exo/artifacts/yaml.go

package artifacts

import (
	"fmt"
	"os"
	"time"

	"git.wntrmute.dev/kyle/exo/blob"
	"git.wntrmute.dev/kyle/exo/core"
	"git.wntrmute.dev/kyle/exo/db"
	"gopkg.in/yaml.v3"
)

// MetadataYAML is the YAML representation of metadata entries.
type MetadataYAML []struct {
	Key      string `yaml:"key"`
	Contents string `yaml:"contents"`
	Type     string `yaml:"type"`
}

// ToStd converts MetadataYAML to core.Metadata.
func (my MetadataYAML) ToStd() core.Metadata {
	if my == nil {
		return core.Metadata{}
	}
	metadata := core.Metadata{}
	for _, entry := range my {
		metadata[entry.Key] = core.Value{Contents: entry.Contents, Type: entry.Type}
	}
	return metadata
}

// CitationYAML is the YAML representation of a citation.
type CitationYAML struct {
	ID        string       `yaml:"id"`
	DOI       string       `yaml:"doi"`
	Title     string       `yaml:"title"`
	Year      int          `yaml:"year"`
	Published string       `yaml:"published"`
	Authors   []string     `yaml:"authors"`
	Publisher *Publisher   `yaml:"publisher"`
	Source    string       `yaml:"source"`
	Abstract  string       `yaml:"abstract"`
	Metadata  MetadataYAML `yaml:"metadata"`
}

// ToStd converts a CitationYAML to a Citation.
func (cy *CitationYAML) ToStd() (*Citation, error) {
	if cy == nil {
		return nil, nil
	}

	cite := &Citation{
		ID:        cy.ID,
		DOI:       cy.DOI,
		Title:     cy.Title,
		Year:      cy.Year,
		Authors:   cy.Authors,
		Publisher: cy.Publisher,
		Source:    cy.Source,
		Abstract:  cy.Abstract,
		Metadata:  cy.Metadata.ToStd(),
	}

	if cy.Published != "" {
		var err error
		cite.Published, err = db.FromDBTime(cy.Published, nil)
		if err != nil {
			return nil, fmt.Errorf("artifacts: failed to parse citation published date: %w", err)
		}
	}

	return cite, nil
}

// BlobHeaderYAML is the YAML representation of a blob reference.
type BlobHeaderYAML struct {
	Format string `yaml:"format"`
	Path   string `yaml:"path"`
}

// SnapshotYAML is the YAML representation of a snapshot.
type SnapshotYAML struct {
	ID        string           `yaml:"id"`
	StoreDate int64            `yaml:"stored"`
	Datetime  string           `yaml:"datetime"`
	Citation  *CitationYAML    `yaml:"citation"`
	Source    string           `yaml:"source"`
	Blobs     []BlobHeaderYAML `yaml:"blobs"`
	Metadata  MetadataYAML     `yaml:"metadata"`
}

// ToStd converts a SnapshotYAML to a Snapshot, reading blob data from files.
func (syml SnapshotYAML) ToStd(artifactID string, parentCitation *Citation) (*Snapshot, error) {
	cite, err := syml.Citation.ToStd()
	if err != nil {
		return nil, err
	}

	snap := &Snapshot{
		ArtifactID: artifactID,
		ID:         syml.ID,
		StoreDate:  time.Unix(syml.StoreDate, 0),
		Citation:   cite,
		Source:     syml.Source,
		Blobs:      map[MIME]*BlobRef{},
		Metadata:   syml.Metadata.ToStd(),
	}

	snap.Datetime, err = db.FromDBTime(syml.Datetime, nil)
	if err != nil {
		return nil, err
	}

	// Inherit from parent citation if snapshot citation is nil or partial.
	if snap.Citation == nil {
		snap.Citation = parentCitation
	} else if parentCitation != nil {
		snap.Citation.Update(parentCitation)
	}

	for _, bh := range syml.Blobs {
		data, err := os.ReadFile(bh.Path)
		if err != nil {
			return nil, fmt.Errorf("artifacts: failed to read blob file %q: %w", bh.Path, err)
		}
		id := blob.HashData(data)
		snap.Blobs[MIME(bh.Format)] = &BlobRef{
			SnapshotID: syml.ID,
			ID:         id,
			Format:     MIME(bh.Format),
			Data:       data,
		}
	}

	return snap, nil
}

// ArtifactYAML is the YAML representation of a complete artifact with snapshots.
type ArtifactYAML struct {
	ID         string            `yaml:"id"`
	Type       string            `yaml:"type"`
	Citation   *CitationYAML     `yaml:"citation"`
	Latest     string            `yaml:"latest"`
	History    map[string]string `yaml:"history"`
	Tags       []string          `yaml:"tags"`
	Categories []string          `yaml:"categories"`
	Metadata   MetadataYAML      `yaml:"metadata"`
	Snapshots  []SnapshotYAML    `yaml:"snapshots"`
}

// ToStd converts an ArtifactYAML to an Artifact and its Snapshots.
func (ayml *ArtifactYAML) ToStd() (*Artifact, []*Snapshot, error) {
	cite, err := ayml.Citation.ToStd()
	if err != nil {
		return nil, nil, err
	}

	art := &Artifact{
		ID:         ayml.ID,
		Type:       ArtifactType(ayml.Type),
		Citation:   cite,
		History:    map[time.Time]string{},
		Tags:       core.MapFromList(ayml.Tags),
		Categories: core.MapFromList(ayml.Categories),
		Metadata:   ayml.Metadata.ToStd(),
	}

	if ayml.Latest != "" {
		art.Latest, err = db.FromDBTime(ayml.Latest, nil)
		if err != nil {
			return nil, nil, err
		}
	}

	for timestamp, id := range ayml.History {
		datetime, err := db.FromDBTime(timestamp, nil)
		if err != nil {
			return nil, nil, err
		}
		art.History[datetime] = id
	}

	var snaps []*Snapshot
	for _, syml := range ayml.Snapshots {
		snap, err := syml.ToStd(ayml.ID, art.Citation)
		if err != nil {
			return nil, nil, err
		}
		snaps = append(snaps, snap)
	}

	return art, snaps, nil
}

// LoadArtifactFromYAML reads and parses an artifact YAML file.
func LoadArtifactFromYAML(path string) (*ArtifactYAML, error) {
	data, err := os.ReadFile(path) //nolint:gosec // path is a user-provided file for import
	if err != nil {
		return nil, fmt.Errorf("artifacts: failed to read YAML file %q: %w", path, err)
	}

	ay := &ArtifactYAML{}
	if err := yaml.Unmarshal(data, ay); err != nil {
		return nil, fmt.Errorf("artifacts: failed to parse YAML file %q: %w", path, err)
	}

	return ay, nil
}