path: root/internal/liturgy/fetch.go

package liturgy

import (
	"encoding/json"
	"fmt"
	"io"
	"net/http"
	"os"
	"path/filepath"
	"regexp"
	"time"
)

// baseURL is the niedziela.pl URL template ("%s" is the date, YYYY-MM-DD).
// It is a package var so tests can point it at an httptest server.
var baseURL = "https://niezbednik.niedziela.pl/liturgia/%s/Ewangelia"

// SetBaseURL overrides the fetch URL template used by Load. It exists so
// tests in other packages (e.g. internal/readings) can point Load at an
// httptest server; production code must never call it.
func SetBaseURL(url string) {
	baseURL = url
}

// userAgent is sent on every fetch; the site serves a different (broken)
// page to non-browser clients without it.
const userAgent = "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 " +
	"(KHTML, like Gecko) Chrome/124.0 Safari/537.36"

// publishedRe matches the tab-pane id niedziela.pl gives a fully published
// day's page (e.g. "tabnowy0all"). Its absence -- together with a "Przykro
// nam" placeholder -- marks a date that has not been published yet; see
// ewangelia.py's fetch() for the original behaviour this mirrors.
var publishedRe = regexp.MustCompile(`id="\w*0all"`)

// dateRe is the same YYYY-MM-DD shape internal/cli's dateRe validates
// against. Load checks opts.Date against it before building any filesystem
// path (jsonPath/htmlPath below are built by string concatenation, so an
// unvalidated Date is a path-traversal vector) -- defense-in-depth so every
// caller (web, cli, tui) is protected even if a future caller forgets to
// validate its own input first.
var dateRe = regexp.MustCompile(`^\d{4}-\d{2}-\d{2}$`)

// cacheFileRe matches the date-prefixed cache file names Load/Harvest/tradlit
// write into CacheDir(): "<YYYY-MM-DD>.html", "<YYYY-MM-DD>.json", and
// "<YYYY-MM-DD>.trad.<lang>.json" (the traditional-lectionary cache; see
// internal/tradlit). CleanCache uses it to tell cache entries apart from
// anything else that might be sitting in the directory, and to recover the
// date (group 1) for the age check regardless of which cache file it is.
var cacheFileRe = regexp.MustCompile(`^(\d{4}-\d{2}-\d{2})\.[a-z0-9.]+$`)

// Options controls how Load resolves a day's readings.
type Options struct {
	// Date is the day to load, formatted YYYY-MM-DD.
	Date string
	// Refresh bypasses both cache layers and re-fetches from the network.
	Refresh bool
	// Offline restricts Load to previously cached/harvested data (see
	// LoadOffline), never hitting the network.
	Offline bool
}

// CacheDir is where the HTML/JSON cache layers live:
// ${XDG_CACHE_HOME:-~/.cache}/lectio/
// Exported so internal/tradlit shares the same cache root for the
// traditional lectionary's propers.
func CacheDir() string {
	base := os.Getenv("XDG_CACHE_HOME")
	if base == "" {
		home, err := os.UserHomeDir()
		if err != nil {
			home = "."
		}
		base = filepath.Join(home, ".cache")
	}
	return filepath.Join(base, "lectio")
}

// Load returns the day's reading sections, from the modern (niedziela.pl)
// source, preferring cache over network:
//
//  0. If Offline, skip straight to LoadOffline (the harvested sigla TSV).
//  1. Unless Refresh, the parsed JSON cache ({date}.json).
//  2. Unless Refresh, the raw HTML cache ({date}.html) -- parsed, and the
//     result written to the JSON cache for next time.
//  3. Otherwise, fetch the page over the network, parse it, and -- only if
//     the page is fully published -- write both cache layers. A fetch error
//     here (e.g. no network) falls back to LoadOffline for this date if it
//     has been harvested, and only surfaces the original fetch error if
//     that fallback also fails.
func Load(opts Options) ([]Section, error) {
	if !dateRe.MatchString(opts.Date) {
		return nil, fmt.Errorf("invalid date %q: want YYYY-MM-DD", opts.Date)
	}

	if opts.Offline {
		return LoadOffline(opts.Date)
	}

	dir := CacheDir()
	jsonPath := filepath.Join(dir, opts.Date+".json")
	htmlPath := filepath.Join(dir, opts.Date+".html")

	if !opts.Refresh {
		if secs, err := loadJSONCache(jsonPath); err == nil {
			return secs, nil
		}
		if page, err := os.ReadFile(htmlPath); err == nil {
			secs, err := Parse(string(page))
			if err != nil {
				return nil, err
			}
			writeJSONCache(jsonPath, secs)
			return secs, nil
		}
	}

	page, err := fetch(opts.Date)
	if err != nil {
		// Network is unreachable: fall back to a prior harvest of this date
		// if there is one, rather than failing outright.
		if secs, offErr := LoadOffline(opts.Date); offErr == nil {
			return secs, nil
		}
		return nil, err
	}

	// Parse itself reports an unpublished date as "no reading published for
	// this date yet" (via the "Przykro nam" placeholder), so an unpublished
	// page is neither cached nor returned as sections here.
	secs, err := Parse(page)
	if err != nil {
		return nil, err
	}

	// Cache only fully-published pages, so an as-yet-unpublished future date
	// keeps being retried instead of caching a "no reading" placeholder.
	if publishedRe.MatchString(page) {
		if err := os.MkdirAll(dir, 0o755); err == nil {
			_ = os.WriteFile(htmlPath, []byte(page), 0o644)
			writeJSONCache(jsonPath, secs)
		}
	}

	return secs, nil
}

// CleanCache removes cached readings whose date is before `before` from
// CacheDir(). It matches only date-prefixed cache files (see cacheFileRe:
// "<YYYY-MM-DD>.html", ".json", or the traditional lectionary's
// ".trad.<lang>.json"); anything else in the directory (e.g. a stray
// notes.txt, or the sigla store, which lives elsewhere entirely) is left
// alone. A missing cache dir is not an error -- it just means there is
// nothing to clean yet.
func CleanCache(before time.Time) (removed int, freed int64, err error) {
	dir := CacheDir()
	entries, err := os.ReadDir(dir)
	if err != nil {
		if os.IsNotExist(err) {
			return 0, 0, nil
		}
		return 0, 0, err
	}

	for _, entry := range entries {
		if entry.IsDir() {
			continue
		}
		m := cacheFileRe.FindStringSubmatch(entry.Name())
		if m == nil {
			continue
		}
		date, perr := time.Parse("2006-01-02", m[1])
		if perr != nil || !date.Before(before) {
			continue
		}

		path := filepath.Join(dir, entry.Name())
		info, serr := os.Stat(path)
		if serr != nil {
			return removed, freed, serr
		}
		if rerr := os.Remove(path); rerr != nil {
			return removed, freed, rerr
		}
		removed++
		freed += info.Size()
	}
	return removed, freed, nil
}

// loadJSONCache reads and unmarshals the parsed-sections cache file.
func loadJSONCache(path string) ([]Section, error) {
	data, err := os.ReadFile(path)
	if err != nil {
		return nil, err
	}
	var secs []Section
	if err := json.Unmarshal(data, &secs); err != nil {
		return nil, err
	}
	return secs, nil
}

// writeJSONCache best-effort writes the parsed sections cache; a failure to
// cache should never fail the load itself.
func writeJSONCache(path string, secs []Section) {
	data, err := json.Marshal(secs)
	if err != nil {
		return
	}
	_ = os.WriteFile(path, data, 0o644)
}

// fetch GETs the day's page from baseURL with the browser User-Agent.
// Whether the page is actually published is left to the caller (Parse
// detects an unpublished date; Load re-checks the tab id to decide whether
// the result is cache-worthy).
func fetch(dateStr string) (string, error) {
	url := fmt.Sprintf(baseURL, dateStr)
	req, err := http.NewRequest(http.MethodGet, url, nil)
	if err != nil {
		return "", err
	}
	req.Header.Set("User-Agent", userAgent)

	client := &http.Client{Timeout: 20 * time.Second}
	resp, err := client.Do(req)
	if err != nil {
		return "", fmt.Errorf("failed to fetch %s: %w", url, err)
	}
	defer resp.Body.Close()

	body, err := io.ReadAll(resp.Body)
	if err != nil {
		return "", fmt.Errorf("failed to read response from %s: %w", url, err)
	}
	return string(body), nil
}