bv_core/
manifest.rs

1use std::collections::HashMap;
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                    let best_vram_gb = (best_vram_mb as f64 / 1024.0).floor() as u32;
132                    if best_vram_gb < min_vram {
133                        out.push(HardwareMismatch::InsufficientVram {
134                            required_gb: min_vram,
135                            available_gb: best_vram_gb,
136                        });
137                    }
138                }
139                if let Some(min_cuda) = &gpu_req.cuda_version {
140                    let best_cuda = detected
141                        .gpus
142                        .iter()
143                        .filter_map(|g| g.cuda_version.as_ref())
144                        .max();
145                    match best_cuda {
146                        None => out.push(HardwareMismatch::NoCuda {
147                            required: min_cuda.clone(),
148                        }),
149                        Some(avail) if avail < min_cuda => {
150                            out.push(HardwareMismatch::CudaTooOld {
151                                required: min_cuda.clone(),
152                                available: avail.clone(),
153                            });
154                        }
155                        _ => {}
156                    }
157                }
158            }
159        }
160
161        if let Some(min_ram) = self.ram_gb {
162            let avail = detected.ram_gb();
163            if avail < min_ram {
164                out.push(HardwareMismatch::InsufficientRam {
165                    required_gb: min_ram,
166                    available_gb: avail,
167                });
168            }
169        }
170
171        if let Some(min_disk) = self.disk_gb {
172            let avail = detected.disk_free_gb();
173            if avail < min_disk {
174                out.push(HardwareMismatch::InsufficientDisk {
175                    required_gb: min_disk,
176                    available_gb: avail,
177                });
178            }
179        }
180
181        out
182    }
183}
184
185#[derive(Debug, Clone, Serialize, Deserialize)]
186pub struct ImageSpec {
187    /// Runtime backend, e.g. `"docker"` or `"apptainer"`.
188    pub backend: String,
189    /// Canonical OCI reference, e.g. `"biocontainers/bwa:0.7.17"`.
190    pub reference: String,
191    /// Optional pinned digest for reproducibility.
192    pub digest: Option<String>,
193}
194
195#[derive(Debug, Clone, Serialize, Deserialize)]
196pub struct ReferenceDataSpec {
197    pub id: String,
198    pub version: String,
199    pub required: bool,
200    /// Container path where the dataset directory is mounted read-only.
201    #[serde(default, skip_serializing_if = "Option::is_none")]
202    pub mount_path: Option<String>,
203    /// Approximate compressed size in bytes.
204    #[serde(default, skip_serializing_if = "Option::is_none")]
205    pub size_bytes: Option<u64>,
206}
207
208/// Typed I/O port declaration for a tool.
209#[derive(Debug, Clone, Serialize, Deserialize)]
210pub struct IoSpec {
211    pub name: String,
212    /// Type reference, e.g. `"fasta"` or `"fasta[protein]"`.
213    #[serde(rename = "type")]
214    pub r#type: TypeRef,
215    /// How many values this port accepts.
216    #[serde(default)]
217    pub cardinality: Cardinality,
218    /// Absolute path inside the container where this value is mounted.
219    #[serde(default, skip_serializing_if = "Option::is_none")]
220    pub mount: Option<PathBuf>,
221    #[serde(default, skip_serializing_if = "Option::is_none")]
222    pub description: Option<String>,
223    #[serde(default, skip_serializing_if = "Option::is_none")]
224    pub default: Option<String>,
225}
226
227#[derive(Debug, Clone, Serialize, Deserialize)]
228pub struct EntrypointSpec {
229    pub command: String,
230    pub args_template: Option<String>,
231    #[serde(default)]
232    pub env: HashMap<String, String>,
233}
234
235/// Binary names that the tool's container exposes on PATH.
236///
237/// Omitting this block defaults to `exposed = [entrypoint.command]` for
238/// single-binary tools that do not need to declare anything extra.
239#[derive(Debug, Clone, Serialize, Deserialize)]
240pub struct BinariesSpec {
241    pub exposed: Vec<String>,
242}
243
244/// Canonical inputs and expected outputs used by the conformance test runner.
245#[derive(Debug, Clone, Serialize, Deserialize)]
246pub struct TestSpec {
247    /// Map of port name to a `test://` URI identifying the canonical input.
248    #[serde(default)]
249    pub inputs: std::collections::HashMap<String, String>,
250    /// Port names whose output files must exist and pass type-level checks.
251    #[serde(default)]
252    pub expected_outputs: Vec<String>,
253    /// Additional CLI args appended to the entrypoint during test runs.
254    #[serde(default)]
255    pub extra_args: Vec<String>,
256    /// Seconds before the conformance run is killed.
257    #[serde(default = "default_timeout")]
258    pub timeout_seconds: u64,
259    /// When true, skipped in fast CI and run only on a separate slow schedule.
260    #[serde(default)]
261    pub slow: bool,
262}
263
264fn default_timeout() -> u64 {
265    60
266}
267
268/// Optional Sigstore/cosign signature metadata.
269#[derive(Debug, Clone, Serialize, Deserialize)]
270pub struct SignatureSpec {
271    /// `"sigstore"` to verify the OCI image signature with cosign.
272    #[serde(default, skip_serializing_if = "Option::is_none")]
273    pub image: Option<String>,
274    /// `"sigstore"` to verify the manifest's commit signature.
275    #[serde(default, skip_serializing_if = "Option::is_none")]
276    pub manifest: Option<String>,
277}
278
279#[derive(Debug, Clone, Serialize, Deserialize)]
280pub struct ToolManifest {
281    pub id: String,
282    pub version: String,
283    pub description: Option<String>,
284    pub homepage: Option<String>,
285    pub license: Option<String>,
286    /// Governance tier. Defaults to `community` for new submissions.
287    #[serde(default)]
288    pub tier: Tier,
289    /// GitHub handles of maintainers, e.g. `"github:alice"`.
290    #[serde(default, skip_serializing_if = "Vec::is_empty")]
291    pub maintainers: Vec<String>,
292    /// Set to `true` when a tool is superseded or no longer maintained.
293    #[serde(default, skip_serializing_if = "std::ops::Not::not")]
294    pub deprecated: bool,
295    pub image: ImageSpec,
296    pub hardware: HardwareSpec,
297    #[serde(default)]
298    pub reference_data: HashMap<String, ReferenceDataSpec>,
299    /// Typed inputs. Optional; manifests without this section parse unchanged.
300    #[serde(default)]
301    pub inputs: Vec<IoSpec>,
302    /// Typed outputs. Optional; manifests without this section parse unchanged.
303    #[serde(default)]
304    pub outputs: Vec<IoSpec>,
305    pub entrypoint: EntrypointSpec,
306    /// Binary names this tool exposes on PATH inside its container.
307    /// Omit for single-binary tools; defaults to `[entrypoint.command]`.
308    #[serde(default, skip_serializing_if = "Option::is_none")]
309    pub binaries: Option<BinariesSpec>,
310    /// Conformance test block; used by `bv conformance check`.
311    #[serde(default, skip_serializing_if = "Option::is_none")]
312    pub test: Option<TestSpec>,
313    /// Sigstore / cosign signature declarations.
314    #[serde(default, skip_serializing_if = "Option::is_none")]
315    pub signatures: Option<SignatureSpec>,
316}
317
318impl ToolManifest {
319    pub fn has_typed_io(&self) -> bool {
320        !self.inputs.is_empty() || !self.outputs.is_empty()
321    }
322
323    /// Returns the effective list of binary names this tool exposes.
324    ///
325    /// When no `[tool.binaries]` block is present, defaults to
326    /// `[entrypoint.command]` (the basename component for path-style commands).
327    pub fn effective_binaries(&self) -> Vec<&str> {
328        match &self.binaries {
329            Some(b) => b.exposed.iter().map(|s| s.as_str()).collect(),
330            None => {
331                let cmd = &self.entrypoint.command;
332                let name = cmd
333                    .rfind('/')
334                    .map(|i| &cmd[i + 1..])
335                    .unwrap_or(cmd.as_str());
336                vec![name]
337            }
338        }
339    }
340}
341
342/// Top-level manifest, corresponding to a single `.toml` file in the registry.
343#[derive(Debug, Clone, Serialize, Deserialize)]
344pub struct Manifest {
345    pub tool: ToolManifest,
346}
347
348#[derive(Debug)]
349pub struct ValidationError {
350    pub field: String,
351    pub message: String,
352}
353
354impl fmt::Display for ValidationError {
355    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
356        write!(f, "{}: {}", self.field, self.message)
357    }
358}
359
360impl Manifest {
361    pub fn from_toml_str(s: &str) -> Result<Self> {
362        let m: Manifest = toml::from_str(s).map_err(|e| BvError::ManifestParse(e.to_string()))?;
363        m.validate_types()?;
364        Ok(m)
365    }
366
367    pub fn to_toml_string(&self) -> Result<String> {
368        toml::to_string_pretty(self).map_err(|e| BvError::ManifestParse(e.to_string()))
369    }
370
371    /// Validates that all TypeRefs in inputs/outputs exist in the bv-types vocabulary.
372    fn validate_types(&self) -> Result<()> {
373        let t = &self.tool;
374        for (side, specs) in [("inputs", &t.inputs), ("outputs", &t.outputs)] {
375            for spec in specs {
376                let id = spec.r#type.base_id();
377                if bv_types::lookup(id).is_none() {
378                    let suggestion = bv_types::suggest(id)
379                        .map(|s| format!(", did you mean `{s}`?"))
380                        .unwrap_or_default();
381                    return Err(BvError::ManifestParse(format!(
382                        "tool.{side}[{}]: unknown type `{id}`{suggestion}",
383                        spec.name
384                    )));
385                }
386            }
387        }
388        Ok(())
389    }
390
391    /// Returns a list of validation errors, or `Ok(())` if the manifest is valid.
392    pub fn validate(&self) -> std::result::Result<(), Vec<ValidationError>> {
393        let mut errors = Vec::new();
394        let t = &self.tool;
395
396        if t.id.is_empty() {
397            errors.push(ValidationError {
398                field: "tool.id".into(),
399                message: "must not be empty".into(),
400            });
401        }
402        if t.version.is_empty() {
403            errors.push(ValidationError {
404                field: "tool.version".into(),
405                message: "must not be empty".into(),
406            });
407        }
408        if t.image.backend.is_empty() {
409            errors.push(ValidationError {
410                field: "tool.image.backend".into(),
411                message: "must not be empty".into(),
412            });
413        }
414        if t.image.reference.is_empty() {
415            errors.push(ValidationError {
416                field: "tool.image.reference".into(),
417                message: "must not be empty".into(),
418            });
419        }
420        if t.entrypoint.command.is_empty() {
421            errors.push(ValidationError {
422                field: "tool.entrypoint.command".into(),
423                message: "must not be empty".into(),
424            });
425        }
426
427        for spec in &t.inputs {
428            if let Some(mount) = &spec.mount
429                && !mount.is_absolute()
430            {
431                errors.push(ValidationError {
432                    field: format!("tool.inputs[{}].mount", spec.name),
433                    message: "must be an absolute path".into(),
434                });
435            }
436        }
437        for spec in &t.outputs {
438            if let Some(mount) = &spec.mount
439                && !mount.is_absolute()
440            {
441                errors.push(ValidationError {
442                    field: format!("tool.outputs[{}].mount", spec.name),
443                    message: "must be an absolute path".into(),
444                });
445            }
446        }
447
448        if let Some(binaries) = &t.binaries {
449            let mut seen = std::collections::HashSet::new();
450            for name in &binaries.exposed {
451                if !seen.insert(name.as_str()) {
452                    errors.push(ValidationError {
453                        field: "tool.binaries.exposed".into(),
454                        message: format!("duplicate binary name '{name}'"),
455                    });
456                }
457            }
458            if !binaries.exposed.is_empty() {
459                let cmd = &t.entrypoint.command;
460                let basename = cmd.rfind('/').map(|i| &cmd[i + 1..]).unwrap_or(cmd);
461                if !binaries.exposed.iter().any(|b| b == basename) {
462                    errors.push(ValidationError {
463                        field: "tool.binaries.exposed".into(),
464                        message: format!(
465                            "entrypoint command '{basename}' must be listed in exposed"
466                        ),
467                    });
468                }
469            }
470        }
471
472        if errors.is_empty() {
473            Ok(())
474        } else {
475            Err(errors)
476        }
477    }
478}
479
480#[cfg(test)]
481mod tests {
482    use super::*;
483
484    const SAMPLE: &str = r#"
485[tool]
486id = "bwa"
487version = "0.7.17"
488description = "BWA short-read aligner"
489homepage = "http://bio-bwa.sourceforge.net/"
490license = "GPL-3.0"
491
492[tool.image]
493backend = "docker"
494reference = "biocontainers/bwa:0.7.17--h5bf99c6_8"
495digest = "sha256:abcdef1234567890abcdef1234567890abcdef1234567890abcdef1234567890ab"
496
497[tool.hardware]
498cpu_cores = 8
499ram_gb = 32.0
500disk_gb = 50.0
501
502[tool.hardware.gpu]
503required = false
504
505[[tool.inputs]]
506name = "reads_r1"
507type = "fastq"
508cardinality = "one"
509description = "Forward reads"
510
511[[tool.inputs]]
512name = "reads_r2"
513type = "fastq"
514cardinality = "optional"
515description = "Reverse reads (paired-end)"
516
517[[tool.outputs]]
518name = "alignment"
519type = "bam"
520description = "Aligned reads"
521
522[tool.entrypoint]
523command = "bwa"
524args_template = "mem -t {cpu_cores} {reference} {reads_r1} {reads_r2}"
525
526[tool.entrypoint.env]
527MALLOC_ARENA_MAX = "4"
528"#;
529
530    const SAMPLE_NO_IO: &str = r#"
531[tool]
532id = "mytool"
533version = "1.0.0"
534
535[tool.image]
536backend = "docker"
537reference = "example/mytool:1.0.0"
538
539[tool.hardware]
540
541[tool.entrypoint]
542command = "mytool"
543"#;
544
545    #[test]
546    fn round_trip() {
547        let manifest = Manifest::from_toml_str(SAMPLE).expect("parse failed");
548        assert_eq!(manifest.tool.id, "bwa");
549        assert_eq!(manifest.tool.version, "0.7.17");
550        assert_eq!(manifest.tool.image.backend, "docker");
551        assert_eq!(manifest.tool.inputs.len(), 2);
552        assert_eq!(manifest.tool.outputs.len(), 1);
553        assert_eq!(manifest.tool.inputs[0].cardinality, Cardinality::One);
554        assert_eq!(manifest.tool.inputs[1].cardinality, Cardinality::Optional);
555
556        let serialised = manifest.to_toml_string().expect("serialise failed");
557        let reparsed = Manifest::from_toml_str(&serialised).expect("reparse failed");
558        assert_eq!(reparsed.tool.id, manifest.tool.id);
559        assert_eq!(reparsed.tool.version, manifest.tool.version);
560    }
561
562    #[test]
563    fn no_io_parses_unchanged() {
564        let m = Manifest::from_toml_str(SAMPLE_NO_IO).expect("parse failed");
565        assert!(m.tool.inputs.is_empty());
566        assert!(m.tool.outputs.is_empty());
567        assert!(!m.tool.has_typed_io());
568    }
569
570    #[test]
571    fn typeref_params_parsed() {
572        let s = r#"
573[tool]
574id = "t"
575version = "1.0.0"
576
577[tool.image]
578backend = "docker"
579reference = "example/t:1.0.0"
580
581[tool.hardware]
582
583[[tool.inputs]]
584name = "seqs"
585type = "fasta[protein]"
586cardinality = "one"
587
588[tool.entrypoint]
589command = "t"
590"#;
591        let m = Manifest::from_toml_str(s).unwrap();
592        assert_eq!(m.tool.inputs[0].r#type.params, vec!["protein"]);
593    }
594
595    #[test]
596    fn unknown_type_error() {
597        let s = r#"
598[tool]
599id = "t"
600version = "1.0.0"
601
602[tool.image]
603backend = "docker"
604reference = "example/t:1.0.0"
605
606[tool.hardware]
607
608[[tool.inputs]]
609name = "seqs"
610type = "protien_fasta"
611cardinality = "one"
612
613[tool.entrypoint]
614command = "t"
615"#;
616        let err = Manifest::from_toml_str(s).unwrap_err();
617        let msg = err.to_string();
618        assert!(msg.contains("unknown type"), "got: {msg}");
619    }
620
621    #[test]
622    fn cuda_version_ordering() {
623        let v12_1: CudaVersion = "12.1".parse().unwrap();
624        let v12_4: CudaVersion = "12.4".parse().unwrap();
625        let v13_0: CudaVersion = "13.0".parse().unwrap();
626        assert!(v12_1 < v12_4);
627        assert!(v12_4 < v13_0);
628        assert_eq!(v12_1, "12.1".parse::<CudaVersion>().unwrap());
629    }
630
631    #[test]
632    fn validate_catches_empty_id() {
633        let mut manifest = Manifest::from_toml_str(SAMPLE).unwrap();
634        manifest.tool.id = String::new();
635        let errs = manifest.validate().unwrap_err();
636        assert!(errs.iter().any(|e| e.field == "tool.id"));
637    }
638
639    #[test]
640    fn registry_manifests_parse() {
641        let registry = concat!(env!("CARGO_MANIFEST_DIR"), "/../../bv-registry/tools");
642        let Ok(read) = std::fs::read_dir(registry) else {
643            return;
644        };
645        for entry in read {
646            let tool_dir = entry.unwrap().path();
647            if !tool_dir.is_dir() {
648                continue;
649            }
650            for version_entry in std::fs::read_dir(&tool_dir).unwrap() {
651                let path = version_entry.unwrap().path();
652                if path.extension().is_some_and(|e| e == "toml") {
653                    let s = std::fs::read_to_string(&path)
654                        .unwrap_or_else(|_| panic!("failed to read {}", path.display()));
655                    Manifest::from_toml_str(&s)
656                        .unwrap_or_else(|e| panic!("{}: {e}", path.display()));
657                }
658            }
659        }
660    }
661}
bv_core/manifest.rs

bv_core/
manifest.rs
