commit caba3915467df869afceb62b12b2a1570c8dc17f · edavis.dev/monarch

+21 -3

cmd/monarch/census.go

···

       17
        
       	cursor   *CursorService

     

       18
        
       	backfill *backfill.Backfiller

     

       19
        
       

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       20
        
       	storeLk sync.Mutex

     

       21
        
       }

     

       22
        
       

     
···

       26
        
       

     

       27
        
       func NewCensusService(cursorSvc *CursorService, backfillSvc *backfill.Backfiller) *CensusService {

     

       28
        
       	return &CensusService{

     

       29
       -
       		cursor:   cursorSvc,

     

       30
       -
       		backfill: backfillSvc,

     

       0
        
       
     

       31
        
       	}

     

       32
        
       }

     

       33
        
       

     
···

       65
        
       		}

     

       66
        
       

     

       67
        
       		for _, host := range res.Hosts {

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       68
        
       			sem.Acquire(ctx, 1)

     

       69
        
       			wg.Add(1) // TODO wg.Go

     

       70
        
       			go func() {

     
···

       149
        
       	}

     

       150
        
       

     

       151
        
       	slog.Info("finished listing repos", "host", host)

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       152
        
       }

     

       153
        
       

     

       154
        
       func (cs *CensusService) Start(ctx context.Context, cctx *cli.Context) {

     
···

       156
        
       	cs.listHosts(ctx, cctx)

     

       157
        
       

     

       158
        
       	slog.Info("finished with initial refresh, starting ticker")

     

       159
       -
       	t := time.NewTicker(time.Hour)

     

       160
        
       	defer t.Stop()

     

       161
        
       

     

       162
        
       	for {

···

       17
        
       	cursor   *CursorService

     

       18
        
       	backfill *backfill.Backfiller

     

       19
        
       

     

       20
       +
       	seenHosts map[string]bool

     

       21
       +
       	seenLk    sync.Mutex

     

       22
       +
       

     

       23
        
       	storeLk sync.Mutex

     

       24
        
       }

     

       25
        
       

     
···

       29
        
       

     

       30
        
       func NewCensusService(cursorSvc *CursorService, backfillSvc *backfill.Backfiller) *CensusService {

     

       31
        
       	return &CensusService{

     

       32
       +
       		cursor:    cursorSvc,

     

       33
       +
       		backfill:  backfillSvc,

     

       34
       +
       		seenHosts: make(map[string]bool),

     

       35
        
       	}

     

       36
        
       }

     

       37
        
       

     
···

       69
        
       		}

     

       70
        
       

     

       71
        
       		for _, host := range res.Hosts {

     

       72
       +
       			// don't reprocess hosts already handled

     

       73
       +
       			cs.seenLk.Lock()

     

       74
       +
       			_, ok := cs.seenHosts[host.Hostname]

     

       75
       +
       			cs.seenLk.Unlock()

     

       76
       +
       			if ok {

     

       77
       +
       				slog.Info("already processed host, skipping")

     

       78
       +
       				continue

     

       79
       +
       			}

     

       80
       +
       

     

       81
        
       			sem.Acquire(ctx, 1)

     

       82
        
       			wg.Add(1) // TODO wg.Go

     

       83
        
       			go func() {

     
···

       162
        
       	}

     

       163
        
       

     

       164
        
       	slog.Info("finished listing repos", "host", host)

     

       165
       +
       

     

       166
       +
       	cs.seenLk.Lock()

     

       167
       +
       	defer cs.seenLk.Unlock()

     

       168
       +
       

     

       169
       +
       	cs.seenHosts[host] = true

     

       170
        
       }

     

       171
        
       

     

       172
        
       func (cs *CensusService) Start(ctx context.Context, cctx *cli.Context) {

     
···

       174
        
       	cs.listHosts(ctx, cctx)

     

       175
        
       

     

       176
        
       	slog.Info("finished with initial refresh, starting ticker")

     

       177
       +
       	t := time.NewTicker(time.Minute * 5)

     

       178
        
       	defer t.Stop()

     

       179
        
       

     

       180
        
       	for {

+1 -6

cmd/monarch/cursors.go

···

       33
        
       		select {

     

       34
        
       		case <-ctx.Done():

     

       35
        
       			slog.Info("stopping cursor checkpointer", "err", ctx.Err())

     

       36
       -
       

     

       37
       -
       			slog.Info("persisting firehose cursor before exit", "seq", cs.firehoseSeq)

     

       38
       -
       			if err := cs.PersistFirehoseCursor(); err != nil {

     

       39
       -
       				slog.Error("error persisting firehose cursor", "err", err)

     

       40
       -
       			}

     

       41
       -
       

     

       42
        
       			return

     

       43
        
       		case <-t.C:

     

       44
        
       		}

···

       33
        
       		select {

     

       34
        
       		case <-ctx.Done():

     

       35
        
       			slog.Info("stopping cursor checkpointer", "err", ctx.Err())

     

       36
       +
       			cs.PersistFirehoseCursor() // one final save

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       37
        
       			return

     

       38
        
       		case <-t.C:

     

       39
        
       		}

+44 -114

cmd/monarch/handlers.go

···

       1
        
       package main

     

       2
        
       

     

       3
        
       import (

     

       0
        
       
     

       4
        
       	"context"

     

       0
        
       
     

       5
        
       	"fmt"

     

       6
        
       

     

       0
        
       
     

       7
        
       	"github.com/bluesky-social/indigo/atproto/syntax"

     

       8
        
       	"github.com/ipfs/go-cid"

     

       9
        
       	"gorm.io/gorm"

     

       10
       -
       	"gorm.io/gorm/clause"

     

       11
       -
       	"tangled.sh/edavis.dev/monarch/models"

     

       12
        
       )

     

       13
        
       

     

       14
        
       type Action int

     
···

       22
        
       	store *gorm.DB

     

       23
        
       }

     

       24
        
       

     

       25
       -
       func NewHandlerService(store *gorm.DB) *HandlerService {

     

       26
       -
       	migrations := []any{

     

       27
       -
       		// app.bsky.*

     

       28
       -
       		&models.ActorProfile{},

     

       29
       -
       		&models.ActorProfile_Label{},

     

       30
       -
       		&models.ActorStatus{},

     

       0
        
       
     

       31
        
       

     

       32
       -
       		&models.FeedGenerator{},

     

       33
       -
       		&models.FeedGenerator_Label{},

     

       34
       -
       		&models.FeedLike{},

     

       35
       -
       		&models.FeedPost{},

     

       36
       -
       		&models.FeedPost_Label{},

     

       37
       -
       		&models.FeedPost_Lang{},

     

       38
       -
       		&models.FeedPost_Tag{},

     

       39
       -
       		&models.FeedPostgate{},

     

       40
       -
       		&models.FeedPostgate_DetachedEmbeddingUri{},

     

       41
       -
       		&models.FeedPostgate_EmbeddingRule{},

     

       42
       -
       		&models.FeedRepost{},

     

       43
       -
       		&models.FeedThreadgate{},

     

       44
       -
       		&models.FeedThreadgate_AllowRule{},

     

       45
       -
       		&models.FeedThreadgate_HiddenReply{},

     

       46
       -
       

     

       47
       -
       		&models.GraphBlock{},

     

       48
       -
       		&models.GraphFollow{},

     

       49
       -
       		&models.GraphList{},

     

       50
       -
       		&models.GraphListblock{},

     

       51
       -
       		&models.GraphListitem{},

     

       52
       -
       		&models.GraphStarterpack{},

     

       53
       -
       		&models.GraphStarterpack_Feed{},

     

       54
       -
       		&models.GraphVerification{},

     

       55
       -
       

     

       56
       -
       		&models.LabelerService{},

     

       57
       -
       		&models.LabelerService_Label{},

     

       58
       -
       		&models.LabelerService_Policy{},

     

       59
       -
       		&models.LabelerService_ReasonType{},

     

       60
       -
       		&models.LabelerService_SubjectCollection{},

     

       61
       -
       		&models.LabelerService_SubjectType{},

     

       62
       -
       

     

       63
       -
       		// chat.bsky.*

     

       64
       -
       		&models.ActorDeclaration{},

     

       65
       -
       	}

     

       66
       -
       	for _, migration := range migrations {

     

       67
       -
       		store.AutoMigrate(migration)

     

       68
       -
       	}

     

       69
        
       

     

       70
        
       	return &HandlerService{

     

       71
        
       		store: store,

     

       72
        
       	}

     

       73
        
       }

     

       74
        
       

     

       75
       -
       type helper struct {

     

       76
       -
       	db     *gorm.DB

     

       77
       -
       	uri    syntax.ATURI

     

       78
       -
       	record []byte

     

       79
       -
       	action Action

     

       80
       -
       }

     

       81
       -
       

     

       82
       -
       func NewHelper(db *gorm.DB, uri syntax.ATURI, record []byte, action Action) *helper {

     

       83
       -
       	return &helper{db, uri, record, action}

     

       84
       -
       }

     

       85
       -
       

     

       86
       -
       func upsertRecord[T any](maker func(syntax.ATURI, []byte) T, opts *helper) error {

     

       87
       -
       	obj := maker(opts.uri, opts.record)

     

       88
       -
       	switch opts.action {

     

       89
       -
       	case ActionCreate:

     

       90
       -
       		if err := opts.db.Omit(clause.Associations).Create(obj).Error; err != nil {

     

       91
       -
       			return fmt.Errorf("error inserting %s record: %w", opts.uri.Collection(), err)

     

       92
       -
       		}

     

       93
       -
       	case ActionUpdate:

     

       94
       -
       		if err := opts.db.Save(obj).Error; err != nil {

     

       95
       -
       			return fmt.Errorf("error updating %s record: %w", opts.uri.Collection(), err)

     

       96
       -
       		}

     

       97
       -
       	}

     

       98
       -
       	return nil

     

       99
        
       }

     

       100
        
       

     

       101
        
       func (hs *HandlerService) HandleUpsert(ctx context.Context, repo string, rev string, path string, rec *[]byte, cid *cid.Cid, action Action) error {

     
···

       104
        
       		return fmt.Errorf("error parsing at-uri: %w", err)

     

       105
        
       	}

     

       106
        
       

     

       107
       -
       	opts := NewHelper(hs.store, uri, *rec, action)

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       108
        
       

     

       0
        
       
     

       109
        
       	switch uri.Collection() {

     

       110
        
       	case syntax.NSID("app.bsky.actor.profile"):

     

       111
       -
       		return upsertRecord(models.NewActorProfile, opts)

     

       112
       -
       

     

       113
       -
       	case syntax.NSID("app.bsky.actor.status"):

     

       114
       -
       		return upsertRecord(models.NewActorStatus, opts)

     

       115
        
       

     

       116
        
       	case syntax.NSID("app.bsky.feed.generator"):

     

       117
       -
       		return upsertRecord(models.NewFeedGenerator, opts)

     

       118
       -
       

     

       119
       -
       	case syntax.NSID("app.bsky.feed.like"):

     

       120
       -
       		return upsertRecord(models.NewFeedLike, opts)

     

       121
       -
       

     

       122
       -
       	case syntax.NSID("app.bsky.feed.post"):

     

       123
       -
       		return upsertRecord(models.NewFeedPost, opts)

     

       124
       -
       

     

       125
       -
       	case syntax.NSID("app.bsky.feed.postgate"):

     

       126
       -
       		return upsertRecord(models.NewFeedPostgate, opts)

     

       127
       -
       

     

       128
       -
       	case syntax.NSID("app.bsky.feed.repost"):

     

       129
       -
       		return upsertRecord(models.NewFeedRepost, opts)

     

       130
       -
       

     

       131
       -
       	case syntax.NSID("app.bsky.feed.threadgate"):

     

       132
       -
       		return upsertRecord(models.NewFeedThreadgate, opts)

     

       133
       -
       

     

       134
       -
       	case syntax.NSID("app.bsky.graph.block"):

     

       135
       -
       		return upsertRecord(models.NewGraphBlock, opts)

     

       136
       -
       

     

       137
       -
       	case syntax.NSID("app.bsky.graph.follow"):

     

       138
       -
       		return upsertRecord(models.NewGraphFollow, opts)

     

       139
       -
       

     

       140
       -
       	case syntax.NSID("app.bsky.graph.list"):

     

       141
       -
       		return upsertRecord(models.NewGraphList, opts)

     

       142
       -
       

     

       143
       -
       	case syntax.NSID("app.bsky.graph.listblock"):

     

       144
       -
       		return upsertRecord(models.NewGraphListblock, opts)

     

       145
       -
       

     

       146
       -
       	case syntax.NSID("app.bsky.graph.listitem"):

     

       147
       -
       		return upsertRecord(models.NewGraphListitem, opts)

     

       148
       -
       

     

       149
       -
       	case syntax.NSID("app.bsky.graph.starterpack"):

     

       150
       -
       		return upsertRecord(models.NewGraphStarterpack, opts)

     

       151
       -
       

     

       152
       -
       	case syntax.NSID("app.bsky.graph.verification"):

     

       153
       -
       		return upsertRecord(models.NewGraphVerification, opts)

     

       154
        
       

     

       155
        
       	case syntax.NSID("app.bsky.labeler.service"):

     

       156
       -
       		return upsertRecord(models.NewLabelerService, opts)

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       157
        
       

     

       158
       -
       	case syntax.NSID("chat.bsky.actor.declaration"):

     

       159
       -
       		return upsertRecord(models.NewActorDeclaration, opts)

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       160
        
       	}

     

       161
        
       

     

       162
        
       	return nil

···

       1
        
       package main

     

       2
        
       

     

       3
        
       import (

     

       4
       +
       	"bytes"

     

       5
        
       	"context"

     

       6
       +
       	"encoding/json"

     

       7
        
       	"fmt"

     

       8
        
       

     

       9
       +
       	appbsky "github.com/bluesky-social/indigo/api/bsky"

     

       10
        
       	"github.com/bluesky-social/indigo/atproto/syntax"

     

       11
        
       	"github.com/ipfs/go-cid"

     

       12
        
       	"gorm.io/gorm"

     

       0
        
       
     

       0
        
       
     

       13
        
       )

     

       14
        
       

     

       15
        
       type Action int

     
···

       23
        
       	store *gorm.DB

     

       24
        
       }

     

       25
        
       

     

       26
       +
       type Record struct {

     

       27
       +
       	ID         int `gorm:"primaryKey"`

     

       28
       +
       	Repo       string

     

       29
       +
       	Collection string

     

       30
       +
       	Key        string

     

       31
       +
       	Data       json.RawMessage `gorm:"type:jsonb"`

     

       32
       +
       }

     

       33
        
       

     

       34
       +
       func NewHandlerService(store *gorm.DB) *HandlerService {

     

       35
       +
       	store.AutoMigrate(&Record{})

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       36
        
       

     

       37
        
       	return &HandlerService{

     

       38
        
       		store: store,

     

       39
        
       	}

     

       40
        
       }

     

       41
        
       

     

       42
       +
       var trackedCollections = map[syntax.NSID]bool{

     

       43
       +
       	syntax.NSID("app.bsky.actor.profile"):      false,

     

       44
       +
       	syntax.NSID("app.bsky.feed.generator"):     true,

     

       45
       +
       	syntax.NSID("app.bsky.labeler.service"):    true,

     

       46
       +
       	syntax.NSID("app.bsky.graph.list"):         true,

     

       47
       +
       	syntax.NSID("app.bsky.graph.verification"): true,

     

       48
       +
       	syntax.NSID("app.bsky.graph.starterpack"):  true,

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       49
        
       }

     

       50
        
       

     

       51
        
       func (hs *HandlerService) HandleUpsert(ctx context.Context, repo string, rev string, path string, rec *[]byte, cid *cid.Cid, action Action) error {

     
···

       54
        
       		return fmt.Errorf("error parsing at-uri: %w", err)

     

       55
        
       	}

     

       56
        
       

     

       57
       +
       	tracked := trackedCollections[uri.Collection()]

     

       58
       +
       	if !tracked {

     

       59
       +
       		return nil

     

       60
       +
       	}

     

       61
        
       

     

       62
       +
       	var body []byte

     

       63
        
       	switch uri.Collection() {

     

       64
        
       	case syntax.NSID("app.bsky.actor.profile"):

     

       65
       +
       		var out appbsky.ActorProfile

     

       66
       +
       		out.UnmarshalCBOR(bytes.NewReader(*rec))

     

       67
       +
       		body, err = json.Marshal(out)

     

       0
        
       
     

       68
        
       

     

       69
        
       	case syntax.NSID("app.bsky.feed.generator"):

     

       70
       +
       		var out appbsky.FeedGenerator

     

       71
       +
       		out.UnmarshalCBOR(bytes.NewReader(*rec))

     

       72
       +
       		body, err = json.Marshal(out)

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       73
        
       

     

       74
        
       	case syntax.NSID("app.bsky.labeler.service"):

     

       75
       +
       		var out appbsky.LabelerService

     

       76
       +
       		out.UnmarshalCBOR(bytes.NewReader(*rec))

     

       77
       +
       		body, err = json.Marshal(out)

     

       78
       +
       	}

     

       79
        
       

     

       80
       +
       	switch action {

     

       81
       +
       	case ActionCreate:

     

       82
       +
       		if err := hs.store.Create(&Record{

     

       83
       +
       			Repo:       repo,

     

       84
       +
       			Collection: string(uri.Collection()),

     

       85
       +
       			Key:        string(uri.RecordKey()),

     

       86
       +
       			Data:       json.RawMessage(body),

     

       87
       +
       		}).Error; err != nil {

     

       88
       +
       			return err

     

       89
       +
       		}

     

       90
        
       	}

     

       91
        
       

     

       92
        
       	return nil