[wasm-metadata] add support for OCI licenses (#1937)

Author: yoshuawuyts

crates/wasm-metadata/src/add_metadata.rs

@@ -1,4 +1,4 @@
-use crate::{rewrite_wasm, Author, Description, Producers, RegistryMetadata};
+use crate::{rewrite_wasm, Author, Description, Licenses, Producers, RegistryMetadata};
 
 use anyhow::Result;
 
@@ -34,6 +34,10 @@ pub struct AddMetadata {
     #[cfg_attr(feature = "clap", clap(long, value_name = "NAME"))]
     pub description: Option<Description>,
 
+    /// License(s) under which contained software is distributed as an SPDX License Expression.
+    #[cfg_attr(feature = "clap", clap(long, value_name = "NAME"))]
+    pub licenses: Option<Licenses>,
+
     /// Add an registry metadata to the registry-metadata section
     #[cfg_attr(feature="clap", clap(long, value_parser = parse_registry_metadata_value, value_name="PATH"))]
     pub registry_metadata: Option<RegistryMetadata>,
@@ -65,6 +69,7 @@ impl AddMetadata {
             &Producers::from_meta(self),
             &self.author,
             &self.description,
+            &self.licenses,
             self.registry_metadata.as_ref(),
             input,
         )

crates/wasm-metadata/src/lib.rs

@@ -5,7 +5,7 @@
 pub use add_metadata::AddMetadata;
 pub use metadata::Metadata;
 pub use names::{ComponentNames, ModuleNames};
-pub use oci_annotations::{Author, Description};
+pub use oci_annotations::{Author, Description, Licenses};
 pub use producers::{Producers, ProducersField};
 pub use registry::{CustomLicense, Link, LinkType, RegistryMetadata};
 

crates/wasm-metadata/src/metadata.rs

@@ -4,7 +4,9 @@ use std::fmt;
 use std::ops::Range;
 use wasmparser::{KnownCustom, Parser, Payload::*};
 
-use crate::{Author, ComponentNames, Description, ModuleNames, Producers, RegistryMetadata};
+use crate::{
+    Author, ComponentNames, Description, Licenses, ModuleNames, Producers, RegistryMetadata,
+};
 
 /// A tree of the metadata found in a WebAssembly binary.
 #[derive(Debug, Serialize)]
@@ -22,6 +24,8 @@ pub enum Metadata {
         author: Option<Author>,
         /// Human-readable description of the binary
         description: Option<Description>,
+        /// License(s) under which contained software is distributed as an SPDX License Expression.
+        licenses: Option<Licenses>,
         /// All child modules and components inside the component.
         children: Vec<Box<Metadata>>,
         /// Byte range of the module in the parent binary
@@ -39,6 +43,8 @@ pub enum Metadata {
         author: Option<Author>,
         /// Human-readable description of the binary
         description: Option<Description>,
+        /// License(s) under which contained software is distributed as an SPDX License Expression.
+        licenses: Option<Licenses>,
         /// Byte range of the module in the parent binary
         range: Range<usize>,
     },
@@ -128,6 +134,13 @@ impl Metadata {
                             Metadata::Component { description, .. } => *description = Some(a),
                         }
                     }
+                    KnownCustom::Unknown if c.name() == "licenses" => {
+                        let a = Licenses::parse_custom_section(&c)?;
+                        match metadata.last_mut().expect("non-empty metadata stack") {
+                            Metadata::Module { licenses, .. } => *licenses = Some(a),
+                            Metadata::Component { licenses, .. } => *licenses = Some(a),
+                        }
+                    }
                     _ => {}
                 },
                 _ => {}
@@ -144,6 +157,7 @@ impl Metadata {
             producers: None,
             author: None,
             description: None,
+            licenses: None,
             registry_metadata: None,
             children: Vec::new(),
             range,
@@ -156,6 +170,7 @@ impl Metadata {
             producers: None,
             author: None,
             description: None,
+            licenses: None,
             registry_metadata: None,
             range,
         }

crates/wasm-metadata/src/oci_annotations/licenses.rs

@@ -0,0 +1,122 @@
+use std::borrow::Cow;
+use std::fmt::{self, Display};
+use std::str::FromStr;
+
+use anyhow::{ensure, Error, Result};
+use serde::Serialize;
+use wasm_encoder::{ComponentSection, CustomSection, Encode, Section};
+use wasmparser::CustomSectionReader;
+
+/// License(s) under which contained software is distributed as an SPDX License Expression.
+#[derive(Debug, Clone, PartialEq)]
+pub struct Licenses(CustomSection<'static>);
+
+impl Licenses {
+    /// Create a new instance of `Licenses`.
+    pub fn new(s: &str) -> Result<Self> {
+        Ok(spdx::Expression::parse(s)?.into())
+    }
+
+    /// Parse a `licenses` custom section from a wasm binary.
+    pub(crate) fn parse_custom_section(reader: &CustomSectionReader<'_>) -> Result<Self> {
+        ensure!(
+            reader.name() == "licenses",
+            "The `licenses` custom section should have a name of 'license'"
+        );
+        let data = String::from_utf8(reader.data().to_owned())?;
+        Self::new(&data)
+    }
+}
+
+impl FromStr for Licenses {
+    type Err = Error;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        Self::new(s)
+    }
+}
+
+impl From<spdx::Expression> for Licenses {
+    fn from(expression: spdx::Expression) -> Self {
+        Self(CustomSection {
+            name: "licenses".into(),
+            data: Cow::Owned(expression.to_string().into_bytes()),
+        })
+    }
+}
+
+impl Serialize for Licenses {
+    fn serialize<S>(&self, serializer: S) -> std::result::Result<S::Ok, S::Error>
+    where
+        S: serde::Serializer,
+    {
+        serializer.serialize_str(&self.to_string())
+    }
+}
+
+impl Display for Licenses {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        // NOTE: this will never panic since we always guarantee the data is
+        // encoded as utf8, even if we internally store it as [u8].
+        let data = String::from_utf8(self.0.data.to_vec()).unwrap();
+        write!(f, "{data}")
+    }
+}
+
+impl ComponentSection for Licenses {
+    fn id(&self) -> u8 {
+        ComponentSection::id(&self.0)
+    }
+}
+
+impl Section for Licenses {
+    fn id(&self) -> u8 {
+        Section::id(&self.0)
+    }
+}
+
+impl Encode for Licenses {
+    fn encode(&self, sink: &mut Vec<u8>) {
+        self.0.encode(sink);
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+    use wasm_encoder::Component;
+    use wasmparser::Payload;
+
+    #[test]
+    fn roundtrip() {
+        let mut component = Component::new();
+        component.section(
+            &Licenses::new("Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT").unwrap(),
+        );
+        let component = component.finish();
+
+        let mut parsed = false;
+        for section in wasmparser::Parser::new(0).parse_all(&component) {
+            if let Payload::CustomSection(reader) = section.unwrap() {
+                let description = Licenses::parse_custom_section(&reader).unwrap();
+                assert_eq!(
+                    description.to_string(),
+                    "Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT"
+                );
+                parsed = true;
+            }
+        }
+        assert!(parsed);
+    }
+
+    #[test]
+    fn serialize() {
+        let description =
+            Licenses::new("Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT").unwrap();
+        let json = serde_json::to_string(&description).unwrap();
+        assert_eq!(
+            r#""Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT""#,
+            json
+        );
+    }
+}

crates/wasm-metadata/src/oci_annotations/mod.rs

@@ -17,6 +17,8 @@
 
 pub use author::Author;
 pub use description::Description;
+pub use licenses::Licenses;
 
 mod author;
 mod description;
+mod licenses;

crates/wasm-metadata/src/producers.rs

@@ -148,7 +148,7 @@ impl Producers {
     /// Merge into an existing wasm module. Rewrites the module with this producers section
     /// merged into its existing one, or adds this producers section if none is present.
     pub fn add_to_wasm(&self, input: &[u8]) -> Result<Vec<u8>> {
-        rewrite_wasm(&None, self, &None, &None, None, input)
+        rewrite_wasm(&None, self, &None, &None, &None, None, input)
     }
 
     pub(crate) fn display(&self, f: &mut fmt::Formatter, indent: usize) -> fmt::Result {

crates/wasm-metadata/src/registry.rs

@@ -44,7 +44,15 @@ impl RegistryMetadata {
     /// Merge into an existing wasm module. Rewrites the module with this registry-metadata section
     /// overwriting its existing one, or adds this registry-metadata section if none is present.
     pub fn add_to_wasm(&self, input: &[u8]) -> Result<Vec<u8>> {
-        rewrite_wasm(&None, &Producers::empty(), &None, &None, Some(&self), input)
+        rewrite_wasm(
+            &None,
+            &Producers::empty(),
+            &None,
+            &None,
+            &None,
+            Some(&self),
+            input,
+        )
     }
 
     /// Parse a Wasm binary and extract the `Registry` section, if there is any.

crates/wasm-metadata/src/rewrite.rs

@@ -1,4 +1,6 @@
-use crate::{Author, ComponentNames, Description, ModuleNames, Producers, RegistryMetadata};
+use crate::{
+    Author, ComponentNames, Description, Licenses, ModuleNames, Producers, RegistryMetadata,
+};
 use anyhow::Result;
 use std::borrow::Cow;
 use std::mem;
@@ -11,6 +13,7 @@ pub(crate) fn rewrite_wasm(
     add_producers: &Producers,
     add_author: &Option<Author>,
     add_description: &Option<Description>,
+    add_licenses: &Option<Licenses>,
     add_registry_metadata: Option<&RegistryMetadata>,
     input: &[u8],
 ) -> Result<Vec<u8>> {
@@ -106,6 +109,13 @@ pub(crate) fn rewrite_wasm(
                             continue;
                         }
                     }
+                    KnownCustom::Unknown if c.name() == "licenses" => {
+                        if add_licenses.is_none() {
+                            let licenses = Licenses::parse_custom_section(c)?;
+                            licenses.append_to(&mut output);
+                            continue;
+                        }
+                    }
                     _ => {}
                 }
             }
@@ -141,6 +151,9 @@ pub(crate) fn rewrite_wasm(
     if let Some(description) = add_description {
         description.append_to(&mut output);
     }
+    if let Some(licenses) = add_licenses {
+        licenses.append_to(&mut output);
+    }
     if add_registry_metadata.is_some() {
         let registry_metadata = wasm_encoder::CustomSection {
             name: Cow::Borrowed("registry-metadata"),

crates/wasm-metadata/tests/component.rs

@@ -13,6 +13,9 @@ fn add_to_empty_component() {
         sdk: vec![],
         author: Some(Author::new("Chashu Cat")),
         description: Some(Description::new("Chashu likes tuna")),
+        licenses: Some(
+            Licenses::new("Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT").unwrap(),
+        ),
         registry_metadata: Some(RegistryMetadata {
             authors: Some(vec!["foo".to_owned()]),
             description: Some("foo bar baz".to_owned()),
@@ -46,6 +49,7 @@ fn add_to_empty_component() {
             registry_metadata,
             author,
             description,
+            licenses,
             children,
             range,
         } => {
@@ -63,6 +67,10 @@ fn add_to_empty_component() {
 
             assert_eq!(author.unwrap(), Author::new("Chashu Cat"));
             assert_eq!(description.unwrap(), Description::new("Chashu likes tuna"));
+            assert_eq!(
+                licenses.unwrap(),
+                Licenses::new("Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT").unwrap()
+            );
 
             let registry_metadata = registry_metadata.unwrap();
 
@@ -106,7 +114,7 @@ fn add_to_empty_component() {
             );
 
             assert_eq!(range.start, 0);
-            assert_eq!(range.end, 485);
+            assert_eq!(range.end, 547);
         }
         _ => panic!("metadata should be component"),
     }
@@ -123,6 +131,9 @@ fn add_to_nested_component() {
         sdk: vec![],
         author: Some(Author::new("Chashu Cat")),
         description: Some(Description::new("Chashu likes tuna")),
+        licenses: Some(
+            Licenses::new("Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT").unwrap(),
+        ),
         registry_metadata: Some(RegistryMetadata {
             authors: Some(vec!["Foo".to_owned()]),
             ..Default::default()
@@ -169,6 +180,7 @@ fn add_to_nested_component() {
                     name,
                     producers,
                     author,
+                    licenses,
                     registry_metadata,
                     range,
                     description,
@@ -186,6 +198,13 @@ fn add_to_nested_component() {
 
                     assert_eq!(author, &Some(Author::new("Chashu Cat")));
                     assert_eq!(description, &Some(Description::new("Chashu likes tuna")));
+                    assert_eq!(
+                        licenses,
+                        &Some(
+                            Licenses::new("Apache-2.0 WITH LLVM-exception OR Apache-2.0 OR MIT")
+                                .unwrap()
+                        )
+                    );
 
                     let registry_metadata = registry_metadata.as_ref().unwrap();
                     assert_eq!(
@@ -194,7 +213,7 @@ fn add_to_nested_component() {
                     );
 
                     assert_eq!(range.start, 11);
-                    assert_eq!(range.end, 174);
+                    assert_eq!(range.end, 236);
                 }
                 _ => panic!("child is a module"),
             }