patchwork

// Package engine applies Patchwork's publish rules: it turns observed file
// states into snapshots, and decides which snapshots become public
// revisions.
package engine

import (
	"context"
	"encoding/json"
	"errors"
	"fmt"
	"path"
	"strings"
	"time"
	"unicode/utf8"

	"git.pablomurad.com/pablo/patchwork/internal/config"
	"git.pablomurad.com/pablo/patchwork/internal/diff"
	"git.pablomurad.com/pablo/patchwork/internal/doc"
	"git.pablomurad.com/pablo/patchwork/internal/slug"
	"git.pablomurad.com/pablo/patchwork/internal/store"
)

// Action is what ingesting one observation did.
type Action string

const (
	Unchanged Action = "unchanged" // same text as the last snapshot
	Baseline  Action = "baseline"  // first sight of the file: version 1
	Silent    Action = "silent"    // stored as history only
	Published Action = "published" // a new public revision
	Amended   Action = "amended"   // the last revision's note was rewritten
	Folded    Action = "folded"    // a quick follow-up joined the last revision
	Removed   Action = "removed"   // the file is gone
)

// Observation is the content of one file at one moment.
type Observation struct {
	Path  string // relative to the source root, '/'-separated
	Raw   []byte
	Mtime time.Time
	// At overrides the observation time (snapshot --at); zero means now.
	At time.Time
}

// Result reports what happened to one file.
type Result struct {
	Path     string
	Slug     string
	Action   Action
	Number   int // the revision created or changed; 1 for a baseline
	Note     string
	Restored bool // the page had been missing or removed
	Warnings []string
}

// Engine ingests observations into a store.
type Engine struct {
	Store  *store.Store
	Cfg    *config.Config
	Source string
	Now    func() time.Time
	// NewUID makes page uids; tests replace it to get stable feed ids.
	NewUID func() string

	slugRule slug.Rule
	opts     doc.Options
	loc      *time.Location
}

// New builds an engine for one source.
func New(st *store.Store, cfg *config.Config) (*Engine, error) {
	rule, err := slug.Lookup(cfg.Source.Slug)
	if err != nil {
		return nil, err
	}
	return &Engine{
		Store: st, Cfg: cfg, Source: cfg.Source.Mode, Now: time.Now, NewUID: store.NewUID,
		slugRule: rule, opts: cfg.DocOptions(), loc: cfg.Location(),
	}, nil
}

// Ingest stores an observation and applies the publish rules. With dryRun
// nothing is written, but the result says what would have happened.
func (e *Engine) Ingest(ctx context.Context, obs Observation, dryRun bool) (Result, error) {
	var res Result
	err := e.Store.Tx(ctx, dryRun, func(q *store.Queries) error {
		var err error
		res, err = e.ingest(ctx, q, obs)
		return err
	})
	return res, err
}

func (e *Engine) ingest(ctx context.Context, q *store.Queries, obs Observation) (Result, error) {
	at := obs.At
	if at.IsZero() {
		at = e.Now()
	}
	d := doc.Parse(obs.Raw, e.opts)
	res := Result{Path: obs.Path, Note: d.Note}
	res.Warnings = e.check(d, obs)
	kind, maturity := e.enum(d, &res)

	page, err := q.PageByPath(ctx, e.Source, obs.Path)
	if errors.Is(err, store.ErrNotFound) {
		return e.baseline(ctx, q, obs, d, at, kind, maturity, res)
	}
	if err != nil {
		return res, err
	}
	res.Slug = page.Slug

	if page.RemovedAt != nil || page.MissingScans > 0 {
		res.Restored = page.RemovedAt != nil
		page.RemovedAt, page.MissingScans = nil, 0
	}
	page.Slug = e.slugRule(obs.Path)
	// The fallback stays whatever the baseline settled on: later uploads
	// change the mtime, not the day the page was planted.
	page.OriginalDate = e.originalDate(d, obs.Path, page.OriginalDate)
	res.Slug = page.Slug
	if err := q.UpdatePage(ctx, page); err != nil {
		return res, err
	}

	prev, err := q.LastSnapshot(ctx, page.ID)
	if err != nil {
		return res, err
	}
	if prev.RawHash == d.RawHash {
		res.Action = Unchanged
		return res, nil
	}
	snap := e.snapshot(page.ID, obs, d, at, kind, maturity)
	if err := q.InsertSnapshot(ctx, snap); err != nil {
		return res, err
	}

	last, err := q.LastRevision(ctx, page.ID)
	if err != nil {
		return res, fmt.Errorf("%s has no revision: %w", obs.Path, err)
	}
	lastSnap, err := q.Snapshot(ctx, last.SnapshotID)
	if err != nil {
		return res, err
	}

	noteNew := d.Note != "" && d.Note != last.Note
	sameContent := d.ContentHash == lastSnap.ContentHash && maturity == last.Maturity
	fold := e.Cfg.Publish.FoldWindow.Duration

	switch {
	case noteNew && sameContent && last.Baseline():
		res.Action = Silent
		res.Warnings = append(res.Warnings, "new revision note but no visible change since the first version; nothing published")

	case noteNew && sameContent:
		// Only the note changed: the author is fixing the note.
		last.Note, last.Kind, last.SnapshotID, last.UpdatedAt = d.Note, kind, snap.ID, at
		if err := q.UpdateRevision(ctx, last); err != nil {
			return res, err
		}
		res.Action, res.Number = Amended, last.Number

	case noteNew:
		n, err := q.MaxRevisionNumber(ctx, page.ID)
		if err != nil {
			return res, err
		}
		rev := &store.Revision{
			PageID: page.ID, Number: n + 1, SnapshotID: snap.ID,
			Note: d.Note, Kind: kind, Maturity: maturity, PublishedAt: at, UpdatedAt: at,
		}
		e.setDiff(rev, lastSnap, d)
		if err := q.InsertRevision(ctx, rev); err != nil {
			return res, err
		}
		res.Action, res.Number = Published, rev.Number

	case d.Note != "" && d.Note == last.Note && !last.Baseline() && fold > 0 && at.Sub(last.PublishedAt) <= fold:
		// Same note, shortly after publishing: the typo spotted right
		// after the upload belongs to that revision.
		base, err := q.RevisionBefore(ctx, page.ID, last.Number)
		if err != nil {
			return res, err
		}
		baseSnap, err := q.Snapshot(ctx, base.SnapshotID)
		if err != nil {
			return res, err
		}
		last.SnapshotID, last.Kind, last.Maturity, last.UpdatedAt = snap.ID, kind, maturity, at
		e.setDiff(last, baseSnap, d)
		if err := q.UpdateRevision(ctx, last); err != nil {
			return res, err
		}
		res.Action, res.Number = Folded, last.Number

	default:
		res.Action = Silent
	}
	return res, nil
}

func (e *Engine) baseline(ctx context.Context, q *store.Queries, obs Observation, d *doc.Doc, at time.Time, kind, maturity string, res Result) (Result, error) {
	page := &store.Page{
		UID: e.NewUID(), Source: e.Source, Path: obs.Path,
		Slug: e.slugRule(obs.Path), FirstSeenAt: at,
	}
	page.OriginalDate = e.originalDate(d, obs.Path, e.baselineDate(obs, at))
	if page.Slug == "" {
		res.Warnings = append(res.Warnings, "the file name produces an empty slug; the page has no URL")
	}
	if err := q.InsertPage(ctx, page); err != nil {
		return res, err
	}
	snap := e.snapshot(page.ID, obs, d, at, kind, maturity)
	if err := q.InsertSnapshot(ctx, snap); err != nil {
		return res, err
	}
	// The baseline keeps the note already in the file, unpublished, so a
	// later upload that still carries it doesn't announce an old note.
	rev := &store.Revision{
		PageID: page.ID, Number: 1, SnapshotID: snap.ID,
		Note: d.Note, Kind: kind, Maturity: maturity, PublishedAt: at, UpdatedAt: at,
	}
	if err := q.InsertRevision(ctx, rev); err != nil {
		return res, err
	}
	res.Slug, res.Action, res.Number = page.Slug, Baseline, 1
	return res, nil
}

func (e *Engine) snapshot(pageID int64, obs Observation, d *doc.Doc, at time.Time, kind, maturity string) *store.Snapshot {
	fm, err := json.Marshal(d.Frontmatter)
	if err != nil {
		fm = []byte("{}")
	}
	s := &store.Snapshot{
		PageID: pageID, ObservedAt: at, RawHash: d.RawHash, ContentHash: d.ContentHash,
		Raw: d.Text, FrontmatterJSON: string(fm), Title: d.Title, Note: d.Note, Kind: kind, Maturity: maturity,
	}
	if !obs.Mtime.IsZero() {
		m := obs.Mtime
		s.SourceMtime = &m
	}
	return s
}

func (e *Engine) setDiff(rev *store.Revision, base *store.Snapshot, cur *doc.Doc) {
	full := diff.Compare(doc.Parse([]byte(base.Raw), e.opts), cur, e.opts.DiffKeys)
	b, err := json.Marshal(full)
	if err != nil {
		panic(err) // the diff types always marshal
	}
	rev.DiffJSON, rev.DiffVersion = string(b), diff.Version
	rev.WordsAdded, rev.WordsRemoved = full.Stats.WordsAdded, full.Stats.WordsRemoved
}

func (e *Engine) check(d *doc.Doc, obs Observation) []string {
	var w []string
	if !utf8.Valid(obs.Raw) {
		w = append(w, "the file is not valid UTF-8")
	}
	if d.FrontmatterErr != nil {
		w = append(w, "frontmatter is not valid YAML, so no revision note is read: "+firstLine(d.FrontmatterErr.Error()))
	}
	return w
}

// enum keeps only declared values of revision_kind and maturity; anything
// else is dropped with a warning rather than published.
func (e *Engine) enum(d *doc.Doc, res *Result) (kind, maturity string) {
	kind, maturity = d.Kind, d.Maturity
	if !doc.Valid(kind, doc.Kinds) {
		res.Warnings = append(res.Warnings, fmt.Sprintf("%s %q is not one of %s; ignored", e.opts.KindKey, kind, strings.Join(doc.Kinds, ", ")))
		kind = ""
	}
	if !doc.Valid(maturity, doc.Maturities) {
		res.Warnings = append(res.Warnings, fmt.Sprintf("%s %q is not one of %s; ignored", e.opts.MaturityKey, maturity, strings.Join(doc.Maturities, ", ")))
		maturity = ""
	}
	return kind, maturity
}

// originalDate is the page's first publication date: frontmatter first,
// then a date at the start of the file name, then fallback.
func (e *Engine) originalDate(d *doc.Doc, relPath string, fallback string) string {
	switch v := d.Frontmatter[e.Cfg.Frontmatter.DateKey].(type) {
	case time.Time:
		// YAML reads `date: 2026-06-21` as midnight UTC; that is a calendar
		// date, not an instant, so it must not shift into the site's zone.
		if v.Hour() == 0 && v.Minute() == 0 && v.Second() == 0 && v.Location() == time.UTC {
			return v.Format("2006-01-02")
		}
		return v.In(e.loc).Format("2006-01-02")
	case string:
		s := strings.TrimSpace(v)
		if t, err := time.Parse("2006-01-02", s); err == nil {
			return t.Format("2006-01-02")
		}
		if t, err := time.Parse(time.RFC3339, s); err == nil {
			return t.In(e.loc).Format("2006-01-02")
		}
	}
	if date, ok := slug.DatePrefix(path.Base(relPath)); ok {
		return date
	}
	return fallback
}

// baselineDate is the fallback original date of a page seen for the first
// time: the file's modification time when it's older than the observation.
// When Patchwork starts on an existing site, that is when each page was
// written or uploaded; the first scan's own date would give every undated
// page the day of the deploy.
func (e *Engine) baselineDate(obs Observation, at time.Time) string {
	t := at
	if !obs.Mtime.IsZero() && obs.Mtime.Before(at) {
		t = obs.Mtime
	}
	return t.In(e.loc).Format("2006-01-02")
}

func firstLine(s string) string {
	line, _, _ := strings.Cut(s, "\n")
	return line
}