bv_core/
manifest.rs

1use std::collections::BTreeMap;
2use std::fmt;
3use std::path::PathBuf;
4use std::str::FromStr;
5
6use serde::{Deserialize, Serialize};
7
8use bv_types::{Cardinality, TypeRef};
9
10use crate::error::{BvError, Result};
11
12/// Quality and governance tier for a tool in the registry.
13#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default)]
14#[serde(rename_all = "lowercase")]
15pub enum Tier {
16    /// Typed I/O complete, conformance tests pass, from a recognized publisher, actively maintained.
17    Core,
18    /// Typed I/O present (may be partial), basic checks pass.
19    #[default]
20    Community,
21    /// Basic checks pass; may lack typed I/O. Hidden from default search results.
22    Experimental,
23}
24
25impl Tier {
26    pub fn as_str(&self) -> &'static str {
27        match self {
28            Tier::Core => "core",
29            Tier::Community => "community",
30            Tier::Experimental => "experimental",
31        }
32    }
33}
34
35impl fmt::Display for Tier {
36    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
37        f.write_str(self.as_str())
38    }
39}
40
41/// Structured CUDA version with ordering (`12.1 < 12.4 < 13.0`).
42#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord)]
43pub struct CudaVersion {
44    pub major: u32,
45    pub minor: u32,
46}
47
48impl fmt::Display for CudaVersion {
49    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
50        write!(f, "{}.{}", self.major, self.minor)
51    }
52}
53
54impl FromStr for CudaVersion {
55    type Err = String;
56
57    fn from_str(s: &str) -> std::result::Result<Self, Self::Err> {
58        let (maj, min) = s
59            .split_once('.')
60            .ok_or_else(|| format!("expected 'major.minor', got '{s}'"))?;
61        Ok(CudaVersion {
62            major: maj
63                .parse()
64                .map_err(|_| format!("invalid major version '{maj}'"))?,
65            minor: min
66                .parse()
67                .map_err(|_| format!("invalid minor version '{min}'"))?,
68        })
69    }
70}
71
72impl TryFrom<String> for CudaVersion {
73    type Error = String;
74    fn try_from(s: String) -> std::result::Result<Self, Self::Error> {
75        s.parse()
76    }
77}
78
79impl From<CudaVersion> for String {
80    fn from(v: CudaVersion) -> String {
81        v.to_string()
82    }
83}
84
85impl Serialize for CudaVersion {
86    fn serialize<S: serde::Serializer>(&self, s: S) -> std::result::Result<S::Ok, S::Error> {
87        s.serialize_str(&self.to_string())
88    }
89}
90
91impl<'de> Deserialize<'de> for CudaVersion {
92    fn deserialize<D: serde::Deserializer<'de>>(d: D) -> std::result::Result<Self, D::Error> {
93        let s = String::deserialize(d)?;
94        s.parse().map_err(serde::de::Error::custom)
95    }
96}
97
98#[derive(Debug, Clone, Serialize, Deserialize)]
99pub struct GpuSpec {
100    pub required: bool,
101    pub min_vram_gb: Option<u32>,
102    pub cuda_version: Option<CudaVersion>,
103}
104
105#[derive(Debug, Clone, Serialize, Deserialize)]
106pub struct HardwareSpec {
107    pub gpu: Option<GpuSpec>,
108    pub cpu_cores: Option<u32>,
109    pub ram_gb: Option<f64>,
110    pub disk_gb: Option<f64>,
111}
112
113impl HardwareSpec {
114    /// Check this manifest's requirements against the host's detected hardware.
115    /// Returns every requirement that is not satisfied.
116    pub fn check_against(
117        &self,
118        detected: &crate::hardware::DetectedHardware,
119    ) -> Vec<crate::hardware::HardwareMismatch> {
120        use crate::hardware::HardwareMismatch;
121        let mut out = Vec::new();
122
123        if let Some(gpu_req) = &self.gpu
124            && gpu_req.required
125        {
126            if detected.gpus.is_empty() {
127                out.push(HardwareMismatch::NoGpu);
128            } else {
129                if let Some(min_vram) = gpu_req.min_vram_gb {
130                    let best_vram_mb = detected.gpus.iter().map(|g| g.vram_mb).max().unwrap_or(0);
131                    // Round to nearest GiB instead of floor: nvidia-smi
132                    // typically reports just-under marketing capacity (e.g.
133                    // 24268 MiB on a "24 GB" RTX 3090). flooring made
134                    // min_vram_gb=24 spuriously fail on real hardware.
135                    let best_vram_gb = ((best_vram_mb as f64) / 1024.0).round() as u32;
136                    if best_vram_gb < min_vram {
137                        out.push(HardwareMismatch::InsufficientVram {
138                            required_gb: min_vram,
139                            available_gb: best_vram_gb,
140                        });
141                    }
142                }
143                if let Some(min_cuda) = &gpu_req.cuda_version {
144                    let best_cuda = detected
145                        .gpus
146                        .iter()
147                        .filter_map(|g| g.cuda_version.as_ref())
148                        .max();
149                    match best_cuda {
150                        None => out.push(HardwareMismatch::NoCuda {
151                            required: min_cuda.clone(),
152                        }),
153                        Some(avail) if avail < min_cuda => {
154                            out.push(HardwareMismatch::CudaTooOld {
155                                required: min_cuda.clone(),
156                                available: avail.clone(),
157                            });
158                        }
159                        _ => {}
160                    }
161                }
162            }
163        }
164
165        if let Some(min_ram) = self.ram_gb {
166            let avail = detected.ram_gb();
167            if avail < min_ram {
168                out.push(HardwareMismatch::InsufficientRam {
169                    required_gb: min_ram,
170                    available_gb: avail,
171                });
172            }
173        }
174
175        if let Some(min_disk) = self.disk_gb {
176            let avail = detected.disk_free_gb();
177            if avail < min_disk {
178                out.push(HardwareMismatch::InsufficientDisk {
179                    required_gb: min_disk,
180                    available_gb: avail,
181                });
182            }
183        }
184
185        out
186    }
187}
188
189#[derive(Debug, Clone, Serialize, Deserialize)]
190pub struct ImageSpec {
191    /// Runtime backend, e.g. `"docker"` or `"apptainer"`.
192    pub backend: String,
193    /// Canonical OCI reference, e.g. `"biocontainers/bwa:0.7.17"`.
194    pub reference: String,
195    /// Optional pinned digest for reproducibility.
196    pub digest: Option<String>,
197}
198
199#[derive(Debug, Clone, Serialize, Deserialize)]
200pub struct ReferenceDataSpec {
201    pub id: String,
202    pub version: String,
203    pub required: bool,
204    /// Container path where the dataset directory is mounted read-only.
205    #[serde(default, skip_serializing_if = "Option::is_none")]
206    pub mount_path: Option<String>,
207    /// Approximate compressed size in bytes.
208    #[serde(default, skip_serializing_if = "Option::is_none")]
209    pub size_bytes: Option<u64>,
210}
211
212/// Typed I/O port declaration for a tool.
213#[derive(Debug, Clone, Serialize, Deserialize)]
214pub struct IoSpec {
215    pub name: String,
216    /// Type reference, e.g. `"fasta"` or `"fasta[protein]"`.
217    #[serde(rename = "type")]
218    pub r#type: TypeRef,
219    /// How many values this port accepts.
220    #[serde(default)]
221    pub cardinality: Cardinality,
222    /// Absolute path inside the container where this value is mounted.
223    #[serde(default, skip_serializing_if = "Option::is_none")]
224    pub mount: Option<PathBuf>,
225    #[serde(default, skip_serializing_if = "Option::is_none")]
226    pub description: Option<String>,
227    #[serde(default, skip_serializing_if = "Option::is_none")]
228    pub default: Option<String>,
229}
230
231#[derive(Debug, Clone, Serialize, Deserialize)]
232pub struct EntrypointSpec {
233    pub command: String,
234    pub args_template: Option<String>,
235    #[serde(default)]
236    pub env: BTreeMap<String, String>,
237}
238
239/// Binary names that the tool's container exposes on PATH.
240///
241/// Omitting this block defaults to `exposed = [entrypoint.command]` for
242/// single-binary tools that do not need to declare anything extra.
243#[derive(Debug, Clone, Serialize, Deserialize)]
244pub struct BinariesSpec {
245    pub exposed: Vec<String>,
246}
247
248/// Per-tool overrides for `bv conformance`'s smoke check.
249///
250/// The smoke check tries a small set of probe args (`--version`, `-version`,
251/// `--help`, `-h`, `-v`, `version`) against every binary the tool exposes,
252/// and counts a binary as alive if any probe produces output or exits 0.
253/// Most tools don't need a `[tool.smoke]` block at all; this is the escape
254/// hatch for the unusual cases.
255#[derive(Debug, Clone, Default, Serialize, Deserialize)]
256pub struct SmokeSpec {
257    /// Override probe args for specific binaries, e.g. `{ "blastn" = "-version" }`.
258    /// Each value is a single command-line argument (or empty string for "run
259    /// the binary with no args"). When set, only this probe is tried for that
260    /// binary; the default list is bypassed.
261    #[serde(default, skip_serializing_if = "std::collections::BTreeMap::is_empty")]
262    pub probes: std::collections::BTreeMap<String, String>,
263    /// Binaries to skip entirely (daemons, "no non-destructive invocation"
264    /// tools, etc.). Listed binaries still appear in `[tool.binaries]` and
265    /// get shims; conformance just doesn't probe them.
266    #[serde(default, skip_serializing_if = "Vec::is_empty")]
267    pub skip: Vec<String>,
268}
269
270#[allow(dead_code)]
271fn default_timeout() -> u64 {
272    60
273}
274
275/// Optional Sigstore/cosign signature metadata.
276#[derive(Debug, Clone, Serialize, Deserialize)]
277pub struct SignatureSpec {
278    /// `"sigstore"` to verify the OCI image signature with cosign.
279    #[serde(default, skip_serializing_if = "Option::is_none")]
280    pub image: Option<String>,
281    /// `"sigstore"` to verify the manifest's commit signature.
282    #[serde(default, skip_serializing_if = "Option::is_none")]
283    pub manifest: Option<String>,
284}
285
286#[derive(Debug, Clone, Serialize, Deserialize)]
287pub struct ToolManifest {
288    pub id: String,
289    pub version: String,
290    pub description: Option<String>,
291    pub homepage: Option<String>,
292    pub license: Option<String>,
293    /// Governance tier. Defaults to `community` for new submissions.
294    #[serde(default)]
295    pub tier: Tier,
296    /// GitHub handles of maintainers, e.g. `"github:alice"`.
297    #[serde(default, skip_serializing_if = "Vec::is_empty")]
298    pub maintainers: Vec<String>,
299    /// Set to `true` when a tool is superseded or no longer maintained.
300    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
301    pub deprecated: bool,
302    pub image: ImageSpec,
303    pub hardware: HardwareSpec,
304    #[serde(default)]
305    pub reference_data: BTreeMap<String, ReferenceDataSpec>,
306    /// Typed inputs. Optional; manifests without this section parse unchanged.
307    #[serde(default)]
308    pub inputs: Vec<IoSpec>,
309    /// Typed outputs. Optional; manifests without this section parse unchanged.
310    #[serde(default)]
311    pub outputs: Vec<IoSpec>,
312    /// Default invocation. Required unless `[tool.subcommands]` is non-empty;
313    /// see `validate()`. Multi-script tools may omit this entirely.
314    #[serde(default, skip_serializing_if = "Option::is_none")]
315    pub entrypoint: Option<EntrypointSpec>,
316    /// Tool-namespaced launchers. Reachable as `bv run <toolid> <name> ...args`.
317    /// Each value is the literal argv prefix; user args are appended verbatim.
318    /// Unlike `[tool.binaries]`, names are not exposed on PATH or in the global
319    /// binary index, so generic names (`train`, `eval`) are safe.
320    #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
321    pub subcommands: BTreeMap<String, Vec<String>>,
322    /// Container paths the tool writes to during normal execution and that
323    /// should therefore be bound to writable host directories. Critical on
324    /// apptainer (read-only SIF root), nice-to-have on docker (lets caches
325    /// outlive `docker rm`). Tool authors declare these; users can override
326    /// the host side via `[[cache]]` in `bv.toml`.
327    #[serde(default, skip_serializing_if = "Vec::is_empty")]
328    pub cache_paths: Vec<String>,
329    /// Binary names this tool exposes on PATH inside its container.
330    /// Omit for single-binary tools; defaults to `[entrypoint.command]`.
331    #[serde(default, skip_serializing_if = "Option::is_none")]
332    pub binaries: Option<BinariesSpec>,
333    /// Smoke-check overrides; consulted by `bv conformance` for unusual binaries.
334    #[serde(default, skip_serializing_if = "Option::is_none")]
335    pub smoke: Option<SmokeSpec>,
336    /// Sigstore / cosign signature declarations.
337    #[serde(default, skip_serializing_if = "Option::is_none")]
338    pub signatures: Option<SignatureSpec>,
339}
340
341impl ToolManifest {
342    pub fn has_typed_io(&self) -> bool {
343        !self.inputs.is_empty() || !self.outputs.is_empty()
344    }
345
346    /// Returns the effective list of binary names this tool exposes.
347    ///
348    /// When `[tool.binaries]` is absent, defaults to the entrypoint command's
349    /// basename. Multi-script tools without an entrypoint expose no binaries
350    /// (their subcommands stay namespaced under the tool id).
351    pub fn effective_binaries(&self) -> Vec<&str> {
352        if let Some(b) = &self.binaries {
353            return b.exposed.iter().map(|s| s.as_str()).collect();
354        }
355        let Some(ep) = &self.entrypoint else {
356            return vec![];
357        };
358        let cmd = &ep.command;
359        let name = cmd
360            .rfind('/')
361            .map(|i| &cmd[i + 1..])
362            .unwrap_or(cmd.as_str());
363        vec![name]
364    }
365}
366
367/// Top-level manifest, corresponding to a single `.toml` file in the registry.
368#[derive(Debug, Clone, Serialize, Deserialize)]
369pub struct Manifest {
370    pub tool: ToolManifest,
371}
372
373#[derive(Debug)]
374pub struct ValidationError {
375    pub field: String,
376    pub message: String,
377}
378
379impl fmt::Display for ValidationError {
380    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
381        write!(f, "{}: {}", self.field, self.message)
382    }
383}
384
385impl Manifest {
386    pub fn from_toml_str(s: &str) -> Result<Self> {
387        let m: Manifest = toml::from_str(s).map_err(|e| BvError::ManifestParse(e.to_string()))?;
388        m.validate_types()?;
389        if let Err(errs) = m.validate() {
390            let combined = errs
391                .iter()
392                .map(|e| e.to_string())
393                .collect::<Vec<_>>()
394                .join("; ");
395            return Err(BvError::ManifestParse(format!(
396                "manifest validation failed: {combined}"
397            )));
398        }
399        Ok(m)
400    }
401
402    pub fn to_toml_string(&self) -> Result<String> {
403        toml::to_string_pretty(self).map_err(|e| BvError::ManifestParse(e.to_string()))
404    }
405
406    /// Validates that all TypeRefs in inputs/outputs exist in the bv-types vocabulary.
407    fn validate_types(&self) -> Result<()> {
408        let t = &self.tool;
409        for (side, specs) in [("inputs", &t.inputs), ("outputs", &t.outputs)] {
410            for spec in specs {
411                let id = spec.r#type.base_id();
412                if bv_types::lookup(id).is_none() {
413                    let suggestion = bv_types::suggest(id)
414                        .map(|s| format!(", did you mean `{s}`?"))
415                        .unwrap_or_default();
416                    return Err(BvError::ManifestParse(format!(
417                        "tool.{side}[{}]: unknown type `{id}`{suggestion}",
418                        spec.name
419                    )));
420                }
421            }
422        }
423        Ok(())
424    }
425
426    /// Returns a list of validation errors, or `Ok(())` if the manifest is valid.
427    pub fn validate(&self) -> std::result::Result<(), Vec<ValidationError>> {
428        let mut errors = Vec::new();
429        let t = &self.tool;
430
431        if t.id.is_empty() {
432            errors.push(ValidationError {
433                field: "tool.id".into(),
434                message: "must not be empty".into(),
435            });
436        }
437        if t.version.is_empty() {
438            errors.push(ValidationError {
439                field: "tool.version".into(),
440                message: "must not be empty".into(),
441            });
442        }
443        if t.image.backend.is_empty() {
444            errors.push(ValidationError {
445                field: "tool.image.backend".into(),
446                message: "must not be empty".into(),
447            });
448        }
449        if t.image.reference.is_empty() {
450            errors.push(ValidationError {
451                field: "tool.image.reference".into(),
452                message: "must not be empty".into(),
453            });
454        }
455        match (&t.entrypoint, t.subcommands.is_empty()) {
456            (None, true) => errors.push(ValidationError {
457                field: "tool.entrypoint".into(),
458                message: "must declare either [tool.entrypoint] or [tool.subcommands]".into(),
459            }),
460            (Some(ep), _) if ep.command.is_empty() => errors.push(ValidationError {
461                field: "tool.entrypoint.command".into(),
462                message: "must not be empty".into(),
463            }),
464            _ => {}
465        }
466
467        for (name, cmd) in &t.subcommands {
468            if name.is_empty() {
469                errors.push(ValidationError {
470                    field: "tool.subcommands".into(),
471                    message: "subcommand name must not be empty".into(),
472                });
473                continue;
474            }
475            if name.starts_with('-') {
476                errors.push(ValidationError {
477                    field: format!("tool.subcommands.{name}"),
478                    message: "subcommand name must not start with '-'".into(),
479                });
480            }
481            if cmd.is_empty() {
482                errors.push(ValidationError {
483                    field: format!("tool.subcommands.{name}"),
484                    message: "command vector must not be empty".into(),
485                });
486            }
487        }
488
489        for spec in &t.inputs {
490            if let Some(mount) = &spec.mount
491                && !mount.is_absolute()
492            {
493                errors.push(ValidationError {
494                    field: format!("tool.inputs[{}].mount", spec.name),
495                    message: "must be an absolute path".into(),
496                });
497            }
498        }
499        for spec in &t.outputs {
500            if let Some(mount) = &spec.mount
501                && !mount.is_absolute()
502            {
503                errors.push(ValidationError {
504                    field: format!("tool.outputs[{}].mount", spec.name),
505                    message: "must be an absolute path".into(),
506                });
507            }
508        }
509
510        if let Some(binaries) = &t.binaries {
511            let mut seen = std::collections::HashSet::new();
512            for name in &binaries.exposed {
513                if !seen.insert(name.as_str()) {
514                    errors.push(ValidationError {
515                        field: "tool.binaries.exposed".into(),
516                        message: format!("duplicate binary name '{name}'"),
517                    });
518                }
519            }
520            if !binaries.exposed.is_empty()
521                && let Some(ep) = &t.entrypoint
522            {
523                let cmd = &ep.command;
524                let basename = cmd.rfind('/').map(|i| &cmd[i + 1..]).unwrap_or(cmd);
525                if !binaries.exposed.iter().any(|b| b == basename) {
526                    errors.push(ValidationError {
527                        field: "tool.binaries.exposed".into(),
528                        message: format!(
529                            "entrypoint command '{basename}' must be listed in exposed"
530                        ),
531                    });
532                }
533            }
534        }
535
536        if errors.is_empty() {
537            Ok(())
538        } else {
539            Err(errors)
540        }
541    }
542}
543
544#[cfg(test)]
545mod tests {
546    use super::*;
547
548    const SAMPLE: &str = r#"
549[tool]
550id = "bwa"
551version = "0.7.17"
552description = "BWA short-read aligner"
553homepage = "http://bio-bwa.sourceforge.net/"
554license = "GPL-3.0"
555
556[tool.image]
557backend = "docker"
558reference = "biocontainers/bwa:0.7.17--h5bf99c6_8"
559digest = "sha256:abcdef1234567890abcdef1234567890abcdef1234567890abcdef1234567890ab"
560
561[tool.hardware]
562cpu_cores = 8
563ram_gb = 32.0
564disk_gb = 50.0
565
566[tool.hardware.gpu]
567required = false
568
569[[tool.inputs]]
570name = "reads_r1"
571type = "fastq"
572cardinality = "one"
573description = "Forward reads"
574
575[[tool.inputs]]
576name = "reads_r2"
577type = "fastq"
578cardinality = "optional"
579description = "Reverse reads (paired-end)"
580
581[[tool.outputs]]
582name = "alignment"
583type = "bam"
584description = "Aligned reads"
585
586[tool.entrypoint]
587command = "bwa"
588args_template = "mem -t {cpu_cores} {reference} {reads_r1} {reads_r2}"
589
590[tool.entrypoint.env]
591MALLOC_ARENA_MAX = "4"
592"#;
593
594    const SAMPLE_NO_IO: &str = r#"
595[tool]
596id = "mytool"
597version = "1.0.0"
598
599[tool.image]
600backend = "docker"
601reference = "example/mytool:1.0.0"
602
603[tool.hardware]
604
605[tool.entrypoint]
606command = "mytool"
607"#;
608
609    #[test]
610    fn round_trip() {
611        let manifest = Manifest::from_toml_str(SAMPLE).expect("parse failed");
612        assert_eq!(manifest.tool.id, "bwa");
613        assert_eq!(manifest.tool.version, "0.7.17");
614        assert_eq!(manifest.tool.image.backend, "docker");
615        assert_eq!(manifest.tool.inputs.len(), 2);
616        assert_eq!(manifest.tool.outputs.len(), 1);
617        assert_eq!(manifest.tool.inputs[0].cardinality, Cardinality::One);
618        assert_eq!(manifest.tool.inputs[1].cardinality, Cardinality::Optional);
619
620        let serialised = manifest.to_toml_string().expect("serialise failed");
621        let reparsed = Manifest::from_toml_str(&serialised).expect("reparse failed");
622        assert_eq!(reparsed.tool.id, manifest.tool.id);
623        assert_eq!(reparsed.tool.version, manifest.tool.version);
624    }
625
626    /// Regression: HashMap-backed fields produced non-deterministic TOML
627    /// output, breaking lockfile drift detection. Re-serializing the same
628    /// manifest must always yield identical bytes.
629    #[test]
630    fn to_toml_string_is_deterministic_with_subcommands() {
631        let s = r#"
632[tool]
633id = "multi"
634version = "1.0.0"
635
636[tool.image]
637backend = "docker"
638reference = "example/multi:1.0.0"
639
640[tool.hardware]
641
642[tool.entrypoint]
643command = "main"
644
645[tool.subcommands]
646zebra = ["script_z.py"]
647alpha = ["script_a.py"]
648mango = ["python", "-m", "scripts.mango"]
649beta = ["script_b.py"]
650"#;
651        let m = Manifest::from_toml_str(s).expect("parse");
652        let a = m.to_toml_string().unwrap();
653        // Re-serialize many times to make iteration-order luck unlikely.
654        for _ in 0..32 {
655            assert_eq!(a, m.to_toml_string().unwrap(), "non-deterministic output");
656        }
657        // And the keys must appear in lexicographic order (BTreeMap).
658        let alpha = a.find("alpha = ").unwrap();
659        let beta = a.find("beta = ").unwrap();
660        let mango = a.find("mango = ").unwrap();
661        let zebra = a.find("zebra = ").unwrap();
662        assert!(alpha < beta && beta < mango && mango < zebra);
663    }
664
665    #[test]
666    fn no_io_parses_unchanged() {
667        let m = Manifest::from_toml_str(SAMPLE_NO_IO).expect("parse failed");
668        assert!(m.tool.inputs.is_empty());
669        assert!(m.tool.outputs.is_empty());
670        assert!(!m.tool.has_typed_io());
671    }
672
673    #[test]
674    fn typeref_params_parsed() {
675        let s = r#"
676[tool]
677id = "t"
678version = "1.0.0"
679
680[tool.image]
681backend = "docker"
682reference = "example/t:1.0.0"
683
684[tool.hardware]
685
686[[tool.inputs]]
687name = "seqs"
688type = "fasta[protein]"
689cardinality = "one"
690
691[tool.entrypoint]
692command = "t"
693"#;
694        let m = Manifest::from_toml_str(s).unwrap();
695        assert_eq!(m.tool.inputs[0].r#type.params, vec!["protein"]);
696    }
697
698    #[test]
699    fn unknown_type_error() {
700        let s = r#"
701[tool]
702id = "t"
703version = "1.0.0"
704
705[tool.image]
706backend = "docker"
707reference = "example/t:1.0.0"
708
709[tool.hardware]
710
711[[tool.inputs]]
712name = "seqs"
713type = "protien_fasta"
714cardinality = "one"
715
716[tool.entrypoint]
717command = "t"
718"#;
719        let err = Manifest::from_toml_str(s).unwrap_err();
720        let msg = err.to_string();
721        assert!(msg.contains("unknown type"), "got: {msg}");
722    }
723
724    #[test]
725    fn cuda_version_ordering() {
726        let v12_1: CudaVersion = "12.1".parse().unwrap();
727        let v12_4: CudaVersion = "12.4".parse().unwrap();
728        let v13_0: CudaVersion = "13.0".parse().unwrap();
729        assert!(v12_1 < v12_4);
730        assert!(v12_4 < v13_0);
731        assert_eq!(v12_1, "12.1".parse::<CudaVersion>().unwrap());
732    }
733
734    #[test]
735    fn subcommands_only_parses() {
736        let s = r#"
737[tool]
738id = "genie2"
739version = "1.0.0"
740
741[tool.image]
742backend = "docker"
743reference = "ghcr.io/example/genie2:1.0.0"
744
745[tool.hardware]
746
747[tool.subcommands]
748train                = ["python", "genie/train.py"]
749sample_unconditional = ["python", "genie/sample_unconditional.py"]
750"#;
751        let m = Manifest::from_toml_str(s).unwrap();
752        assert!(m.tool.entrypoint.is_none());
753        assert_eq!(m.tool.subcommands.len(), 2);
754        assert_eq!(
755            m.tool.subcommands.get("train").unwrap(),
756            &vec!["python".to_string(), "genie/train.py".to_string()]
757        );
758        m.validate().expect("subcommand-only manifest is valid");
759        // No entrypoint and no [tool.binaries] => no exposed binaries.
760        assert!(m.tool.effective_binaries().is_empty());
761    }
762
763    #[test]
764    fn validate_requires_entrypoint_or_subcommands() {
765        let s = r#"
766[tool]
767id = "broken"
768version = "1.0.0"
769
770[tool.image]
771backend = "docker"
772reference = "example/broken:1.0.0"
773
774[tool.hardware]
775"#;
776        // from_toml_str now runs validate(); manifest with neither entrypoint
777        // nor subcommands must be rejected at parse time.
778        let err = Manifest::from_toml_str(s).unwrap_err();
779        assert!(
780            err.to_string().contains("tool.entrypoint"),
781            "expected entrypoint-or-subcommands error, got: {err}"
782        );
783    }
784
785    #[test]
786    fn validate_rejects_dash_prefixed_subcommand() {
787        let s = r#"
788[tool]
789id = "t"
790version = "1.0.0"
791
792[tool.image]
793backend = "docker"
794reference = "example/t:1.0.0"
795
796[tool.hardware]
797
798[tool.subcommands]
799"-bad" = ["python", "x.py"]
800"#;
801        let err = Manifest::from_toml_str(s).unwrap_err();
802        assert!(err.to_string().contains("-bad"), "got: {err}");
803    }
804
805    #[test]
806    fn subcommands_round_trip() {
807        let s = r#"
808[tool]
809id = "t"
810version = "1.0.0"
811
812[tool.image]
813backend = "docker"
814reference = "example/t:1.0.0"
815
816[tool.hardware]
817
818[tool.subcommands]
819go = ["python", "main.py"]
820"#;
821        let m = Manifest::from_toml_str(s).unwrap();
822        let serialised = m.to_toml_string().unwrap();
823        let reparsed = Manifest::from_toml_str(&serialised).unwrap();
824        assert_eq!(reparsed.tool.subcommands.len(), 1);
825    }
826
827    #[test]
828    fn validate_catches_empty_id() {
829        let mut manifest = Manifest::from_toml_str(SAMPLE).unwrap();
830        manifest.tool.id = String::new();
831        let errs = manifest.validate().unwrap_err();
832        assert!(errs.iter().any(|e| e.field == "tool.id"));
833    }
834
835    #[test]
836    fn registry_manifests_parse() {
837        let registry = concat!(env!("CARGO_MANIFEST_DIR"), "/../../bv-registry/tools");
838        let Ok(read) = std::fs::read_dir(registry) else {
839            return;
840        };
841        for entry in read {
842            let tool_dir = entry.unwrap().path();
843            if !tool_dir.is_dir() {
844                continue;
845            }
846            for version_entry in std::fs::read_dir(&tool_dir).unwrap() {
847                let path = version_entry.unwrap().path();
848                if path.extension().is_some_and(|e| e == "toml") {
849                    let s = std::fs::read_to_string(&path)
850                        .unwrap_or_else(|_| panic!("failed to read {}", path.display()));
851                    Manifest::from_toml_str(&s)
852                        .unwrap_or_else(|e| panic!("{}: {e}", path.display()));
853                }
854            }
855        }
856    }
857}
bv_core/manifest.rs

bv_core/
manifest.rs
