refactor fetch functions to insert into FullData

This commit is contained in:
nytpu
2021-03-18 10:19:07 -06:00
parent e1a5c68ec7
commit 5ba676ed5d
4 changed files with 104 additions and 127 deletions
+37 -42
View File
@@ -9,68 +9,63 @@ package fetch
import ( import (
"fmt" "fmt"
"net/url" "net/url"
"reflect"
"golang.nytpu.com/comitium/core" "golang.nytpu.com/comitium/core"
) )
// NewFeed will fetch and create a new core.Feed given any url // Feed will fetch and insert a new core.Feed into a core.FullData given any url
func NewFeed(remoteURL *url.URL, title string) (*core.Feed, error) { // (or update if it's preexisting)
switch remoteURL.Scheme { func Feed(data *core.FullData, remote *url.URL, title string) error {
switch remote.Scheme {
case "gemini": case "gemini":
return newGeminiFeed(remoteURL, title) return geminiFeed(data, remote, title)
case "http", "https": case "http", "https":
return newHTTPFeed(remoteURL, title) return httpFeed(data, remote, title)
case "gopher", "gophers": case "gopher", "gophers":
return newGopherFeed(remoteURL, title) return gopherFeed(data, remote, title)
default: default:
return nil, fmt.Errorf("Unsupported protocol '%s'", remoteURL.Scheme) return fmt.Errorf("Unsupported protocol '%s'", remote.Scheme)
} }
} }
// NewPage will fetch and create a new core.Page given any url // Page will fetch and insert a new core.Page into a core.FullData given any url
func NewPage(remoteURL *url.URL, title string) (*core.Page, error) { // (or update if it's preexisting)
switch remoteURL.Scheme { func Page(data *core.FullData, remote *url.URL, title string) error {
switch remote.Scheme {
case "gemini": case "gemini":
return newGeminiPage(remoteURL, title) return geminiPage(data, remote, title)
case "http", "https": case "http", "https":
return newHTTPPage(remoteURL, title) return httpPage(data, remote, title)
case "gopher", "gophers": case "gopher", "gophers":
return newGopherPage(remoteURL, title) return gopherPage(data, remote, title)
default: default:
return nil, fmt.Errorf("Unsupported protocol '%s'", remoteURL.Scheme) return fmt.Errorf("Unsupported protocol '%s'", remote.Scheme)
} }
} }
// UpdateFeed will check an existing core.Feed for updates // InsertFeed will insert or update a core.Feed in a core.FullData
func UpdateFeed(f *core.Feed) error { func InsertFeed(data *core.FullData, new *core.Feed, remote *url.URL) {
switch f.FeedLink.Scheme { data.FeedsMu.Lock()
case "gemini": oldFeed, ok := data.Feeds[*remote]
// return updateGeminiFeed(f) if !ok || !reflect.DeepEqual(new, oldFeed) {
// TODO // Feeds are different, or there was never an old one
fallthrough data.Feeds[*remote] = new
case "http", "https": data.FeedsMu.Unlock()
// return updateHTTPFeed(f) } else {
// TODO data.FeedsMu.Unlock()
fallthrough
case "gopher", "gophers":
return updateGopherFeed(f)
default:
return fmt.Errorf("Unsupported protocol '%s'", f.FeedLink.Scheme)
} }
return nil
} }
// UpdatePage will check an existing core.Page for updages // InsertFeed will insert or update a core.Page in a core.FullData
func UpdatePage(p *core.Page) error { func InsertPage(data *core.FullData, new *core.Page, remote *url.URL) {
switch p.Link.Scheme { data.PagesMu.Lock()
case "gemini": _, ok := data.Pages[*remote]
return updateGeminiPage(p) if !ok || data.Pages[*remote].Hash != new.Hash {
case "http", "https": // pages are different, or there was never an old one
return updateHTTPPage(p) data.Pages[*remote] = new
case "gopher", "gophers": data.PagesMu.Unlock()
return updateGopherPage(p) } else {
default: data.PagesMu.Unlock()
return fmt.Errorf("Unsupported protocol '%s'", p.Link.Scheme)
} }
return nil
} }
+31 -31
View File
@@ -27,7 +27,7 @@ import (
) )
// fetchGemini will make a request for a gemini resource // fetchGemini will make a request for a gemini resource
func fetchGemini(remoteURL *url.URL) (*gemini.Response, error) { func fetchGemini(remote *url.URL) (*gemini.Response, error) {
client := &gemini.Client{} client := &gemini.Client{}
resp, err := client.Get(context.Background(), remoteURL.String()) resp, err := client.Get(context.Background(), remoteURL.String())
if err != nil { if err != nil {
@@ -40,17 +40,18 @@ func fetchGemini(remoteURL *url.URL) (*gemini.Response, error) {
return resp, nil return resp, nil
} }
// newGeminiFeed will fetch and create a new core.Feed given a gemini url // geminiFeed will fetch and insert a new core.Feed into a core.FullData given
func newGeminiFeed(remoteURL *url.URL, title string) (*core.Feed, error) { // a gemini url (or update if it's preexisting)
resp, err := fetchGemini(remoteURL) func geminiFeed(data *core.FullData, remote *url.URL, title string) error {
resp, err := fetchGemini(remote)
if err != nil { if err != nil {
return nil, err return err
} }
defer resp.Body.Close() defer resp.Body.Close()
mimetype, _, err := mime.ParseMediaType(resp.Meta) mimetype, _, err := mime.ParseMediaType(resp.Meta)
if err != nil { if err != nil {
return nil, fmt.Errorf("Unintelligible content type: %s", resp.Meta) return fmt.Errorf("Unintelligible content type: %s", resp.Meta)
} }
reader := io.LimitReader(resp.Body, 1073741824) // 1 GiB reader := io.LimitReader(resp.Body, 1073741824) // 1 GiB
@@ -59,11 +60,12 @@ func newGeminiFeed(remoteURL *url.URL, title string) (*core.Feed, error) {
case "text/gemini": case "text/gemini":
var feed core.Feed var feed core.Feed
feed.Title = title // will be set later if given title is "" feed.Title = title // will be set later if given title is ""
feed.Link = remoteURL feed.Link = remote
feed.FeedLink = remoteURL feed.FeedLink = remote
text, err := gemini.ParseText(resp.Body) text, err := gemini.ParseText(resp.Body)
if err != nil { if err != nil {
return nil, err return err
} }
for _, line := range text { for _, line := range text {
switch line := line.(type) { switch line := line.(type) {
@@ -84,7 +86,7 @@ func newGeminiFeed(remoteURL *url.URL, title string) (*core.Feed, error) {
if err != nil { if err != nil {
continue continue
} }
link = remoteURL.ResolveReference(link) link = remote.ResolveReference(link)
item := &core.Item{} item := &core.Item{}
item.Title = strings.TrimLeft(line.Name[10:], ": ") item.Title = strings.TrimLeft(line.Name[10:], ": ")
item.Published = date item.Published = date
@@ -92,7 +94,8 @@ func newGeminiFeed(remoteURL *url.URL, title string) (*core.Feed, error) {
feed.Items = append(feed.Items, item) feed.Items = append(feed.Items, item)
} }
} }
return &feed, nil InsertFeed(data, &feed, remote)
return nil
case "text/xml", case "text/xml",
"application/rss+xml", "application/rss+xml",
"application/atom+xml", "application/atom+xml",
@@ -100,30 +103,22 @@ func newGeminiFeed(remoteURL *url.URL, title string) (*core.Feed, error) {
"application/json": "application/json":
gofeed, err := gofeed.NewParser().Parse(reader) gofeed, err := gofeed.NewParser().Parse(reader)
if err != nil { if err != nil {
return nil, err return err
} }
feed, err := core.GofeedConvert(gofeed, title) feed, err := core.GofeedConvert(gofeed, title)
return feed, err
default:
return nil, fmt.Errorf("Cannot interpret %s as a feed", resp.Meta)
}
}
// newGeminiPage will create a new core.Page given a gemini url
func newGeminiPage(remoteURL *url.URL, title string) (*core.Page, error) {
var page core.Page
page.Title = title
page.Link = remoteURL
err := updateGeminiPage(&page)
if err != nil { if err != nil {
return nil, err return err
}
InsertFeed(data, feed, remote)
return nil
default:
return fmt.Errorf("Cannot interpret %s as a feed", resp.Meta)
} }
return &page, nil
} }
// updateGeminiPage will check an existing core.Page for changes // geminiPage will fetch and insert a new core.Page into a core.FullData given
func updateGeminiPage(p *core.Page) error { // a gemini url (or update if it's preexisting)
func geminiPage(data *core.FullData, remote *url.URL, title string) error {
resp, err := fetchGemini(p.Link) resp, err := fetchGemini(p.Link)
if err != nil { if err != nil {
return err return err
@@ -132,14 +127,19 @@ func updateGeminiPage(p *core.Page) error {
reader := io.LimitReader(resp.Body, 1073741824) // 1 GiB reader := io.LimitReader(resp.Body, 1073741824) // 1 GiB
var page core.Page
page.Title = title
page.Link = remote
h := sha256.New() h := sha256.New()
if _, err := io.Copy(h, reader); err != nil { if _, err := io.Copy(h, reader); err != nil {
return err return err
} }
newHash := fmt.Sprintf("%x", h.Sum(nil)) newHash := fmt.Sprintf("%x", h.Sum(nil))
if newHash != p.Hash { if newHash != p.Hash {
p.Hash = newHash page.Hash = newHash
p.Updated = time.Now() page.Updated = time.Now()
} }
InsertPage(data, &page, remote)
return nil return nil
} }
+14 -29
View File
@@ -12,7 +12,6 @@ import (
"net/url" "net/url"
"time" "time"
//"github.com/mmcdole/gofeed"
"github.com/prologic/go-gopher" "github.com/prologic/go-gopher"
"golang.nytpu.com/comitium/core" "golang.nytpu.com/comitium/core"
@@ -27,34 +26,17 @@ func fetchGopher(remoteURL *url.URL) (*gopher.Response, error) {
return resp, nil return resp, nil
} }
// newGopherFeed will fetch and create a new core.Feed given a gopher url // gopherFeed will fetch and insert a new core.Feed into a core.FullData given a
func newGopherFeed(remoteURL *url.URL, title string) (*core.Feed, error) { // gopher url (or update if it's preexisting)
// TODO func gopherFeed(data *core.FullData, remote *url.URL, title string) error {
return nil, fmt.Errorf("Feeds served over gopher are currently unsupported.")
}
// updateGopherFeed will check an existing core.Feed for updates
func updateGopherFeed(*core.Feed) error {
// TODO // TODO
return fmt.Errorf("Feeds served over gopher are currently unsupported.") return fmt.Errorf("Feeds served over gopher are currently unsupported.")
} }
// newGopherPage will create a new core.Page given a gopher url // gopherPage will fetch and insert a new core.Page into a core.FullData given a
func newGopherPage(remoteURL *url.URL, title string) (*core.Page, error) { // gopher url (or update if it's preexisting)
var page core.Page func gopherPage(data *core.FullData, remote *url.URL, title string) error {
page.Title = title resp, err := fetchGopher(remote)
page.Link = remoteURL
err := updateGopherPage(&page)
if err != nil {
return nil, err
}
return &page, nil
}
// updateGopherPage wll check an existing core.Page for changes
func updateGopherPage(p *core.Page) error {
resp, err := fetchGopher(p.Link)
if err != nil { if err != nil {
return err return err
} }
@@ -71,14 +53,17 @@ func updateGopherPage(p *core.Page) error {
reader = io.LimitReader(resp.Body, 1073741824) // 1 GiB reader = io.LimitReader(resp.Body, 1073741824) // 1 GiB
} }
var page core.Page
page.Title = title
page.Link = remote
h := sha256.New() h := sha256.New()
if _, err := io.Copy(h, reader); err != nil { if _, err := io.Copy(h, reader); err != nil {
return err return err
} }
newHash := fmt.Sprintf("%x", h.Sum(nil)) newHash := fmt.Sprintf("%x", h.Sum(nil))
if newHash != p.Hash { page.Hash = newHash
p.Hash = newHash page.Updated = time.Now()
p.Updated = time.Now()
} InsertPage(data, &page, remote)
return nil return nil
} }
+23 -26
View File
@@ -48,44 +48,38 @@ func fetchHTTP(remoteURL *url.URL) (*http.Response, error) {
return resp, nil return resp, nil
} }
// newHTTPFeed will fetch and create a new core.Feed given an http url // httpFeed will fetch and insert a new core.Feed into a core.FullData given an
func newHTTPFeed(remoteURL *url.URL, title string) (*core.Feed, error) { // http url (or update if it's preexisting)
resp, err := fetchHTTP(remoteURL) func httpFeed(data *core.FullData, remote *url.URL, title string) error {
resp, err := fetchHTTP(remote)
if err != nil { if err != nil {
return nil, err return err
} }
defer resp.Body.Close() defer resp.Body.Close()
if resp.Header.Get("Content-Type") == "text/html" { if resp.Header.Get("Content-Type") == "text/html" {
// TODO // TODO
return nil, fmt.Errorf("Extracting feed from HTML pages is unimplemented") return fmt.Errorf("Extracting feed from HTML pages is unimplemented")
} }
reader := io.LimitReader(resp.Body, 1073741824) // 1 GiB reader := io.LimitReader(resp.Body, 1073741824) // 1 GiB
gofeed, err := gofeed.NewParser().Parse(reader) gofeed, err := gofeed.NewParser().Parse(reader)
if err != nil { if err != nil {
return nil, err return err
} }
feed, err := core.GofeedConvert(gofeed, title) feed, err := core.GofeedConvert(gofeed, title)
return feed, err
}
// newHTTPPage will create a new core.Page given an http url
func newHTTPPage(remoteURL *url.URL, title string) (*core.Page, error) {
var page core.Page
page.Title = title
page.Link = remoteURL
err := updateHTTPPage(&page)
if err != nil { if err != nil {
return nil, err return err
}
return &page, nil
} }
// updateHTTPPage will check an existing core.Page for changes InsertFeed(data, feed, remote)
func updateHTTPPage(p *core.Page) error { return nil
resp, err := fetchHTTP(p.Link) }
// httpPage will fetch and insert a new core.Page into a core.FullData given an
// http url (or update if it's preexisting)
func httpPage(data *core.FullData, remote *url.URL, title string) error {
resp, err := fetchHTTP(remote)
if err != nil { if err != nil {
return err return err
} }
@@ -93,14 +87,17 @@ func updateHTTPPage(p *core.Page) error {
reader := io.LimitReader(resp.Body, 1073741824) // 1 GiB reader := io.LimitReader(resp.Body, 1073741824) // 1 GiB
var page core.Page
page.Title = title
page.Link = remote
h := sha256.New() h := sha256.New()
if _, err := io.Copy(h, reader); err != nil { if _, err := io.Copy(h, reader); err != nil {
return err return err
} }
newHash := fmt.Sprintf("%x", h.Sum(nil)) newHash := fmt.Sprintf("%x", h.Sum(nil))
if newHash != p.Hash { page.Hash = newHash
p.Hash = newHash page.Updated = time.Now()
p.Updated = time.Now()
} InsertPage(data, &page, remote)
return nil return nil
} }