285 lines
5.2 KiB
Go
285 lines
5.2 KiB
Go
package main
|
|
|
|
import (
|
|
"cmp"
|
|
"context"
|
|
"database/sql"
|
|
"fmt"
|
|
"iter"
|
|
"net/http"
|
|
"net/url"
|
|
"slices"
|
|
"strings"
|
|
"time"
|
|
|
|
_ "embed"
|
|
|
|
"go.yarn.social/lextwt"
|
|
"go.yarn.social/types"
|
|
)
|
|
|
|
type Feed struct {
|
|
FeedID uuid
|
|
FetchURI string
|
|
URI string
|
|
Nick string
|
|
LastScanOn sql.NullTime
|
|
RefreshRate int
|
|
|
|
LastModified sql.NullTime
|
|
LastError sql.NullString
|
|
ETag sql.NullString
|
|
|
|
Version string
|
|
DiscloseFeedURL string
|
|
DiscloseNick string
|
|
FirstFetch bool
|
|
|
|
State State
|
|
}
|
|
|
|
type State string
|
|
|
|
const (
|
|
PermanentlyDead State = "permanantly-dead"
|
|
Frozen State = "frozen"
|
|
Cold State = "cold"
|
|
Warm State = "warm"
|
|
Hot State = "hot"
|
|
)
|
|
|
|
var (
|
|
//go:embed init.sql
|
|
initSQL string
|
|
|
|
insertFeed = `
|
|
insert into feeds
|
|
(feed_id, uri, nick, last_scan_on, refresh_rate)
|
|
values (?, ?, ?, ?, ?)
|
|
ON CONFLICT (feed_id) DO NOTHING
|
|
`
|
|
|
|
insertTwt = `
|
|
insert into twts
|
|
(feed_id, hash, conv, dt, text, mentions, tags)
|
|
values (?, ?, ?, ?, ?, ?, ?)
|
|
ON CONFLICT (feed_id, hash) DO NOTHING
|
|
`
|
|
|
|
fetchFeeds = `
|
|
select
|
|
feed_id,
|
|
uri,
|
|
nick,
|
|
last_scan_on,
|
|
refresh_rate,
|
|
last_modified_on,
|
|
last_etag
|
|
from feeds
|
|
where datetime(last_scan_on, '+'||refresh_rate||' seconds') < datetime(current_timestamp, '+10 minutes')
|
|
`
|
|
updateFeed = `
|
|
update feeds set
|
|
last_scan_on = ?,
|
|
refresh_rate = ?,
|
|
last_modified_on = ?,
|
|
last_etag = ?,
|
|
last_error = ?
|
|
where feed_id = ?
|
|
`
|
|
)
|
|
|
|
func (f *Feed) Save(ctx context.Context, db *sql.DB) error {
|
|
fmt.Println(f.FetchURI, " ", f.LastModified, " ", f.LastError)
|
|
_, err := db.ExecContext(
|
|
ctx,
|
|
updateFeed,
|
|
f.LastScanOn,
|
|
f.RefreshRate,
|
|
f.LastModified,
|
|
f.ETag,
|
|
f.LastError,
|
|
f.FeedID,
|
|
)
|
|
return err
|
|
}
|
|
|
|
func (f *Feed) Scan(res interface{ Scan(...any) error }) error {
|
|
f.State = "load"
|
|
var err error
|
|
|
|
f.Version = "0.0.1"
|
|
err = res.Scan(
|
|
&f.FeedID,
|
|
&f.URI,
|
|
&f.Nick,
|
|
&f.LastScanOn,
|
|
&f.RefreshRate,
|
|
&f.LastModified,
|
|
&f.ETag,
|
|
)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
if !f.LastScanOn.Valid {
|
|
f.FirstFetch = true
|
|
f.LastScanOn.Time = time.Now()
|
|
f.LastScanOn.Valid = true
|
|
} else {
|
|
f.LastScanOn.Time = f.LastScanOn.Time.Add(time.Duration(f.RefreshRate) * time.Second)
|
|
}
|
|
|
|
f.FetchURI = f.URI
|
|
return err
|
|
}
|
|
|
|
func loadFeeds(ctx context.Context, db *sql.DB) (iter.Seq[Feed], error) {
|
|
var err error
|
|
var res *sql.Rows
|
|
|
|
res, err = db.QueryContext(ctx, fetchFeeds)
|
|
|
|
if err != nil {
|
|
return slices.Values([]Feed{}), err
|
|
}
|
|
|
|
return func(yield func(Feed) bool) {
|
|
for res.Next() {
|
|
var f Feed
|
|
err = f.Scan(res)
|
|
if err != nil {
|
|
return
|
|
}
|
|
if !yield(f) {
|
|
return
|
|
}
|
|
}
|
|
}, err
|
|
}
|
|
|
|
func storeFeed(db *sql.DB, f types.TwtFile) error {
|
|
loadTS := time.Now()
|
|
refreshRate := 600
|
|
|
|
feedID := urlNS.UUID5(cmp.Or(f.Twter().HashingURI, f.Twter().URI))
|
|
|
|
tx, err := db.Begin()
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
followers := f.Info().GetAll("follow")
|
|
followMap := make(map[string]string, len(followers))
|
|
for _, f := range f.Info().GetAll("follow") {
|
|
nick, uri, ok := strings.Cut(f.Value(), "http")
|
|
if !ok {
|
|
continue
|
|
}
|
|
nick = strings.TrimSpace(nick)
|
|
uri = "http" + strings.TrimSpace(uri)
|
|
|
|
if _, err := url.Parse(uri); err != nil {
|
|
continue
|
|
}
|
|
|
|
followMap[nick] = uri
|
|
}
|
|
|
|
defer tx.Rollback()
|
|
|
|
_, err = tx.Exec(
|
|
insertFeed,
|
|
feedID,
|
|
f.Twter().HashingURI,
|
|
f.Twter().DomainNick(),
|
|
loadTS,
|
|
refreshRate,
|
|
)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
for _, twt := range f.Twts() {
|
|
mentions := make(uuids, 0, len(twt.Mentions()))
|
|
for _, mention := range twt.Mentions() {
|
|
followMap[mention.Twter().Nick] = mention.Twter().URI
|
|
mentions = append(mentions, urlNS.UUID5(mention.Twter().URI))
|
|
}
|
|
|
|
tags := make(strList, 0, len(twt.Tags()))
|
|
for _, tag := range twt.Tags() {
|
|
tags = append(tags, tag.Text())
|
|
}
|
|
|
|
subject := twt.Subject()
|
|
subjectTag := ""
|
|
if subject != nil {
|
|
if tag, ok := subject.Tag().(*lextwt.Tag); ok && tag != nil {
|
|
subjectTag = tag.Text()
|
|
}
|
|
}
|
|
|
|
_, err = tx.Exec(
|
|
insertTwt,
|
|
feedID,
|
|
twt.Hash(),
|
|
subjectTag,
|
|
twt.Created(),
|
|
fmt.Sprint(twt),
|
|
mentions.ToStrList(),
|
|
tags,
|
|
)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
}
|
|
|
|
for nick, uri := range followMap {
|
|
_, err = tx.Exec(
|
|
insertFeed,
|
|
urlNS.UUID5(uri),
|
|
uri,
|
|
nick,
|
|
nil,
|
|
refreshRate,
|
|
)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
}
|
|
|
|
return tx.Commit()
|
|
}
|
|
|
|
func (feed *Feed) MakeHTTPRequest(ctx context.Context) (*http.Request, error) {
|
|
feed.State = "fetch"
|
|
if strings.Contains(feed.FetchURI, "lublin.se") {
|
|
return nil, fmt.Errorf("%w: permaban: %s", ErrPermanentlyDead, feed.URI)
|
|
}
|
|
|
|
req, err := http.NewRequestWithContext(ctx, "GET", feed.FetchURI, nil)
|
|
if err != nil {
|
|
return nil, fmt.Errorf("creating HTTP request failed: %w", err)
|
|
}
|
|
|
|
req.Header.Add("Accept", "text/plain")
|
|
|
|
if !feed.LastModified.Valid {
|
|
req.Header.Add("If-Modified-Since", feed.LastModified.Time.Format(http.TimeFormat))
|
|
}
|
|
|
|
if feed.ETag.Valid {
|
|
req.Header.Add("If-None-Match", feed.ETag.String)
|
|
}
|
|
|
|
if feed.DiscloseFeedURL != "" && feed.DiscloseNick != "" {
|
|
req.Header.Set("User-Agent", fmt.Sprintf("xt/%s (+%s; @%s)",
|
|
feed.Version, feed.DiscloseFeedURL, feed.DiscloseNick))
|
|
} else {
|
|
req.Header.Set("User-Agent", fmt.Sprintf("xt/%s", feed.Version))
|
|
}
|
|
|
|
return req, nil
|
|
}
|