_writer = PdfWriter(version="1.5")
pdf_group = PdfDict()
pdf_group.indirect = True
pdf_group.CS = PdfName.DeviceRGB
pdf_group.I = PdfBool(True)
pdf_group.S = PdfName.Transparency
pdf_font_mapping = PdfDict()
pdf_font_mapping.indirect = True
pdf_font_mapping.F1 = self._build_font()
for _ in self._pages:
pdf_page = PdfDict()
pdf_page.Type = PdfName.Page
pdf_writer.addpage(pdf_page)
# pdfrw makes a internal copy of the pages
# use the copy so that references to pages in links are correct
pdf_pages = list(pdf_writer.pagearray)
srgb_colorspace = PdfDict()
srgb_colorspace.indirect = True
srgb_colorspace.N = 3 # Number of components (red, green, blue)
with open(SRGB_ICC_FILENAME, "rb") as f:
srgb_colorspace_stream = f.read()
srgb_colorspace.Filter = [PdfName.FlateDecode]
srgb_colorspace.stream = zlib.compress(srgb_colorspace_stream,
9).decode("latin-1")
srgb_colorspace.Length1 = len(srgb_colorspace_stream)
default_rgb_colorspace = PdfArray([PdfName.ICCBased, srgb_colorspace])
default_rgb_colorspace.indirect = True
# Handle all pages in parallel
@asyncio.coroutine
def make_page(page, pdf_page, psem):
# Prepare everything in parallel
@asyncio.coroutine
def get_pdf_thumbnail(psem):
if page.thumbnail is None:
return None
return (yield from page.thumbnail.pdf_thumbnail(psem))
@asyncio.coroutine
def get_pdf_background(psem):
if page.background is None:
return None
return (yield from page.background.pdf_image(psem))
@asyncio.coroutine
def get_pdf_mask(foreground, psem):
if foreground.color is not None:
return None
return (yield from foreground.pdf_mask(psem))
pdf_thumbnail, pdf_background, pdf_foregrounds, pdf_masks = (
yield from asyncio.gather(
get_pdf_thumbnail(psem), get_pdf_background(psem),
asyncio.gather(
*[fg.pdf_image(psem) for fg in page.foreground]),
asyncio.gather(
*[get_pdf_mask(fg, psem) for fg in page.foreground])))
pdf_page.MediaBox = PdfArray(
[0, 0, PdfNumber(page.width),
PdfNumber(page.height)])
pdf_page.Group = pdf_group
pdf_resources = PdfDict()
pdf_colorspace = PdfDict()
pdf_colorspace.DefaultRGB = default_rgb_colorspace
pdf_resources.ColorSpace = pdf_colorspace
pdf_xobject = PdfDict()
if pdf_thumbnail is not None:
pdf_page.Thumb = pdf_thumbnail
im_index = 0
# Save graphics state and scale unity rectangle to page size
matrix = TransformationMatrix()
matrix.scale(page.width, page.height)
before_graphics = ("q\n" + "%s cm\n" % matrix.to_pdf())
after_graphics = "\nQ\n"
contents = ""
graphics = ""
current_color = None
if page.color != self._factory.WHITE:
if current_color != page.color:
current_color = page.color
graphics += page.color.to_pdf() + " rg "
graphics += ("0 0 1 1 re " + "f\n")
if pdf_background is not None:
pdf_xobject[PdfName("Im%d" % im_index)] = pdf_background
graphics += "/Im%d Do\n" % im_index
im_index += 1
for foreground, pdf_foreground, pdf_mask in zip(
page.foreground, pdf_foregrounds, pdf_masks):
if pdf_mask is not None:
pdf_xobject[PdfName("Im%d" % im_index)] = pdf_mask
im_index += 1
pdf_xobject[PdfName("Im%d" % im_index)] = pdf_foreground
if (foreground.color is not None
and current_color != foreground.color):
current_color = foreground.color
graphics += foreground.color.to_pdf() + " rg "
graphics += "/Im%d Do\n" % im_index
im_index += 1
if graphics:
contents += (before_graphics + graphics.rstrip(" \n") +
after_graphics)
current_color = None
before_text = ("BT\n" + "/F1 1 Tf 3 Tr\n")
after_text = "\nET\n"
text = ""
pdf_annots = []
for t in page.text:
if t.text:
matrix = TransformationMatrix()
# Glyph size is 0.5 x 1
matrix.scale(2 / len(t.text), 1)
matrix.translate(-0.5, -0.5)
if t.direction == "ltr":
pass
elif t.direction == "rtl":
matrix.translate(0, -1)
elif t.direction == "ttb":
matrix.rotate(90)
matrix.rotate(-t.rotation)
matrix.translate(0.5, 0.5)
matrix.scale(t.width, t.height)
matrix.translate(t.x, t.y)
text += "%s Tm %s Tj\n" % (
matrix.to_pdf(), PdfString().from_bytes(
t.text.encode("utf-16-be"), bytes_encoding="hex"))
if t.external_link is not None or t.internal_link is not None:
pdf_annot = PdfDict()
pdf_annots.append(pdf_annot)
pdf_annot.Type = PdfName.Annot
pdf_annot.Subtype = PdfName.Link
pdf_annot.Border = [0, 0, 0]
pdf_annot.Rect = [
PdfNumber(t.x),
PdfNumber(t.y),
PdfNumber(t.x + t.width),
PdfNumber(t.y + t.height)
]
if t.external_link is not None:
pdf_a = PdfDict()
pdf_annot.A = pdf_a
pdf_a.Type = PdfName.Action
pdf_a.S = PdfName.URI
pdf_a.URI = t.external_link.decode("latin-1")
if t.internal_link is not None:
pdf_target_page = pdf_pages[t.internal_link[0]]
target_x, target_y = t.internal_link[1]
pdf_annot.Dest = [
pdf_target_page, PdfName.XYZ,
PdfNumber(target_x),
PdfNumber(target_y), 0
]
text = text.rstrip(" \n")
if text:
pdf_resources.Font = pdf_font_mapping
contents += (before_text + text + after_text)
contents = contents.rstrip(" \n")
if contents:
pdf_contents = PdfDict()
pdf_contents.indirect = True
pdf_page.Contents = pdf_contents
if COMPRESS_PAGE_CONTENTS:
pdf_contents.Filter = [PdfName.FlateDecode]
pdf_contents.stream = zlib.compress(
contents.encode("latin-1"), 9).decode("latin-1")
else:
pdf_contents.stream = contents
if pdf_annots:
pdf_page.Annots = pdf_annots
if pdf_xobject:
pdf_resources.XObject = pdf_xobject
if pdf_resources:
pdf_page.Resources = pdf_resources
# Report progress
nonlocal finished_pages
finished_pages += 1
if progress_cb:
progress_cb(finished_pages / len(self._pages))
finished_pages = 0
yield from asyncio.gather(*[
make_page(page, pdf_page, process_semaphore)
for page, pdf_page in zip(self._pages, pdf_pages)
])
trailer = pdf_writer.trailer
document_id = PdfString().from_bytes(os.urandom(16))
trailer.ID = [document_id, document_id]
mark_info = PdfDict()
mark_info.Marked = PdfBool(True)
trailer.Root.MarkInfo = mark_info
struct_tree_root = PdfDict()
struct_tree_root.Type = PdfName.StructTreeRoot
trailer.Root.StructTreeRoot = struct_tree_root
metadata = PdfDict()
metadata.indirect = True
metadata.Type = PdfName.Metadata
metadata.Subtype = PdfName.XML
xmp = XMPMeta()
xmp.set_property(XMP_NS_PDFA_ID, "part", "2")
xmp.set_property(XMP_NS_PDFA_ID, "conformance", "A")
metadata_stream = xmp.serialize_to_str().encode("utf-8")
metadata.Filter = [PdfName.FlateDecode]
metadata.stream = zlib.compress(metadata_stream, 9).decode("latin-1")
metadata.Length1 = len(metadata_stream)
trailer.Root.Metadata = metadata
with TemporaryDirectory(prefix="djpdf-") as temp_dir:
pdf_writer.write(path.join(temp_dir, "temp.pdf"))
cmd = [
QPDF_CMD, "--stream-data=preserve",
"--object-streams=preserve", "--normalize-content=n",
"--newline-before-endstream"
]
if LINEARIZE_PDF:
cmd.extend(["--linearize"])
cmd.extend([
path.abspath(path.join(temp_dir, "temp.pdf")),
path.abspath(outfile)
])
yield from run_command_async(cmd, process_semaphore)
def _build_font():
with open(FONT_FILENAME, "rb") as f:
embedded_font_stream = f.read()
embedded_font = PdfDict()
embedded_font.indirect = True
embedded_font.Filter = [PdfName.FlateDecode]
embedded_font.stream = zlib.compress(embedded_font_stream,
9).decode("latin-1")
embedded_font.Length1 = len(embedded_font_stream)
font_descriptor = PdfDict()
font_descriptor.indirect = True
font_descriptor.Ascent = 1000
font_descriptor.CapHeight = 1000
font_descriptor.Descent = -1
font_descriptor.Flags = 5 # FixedPitch + Symbolic
font_descriptor.FontBBox = PdfArray([0, 0, 1000, 500])
font_descriptor.FontFile2 = embedded_font
font_descriptor.FontName = PdfName.GlyphLessFont
font_descriptor.ItalicAngle = 0
font_descriptor.StemV = 80
font_descriptor.Type = PdfName.FontDescriptor
# Map everything to glyph 1
cid_to_gid_map_stream = b"\0\1" * (1 << 16)
cid_to_gid_map = PdfDict()
cid_to_gid_map.indirect = True
cid_to_gid_map.Filter = [PdfName.FlateDecode]
cid_to_gid_map.stream = zlib.compress(cid_to_gid_map_stream,
9).decode("latin-1")
cid_to_gid_map.Length1 = len(cid_to_gid_map_stream)
cid_system_info = PdfDict()
cid_system_info.Ordering = PdfString.from_unicode("Identity")
cid_system_info.Registry = PdfString.from_unicode("Adobe")
cid_system_info.Supplement = 0
cid_font = PdfDict()
cid_font.indirect = True
cid_font.CIDToGIDMap = cid_to_gid_map
cid_font.BaseFont = PdfName.GlyphLessFont
cid_font.CIDSystemInfo = cid_system_info
cid_font.FontDescriptor = font_descriptor
cid_font.Subtype = PdfName.CIDFontType2
cid_font.Type = PdfName.Font
cid_font.DW = 500
with open(UNICODE_CMAP_FILENAME, "rb") as f:
unicode_cmap_stream = f.read()
unicode_cmap = PdfDict()
unicode_cmap.indirect = True
unicode_cmap.Filter = [PdfName.FlateDecode]
unicode_cmap.stream = zlib.compress(unicode_cmap_stream,
9).decode("latin-1")
font = PdfDict()
font.indirect = True
font.BaseFont = PdfName.GlyphLessFont
font.DescendantFonts = PdfArray([cid_font])
font.Encoding = PdfName("Identity-H")
font.Subtype = PdfName.Type0
font.ToUnicode = unicode_cmap
font.Type = PdfName.Font
return font