clusterflux-public/crates/clusterflux-cli/src/build.rs

use std::path::{Path, PathBuf};
use std::process::Command;

use anyhow::{bail, Context, Result};
use clusterflux_core::Digest;
use serde_json::{json, Value};
use wasmparser::{Parser, Payload};

use crate::errors::cli_error_summary_for_category;
use crate::{bundle_inspection, BuildArgs, BundleInspectArgs};

pub(crate) fn build_report(args: BuildArgs, cwd: PathBuf) -> Result<Value> {
    let inspection = bundle_inspection(
        BundleInspectArgs {
            project: args.project.clone(),
            source_provider: args.source_provider.clone(),
            disabled_source_providers: args.disabled_source_providers.clone(),
            json: true,
        },
        cwd,
    )?;
    let diagnostics = inspection.pre_schedule_diagnostics.clone();
    let blocking_diagnostic = diagnostics
        .iter()
        .find(|diagnostic| diagnostic.severity == "error")
        .cloned();
    if let Some(diagnostic) = blocking_diagnostic {
        let machine_category = match diagnostic.category.as_str() {
            "environment" => "environment",
            "source_provider" | "capability" => "capability",
            _ => "unknown",
        };
        return Ok(json!({
            "command": "build",
            "status": "blocked_before_schedule",
            "bundle": inspection,
            "diagnostics": diagnostics,
            "contains_full_repository_upload": false,
            "content_addressed": false,
            "debug_metadata_available": false,
            "scheduled_work": false,
            "machine_error": cli_error_summary_for_category(machine_category, &diagnostic.message),
        }));
    }

    let mut wasm = compile_project_wasm(&inspection.project)?;
    let environment_manifest = serde_json::to_vec(&inspection.metadata.environments)?;
    append_custom_section(
        &mut wasm.bytes,
        "clusterflux.environments",
        &environment_manifest,
    );
    let bundle_digest = Digest::sha256(&wasm.bytes);
    let task_descriptors = descriptor_records(&wasm.bytes, "clusterflux.tasks")?;
    let entrypoint_descriptors = descriptor_records(&wasm.bytes, "clusterflux.entrypoints")?;
    if task_descriptors.is_empty() {
        bail!(
            "compiled Wasm module contains no #[clusterflux::task] descriptors; annotate at least one exported task"
        );
    }
    if entrypoint_descriptors.is_empty() {
        bail!(
            "compiled Wasm module contains no #[clusterflux::main] descriptors; annotate at least one entrypoint"
        );
    }
    let output = args.output.unwrap_or_else(|| {
        inspection.project.join(".clusterflux/build").join(
            bundle_digest
                .as_str()
                .trim_start_matches("sha256:")
                .get(..16)
                .unwrap_or("bundle"),
        )
    });
    let bundle_artifact = write_bundle(
        &output,
        &wasm,
        &bundle_digest,
        &inspection,
        &task_descriptors,
        &entrypoint_descriptors,
    )?;

    Ok(json!({
        "command": "build",
        "status": "built",
        "bundle": inspection,
        "bundle_artifact": bundle_artifact,
        "diagnostics": diagnostics,
        "contains_full_repository_upload": false,
        "content_addressed": true,
        "debug_metadata_available": true,
        "scheduled_work": false,
    }))
}

fn append_custom_section(module: &mut Vec<u8>, name: &str, data: &[u8]) {
    let mut section = Vec::new();
    encode_unsigned_leb(name.len() as u64, &mut section);
    section.extend_from_slice(name.as_bytes());
    section.extend_from_slice(data);
    module.push(0);
    encode_unsigned_leb(section.len() as u64, module);
    module.extend_from_slice(&section);
}

fn encode_unsigned_leb(mut value: u64, output: &mut Vec<u8>) {
    loop {
        let mut byte = (value & 0x7f) as u8;
        value >>= 7;
        if value != 0 {
            byte |= 0x80;
        }
        output.push(byte);
        if value == 0 {
            break;
        }
    }
}

struct CompiledWasm {
    bytes: Vec<u8>,
    package: String,
    target: String,
    source_path: PathBuf,
}

fn compile_project_wasm(project: &Path) -> Result<CompiledWasm> {
    let manifest = project.join("Cargo.toml");
    let metadata_output = Command::new("cargo")
        .args([
            "metadata",
            "--format-version",
            "1",
            "--no-deps",
            "--manifest-path",
        ])
        .arg(&manifest)
        .output()
        .with_context(|| format!("failed to run cargo metadata for {}", manifest.display()))?;
    if !metadata_output.status.success() {
        bail!(
            "cargo metadata failed for {}: {}",
            manifest.display(),
            String::from_utf8_lossy(&metadata_output.stderr).trim()
        );
    }
    let metadata: Value = serde_json::from_slice(&metadata_output.stdout)?;
    let canonical_manifest = std::fs::canonicalize(&manifest)?;
    let package = metadata["packages"]
        .as_array()
        .and_then(|packages| {
            packages.iter().find(|package| {
                package["manifest_path"]
                    .as_str()
                    .and_then(|path| std::fs::canonicalize(path).ok())
                    .as_ref()
                    == Some(&canonical_manifest)
            })
        })
        .context("cargo metadata did not return the requested project package")?;
    let package_name = package["name"]
        .as_str()
        .context("cargo package name missing")?;
    let target = package["targets"]
        .as_array()
        .and_then(|targets| {
            targets.iter().find(|target| {
                target["crate_types"]
                    .as_array()
                    .is_some_and(|types| types.iter().any(|kind| kind == "cdylib"))
            })
        })
        .context("Clusterflux project library must include crate-type = [\"cdylib\"]")?;
    let target_name = target["name"]
        .as_str()
        .context("cargo target name missing")?;
    let build = Command::new("cargo")
        // The MVP transports one Wasm bundle in a bounded control frame. These are
        // ordinary Cargo release-profile settings, applied only to the guest build,
        // that keep the product SDK/runtime inside that accepted boundary.
        .env("CARGO_PROFILE_RELEASE_OPT_LEVEL", "z")
        .env("CARGO_PROFILE_RELEASE_LTO", "thin")
        .env("CARGO_PROFILE_RELEASE_CODEGEN_UNITS", "1")
        .args([
            "build",
            "--quiet",
            "--release",
            "--target",
            "wasm32-unknown-unknown",
            "--lib",
            "--manifest-path",
        ])
        .arg(&manifest)
        .output()
        .with_context(|| format!("failed to compile {} to Wasm", manifest.display()))?;
    if !build.status.success() {
        bail!(
            "Wasm bundle compilation failed for {}: {}",
            manifest.display(),
            String::from_utf8_lossy(&build.stderr).trim()
        );
    }
    let target_directory = metadata["target_directory"]
        .as_str()
        .context("cargo target_directory missing")?;
    let source_path = Path::new(target_directory)
        .join("wasm32-unknown-unknown/release")
        .join(format!("{}.wasm", target_name.replace('-', "_")));
    let bytes = std::fs::read(&source_path)
        .with_context(|| format!("compiled Wasm module missing at {}", source_path.display()))?;
    Ok(CompiledWasm {
        bytes,
        package: package_name.to_owned(),
        target: target_name.to_owned(),
        source_path,
    })
}

fn descriptor_records(module: &[u8], section_name: &str) -> Result<Vec<Value>> {
    let mut records: Vec<Value> = Vec::new();
    for payload in Parser::new(0).parse_all(module) {
        let Payload::CustomSection(section) = payload? else {
            continue;
        };
        if section.name() != section_name {
            continue;
        }
        for record in section
            .data()
            .split(|byte| *byte == b'\n' || *byte == 0)
            .filter(|record| !record.is_empty())
        {
            records.push(serde_json::from_slice(record).with_context(|| {
                format!("invalid descriptor record in Wasm custom section {section_name}")
            })?);
        }
    }
    records.sort_by(|left, right| left["name"].as_str().cmp(&right["name"].as_str()));
    Ok(records)
}

fn write_bundle(
    output: &Path,
    wasm: &CompiledWasm,
    bundle_digest: &Digest,
    inspection: &crate::bundle::BundleInspection,
    tasks: &[Value],
    entrypoints: &[Value],
) -> Result<Value> {
    std::fs::create_dir_all(output)?;
    let module_path = output.join("module.wasm");
    let task_path = output.join("task-descriptors.json");
    let entrypoint_path = output.join("entrypoints.json");
    let environment_path = output.join("environments.json");
    let source_path = output.join("source-provider.json");
    let vfs_path = output.join("vfs-seed.json");
    let debug_path = output.join("debug-metadata.json");
    let manifest_path = output.join("manifest.json");
    std::fs::write(&module_path, &wasm.bytes)?;
    write_json(&task_path, &json!(tasks))?;
    write_json(&entrypoint_path, &json!(entrypoints))?;
    write_json(&environment_path, &json!(inspection.metadata.environments))?;
    write_json(&source_path, &json!(inspection.source_provider_manifest))?;
    write_json(
        &vfs_path,
        &json!({
            "epoch": 0,
            "mounts": ["/vfs/artifacts", "/vfs/sources", "/vfs/blobs"],
            "large_bytes_embedded": false,
        }),
    )?;
    write_json(&debug_path, &json!(inspection.metadata.debug_metadata))?;
    let manifest = json!({
        "kind": "clusterflux-bundle",
        "format_version": 1,
        "package": wasm.package,
        "target": wasm.target,
        "bundle_digest": bundle_digest,
        "module": "module.wasm",
        "module_size_bytes": wasm.bytes.len(),
        "task_descriptors": "task-descriptors.json",
        "entrypoints": "entrypoints.json",
        "environments": "environments.json",
        "source_provider": "source-provider.json",
        "vfs_seed": "vfs-seed.json",
        "debug_metadata": "debug-metadata.json",
        "required_capabilities": tasks.iter().flat_map(|task| {
            task["required_capabilities"].as_array().into_iter().flatten().cloned()
        }).collect::<Vec<_>>(),
        "metadata_identity": inspection.metadata.identity,
        "coordinator_receives_source_bytes_by_default": false,
        "embeds_full_repository": false,
    });
    write_json(&manifest_path, &manifest)?;
    Ok(json!({
        "directory": output,
        "manifest": manifest_path,
        "module": module_path,
        "compiled_module_source": wasm.source_path,
        "bundle_digest": bundle_digest,
        "module_size_bytes": wasm.bytes.len(),
        "task_descriptor_count": tasks.len(),
        "entrypoint_count": entrypoints.len(),
        "files": [
            "manifest.json",
            "module.wasm",
            "task-descriptors.json",
            "entrypoints.json",
            "environments.json",
            "source-provider.json",
            "vfs-seed.json",
            "debug-metadata.json",
        ],
    }))
}

fn write_json(path: &Path, value: &Value) -> Result<()> {
    let mut bytes = serde_json::to_vec_pretty(value)?;
    bytes.push(b'\n');
    std::fs::write(path, bytes).with_context(|| format!("failed to write {}", path.display()))
}