Improved caching

3 files changed, 297 insertions, 21 deletions

diff --git a/internal/feed/cache.go b/internal/feed/cache.go
new file mode 100644
index 0000000..4f27144
--- /dev/null
+++ b/internal/feed/cache.go
@@ -0,0 +1,230 @@
+package feed
+
+import (
+	"bufio"
+	"crypto/sha256"
+	"encoding/gob"
+	"errors"
+	"fmt"
+	"os"
+	"time"
+
+	"github.com/Necoro/feed2imap-go/internal/log"
+)
+
+const (
+	currentVersion byte   = 1
+	startFeedId    uint64 = 1
+)
+
+type Cache interface {
+	findItem(*Feed) CachedFeed
+	Version() byte
+	transformToCurrent() (Cache, error)
+}
+
+type feedId uint64
+
+type feedDescriptor struct {
+	Name string
+	Url  string
+}
+
+type CachedFeed interface {
+	Checked(withFailure bool)
+	Failures() uint
+}
+
+type cachedFeed struct {
+	LastCheck   time.Time
+	NumFailures uint // can't be named `Failures` b/c it'll collide with the interface
+	Items       []cachedItem
+}
+
+func (cf *cachedFeed) Checked(withFailure bool) {
+	cf.LastCheck = time.Now()
+	if withFailure {
+		cf.NumFailures++
+	} else {
+		cf.NumFailures = 0
+	}
+}
+
+func (cf *cachedFeed) Failures() uint {
+	return cf.NumFailures
+}
+
+type itemHash [sha256.Size]byte
+
+type cachedItem struct {
+	Uid     string
+	Title   string
+	Link    string
+	Date    time.Time
+	Updated time.Time
+	Creator string
+	Hash    itemHash
+}
+
+type v1Cache struct {
+	version byte
+	Ids     map[feedDescriptor]feedId
+	NextId  uint64
+	Feeds   map[feedId]*cachedFeed
+}
+
+func (cache *v1Cache) Version() byte {
+	return cache.version
+}
+
+func New() Cache {
+	cache := v1Cache{
+		Ids:    map[feedDescriptor]feedId{},
+		Feeds:  map[feedId]*cachedFeed{},
+		NextId: startFeedId,
+	}
+	cache.version = currentVersion
+	return &cache
+}
+
+func cacheForVersion(version byte) (Cache, error) {
+	switch version {
+	case 1:
+		return New(), nil
+	default:
+		return nil, fmt.Errorf("unknown cache version '%d'", version)
+	}
+}
+
+func (cache *v1Cache) transformToCurrent() (Cache, error) {
+	return cache, nil
+}
+
+func (cache *v1Cache) getItem(id feedId) CachedFeed {
+	feed, ok := cache.Feeds[id]
+	if !ok {
+		feed = &cachedFeed{}
+		cache.Feeds[id] = feed
+	}
+	return feed
+}
+
+func (cache *v1Cache) findItem(feed *Feed) CachedFeed {
+	if feed.cached != nil {
+		return feed.cached.(*cachedFeed)
+	}
+
+	fId := feedDescriptor{Name: feed.Name, Url: feed.Url}
+	id, ok := cache.Ids[fId]
+	if !ok {
+		var otherId feedDescriptor
+		changed := false
+		for otherId, id = range cache.Ids {
+			if otherId.Name == fId.Name {
+				log.Warnf("Feed %s seems to have changed URLs: New '%s', old '%s'. Updating.",
+					fId.Name, fId.Url, otherId.Url)
+				changed = true
+				break
+			} else if otherId.Url == fId.Url {
+				log.Warnf("Feed with URL '%s' seems to have changed its name: New '%s', old '%s'. Updating",
+					fId.Url, fId.Name, otherId.Name)
+				changed = true
+				break
+			}
+		}
+		if changed {
+			delete(cache.Ids, otherId)
+		} else {
+			id = feedId(cache.NextId)
+			cache.NextId++
+		}
+
+		cache.Ids[fId] = id
+	}
+
+	item := cache.getItem(id)
+	feed.cached = item
+	return item
+}
+
+func (feeds *Feeds) StoreCache(fileName string) error {
+	cache := feeds.cache
+	if cache == nil {
+		return fmt.Errorf("trying to store nil cache")
+	}
+	if cache.Version() != currentVersion {
+		return fmt.Errorf("trying to store cache with unsupported version '%d' (current: '%d')", cache.Version(), currentVersion)
+	}
+
+	f, err := os.Create(fileName)
+	if err != nil {
+		return fmt.Errorf("trying to store cache to '%s': %w", fileName, err)
+	}
+	defer f.Close()
+
+	writer := bufio.NewWriter(f)
+	if err = writer.WriteByte(currentVersion); err != nil {
+		return fmt.Errorf("writing to '%s': %w", fileName, err)
+	}
+
+	encoder := gob.NewEncoder(writer)
+	if err = encoder.Encode(cache); err != nil {
+		return fmt.Errorf("encoding cache: %w", err)
+	}
+
+	writer.Flush()
+	log.Printf("Stored cache to '%s'.", fileName)
+
+	return nil
+}
+
+func (feeds *Feeds) LoadCache(fileName string) error {
+	cache, err := loadCache(fileName)
+	if err != nil {
+		return err
+	}
+	feeds.cache = cache
+
+	for _, feed := range feeds.feeds {
+		feed.cached = cache.findItem(feed)
+	}
+	return nil
+}
+
+func loadCache(fileName string) (Cache, error) {
+	f, err := os.Open(fileName)
+	if err != nil {
+		if errors.Is(err, os.ErrNotExist) {
+			// no cache there yet -- make new
+			return New(), nil
+		}
+		return nil, fmt.Errorf("opening cache at '%s': %w", fileName, err)
+	}
+	defer f.Close()
+
+	log.Printf("Loading cache from '%s'", fileName)
+
+	reader := bufio.NewReader(f)
+	version, err := reader.ReadByte()
+	if err != nil {
+		return nil, fmt.Errorf("reading from '%s': %w", fileName, err)
+	}
+
+	cache, err := cacheForVersion(version)
+	if err != nil {
+		return nil, err
+	}
+
+	decoder := gob.NewDecoder(reader)
+	if err = decoder.Decode(cache); err != nil {
+		return nil, fmt.Errorf("decoding for version '%d' from '%s': %w", version, fileName, err)
+	}
+
+	if cache, err = cache.transformToCurrent(); err != nil {
+		return nil, fmt.Errorf("cannot transform from version %d to %d: %w", version, currentVersion, err)
+	}
+
+	log.Printf("Loaded cache (version %d), transformed to version %d.", version, currentVersion)
+
+	return cache, nil
+}
diff --git a/internal/feed/feed.go b/internal/feed/feed.go
index cd906a2..5af4188 100644
--- a/internal/feed/feed.go
+++ b/internal/feed/feed.go
@@ -3,10 +3,13 @@ package feed
 import (
 	"fmt"
 	"strings"
+	"sync"
+	"time"
 
 	"github.com/mmcdole/gofeed"
 
 	"github.com/Necoro/feed2imap-go/internal/config"
+	"github.com/Necoro/feed2imap-go/internal/log"
 )
 
 type Feed struct {
@@ -14,8 +17,9 @@ type Feed struct {
 	Target []string
 	Url    string
 	config.Options
-	feed  *gofeed.Feed
-	items []feeditem
+	feed   *gofeed.Feed
+	items  []feeditem
+	cached CachedFeed
 }
 
 type feeditem struct {
@@ -23,9 +27,18 @@ type feeditem struct {
 	*gofeed.Item
 }
 
-type Feeds map[string]*Feed
+type Feeds struct {
+	feeds map[string]*Feed
+	cache Cache
+}
+
+func NewFeeds() *Feeds {
+	return &Feeds{
+		feeds: map[string]*Feed{},
+	}
+}
 
-func (f Feeds) String() string {
+func (feeds *Feeds) String() string {
 	var b strings.Builder
 	app := func(a ...interface{}) {
 		_, _ = fmt.Fprint(&b, a...)
@@ -33,7 +46,7 @@ func (f Feeds) String() string {
 	app("Feeds [")
 
 	first := true
-	for k, v := range f {
+	for k, v := range feeds.feeds {
 		if !first {
 			app(", ")
 		}
@@ -49,3 +62,44 @@ func (f Feeds) String() string {
 
 	return b.String()
 }
+
+func (feeds *Feeds) Len() int {
+	return len(feeds.feeds)
+}
+
+func (feeds *Feeds) Contains(name string) bool {
+	_, ok := feeds.feeds[name]
+	return ok
+}
+
+func (feeds *Feeds) Set(name string, feed *Feed) {
+	feeds.feeds[name] = feed
+}
+
+func (feeds *Feeds) Foreach(f func(*Feed)) {
+	for _, feed := range feeds.feeds {
+		f(feed)
+	}
+}
+
+func (feeds *Feeds) ForeachGo(goFunc func(*Feed, *sync.WaitGroup)) {
+	var wg sync.WaitGroup
+	wg.Add(feeds.Len())
+
+	for _, feed := range feeds.feeds {
+		go goFunc(feed, &wg)
+	}
+	wg.Wait()
+}
+
+func (feed *Feed) NeedsUpdate(updateTime time.Time) bool {
+	if !updateTime.IsZero() && int(time.Since(updateTime).Hours()) >= feed.MinFreq {
+		log.Printf("Feed '%s' does not need updating, skipping.", feed.Name)
+		return false
+	}
+	return true
+}
+
+func (feed *Feed) Success() bool {
+	return feed.feed != nil
+}
diff --git a/internal/feed/parse.go b/internal/feed/parse.go
index 35a7596..6deebb2 100644
--- a/internal/feed/parse.go
+++ b/internal/feed/parse.go
@@ -32,7 +32,7 @@ func parseFeed(feed *Feed) error {
 	return nil
 }
 
-func handleFeed(feed *Feed, group *sync.WaitGroup, success chan<- bool) {
+func handleFeed(feed *Feed, group *sync.WaitGroup) {
 	defer group.Done()
 	log.Printf("Fetching %s from %s", feed.Name, feed.Url)
 
@@ -40,25 +40,17 @@ func handleFeed(feed *Feed, group *sync.WaitGroup, success chan<- bool) {
 	if err != nil {
 		log.Error(err)
 	}
-	success <- err == nil
 }
 
-func Parse(feeds Feeds) int {
-	var wg sync.WaitGroup
-	wg.Add(len(feeds))
-
-	success := make(chan bool, len(feeds))
-
-	for _, feed := range feeds {
-		go handleFeed(feed, &wg, success)
-	}
-
-	wg.Wait()
-	close(success)
+func (feeds Feeds) Parse() int {
+	feeds.ForeachGo(handleFeed)
 
 	ctr := 0
-	for s := range success {
-		if s {
+	for _, feed := range feeds.feeds {
+		success := feed.Success()
+		feed.cached.Checked(!success)
+
+		if success {
 			ctr++
 		}
 	}