teamy-phon-codegen 0.2.0-rc.9

//! Deterministic Java 17 schema-table generation.
//!
//! Vox's Java target owns ergonomic DTOs and adapters. This lower-level target
//! emits the canonical Phon bytes and root ids those adapters are derived from.

use std::fmt::Write;

use phon_schema::schema_bundle_to_bytes;

use crate::Module;

#[must_use]
pub fn render_schema_table(module: &Module, package: &str, class_name: &str) -> String {
    let mut out = format!(
        "// @generated by phon-codegen; DO NOT EDIT.\npackage {package};\n\n\
         public final class {class_name} {{\n"
    );
    for root in &module.roots {
        let _ = writeln!(
            out,
            "  public static final long {}_SCHEMA_ID = 0x{:016x}L;",
            java_constant(&root.name),
            root.id.as_u64()
        );
    }
    let bytes = schema_bundle_to_bytes(&module.schemas).expect("validated codegen schema bundle");
    out.push_str("  public static final byte[] SCHEMA_BUNDLE = schemaBundle();\n");
    out.push_str(&render_byte_factory("schemaBundle", &bytes));
    let _ = writeln!(out, "  private {class_name}() {{}}");
    out.push_str("}\n");
    out
}

/// Emit bounded UTF-8 constants instead of one bytecode instruction per byte.
/// 16 KiB chunks stay below the JVM's 65,535-byte constant and method limits,
/// including for bundles at the runtime's 64 MiB admission limit.
#[must_use]
pub fn render_byte_factory(name: &str, bytes: &[u8]) -> String {
    let mut out = format!(
        "  private static byte[] {name}() {{\n    var out = new java.io.ByteArrayOutputStream({});\n",
        bytes.len()
    );
    for chunk in bytes.chunks(16 * 1024) {
        out.push_str("    out.writeBytes(java.util.HexFormat.of().parseHex(\"");
        for byte in chunk {
            let _ = write!(out, "{byte:02x}");
        }
        out.push_str("\"));\n");
    }
    out.push_str("    return out.toByteArray();\n  }\n");
    out
}

fn java_constant(name: &str) -> String {
    name.chars()
        .flat_map(|character| {
            if character.is_ascii_uppercase() {
                vec!['_', character]
            } else if character.is_ascii_alphanumeric() {
                vec![character.to_ascii_uppercase()]
            } else {
                vec!['_']
            }
        })
        .collect::<String>()
        .trim_start_matches('_')
        .to_string()
}

#[cfg(test)]
mod tests {
    use facet::Facet;

    use crate::Builder;

    use super::render_schema_table;

    #[derive(Facet)]
    struct Message {
        text: String,
    }

    #[test]
    fn large_byte_factory_compiles_and_preserves_every_byte() {
        let bytes: Vec<u8> = (0..262_147).map(|i| (i % 251) as u8).collect();
        let factory = super::render_byte_factory("bundle", &bytes);
        assert_eq!(factory, super::render_byte_factory("bundle", &bytes));
        let directory =
            std::env::temp_dir().join(format!("phon-java-bundle-{}", std::process::id()));
        std::fs::create_dir_all(&directory).unwrap();
        let source = format!(
            "public class LargeBundle {{ {factory} public static void main(String[] args) throws Exception {{ java.nio.file.Files.write(java.nio.file.Path.of(args[0]), bundle()); }} }}"
        );
        std::fs::write(directory.join("LargeBundle.java"), source).unwrap();
        let compile = std::process::Command::new("javac")
            .args(["--release", "17"])
            .arg(directory.join("LargeBundle.java"))
            .output()
            .expect("JDK required for Java generation qualification");
        assert!(
            compile.status.success(),
            "{}",
            String::from_utf8_lossy(&compile.stderr)
        );
        let output = directory.join("bundle.bin");
        assert!(
            std::process::Command::new("java")
                .arg("-cp")
                .arg(&directory)
                .arg("LargeBundle")
                .arg(&output)
                .status()
                .unwrap()
                .success()
        );
        assert_eq!(std::fs::read(output).unwrap(), bytes);
        std::fs::remove_dir_all(directory).unwrap();
    }

    #[test]
    fn schema_table_is_deterministic_and_contains_canonical_root() {
        let module = Builder::new()
            .add::<Message>()
            .build()
            .expect("derive schema");
        let first = render_schema_table(&module, "org.facet.phon.generated", "FixtureSchemas");
        let second = render_schema_table(&module, "org.facet.phon.generated", "FixtureSchemas");
        assert_eq!(first, second);
        assert!(first.contains("MESSAGE_SCHEMA_ID = 0x"));
        assert!(first.contains("HexFormat.of().parseHex"));
        assert!(first.contains("private FixtureSchemas()"));
    }
}