job-search/internal/mock/findatender/store.go

package mock

import (
	"bufio"
	"bytes"
	"encoding/json"
	"errors"
	"fmt"
	"log/slog"
	"math/rand"
	"os"
	"path/filepath"
	"sort"
	"strconv"
	"strings"
	"sync"
	"time"

	"git.avsdev.uk/Go/job-search/internal/mock/findatender/ocds"
)

// Store keeps every release in memory and appends each new one to a JSON Lines
// file, so releases survive restarts. Releases are kept in date order.
type Store struct {
	mu         sync.RWMutex
	file       *os.File
	releases   []ocds.Release
	times      []time.Time
	latest     map[string]int // ocid -> index of its most recent release
	lastNumber map[int]int    // year -> highest notice number used

	// Procedures not yet awarded, as a slice plus positions so that adding,
	// removing and picking one at random are all constant time.
	open         []string
	openPosition map[string]int

	opts options
}

// OpenStore loads the releases stored at path, creating the file and its
// directory if needed, and opens it for appending. Lines that can't be read
// are skipped, releases are put back in date order if needed, and a record
// left incomplete by a crash is ended so new ones start on a fresh line.
// It logs to slog.Default() unless given WithLogger.
func OpenStore(path string, opts ...Option) (*Store, error) {
	store := &Store{
		latest:       map[string]int{},
		lastNumber:   map[int]int{},
		openPosition: map[string]int{},
		opts:         applyOptions(opts),
	}

	if err := os.MkdirAll(filepath.Dir(path), 0o750); err != nil {
		return nil, err
	}

	releases, err := readReleases(store.opts.logger("store"), path)
	if err != nil {
		return nil, err
	}

	// The file is written in order, but sort anyway in case it was hand-edited.
	sort.SliceStable(releases, func(i, other int) bool {
		return releases[i].Date.Before(releases[other].Date)
	})
	for _, release := range releases {
		store.releases = append(store.releases, release)
		store.times = append(store.times, release.Date)
	}
	for i, release := range store.releases {
		store.index(i, release)
	}

	if store.file, err = openForAppend(path); err != nil {
		return nil, err
	}

	return store, nil
}

// readReleases reads the releases stored at path, skipping blank lines and
// lines that can't be decoded (which are logged to logger). A missing file
// holds no releases.
func readReleases(logger *slog.Logger, path string) ([]ocds.Release, error) {
	input, err := os.Open(filepath.Clean(path))
	if os.IsNotExist(err) {
		return nil, nil
	}
	if err != nil {
		return nil, err
	}
	defer input.Close() //nolint:errcheck // read only

	var releases []ocds.Release
	scanner := bufio.NewScanner(input)
	scanner.Buffer(make([]byte, 1<<20), 64<<20)
	for line := 1; scanner.Scan(); line++ {
		data := bytes.TrimSpace(scanner.Bytes())
		if len(data) == 0 {
			continue
		}
		var release ocds.Release
		if err := json.Unmarshal(data, &release); err != nil {
			logger.Warn("skipping unreadable line", "line", line, "error", err)
			continue
		}
		releases = append(releases, release)
	}
	if err := scanner.Err(); err != nil {
		return nil, fmt.Errorf("reading %s: %w", path, err)
	}

	return releases, nil
}

// openForAppend opens path for appending releases. If a previous run died
// mid-write, it ends the partial record so the next one starts on a fresh
// line.
func openForAppend(path string) (*os.File, error) {
	file, err := os.OpenFile(
		filepath.Clean(path), os.O_CREATE|os.O_RDWR|os.O_APPEND, 0o600,
	)
	if err != nil {
		return nil, err
	}

	if info, err := file.Stat(); err == nil && info.Size() > 0 {
		last := make([]byte, 1)
		_, readErr := file.ReadAt(last, info.Size()-1)
		if readErr == nil && last[0] != '\n' {
			if _, err := file.Write([]byte("\n")); err != nil {
				return nil, errors.Join(
					fmt.Errorf("repairing %s: %w", path, err), file.Close(),
				)
			}
		}
	}

	return file, nil
}

// index records release, at position i, in the lookups: the ocid's latest
// release, whether its procedure is still open (an award closes it), and the
// highest notice number used each year.
func (s *Store) index(i int, release ocds.Release) {
	s.latest[release.OCID] = i

	if len(release.Tag) > 0 && release.Tag[0] == "award" {
		s.closeProcedure(release.OCID)
	} else if _, ok := s.openPosition[release.OCID]; !ok {
		s.openPosition[release.OCID] = len(s.open)
		s.open = append(s.open, release.OCID)
	}

	if numberText, year, ok := strings.Cut(release.ID, "-"); ok {
		number, numberErr := strconv.Atoi(numberText)
		yearNumber, yearErr := strconv.Atoi(year)
		if numberErr == nil && yearErr == nil && number > s.lastNumber[yearNumber] {
			s.lastNumber[yearNumber] = number
		}
	}
}

// closeProcedure removes ocid from the open procedures, if it is there.
func (s *Store) closeProcedure(ocid string) {
	position, ok := s.openPosition[ocid]
	if !ok {
		return
	}

	last := len(s.open) - 1
	s.open[position] = s.open[last]
	s.openPosition[s.open[position]] = position
	s.open = s.open[:last]
	delete(s.openPosition, ocid)
}

// Append persists a release and flushes it to disk.
func (s *Store) Append(release ocds.Release) error {
	return s.append(release, true)
}

// AppendNoSync persists a release without an fsync; call Sync afterwards.
// Used for bulk catch-up, where a sync per record would be very slow.
func (s *Store) AppendNoSync(release ocds.Release) error {
	return s.append(release, false)
}

// Sync flushes appended releases to disk. Call it after AppendNoSync.
func (s *Store) Sync() error {
	s.mu.Lock()
	defer s.mu.Unlock()
	return s.file.Sync()
}

// append stores release, syncing the file to disk if sync is true. A date
// earlier than the newest stored release (for example after a clock change) is
// moved up so the order stays valid.
func (s *Store) append(release ocds.Release, sync bool) error {
	s.mu.Lock()
	defer s.mu.Unlock()

	release.Date = release.Date.UTC()
	if count := len(s.times); count > 0 && release.Date.Before(s.times[count-1]) {
		release.Date = s.times[count-1]
	}

	data, err := json.Marshal(release)
	if err != nil {
		return err
	}
	if _, err := s.file.Write(append(data, '\n')); err != nil {
		return err
	}
	if sync {
		if err := s.file.Sync(); err != nil {
			return err
		}
	}

	s.releases = append(s.releases, release)
	s.times = append(s.times, release.Date)
	s.index(len(s.releases)-1, release)

	return nil
}

// Page returns up to limit releases dated in [from, upper], starting no earlier
// than position start, plus the position to continue from and whether more
// exist.
func (s *Store) Page(
	from, upper time.Time,
	start, limit int,
) ([]ocds.Release, int, bool) {
	s.mu.RLock()
	defer s.mu.RUnlock()

	i := sort.Search(len(s.times), func(i int) bool {
		return !s.times[i].Before(from)
	})
	if start > i {
		i = start
	}

	out := make([]ocds.Release, 0, limit)
	for i < len(s.releases) && len(out) < limit && !s.times[i].After(upper) {
		out = append(out, s.releases[i])
		i++
	}

	more := i < len(s.releases) && !s.times[i].After(upper)

	return out, i, more
}

// Last returns the date of the newest release, or false if the store is
// empty.
func (s *Store) Last() (time.Time, bool) {
	s.mu.RLock()
	defer s.mu.RUnlock()
	if len(s.times) == 0 {
		return time.Time{}, false
	}
	return s.times[len(s.times)-1], true
}

// Len returns the number of stored releases.
func (s *Store) Len() int {
	s.mu.RLock()
	defer s.mu.RUnlock()
	return len(s.releases)
}

// HasOCID reports whether any release of the procedure ocid is stored.
func (s *Store) HasOCID(ocid string) bool {
	s.mu.RLock()
	defer s.mu.RUnlock()
	_, ok := s.latest[ocid]
	return ok
}

// NextNoticeID returns the next FTS-style notice id ("000123-2026") for year.
func (s *Store) NextNoticeID(year int) string {
	s.mu.RLock()
	defer s.mu.RUnlock()
	return fmt.Sprintf("%06d-%d", s.lastNumber[year]+1, year)
}

// RandomOpen picks a procedure that has not been awarded and whose latest
// release is after since. Procedures idle since before then are dropped from
// the open set, as if they had been abandoned.
func (s *Store) RandomOpen(
	rng *rand.Rand, since time.Time,
) (ocds.Release, bool) {
	s.mu.Lock()
	defer s.mu.Unlock()
	for len(s.open) > 0 {
		ocid := s.open[rng.Intn(len(s.open))]
		i := s.latest[ocid]
		if s.times[i].After(since) {
			return s.releases[i], true
		}
		s.closeProcedure(ocid)
	}
	return ocds.Release{}, false
}

// Close closes the store's file. The store must not be used afterwards.
func (s *Store) Close() error {
	s.mu.Lock()
	defer s.mu.Unlock()
	return s.file.Close()
}