diff --git a/FULL_HELP_DOCS.md b/FULL_HELP_DOCS.md index 109e02ebae..05d09ee8e2 100644 --- a/FULL_HELP_DOCS.md +++ b/FULL_HELP_DOCS.md @@ -422,6 +422,19 @@ To view the commands that will be executed, without executing them, use the --pr - `--print-commands-only` — Print commands to build without executing them + Incompatible with `--verifiable`, which compiles from a throwaway extracted-archive tempdir that only exists for the build, so a printed command bind-mounting it could never be replayed. + +###### **Verifiable Options:** + +- `--verifiable` — Produce a SEP-58 verifiable (reproducible) build. + + Snapshots the working tree into a byte-reproducible source archive, builds it in a digest-pinned container image, and records provenance meta (bldimg, source_uri, source_sha256, bldopt) into the wasm so a third party can reproduce the exact bytes. Implies `--locked`. Requires a clean git tree. Requires `--image` pinned by digest (`/@sha256:<64-hex>`) so the recorded `bldimg` names the exact bytes. + + Incompatible with `--print-commands-only`: a verifiable build compiles from an extracted-archive tempdir that only exists for the build, so a printed command bind-mounting it could never be replayed. + +- `--source-sha256 ` — Pin the SEP-58 source_sha256 of the generated archive (64-char lower-case hex). The build fails if the archive hashes to a different value +- `--source-uri ` — Record a SEP-58 source_uri where the source archive can be fetched (a URI with a scheme, e.g. https://example.com/src.tar.gz) + ## `stellar contract archive` Generate the reproducible source archive used by verifiable builds diff --git a/cmd/crates/soroban-test/tests/it/build.rs b/cmd/crates/soroban-test/tests/it/build.rs index b5c002a8b0..060ed8cf02 100644 --- a/cmd/crates/soroban-test/tests/it/build.rs +++ b/cmd/crates/soroban-test/tests/it/build.rs @@ -1081,6 +1081,9 @@ fn build_always_injects_cli_version() { ); } +const ZERO_DIGEST: &str = + "docker.io/stellar/stellar-cli@sha256:0000000000000000000000000000000000000000000000000000000000000000"; + // Convenience: drive a git command in a fixture directory. fn git_in(dir: &Path, args: &[&str]) { std::process::Command::new("git") @@ -1104,6 +1107,146 @@ fn fresh_workspace() -> (TempDir, PathBuf) { (temp, workspace) } +// `--verifiable` cannot accept reserved `--meta` keys that the cli writes itself. +#[test] +fn verifiable_meta_conflict_errors() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .arg("--source-sha256") + .arg("a".repeat(64)) + .arg("--meta") + .arg("bldimg=not-allowed") + .assert() + .failure() + .stderr(predicate::str::contains("reserved key: bldimg")); +} + +// A verifiable build compiles from a throwaway extracted-archive tempdir, so a +// `--print-commands-only` command bind-mounting it could never be replayed; +// clap rejects the combination up front. +#[test] +fn verifiable_rejects_print_commands_only() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .arg("--print-commands-only") + .assert() + .failure() + .stderr(predicate::str::contains("cannot be used with")); +} + +// `--image` is validated against the SEP-58 bldimg regex; tag-only refs fail. +#[test] +fn verifiable_image_must_be_digest_pinned() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg("docker.io/stellar/stellar-cli:latest") + .arg("--source-sha256") + .arg("a".repeat(64)) + .assert() + .failure() + .stderr(predicate::str::contains("bldimg format")); +} + +// SEP-58 metadata must be ASCII; a non-ASCII `--image` is rejected before the +// bldimg format check. +#[test] +fn verifiable_image_must_be_ascii() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + let non_ascii = format!("localhost:5000/café@sha256:{}", "0".repeat(64)); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(non_ascii) + .arg("--source-sha256") + .arg("a".repeat(64)) + .assert() + .failure() + .stderr(predicate::str::contains("must be ASCII")); +} + +// SEP-58 bldimg requires an explicit registry host (e.g. `docker.io/...`). +// Implicit Docker-Hub-style short refs are rejected. +#[test] +fn verifiable_image_requires_explicit_registry_host() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + let short_ref = format!("stellar/stellar-cli@sha256:{}", "0".repeat(64)); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(short_ref) + .arg("--source-sha256") + .arg("a".repeat(64)) + .assert() + .failure() + .stderr(predicate::str::contains("bldimg format")); +} + +// `--verifiable` always generates the source archive (and computes +// source_sha256) before the docker stage, so the "Wrote source archive" line +// appears even though the build then fails to reach a real image. +#[test] +fn verifiable_always_writes_source_archive() { + let sandbox = TestEnv::default(); + let (_temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); + + sandbox + .new_assert_cmd("contract") + .current_dir(workspace.join("contracts").join("add")) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .assert() + .failure() + .stderr( + predicate::str::contains("Wrote source archive") + .and(predicate::str::contains("source_sha256")), + ); +} + // `contract archive --out-file` writes the gzipped tarball and prints its // source_sha256. #[test] @@ -1285,3 +1428,73 @@ fn contract_archive_dirty_tree_errors() { "no archive should be written for a dirty tree" ); } + +// `--source-sha256` value must match the 64-hex regex. +#[test] +fn verifiable_source_sha256_format_errors() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .arg("--source-sha256") + .arg("not-a-sha") + .assert() + .failure() + .stderr(predicate::str::contains("source_sha256 format")); +} + +// `--source-uri` value must be a URI with a scheme. +#[test] +fn verifiable_source_uri_format_errors() { + let sandbox = TestEnv::default(); + let cargo_dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")); + let fixture_path = cargo_dir.join("tests/fixtures/workspace/contracts/add"); + + sandbox + .new_assert_cmd("contract") + .current_dir(fixture_path) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .arg("--source-sha256") + .arg("a".repeat(64)) + .arg("--source-uri") + .arg("not a uri") + .assert() + .failure() + .stderr(predicate::str::contains("source_uri format")); +} + +// A dirty git tree is a hard fail under `--verifiable` (the recorded +// source_sha256 would not describe the bytes built). +#[test] +fn verifiable_dirty_tree_errors() { + let sandbox = TestEnv::default(); + let (_temp, workspace) = fresh_workspace(); + git_in(&workspace, &["init", "-q", "-b", "main"]); + git_in(&workspace, &["add", "-A"]); + git_in(&workspace, &["commit", "-q", "-m", "init"]); + // Dirty the tree after committing so status is non-empty. + std::fs::write(workspace.join("dirty.txt"), b"uncommitted").unwrap(); + + sandbox + .new_assert_cmd("contract") + .current_dir(workspace.join("contracts").join("add")) + .arg("build") + .arg("--verifiable") + .arg("--image") + .arg(ZERO_DIGEST) + .arg("--source-sha256") + .arg("a".repeat(64)) + .assert() + .failure() + .stderr(predicate::str::contains("dirty").or(predicate::str::contains("clean tree"))); +} diff --git a/cmd/soroban-cli/src/commands/container/shared.rs b/cmd/soroban-cli/src/commands/container/shared.rs index 3bbd073918..4b67b7764a 100644 --- a/cmd/soroban-cli/src/commands/container/shared.rs +++ b/cmd/soroban-cli/src/commands/container/shared.rs @@ -1,6 +1,8 @@ use core::fmt; +use std::process::Stdio; use clap::ValueEnum; +use tokio::io::{AsyncBufReadExt, AsyncReadExt, BufReader}; use tokio::process::Command; use crate::print::Print; @@ -20,6 +22,9 @@ pub enum Error { program: String, source: std::io::Error, }, + + #[error("could not pull image {image}: {stderr}")] + PullImageFailed { image: String, stderr: String }, } /// Container runtime to shell out to. @@ -261,6 +266,59 @@ impl Args { }; cmd } + + /// Pull `image`, streaming the engine's high-level status lines ("Pulling + /// from", "Digest", "Status") through `print`. Per-layer progress written to + /// stderr is captured rather than shown and surfaced only when the pull + /// fails, as `PullImageFailed` — callers that need to explain a failed pull + /// (e.g. the verifiable build's tag-listing hint) rely on that captured text. + /// A missing engine binary surfaces via `io_error` as `NotFound`. + pub(crate) async fn pull_image(&self, image: &str, print: &Print) -> Result<(), Error> { + let mut child = self + .pull_command(image) + .stdout(Stdio::piped()) + .stderr(Stdio::piped()) + .spawn() + .map_err(|e| self.io_error(e))?; + + let stdout = child.stdout.take(); + let stderr = child.stderr.take(); + + let stream_stdout = async { + if let Some(stdout) = stdout { + let mut lines = BufReader::new(stdout).lines(); + while let Ok(Some(line)) = lines.next_line().await { + if line.contains("Pulling from") + || line.contains("Digest") + || line.contains("Status") + { + print.infoln(line); + } + } + } + }; + + let capture_stderr = async { + let mut buf = String::new(); + if let Some(mut stderr) = stderr { + let _ = stderr.read_to_string(&mut buf).await; + } + buf + }; + + // Drain both pipes concurrently so a full stderr buffer can't deadlock + // the child while we're reading stdout. + let ((), stderr) = tokio::join!(stream_stdout, capture_stderr); + + if child.wait().await.map_err(|e| self.io_error(e))?.success() { + Ok(()) + } else { + Err(Error::PullImageFailed { + image: image.to_string(), + stderr: stderr.trim().to_string(), + }) + } + } } /// Resource limits for commands that *run* a container (e.g. `container start`). @@ -498,7 +556,7 @@ mod test { let not_found = std::io::Error::from(std::io::ErrorKind::NotFound); match args(None, Some(Engine::AppleContainer)).io_error(not_found) { Error::NotFound { program, .. } => assert_eq!(program, "container"), - Error::Command { .. } => panic!("expected NotFound, got Command"), + other => panic!("expected NotFound, got {other:?}"), } } diff --git a/cmd/soroban-cli/src/commands/contract/build.rs b/cmd/soroban-cli/src/commands/contract/build.rs index 23b45af6e4..6d222aeec3 100644 --- a/cmd/soroban-cli/src/commands/contract/build.rs +++ b/cmd/soroban-cli/src/commands/contract/build.rs @@ -23,7 +23,7 @@ use crate::utils::XDR_DEPTH_LIMIT; use crate::{ commands::{ container::shared::{Args as ContainerArgs, RunArgs as ContainerRunArgs}, - global, version, HEADING_CONTAINER, + global, version, HEADING_CONTAINER, HEADING_VERIFIABLE, }, print::Print, wasm, @@ -31,6 +31,7 @@ use crate::{ pub mod container; pub(crate) mod source_archive; +pub mod verifiable; /// A built WASM artifact with its package name and file path. #[derive(Debug, Clone)] @@ -100,7 +101,11 @@ pub struct Cmd { pub locked: bool, /// Print commands to build without executing them - #[arg(long, conflicts_with = "out_dir", help_heading = "Other")] + /// + /// Incompatible with `--verifiable`, which compiles from a throwaway + /// extracted-archive tempdir that only exists for the build, so a printed + /// command bind-mounting it could never be replayed. + #[arg(long, conflicts_with_all = ["out_dir", "verifiable"], help_heading = "Other")] pub print_commands_only: bool, /// Build inside this container image (e.g. @@ -122,6 +127,37 @@ pub struct Cmd { #[arg(long, requires = "image", help_heading = HEADING_CONTAINER)] pub pull: bool, + /// Produce a SEP-58 verifiable (reproducible) build. + /// + /// Snapshots the working tree into a byte-reproducible source archive, + /// builds it in a digest-pinned container image, and records provenance meta + /// (bldimg, source_uri, source_sha256, bldopt) into the wasm so a third + /// party can reproduce the exact bytes. Implies `--locked`. Requires a clean + /// git tree. Requires `--image` pinned by digest + /// (`/@sha256:<64-hex>`) so the recorded `bldimg` names + /// the exact bytes. + /// + /// Incompatible with `--print-commands-only`: a verifiable build compiles + /// from an extracted-archive tempdir that only exists for the build, so a + /// printed command bind-mounting it could never be replayed. + #[arg(long, requires = "image", help_heading = HEADING_VERIFIABLE)] + pub verifiable: bool, + + /// Pin the SEP-58 source_sha256 of the generated archive (64-char lower-case + /// hex). The build fails if the archive hashes to a different value. + #[arg(long, requires = "verifiable", help_heading = HEADING_VERIFIABLE)] + pub source_sha256: Option, + + /// Record a SEP-58 source_uri where the source archive can be fetched (a URI + /// with a scheme, e.g. https://example.com/src.tar.gz). + #[arg( + long, + requires = "verifiable", + requires = "source_sha256", + help_heading = HEADING_VERIFIABLE + )] + pub source_uri: Option, + #[command(flatten)] pub build_args: BuildArgs, @@ -245,6 +281,9 @@ pub enum Error { #[error(transparent)] Container(#[from] container::Error), + + #[error(transparent)] + Verifiable(#[from] verifiable::Error), } pub(crate) const WASM_TARGET: &str = "wasm32v1-none"; @@ -265,6 +304,9 @@ impl Default for Cmd { print_commands_only: false, image: None, pull: false, + verifiable: false, + source_sha256: None, + source_uri: None, build_args: BuildArgs::default(), container_args: ContainerArgs::default(), run_args: ContainerRunArgs::default(), @@ -278,7 +320,14 @@ impl Cmd { pub async fn run(&self, global_args: &global::Args) -> Result, Error> { let print = Print::new(global_args.quiet); - // When an image is given, build inside that container instead of locally. + // A verifiable build archives the source and builds it in a + // digest-pinned container, recording SEP-58 provenance meta. + if self.verifiable { + return verifiable::run(self, global_args, &print).await; + } + + // When an image is given (without --verifiable), build inside that + // container instead of locally. if self.image.is_some() { return container::run(self, global_args, &print).await; } diff --git a/cmd/soroban-cli/src/commands/contract/build/container.rs b/cmd/soroban-cli/src/commands/contract/build/container.rs index c6299e9985..30ec3b1d1c 100644 --- a/cmd/soroban-cli/src/commands/contract/build/container.rs +++ b/cmd/soroban-cli/src/commands/contract/build/container.rs @@ -27,18 +27,18 @@ use super::{get_wasm_target, BuiltContract, Cmd, WASM_TARGET, WASM_TARGET_OLD}; /// First CLI release whose `contract build` accepts `--locked` (added in cli /// v25.2.0). Older images reject it, so it's dropped (with a warning) on anything /// older, matching the version detected from the image's own `version` output. -const LOCKED_MIN: &str = "25.2.0"; +pub(super) const LOCKED_MIN: &str = "25.2.0"; /// First CLI release whose `contract build` has the `--optimize` flag at all. /// Older images reject it, so — since optimization is on by default — this is the /// effective minimum supported image. We probe the image's `version` and skip the /// flag (with a warning) on anything older. -const OPTIMIZE_FLAG_MIN: &str = "23.2.0"; +pub(super) const OPTIMIZE_FLAG_MIN: &str = "23.2.0"; /// First CLI release whose `contract build` accepts `--optimize=false` as an /// explicit value. Images between [`OPTIMIZE_FLAG_MIN`] and this default to *not* /// optimizing, so for them we forward nothing to get an unoptimized build. -const OPTIMIZE_NEW_SYNTAX_MIN: &str = "26.1.0"; +pub(super) const OPTIMIZE_NEW_SYNTAX_MIN: &str = "26.1.0"; #[derive(thiserror::Error, Debug)] pub enum Error { @@ -156,14 +156,18 @@ pub async fn run( let container_cmds: Vec> = targets .iter() .map(|target| { + // Plain container builds forward the user's `--locked` (when the + // image accepts it) and don't record bldopts. forwarded_build_args( cmd, &workspace_root, *target, - supports_locked, + cmd.locked && supports_locked, supports_optimize_flag, supports_optimize_false, + false, ) + .0 }) .collect(); @@ -201,6 +205,7 @@ pub async fn run( &docker, &cmd.run_args, &bin, + "stellar-contract-build", print, print_only, ) @@ -211,10 +216,10 @@ pub async fn run( return Ok(Vec::new()); } - collect_built_contracts(cmd, &md, &workspace_root) + collect_built_contracts(cmd, &md, &workspace_root, None) } -fn metadata(cmd: &Cmd) -> Result { +pub(super) fn metadata(cmd: &Cmd) -> Result { let mut mc = MetadataCommand::new(); mc.no_deps(); if let Some(p) = &cmd.manifest_path { @@ -227,7 +232,7 @@ fn metadata(cmd: &Cmd) -> Result Vec { +pub(super) fn resolve_packages(cmd: &Cmd, md: &cargo_metadata::Metadata) -> Vec { if let Some(pkg) = &cmd.package { return vec![pkg.clone()]; } @@ -248,13 +253,16 @@ fn resolve_packages(cmd: &Cmd, md: &cargo_metadata::Metadata) -> Vec { } /// The `contract build …` argv forwarded to the container, mirroring the local -/// build's flags. `--manifest-path` is relativized against the workspace root so -/// it's valid inside `/source`. `--out-dir` is deliberately omitted — artifacts -/// are collected on the host from the mounted `target/`. +/// build's flags, plus (when `record_bldopts`) the shell-escaped `bldopt` +/// strings recorded into SEP-58 metadata by verifiable builds. `--manifest-path` +/// is relativized against the workspace/source root so it's valid inside +/// `/source`. `--out-dir` is deliberately omitted — artifacts are collected on +/// the host from the mounted `target/`. /// -/// `supports_locked`: whether the container's `contract build` accepts `--locked` -/// (added in cli 25.2.0). When false, the user's `--locked` is dropped rather -/// than forwarded to an image that would reject it. +/// `include_locked`: whether to add `--locked`. A plain container build passes +/// `cmd.locked && supports_locked` (the user's flag, when the image accepts it); +/// a verifiable build implies it (`supports_locked`). Either way it's dropped on +/// images too old to accept the flag (added in cli 25.2.0). /// /// `supports_optimize_flag`: whether the container's cli has the `--optimize` /// flag at all (added in cli 23.2.0). When false, nothing about optimize is @@ -264,18 +272,47 @@ fn resolve_packages(cmd: &Cmd, md: &cargo_metadata::Metadata) -> Vec { /// `--optimize=false` (added in cli 26.1.0). When false and the user disabled /// optimization, nothing is forwarded — the older cli defaults to not /// optimizing, and passing `--optimize=false` there would fail. -fn forwarded_build_args( +/// +/// `record_bldopts`: when true, every forwarded build-affecting flag is also +/// captured as a `bldopt` (its value shell-escaped once, at the source, so each +/// recorded option is valid shell on its own) for the verifiable build's SEP-58 +/// metadata. A plain container build passes false and ignores the second tuple +/// element. +#[allow(clippy::too_many_arguments, clippy::fn_params_excessive_bools)] +pub(super) fn forwarded_build_args( cmd: &Cmd, workspace_root: &Path, package: Option<&str>, - supports_locked: bool, + include_locked: bool, supports_optimize_flag: bool, supports_optimize_false: bool, -) -> Vec { + record_bldopts: bool, +) -> (Vec, Vec) { let mut args = vec!["contract".to_string(), "build".to_string()]; + let mut bldopts: Vec = Vec::new(); + + // Record a build option. `None` means a bare flag (`--locked`); `Some(v)` + // means `--flag=v`. The forwarded copy keeps the value raw (the container + // gets it as argv, and `compose_shell_command` re-escapes it for the + // multi-package `sh -c`); the bldopt copy shell-escapes only the value side, + // once, so every recorded option is valid shell on its own — e.g. + // `--meta=note='added on build'`, never `'--meta=note=added on build'`. + let mut record = |key: &str, value: Option<&str>| { + if let Some(v) = value { + args.push(format!("{key}={v}")); + if record_bldopts { + bldopts.push(format!("{key}={}", shell_escape::escape(v.into()))); + } + } else { + args.push(key.to_string()); + if record_bldopts { + bldopts.push(key.to_string()); + } + } + }; - if cmd.locked && supports_locked { - args.push("--locked".to_string()); + if include_locked { + record("--locked", None); } if let Some(path) = &cmd.manifest_path { let abs = std::path::absolute(path).unwrap_or_else(|_| path.clone()); @@ -283,25 +320,25 @@ fn forwarded_build_args( .strip_prefix(workspace_root) .map(Path::to_path_buf) .unwrap_or(abs); - args.push(format!("--manifest-path={}", rel.to_slash_lossy())); + record("--manifest-path", Some(rel.to_slash_lossy().as_ref())); } if cmd.profile != "release" { - args.push(format!("--profile={}", cmd.profile)); + record("--profile", Some(cmd.profile.as_str())); } if let Some(features) = &cmd.features { - args.push(format!("--features={features}")); + record("--features", Some(features.as_str())); } if cmd.all_features { - args.push("--all-features".to_string()); + record("--all-features", None); } if cmd.no_default_features { - args.push("--no-default-features".to_string()); + record("--no-default-features", None); } if let Some(pkg) = package { - args.push(format!("--package={pkg}")); + record("--package", Some(pkg)); } for (k, v) in &cmd.build_args.meta { - args.push(format!("--meta={k}={v}")); + record(&format!("--meta={k}"), Some(v.as_str())); } // Optimization is forwarded per the image's cli version. To enable it, bare // `--optimize` on images >= v23.2.0 (older images lack the flag entirely, so @@ -309,13 +346,13 @@ fn forwarded_build_args( // older ones default to not optimizing, so forwarding nothing matches. if cmd.build_args.optimize { if supports_optimize_flag { - args.push("--optimize".to_string()); + record("--optimize", None); } } else if supports_optimize_false { - args.push("--optimize=false".to_string()); + record("--optimize", Some("false")); } - args + (args, bldopts) } async fn pull_image(docker: &shared::Args, image: &str, print: &Print) -> Result<(), Error> { @@ -364,18 +401,18 @@ async fn run_probe( /// Facts probed from the image before building, gathered in one throwaway /// container to avoid a round-trip per fact. -struct ImageProbe { +pub(super) struct ImageProbe { /// CLI binary on the image's PATH — `stellar` (v21.0.0+) or `soroban` /// (older). Used when invoking the CLI by name in the chained multi-build /// command; the single-build path uses the image's entrypoint instead. - bin: String, + pub(super) bin: String, /// Parsed CLI version, or `None` when the image reported no parseable version /// (treated as a current image by the caller). - version: Option, + pub(super) version: Option, /// The image's default rustup toolchain (e.g. /// `1.97.1-aarch64-unknown-linux-gnu`), pinned into `RUSTUP_TOOLCHAIN`. /// Guaranteed non-empty — the probe hard-fails when it can't be determined. - toolchain: String, + pub(super) toolchain: String, } /// Probe the image once for everything the build needs: the CLI binary name, its @@ -384,7 +421,7 @@ struct ImageProbe { /// already require) that detects the binary, then reports each fact on its own /// tagged line so the combined stdout can be split apart. Hard-fails when no /// default toolchain can be determined, rather than building unpinned. -async fn probe_image(image: &str, docker: &shared::Args) -> Result { +pub(super) async fn probe_image(image: &str, docker: &shared::Args) -> Result { // Detect the binary first, then run `$bin version` (version on its first // line) and `rustup default` (the toolchain name). Tag each line so we can // pick the values back out regardless of any extra output. @@ -446,7 +483,7 @@ fn parse_default_toolchain(stdout: &str) -> Option { } #[allow(clippy::too_many_arguments)] -async fn run_in_container( +pub(super) async fn run_in_container( image: &str, workspace_root: &Path, container_cmds: &[Vec], @@ -454,6 +491,7 @@ async fn run_in_container( docker: &shared::Args, run_args: &shared::RunArgs, bin: &str, + container_name_prefix: &str, print: &Print, print_only: bool, ) -> Result<(), Error> { @@ -513,7 +551,7 @@ async fn run_in_container( // per invocation so concurrent builds don't collide, and kept out of the // reproduce line where a fixed name would clash on re-run. let container_name = format!( - "stellar-contract-build-{}-{:08x}", + "{container_name_prefix}-{}-{:08x}", std::process::id(), rand::random::() ); @@ -698,17 +736,39 @@ fn newest_existing_artifact(candidates: &[PathBuf]) -> Option { .cloned() } -/// Collect the built wasm from the mounted `target/`. Because the working tree -/// was bind-mounted, the container writes artifacts straight to the host under -/// `/target///`. The container's rust toolchain -/// decides the target triple, so both known triples are probed. Copies to -/// `--out-dir` when set. -fn collect_built_contracts( +/// Collect the built wasm artifacts. Package names and the host target dir come +/// from host `cargo metadata`. +/// +/// `extracted_root` is `None` for a plain container build: the working tree was +/// bind-mounted, so the container wrote artifacts straight to the host target +/// dir and they're read (and optionally copied to `--out-dir`) in place. It's +/// `Some(er)` for a verifiable build, where the container built from an +/// extracted-archive tempdir; the artifacts then live under that tree's target +/// dir and must be copied back to the host target dir (or `--out-dir`) before +/// the tempdir drops. `source_root` is the host source root the extracted tree +/// mirrors, so the target dir's position relative to it carries over. +/// +/// The container's rust toolchain decides the target triple, so both known +/// triples are probed and the *freshest* artifact wins (an earlier build into +/// the other triple can leave a stale wasm behind). +pub(super) fn collect_built_contracts( cmd: &Cmd, md: &cargo_metadata::Metadata, - workspace_root: &Path, + source_root: &Path, + extracted_root: Option<&Path>, ) -> Result, super::Error> { - let target_root = workspace_root.join("target"); + // Where the user's artifacts ultimately belong (and where a verifiable build's + // wasm is copied back to): the workspace's real target dir, which may be moved + // by a `.cargo/config.toml` `target-dir` or host `CARGO_TARGET_DIR`. + let host_target = md.target_directory.as_std_path(); + + // Where the build ACTUALLY wrote artifacts. The container is always forced to + // `CARGO_TARGET_DIR=/source/target` (independent of the host's target-dir + // config), which on the host is `/target` — the extracted + // archive root for a verifiable build, else the source (workspace) root. Basing + // this on `host_target` instead would miss the artifacts whenever a custom + // target-dir moves it away from `/target`. + let src_target = extracted_root.unwrap_or(source_root).join("target"); let mut out = Vec::new(); for p in &md.packages { @@ -728,30 +788,48 @@ fn collect_built_contracts( } let file = format!("{}.wasm", p.name.replace('-', "_")); - // The container may build for either wasm target depending on its rust - // version, so probe both triple dirs. Pick the *freshest* rather than the - // first that exists: an earlier build into the other triple can leave a - // stale wasm behind, and selecting by existence alone would return it. - // Fall back to the current host default for the reported path when the - // build produced nothing. - let candidates: Vec = [WASM_TARGET, WASM_TARGET_OLD] + // Probe both triple dirs (the container's rust version decides which), + // picking the freshest by mtime rather than the first that exists. The + // chosen triple's relative path is reused to mirror the layout when + // copying a verifiable build's artifact back to the host target dir. + let rel_candidates: Vec = [WASM_TARGET, WASM_TARGET_OLD] .iter() - .map(|triple| target_root.join(triple).join(&cmd.profile).join(&file)) + .map(|triple| Path::new(triple).join(&cmd.profile).join(&file)) .collect(); - let src = newest_existing_artifact(&candidates).unwrap_or_else(|| { - let triple = get_wasm_target().unwrap_or_else(|_| WASM_TARGET.to_string()); - target_root.join(triple).join(&cmd.profile).join(&file) - }); + let abs_candidates: Vec = rel_candidates + .iter() + .map(|rel| src_target.join(rel)) + .collect(); + let chosen_rel = newest_existing_artifact(&abs_candidates) + .and_then(|src| src.strip_prefix(&src_target).ok().map(Path::to_path_buf)) + .unwrap_or_else(|| { + let triple = get_wasm_target().unwrap_or_else(|_| WASM_TARGET.to_string()); + Path::new(&triple).join(&cmd.profile).join(&file) + }); + let src = src_target.join(&chosen_rel); + + // Destination: `--out-dir` wins; else if the build ran in an extracted + // tempdir, copy into the host target dir so the artifact survives the + // tempdir drop; else leave it in place (already on the host). + let dest = if let Some(out_dir) = &cmd.out_dir { + Some(out_dir.join(&file)) + } else if extracted_root.is_some() { + Some(host_target.join(&chosen_rel)) + } else { + None + }; - let path = if let Some(out_dir) = &cmd.out_dir { - std::fs::create_dir_all(out_dir).map_err(super::Error::CreatingOutDir)?; - let dest = out_dir.join(&file); - if src.exists() { + let path = match dest { + Some(dest) if src.exists() => { + if let Some(parent) = dest.parent() { + std::fs::create_dir_all(parent).map_err(super::Error::CreatingOutDir)?; + } std::fs::copy(&src, &dest).map_err(super::Error::CopyingWasmFile)?; + dest } - dest - } else { - src + // Source missing: report the intended dest (matches prior leniency). + Some(dest) => dest, + None => src, }; out.push(BuiltContract { @@ -778,12 +856,14 @@ mod tests { #[test] fn forwarded_build_args_defaults() { let cmd = Cmd::default(); - let args = forwarded_build_args(&cmd, &ws(), None, true, true, true); + let (args, bldopts) = forwarded_build_args(&cmd, &ws(), None, false, true, true, false); assert_eq!(args[..2], ["contract".to_string(), "build".to_string()]); // Default optimize=true → bare `--optimize`; no `--locked` unless asked. assert!(args.contains(&"--optimize".to_string())); assert!(!args.iter().any(|a| a == "--locked")); assert!(!args.iter().any(|a| a.starts_with("--package"))); + // Plain container builds don't record bldopts. + assert!(bldopts.is_empty()); } #[test] @@ -792,19 +872,21 @@ mod tests { locked: true, ..Cmd::default() }; - let args = forwarded_build_args(&cmd, &ws(), Some("contract-a"), true, true, true); + let (args, _) = + forwarded_build_args(&cmd, &ws(), Some("contract-a"), true, true, true, false); assert!(args.contains(&"--locked".to_string())); assert!(args.contains(&"--package=contract-a".to_string())); } #[test] fn forwarded_build_args_drops_locked_when_unsupported() { - // User asked for --locked but the image's cli doesn't accept it. + // User asked for --locked but the image's cli doesn't accept it, so the + // caller passes include_locked=false (cmd.locked && supports_locked). let cmd = Cmd { locked: true, ..Cmd::default() }; - let args = forwarded_build_args(&cmd, &ws(), None, false, true, true); + let (args, _) = forwarded_build_args(&cmd, &ws(), None, false, true, true, false); assert!(!args.iter().any(|a| a == "--locked")); } @@ -814,7 +896,7 @@ mod tests { // though optimize defaults to true. let cmd = Cmd::default(); assert!(cmd.build_args.optimize); - let args = forwarded_build_args(&cmd, &ws(), None, true, false, false); + let (args, _) = forwarded_build_args(&cmd, &ws(), None, false, false, false, false); assert!(!args.iter().any(|a| a.starts_with("--optimize"))); } @@ -834,7 +916,7 @@ mod tests { }, ..Cmd::default() }; - let args = forwarded_build_args(&cmd, &ws(), None, true, true, true); + let (args, _) = forwarded_build_args(&cmd, &ws(), None, false, true, true, false); assert!(args.contains(&"--profile=dev".to_string())); assert!(args.contains(&"--features=a,b".to_string())); assert!(args.contains(&"--all-features".to_string())); @@ -844,6 +926,32 @@ mod tests { assert!(args.contains(&"--optimize=false".to_string())); } + #[test] + fn forwarded_build_args_records_bldopts_when_requested() { + // Verifiable builds pass record_bldopts=true and include_locked=true, + // capturing each forwarded flag as a shell-escaped bldopt. + let cmd = Cmd { + features: Some("a,b".to_string()), + build_args: BuildArgs { + meta: vec![("note".to_string(), "added on build".to_string())], + optimize: true, + }, + ..Cmd::default() + }; + let (forwarded, bldopts) = + forwarded_build_args(&cmd, &ws(), Some("contract-a"), true, true, true, true); + assert!(forwarded.contains(&"--locked".to_string())); + assert!(forwarded.contains(&"--meta=note=added on build".to_string())); + assert!(bldopts.contains(&"--locked".to_string())); + assert!(bldopts.contains(&"--features=a,b".to_string())); + assert!(bldopts.contains(&"--package=contract-a".to_string())); + // Only the value side is shell-escaped, and each bldopt is one token. + assert!(bldopts.contains(&"--meta=note='added on build'".to_string())); + for o in &bldopts { + assert_eq!(shlex::split(o).expect("valid shell").len(), 1, "{o}"); + } + } + #[test] fn forwarded_build_args_optimize_false_old_image_forwards_nothing() { // Old image defaults to not optimizing and rejects `--optimize=false`, @@ -855,7 +963,7 @@ mod tests { }, ..Cmd::default() }; - let args = forwarded_build_args(&cmd, &ws(), None, true, true, false); + let (args, _) = forwarded_build_args(&cmd, &ws(), None, false, true, false, false); assert!(!args.iter().any(|a| a.starts_with("--optimize"))); } @@ -865,7 +973,7 @@ mod tests { manifest_path: Some(PathBuf::from("/tmp/ws/contracts/add/Cargo.toml")), ..Cmd::default() }; - let args = forwarded_build_args(&cmd, &ws(), None, true, true, true); + let (args, _) = forwarded_build_args(&cmd, &ws(), None, false, true, true, false); assert!(args.contains(&"--manifest-path=contracts/add/Cargo.toml".to_string())); } diff --git a/cmd/soroban-cli/src/commands/contract/build/source_archive.rs b/cmd/soroban-cli/src/commands/contract/build/source_archive.rs index 9afc657a82..588f8ce384 100644 --- a/cmd/soroban-cli/src/commands/contract/build/source_archive.rs +++ b/cmd/soroban-cli/src/commands/contract/build/source_archive.rs @@ -350,21 +350,21 @@ fn gzip(bytes: &[u8]) -> Result, Error> { }) } +/// Decompress gzip and unpack the tar into `dest`. Entries are `source/…`, so +/// they land at `/source/…`. +pub(crate) fn unpack_targz(bytes: &[u8], dest: &Path) -> Result<(), Error> { + let dec = flate2::read::GzDecoder::new(bytes); + tar::Archive::new(dec) + .unpack(dest) + .map_err(Error::ArchiveExtract) +} + #[cfg(test)] mod tests { use super::*; use crate::config::locator::enforce_hardened_tree; use sha2::{Digest, Sha256}; - /// Decompress gzip and unpack the tar into `dest`. Entries are `source/…`, - /// so they land at `/source/…`. - fn unpack_targz(bytes: &[u8], dest: &Path) -> Result<(), Error> { - let dec = flate2::read::GzDecoder::new(bytes); - tar::Archive::new(dec) - .unpack(dest) - .map_err(Error::ArchiveExtract) - } - #[test] fn is_warned_matches_names_and_dotted_suffixes() { use std::ffi::OsStr; diff --git a/cmd/soroban-cli/src/commands/contract/build/verifiable.rs b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs new file mode 100644 index 0000000000..06153c3b18 --- /dev/null +++ b/cmd/soroban-cli/src/commands/contract/build/verifiable.rs @@ -0,0 +1,577 @@ +//! Verifiable (SEP-58 reproducible) contract builds. +//! +//! Triggered by `stellar contract build --verifiable`. Unlike the plain +//! `--image` container build, this snapshots the working tree into a +//! byte-reproducible source archive, hashes it (`source_sha256`), extracts it +//! into a permission-hardened tempdir, and builds *that* in a digest-pinned +//! image — recording SEP-58 provenance meta (`bldimg`, `source_uri`, +//! `source_sha256`, `bldopt`) into the wasm so a third party can reproduce the +//! exact bytes. +//! +//! The container execution machinery (image probe, `run_in_container`, +//! reproduce lines, artifact collection) is shared with +//! [`super::container`]; this module adds the archive and the SEP-58 metadata +//! on top. The build image is the user-supplied, digest-pinned `--image`. + +use std::path::{Path, PathBuf}; + +use regex::Regex; +use semver::Version; +use sha2::{Digest, Sha256}; +use soroban_spec_tools::sanitize; + +use crate::{ + commands::{ + container::shared::{self, Error as ConnectionError}, + global, + }, + config::{ + data, + locator::{enforce_hardened_tree, write_hardened_file}, + }, + print::Print, +}; + +use super::{container, source_archive, BuiltContract, Cmd}; + +const RESERVED_META_KEYS: &[&str] = &["bldimg", "source_uri", "source_sha256", "bldopt"]; + +#[derive(thiserror::Error, Debug)] +pub enum Error { + #[error(transparent)] + DockerConnection(#[from] ConnectionError), + + #[error("--image value {value:?} does not match the SEP-58 bldimg format `/@sha256:<64-hex>`. Examples: docker.io/stellar/stellar-cli@sha256:<64-hex>, localhost:5000/foo@sha256:<64-hex>. Tag-only refs and implicit Docker-Hub short refs are not accepted.")] + BldimgFormat { value: String }, + + #[error(transparent)] + SourceArchive(#[from] source_archive::Error), + + #[error( + "the cli sets bldimg, source_uri, source_sha256, and bldopt automatically when --verifiable is used; remove them from --meta. Got reserved key: {key}" + )] + ReservedMetaKey { key: String }, + + #[error("--source-sha256 value {value:?} does not match the SEP-58 source_sha256 format `^[0-9a-f]{{64}}$` (64-char lower-case hex).")] + SourceSha256Format { value: String }, + + #[error("--source-uri value {value:?} does not match the SEP-58 source_uri format `^[a-zA-Z][a-zA-Z0-9+.-]*:\\S+$` (a URI with a scheme, e.g. https://example.com/src.tar.gz).")] + SourceUriFormat { value: String }, + + #[error("--source-sha256 {provided} does not match the SHA-256 of the generated archive {computed}. Omit --source-sha256 to record the computed value, or fix the value.")] + SourceSha256Mismatch { provided: String, computed: String }, + + #[error("SEP-58 metadata must be ASCII, but {field} contains non-ASCII characters: {value}. Use an ASCII value (e.g. a punycode/percent-encoded URI).")] + NonAsciiMeta { field: String, value: String }, + + #[error(transparent)] + Data(#[from] data::Error), +} + +pub async fn run( + cmd: &Cmd, + global_args: &global::Args, + print: &Print, +) -> Result, super::Error> { + let _ = global_args; + + // Stage 1: pure validation, no I/O. + for (k, _) in &cmd.build_args.meta { + if RESERVED_META_KEYS.iter().any(|r| r == k) { + return Err(Error::ReservedMetaKey { key: k.clone() }.into()); + } + } + if let Some(img) = &cmd.image { + validate_image(img)?; + } + + // Stage 2: local filesystem + git, no network. + validate_source_formats(cmd)?; + + // The source root is the current working directory: it's archived, + // bind-mounted into the container, and the `--manifest-path` bldopt is + // relativized against it. Run from the project/workspace root you want built. + let source_root = source_archive::resolve_source_root(); + + // The archive is the working tree, so refuse a dirty repo: a verifiable build + // should be deliberate, off a committed state, not whatever happens to be on + // disk. Skipped when the source root isn't a git repo. + source_archive::ensure_clean_tree(&source_root, print).map_err(Error::from)?; + + // Build the source archive, record its hash, and build from the *extracted* + // archive (in a hardened tempdir) so the wasm is produced from exactly the + // bytes that were hashed. + let resolved = { + let a = resolve_archive(cmd, &source_root, print)?; + // The extracted `source/` dir mirrors `source_root` exactly and is both + // the container mount and the tree the build writes `target/` into. + let mount_root = a.extracted_root.join("source"); + ResolvedSource { + source_sha256: a.source_sha256, + extracted_root: Some(mount_root.clone()), + mount_root, + _tmp: Some(a.tmp), + } + }; + + let source_ids = SourceIds { + source_uri: cmd.source_uri.clone(), + source_sha256: Some(resolved.source_sha256.clone()), + }; + + // Stage 3: the container engine. + let docker = cmd.container_args.clone(); + docker.warn_if_host_ignored(print); + let image_ref = resolve_image(cmd, &docker, print).await?; + + // Probe the pinned image once for its cli binary, version, and default + // toolchain (shared with the plain container build), then gate flags on the + // reported version. + let probe = container::probe_image(&image_ref, &docker).await?; + let cli_version = probe.version.clone(); + let at_least = |min: &str| { + cli_version + .as_ref() + .is_none_or(|v| *v >= Version::parse(min).unwrap()) + }; + let supports_locked = at_least(container::LOCKED_MIN); + let supports_optimize_flag = at_least(container::OPTIMIZE_FLAG_MIN); + let supports_optimize_false = at_least(container::OPTIMIZE_NEW_SYNTAX_MIN); + + // `--locked` is implied by `--verifiable` (a reproducible build should pin + // the lockfile), but it was only added to `contract build` in cli 25.2.0. + if supports_locked { + if !cmd.locked { + print.infoln("Implying --locked because --verifiable was passed"); + } + } else { + print.warnln( + "The build image's `contract build` does not support --locked; \ + building without it. Dependency drift may affect reproducibility.", + ); + } + + // Resolve host `cargo metadata` once and reuse it for package selection and + // artifact collection, mirroring the plain container build. + let md = container::metadata(cmd).map_err(container::Error::Metadata)?; + + // Build once per package, each with its own `--package` forwarded and + // recorded as a `bldopt`, so every wasm is independently reproducible. + let packages = container::resolve_packages(cmd, &md); + if cmd.package.is_none() && !packages.is_empty() { + print.infoln(format!("Inferred packages: {}", packages.join(", "))); + } + let targets: Vec> = if packages.is_empty() { + vec![None] + } else { + packages.iter().map(|p| Some(p.as_str())).collect() + }; + let container_cmds: Vec> = targets + .iter() + .map(|target| { + // Verifiable implies `--locked` (when supported) and records every + // build-affecting flag as a `bldopt`. + let (mut args, bldopts) = container::forwarded_build_args( + cmd, + &source_root, + *target, + supports_locked, + supports_optimize_flag, + supports_optimize_false, + true, + ); + args.extend(build_metadata_args(&image_ref, &source_ids, &bldopts)); + args + }) + .collect(); + + // Pin the target dir to a known location under the mount, and the image's + // own default toolchain so a `rust-toolchain.toml` in the source can't + // redirect the build to a toolchain rustup would then try to install. + let mut env = vec!["CARGO_TARGET_DIR=/source/target".to_string()]; + print.infoln(format!("Using Rust toolchain {}", probe.toolchain)); + env.push(format!("RUSTUP_TOOLCHAIN={}", probe.toolchain)); + + container::run_in_container( + &image_ref, + &resolved.mount_root, + &container_cmds, + &env, + &docker, + &cmd.run_args, + &probe.bin, + "stellar-verifiable-build", + print, + cmd.print_commands_only, + ) + .await?; + + // Nothing was built when only printing the command. + if cmd.print_commands_only { + return Ok(Vec::new()); + } + + container::collect_built_contracts(cmd, &md, &source_root, resolved.extracted_root.as_deref()) +} + +/// The recorded `source_sha256`, the directory bind-mounted at `/source`, the +/// extracted-archive root, and its tempdir guard — held so the temp dir +/// outlives the container build and artifact collection. +struct ResolvedSource { + source_sha256: String, + mount_root: PathBuf, + extracted_root: Option, + _tmp: Option, +} + +/// Source-identification fields recorded as SEP-58 meta. `source_sha256` is +/// always `Some` by the time these are built in `run()` (computed from the +/// generated archive). `source_uri` is `Some` only when the user passed +/// `--source-uri`. +#[derive(Debug, Default, Clone)] +struct SourceIds { + source_uri: Option, + source_sha256: Option, +} + +/// Format-validate the user-supplied source flags. Both are optional under +/// `--verifiable`; `--source-sha256`, when present, is validated as a pin in +/// `resolve_archive`. +fn validate_source_formats(cmd: &Cmd) -> Result<(), Error> { + if let Some(sha) = &cmd.source_sha256 { + if !source_sha256_regex().is_match(sha) { + return Err(Error::SourceSha256Format { value: sha.clone() }); + } + } + if let Some(uri) = &cmd.source_uri { + // SEP-58 metadata is ASCII, but the URI regex's `\S` is Unicode-aware, so + // guard explicitly before the format check. + if !uri.is_ascii() { + return Err(Error::NonAsciiMeta { + field: "source_uri".to_string(), + value: sanitize(uri), + }); + } + if !source_uri_regex().is_match(uri) { + return Err(Error::SourceUriFormat { value: uri.clone() }); + } + } + Ok(()) +} + +/// Validate the SEP-58 `bldimg` (`--image`): it must be ASCII (the format regex +/// is Unicode-aware, so guard explicitly) and match the digest-pinned format. +/// The offending value is sanitized for display. +fn validate_image(image: &str) -> Result<(), Error> { + if !image.is_ascii() { + return Err(Error::NonAsciiMeta { + field: "bldimg".to_string(), + value: sanitize(image), + }); + } + if !bldimg_regex().is_match(image) { + return Err(Error::BldimgFormat { + value: image.to_string(), + }); + } + Ok(()) +} + +/// Outcome of archiving: the generated archive's SHA-256 and the directory it +/// was extracted into (held alive by `tmp`). +struct ArchiveResult { + source_sha256: String, + extracted_root: PathBuf, + tmp: tempfile::TempDir, +} + +/// Build the source archive, record its hash, write it to the managed archives +/// dir (content-addressed, so the bytes are available to upload for +/// `--source-uri`), and extract it into a permission-hardened tempdir that the +/// container then builds from. +fn resolve_archive(cmd: &Cmd, source_root: &Path, print: &Print) -> Result { + let bytes = source_archive::build_source_archive(source_root, print, true, None)?; + let computed = hex::encode(Sha256::digest(&bytes)); + + // If the user pinned a hash, it must match what we produced. + if let Some(provided) = &cmd.source_sha256 { + if provided != &computed { + return Err(Error::SourceSha256Mismatch { + provided: provided.clone(), + computed, + }); + } + } + + // Content-addressed name under the managed archives dir. The archive is the + // whole working tree, so it can hold private source or an unignored `.env`; + // write it `0600` (never the umask default `0644`) so it isn't world-readable. + let out_path = data::archives_dir()?.join(format!("{computed}.tar.gz")); + if let Some(parent) = out_path.parent() { + std::fs::create_dir_all(parent).map_err(|source| source_archive::Error::ArchiveWrite { + path: out_path.clone(), + source, + })?; + } + write_hardened_file(&out_path, &bytes).map_err(|source| { + source_archive::Error::ArchiveWrite { + path: out_path.clone(), + source, + } + })?; + print.infoln(format!( + "Wrote source archive {} (source_sha256 {computed})", + out_path.display() + )); + + // Extract and harden, then build from the extracted copy so the wasm is + // produced from exactly the archived bytes. + // + // Extract under the data dir, NOT the OS temp dir: on macOS `$TMPDIR` lives + // under /var/folders, which container VMs (Docker Desktop, Colima, …) don't + // share by default, so a bind mount of it would be empty inside the + // container. The data dir lives under the user's home, which is shared. + let base = data::data_local_dir()?; + std::fs::create_dir_all(&base).map_err(|source| source_archive::Error::ArchiveWrite { + path: base.clone(), + source, + })?; + let tmp = tempfile::Builder::new() + .prefix("verifiable-src-") + .tempdir_in(&base) + .map_err(source_archive::Error::ArchiveExtract)?; + source_archive::unpack_targz(&bytes, tmp.path())?; + enforce_hardened_tree(tmp.path()).map_err(source_archive::Error::ArchiveExtract)?; + + let extracted_root = tmp.path().to_path_buf(); + Ok(ArchiveResult { + source_sha256: computed, + extracted_root, + tmp, + }) +} + +fn bldimg_regex() -> Regex { + Regex::new(r"^(?:localhost(?::\d+)?|[^\s@/]*[.:][^\s@/]*)/[^\s@]+@sha256:[0-9a-f]{64}$") + .unwrap() +} + +fn source_sha256_regex() -> Regex { + Regex::new(r"^[0-9a-f]{64}$").unwrap() +} + +fn source_uri_regex() -> Regex { + Regex::new(r"^[a-zA-Z][a-zA-Z0-9+.-]*:\S+$").unwrap() +} + +/// Emit the SEP-58 `--meta` pairs recorded into the wasm: `bldimg` (the pinned +/// image digest) first, then `source_uri`/`source_sha256` when present, then one +/// `bldopt` per recorded build option. The bldopts already arrive as valid shell +/// (escaped at the source in `forwarded_build_args`), so a verifier reconstructs +/// the build by joining the recorded values and running them through a shell. +fn build_metadata_args(image_ref: &str, ids: &SourceIds, bldopts: &[String]) -> Vec { + let mut out = Vec::new(); + + let push = |out: &mut Vec, key: &str, val: &str| { + out.push("--meta".to_string()); + out.push(format!("{key}={val}")); + }; + + push(&mut out, "bldimg", image_ref); + + if let Some(v) = &ids.source_uri { + push(&mut out, "source_uri", v); + } + if let Some(v) = &ids.source_sha256 { + push(&mut out, "source_sha256", v); + } + + for o in bldopts { + push(&mut out, "bldopt", o); + } + + out +} + +/// The image to build in and record as `bldimg`: the user-supplied, +/// digest-pinned `--image` (required by clap and already format-validated in +/// `run`). It's a content-addressed ref, so it names the exact bytes as-is; +/// `--pull` refreshes it up front, otherwise a missing image is fetched by the +/// run itself, matching the plain container build. +async fn resolve_image(cmd: &Cmd, docker: &shared::Args, print: &Print) -> Result { + let image = cmd + .image + .clone() + .expect("--image is required with --verifiable (enforced by clap)"); + if cmd.pull { + docker.pull_image(&image, print).await?; + } + Ok(image) +} + +#[cfg(test)] +mod tests { + use super::*; + + fn pairs(args: &[String]) -> Vec<(&str, &str)> { + args.chunks(2) + .map(|c| (c[0].as_str(), c[1].as_str())) + .collect() + } + + #[test] + fn build_metadata_args_uri_and_sha256() { + let ids = SourceIds { + source_uri: Some("https://example.com/src.tar.gz".to_string()), + source_sha256: Some("a".repeat(64)), + }; + let m = build_metadata_args( + "docker.io/stellar/stellar-cli@sha256:abc", + &ids, + &["--locked".to_string(), "--features=a".to_string()], + ); + let p = pairs(&m); + // bldimg first; source_uri then source_sha256; bldopts last. + assert_eq!( + p[0], + ("--meta", "bldimg=docker.io/stellar/stellar-cli@sha256:abc") + ); + assert_eq!( + p[1], + ("--meta", "source_uri=https://example.com/src.tar.gz") + ); + assert_eq!(p[2].0, "--meta"); + assert!(p[2].1.starts_with("source_sha256=")); + assert_eq!(p[3], ("--meta", "bldopt=--locked")); + assert_eq!(p[4], ("--meta", "bldopt=--features=a")); + } + + #[test] + fn build_metadata_args_sha256_only_omits_uri() { + let ids = SourceIds { + source_sha256: Some("f".repeat(64)), + ..SourceIds::default() + }; + let m = build_metadata_args("docker.io/stellar/stellar-cli@sha256:abc", &ids, &[]); + assert!(m + .iter() + .any(|s| s == &format!("source_sha256={}", "f".repeat(64)))); + assert!(!m.iter().any(|s| s.starts_with("source_uri="))); + } + + #[test] + fn validate_source_formats_rejects_bad_sha256() { + let cmd = Cmd { + source_sha256: Some("not-a-sha".to_string()), + ..Cmd::default() + }; + let err = validate_source_formats(&cmd).unwrap_err(); + assert!(matches!(err, Error::SourceSha256Format { .. })); + } + + #[test] + fn validate_source_formats_rejects_bad_uri() { + let cmd = Cmd { + source_uri: Some("not a uri".to_string()), // no scheme + source_sha256: Some("a".repeat(64)), + ..Cmd::default() + }; + let err = validate_source_formats(&cmd).unwrap_err(); + assert!(matches!(err, Error::SourceUriFormat { .. })); + } + + // SEP-58 metadata must be ASCII; the URI regex's `\S` is Unicode-aware, so a + // non-ASCII (but otherwise well-formed) URI must still be rejected. + #[test] + fn validate_image_checks_ascii_then_format() { + // Non-ASCII registry/repo → rejected before the format check. + let err = + validate_image(&format!("localhost:5000/café@sha256:{}", "0".repeat(64))).unwrap_err(); + assert!(matches!(err, Error::NonAsciiMeta { .. }), "got {err:?}"); + // ASCII but tag-only → format error. + let err = validate_image("docker.io/stellar/stellar-cli:latest").unwrap_err(); + assert!(matches!(err, Error::BldimgFormat { .. }), "got {err:?}"); + // ASCII, digest-pinned → ok. + validate_image(&format!( + "docker.io/stellar/stellar-cli@sha256:{}", + "a".repeat(64) + )) + .unwrap(); + } + + #[test] + fn validate_source_formats_rejects_non_ascii_uri() { + let cmd = Cmd { + source_uri: Some("https://例.example/src.tar.gz".to_string()), + source_sha256: Some("a".repeat(64)), + ..Cmd::default() + }; + let err = validate_source_formats(&cmd).unwrap_err(); + assert!(matches!(err, Error::NonAsciiMeta { .. }), "got {err:?}"); + } + + #[test] + fn validate_source_formats_accepts_valid_and_absent() { + // Both absent is fine here — requiredness is enforced by clap/run(). + validate_source_formats(&Cmd::default()).unwrap(); + let cmd = Cmd { + source_uri: Some("https://example.com/src.tar.gz".to_string()), + source_sha256: Some("f".repeat(64)), + ..Cmd::default() + }; + validate_source_formats(&cmd).unwrap(); + } + + #[test] + fn bldimg_regex_accepts_docker_hub_full_ref() { + assert!(bldimg_regex().is_match(&format!( + "docker.io/stellar/stellar-cli@sha256:{}", + "a".repeat(64) + ))); + } + + #[test] + fn bldimg_regex_accepts_localhost_registry() { + assert!(bldimg_regex().is_match(&format!("localhost:5000/foo@sha256:{}", "0".repeat(64)))); + } + + #[test] + fn bldimg_regex_rejects_implicit_hub_short_ref() { + // Implicit Docker Hub short ref: no registry host prefix. + assert!(!bldimg_regex().is_match(&format!("stellar/stellar-cli@sha256:{}", "a".repeat(64)))); + } + + #[test] + fn bldimg_regex_rejects_tag_only() { + assert!(!bldimg_regex().is_match("docker.io/stellar/stellar-cli:latest")); + } + + #[test] + fn bldimg_regex_rejects_short_sha() { + assert!(!bldimg_regex().is_match("docker.io/stellar/stellar-cli@sha256:abc")); + } + + #[test] + fn source_sha256_regex_matches_64_hex() { + assert!(source_sha256_regex().is_match(&"f".repeat(64))); + assert!(!source_sha256_regex().is_match(&"f".repeat(63))); + assert!(!source_sha256_regex().is_match(&"F".repeat(64))); // upper-case rejected + } + + #[test] + fn source_uri_regex_accepts_any_scheme() { + assert!(source_uri_regex().is_match("https://example.com/src.tar.gz")); + assert!(source_uri_regex().is_match("http://example.com/foo.git")); + assert!(source_uri_regex().is_match("ipfs://Qm...abc")); + assert!(source_uri_regex().is_match("github:foo/bar")); + assert!(!source_uri_regex().is_match("foo/bar")); // no scheme + assert!(!source_uri_regex().is_match("https://has space")); // whitespace + } + + #[test] + fn reserved_meta_keys_list() { + for key in ["bldimg", "source_uri", "source_sha256", "bldopt"] { + assert!(RESERVED_META_KEYS.contains(&key)); + } + } +} diff --git a/cmd/soroban-cli/src/commands/mod.rs b/cmd/soroban-cli/src/commands/mod.rs index 4d82ce0890..2e879e8f0d 100644 --- a/cmd/soroban-cli/src/commands/mod.rs +++ b/cmd/soroban-cli/src/commands/mod.rs @@ -33,6 +33,7 @@ pub const HEADING_GLOBAL: &str = "Global Options"; pub const HEADING_SIGNING: &str = "Signing Options"; pub const HEADING_TRANSACTION: &str = "Transaction Options"; pub const HEADING_CONTAINER: &str = "Container Options"; +pub const HEADING_VERIFIABLE: &str = "Verifiable Options"; const ABOUT: &str = "Work seamlessly with Stellar accounts, contracts, and assets from the command line. diff --git a/cmd/soroban-cli/src/config/data.rs b/cmd/soroban-cli/src/config/data.rs index db123741a3..b726ab838d 100644 --- a/cmd/soroban-cli/src/config/data.rs +++ b/cmd/soroban-cli/src/config/data.rs @@ -59,6 +59,12 @@ pub fn bucket_dir() -> Result { Ok(dir) } +pub fn archives_dir() -> Result { + let dir = data_local_dir()?.join("archives"); + std::fs::create_dir_all(&dir)?; + Ok(dir) +} + pub fn write(action: Action, rpc_url: &Url) -> Result { let data = Data { action, @@ -213,6 +219,18 @@ mod test { use crate::test_utils::with_env_set; use serial_test::serial; + #[test] + #[serial] + fn archives_dir_under_data_home_and_created() { + let t = assert_fs::TempDir::new().unwrap(); + with_env_set("STELLAR_DATA_HOME", t.path(), || { + let dir = archives_dir().unwrap(); + assert!(dir.ends_with("archives")); + assert!(dir.starts_with(t.path())); + assert!(dir.is_dir(), "archives_dir() should create the directory"); + }); + } + #[test] #[serial] fn test_write_read() {