feat: c2patool to read cawg data (#904)

* ci: Add debug statements * Omit additional paddings * feat: Omit more padding tags * ci: add explanation * ci: Catch result strings before they get returned * ci: sdk catch result string before returning them * ci: look into parsing stuff * ci: WIP for cawg display * ci: Replace signature with hash * ci: Replace signature with hash * ci: Clean up leftover debug code * ci: Parse cawg data in cli, introduce tokio * ci: WIP * ci: WIP * ci: Up some deps * ci: WIP * ci: WIP * feat: Add methods to reader to decorate the json * ci: Got the cleaned content signature * ci: Some clean up * ci: Some clean up * ci: WIP * feat: cawg parsing hack * feat: move display * feat: Refactor * feat: Refactor * feat: Refactor * ci: Tmp check all other tests * ci: Tmp check all other tests * Debug * fix: fix mistake of returning too early * fix: Remove debug file * fix: Remove debug file * fix: Change error handling
contentauth · Feb 6, 2025 · b1990a3 · b1990a3
1 parent 4cb20e3
commit b1990a3
Show file tree

Hide file tree

Showing 13 changed files with 184 additions and 14 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Makefile b/Makefile
@@ -31,6 +31,8 @@ test-wasm:
 test-wasm-web:
 	cd sdk && wasm-pack test --chrome --headless -- --features="serialize_thumbnails"
 
+test-no-wasm: check-format check-docs clippy test-local
+
 # Full local validation, build and test all features including wasm
 # Run this before pushing a PR to pre-validate
 test: check-format check-docs clippy test-local test-wasm-web

diff --git a/cawg_identity/src/claim_aggregation/ica_signature_verifier.rs b/cawg_identity/src/claim_aggregation/ica_signature_verifier.rs
@@ -153,7 +153,7 @@ impl SignatureVerifier for IcaSignatureVerifier {
                     ));
                 };
 
-                dbg!(&jwk_prop);
+                // dbg!(&jwk_prop);
 
                 // OMG SO HACKY!
                 let Ok(jwk_json) = serde_json::to_string_pretty(jwk_prop) else {

diff --git a/cawg_identity/src/claim_aggregation/w3c_vc/did_web.rs b/cawg_identity/src/claim_aggregation/w3c_vc/did_web.rs
@@ -68,7 +68,7 @@ pub(crate) async fn resolve(did: &Did<'_>) -> Result<DidDocument, DidWebError> {
 
     let method_specific_id = did.method_specific_id();
 
-    dbg!(method_specific_id);
+    //dbg!(method_specific_id);
 
     let url = to_url(method_specific_id)?;
     // TODO: https://w3c-ccg.github.io/did-method-web/#in-transit-security

diff --git a/cawg_identity/src/claim_aggregation/w3c_vc/serialization.rs b/cawg_identity/src/claim_aggregation/w3c_vc/serialization.rs
@@ -57,8 +57,6 @@ pub(crate) mod one_or_many {
         where
             M: de::MapAccess<'de>,
         {
-            eprintln!("Yo!");
-
             let one = Deserialize::deserialize(de::value::MapAccessDeserializer::new(map))?;
 
             Ok(nev!(one))

diff --git a/cli/Cargo.toml b/cli/Cargo.toml
@@ -22,6 +22,7 @@ unexpected_cfgs = { level = "warn", check-cfg = ['cfg(test)'] }
 [dependencies]
 anyhow = "1.0"
 atree = "0.5.2"
+cawg-identity = { path = "../cawg_identity"}
 c2pa = { path = "../sdk", version = "0.45.1", features = [
 	"fetch_remote_manifests",
 	"file_io",
@@ -37,6 +38,7 @@ serde = { version = "1.0", features = ["derive"] }
 serde_derive = "1.0"
 serde_json = "1.0"
 tempfile = "3.3"
+tokio = { version = "1.42", features = ["full"] }
 treeline = "0.1.0"
 pem = "3.0.3"
 openssl = { version = "0.10.61", features = ["vendored"] }

diff --git a/cli/src/main.rs b/cli/src/main.rs
@@ -27,10 +27,13 @@ use std::{
 
 use anyhow::{anyhow, bail, Context, Result};
 use c2pa::{Builder, ClaimGeneratorInfo, Error, Ingredient, ManifestDefinition, Reader, Signer};
+use cawg_identity::{claim_aggregation::IcaSignatureVerifier, IdentityAssertion};
 use clap::{Parser, Subcommand};
 use log::debug;
 use serde::Deserialize;
+use serde_json::{Map, Value};
 use signer::SignConfig;
+use tokio::runtime::Runtime;
 use url::Url;
 
 use crate::{
@@ -424,6 +427,144 @@ fn verify_fragmented(init_pattern: &Path, frag_pattern: &Path) -> Result<Vec<Rea
     Ok(readers)
 }
 
+fn decorate_json_display(reader: Reader, tokio_runtime: &Runtime) -> String {
+    let mut reader_content = match reader.json_value_map() {
+        Ok(mapped_json) => mapped_json,
+        Err(_) => {
+            println!("Could not parse manifest store JSON content");
+            return String::new();
+        }
+    };
+
+    let json_content = match reader_content.get_mut("manifests") {
+        Some(json) => json,
+        None => {
+            println!("No JSON to parse in manifest store (key: manifests)");
+            return String::new();
+        }
+    };
+
+    // Update manifests with CAWG details
+    if let Value::Object(map) = json_content {
+        // Iterate over the key-value pairs
+        for (key, value) in &mut *map {
+            // Get additional CAWG details
+            let current_manifest: &c2pa::Manifest = reader.get_manifest(key).unwrap();
+
+            // Replace the signature content of the assertion with parsed details by CAWG
+            let assertions = value.get_mut("assertions").unwrap();
+            let assertions_array = assertions.as_array_mut().unwrap();
+            for assertion in assertions_array {
+                let label = assertion.get("label").unwrap().to_string();
+
+                // for CAWG assertions, further parse the signature
+                if label.contains("cawg.identity") {
+                    let parsed_cawg_json_string =
+                        match get_cawg_details_for_manifest(current_manifest, tokio_runtime) {
+                            Some(parsed_cawg_json_string) => parsed_cawg_json_string,
+                            None => {
+                                println!("Could not parse CAWG details for manifest");
+                                continue;
+                            }
+                        };
+
+                    let assertion_data: &mut serde_json::Value = assertion.get_mut("data").unwrap();
+                    assertion_data["signature"] =
+                        serde_json::from_str(&parsed_cawg_json_string).unwrap();
+                }
+            }
+        }
+    } else {
+        println!("Could not parse manifest store JSON content");
+    }
+
+    match serde_json::to_string_pretty(&reader_content) {
+        Ok(decorated_result) => decorated_result,
+        Err(err) => {
+            println!(
+                "Could not parse manifest store JSON content with additional CAWG details: {:?}",
+                err
+            );
+            String::new()
+        }
+    }
+}
+
+/// Parse additional CAWG details from the manifest store to update displayed results.
+/// As CAWG mostly async, this will block on network requests for checks using a tokio runtime.
+fn get_cawg_details_for_manifest(
+    manifest: &c2pa::Manifest,
+    tokio_runtime: &Runtime,
+) -> Option<String> {
+    let ia_iter = IdentityAssertion::from_manifest(manifest);
+
+    // TODO: Determine what should happen when multiple identities are reported (currently only 1 is supported)
+    let mut parsed_cawg_json = String::new();
+
+    ia_iter.for_each(|ia| {
+        let identity_assertion = match ia {
+            Ok(ia) => ia,
+            Err(err) => {
+                println!("Could not parse identity assertion: {:?}", err);
+                return;
+            }
+        };
+
+        let isv = IcaSignatureVerifier {};
+        let ica_validated = tokio_runtime.block_on(identity_assertion.validate(manifest, &isv));
+        let ica = match ica_validated {
+            Ok(ica) => ica,
+            Err(err) => {
+                println!("Could not validate identity assertion: {:?}", err);
+                return;
+            }
+        };
+
+        parsed_cawg_json = serde_json::to_string(&ica).unwrap();
+    });
+
+    // Get the JSON as mutable, so we can further parse and format
+    let maybe_map = serde_json::from_str(parsed_cawg_json.as_str());
+    let mut map: Map<String, Value> = match maybe_map {
+        Ok(map) => map,
+        Err(err) => {
+            println!("Could not parse CAWG details for manifest: {:?}", err);
+            return None;
+        }
+    };
+
+    // Get the credentials subject information...
+    let credentials_subject = map.get_mut("credentialSubject");
+    let credentials_subject = match credentials_subject {
+        Some(credentials_subject) => credentials_subject,
+        None => {
+            println!("Could not find credentialSubject in CAWG details for manifest");
+            return None;
+        }
+    };
+    let credentials_subject_as_obj = credentials_subject.as_object_mut();
+    let credential_subject_details = match credentials_subject_as_obj {
+        Some(credentials_subject) => credentials_subject,
+        None => {
+            println!("Could not parse credential subject as object in CAWG details for manifest");
+            return None;
+        }
+    };
+    // As per design CAWG has some repetition between assertion an signature (c2paAsset field)
+    // so we remove the c2paAsset field from the credential subject details too
+    credential_subject_details.remove("c2paAsset");
+
+    // return the for-display json-formatted string
+    let serialized_content = serde_json::to_string(&map);
+    match serialized_content {
+        Ok(serialized_content) => Some(serialized_content),
+        Err(err) => {
+            println!("Could not parse CAWG details for manifest: {:?}", err);
+            None
+        }
+    }
+}
+
 fn main() -> Result<()> {
     let args = CliArgs::parse();
 
@@ -464,6 +605,9 @@ fn main() -> Result<()> {
     // configure the SDK
     configure_sdk(&args).context("Could not configure c2pa-rs")?;
 
+    // configure tokio runtime for blocking operations
+    let tokio_runtime: Runtime = Runtime::new()?;
+
     // Remove manifest needs to also remove XMP provenance
     // if args.remove_manifest {
     //     match args.output {
@@ -674,10 +818,9 @@ fn main() -> Result<()> {
             Ingredient::from_file(&args.path).map_err(special_errs)?
         )
     } else if args.detailed {
-        println!(
-            "{:#?}",
-            Reader::from_file(&args.path).map_err(special_errs)?
-        )
+        println!("## TMN-Debug ~ cli#main ~ Here we read the detailed edition");
+        let reader = Reader::from_file(&args.path).map_err(special_errs)?;
+        println!("{:#?}", reader)
     } else if let Some(Commands::Fragment {
         fragments_glob: Some(fg),
     }) = &args.command
@@ -689,7 +832,9 @@ fn main() -> Result<()> {
             println!("{} Init manifests validated", stores.len());
         }
     } else {
-        println!("{}", Reader::from_file(&args.path).map_err(special_errs)?)
+        let reader: Reader = Reader::from_file(&args.path).map_err(special_errs)?;
+        let stringified_decorated_json = decorate_json_display(reader, &tokio_runtime);
+        println!("{}", stringified_decorated_json);
     }
 
     Ok(())

diff --git a/cli/tests/fixtures/C_with_CAWG_data.jpg b/cli/tests/fixtures/C_with_CAWG_data.jpg
diff --git a/sdk/Cargo.toml b/sdk/Cargo.toml
@@ -115,7 +115,7 @@ serde_bytes = "0.11.5"
 serde_cbor = "0.11.1"
 serde_derive = "1.0.197"
 serde_json = { version = "1.0.117", features = ["preserve_order"] }
-serde_with = "3.11.0"
+serde_with = { version = "3.12.0" }
 serde-transcode = "1.1.1"
 sha1 = "0.10.6"
 sha2 = "0.10.6"

diff --git a/sdk/src/error.rs b/sdk/src/error.rs
@@ -252,6 +252,9 @@ pub enum Error {
     #[error("The Verifiable Content structure is not valid")]
     VerifiableCredentialInvalid,
 
+    #[error("error while serializing to JSON: {0}")]
+    JsonSerializationError(String),
+
     /// Could not parse ECDSA signature. (Only appears when using WASM web crypto.)
     #[error("could not parse ECDSA signature")]
     InvalidEcdsaSignature,

diff --git a/sdk/src/hashed_uri.rs b/sdk/src/hashed_uri.rs
@@ -112,7 +112,7 @@ mod tests {
     fn impl_clone() {
         let h = HashedUri::new(
             "self#jumbf=c2pa/urn:uuid:F9168C5E-CEB2-4faa-B6BF-329BF39FA1E4/c2pa.assertions/c2pa.hash.data".to_owned(),
-             Some("sha256".to_owned()),
+            Some("sha256".to_owned()),
             &hex!("53d1b2cf4e6d9a97ed9281183fa5d836c32751b9d2fca724b40836befee7d67f"),
         );
 
@@ -124,7 +124,7 @@ mod tests {
     fn impl_debug() {
         let h = HashedUri::new(
             "self#jumbf=c2pa/urn:uuid:F9168C5E-CEB2-4faa-B6BF-329BF39FA1E4/c2pa.assertions/c2pa.hash.data".to_owned(),
-             Some("sha256".to_owned()),
+            Some("sha256".to_owned()),
             &hex!("53d1b2cf4e6d9a97ed9281183fa5d836c32751b9d2fca724b40836befee7d67f"),
         );
 
@@ -135,7 +135,7 @@ mod tests {
     fn impl_display() {
         let h = HashedUri::new(
             "self#jumbf=c2pa/urn:uuid:F9168C5E-CEB2-4faa-B6BF-329BF39FA1E4/c2pa.assertions/c2pa.hash.data".to_owned(),
-             Some("sha256".to_owned()),
+            Some("sha256".to_owned()),
             &hex!("53d1b2cf4e6d9a97ed9281183fa5d836c32751b9d2fca724b40836befee7d67f"),
         );
 

diff --git a/sdk/src/manifest_store.rs b/sdk/src/manifest_store.rs
@@ -589,7 +589,13 @@ impl std::fmt::Display for ManifestStore {
         }
 
         json = b64_tag(json, "hash");
+
+        // list of tags to omit (padding tags)
+        // Reason of padding, see note at:
+        // https://c2pa.org/specifications/specifications/2.1/specs/C2PA_Specification.html#_going_back_and_filling_in
         json = omit_tag(json, "pad");
+        json = omit_tag(json, "pad1");
+        json = omit_tag(json, "pad2");
 
         f.write_str(&json)
     }

diff --git a/sdk/src/reader.rs b/sdk/src/reader.rs
@@ -23,6 +23,7 @@ use c2pa_status_tracker::DetailedStatusTracker;
 #[cfg(feature = "json_schema")]
 use schemars::JsonSchema;
 use serde::{Deserialize, Serialize};
+use serde_json::{Map, Value};
 
 #[cfg(feature = "file_io")]
 use crate::error::Error;
@@ -241,11 +242,22 @@ impl Reader {
         })
     }
 
-    /// Get the manifest store as a JSON string
+    /// Get the manifest store as a JSON string.
     pub fn json(&self) -> String {
         self.manifest_store.to_string()
     }
 
+    /// Get the manifest store as a serde serialized JSON value map.
+    pub fn json_value_map(&self) -> Result<Map<String, Value>> {
+        let reader_as_json = self.json();
+        let reader_as_json_str = reader_as_json.as_str();
+        let mapped_json = serde_json::from_str(reader_as_json_str);
+        match mapped_json {
+            Ok(mapped_json) => Ok(mapped_json),
+            Err(err) => Err(crate::Error::JsonSerializationError(err.to_string())),
+        }
+    }
+
     /// Get the [`ValidationStatus`] array of the manifest store if it exists.
     /// Call this method to check for validation errors.
     ///