///|
fn pdf_length_name() -> PdfName {
pdf_length_key_name
}
///|
fn pdf_stream_length(object : PdfObject) -> Int raise PdfError {
match object.lookup_immediate(pdf_length_name()) {
Some(PdfInteger(length)) if length >= 0 => length
_ => raise PdfError::StreamLengthExpected
}
}
///|
fn ByteCursor::consume_pdf_ascii(
self : ByteCursor,
values : ArrayView[Int],
error : PdfError,
) -> Unit raise PdfError {
let position = self.absolute_position()
if self.matches_pdf_ascii_at(position, values) {
self.seek(position + values.length())
} else {
raise error
}
}
///|
fn ByteCursor::consume_pdf_stream_line_break(self : ByteCursor) -> Unit {
let value = self.peek_byte()
if value == 13 {
self.nudge()
if self.peek_byte() == 10 {
self.nudge()
}
} else if value == 10 {
self.nudge()
}
}
///|
fn pdf_stream_start_padding_byte(value : Int) -> Bool {
value == 32 || value == 0 || value == 12 || value == 9
}
///|
fn ByteCursor::consume_pdf_stream_start_line_break(self : ByteCursor) -> Unit {
while pdf_stream_start_padding_byte(self.peek_byte()) {
self.nudge()
}
self.consume_pdf_stream_line_break()
}
///|
fn pdf_reader_stream_end_marker_at(data : BytesView, position : Int) -> Bool {
pdf_view_matches_ascii_at(data, position, [
101, 110, 100, 115, 116, 114, 101, 97, 109,
]) ||
pdf_view_matches_ascii_at(data, position, [
10, 101, 110, 100, 115, 116, 114, 101, 97, 109,
]) ||
pdf_view_matches_ascii_at(data, position, [
13, 101, 110, 100, 115, 116, 114, 101, 97, 109,
]) ||
pdf_view_matches_ascii_at(data, position, [
13, 10, 101, 110, 100, 115, 116, 114, 101, 97, 109,
])
}
///|
fn ByteCursor::finish_pdf_stream_object(
self : ByteCursor,
prefix : PdfParsedIndirectObject,
) -> PdfParsedIndirectObject raise PdfError {
let length_result : Result[Int, Error] = try? pdf_stream_length(prefix.object)
match length_result {
Ok(length) => self.finish_pdf_stream_object_with_length(prefix, length)
Err(_) => self.finish_pdf_malformed_stream_object(prefix)
}
}
///|
fn ByteCursor::finish_pdf_stream_object_with_length(
self : ByteCursor,
prefix : PdfParsedIndirectObject,
length : Int,
defer_stream_data? : Bool = false,
) -> PdfParsedIndirectObject raise PdfError {
let stream_keyword_position = self.absolute_position()
self.consume_pdf_ascii(
[115, 116, 114, 101, 97, 109],
PdfError::ParseStreamExpected,
)
self.consume_pdf_stream_start_line_break()
let data_position = self.position()
if self.absolute_position() + length > self.data.length() {
raise PdfError::EndOfInput
}
let stream_data = if defer_stream_data {
self.seek_pdf_absolute(self.absolute_position() + length)
StreamToGet(stream_to_get(self.deep_copy(), data_position, length))
} else {
StreamGot(self.read_bytes(length))
}
if !pdf_reader_stream_end_marker_at(self.data[:], self.absolute_position()) {
self.seek_pdf_absolute(stream_keyword_position)
self.finish_pdf_malformed_stream_object(prefix)
} else {
self.consume_pdf_stream_line_break()
self.drop_pdf_whitespace()
self.consume_pdf_ascii(
[101, 110, 100, 115, 116, 114, 101, 97, 109],
PdfError::ParseEndStreamExpected,
)
self.drop_pdf_whitespace()
self.consume_pdf_ascii(
[101, 110, 100, 111, 98, 106],
PdfError::ParseEndObjectExpected,
)
{
number: prefix.number,
generation: prefix.generation,
object: pdf_stream(prefix.object, stream_data),
}
}
}
///|
fn pdf_reader_stream_data_end_before_marker(
data : BytesView,
marker_position : Int,
) -> Int {
if marker_position > 1 &&
data[marker_position - 1].to_int() == 10 &&
data[marker_position - 2].to_int() == 13 {
marker_position - 2
} else if marker_position > 0 &&
(
data[marker_position - 1].to_int() == 10 ||
data[marker_position - 1].to_int() == 13
) {
marker_position - 1
} else {
marker_position
}
}
///|
fn ByteCursor::finish_pdf_malformed_stream_object(
self : ByteCursor,
prefix : PdfParsedIndirectObject,
) -> PdfParsedIndirectObject raise PdfError {
self.consume_pdf_ascii(
[115, 116, 114, 101, 97, 109],
PdfError::ParseStreamExpected,
)
self.consume_pdf_stream_start_line_break()
let data_start = self.absolute_position()
let endstream = pdf_find_ascii_from(self.data, data_start, [
101, 110, 100, 115, 116, 114, 101, 97, 109,
])
if endstream < 0 {
raise PdfError::ParseEndStreamExpected
}
let data_end = pdf_reader_stream_data_end_before_marker(
self.data[:],
endstream,
)
let stream_data = self.data[data_start:data_end].to_owned()
self.seek_pdf_absolute(endstream + 9)
self.drop_pdf_whitespace()
self.consume_pdf_ascii(
[101, 110, 100, 111, 98, 106],
PdfError::ParseEndObjectExpected,
)
{
number: prefix.number,
generation: prefix.generation,
object: pdf_stream(
prefix.object.add_dict_entry(
pdf_length_name(),
PdfInteger(stream_data.length()),
),
StreamGot(stream_data),
),
}
}
///|
fn ByteCursor::finish_pdf_object_stream_member_with_length(
self : ByteCursor,
prefix : PdfParsedIndirectObject,
length : Int,
) -> PdfParsedIndirectObject raise PdfError {
let stream_keyword_position = self.absolute_position()
self.consume_pdf_ascii(
[115, 116, 114, 101, 97, 109],
PdfError::ParseStreamExpected,
)
self.consume_pdf_stream_start_line_break()
if self.absolute_position() + length > self.data.length() {
raise PdfError::EndOfInput
}
let stream_data = self.read_bytes(length)
if !pdf_reader_stream_end_marker_at(self.data[:], self.absolute_position()) {
self.seek_pdf_absolute(stream_keyword_position)
self.finish_pdf_malformed_object_stream_member(prefix)
} else {
self.consume_pdf_stream_line_break()
self.drop_pdf_whitespace()
self.consume_pdf_ascii(
[101, 110, 100, 115, 116, 114, 101, 97, 109],
PdfError::ParseEndStreamExpected,
)
{
number: prefix.number,
generation: prefix.generation,
object: pdf_stream(prefix.object, StreamGot(stream_data)),
}
}
}
///|
fn ByteCursor::finish_pdf_malformed_object_stream_member(
self : ByteCursor,
prefix : PdfParsedIndirectObject,
) -> PdfParsedIndirectObject raise PdfError {
self.consume_pdf_ascii(
[115, 116, 114, 101, 97, 109],
PdfError::ParseStreamExpected,
)
self.consume_pdf_stream_start_line_break()
let data_start = self.absolute_position()
let endstream = pdf_find_ascii_from(self.data, data_start, [
101, 110, 100, 115, 116, 114, 101, 97, 109,
])
if endstream < 0 {
raise PdfError::ParseEndStreamExpected
}
let data_end = pdf_reader_stream_data_end_before_marker(
self.data[:],
endstream,
)
let stream_data = self.data[data_start:data_end].to_owned()
self.seek_pdf_absolute(endstream + 9)
{
number: prefix.number,
generation: prefix.generation,
object: pdf_stream(
prefix.object.add_dict_entry(
pdf_length_name(),
PdfInteger(stream_data.length()),
),
StreamGot(stream_data),
),
}
}
///|
fn ByteCursor::finish_pdf_object_stream_member(
self : ByteCursor,
prefix : PdfParsedIndirectObject,
) -> PdfParsedIndirectObject raise PdfError {
let stream_position = self.absolute_position()
let length_result : Result[Int, Error] = try? pdf_stream_length(prefix.object)
match length_result {
Ok(length) =>
self.finish_pdf_object_stream_member_with_length(prefix, length) catch {
_ => {
self.seek_pdf_absolute(stream_position)
self.finish_pdf_malformed_object_stream_member(prefix)
}
}
Err(_) => self.finish_pdf_malformed_object_stream_member(prefix)
}
}
///|
fn pdf_parse_object_stream_member_from_view(
data : BytesView,
) -> PdfObject raise PdfError {
let cursor = byte_cursor_of_view(data)
let object = pdf_parse_single_lexeme_object(cursor.lex_pdf_tokens())
if cursor.stopped_at_pdf_stream_candidate() {
match object {
PdfDictionary(_) =>
cursor.finish_pdf_object_stream_member({
number: 0,
generation: 0,
object,
}).object
_ => raise PdfError::ParseSingleObjectExpected
}
} else {
object
}
}
///|
/// Parse a single indirect stream object from owned bytes.
///
/// The input should contain one indirect-object header, a stream dictionary,
/// stream data, `endstream`, and `endobj`.
pub fn pdf_parse_stream_object_from_bytes(
data : PdfBytes,
) -> PdfParsedIndirectObject raise PdfError {
pdf_parse_stream_object_from_view(data)
}
///|
/// Parse a single indirect stream object from a borrowed byte view.
///
/// Malformed `/Length` values are recovered by scanning to `endstream` when
/// possible, matching the stream recovery used by the primitive reader.
pub fn pdf_parse_stream_object_from_view(
data : BytesView,
) -> PdfParsedIndirectObject raise PdfError {
let cursor = byte_cursor_of_view(data)
let prefix = pdf_parse_indirect_object_prefix_lexemes(cursor.lex_pdf_tokens())
cursor.finish_pdf_stream_object(prefix)
}