[wasm-metadata] parse OCI author custom section (bytecodealliance#1925)

* init OCI author parsing

* add serialization

* fix tests

* fix json formatting test

* add author parsing

crates/wasm-encoder/src/core/custom.rs

@@ -3,7 +3,7 @@ use std::borrow::Cow;
 use crate::{encoding_size, Encode, Section, SectionId};
 
 /// A custom section holding arbitrary data.
-#[derive(Clone, Debug)]
+#[derive(Clone, Debug, PartialEq)]
 pub struct CustomSection<'a> {
     /// The name of this custom section.
     pub name: Cow<'a, str>,

crates/wasm-metadata/src/add_metadata.rs

@@ -1,4 +1,4 @@
-use crate::{rewrite_wasm, Producers, RegistryMetadata};
+use crate::{rewrite_wasm, Author, Producers, RegistryMetadata};
 
 use anyhow::Result;
 
@@ -25,6 +25,11 @@ pub struct AddMetadata {
     #[cfg_attr(feature="clap", clap(long, value_parser = parse_key_value, value_name="NAME=VERSION"))]
     pub sdk: Vec<(String, String)>,
 
+    /// Contact details of the people or organization responsible,
+    /// encoded as a freeform string.
+    #[cfg_attr(feature = "clap", clap(long, value_name = "NAME"))]
+    pub author: Option<Author>,
+
     /// 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>,
@@ -54,6 +59,7 @@ impl AddMetadata {
         rewrite_wasm(
             &self.name,
             &Producers::from_meta(self),
+            &self.author,
             self.registry_metadata.as_ref(),
             input,
         )

crates/wasm-metadata/src/lib.rs

@@ -5,6 +5,7 @@
 pub use add_metadata::AddMetadata;
 pub use metadata::Metadata;
 pub use names::{ComponentNames, ModuleNames};
+pub use oci_annotations::Author;
 pub use producers::{Producers, ProducersField};
 pub use registry::{CustomLicense, Link, LinkType, RegistryMetadata};
 
@@ -13,6 +14,7 @@ pub(crate) use rewrite::rewrite_wasm;
 mod add_metadata;
 mod metadata;
 mod names;
+mod oci_annotations;
 mod producers;
 mod registry;
 mod rewrite;

crates/wasm-metadata/src/metadata.rs

@@ -4,7 +4,7 @@ use std::fmt;
 use std::ops::Range;
 use wasmparser::{KnownCustom, Parser, Payload::*};
 
-use crate::{ComponentNames, ModuleNames, Producers, RegistryMetadata};
+use crate::{Author, ComponentNames, ModuleNames, Producers, RegistryMetadata};
 
 /// A tree of the metadata found in a WebAssembly binary.
 #[derive(Debug, Serialize)]
@@ -18,6 +18,8 @@ pub enum Metadata {
         producers: Option<Producers>,
         /// The component's registry metadata section, if any.
         registry_metadata: Option<RegistryMetadata>,
+        /// The component's author section, if any.
+        author: Option<Author>,
         /// All child modules and components inside the component.
         children: Vec<Box<Metadata>>,
         /// Byte range of the module in the parent binary
@@ -31,6 +33,8 @@ pub enum Metadata {
         producers: Option<Producers>,
         /// The module's registry metadata section, if any.
         registry_metadata: Option<RegistryMetadata>,
+        /// The component's author section, if any.
+        author: Option<Author>,
         /// Byte range of the module in the parent binary
         range: Range<usize>,
     },
@@ -106,6 +110,13 @@ impl Metadata {
                             .expect("non-empty metadata stack")
                             .set_registry_metadata(registry);
                     }
+                    KnownCustom::Unknown if c.name() == "author" => {
+                        let a = Author::parse_custom_section(&c)?;
+                        match metadata.last_mut().expect("non-empty metadata stack") {
+                            Metadata::Module { author, .. } => *author = Some(a),
+                            Metadata::Component { author, .. } => *author = Some(a),
+                        }
+                    }
                     _ => {}
                 },
                 _ => {}
@@ -120,6 +131,7 @@ impl Metadata {
         Metadata::Component {
             name: None,
             producers: None,
+            author: None,
             registry_metadata: None,
             children: Vec::new(),
             range,
@@ -130,6 +142,7 @@ impl Metadata {
         Metadata::Module {
             name: None,
             producers: None,
+            author: None,
             registry_metadata: None,
             range,
         }

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

@@ -0,0 +1,111 @@
+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;
+
+/// Contact details of the people or organization responsible,
+/// encoded as a freeform string.
+#[derive(Debug, Clone, PartialEq)]
+pub struct Author(CustomSection<'static>);
+
+impl Author {
+    /// Create a new instance of `Author`.
+    pub fn new<S: Into<Cow<'static, str>>>(s: S) -> Self {
+        Self(CustomSection {
+            name: "author".into(),
+            data: match s.into() {
+                Cow::Borrowed(s) => Cow::Borrowed(s.as_bytes()),
+                Cow::Owned(s) => Cow::Owned(s.into()),
+            },
+        })
+    }
+
+    /// Parse an `author` custom section from a wasm binary.
+    pub(crate) fn parse_custom_section(reader: &CustomSectionReader<'_>) -> Result<Self> {
+        ensure!(
+            reader.name() == "author",
+            "The `author` custom section should have a name of 'author'"
+        );
+        let data = String::from_utf8(reader.data().to_owned())?;
+        Ok(Self::new(data))
+    }
+}
+
+impl FromStr for Author {
+    type Err = Error;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        Ok(Self::new(s.to_owned()))
+    }
+}
+
+impl Serialize for Author {
+    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 Author {
+    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 Author {
+    fn id(&self) -> u8 {
+        ComponentSection::id(&self.0)
+    }
+}
+
+impl Section for Author {
+    fn id(&self) -> u8 {
+        Section::id(&self.0)
+    }
+}
+
+impl Encode for Author {
+    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(&Author::new("Nori Cat"));
+        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 author = Author::parse_custom_section(&reader).unwrap();
+                assert_eq!(author.to_string(), "Nori Cat");
+                parsed = true;
+            }
+        }
+        assert!(parsed);
+    }
+
+    #[test]
+    fn serialize() {
+        let author = Author::new("Chashu Cat");
+        let json = serde_json::to_string(&author).unwrap();
+        assert_eq!(r#""Chashu Cat""#, json);
+    }
+}

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

@@ -0,0 +1,20 @@
+//! Annotations following the [OCI Annotations Spec].
+//!
+//! The fields of these annotations are encoded into custom sections of
+//! component binaries, and are explicitly compatible with the OCI Annotations
+//! Spec. That enables Compontents to be encoded to OCI and back without needing
+//! to perform any additional parsing. This greatly simplifies adding metadata to
+//! component registries, since language-native component toolchains can encode them
+//! directly into components. Which in turn can be picked up by Component-to-OCI
+//! tooling to take those annotations and display them in a way that registries can
+//! understand.
+//!
+//! For the files in this submodule that means we want to be explicitly
+//! compatible with the OCI Annotations specification. Any deviation in our
+//! parsing rules from the spec should be considered a bug we have to fix.
+//!
+//! [OCI Annotations Spec]: https://specs.opencontainers.org/image-spec/annotations/
+
+pub use author::Author;
+
+mod author;

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, input)
+        rewrite_wasm(&None, self, &None, None, input)
     }
 
     pub(crate) fn display(&self, f: &mut fmt::Formatter, indent: usize) -> fmt::Result {

crates/wasm-metadata/src/registry.rs

@@ -44,7 +44,7 @@ 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(), Some(&self), input)
+        rewrite_wasm(&None, &Producers::empty(), &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 +1,4 @@
-use crate::{ComponentNames, ModuleNames, Producers, RegistryMetadata};
+use crate::{Author, ComponentNames, ModuleNames, Producers, RegistryMetadata};
 use anyhow::Result;
 use std::borrow::Cow;
 use std::mem;
@@ -9,6 +9,7 @@ use wasmparser::{KnownCustom, Parser, Payload::*};
 pub(crate) fn rewrite_wasm(
     add_name: &Option<String>,
     add_producers: &Producers,
+    add_author: &Option<Author>,
     add_registry_metadata: Option<&RegistryMetadata>,
     input: &[u8],
 ) -> Result<Vec<u8>> {
@@ -90,6 +91,13 @@ pub(crate) fn rewrite_wasm(
                             continue;
                         }
                     }
+                    KnownCustom::Unknown if c.name() == "author" => {
+                        if add_author.is_none() {
+                            let author = Author::parse_custom_section(c)?;
+                            author.append_to(&mut output);
+                            continue;
+                        }
+                    }
                     _ => {}
                 }
             }
@@ -119,6 +127,9 @@ pub(crate) fn rewrite_wasm(
         // Encode into output:
         producers.section().append_to(&mut output);
     }
+    if let Some(author) = add_author {
+        author.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

@@ -11,6 +11,7 @@ fn add_to_empty_component() {
         language: vec![("bar".to_owned(), "1.0".to_owned())],
         processed_by: vec![("baz".to_owned(), "1.0".to_owned())],
         sdk: vec![],
+        author: Some(Author::new("Chashu Cat")),
         registry_metadata: Some(RegistryMetadata {
             authors: Some(vec!["foo".to_owned()]),
             description: Some("foo bar baz".to_owned()),
@@ -42,6 +43,7 @@ fn add_to_empty_component() {
             name,
             producers,
             registry_metadata,
+            author,
             children,
             range,
         } => {
@@ -57,6 +59,8 @@ fn add_to_empty_component() {
                 "1.0"
             );
 
+            assert_eq!(author.unwrap(), Author::new("Chashu Cat"));
+
             let registry_metadata = registry_metadata.unwrap();
 
             assert!(registry_metadata.validate().is_ok());
@@ -99,7 +103,7 @@ fn add_to_empty_component() {
             );
 
             assert_eq!(range.start, 0);
-            assert_eq!(range.end, 435);
+            assert_eq!(range.end, 454);
         }
         _ => panic!("metadata should be component"),
     }
@@ -114,6 +118,7 @@ fn add_to_nested_component() {
         language: vec![("bar".to_owned(), "1.0".to_owned())],
         processed_by: vec![("baz".to_owned(), "1.0".to_owned())],
         sdk: vec![],
+        author: Some(Author::new("Chashu Cat")),
         registry_metadata: Some(RegistryMetadata {
             authors: Some(vec!["Foo".to_owned()]),
             ..Default::default()
@@ -159,6 +164,7 @@ fn add_to_nested_component() {
                 Metadata::Module {
                     name,
                     producers,
+                    author,
                     registry_metadata,
                     range,
                 } => {
@@ -173,14 +179,16 @@ fn add_to_nested_component() {
                         "1.0"
                     );
 
+                    assert_eq!(author, &Some(Author::new("Chashu Cat")));
+
                     let registry_metadata = registry_metadata.as_ref().unwrap();
                     assert_eq!(
                         registry_metadata.authors.as_ref().unwrap(),
                         &["Foo".to_owned()]
                     );
 
-                    assert_eq!(range.start, 10);
-                    assert_eq!(range.end, 123);
+                    assert_eq!(range.start, 11);
+                    assert_eq!(range.end, 143);
                 }
                 _ => panic!("child is a module"),
             }

crates/wasm-metadata/tests/module.rs

@@ -11,6 +11,7 @@ fn add_to_empty_module() {
         language: vec![("bar".to_owned(), "1.0".to_owned())],
         processed_by: vec![("baz".to_owned(), "1.0".to_owned())],
         sdk: vec![],
+        author: Some(Author::new("Chashu Cat")),
         registry_metadata: Some(RegistryMetadata {
             authors: Some(vec!["foo".to_owned()]),
             description: Some("foo bar baz".to_owned()),
@@ -41,6 +42,7 @@ fn add_to_empty_module() {
         Metadata::Module {
             name,
             producers,
+            author,
             registry_metadata,
             range,
         } => {
@@ -55,6 +57,8 @@ fn add_to_empty_module() {
                 "1.0"
             );
 
+            assert_eq!(author.unwrap(), Author::new("Chashu Cat"));
+
             let registry_metadata = registry_metadata.unwrap();
 
             assert!(registry_metadata.validate().is_ok());
@@ -97,7 +101,7 @@ fn add_to_empty_module() {
             );
 
             assert_eq!(range.start, 0);
-            assert_eq!(range.end, 425);
+            assert_eq!(range.end, 444);
         }
         _ => panic!("metadata should be module"),
     }