def update_annotation(annotation, options):
import re
from pdfrw.objects import PdfString
# Contents holds a plain-text representation of the annotation
# content, such as for accessibility. All annotation types may
# have a Contents. NM holds the "annotation name" which also
# could have redactable text, I suppose. Markup annotations have
# "T" fields that hold a title / text label. Subj holds a
# comment subject. CA, RC, and AC are used in widget annotations.
for string_field in ("Contents", "NM", "T", "Subj", "CA", "RC", "AC"):
if getattr(annotation, string_field):
value = getattr(annotation, string_field).to_unicode()
for pattern, function in options.content_filters:
value = pattern.sub(function, value)
setattr(annotation, string_field, PdfString.from_unicode(value))
# A rich-text stream. Not implemented. Bail so that we don't
# accidentally leak something that should be redacted.
if annotation.RC:
raise ValueError(
"Annotation rich-text streams (Annot/RC) are not supported.")
# An action, usually used for links.
if annotation.A:
update_annotation_action(annotation, annotation.A, options)
if annotation.PA:
update_annotation_action(annotation, annotation.PA, options)
# If set, another annotation.
if annotation.Popup:
update_annotation(annotation.Popup, options)
def update_annotation_action(annotation, action, options):
from pdfrw.objects import PdfString
if action.URI and options.link_filters:
value = action.URI.to_unicode()
for func in options.link_filters:
value = func(value, annotation)
if value is None:
# Remove annotation by supressing the action.
action.URI = None
else:
action.URI = PdfString.from_unicode(value)
if action.Next:
# May be an Action or array of Actions to execute next.
next_action = action.Next
if isinstance(action.Next, dict):
next_action = [action.Next]
for a in next_action:
update_annotation_action(annotation, a, options)
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
def update_metadata(trailer, options):
# Update the PDF's Document Information Dictionary, which contains keys like
# Title, Author, Subject, Keywords, Creator, Producer, CreationDate, and ModDate
# (the latter two containing Date values, the rest strings).
import codecs
from pdfrw.objects import PdfString, PdfName
# Create the metadata dict if it doesn't exist, since the caller may be adding fields.
if not trailer.Info:
trailer.Info = PdfDict()
# Get a list of all metadata fields that exist in the PDF plus any fields
# that there are metadata filters for (since they may insert field values).
keys = set(str(k)[1:] for k in trailer.Info.keys()) \
| set(k for k in options.metadata_filters.keys() if k not in ("DEFAULT", "ALL"))
# Update each metadata field.
for key in keys:
# Get the functions to apply to this field.
functions = options.metadata_filters.get(key)
if functions is None:
# If nothing is defined for this field, use the DEFAULT functions.
functions = options.metadata_filters.get("DEFAULT", [])
# Append the ALL functions.
functions += options.metadata_filters.get("ALL", [])
# Run the functions on any existing values.
value = trailer.Info[PdfName(key)]
for f in functions:
# Before passing to the function, convert from a PdfString to a Python string.
if isinstance(value, PdfString):
# decode from PDF's "(...)" syntax.
value = value.decode()
# Filter the value.
value = f(value)
# Convert Python data type to PdfString.
if isinstance(value, str) or (sys.version_info < (3, )
and isinstance(value, unicode)):
# Convert string to a PdfString instance.
value = PdfString.from_unicode(value)
elif isinstance(value, datetime):
# Convert datetime into a PDF "D" string format.
value = value.strftime("%Y%m%d%H%M%S%z")
if len(value) == 19:
# If TZ info was included, add an apostrophe between the hour/minutes offsets.
value = value[:17] + "'" + value[17:]
value = PdfString("(D:%s)" % value)
elif value is None:
# delete the metadata value
pass
else:
raise ValueError(
"Invalid type of value returned by metadata_filter function. %s was returned by %s."
% (repr(value), f.__name__ or "anonymous function"))
# Replace value.
trailer.Info[PdfName(key)] = value
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