🚧 Sorting, update funcs, mutexes

amfora.go

@@ -39,7 +39,7 @@ func main() {
 	}
 	err = feeds.Init()
 	if err != nil {
-		fmt.Fprintf(os.Stderr, "Config error: %v\n", err)
+		fmt.Fprintf(os.Stderr, "feeds.json error: %v\n", err)
 		os.Exit(1)
 	}
 

feeds/feeds.go

@@ -1,45 +1,32 @@
 package feeds
 
 import (
+	"crypto/sha256"
 	"encoding/json"
 	"errors"
 	"fmt"
 	"io"
+	"mime"
 	"os"
+	"path"
 	"sort"
 	"strings"
+	"sync"
+	"time"
 
+	"github.com/makeworld-the-better-one/amfora/client"
 	"github.com/makeworld-the-better-one/amfora/config"
+	"github.com/makeworld-the-better-one/go-gemini"
 	"github.com/mmcdole/gofeed"
 )
 
-/*
-Example JSON.
-{
-	"feeds": {
-		"url1": <gofeed.Feed>,
-		"url2: <gofeed.Feed>"
-	},
-	"pages": {
-		"url1": "hash",
-		"url2": "hash"
-	}
-}
+var (
+	ErrSaving     = errors.New("couldn't save JSON to disk")
+	ErrNotSuccess = errors.New("status 20 not returned")
+	ErrNotFeed    = errors.New("not a valid feed")
+)
 
-"pages" are the pages tracked for changes that aren't feeds.
-The hash is SHA-256.
-
-*/
-
-// Decoded JSON
-type feedJson struct {
-	Feeds map[string]*gofeed.Feed `json:"feeds"`
-	Pages map[string]string       `json:"pages"`
-}
-
-var data feedJson
-
-var ErrSaving = errors.New("couldn't save JSON to disk")
+var writeMu = sync.Mutex{}
 
 // Init should be called after config.Init.
 func Init() error {
@@ -51,20 +38,28 @@ func Init() error {
 		return fmt.Errorf("feeds json is corrupted: %v", err)
 	}
 	return nil
+
+	// TODO: Start pulling all feeds in another thread
 }
 
-// IsTracked returns true of the feed/page URL is already being tracked.
+// IsTracked returns true if the feed/page URL is already being tracked.
 func IsTracked(url string) bool {
+	data.feedMu.RLock()
 	for u := range data.Feeds {
 		if url == u {
+			data.feedMu.RUnlock()
 			return true
 		}
 	}
+	data.feedMu.RUnlock()
+	data.pageMu.RLock()
 	for u := range data.Pages {
 		if url == u {
+			data.pageMu.RUnlock()
 			return true
 		}
 	}
+	data.pageMu.RUnlock()
 	return false
 }
 
@@ -79,10 +74,16 @@ func GetFeed(mediatype, filename string, r io.Reader) (*gofeed.Feed, bool) {
 		return nil, false
 	}
 	feed, err := gofeed.NewParser().Parse(r)
+	if feed == nil {
+		return nil, false
+	}
 	return feed, err == nil
 }
 
 func writeJson() error {
+	writeMu.Lock()
+	defer writeMu.Unlock()
+
 	f, err := os.OpenFile(config.FeedPath, os.O_WRONLY|os.O_CREATE, 0666)
 	if err != nil {
 		return err
@@ -91,27 +92,126 @@ func writeJson() error {
 	enc := json.NewEncoder(f)
 	enc.SetEscapeHTML(false)
 	enc.SetIndent("", "  ")
+
+	data.feedMu.Lock()
+	data.pageMu.Lock()
 	err = enc.Encode(&data)
+	data.feedMu.Unlock()
+	data.pageMu.Unlock()
+
 	return err
 }
 
 // AddFeed stores a feed.
+// It can be used to update a feed for a URL, although the package
+// will handle that on its own.
 func AddFeed(url string, feed *gofeed.Feed) error {
+	if feed == nil {
+		panic("feed is nil")
+	}
+
 	sort.Sort(feed)
+	// Remove any content to save memory and disk space
+	for _, item := range feed.Items {
+		item.Content = ""
+	}
+
+	data.feedMu.Lock()
 	data.Feeds[url] = feed
+	data.feedMu.Unlock()
+
 	err := writeJson()
 	if err != nil {
+		// Don't use in-memory if it couldn't be saved
+		data.feedMu.Lock()
+		delete(data.Feeds, url)
+		data.feedMu.Unlock()
 		return ErrSaving
 	}
 	return nil
 }
 
 // AddPage stores a page URL to track for changes.
+// Do not use it to update a page, as it only resets the hash.
 func AddPage(url string) error {
-	data.Pages[url] = "" // No hash yet
+	data.pageMu.Lock()
+	data.Pages[url] = &pageJson{} // No hash yet
+	data.pageMu.Unlock()
+
 	err := writeJson()
 	if err != nil {
+		// Don't use in-memory if it couldn't be saved
+		data.pageMu.Lock()
+		delete(data.Pages, url)
+		data.pageMu.Unlock()
 		return ErrSaving
 	}
 	return nil
 }
+
+func updateFeed(url string) error {
+	res, err := client.Fetch(url)
+	if err != nil {
+		if res != nil {
+			res.Body.Close()
+		}
+		return err
+	}
+	defer res.Body.Close()
+
+	if res.Status != gemini.StatusSuccess {
+		return ErrNotSuccess
+	}
+	mediatype, _, err := mime.ParseMediaType(res.Meta)
+	if err != nil {
+		return err
+	}
+	filename := path.Base(url)
+	feed, ok := GetFeed(mediatype, filename, res.Body)
+	if !ok {
+		return ErrNotFeed
+	}
+	return AddFeed(url, feed)
+}
+
+func updatePage(url string) error {
+	res, err := client.Fetch(url)
+	if err != nil {
+		if res != nil {
+			res.Body.Close()
+		}
+		return err
+	}
+	defer res.Body.Close()
+
+	if res.Status != gemini.StatusSuccess {
+		return ErrNotSuccess
+	}
+	h := sha256.New()
+	if _, err := io.Copy(h, res.Body); err != nil {
+		return err
+	}
+	data.pageMu.Lock()
+	data.Pages[url] = &pageJson{
+		Hash:    fmt.Sprintf("%x", h.Sum(nil)),
+		Updated: time.Now().UTC(),
+	}
+	data.pageMu.Unlock()
+
+	err = writeJson()
+	if err != nil {
+		// Don't use in-memory if it couldn't be saved
+		data.pageMu.Lock()
+		delete(data.Pages, url)
+		data.pageMu.Unlock()
+		return err
+	}
+
+	return nil
+}
+
+// updateAll updates all feeds and pages.
+// It should run in goroutine at a regular interval.
+func updateAll() {
+
+}

feeds/structs.go

@@ -0,0 +1,83 @@
+package feeds
+
+import (
+	"sync"
+	"time"
+
+	"github.com/mmcdole/gofeed"
+)
+
+/*
+Example JSON.
+{
+	"feeds": {
+		"url1": <gofeed.Feed>,
+		"url2": <gofeed.Feed>,
+	},
+	"pages": {
+		"url1": {
+			"hash": <hash>,
+			"updated": <time>
+		},
+		"url2": {
+			"hash": <hash>,
+			"updated": <time>
+		}
+	}
+}
+
+"pages" are the pages tracked for changes that aren't feeds.
+The hash used is SHA-256.
+The time is in RFC 3339 format, preferably in the UTC timezone.
+*/
+
+// Decoded JSON
+type jsonData struct {
+	feedMu sync.RWMutex
+	pageMu sync.RWMutex
+	Feeds  map[string]*gofeed.Feed `json:"feeds,omitempty"`
+	Pages  map[string]*pageJson    `json:"pages,omitempty"`
+}
+
+type pageJson struct {
+	Hash    string    `json:"hash"`
+	Updated time.Time `json:"updated"`
+}
+
+var data jsonData // Global instance of jsonData - loaded from JSON and used
+
+// PageEntry is a single item on a feed page.
+// It is used both for tracked feeds and pages.
+type PageEntry struct {
+	Author    string
+	Title     string
+	URL       string
+	Published time.Time
+}
+
+// PageEntries is new-to-old list of Entry structs, used to create a feed page.
+// It should always be assumed to be sorted when used in other packages.
+type PageEntries struct {
+	sync.RWMutex
+	Entries []*PageEntry
+}
+
+// Implement sort.Interface
+
+func (e *PageEntries) Len() int {
+	e.RLock()
+	defer e.RUnlock()
+	return len(e.Entries)
+}
+
+func (e *PageEntries) Less(i, j int) bool {
+	e.RLock()
+	defer e.RUnlock()
+	return e.Entries[i].Published.Before(e.Entries[j].Published)
+}
+
+func (e *PageEntries) Swap(i, j int) {
+	e.Lock()
+	e.Entries[i], e.Entries[j] = e.Entries[j], e.Entries[i]
+	e.Unlock()
+}