path: root/internal/feed/cache_v1.go

package feed

import (
	"crypto/sha256"
	"encoding/base64"
	"encoding/hex"
	"fmt"
	"strconv"
	"strings"
	"time"

	"github.com/google/uuid"

	"github.com/Necoro/feed2imap-go/pkg/log"
	"github.com/Necoro/feed2imap-go/pkg/util"
)

const (
	v1Version   Version = 1
	startFeedId uint64  = 1
)

type feedId uint64

func (id feedId) String() string {
	return strconv.FormatUint(uint64(id), 16)
}

func idFromString(s string) feedId {
	id, _ := strconv.ParseUint(s, 16, 64)
	return feedId(id)
}

type v1Cache struct {
	Ids    map[feedDescriptor]feedId
	NextId uint64
	Feeds  map[feedId]*cachedFeed
}

type cachedFeed struct {
	id           feedId // not saved, has to be set on loading
	LastCheck    time.Time
	currentCheck time.Time
	NumFailures  int // can't be named `Failures` b/c it'll collide with the interface
	Items        []cachedItem
	newItems     []cachedItem
}

type itemHash [sha256.Size]byte

func (h itemHash) String() string {
	return hex.EncodeToString(h[:])
}

type cachedItem struct {
	Guid         string
	Title        string
	Link         string
	Date         time.Time
	UpdatedCache time.Time
	Hash         itemHash
	ID           uuid.UUID
}

func (item cachedItem) String() string {
	return fmt.Sprintf(`{
  ID: %s
  Title: %q
  Guid: %q
  Link: %q
  Date: %s
  Hash: %s
}`,
		base64.RawURLEncoding.EncodeToString(item.ID[:]),
		item.Title, item.Guid, item.Link, util.TimeFormat(item.Date), item.Hash)
}

func (cf *cachedFeed) Checked(withFailure bool) {
	cf.currentCheck = time.Now()
	if withFailure {
		cf.NumFailures++
	} else {
		cf.NumFailures = 0
	}
}

func (cf *cachedFeed) Commit() {
	if cf.newItems != nil {
		cf.Items = cf.newItems
		cf.newItems = nil
	}
	cf.LastCheck = cf.currentCheck
}

func (cf *cachedFeed) Failures() int {
	return cf.NumFailures
}

func (cf *cachedFeed) Last() time.Time {
	return cf.LastCheck
}

func (cf *cachedFeed) ID() string {
	return cf.id.String()
}

func (cache *v1Cache) Version() Version {
	return v1Version
}

func (cache *v1Cache) Info() string {
	b := strings.Builder{}
	for descr, id := range cache.Ids {
		b.WriteString(fmt.Sprintf("%3s: %s (%s)\n", id.String(), descr.Name, descr.Url))
	}
	return b.String()
}

func (cache *v1Cache) SpecificInfo(i interface{}) string {
	id := idFromString(i.(string))

	b := strings.Builder{}
	feed := cache.Feeds[id]

	for descr, fId := range cache.Ids {
		if id == fId {
			b.WriteString(descr.Name)
			b.WriteString(" -- ")
			b.WriteString(descr.Url)
			b.WriteByte('\n')
			break
		}
	}

	b.WriteString(fmt.Sprintf(`
Last Check: %s
Num Failures: %d
Num Items: %d
`,
		util.TimeFormat(feed.LastCheck),
		feed.NumFailures,
		len(feed.Items)))

	for _, item := range feed.Items {
		b.WriteString("\n--------------------\n")
		b.WriteString(item.String())
	}
	return b.String()
}

func newV1Cache() *v1Cache {
	cache := v1Cache{
		Ids:    map[feedDescriptor]feedId{},
		Feeds:  map[feedId]*cachedFeed{},
		NextId: startFeedId,
	}
	return &cache
}

func (cache *v1Cache) transformToCurrent() (CacheImpl, error) {
	return cache, nil
}

func (cache *v1Cache) getItem(id feedId) CachedFeed {
	feed, ok := cache.Feeds[id]
	if !ok {
		feed = &cachedFeed{}
		cache.Feeds[id] = feed
	}
	feed.id = id
	return feed
}

func (cache *v1Cache) findItem(feed *Feed) CachedFeed {
	if feed.cached != nil {
		return feed.cached.(*cachedFeed)
	}

	fDescr := feed.descriptor()
	id, ok := cache.Ids[fDescr]
	if !ok {
		var otherId feedDescriptor
		changed := false
		for otherId, id = range cache.Ids {
			if otherId.Name == fDescr.Name {
				log.Warnf("Feed %s seems to have changed URLs: new '%s', old '%s'. Updating.",
					fDescr.Name, fDescr.Url, otherId.Url)
				changed = true
				break
			} else if otherId.Url == fDescr.Url {
				log.Warnf("Feed with URL '%s' seems to have changed its name: new '%s', old '%s'. Updating.",
					fDescr.Url, fDescr.Name, otherId.Name)
				changed = true
				break
			}
		}
		if changed {
			delete(cache.Ids, otherId)
		} else {
			id = feedId(cache.NextId)
			cache.NextId++
		}

		cache.Ids[fDescr] = id
	}

	item := cache.getItem(id)
	feed.cached = item
	return item
}

func (item *item) newCachedItem() cachedItem {
	var ci cachedItem

	ci.ID = item.itemId
	ci.Title = item.Item.Title
	ci.Link = item.Item.Link
	if item.DateParsed() != nil {
		ci.Date = *item.DateParsed()
	}
	ci.Guid = item.Item.GUID

	contentByte := []byte(item.Item.Description + item.Item.Content)
	ci.Hash = sha256.Sum256(contentByte)

	return ci
}

func (item *cachedItem) similarTo(other *cachedItem, ignoreHash bool) bool {
	return other.Title == item.Title &&
		other.Link == item.Link &&
		other.Date.Equal(item.Date) &&
		(ignoreHash || other.Hash == item.Hash)
}

func (cf *cachedFeed) deleteItem(index int) {
	copy(cf.Items[index:], cf.Items[index+1:])
	cf.Items[len(cf.Items)-1] = cachedItem{}
	cf.Items = cf.Items[:len(cf.Items)-1]
}

func (cf *cachedFeed) filterItems(items []item, ignoreHash, alwaysNew bool) []item {
	if len(items) == 0 {
		return items
	}

	cacheItems := make(map[cachedItem]*item, len(items))
	for idx := range items {
		// remove complete duplicates on the go
		cacheItems[items[idx].newCachedItem()] = &items[idx]
	}
	log.Debugf("%d items after deduplication", len(cacheItems))

	filtered := make([]item, 0, len(items))
	cacheadd := make([]cachedItem, 0, len(items))