path: root/internal/liturgy/store.go

package liturgy

import (
	"fmt"
	"os"
	"path/filepath"
	"sort"
	"strings"
	"time"
)

// harvestRetries is how many times Harvest attempts a single date's fetch
// before treating it as a genuine network failure rather than transient
// hiccup; harvestRetryDelay is the backoff slept between attempts. Both are
// package vars so tests can shrink the delay instead of waiting on it.
var (
	harvestRetries    = 3
	harvestRetryDelay = 2 * time.Second
)

// siglaRow is one line of the sigla TSV: a date's section label and the
// scripture citation extracted from its heading.
type siglaRow struct {
	label, citation string
}

// siglaPath is the persistent sigla store written by Harvest and read by
// LoadOffline: ${XDG_DATA_HOME:-~/.local/share}/lectio/sigla.tsv
func siglaPath() string {
	base := os.Getenv("XDG_DATA_HOME")
	if base == "" {
		home, err := os.UserHomeDir()
		if err != nil {
			home = "."
		}
		base = filepath.Join(home, ".local", "share")
	}
	return filepath.Join(base, "lectio", "sigla.tsv")
}

// sectionLabel derives the short label Harvest stores alongside a citation
// from a section's full heading, e.g. "Ewangelia (J 20, 1. 11-18)" ->
// "Ewangelia". It strips the same trailing "(...)" citation ExtractCitation
// reads, so the two stay in sync.
func sectionLabel(heading string) string {
	loc := citationRe.FindStringIndex(heading)
	if loc == nil {
		return strings.TrimSpace(heading)
	}
	return strings.TrimSpace(heading[:loc[0]])
}

// readSigla loads the sigla TSV into date -> rows. A missing file is not an
// error -- it just means nothing has been harvested yet.
func readSigla(path string) (map[string][]siglaRow, error) {
	data, err := os.ReadFile(path)
	if err != nil {
		if os.IsNotExist(err) {
			return map[string][]siglaRow{}, nil
		}
		return nil, err
	}
	rows := map[string][]siglaRow{}
	for _, line := range strings.Split(string(data), "\n") {
		if line == "" {
			continue
		}
		fields := strings.SplitN(line, "\t", 3)
		if len(fields) != 3 {
			continue
		}
		date := fields[0]
		rows[date] = append(rows[date], siglaRow{label: fields[1], citation: fields[2]})
	}
	return rows, nil
}

// writeSigla writes date -> rows back out as the sigla TSV, sorted by date
// for a deterministic, diffable file.
func writeSigla(path string, byDate map[string][]siglaRow) error {
	if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
		return err
	}
	dates := make([]string, 0, len(byDate))
	for d := range byDate {
		dates = append(dates, d)
	}
	sort.Strings(dates)

	var b strings.Builder
	for _, date := range dates {
		for _, r := range byDate[date] {
			fmt.Fprintf(&b, "%s\t%s\t%s\n", date, r.label, r.citation)
		}
	}
	return os.WriteFile(path, []byte(b.String()), 0o644)
}

// Harvest walks dates forward from fromDate, fetching and parsing each day's
// page and recording every section's citation to the sigla TSV, for up to
// maxDays days (0 = walk until the unpublished horizon). It stops cleanly
// (nil error) the first time a date fails to PARSE -- niedziela.pl's
// "Przykro nam" placeholder (or any other parse failure) marks the horizon
// the site hasn't published past yet, not an error to report. A date that
// fails to FETCH, by contrast, is a transient network problem, not the
// horizon: Harvest retries it a few times (harvestRetries, backing off
// harvestRetryDelay between attempts) and, if it still fails, stops and
// returns an error -- but only after saving whatever was harvested up to
// that point, so the caller never loses progress to a blip.
//
// Re-harvesting a date replaces its rows in the TSV rather than duplicating
// them, so running Harvest again over an already-harvested range is safe.
// It also warms the HTML/JSON cache for every date it successfully harvests.
//
// It returns how many days were harvested and the furthest (most recent)
// date reached, alongside any fetch error (nil on a clean parse-horizon
// stop or on reaching maxDays).
func Harvest(fromDate string, maxDays int) (added int, furthest string, err error) {
	start, err := time.Parse("2006-01-02", fromDate)
	if err != nil {
		return 0, "", fmt.Errorf("invalid date %q: %w", fromDate, err)
	}

	path := siglaPath()
	byDate, err := readSigla(path)
	if err != nil {
		return 0, "", err
	}

	dir := cacheDir()
	day := start
	var harvestErr error
	for i := 0; maxDays == 0 || i < maxDays; i++ {
		dateStr := day.Format("2006-01-02")

		var page string
		var ferr error
		for attempt := 1; attempt <= harvestRetries; attempt++ {
			page, ferr = fetch(dateStr)
			if ferr == nil {
				break
			}
			if attempt < harvestRetries {
				time.Sleep(harvestRetryDelay)
			}
		}
		if ferr != nil {
			// A genuine network/transport error, not the horizon: don't
			// silently stop as if the site simply hadn't published this
			// date yet. Record it and stop walking, but writeSigla below
			// still runs so progress made so far isn't lost.
			harvestErr = fmt.Errorf("harvest interrupted at %s: %w", dateStr, ferr)
			break
		}
		secs, perr := Parse(page)
		if perr != nil {
			break // unpublished horizon (or unparsable page): stop walking, cleanly
		}

		var rows []siglaRow
		for _, s := range secs {
			citation, cerr := ExtractCitation(s.Heading)
			if cerr != nil {
				continue
			}
			rows = append(rows, siglaRow{label: sectionLabel(s.Heading), citation: citation})
		}
		byDate[dateStr] = rows

		if publishedRe.MatchString(page) {
			if mkErr := os.MkdirAll(dir, 0o755); mkErr == nil {
				_ = os.WriteFile(filepath.Join(dir, dateStr+".html"), []byte(page), 0o644)
				writeJSONCache(filepath.Join(dir, dateStr+".json"), secs)
			}
		}

		added++
		furthest = dateStr
		day = day.AddDate(0, 0, 1)
	}

	werr := writeSigla(path, byDate)
	if harvestErr != nil {
		return added, furthest, harvestErr
	}
	if werr != nil {
		return added, furthest, werr
	}
	return added, furthest, nil
}

// LoadOffline builds a day's sections purely from the harvested sigla TSV:
// Heading is the stored section label, Citation the stored citation, and
// Paragraphs empty (no reading text is harvested, only the scripture
// reference). It errors clearly if the date has not been harvested.
func LoadOffline(date string) ([]Section, error) {
	byDate, err := readSigla(siglaPath())
	if err != nil {
		return nil, err
	}
	rows, ok := byDate[date]
	if !ok || len(rows) == 0 {
		return nil, fmt.Errorf("%s not harvested; run 'lectio update' while online first", date)
	}
	secs := make([]Section, 0, len(rows))
	czytanieCount := 0
	for _, r := range rows {
		secs = append(secs, Section{
			Heading:  r.label,
			Citation: r.citation,
			PartID:   partID(r.label, &czytanieCount),
		})
	}
	return secs, nil
}