aboutsummaryrefslogtreecommitdiff
path: root/internal/feed
diff options
context:
space:
mode:
authorRené 'Necoro' Neumann <necoro@necoro.eu>2021-02-22 22:54:43 +0100
committerRené 'Necoro' Neumann <necoro@necoro.eu>2021-02-22 22:54:43 +0100
commit7106d5a6e7585dce5fdd552cca30063dd352dc23 (patch)
tree88a9cb7150f86fcadb8f87e3a6d0892bf68c2251 /internal/feed
parentfb2aa9b1f04d509c8215c1fa6505a144482c343d (diff)
downloadfeed2imap-go-7106d5a6e7585dce5fdd552cca30063dd352dc23.tar.gz
feed2imap-go-7106d5a6e7585dce5fdd552cca30063dd352dc23.tar.bz2
feed2imap-go-7106d5a6e7585dce5fdd552cca30063dd352dc23.zip
Split cache and feed packages
Diffstat (limited to '')
-rw-r--r--internal/feed/cache/cache.go (renamed from internal/feed/cache.go)31
-rw-r--r--internal/feed/cache/cache_v1.go (renamed from internal/feed/cache_v1.go)63
-rw-r--r--internal/feed/cache/state.go146
-rw-r--r--internal/feed/feed.go86
-rw-r--r--internal/feed/item.go53
-rw-r--r--internal/feed/mail.go24
-rw-r--r--internal/feed/parse.go20
-rw-r--r--internal/feed/state.go184
-rw-r--r--internal/feed/template/html.tpl4
-rw-r--r--internal/feed/template/text.tpl2
10 files changed, 319 insertions, 294 deletions
diff --git a/internal/feed/cache.go b/internal/feed/cache/cache.go
index 5f5c394..e7fe673 100644
--- a/internal/feed/cache.go
+++ b/internal/feed/cache/cache.go
@@ -1,4 +1,4 @@
-package feed
+package cache
import (
"bufio"
@@ -11,6 +11,7 @@ import (
"github.com/nightlyone/lockfile"
+ "github.com/Necoro/feed2imap-go/internal/feed"
"github.com/Necoro/feed2imap-go/pkg/log"
)
@@ -20,30 +21,36 @@ const (
currentVersion Version = 1
)
-type CacheImpl interface {
- findItem(*Feed) CachedFeed
+type Impl interface {
+ cachedFeed(*feed.Feed) CachedFeed
+ transformToCurrent() (Impl, error)
Version() Version
Info() string
SpecificInfo(interface{}) string
- transformToCurrent() (CacheImpl, error)
}
type Cache struct {
- CacheImpl
+ Impl
lock lockfile.Lockfile
locked bool
}
type CachedFeed interface {
+ // Checked marks the feed as being a failure or a success on last check.
Checked(withFailure bool)
+ // Failures of this feed up to now.
Failures() int
+ // The Last time, this feed has been checked
Last() time.Time
- ID() string
- filterItems(items []item, ignoreHash bool, alwaysNew bool) []item
+ // Filter the given items against the cached items.
+ Filter(items []feed.Item, ignoreHash bool, alwaysNew bool) []feed.Item
+ // Commit any changes done to the cache state.
Commit()
+ // The Feed, that is cached.
+ Feed() *feed.Feed
}
-func cacheForVersion(version Version) (CacheImpl, error) {
+func cacheForVersion(version Version) (Impl, error) {
switch version {
case v1Version:
return newV1Cache(), nil
@@ -78,7 +85,7 @@ func lock(fileName string) (lock lockfile.Lockfile, err error) {
}
func (cache *Cache) store(fileName string) error {
- if cache.CacheImpl == nil {
+ if cache.Impl == nil {
return fmt.Errorf("trying to store nil cache")
}
if cache.Version() != currentVersion {
@@ -97,7 +104,7 @@ func (cache *Cache) store(fileName string) error {
}
encoder := gob.NewEncoder(writer)
- if err = encoder.Encode(cache.CacheImpl); err != nil {
+ if err = encoder.Encode(cache.Impl); err != nil {
return fmt.Errorf("encoding cache: %w", err)
}
@@ -123,8 +130,8 @@ func newCache() (Cache, error) {
return Cache{}, err
}
return Cache{
- CacheImpl: cache,
- locked: false,
+ Impl: cache,
+ locked: false,
}, nil
}
diff --git a/internal/feed/cache_v1.go b/internal/feed/cache/cache_v1.go
index a80e81c..d754b00 100644
--- a/internal/feed/cache_v1.go
+++ b/internal/feed/cache/cache_v1.go
@@ -1,4 +1,4 @@
-package feed
+package cache
import (
"crypto/sha256"
@@ -11,6 +11,7 @@ import (
"github.com/google/uuid"
+ "github.com/Necoro/feed2imap-go/internal/feed"
"github.com/Necoro/feed2imap-go/pkg/log"
"github.com/Necoro/feed2imap-go/pkg/util"
)
@@ -32,12 +33,13 @@ func idFromString(s string) feedId {
}
type v1Cache struct {
- Ids map[feedDescriptor]feedId
+ Ids map[feed.Descriptor]feedId
NextId uint64
Feeds map[feedId]*cachedFeed
}
type cachedFeed struct {
+ feed *feed.Feed
id feedId // not saved, has to be set on loading
LastCheck time.Time
currentCheck time.Time
@@ -100,8 +102,8 @@ func (cf *cachedFeed) Last() time.Time {
return cf.LastCheck
}
-func (cf *cachedFeed) ID() string {
- return cf.id.String()
+func (cf *cachedFeed) Feed() *feed.Feed {
+ return cf.feed
}
func (cache *v1Cache) Version() Version {
@@ -150,18 +152,18 @@ Num Items: %d
func newV1Cache() *v1Cache {
cache := v1Cache{
- Ids: map[feedDescriptor]feedId{},
+ Ids: map[feed.Descriptor]feedId{},
Feeds: map[feedId]*cachedFeed{},
NextId: startFeedId,
}
return &cache
}
-func (cache *v1Cache) transformToCurrent() (CacheImpl, error) {
+func (cache *v1Cache) transformToCurrent() (Impl, error) {
return cache, nil
}
-func (cache *v1Cache) getItem(id feedId) CachedFeed {
+func (cache *v1Cache) getItem(id feedId) *cachedFeed {
feed, ok := cache.Feeds[id]
if !ok {
feed = &cachedFeed{}
@@ -171,15 +173,11 @@ func (cache *v1Cache) getItem(id feedId) CachedFeed {
return feed
}
-func (cache *v1Cache) findItem(feed *Feed) CachedFeed {
- if feed.cached != nil {
- return feed.cached.(*cachedFeed)
- }
-
- fDescr := feed.descriptor()
+func (cache *v1Cache) cachedFeed(f *feed.Feed) CachedFeed {
+ fDescr := f.Descriptor()
id, ok := cache.Ids[fDescr]
if !ok {
- var otherId feedDescriptor
+ var otherId feed.Descriptor
changed := false
for otherId, id = range cache.Ids {
if otherId.Name == fDescr.Name {
@@ -204,15 +202,16 @@ func (cache *v1Cache) findItem(feed *Feed) CachedFeed {
cache.Ids[fDescr] = id
}
- item := cache.getItem(id)
- feed.cached = item
- return item
+ cf := cache.getItem(id)
+ cf.feed = f
+ f.SetExtID(id)
+ return cf
}
-func (item *item) newCachedItem() cachedItem {
+func newCachedItem(item *feed.Item) cachedItem {
var ci cachedItem
- ci.ID = item.itemId
+ ci.ID = item.ID
ci.Title = item.Item.Title
ci.Link = item.Item.Link
if item.DateParsed() != nil {
@@ -239,27 +238,27 @@ func (cf *cachedFeed) deleteItem(index int) {
cf.Items = cf.Items[:len(cf.Items)-1]
}
-func (cf *cachedFeed) filterItems(items []item, ignoreHash, alwaysNew bool) []item {
+func (cf *cachedFeed) Filter(items []feed.Item, ignoreHash, alwaysNew bool) []feed.Item {
if len(items) == 0 {
return items
}
- cacheItems := make(map[cachedItem]*item, len(items))
+ cacheItems := make(map[cachedItem]*feed.Item, len(items))
for idx := range items {
// remove complete duplicates on the go
- cacheItems[items[idx].newCachedItem()] = &items[idx]
+ cacheItems[newCachedItem(&items[idx])] = &items[idx]
}
log.Debugf("%d items after deduplication", len(cacheItems))
- filtered := make([]item, 0, len(items))
+ filtered := make([]feed.Item, 0, len(items))
cacheadd := make([]cachedItem, 0, len(items))
- app := func(item *item, ci cachedItem, oldIdx *int) {
+ app := func(item *feed.Item, ci cachedItem, oldIdx *int) {
if oldIdx != nil {
- item.updateOnly = true
+ item.UpdateOnly = true
prevId := cf.Items[*oldIdx].ID
ci.ID = prevId
- item.itemId = prevId
- log.Debugf("oldIdx: %d, prevId: %s, item.id: %s", *oldIdx, prevId, item.id())
+ item.ID = prevId
+ log.Debugf("oldIdx: %d, prevId: %s, item.id: %s", *oldIdx, prevId, item.Id())
cf.deleteItem(*oldIdx)
}
filtered = append(filtered, *item)
@@ -275,7 +274,7 @@ CACHE_ITEMS:
if oldItem.Guid == ci.Guid {
log.Debugf("Guid matches with: %s", oldItem)
if !oldItem.similarTo(&ci, ignoreHash) {
- item.addReason("guid (upd)")
+ item.AddReason("guid (upd)")
app(item, ci, &idx)
} else {
log.Debugf("Similar, ignoring item %s", base64.RawURLEncoding.EncodeToString(oldItem.ID[:]))
@@ -286,7 +285,7 @@ CACHE_ITEMS:
}
log.Debug("Found no matching GUID, including.")
- item.addReason("guid")
+ item.AddReason("guid")
app(item, ci, nil)
continue
}
@@ -300,11 +299,11 @@ CACHE_ITEMS:
if oldItem.Link == ci.Link {
if alwaysNew {
log.Debugf("Link matches, but `always-new`.")
- item.addReason("always-new")
+ item.AddReason("always-new")
continue
}
log.Debugf("Link matches, updating: %s", oldItem)
- item.addReason("link (upd)")
+ item.AddReason("link (upd)")
app(item, ci, &idx)
continue CACHE_ITEMS
@@ -312,7 +311,7 @@ CACHE_ITEMS:
}
log.Debugf("No match found, inserting.")
- item.addReason("new")
+ item.AddReason("new")
app(item, ci, nil)
}
diff --git a/internal/feed/cache/state.go b/internal/feed/cache/state.go
new file mode 100644
index 0000000..c2e7de8
--- /dev/null
+++ b/internal/feed/cache/state.go
@@ -0,0 +1,146 @@
+package cache
+
+import (
+ "sync"
+
+ "github.com/Necoro/feed2imap-go/internal/feed"
+ "github.com/Necoro/feed2imap-go/pkg/config"
+ "github.com/Necoro/feed2imap-go/pkg/log"
+)
+
+type State struct {
+ feeds map[string]*feed.Feed
+ cachedFeeds map[string]CachedFeed
+ cache Cache
+ cfg *config.Config
+}
+
+func (state *State) Foreach(f func(CachedFeed)) {
+ for _, feed := range state.cachedFeeds {
+ f(feed)
+ }
+}
+
+func (state *State) ForeachGo(goFunc func(CachedFeed)) {
+ var wg sync.WaitGroup
+ wg.Add(len(state.feeds))
+
+ f := func(feed CachedFeed, wg *sync.WaitGroup) {
+ goFunc(feed)
+ wg.Done()
+ }
+
+ for _, feed := range state.cachedFeeds {
+ go f(feed, &wg)
+ }
+ wg.Wait()
+}
+
+func (state *State) LoadCache(fileName string, forceNew bool) error {
+ var (
+ cache Cache
+ err error
+ )
+
+ if forceNew {
+ cache, err = newCache()
+ } else {
+ cache, err = LoadCache(fileName)
+ }
+
+ if err != nil {
+ return err
+ }
+ state.cache = cache
+
+ for name, feed := range state.feeds {
+ state.cachedFeeds[name] = cache.cachedFeed(feed)
+ }
+
+ // state.feeds should not be used after loading the cache --> enforce a panic
+ state.feeds = nil
+
+ return nil
+}
+
+func (state *State) StoreCache(fileName string) error {
+ return state.cache.store(fileName)
+}
+
+func (state *State) UnlockCache() {
+ _ = state.cache.Unlock()
+}
+
+func (state *State) Fetch() int {
+ state.ForeachGo(handleFeed)
+
+ ctr := 0
+ for _, cf := range state.cachedFeeds {
+ success := cf.Feed().FetchSuccessful()
+ cf.Checked(!success)
+
+ if success {
+ ctr++
+ }
+ }
+
+ return ctr
+}
+
+func handleFeed(cf CachedFeed) {
+ feed := cf.Feed()
+ log.Printf("Fetching %s from %s", feed.Name, feed.Url)
+
+ err := feed.Parse()
+ if err != nil {
+ if feed.Url == "" || cf.Failures() >= feed.Global.MaxFailures {
+ log.Error(err)
+ } else {
+ log.Print(err)
+ }
+ }
+}
+
+func filterFeed(cf CachedFeed) {
+ cf.Feed().Filter(cf.Filter)
+}
+
+func (state *State) Filter() {
+ if log.IsDebug() {
+ // single threaded for better output
+ state.Foreach(filterFeed)
+ } else {
+ state.ForeachGo(filterFeed)
+ }
+}
+
+func NewState(cfg *config.Config) (*State, error) {
+ state := State{
+ feeds: map[string]*feed.Feed{},
+ cachedFeeds: map[string]CachedFeed{},
+ cache: Cache{}, // loaded later on
+ cfg: cfg,
+ }
+
+ for name, parsedFeed := range cfg.Feeds {
+ feed, err := feed.Create(parsedFeed, cfg.GlobalOptions)
+ if err != nil {
+ return nil, err
+ }
+ state.feeds[name] = feed
+ }
+
+ return &state, nil
+}
+
+func (state *State) RemoveUndue() {
+ for name, feed := range state.cachedFeeds {
+ if feed.Feed().Disable || !feed.Feed().NeedsUpdate(feed.Last()) {
+ delete(state.cachedFeeds, name)
+ }
+ }
+}
+
+func (state *State) NumFeeds() int {
+ return len(state.cachedFeeds)
+}
diff --git a/internal/feed/feed.go b/internal/feed/feed.go
index de28ef8..431eae3 100644
--- a/internal/feed/feed.go
+++ b/internal/feed/feed.go
@@ -16,24 +16,30 @@ type Feed struct {
*config.Feed
feed *gofeed.Feed
filter *filter.Filter
- items []item
- cached CachedFeed
+ items []Item
Global config.GlobalOptions
+ extID FeedID
}
-type feedDescriptor struct {
+type FeedID interface {
+ String() string
+}
+
+type FilterFunc func(items []Item, ignHash, alwaysNew bool) []Item
+
+type Descriptor struct {
Name string
Url string
}
-func (feed *Feed) descriptor() feedDescriptor {
+func (feed *Feed) Descriptor() Descriptor {
var url string
if feed.Url != "" {
url = feed.Url
} else {
url = "exec://" + strings.Join(feed.Exec, "/")
}
- return feedDescriptor{
+ return Descriptor{
Name: feed.Name,
Url: url,
}
@@ -54,12 +60,6 @@ func (feed *Feed) FetchSuccessful() bool {
return feed.feed != nil
}
-func (feed *Feed) MarkSuccess() {
- if feed.cached != nil {
- feed.cached.Commit()
- }
-}
-
func Create(parsedFeed *config.Feed, global config.GlobalOptions) (*Feed, error) {
var itemFilter *filter.Filter
var err error
@@ -70,3 +70,67 @@ func Create(parsedFeed *config.Feed, global config.GlobalOptions) (*Feed, error)
}
return &Feed{Feed: parsedFeed, Global: global, filter: itemFilter}, nil
}
+
+func (feed *Feed) filterItems() []Item {
+ if feed.filter == nil {
+ return feed.items
+ }
+
+ items := make([]Item, 0, len(feed.items))
+
+ for _, item := range feed.items {
+ res, err := feed.filter.Run(item.Item)
+ if err != nil {
+ log.Errorf("Feed %s: Item %s: Error applying item filter: %s", feed.Name, printItem(item.Item), err)
+ res = true // include
+ }
+
+ if res {
+ if log.IsDebug() {
+ log.Debugf("Filter '%s' matches for item %s", feed.ItemFilter, printItem(item.Item))
+ }
+ items = append(items, item)
+ } else if log.IsDebug() { // printItem is not for free
+ log.Debugf("Filter '%s' does not match for item %s", feed.ItemFilter, printItem(item.Item))
+ }
+ }
+ return items
+}
+
+func (feed *Feed) Filter(filter FilterFunc) {
+ if len(feed.items) > 0 {
+ origLen := len(feed.items)
+
+ log.Debugf("Filtering %s. Starting with %d items", feed.Name, origLen)
+
+ items := feed.filterItems()
+ newLen := len(items)
+ if newLen < origLen {
+ log.Printf("Item filter on %s: Reduced from %d to %d items.", feed.Name, origLen, newLen)
+ origLen = newLen
+ }
+
+ feed.items = filter(items, feed.IgnHash, feed.AlwaysNew)
+
+ newLen = len(feed.items)
+ if newLen < origLen {
+ log.Printf("Filtered %s. Reduced from %d to %d items.", feed.Name, origLen, newLen)
+ } else {
+ log.Printf("Filtered %s, no reduction.", feed.Name)
+ }
+
+ } else {
+ log.Debugf("No items for %s. No filtering.", feed.Name)
+ }
+}
+
+func (feed *Feed) SetExtID(extID FeedID) {
+ feed.extID = extID
+}
+
+func (feed *Feed) id() string {
+ if feed.extID == nil {
+ return feed.Name
+ }
+ return feed.extID.String()
+}
diff --git a/internal/feed/item.go b/internal/feed/item.go
index 3cb9089..39f41ba 100644
--- a/internal/feed/item.go
+++ b/internal/feed/item.go
@@ -2,6 +2,7 @@ package feed
import (
"encoding/base64"
+ "encoding/json"
"fmt"
"time"
@@ -17,26 +18,26 @@ type feedImage struct {
mime string
}
-type item struct {
- *gofeed.Item
- Feed *gofeed.Feed
- feed *Feed
- Body string
- TextBody string
- updateOnly bool
- reasons []string
- images []feedImage
- itemId uuid.UUID
+type Item struct {
+ *gofeed.Item // access fields implicitly
+ Feed *gofeed.Feed // named explicitly to not shadow common fields with Item
+ feed *Feed
+ Body string
+ TextBody string
+ UpdateOnly bool
+ reasons []string
+ images []feedImage
+ ID uuid.UUID
}
-func (item *item) DateParsed() *time.Time {
+func (item *Item) DateParsed() *time.Time {
if item.UpdatedParsed == nil || item.UpdatedParsed.IsZero() {
return item.PublishedParsed
}
return item.UpdatedParsed
}
-func (item *item) Date() string {
+func (item *Item) Date() string {
if item.Updated == "" {
return item.Published
}
@@ -44,14 +45,14 @@ func (item *item) Date() string {
}
// Creator returns the name of the creating author.
-func (item *item) Creator() string {
+func (item *Item) Creator() string {
if item.Author != nil {
return item.Author.Name
}
return ""
}
-func (item *item) FeedLink() string {
+func (item *Item) FeedLink() string {
if item.Feed.FeedLink != "" {
// the one in the feed itself
return item.Feed.FeedLink
@@ -60,31 +61,37 @@ func (item *item) FeedLink() string {
return item.feed.Url
}
-func (item *item) addReason(reason string) {
+func (item *Item) AddReason(reason string) {
if !util.StrContains(item.reasons, reason) {
item.reasons = append(item.reasons, reason)
}
}
-func (item *item) addImage(img []byte, mime string) int {
+func (item *Item) addImage(img []byte, mime string) int {
i := feedImage{img, mime}
item.images = append(item.images, i)
return len(item.images)
}
-func (item *item) clearImages() {
+func (item *Item) clearImages() {
item.images = []feedImage{}
}
-func (item *item) defaultEmail() string {
+func (item *Item) defaultEmail() string {
return item.feed.Global.DefaultEmail
}
-func (item *item) id() string {
- idStr := base64.RawURLEncoding.EncodeToString(item.itemId[:])
- return item.feed.cached.ID() + "#" + idStr
+func (item *Item) Id() string {
+ idStr := base64.RawURLEncoding.EncodeToString(item.ID[:])
+ return item.feed.id() + "#" + idStr
}
-func (item *item) messageId() string {
- return fmt.Sprintf("<feed#%s@%s>", item.id(), config.Hostname())
+func (item *Item) messageId() string {
+ return fmt.Sprintf("<feed#%s@%s>", item.Id(), config.Hostname())
+}
+
+func printItem(item *gofeed.Item) string {
+ // analogous to gofeed.Feed.String
+ json, _ := json.MarshalIndent(item, "", " ")
+ return string(json)
}
diff --git a/internal/feed/mail.go b/internal/feed/mail.go
index b201d61..636f6a0 100644
--- a/internal/feed/mail.go
+++ b/internal/feed/mail.go
@@ -31,7 +31,7 @@ func address(name, address string) []*mail.Address {
return []*mail.Address{{Name: name, Address: address}}
}
-func (item *item) fromAddress() []*mail.Address {
+func (item *Item) fromAddress() []*mail.Address {
switch {
case item.Author != nil && item.Author.Email != "":
return address(item.Author.Name, item.Author.Email)
@@ -46,11 +46,11 @@ func (item *item) fromAddress() []*mail.Address {
}
}
-func (item *item) toAddress() []*mail.Address {
+func (item *Item) toAddress() []*mail.Address {
return address(item.feed.Name, item.defaultEmail())
}
-func (item *item) buildHeader() message.Header {
+func (item *Item) buildHeader() message.Header {
var h mail.Header
h.SetContentType("multipart/alternative", nil)
h.SetAddressList("From", item.fromAddress())
@@ -58,7 +58,7 @@ func (item *item) buildHeader() message.Header {
h.Set("Message-Id", item.messageId())
h.Set(msg.VersionHeader, version.Version())
h.Set(msg.ReasonHeader, strings.Join(item.reasons, ","))
- h.Set(msg.IdHeader, item.id())
+ h.Set(msg.IdHeader, item.Id())
h.Set(msg.CreateHeader, time.Now().Format(time.RFC1123Z))
if item.GUID != "" {
h.Set(msg.GuidHeader, item.GUID)
@@ -86,7 +86,7 @@ func (item *item) buildHeader() message.Header {
return h.Header
}
-func (item *item) writeContentPart(w *message.Writer, typ string, tpl template.Template) error {
+func (item *Item) writeContentPart(w *message.Writer, typ string, tpl template.Template) error {
var ih message.Header
ih.SetContentType("text/"+typ, map[string]string{"charset": "utf-8"})
ih.SetContentDisposition("inline", nil)
@@ -105,11 +105,11 @@ func (item *item) writeContentPart(w *message.Writer, typ string, tpl template.T
return nil
}
-func (item *item) writeTextPart(w *message.Writer) error {
+func (item *Item) writeTextPart(w *message.Writer) error {
return item.writeContentPart(w, "plain", template.Text)
}
-func (item *item) writeHtmlPart(w *message.Writer) error {
+func (item *Item) writeHtmlPart(w *message.Writer) error {
return item.writeContentPart(w, "html", template.Html)
}
@@ -133,7 +133,7 @@ func (img *feedImage) writeImagePart(w *message.Writer, cid string) error {
return nil
}
-func (item *item) writeToBuffer(b *bytes.Buffer) error {
+func (item *Item) writeToBuffer(b *bytes.Buffer) error {
h := item.buildHeader()
item.buildBody()
@@ -178,7 +178,7 @@ func (item *item) writeToBuffer(b *bytes.Buffer) error {
return nil
}
-func (item *item) message() (msg.Message, error) {
+func (item *Item) message() (msg.Message, error) {
var b bytes.Buffer
if err := item.writeToBuffer(&b); err != nil {
@@ -187,8 +187,8 @@ func (item *item) message() (msg.Message, error) {
msg := msg.Message{
Content: b.String(),
- IsUpdate: item.updateOnly,
- ID: item.id(),
+ IsUpdate: item.UpdateOnly,
+ ID: item.Id(),
}
return msg, nil
@@ -251,7 +251,7 @@ func getBody(content, description string, bodyCfg config.Body) string {
}
}
-func (item *item) buildBody() {
+func (item *Item) buildBody() {
feed := item.feed
var feedUrl *url.URL
diff --git a/internal/feed/parse.go b/internal/feed/parse.go
index b08c286..a2812bd 100644
--- a/internal/feed/parse.go
+++ b/internal/feed/parse.go
@@ -9,10 +9,9 @@ import (
"github.com/mmcdole/gofeed"
"github.com/Necoro/feed2imap-go/internal/http"
- "github.com/Necoro/feed2imap-go/pkg/log"
)
-func (feed *Feed) parse() error {
+func (feed *Feed) Parse() error {
fp := gofeed.NewParser()
var reader io.Reader
@@ -58,22 +57,9 @@ func (feed *Feed) parse() error {
}
feed.feed = parsedFeed
- feed.items = make([]item, len(parsedFeed.Items))
+ feed.items = make([]Item, len(parsedFeed.Items))
for idx, feedItem := range parsedFeed.Items {
- feed.items[idx] = item{Feed: parsedFeed, Item: feedItem, itemId: uuid.New(), feed: feed}
+ feed.items[idx] = Item{Feed: parsedFeed, feed: feed, Item: feedItem, ID: uuid.New()}
}
return cleanup()
}
-
-func handleFeed(feed *Feed) {
- log.Printf("Fetching %s from %s", feed.Name, feed.Url)
-
- err := feed.parse()
- if err != nil {
- if feed.Url == "" || feed.cached.Failures() >= feed.Global.MaxFailures {
- log.Error(err)
- } else {
- log.Print(err)
- }
- }
-}
diff --git a/internal/feed/state.go b/internal/feed/state.go
deleted file mode 100644
index 364616f..0000000
--- a/internal/feed/state.go
+++ /dev/null
@@ -1,184 +0,0 @@
-package feed
-
-import (
- "encoding/json"
- "sync"
-
- "github.com/mmcdole/gofeed"
-
- "github.com/Necoro/feed2imap-go/pkg/config"
- "github.com/Necoro/feed2imap-go/pkg/log"
-)
-
-type State struct {
- feeds map[string]*Feed
- cache Cache
- cfg *config.Config
-}
-
-func (state *State) Foreach(f func(*Feed)) {
- for _, feed := range state.feeds {
- f(feed)
- }
-}
-
-func (state *State) ForeachGo(goFunc func(*Feed)) {
- var wg sync.WaitGroup
- wg.Add(len(state.feeds))
-
- f := func(feed *Feed, wg *sync.WaitGroup) {
- goFunc(feed)
- wg.Done()
- }
-
- for _, feed := range state.feeds {
- go f(feed, &wg)
- }
- wg.Wait()
-}
-
-func (state *State) LoadCache(fileName string, forceNew bool) error {
- var (
- cache Cache
- err error
- )
-
- if forceNew {
- cache, err = newCache()
- } else {
- cache, err = LoadCache(fileName)
- }
-
- if err != nil {
- return err
- }
- state.cache = cache
-
- for _, feed := range state.feeds {
- feed.cached = cache.findItem(feed)
- }
- return nil
-}
-
-func (state *State) StoreCache(fileName string) error {
- return state.cache.store(fileName)
-}
-
-func (state *State) UnlockCache() {
- _ = state.cache.Unlock()
-}
-
-func (state *State) Fetch() int {
- state.ForeachGo(handleFeed)
-
- ctr := 0
- for _, feed := range state.feeds {
- success := feed.FetchSuccessful()
- feed.cached.Checked(!success)
-
- if success {
- ctr++
- }
- }
-
- return ctr
-}
-
-func printItem(item *gofeed.Item) string {
- // analogous to gofeed.Feed.String
- json, _ := json.MarshalIndent(item, "", " ")
- return string(json)
-}
-
-func (feed *Feed) filterItems() []item {
- if feed.filter == nil {
- return feed.items
- }
-
- items := make([]item, 0, len(feed.items))
-
- for _, item := range feed.items {
- res, err := feed.filter.Run(item.Item)
- if err != nil {
- log.Errorf("Feed %s: Item %s: Error applying item filter: %s", feed.Name, printItem(item.Item), err)
- res = true // include
- }
-
- if res {
- if log.IsDebug() {
- log.Debugf("Filter '%s' matches for item %s", feed.ItemFilter, printItem(item.Item))
- }
- items = append(items, item)
- } else if log.IsDebug() { // printItem is not for free
- log.Debugf("Filter '%s' does not match for item %s", feed.ItemFilter, printItem(item.Item))
- }
- }
- return items
-}
-
-func filterFeed(feed *Feed) {
- if len(feed.items) > 0 {
- origLen := len(feed.items)
-
- log.Debugf("Filtering %s. Starting with %d items", feed.Name, origLen)
-
- items := feed.filterItems()
- newLen := len(items)
- if newLen < origLen {
- log.Printf("Item filter on %s: Reduced from %d to %d items.", feed.Name, origLen, newLen)
- origLen = newLen
- }
-
- items = feed.cached.filterItems(items, feed.IgnHash, feed.AlwaysNew)
- feed.items = items
-
- newLen = len(feed.items)
- if newLen < origLen {
- log.Printf("Filtered %s. Reduced from %d to %d items.", feed.Name, origLen, newLen)
- } else {
- log.Printf("Filtered %s, no reduction.", feed.Name)
- }
-
- } else {
- log.Debugf("No items for %s. No filtering.", feed.Name)
- }
-}
-
-func (state *State) Filter() {
- if log.IsDebug() {
- // single threaded for better output
- state.Foreach(filterFeed)
- } else {
- state.ForeachGo(filterFeed)
- }
-}
-
-func NewState(cfg *config.Config) (*State, error) {
- state := State{
- feeds: map[string]*Feed{},
- cache: Cache{}, // loaded later on
- cfg: cfg,
- }
-
- for name, parsedFeed := range cfg.Feeds {
- feed, err := Create(parsedFeed, cfg.GlobalOptions)
- if err != nil {
- return nil, err
- }
- state.feeds[name] = feed
- }
-
- return &state, nil
-}
-
-func (state *State) RemoveUndue() {
- for name, feed := range state.feeds {
- if feed.Disable || !feed.NeedsUpdate(feed.cached.Last()) {
- delete(state.feeds, name)
- }
- }
-}
-
-func (state *State) NumFeeds() int {
- return len(state.feeds)
-}
diff --git a/internal/feed/template/html.tpl b/internal/feed/template/html.tpl
index aa7d341..ed98121 100644
--- a/internal/feed/template/html.tpl
+++ b/internal/feed/template/html.tpl
@@ -1,4 +1,4 @@
-{{- /*gotype:github.com/Necoro/feed2imap-go/internal/feed.feeditem*/ -}}
+{{- /*gotype:github.com/Necoro/feed2imap-go/internal/feed.Item*/ -}}
{{define "bottomLine"}}
{{if .content}}
<tr>
@@ -16,7 +16,7 @@
<td style="text-align: right; padding: 4px"><strong>Feed</strong></td>
<td style="width: 100%; padding: 4px">
{{with .Feed.Link}}<a href="{{.}}">{{end}}
- <strong>{{or .Feed.Title .Feed.Link "Unnammed feed"}}</strong>
+ <strong>{{or .Feed.Title .Feed.Link "Unnamed feed"}}</strong>
{{if .Feed.Link}}</a>{{end}}
</td>
</tr>
diff --git a/internal/feed/template/text.tpl b/internal/feed/template/text.tpl
index 1225902..8ca30ed 100644
--- a/internal/feed/template/text.tpl
+++ b/internal/feed/template/text.tpl
@@ -1,4 +1,4 @@
-{{- /*gotype:github.com/Necoro/feed2imap-go/internal/feed.feeditem*/ -}}
+{{- /*gotype:github.com/Necoro/feed2imap-go/internal/feed.Item*/ -}}
{{- with .Item.Link -}}
<{{.}}>