// Copyright 2022 Adobe. All rights reserved.
// This file is licensed to you under the Apache License,
// Version 2.0 (http://www.apache.org/licenses/LICENSE-2.0)
// or the MIT license (http://opensource.org/licenses/MIT),
// at your option.
// Unless required by applicable law or agreed to in writing,
// this software is distributed on an "AS IS" BASIS, WITHOUT
// WARRANTIES OR REPRESENTATIONS OF ANY KIND, either express or
// implied. See the LICENSE-MIT and LICENSE-APACHE files for the
// specific language governing permissions and limitations under
// each license.
use std::{io::Cursor, str};
use log::error;
use quick_xml::{
events::{BytesStart, Event},
name::QName,
Reader, Writer,
};
use crate::{asset_io::ReadSeek, jumbf_io::get_cailoader_handler, Error, Result};
const RDF_DESCRIPTION: &[u8] = b"rdf:Description";
const XMP_END: &[u8] = b"<?xpacket end=\"w\"?>";
pub const MIN_XMP: &str = r#"<?xpacket begin="" id="W5M0MpCehiHzreSzNTczkc9d"?><x:xmpmeta xmlns:x="adobe:ns:meta/" x:xmptk="XMP Core 6.0.0"><rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"><rdf:Description rdf:about="" xmlns:xmp="http://ns.adobe.com/xap/1.0/" xmlns:xmpMM="http://ns.adobe.com/xap/1.0/mm/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:dcterms="http://purl.org/dc/terms/" xmpMM:DocumentID="xmp.did:cb9f5498-bb58-4572-8043-8c369e6bfb9b" xmpMM:InstanceID="xmp.iid:cb9f5498-bb58-4572-8043-8c369e6bfb9b"> </rdf:Description></rdf:RDF></x:xmpmeta><?xpacket end="w"?>"#;
#[derive(Default)]
pub struct XmpInfo {
pub document_id: Option<String>,
pub instance_id: Option<String>,
pub provenance: Option<String>,
}
impl XmpInfo {
/// search xmp data for provenance, documentID and instanceID
pub fn from_source(source: &mut dyn ReadSeek, format: &str) -> Self {
let xmp = get_cailoader_handler(format).and_then(|cai_loader| {
// read xmp if available
cai_loader.read_xmp(source)
});
// todo: do this in one pass through XMP
let provenance = xmp.as_deref().and_then(extract_provenance);
let document_id = xmp.as_deref().and_then(extract_document_id);
let instance_id = xmp.as_deref().and_then(extract_instance_id);
Self {
document_id,
instance_id,
provenance,
}
}
}
/// Extract an a value from XMP using a key
fn extract_xmp_key(xmp: &str, key: &str) -> Option<String> {
let mut reader = Reader::from_str(xmp);
reader.config_mut().trim_text(true);
loop {
match reader.read_event() {
Ok(Event::Start(ref e)) | Ok(Event::Empty(ref e)) => {
if e.name() == QName(RDF_DESCRIPTION) {
// attribute case
let value = e.attributes().find(|a| {
if let Ok(attribute) = a {
attribute.key == QName(key.as_bytes())
} else {
false
}
});
if let Some(Ok(attribute)) = value {
if let Ok(s) = String::from_utf8(attribute.value.to_vec()) {
return Some(s);
}
}
} else if e.name() == QName(key.as_bytes()) {
// tag case
if let Ok(s) = reader.read_text(e.name()) {
if let Ok(s) = s.decode() {
return Some(s.into_owned());
}
}
}
}
Ok(Event::Eof) => break,
_ => {}
}
}
None
}
// writes the event to the writer)
/// Add a value to XMP using a key, replaces the value if the key exists, or adds it only once
fn add_xmp_key(xmp: &str, key: &str, value: &str) -> Result<String> {
let orig_length = xmp.len();
// Minimal padding should be 2 KB to 4 KB. This is used if no XMP packet end is found.
// If no packet end is found and the original length is less than 4096, the packet will be
// extended.
let mut target_length = orig_length.max(4096);
// Remove the ending xpacket if present for easier manipulation
let xpacket_end = "<?xpacket end";
let xmp_body = if let Some(pos) = xmp.rfind(xpacket_end) {
// Use pos, not orig_length - XMP_END.len(): the trailer may be shorter than XMP_END, which would underflow.
target_length = pos;
xmp[..pos].trim_end()
} else {
xmp
};
let mut reader = Reader::from_str(xmp_body);
// Do not trim text to preserve whitespace
reader.config_mut().trim_text(false);
reader.config_mut().expand_empty_elements = false;
let mut writer = Writer::new(Cursor::new(Vec::new()));
let mut added = false;
loop {
let event = reader
.read_event()
.map_err(|e| Error::XmpReadError(e.to_string()))?;
// println!("{:?}", event);
match event {
Event::Start(ref e) if e.name() == QName(RDF_DESCRIPTION) && !added => {
// creates a new element
let mut elem = BytesStart::from_content(
String::from_utf8_lossy(RDF_DESCRIPTION),
RDF_DESCRIPTION.len(),
);
for attr in e.attributes() {
match attr {
Ok(attr) => {
if attr.key == QName(key.as_bytes()) {
// replace the key/value if it exists
elem.push_attribute((key, value));
added = true;
} else {
// add all other existing elements
elem.extend_attributes([attr]);
}
}
Err(e) => {
error!("Error at position {}", reader.buffer_position());
return Err(Error::XmpReadError(e.to_string()));
}
}
}
if !added {
// didn't exist, so add it
elem.push_attribute((key, value));
added = true;
}
// writes the event to the writer
writer
.write_event(Event::Start(elem))
.map_err(|e| Error::XmpWriteError(e.to_string()))?;
}
Event::Empty(ref e) if e.name() == QName(RDF_DESCRIPTION) && !added => {
// creates a new element
let mut elem = BytesStart::from_content(
String::from_utf8_lossy(RDF_DESCRIPTION),
RDF_DESCRIPTION.len(),
);
for attr in e.attributes() {
match attr {
Ok(attr) => {
if attr.key == QName(key.as_bytes()) {
// replace the key/value if it exists
elem.push_attribute((key, value));
added = true;
} else {
// add all other existing elements
elem.extend_attributes([attr]);
}
}
Err(e) => {
error!("Error at position {}", reader.buffer_position());
return Err(Error::XmpReadError(e.to_string()));
}
}
}
if !added {
// didn't exist, so add it
elem.push_attribute((key, value));
added = true;
}
// writes the event to the writer
writer
.write_event(Event::Empty(elem))
.map_err(|e| Error::XmpWriteError(e.to_string()))?;
}
Event::Eof => break,
e => {
writer
.write_event(e)
.map_err(|e| Error::XmpWriteError(e.to_string()))?;
}
}
}
// Maintain XMP packet length with padding if possible.
let padding_length = target_length.saturating_sub(writer.get_ref().get_ref().len());
write_xmp_padding(&mut writer.get_mut(), padding_length)?;
let result = writer.into_inner().into_inner();
String::from_utf8(result).map_err(|e| Error::XmpWriteError(e.to_string()))
}
/// Remove a value from XMP by key — the inverse of `add_xmp_key`. Strips the
/// key whether it's stored as an `rdf:Description` attribute (how
/// `add_xmp_key` always writes it) or as a standalone child element (how
/// `extract_xmp_key` also knows how to read it, for interop with XMP this
/// crate didn't produce), leaving everything else in the packet untouched.
fn remove_xmp_key(xmp: &str, key: &str) -> Result<String> {
let orig_length = xmp.len();
let mut target_length = orig_length.max(4096);
let xpacket_end = "<?xpacket end";
let xmp_body = if let Some(pos) = xmp.rfind(xpacket_end) {
// See add_xmp_key: use pos to avoid underflowing target_length.
target_length = pos;
xmp[..pos].trim_end()
} else {
xmp
};
let mut reader = Reader::from_str(xmp_body);
reader.config_mut().trim_text(false);
reader.config_mut().expand_empty_elements = false;
let mut writer = Writer::new(Cursor::new(Vec::new()));
// Set while inside a standalone `<key>...</key>` child element, so its
// contents (text, nested events) are swallowed until the matching end.
let mut skipping_key_element = false;
loop {
let event = reader
.read_event()
.map_err(|e| Error::XmpReadError(e.to_string()))?;
match event {
Event::Start(ref e) if e.name() == QName(RDF_DESCRIPTION) => {
let mut elem = BytesStart::from_content(
String::from_utf8_lossy(RDF_DESCRIPTION),
RDF_DESCRIPTION.len(),
);
for attr in e.attributes() {
match attr {
Ok(attr) if attr.key == QName(key.as_bytes()) => {
// drop this attribute
}
Ok(attr) => {
elem.extend_attributes([attr]);
}
Err(e) => {
error!("Error at position {}", reader.buffer_position());
return Err(Error::XmpReadError(e.to_string()));
}
}
}
writer
.write_event(Event::Start(elem))
.map_err(|e| Error::XmpWriteError(e.to_string()))?;
}
Event::Empty(ref e) if e.name() == QName(RDF_DESCRIPTION) => {
let mut elem = BytesStart::from_content(
String::from_utf8_lossy(RDF_DESCRIPTION),
RDF_DESCRIPTION.len(),
);
for attr in e.attributes() {
match attr {
Ok(attr) if attr.key == QName(key.as_bytes()) => {
// drop this attribute
}
Ok(attr) => {
elem.extend_attributes([attr]);
}
Err(e) => {
error!("Error at position {}", reader.buffer_position());
return Err(Error::XmpReadError(e.to_string()));
}
}
}
writer
.write_event(Event::Empty(elem))
.map_err(|e| Error::XmpWriteError(e.to_string()))?;
}
Event::Start(ref e) if e.name() == QName(key.as_bytes()) => {
skipping_key_element = true;
}
Event::Empty(ref e) if e.name() == QName(key.as_bytes()) => {
// self-closing standalone form — nothing further to skip.
}
Event::End(ref e) if skipping_key_element && e.name() == QName(key.as_bytes()) => {
skipping_key_element = false;
}
Event::Eof => break,
_ if skipping_key_element => {
// swallow contents of the standalone element being removed.
}
e => {
writer
.write_event(e)
.map_err(|e| Error::XmpWriteError(e.to_string()))?;
}
}
}
// Maintain XMP packet length with padding if possible.
let padding_length = target_length.saturating_sub(writer.get_ref().get_ref().len());
write_xmp_padding(&mut writer.get_mut(), padding_length)?;
let result = writer.into_inner().into_inner();
String::from_utf8(result).map_err(|e| Error::XmpWriteError(e.to_string()))
}
/// extract the dc:provenance value from xmp
pub fn extract_provenance(xmp: &str) -> Option<String> {
extract_xmp_key(xmp, "dcterms:provenance")
}
/// extract the xmpMM:InstanceID value from xmp
pub(crate) fn extract_instance_id(xmp: &str) -> Option<String> {
extract_xmp_key(xmp, "xmpMM:InstanceID")
}
/// extract the "xmpMM:DocumentID" value from xmp
pub(crate) fn extract_document_id(xmp: &str) -> Option<String> {
extract_xmp_key(xmp, "xmpMM:DocumentID")
}
/// add or replace a dc:provenance value to xmp, including dc:terms if needed
pub fn add_provenance(xmp: &str, provenance: &str) -> Result<String> {
let xmp = add_xmp_key(xmp, "xmlns:dcterms", "http://purl.org/dc/terms/")?;
add_xmp_key(&xmp, "dcterms:provenance", provenance)
}
/// Remove the dc:provenance value from XMP, if present, leaving everything
/// else in the packet untouched. The inverse of `add_provenance`.
pub fn remove_provenance(xmp: &str) -> Result<String> {
remove_xmp_key(xmp, "dcterms:provenance")
}
/// Add or replace the xmpMM:InstanceID value in XMP, including the xmpMM
/// namespace declaration if needed.
pub(crate) fn set_instance_id(xmp: &str, instance_id: &str) -> Result<String> {
let xmp = add_xmp_key(xmp, "xmlns:xmpMM", "http://ns.adobe.com/xap/1.0/mm/")?;
add_xmp_key(&xmp, "xmpMM:InstanceID", instance_id)
}
// According to the XMP Spec, the recommended practice is to use the ASCII space
// character (U+0020) in the appropriate encoding, with a newline about every 100 characters
fn write_xmp_padding<W: std::io::Write>(writer: &mut W, len: usize) -> std::io::Result<()> {
let mut remaining = len;
// Add newline before trailer no matter what.
writer.write_all(b"\n")?;
remaining = remaining.saturating_sub(1);
if remaining > 0 {
// Account for final newline since we are adding whitespace.
remaining -= 1;
while remaining > 0 {
let chunk = remaining.min(99);
writer.write_all(&vec![b' '; chunk])?;
remaining -= chunk;
if remaining > 0 {
writer.write_all(b"\n")?;
remaining -= 1;
}
}
writer.write_all(b"\n")?;
}
writer.write_all(XMP_END)?;
Ok(())
}
#[cfg(test)]
mod tests {
#![allow(clippy::expect_used)]
#![allow(clippy::unwrap_used)]
//use env_logger;
use c2pa_macros::c2pa_test_async;
#[cfg(all(target_arch = "wasm32", not(target_os = "wasi")))]
use wasm_bindgen_test::wasm_bindgen_test;
use super::*;
use crate::{asset_handlers::jpeg_io::JpegIO, asset_io::AssetIO};
const XMP_DATA: &str = r#"<?xpacket begin="" id="W5M0MpCehiHzreSzNTczkc9d"?>
<x:xmpmeta xmlns:x="adobe:ns:meta/" x:xmptk="contentauth">
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
<rdf:Description rdf:about=""
xmlns:xmp="http://ns.adobe.com/xap/1.0/"
xmlns:xmpMM="http://ns.adobe.com/xap/1.0/mm/"
xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:dcterms="http://purl.org/dc/terms/"
xmpMM:DocumentID="xmp.did:cb9f5498-bb58-4572-8043-8c369e6bfb9b"
xmpMM:InstanceID="xmp.iid:cb9f5498-bb58-4572-8043-8c369e6bfb9b"
dcterms:provenance="self#jumbf=c2pa/contentauth:urn:uuid:a58065fb-79ae-4eb3-87b9-a19830860059/c2pa.claim"
dc:format="image/jpeg">
</rdf:Description>
</rdf:RDF>
</x:xmpmeta>"#;
const PROVENANCE: &str =
"self#jumbf=c2pa/contentauth:urn:uuid:a58065fb-79ae-4eb3-87b9-a19830860059/c2pa.claim";
#[test]
fn read_xmp() {
let provenance = extract_provenance(XMP_DATA);
assert_eq!(provenance, Some("self#jumbf=c2pa/contentauth:urn:uuid:a58065fb-79ae-4eb3-87b9-a19830860059/c2pa.claim".to_owned()));
let document_id = extract_document_id(XMP_DATA);
assert_eq!(
document_id,
Some("xmp.did:cb9f5498-bb58-4572-8043-8c369e6bfb9b".to_owned())
);
let instance_id = extract_instance_id(XMP_DATA);
assert_eq!(
instance_id,
Some("xmp.iid:cb9f5498-bb58-4572-8043-8c369e6bfb9b".to_owned())
);
let unicorn = extract_xmp_key(XMP_DATA, "unicorn");
assert_eq!(unicorn, None);
let bad_xmp = extract_xmp_key("bad xmp", "unicorn");
assert_eq!(bad_xmp, None);
}
#[test]
fn add_xmp() {
let xmp = add_provenance(XMP_DATA, PROVENANCE).expect("adding provenance");
let unicorn = extract_provenance(&xmp);
println!("{xmp}");
assert_eq!(unicorn, Some(PROVENANCE.to_string()));
let xmp = add_provenance(MIN_XMP, PROVENANCE).expect("adding provenance");
let unicorn = extract_provenance(&xmp);
println!("{xmp}");
assert_eq!(unicorn, Some(PROVENANCE.to_string()));
}
// Regression test: a truncated trailer shorter than XMP_END must not underflow target_length.
#[test]
fn add_xmp_key_truncated_trailer_does_not_underflow() {
let truncated = "<?xpacket end";
assert_eq!(truncated.len(), 13);
let xmp = add_provenance(truncated, PROVENANCE).expect("adding provenance");
assert_eq!(extract_provenance(&xmp), None);
}
#[test]
fn remove_xmp_key_truncated_trailer_does_not_underflow() {
let truncated = "<?xpacket end";
assert_eq!(truncated.len(), 13);
let xmp = remove_provenance(truncated).expect("removing provenance");
assert_eq!(extract_provenance(&xmp), None);
}
// Control: a complete trailer should still round-trip normally.
#[test]
fn add_xmp_key_full_trailer_still_works() {
let xmp = add_provenance(XMP_DATA, PROVENANCE).expect("adding provenance");
assert_eq!(extract_provenance(&xmp), Some(PROVENANCE.to_string()));
assert!(xmp
.trim_end()
.ends_with(std::str::from_utf8(XMP_END).unwrap()));
}
#[c2pa_test_async]
async fn test_broken_xmp_read_write_stream() {
let source_bytes = include_bytes!("../../tests/fixtures/IMG_0003.jpg");
let test_msg = "https://cai-manifests-stage.adobe.com/manifests/urn-c2pa-0ab6e8b8-5c28-4ef1-8f58-86c21f0349bf-adobe";
let handler = JpegIO::new("");
let assetio_handler = handler.get_handler("jpg");
let remote_ref_handler = assetio_handler.remote_manifest_url_ref().unwrap();
let mut source_stream = Cursor::new(source_bytes.to_vec());
let mut output_stream = Cursor::new(Vec::new());
let original_xmp_length = assetio_handler
.get_reader()
.read_xmp(&mut source_stream)
.unwrap()
.len();
source_stream.set_position(0);
remote_ref_handler
.write_remote_manifest_url(&mut source_stream, &mut output_stream, test_msg)
.unwrap();
output_stream.set_position(0);
// read back in XMP
let read_xmp = assetio_handler
.get_reader()
.read_xmp(&mut output_stream)
.unwrap();
output_stream.set_position(0);
//std::fs::write("../target/xmp_write.jpg", output_stream.into_inner()).unwrap();
assert!(read_xmp.contains(test_msg));
assert_eq!(read_xmp.len(), original_xmp_length);
}
}