commit 4bd418416af3cfe715a58515ef93ce463a32345c · edavis.dev/monarch

-26

cmd/backfiller/backend.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       import (

     

       4
       -
       	"sync"

     

       5
       -
       

     

       6
       -
       	"github.com/bluesky-social/indigo/backfill"

     

       7
       -
       	"gorm.io/gorm"

     

       8
       -
       )

     

       9
       -
       

     

       10
       -
       type Backend struct {

     

       11
       -
       	state *gorm.DB

     

       12
       -
       	data  *gorm.DB

     

       13
       -
       	bf    *backfill.Backfiller

     

       14
       -
       

     

       15
       -
       	firehoseLk  sync.Mutex

     

       16
       -
       	firehoseSeq string

     

       17
       -
       	reposLk     sync.Mutex

     

       18
       -
       	reposSeq    string

     

       19
       -
       }

     

       20
       -
       

     

       21
       -
       func NewBackend(state, data *gorm.DB) *Backend {

     

       22
       -
       	return &Backend{

     

       23
       -
       		state: state,

     

       24
       -
       		data:  data,

     

       25
       -
       	}

     

       26
       -
       }

-66

cmd/backfiller/cursors.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       import (

     

       4
       -
       	"context"

     

       5
       -
       	"errors"

     

       6
       -
       	"fmt"

     

       7
       -
       	"log/slog"

     

       8
       -
       	"time"

     

       9
       -
       

     

       10
       -
       	"gorm.io/gorm"

     

       11
       -
       )

     

       12
       -
       

     

       13
       -
       type cursorRecord struct {

     

       14
       -
       	ID  uint   `gorm:"primaryKey"`

     

       15
       -
       	Key string `gorm:"unique"`

     

       16
       -
       	Val string

     

       17
       -
       }

     

       18
       -
       

     

       19
       -
       func (b *Backend) LoadCursor(key string) (string, error) {

     

       20
       -
       	var rec cursorRecord

     

       21
       -
       	if err := b.state.Where("key = ?", key).First(&rec).Error; err != nil {

     

       22
       -
       		if errors.Is(err, gorm.ErrRecordNotFound) {

     

       23
       -
       			b.state.Create(&cursorRecord{Key: key, Val: ""})

     

       24
       -
       		}

     

       25
       -
       		return "", err

     

       26
       -
       	}

     

       27
       -
       	return rec.Val, nil

     

       28
       -
       }

     

       29
       -
       

     

       30
       -
       func (b *Backend) FlushCursors() error {

     

       31
       -
       	sl := slog.With("source", "flushCursors")

     

       32
       -
       

     

       33
       -
       	b.firehoseLk.Lock()

     

       34
       -
       	sl.Info("persisting firehose cursor", "cursor", b.firehoseSeq)

     

       35
       -
       	if err := b.state.Model(&cursorRecord{}).Where("key = ?", "firehose").Update("val", b.firehoseSeq).Error; err != nil {

     

       36
       -
       		return fmt.Errorf("failed to persist firehose cursor: %w", err)

     

       37
       -
       	}

     

       38
       -
       	b.firehoseLk.Unlock()

     

       39
       -
       

     

       40
       -
       	b.reposLk.Lock()

     

       41
       -
       	sl.Info("persisting repos cursor", "cursor", b.reposSeq)

     

       42
       -
       	if err := b.state.Model(&cursorRecord{}).Where("key = ?", "repos").Update("val", b.reposSeq).Error; err != nil {

     

       43
       -
       		return fmt.Errorf("failed to persist repos cursor: %w", err)

     

       44
       -
       	}

     

       45
       -
       	b.reposLk.Unlock()

     

       46
       -
       

     

       47
       -
       	return nil

     

       48
       -
       }

     

       49
       -
       

     

       50
       -
       func (b *Backend) SyncCursors(ctx context.Context) error {

     

       51
       -
       	for range time.Tick(time.Second * 5) {

     

       52
       -
       		select {

     

       53
       -
       		case <-ctx.Done():

     

       54
       -
       			return nil

     

       55
       -
       		default:

     

       56
       -
       			//

     

       57
       -
       		}

     

       58
       -
       

     

       59
       -
       		if err := b.FlushCursors(); err != nil {

     

       60
       -
       			slog.Error("failed to flush cursors", "err", err)

     

       61
       -
       			return fmt.Errorf("failed to flush cursors: %w", err)

     

       62
       -
       		}

     

       63
       -
       	}

     

       64
       -
       

     

       65
       -
       	return nil

     

       66
       -
       }

-58

cmd/backfiller/database.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       import (

     

       4
       -
       	"fmt"

     

       5
       -
       	"log"

     

       6
       -
       	"log/slog"

     

       7
       -
       	"os"

     

       8
       -
       	"time"

     

       9
       -
       

     

       10
       -
       	"gorm.io/driver/sqlite"

     

       11
       -
       	"gorm.io/gorm"

     

       12
       -
       	"gorm.io/gorm/logger"

     

       13
       -
       )

     

       14
       -
       

     

       15
       -
       func NewDatabase(path string) *gorm.DB {

     

       16
       -
       	sl := slog.With("source", "database")

     

       17
       -
       	l := logger.New(

     

       18
       -
       		log.New(os.Stdout, "\r\n", log.LstdFlags),

     

       19
       -
       		logger.Config{

     

       20
       -
       			SlowThreshold: time.Second,

     

       21
       -
       			Colorful:      false,

     

       22
       -
       		},

     

       23
       -
       	)

     

       24
       -
       	db, err := gorm.Open(sqlite.Open(path), &gorm.Config{

     

       25
       -
       		Logger: l,

     

       26
       -
       	})

     

       27
       -
       	if err != nil {

     

       28
       -
       		sl.Error("failed to open database", "err", err)

     

       29
       -
       	}

     

       30
       -
       	db.Exec("PRAGMA journal_mode=WAL")

     

       31
       -
       	return db

     

       32
       -
       }

     

       33
       -
       

     

       34
       -
       func (b *Backend) CleanlyClose() error {

     

       35
       -
       	closeDb := func(db *gorm.DB) error {

     

       36
       -
       		if err := db.Exec("PRAGMA wal_checkpoint(TRUNCATE)").Error; err != nil {

     

       37
       -
       			return fmt.Errorf("failed checkpointing the WAL: %w", err)

     

       38
       -
       		}

     

       39
       -
       		rawDb, err := db.DB()

     

       40
       -
       		if err != nil {

     

       41
       -
       			return fmt.Errorf("failed getting underlying DB connection: %w", err)

     

       42
       -
       		}

     

       43
       -
       		if err := rawDb.Close(); err != nil {

     

       44
       -
       			return fmt.Errorf("failed closing underlying DB connection: %w", err)

     

       45
       -
       		}

     

       46
       -
       		return nil

     

       47
       -
       	}

     

       48
       -
       

     

       49
       -
       	if err := closeDb(b.state); err != nil {

     

       50
       -
       		return fmt.Errorf("failed to close state database: %w", err)

     

       51
       -
       	}

     

       52
       -
       

     

       53
       -
       	if err := closeDb(b.data); err != nil {

     

       54
       -
       		return fmt.Errorf("failed to close content database: %w", err)

     

       55
       -
       	}

     

       56
       -
       

     

       57
       -
       	return nil

     

       58
       -
       }

-74

cmd/backfiller/handlers.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       import (

     

       4
       -
       	"bytes"

     

       5
       -
       	"context"

     

       6
       -
       	"fmt"

     

       7
       -
       	"log/slog"

     

       8
       -
       	"strconv"

     

       9
       -
       	"strings"

     

       10
       -
       

     

       11
       -
       	comatproto "github.com/bluesky-social/indigo/api/atproto"

     

       12
       -
       	appbsky "github.com/bluesky-social/indigo/api/bsky"

     

       13
       -
       	"github.com/ipfs/go-cid"

     

       14
       -
       )

     

       15
       -
       

     

       16
       -
       type commitHandler func(context.Context, *comatproto.SyncSubscribeRepos_Commit) error

     

       17
       -
       

     

       18
       -
       func (b *Backend) RepoCommitHandler(

     

       19
       -
       	ctx context.Context, evt *comatproto.SyncSubscribeRepos_Commit,

     

       20
       -
       ) error {

     

       21
       -
       	select {

     

       22
       -
       	case <-ctx.Done():

     

       23
       -
       		return nil

     

       24
       -
       	default:

     

       25
       -
       		//

     

       26
       -
       	}

     

       27
       -
       

     

       28
       -
       	b.firehoseLk.Lock()

     

       29
       -
       	b.firehoseSeq = strconv.Itoa(int(evt.Seq))

     

       30
       -
       	b.firehoseLk.Unlock()

     

       31
       -
       

     

       32
       -
       	return b.bf.HandleEvent(ctx, evt)

     

       33
       -
       }

     

       34
       -
       

     

       35
       -
       type handleOpCreateUpdate func(context.Context, string, string, string, *[]byte, *cid.Cid) error

     

       36
       -
       type handleOpDelete func(context.Context, string, string, string) error

     

       37
       -
       

     

       38
       -
       func (b *Backend) HandleCreateOp(ctx context.Context, repo, rev, path string, rec *[]byte, cid *cid.Cid) error {

     

       39
       -
       	if !strings.HasPrefix(path, "app.bsky.feed.generator/") {

     

       40
       -
       		return nil

     

       41
       -
       	}

     

       42
       -
       

     

       43
       -
       	sl := slog.With("source", "HandleCreateOp")

     

       44
       -
       

     

       45
       -
       	var out appbsky.FeedGenerator

     

       46
       -
       	if err := out.UnmarshalCBOR(bytes.NewReader(*rec)); err != nil {

     

       47
       -
       		sl.Error("failed to unmarshal record", "err", err)

     

       48
       -
       		return fmt.Errorf("failed to unmarshal record: %w", err)

     

       49
       -
       	}

     

       50
       -
       

     

       51
       -
       	feedgen := &FeedGenerator{

     

       52
       -
       		AtUri:               fmt.Sprintf("at://%s/%s", repo, path),

     

       53
       -
       		DisplayName:         out.DisplayName,

     

       54
       -
       		FeedService:         out.Did,

     

       55
       -
       		CreatedAt:           out.CreatedAt,

     

       56
       -
       		Description:         out.Description,

     

       57
       -
       		ContentMode:         out.ContentMode,

     

       58
       -
       		AcceptsInteractions: out.AcceptsInteractions,

     

       59
       -
       	}

     

       60
       -
       

     

       61
       -
       	if err := b.data.Model(&FeedGenerator{}).Create(feedgen).Error; err != nil {

     

       62
       -
       		return fmt.Errorf("error adding feedgen to database: %w", err)

     

       63
       -
       	}

     

       64
       -
       

     

       65
       -
       	return nil

     

       66
       -
       }

     

       67
       -
       

     

       68
       -
       func (b *Backend) HandleUpdateOp(ctx context.Context, repo, rev, path string, rec *[]byte, cid *cid.Cid) error {

     

       69
       -
       	return nil

     

       70
       -
       }

     

       71
       -
       

     

       72
       -
       func (b *Backend) HandleDeleteOp(ctx context.Context, repo, rev, path string) error {

     

       73
       -
       	return nil

     

       74
       -
       }

-154

cmd/backfiller/main.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       import (

     

       4
       -
       	"context"

     

       5
       -
       	"log/slog"

     

       6
       -
       	"net/http"

     

       7
       -
       	"os"

     

       8
       -
       	"os/signal"

     

       9
       -
       	"syscall"

     

       10
       -
       	"time"

     

       11
       -
       

     

       12
       -
       	comatproto "github.com/bluesky-social/indigo/api/atproto"

     

       13
       -
       	"github.com/bluesky-social/indigo/backfill"

     

       14
       -
       	"github.com/bluesky-social/indigo/events"

     

       15
       -
       	"github.com/bluesky-social/indigo/events/schedulers/parallel"

     

       16
       -
       

     

       17
       -
       	"github.com/gorilla/websocket"

     

       18
       -
       	"gorm.io/gorm"

     

       19
       -
       )

     

       20
       -
       

     

       21
       -
       func NewBackfiller(

     

       22
       -
       	db *gorm.DB, create handleOpCreateUpdate, update handleOpCreateUpdate, delete handleOpDelete,

     

       23
       -
       ) *backfill.Backfiller {

     

       24
       -
       	opts := &backfill.BackfillOptions{

     

       25
       -
       		// ParallelBackfills:     50,

     

       26
       -
       		// ParallelRecordCreates: 25,

     

       27
       -
       		// SyncRequestsPerSecond: 25,

     

       28
       -
       

     

       29
       -
       		ParallelBackfills:     10,

     

       30
       -
       		ParallelRecordCreates: 1, // sqlite

     

       31
       -
       		SyncRequestsPerSecond: 5,

     

       32
       -
       

     

       33
       -
       		// NSIDFilter: "app.bsky.feed.generator",

     

       34
       -
       		RelayHost: "https://bsky.network",

     

       35
       -
       	}

     

       36
       -
       

     

       37
       -
       	return backfill.NewBackfiller(

     

       38
       -
       		"backfills",

     

       39
       -
       		backfill.NewGormstore(db),

     

       40
       -
       		create, update, delete,

     

       41
       -
       		opts,

     

       42
       -
       	)

     

       43
       -
       }

     

       44
       -
       

     

       45
       -
       func NewFirehose(ctx context.Context, cursor string) *websocket.Conn {

     

       46
       -
       	sl := slog.With("source", "firehose")

     

       47
       -
       	subscribeUrl := "wss://bsky.network/xrpc/com.atproto.sync.subscribeRepos"

     

       48
       -
       	if cursor != "" {

     

       49
       -
       		subscribeUrl += "?cursor=" + cursor

     

       50
       -
       	}

     

       51
       -
       

     

       52
       -
       	conn, _, err := websocket.DefaultDialer.DialContext(ctx, subscribeUrl, http.Header{

     

       53
       -
       		"User-Agent": []string{"backfiller/0.1 (@edavis.dev)"},

     

       54
       -
       	})

     

       55
       -
       	if err != nil {

     

       56
       -
       		sl.Error("failed to connect to relay", "err", err)

     

       57
       -
       	}

     

       58
       -
       

     

       59
       -
       	return conn

     

       60
       -
       }

     

       61
       -
       

     

       62
       -
       func NewScheduler(

     

       63
       -
       	ctx context.Context, commitCallback commitHandler,

     

       64
       -
       ) *parallel.Scheduler {

     

       65
       -
       	rsc := events.RepoStreamCallbacks{

     

       66
       -
       		RepoCommit: func(evt *comatproto.SyncSubscribeRepos_Commit) error {

     

       67
       -
       			return commitCallback(ctx, evt)

     

       68
       -
       		},

     

       69
       -
       	}

     

       70
       -
       

     

       71
       -
       	return parallel.NewScheduler(16, 100, "firehose", rsc.EventHandler)

     

       72
       -
       }

     

       73
       -
       

     

       74
       -
       func main() {

     

       75
       -
       	sl := slog.With("source", "backfiller")

     

       76
       -
       

     

       77
       -
       	streamClosed := make(chan struct{})

     

       78
       -
       	streamCtx, streamCancel := context.WithCancel(context.Background())

     

       79
       -
       

     

       80
       -
       	stateDb := NewDatabase("state.db")

     

       81
       -
       	stateDb.AutoMigrate(&backfill.GormDBJob{})

     

       82
       -
       	stateDb.AutoMigrate(&cursorRecord{})

     

       83
       -
       

     

       84
       -
       	contentDb := NewDatabase("database.db")

     

       85
       -
       	contentDb.AutoMigrate(&FeedGenerator{})

     

       86
       -
       

     

       87
       -
       	backend := NewBackend(stateDb, contentDb)

     

       88
       -
       

     

       89
       -
       	bf := NewBackfiller(stateDb, backend.HandleCreateOp, backend.HandleUpdateOp, backend.HandleDeleteOp)

     

       90
       -
       	go bf.Start()

     

       91
       -
       

     

       92
       -
       	// attach the backfiller to the backend so the curors, the pump,

     

       93
       -
       	// and the commit callback can use it

     

       94
       -
       	backend.bf = bf

     

       95
       -
       

     

       96
       -
       	go backend.SyncCursors(streamCtx)

     

       97
       -
       

     

       98
       -
       	cursor, err := backend.LoadCursor("firehose")

     

       99
       -
       	if err != nil {

     

       100
       -
       		sl.Error("failed loading firehose cursor", "err", err)

     

       101
       -
       	}

     

       102
       -
       	conn := NewFirehose(streamCtx, cursor)

     

       103
       -
       

     

       104
       -
       	sched := NewScheduler(streamCtx, backend.RepoCommitHandler)

     

       105
       -
       	go func() {

     

       106
       -
       		if err := events.HandleRepoStream(streamCtx, conn, sched, sl); err != nil {

     

       107
       -
       			sl.Error("failed to start scheduler", "err", err)

     

       108
       -
       		}

     

       109
       -
       		close(streamClosed)

     

       110
       -
       	}()

     

       111
       -
       

     

       112
       -
       	go func() {

     

       113
       -
       		if err := backend.PumpRepos(streamCtx); err != nil {

     

       114
       -
       			sl.Error("failed pumping repos", "err", err)

     

       115
       -
       		}

     

       116
       -
       	}()

     

       117
       -
       

     

       118
       -
       	quit := make(chan struct{})

     

       119
       -
       	exitSignals := make(chan os.Signal, 1)

     

       120
       -
       	signal.Notify(exitSignals, syscall.SIGINT, syscall.SIGTERM)

     

       121
       -
       	go func() {

     

       122
       -
       		select {

     

       123
       -
       		case sig := <-exitSignals:

     

       124
       -
       			sl.Info("received OS exit signal", "signal", sig)

     

       125
       -
       		case <-streamClosed:

     

       126
       -
       			//

     

       127
       -
       		}

     

       128
       -
       

     

       129
       -
       		conn.Close()

     

       130
       -
       

     

       131
       -
       		streamCancel()

     

       132
       -
       		<-streamClosed

     

       133
       -
       

     

       134
       -
       		time.Sleep(time.Millisecond * 100)

     

       135
       -
       

     

       136
       -
       		endctx, cancel := context.WithTimeout(context.TODO(), time.Minute)

     

       137
       -
       		defer cancel()

     

       138
       -
       		bf.Stop(endctx)

     

       139
       -
       

     

       140
       -
       		close(quit)

     

       141
       -
       	}()

     

       142
       -
       

     

       143
       -
       	<-quit

     

       144
       -
       

     

       145
       -
       	sl.Info("flushing cursors")

     

       146
       -
       	if err := backend.FlushCursors(); err != nil {

     

       147
       -
       		sl.Error("failed to flush cursors on close", "err", err)

     

       148
       -
       	}

     

       149
       -
       

     

       150
       -
       	sl.Info("closing databases")

     

       151
       -
       	if err := backend.CleanlyClose(); err != nil {

     

       152
       -
       		sl.Error("failed to close databases", "err", err)

     

       153
       -
       	}

     

       154
       -
       }

-12

cmd/backfiller/models.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       type FeedGenerator struct {

     

       4
       -
       	ID                  uint   `gorm:"primaryKey"`

     

       5
       -
       	AtUri               string `gorm:"unique"`

     

       6
       -
       	DisplayName         string

     

       7
       -
       	FeedService         string

     

       8
       -
       	Description         *string

     

       9
       -
       	ContentMode         *string

     

       10
       -
       	AcceptsInteractions *bool

     

       11
       -
       	CreatedAt           string

     

       12
       -
       }

-69

cmd/backfiller/pump.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       import (

     

       4
       -
       	"context"

     

       5
       -
       	"fmt"

     

       6
       -
       	"log/slog"

     

       7
       -
       

     

       8
       -
       	"github.com/bluesky-social/indigo/api/atproto"

     

       9
       -
       	"github.com/bluesky-social/indigo/backfill"

     

       10
       -
       	"github.com/bluesky-social/indigo/xrpc"

     

       11
       -
       )

     

       12
       -
       

     

       13
       -
       type jobMaker interface {

     

       14
       -
       	GetOrCreateJob(context.Context, string, string) (backfill.Job, error)

     

       15
       -
       }

     

       16
       -
       

     

       17
       -
       func (b *Backend) PumpRepos(ctx context.Context) error {

     

       18
       -
       	sl := slog.With("source", "pumpRepos")

     

       19
       -
       	bf := b.bf

     

       20
       -
       

     

       21
       -
       	xrpcc := &xrpc.Client{

     

       22
       -
       		Host: "https://bsky.network",

     

       23
       -
       	}

     

       24
       -
       

     

       25
       -
       	jmstore, ok := bf.Store.(jobMaker)

     

       26
       -
       	if !ok {

     

       27
       -
       		return fmt.Errorf("configured job store doesn't support random job creation")

     

       28
       -
       	}

     

       29
       -
       

     

       30
       -
       	curs, err := b.LoadCursor("repos")

     

       31
       -
       	if err != nil {

     

       32
       -
       		sl.Error("failed to load repos cursor", "err", err)

     

       33
       -
       	}

     

       34
       -
       

     

       35
       -
       	for {

     

       36
       -
       		select {

     

       37
       -
       		case <-ctx.Done():

     

       38
       -
       			sl.Info("stopping repo pump")

     

       39
       -
       			return nil

     

       40
       -
       		default:

     

       41
       -
       			//

     

       42
       -
       		}

     

       43
       -
       

     

       44
       -
       		sl.Info("listing repos", "cursor", curs)

     

       45
       -
       		res, err := atproto.SyncListRepos(ctx, xrpcc, curs, 1000)

     

       46
       -
       		if err != nil {

     

       47
       -
       			return fmt.Errorf("error listing repos: %w", err)

     

       48
       -
       		}

     

       49
       -
       

     

       50
       -
       		for _, repo := range res.Repos {

     

       51
       -
       			_, err := jmstore.GetOrCreateJob(ctx, repo.Did, backfill.StateEnqueued)

     

       52
       -
       			if err != nil {

     

       53
       -
       				sl.Warn("failed to create backfill job", "err", err)

     

       54
       -
       				continue

     

       55
       -
       			}

     

       56
       -
       		}

     

       57
       -
       

     

       58
       -
       		if res.Cursor != nil && *res.Cursor != "" {

     

       59
       -
       			curs = *res.Cursor

     

       60
       -
       			b.reposLk.Lock()

     

       61
       -
       			b.reposSeq = curs

     

       62
       -
       			b.reposLk.Unlock()

     

       63
       -
       		} else {

     

       64
       -
       			break

     

       65
       -
       		}

     

       66
       -
       	}

     

       67
       -
       

     

       68
       -
       	return nil

     

       69
       -
       }