comparing b14f9c274ca92b92cd89bd475bb4ea07fde91205 and hailey/nervana-topics on hailey.at/photocopy

+47 -10

backfiller.go

···

       7
        
       	"fmt"

     

       8
        
       	"io"

     

       9
        
       	"net/http"

     

       0
        
       
     

       10
        
       	"strings"

     

       11
        
       	"sync"

     

       12
        
       	"time"

     
···

       20
        
       	"go.uber.org/ratelimit"

     

       21
        
       )

     

       22
        
       

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       23
        
       type RepoDownloader struct {

     

       24
       -
       	clients    map[string]*http.Client

     

       25
       -
       	rateLimits map[string]ratelimit.Limiter

     

       26
       -
       	mu         sync.RWMutex

     

       27
       -
       	p          *Photocopy

     

       0
        
       
     

       28
        
       }

     

       29
        
       

     

       30
        
       func NewRepoDownloader(p *Photocopy) *RepoDownloader {

     

       31
        
       	return &RepoDownloader{

     

       32
       -
       		clients:    make(map[string]*http.Client),

     

       33
       -
       		rateLimits: make(map[string]ratelimit.Limiter),

     

       34
       -
       		p:          p,

     

       0
        
       
     

       35
        
       	}

     

       36
        
       }

     

       37
        
       

     
···

       122
        
       	bs := atproto_repo.NewTinyBlockstore()

     

       123
        
       	cs, err := car.NewCarReader(bytes.NewReader(b))

     

       124
        
       	if err != nil {

     

       0
        
       
     

       125
        
       		return fmt.Errorf("error opening car: %v\n", err)

     

       126
        
       	}

     

       127
        
       

     
···

       134
        
       

     

       135
        
       	r, err := repo.OpenRepo(context.TODO(), bs, cs.Header.Roots[0])

     

       136
        
       	if err != nil || r == nil {

     

       0
        
       
     

       137
        
       		fmt.Printf("could not open repo: %v", err)

     

       138
        
       		return nil

     

       139
        
       	}

     
···

       238
        
       	Status   string `json:"status"`

     

       239
        
       }

     

       240
        
       

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       241
        
       func (p *Photocopy) runBackfiller(ctx context.Context) error {

     

       242
        
       	startTime := time.Now()

     

       243
        
       

     

       244
        
       	fmt.Println("querying clickhouse for dids and services...")

     

       245
        
       

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       246
        
       	var hostsCursor string

     

       247
        
       	var sevs []ListServicesResponseItem

     

       248
        
       	for {

     
···

       299
        
       	downloader := NewRepoDownloader(p)

     

       300
        
       	serviceDids := map[string][]string{}

     

       301
        
       

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       302
        
       	wg := sync.WaitGroup{}

     

       303
        
       	mplk := sync.Mutex{}

     

       304
        
       	for s := range servicesDids {

     
···

       312
        
       			}

     

       313
        
       			dids := []string{}

     

       314
        
       			for _, r := range repos {

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       315
        
       				dids = append(dids, r.Did)

     

       316
        
       			}

     

       317
        
       			mplk.Lock()

     
···

       365
        
       				}

     

       366
        
       

     

       367
        
       				go func(b []byte, did string) {

     

       368
       -
       					if err := p.processRepo(ctx, b, did); err != nil {

     

       369
       -
       						fmt.Printf("error processing backfill record: %v\n", err)

     

       370
       -
       					}

     

       371
        
       				}(b, did)

     

       372
        
       

     

       373
        
       				processed++

···

       7
        
       	"fmt"

     

       8
        
       	"io"

     

       9
        
       	"net/http"

     

       10
       +
       	"runtime"

     

       11
        
       	"strings"

     

       12
        
       	"sync"

     

       13
        
       	"time"

     
···

       21
        
       	"go.uber.org/ratelimit"

     

       22
        
       )

     

       23
        
       

     

       24
       +
       type ProcessJob struct {

     

       25
       +
       	did       string

     

       26
       +
       	repoBytes []byte

     

       27
       +
       }

     

       28
       +
       

     

       29
        
       type RepoDownloader struct {

     

       30
       +
       	clients     map[string]*http.Client

     

       31
       +
       	rateLimits  map[string]ratelimit.Limiter

     

       32
       +
       	processChan chan ProcessJob

     

       33
       +
       	mu          sync.RWMutex

     

       34
       +
       	p           *Photocopy

     

       35
        
       }

     

       36
        
       

     

       37
        
       func NewRepoDownloader(p *Photocopy) *RepoDownloader {

     

       38
        
       	return &RepoDownloader{

     

       39
       +
       		clients:     make(map[string]*http.Client),

     

       40
       +
       		rateLimits:  make(map[string]ratelimit.Limiter),

     

       41
       +
       		p:           p,

     

       42
       +
       		processChan: make(chan ProcessJob, 1000),

     

       43
        
       	}

     

       44
        
       }

     

       45
        
       

     
···

       130
        
       	bs := atproto_repo.NewTinyBlockstore()

     

       131
        
       	cs, err := car.NewCarReader(bytes.NewReader(b))

     

       132
        
       	if err != nil {

     

       133
       +
       		fmt.Println("error opening car", err)

     

       134
        
       		return fmt.Errorf("error opening car: %v\n", err)

     

       135
        
       	}

     

       136
        
       

     
···

       143
        
       

     

       144
        
       	r, err := repo.OpenRepo(context.TODO(), bs, cs.Header.Roots[0])

     

       145
        
       	if err != nil || r == nil {

     

       146
       +
       		fmt.Println("error opening repo", err)

     

       147
        
       		fmt.Printf("could not open repo: %v", err)

     

       148
        
       		return nil

     

       149
        
       	}

     
···

       248
        
       	Status   string `json:"status"`

     

       249
        
       }

     

       250
        
       

     

       251
       +
       func (p *Photocopy) runProcessRepoWorker(ctx context.Context, jobs <-chan ProcessJob) {

     

       252
       +
       	for j := range jobs {

     

       253
       +
       		p.processRepo(ctx, j.repoBytes, j.did)

     

       254
       +
       	}

     

       255
       +
       }

     

       256
       +
       

     

       257
        
       func (p *Photocopy) runBackfiller(ctx context.Context) error {

     

       258
        
       	startTime := time.Now()

     

       259
        
       

     

       260
        
       	fmt.Println("querying clickhouse for dids and services...")

     

       261
        
       

     

       262
       +
       	type alreadyFetchedItem struct {

     

       263
       +
       		Did string `ch:"did"`

     

       264
       +
       	}

     

       265
       +
       	var alreadyFetched []alreadyFetchedItem

     

       266
       +
       	if err := p.conn.Select(ctx, &alreadyFetched, "SELECT DISTINCT(did) FROM default.record WHERE created_at < '2025-07-01'"); err != nil {

     

       267
       +
       		return err

     

       268
       +
       	}

     

       269
       +
       

     

       270
       +
       	alreadyFetchedMap := map[string]bool{}

     

       271
       +
       	for _, d := range alreadyFetched {

     

       272
       +
       		alreadyFetchedMap[d.Did] = true

     

       273
       +
       	}

     

       274
       +
       

     

       275
       +
       	fmt.Println("getting dids")

     

       276
       +
       

     

       277
        
       	var hostsCursor string

     

       278
        
       	var sevs []ListServicesResponseItem

     

       279
        
       	for {

     
···

       330
        
       	downloader := NewRepoDownloader(p)

     

       331
        
       	serviceDids := map[string][]string{}

     

       332
        
       

     

       333
       +
       	for range runtime.NumCPU() / 2 {

     

       334
       +
       		go p.runProcessRepoWorker(ctx, downloader.processChan)

     

       335
       +
       	}

     

       336
       +
       

     

       337
        
       	wg := sync.WaitGroup{}

     

       338
        
       	mplk := sync.Mutex{}

     

       339
        
       	for s := range servicesDids {

     
···

       347
        
       			}

     

       348
        
       			dids := []string{}

     

       349
        
       			for _, r := range repos {

     

       350
       +
       				if alreadyFetchedMap[r.Did] {

     

       351
       +
       					skipped++

     

       352
       +
       					continue

     

       353
       +
       				}

     

       354
        
       				dids = append(dids, r.Did)

     

       355
        
       			}

     

       356
        
       			mplk.Lock()

     
···

       404
        
       				}

     

       405
        
       

     

       406
        
       				go func(b []byte, did string) {

     

       407
       +
       					downloader.processChan <- ProcessJob{repoBytes: b, did: did}

     

       0
        
       
     

       0
        
       
     

       408
        
       				}(b, did)

     

       409
        
       

     

       410
        
       				processed++

-420

cmd/bodega/main.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       import (

     

       4
       -
       	"bytes"

     

       5
       -
       	"context"

     

       6
       -
       	"encoding/json"

     

       7
       -
       	"fmt"

     

       8
       -
       	"io"

     

       9
       -
       	"log/slog"

     

       10
       -
       	"net/http"

     

       11
       -
       	"os"

     

       12
       -
       	"strings"

     

       13
       -
       	"sync"

     

       14
       -
       	"time"

     

       15
       -
       

     

       16
       -
       	"github.com/ClickHouse/clickhouse-go/v2"

     

       17
       -
       	atproto_repo "github.com/bluesky-social/indigo/atproto/repo"

     

       18
       -
       	"github.com/bluesky-social/indigo/atproto/syntax"

     

       19
       -
       	"github.com/bluesky-social/indigo/repo"

     

       20
       -
       	"github.com/bluesky-social/indigo/util"

     

       21
       -
       	"github.com/haileyok/photocopy/clickhouse_inserter"

     

       22
       -
       	"github.com/haileyok/photocopy/models"

     

       23
       -
       	"github.com/ipfs/go-cid"

     

       24
       -
       	"github.com/ipld/go-car"

     

       25
       -
       	_ "github.com/joho/godotenv/autoload"

     

       26
       -
       	"github.com/urfave/cli/v2"

     

       27
       -
       	"go.uber.org/ratelimit"

     

       28
       -
       )

     

       29
       -
       

     

       30
       -
       func main() {

     

       31
       -
       	app := cli.App{

     

       32
       -
       		Name:   "bodega",

     

       33
       -
       		Action: run,

     

       34
       -
       		Flags: []cli.Flag{

     

       35
       -
       			&cli.StringFlag{

     

       36
       -
       				Name:     "clickhouse-addr",

     

       37
       -
       				EnvVars:  []string{"PHOTOCOPY_CLICKHOUSE_ADDR"},

     

       38
       -
       				Required: true,

     

       39
       -
       			},

     

       40
       -
       			&cli.StringFlag{

     

       41
       -
       				Name:     "clickhouse-database",

     

       42
       -
       				EnvVars:  []string{"PHOTOCOPY_CLICKHOUSE_DATABASE"},

     

       43
       -
       				Required: true,

     

       44
       -
       			},

     

       45
       -
       			&cli.StringFlag{

     

       46
       -
       				Name:    "clickhouse-user",

     

       47
       -
       				EnvVars: []string{"PHOTOCOPY_CLICKHOUSE_USER"},

     

       48
       -
       				Value:   "default",

     

       49
       -
       			},

     

       50
       -
       			&cli.StringFlag{

     

       51
       -
       				Name:     "clickhouse-pass",

     

       52
       -
       				EnvVars:  []string{"PHOTOCOPY_CLICKHOUSE_PASS"},

     

       53
       -
       				Required: true,

     

       54
       -
       			},

     

       55
       -
       			&cli.BoolFlag{

     

       56
       -
       				Name:  "debug",

     

       57
       -
       				Value: false,

     

       58
       -
       			},

     

       59
       -
       		},

     

       60
       -
       	}

     

       61
       -
       

     

       62
       -
       	app.Run(os.Args)

     

       63
       -
       }

     

       64
       -
       

     

       65
       -
       type RepoDownloader struct {

     

       66
       -
       	clients    map[string]*http.Client

     

       67
       -
       	rateLimits map[string]ratelimit.Limiter

     

       68
       -
       	mu         sync.RWMutex

     

       69
       -
       }

     

       70
       -
       

     

       71
       -
       func NewRepoDownloader() *RepoDownloader {

     

       72
       -
       	return &RepoDownloader{

     

       73
       -
       		clients:    make(map[string]*http.Client),

     

       74
       -
       		rateLimits: make(map[string]ratelimit.Limiter),

     

       75
       -
       	}

     

       76
       -
       }

     

       77
       -
       

     

       78
       -
       func (rd *RepoDownloader) getClient(service string) *http.Client {

     

       79
       -
       	rd.mu.RLock()

     

       80
       -
       	client, exists := rd.clients[service]

     

       81
       -
       	rd.mu.RUnlock()

     

       82
       -
       

     

       83
       -
       	if exists {

     

       84
       -
       		return client

     

       85
       -
       	}

     

       86
       -
       

     

       87
       -
       	rd.mu.Lock()

     

       88
       -
       	defer rd.mu.Unlock()

     

       89
       -
       

     

       90
       -
       	if client, exists := rd.clients[service]; exists {

     

       91
       -
       		return client

     

       92
       -
       	}

     

       93
       -
       

     

       94
       -
       	client = util.RobustHTTPClient()

     

       95
       -
       	client.Timeout = 30 * time.Minute

     

       96
       -
       	rd.clients[service] = client

     

       97
       -
       	return client

     

       98
       -
       }

     

       99
       -
       

     

       100
       -
       func (rd *RepoDownloader) getRateLimiter(service string) ratelimit.Limiter {

     

       101
       -
       	rd.mu.RLock()

     

       102
       -
       	limiter, exists := rd.rateLimits[service]

     

       103
       -
       	rd.mu.RUnlock()

     

       104
       -
       

     

       105
       -
       	if exists {

     

       106
       -
       		return limiter

     

       107
       -
       	}

     

       108
       -
       

     

       109
       -
       	rd.mu.Lock()

     

       110
       -
       	defer rd.mu.Unlock()

     

       111
       -
       

     

       112
       -
       	if limiter, exists := rd.rateLimits[service]; exists {

     

       113
       -
       		return limiter

     

       114
       -
       	}

     

       115
       -
       

     

       116
       -
       	// 3000 per five minutes

     

       117
       -
       	limiter = ratelimit.New(10)

     

       118
       -
       	rd.rateLimits[service] = limiter

     

       119
       -
       	return limiter

     

       120
       -
       }

     

       121
       -
       

     

       122
       -
       func (rd *RepoDownloader) downloadRepo(service, did string) ([]byte, error) {

     

       123
       -
       	dlurl := fmt.Sprintf("%s/xrpc/com.atproto.sync.getRepo?did=%s", service, did)

     

       124
       -
       

     

       125
       -
       	req, err := http.NewRequestWithContext(context.TODO(), "GET", dlurl, nil)

     

       126
       -
       	if err != nil {

     

       127
       -
       		return nil, fmt.Errorf("failed to create request: %w", err)

     

       128
       -
       	}

     

       129
       -
       

     

       130
       -
       	client := rd.getClient(service)

     

       131
       -
       

     

       132
       -
       	resp, err := client.Do(req)

     

       133
       -
       	if err != nil {

     

       134
       -
       		return nil, fmt.Errorf("failed to download repo: %w", err)

     

       135
       -
       	}

     

       136
       -
       	defer resp.Body.Close()

     

       137
       -
       

     

       138
       -
       	if resp.StatusCode != http.StatusOK {

     

       139
       -
       		return nil, fmt.Errorf("unexpected status code: %d", resp.StatusCode)

     

       140
       -
       	}

     

       141
       -
       

     

       142
       -
       	b, err := io.ReadAll(resp.Body)

     

       143
       -
       	if err != nil {

     

       144
       -
       		return nil, fmt.Errorf("could not read bytes from response: %w", err)

     

       145
       -
       	}

     

       146
       -
       

     

       147
       -
       	return b, nil

     

       148
       -
       }

     

       149
       -
       

     

       150
       -
       func processRepo(b []byte, did string, inserter *clickhouse_inserter.Inserter) error {

     

       151
       -
       	bs := atproto_repo.NewTinyBlockstore()

     

       152
       -
       	cs, err := car.NewCarReader(bytes.NewReader(b))

     

       153
       -
       	if err != nil {

     

       154
       -
       		return fmt.Errorf("error opening car: %v\n", err)

     

       155
       -
       	}

     

       156
       -
       

     

       157
       -
       	currBlock, _ := cs.Next()

     

       158
       -
       	for currBlock != nil {

     

       159
       -
       		bs.Put(context.TODO(), currBlock)

     

       160
       -
       		next, _ := cs.Next()

     

       161
       -
       		currBlock = next

     

       162
       -
       	}

     

       163
       -
       

     

       164
       -
       	r, err := repo.OpenRepo(context.TODO(), bs, cs.Header.Roots[0])

     

       165
       -
       	if err != nil || r == nil {

     

       166
       -
       		fmt.Printf("could not open repo: %v", err)

     

       167
       -
       		return nil

     

       168
       -
       	}

     

       169
       -
       

     

       170
       -
       	if err := r.ForEach(context.TODO(), "", func(key string, cid cid.Cid) error {

     

       171
       -
       		pts := strings.Split(key, "/")

     

       172
       -
       		nsid := pts[0]

     

       173
       -
       		rkey := pts[1]

     

       174
       -
       		cidStr := cid.String()

     

       175
       -
       		b, err := bs.Get(context.TODO(), cid)

     

       176
       -
       		if err != nil {

     

       177
       -
       			return nil

     

       178
       -
       		}

     

       179
       -
       

     

       180
       -
       		var cat time.Time

     

       181
       -
       		tid, err := syntax.ParseTID(rkey)

     

       182
       -
       		if err != nil {

     

       183
       -
       			cat = time.Now()

     

       184
       -
       		} else {

     

       185
       -
       			cat = tid.Time()

     

       186
       -
       		}

     

       187
       -
       

     

       188
       -
       		rec := models.Record{

     

       189
       -
       			Did:        did,

     

       190
       -
       			Rkey:       rkey,

     

       191
       -
       			Collection: nsid,

     

       192
       -
       			Cid:        cidStr,

     

       193
       -
       			Seq:        "",

     

       194
       -
       			Raw:        string(b.RawData()),

     

       195
       -
       			CreatedAt:  cat,

     

       196
       -
       		}

     

       197
       -
       

     

       198
       -
       		inserter.Insert(context.TODO(), rec)

     

       199
       -
       

     

       200
       -
       		return nil

     

       201
       -
       	}); err != nil {

     

       202
       -
       		return fmt.Errorf("erorr traversing records: %v", err)

     

       203
       -
       	}

     

       204
       -
       

     

       205
       -
       	return nil

     

       206
       -
       }

     

       207
       -
       

     

       208
       -
       type ListReposResponse struct {

     

       209
       -
       	Cursor string          `json:"cursor"`

     

       210
       -
       	Repos  []ListReposRepo `json:"repos"`

     

       211
       -
       }

     

       212
       -
       

     

       213
       -
       type ListReposRepo struct {

     

       214
       -
       	Did    string  `json:"did"`

     

       215
       -
       	Head   string  `json:"head"`

     

       216
       -
       	Rev    string  `json:"rev"`

     

       217
       -
       	Active bool    `json:"active"`

     

       218
       -
       	Status *string `json:"status,omitempty"`

     

       219
       -
       }

     

       220
       -
       

     

       221
       -
       func (rd *RepoDownloader) getDidsFromService(ctx context.Context, service string) ([]ListReposRepo, error) {

     

       222
       -
       	var cursor string

     

       223
       -
       	var repos []ListReposRepo

     

       224
       -
       	for {

     

       225
       -
       		req, err := http.NewRequestWithContext(ctx, "GET", fmt.Sprintf("%s/xrpc/com.atproto.sync.listRepos?limit=1000&cursor=%s", service, cursor), nil)

     

       226
       -
       		if err != nil {

     

       227
       -
       			return nil, err

     

       228
       -
       		}

     

       229
       -
       

     

       230
       -
       		rl := rd.getRateLimiter(service)

     

       231
       -
       		rl.Take()

     

       232
       -
       

     

       233
       -
       		cli := rd.getClient(service)

     

       234
       -
       		resp, err := cli.Do(req)

     

       235
       -
       		if err != nil {

     

       236
       -
       			return nil, err

     

       237
       -
       		}

     

       238
       -
       		defer resp.Body.Close()

     

       239
       -
       

     

       240
       -
       		if resp.StatusCode != http.StatusOK {

     

       241
       -
       			return nil, fmt.Errorf("received non-200 response code: %d", resp.StatusCode)

     

       242
       -
       		}

     

       243
       -
       

     

       244
       -
       		var reposResp ListReposResponse

     

       245
       -
       		if err := json.NewDecoder(resp.Body).Decode(&reposResp); err != nil {

     

       246
       -
       			return nil, fmt.Errorf("error decoding repos response: %w", err)

     

       247
       -
       		}

     

       248
       -
       

     

       249
       -
       		repos = append(repos, reposResp.Repos...)

     

       250
       -
       

     

       251
       -
       		if len(reposResp.Repos) != 1000 {

     

       252
       -
       			break

     

       253
       -
       		}

     

       254
       -
       	}

     

       255
       -
       

     

       256
       -
       	return repos, nil

     

       257
       -
       }

     

       258
       -
       

     

       259
       -
       var run = func(cmd *cli.Context) error {

     

       260
       -
       	startTime := time.Now()

     

       261
       -
       

     

       262
       -
       	conn, err := clickhouse.Open(&clickhouse.Options{

     

       263
       -
       		Addr: []string{cmd.String("clickhouse-addr")},

     

       264
       -
       		Auth: clickhouse.Auth{

     

       265
       -
       			Database: cmd.String("clickhouse-database"),

     

       266
       -
       			Username: cmd.String("clickhouse-user"),

     

       267
       -
       			Password: cmd.String("clickhouse-pass"),

     

       268
       -
       		},

     

       269
       -
       	})

     

       270
       -
       	if err != nil {

     

       271
       -
       		return err

     

       272
       -
       	}

     

       273
       -
       	defer conn.Close()

     

       274
       -
       

     

       275
       -
       	fmt.Println("querying clickhouse for dids and services...")

     

       276
       -
       

     

       277
       -
       	type servicesQueryRow struct {

     

       278
       -
       		PlcOpServices []string `ch:"plc_op_services"`

     

       279
       -
       	}

     

       280
       -
       	var servicesQueryRows []servicesQueryRow

     

       281
       -
       	if err := conn.Select(cmd.Context, &servicesQueryRows, `

     

       282
       -
       		SELECT DISTINCT(plc_op_services) FROM default.plc WHERE arrayExists(x -> x LIKE '%.bsky.network', plc_op_services)

     

       283
       -
       		`); err != nil {

     

       284
       -
       		return err

     

       285
       -
       	}

     

       286
       -
       

     

       287
       -
       	servicesDids := map[string][]string{}

     

       288
       -
       	for _, svcs := range servicesQueryRows {

     

       289
       -
       		for _, s := range svcs.PlcOpServices {

     

       290
       -
       			servicesDids[s] = []string{}

     

       291
       -
       		}

     

       292
       -
       	}

     

       293
       -
       

     

       294
       -
       	fmt.Printf("found %d services\n", len(servicesDids))

     

       295
       -
       

     

       296
       -
       	fmt.Printf("getting most recent record for each did...")

     

       297
       -
       	var records []models.Record

     

       298
       -
       	if err := conn.Select(cmd.Context, &records, `

     

       299
       -
       SELECT did, created_at

     

       300
       -
       FROM default.record

     

       301
       -
       QUALIFY row_number() OVER (PARTITION BY did ORDER BY created_at ASC) = 1

     

       302
       -
       		`); err != nil {

     

       303
       -
       		return err

     

       304
       -
       	}

     

       305
       -
       

     

       306
       -
       	fmt.Printf("collecting dids...\n")

     

       307
       -
       

     

       308
       -
       	didCreatedAt := map[string]time.Time{}

     

       309
       -
       	for _, r := range records {

     

       310
       -
       		didCreatedAt[r.Did] = r.CreatedAt

     

       311
       -
       	}

     

       312
       -
       

     

       313
       -
       	inserter, err := clickhouse_inserter.New(context.TODO(), &clickhouse_inserter.Args{

     

       314
       -
       		BatchSize: 100000,

     

       315
       -
       		Logger:    slog.Default(),

     

       316
       -
       		Conn:      conn,

     

       317
       -
       		Query:     "INSERT INTO record (did, rkey, collection, cid, seq, raw, created_at)",

     

       318
       -
       		RateLimit: 2, // two inserts per second in the event of massive repos

     

       319
       -
       	})

     

       320
       -
       	if err != nil {

     

       321
       -
       		return err

     

       322
       -
       	}

     

       323
       -
       

     

       324
       -
       	fmt.Printf("building download buckets...")

     

       325
       -
       

     

       326
       -
       	skipped := 0

     

       327
       -
       	total := 0

     

       328
       -
       	needOlderThan, _ := time.Parse(time.DateTime, "2025-06-28 04:18:22")

     

       329
       -
       	downloader := NewRepoDownloader()

     

       330
       -
       	serviceDids := map[string][]string{}

     

       331
       -
       

     

       332
       -
       	wg := sync.WaitGroup{}

     

       333
       -
       	for s := range servicesDids {

     

       334
       -
       		wg.Add(1)

     

       335
       -
       		go func() {

     

       336
       -
       			defer wg.Done()

     

       337
       -
       			repos, err := downloader.getDidsFromService(context.TODO(), s)

     

       338
       -
       			if err != nil {

     

       339
       -
       				fmt.Printf("error getting dids for services %s: %v", s, err)

     

       340
       -
       				return

     

       341
       -
       			}

     

       342
       -
       			dids := []string{}

     

       343
       -
       			for _, r := range repos {

     

       344
       -
       				lastRecord, exists := didCreatedAt[r.Did]

     

       345
       -
       				if exists && lastRecord.Before(needOlderThan) {

     

       346
       -
       					skipped++

     

       347
       -
       					continue

     

       348
       -
       				}

     

       349
       -
       

     

       350
       -
       				dids = append(dids, r.Did)

     

       351
       -
       			}

     

       352
       -
       			serviceDids[s] = dids

     

       353
       -
       		}()

     

       354
       -
       	}

     

       355
       -
       

     

       356
       -
       	fmt.Println("getting all the repos...")

     

       357
       -
       	wg.Wait()

     

       358
       -
       

     

       359
       -
       	fmt.Printf("Total jobs: %d across %d services \n", total, len(serviceDids))

     

       360
       -
       	fmt.Printf("was able to skip %d repos\n", skipped)

     

       361
       -
       

     

       362
       -
       	for service, dids := range serviceDids {

     

       363
       -
       		if len(dids) < 100 {

     

       364
       -
       			continue

     

       365
       -
       		}

     

       366
       -
       		fmt.Printf("%s: %d jobs\n", service, len(dids))

     

       367
       -
       	}

     

       368
       -
       

     

       369
       -
       	processed := 0

     

       370
       -
       	errored := 0

     

       371
       -
       

     

       372
       -
       	for service, dids := range serviceDids {

     

       373
       -
       		go func() {

     

       374
       -
       			for _, did := range dids {

     

       375
       -
       				ratelimiter := downloader.getRateLimiter(service)

     

       376
       -
       				ratelimiter.Take()

     

       377
       -
       

     

       378
       -
       				b, err := downloader.downloadRepo(service, did)

     

       379
       -
       				if err != nil {

     

       380
       -
       					errored++

     

       381
       -
       					processed++

     

       382
       -
       					continue

     

       383
       -
       				}

     

       384
       -
       

     

       385
       -
       				go func(b []byte, did string, inserter *clickhouse_inserter.Inserter) {

     

       386
       -
       					processRepo(b, did, inserter)

     

       387
       -
       				}(b, did, inserter)

     

       388
       -
       

     

       389
       -
       				processed++

     

       390
       -
       			}

     

       391
       -
       		}()

     

       392
       -
       	}

     

       393
       -
       

     

       394
       -
       	ticker := time.NewTicker(1 * time.Second)

     

       395
       -
       	defer ticker.Stop()

     

       396
       -
       

     

       397
       -
       	for range ticker.C {

     

       398
       -
       		elapsed := time.Since(startTime)

     

       399
       -
       		rate := float64(processed) / elapsed.Seconds()

     

       400
       -
       		remaining := total - processed

     

       401
       -
       

     

       402
       -
       		var eta string

     

       403
       -
       		if rate > 0 {

     

       404
       -
       			etaSeconds := float64(remaining) / rate

     

       405
       -
       			etaDuration := time.Duration(etaSeconds * float64(time.Second))

     

       406
       -
       			eta = fmt.Sprintf(", ETA: %v", etaDuration.Round(time.Second))

     

       407
       -
       		} else {

     

       408
       -
       			eta = ", ETA: calculating..."

     

       409
       -
       		}

     

       410
       -
       

     

       411
       -
       		fmt.Printf("\rProgress: %d/%d processed (%.1f%%), %d skipped, %d errors, %.1f jobs/sec%s",

     

       412
       -
       			processed, total, float64(processed)/float64(total)*100, skipped, errored, rate, eta)

     

       413
       -
       	}

     

       414
       -
       

     

       415
       -
       	fmt.Printf("\nCompleted: %d processed, %d errors\n", processed, errored)

     

       416
       -
       

     

       417
       -
       	inserter.Close(context.TODO())

     

       418
       -
       

     

       419
       -
       	return nil

     

       420
       -
       }

-28

cmd/bodega/plc_models.go

···

       1
       -
       package main

     

       2
       -
       

     

       3
       -
       import (

     

       4
       -
       	"time"

     

       5
       -
       )

     

       6
       -
       

     

       7
       -
       type ClickhousePLCEntry struct {

     

       8
       -
       	Did                 string    `ch:"did"`

     

       9
       -
       	Cid                 string    `ch:"cid"`

     

       10
       -
       	Nullified           bool      `ch:"nullified"`

     

       11
       -
       	CreatedAt           time.Time `ch:"created_at"`

     

       12
       -
       	PlcOpSig            string    `ch:"plc_op_sig"`

     

       13
       -
       	PlcOpPrev           string    `ch:"plc_op_prev"`

     

       14
       -
       	PlcOpType           string    `ch:"plc_op_type"`

     

       15
       -
       	PlcOpServices       []string  `ch:"plc_op_services"`

     

       16
       -
       	PlcOpAlsoKnownAs    []string  `ch:"plc_op_also_known_as"`

     

       17
       -
       	PlcOpRotationKeys   []string  `ch:"plc_op_rotation_keys"`

     

       18
       -
       	PlcTombSig          string    `ch:"plc_tomb_sig"`

     

       19
       -
       	PlcTombPrev         string    `ch:"plc_tomb_prev"`

     

       20
       -
       	PlcTombType         string    `ch:"plc_tomb_type"`

     

       21
       -
       	LegacyOpSig         string    `ch:"legacy_op_sig"`

     

       22
       -
       	LegacyOpPrev        string    `ch:"legacy_op_prev"`

     

       23
       -
       	LegacyOpType        string    `ch:"legacy_op_type"`

     

       24
       -
       	LegacyOpHandle      string    `ch:"legacy_op_handle"`

     

       25
       -
       	LegacyOpService     string    `ch:"legacy_op_service"`

     

       26
       -
       	LegacyOpSigningKey  string    `ch:"legacy_op_signing_key"`

     

       27
       -
       	LegacyOpRecoveryKey string    `ch:"legacy_op_recovery_key"`

     

       28
       -
       }

+10

cmd/photocopy/main.go

···

       70
        
       			&cli.BoolFlag{

     

       71
        
       				Name: "with-backfill",

     

       72
        
       			},

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       73
        
       		},

     

       74
        
       		Commands: cli.Commands{

     

       75
        
       			&cli.Command{

     
···

       121
        
       		ClickhouseUser:       cmd.String("clickhouse-user"),

     

       122
        
       		ClickhousePass:       cmd.String("clickhouse-pass"),

     

       123
        
       		RatelimitBypassKey:   cmd.String("ratelimit-bypass-key"),

     

       0
        
       
     

       0
        
       
     

       124
        
       	})

     

       125
        
       	if err != nil {

     

       126
        
       		panic(err)

···

       70
        
       			&cli.BoolFlag{

     

       71
        
       				Name: "with-backfill",

     

       72
        
       			},

     

       73
       +
       			&cli.StringFlag{

     

       74
       +
       				Name:    "nervana-endpoint",

     

       75
       +
       				EnvVars: []string{"PHOTOCOPY_NERVANA_ENDPOINT"},

     

       76
       +
       			},

     

       77
       +
       			&cli.StringFlag{

     

       78
       +
       				Name:    "nervana-api-key",

     

       79
       +
       				EnvVars: []string{"PHOTOCOPY_NERVANA_API_KEY"},

     

       80
       +
       			},

     

       81
        
       		},

     

       82
        
       		Commands: cli.Commands{

     

       83
        
       			&cli.Command{

     
···

       129
        
       		ClickhouseUser:       cmd.String("clickhouse-user"),

     

       130
        
       		ClickhousePass:       cmd.String("clickhouse-pass"),

     

       131
        
       		RatelimitBypassKey:   cmd.String("ratelimit-bypass-key"),

     

       132
       +
       		NervanaEndpoint:      cmd.String("nervana-endpoint"),

     

       133
       +
       		NervanaApiKey:        cmd.String("nervana-api-key"),

     

       134
        
       	})

     

       135
        
       	if err != nil {

     

       136
        
       		panic(err)

+1 -1

go.mod

···

       13
        
       	github.com/prometheus/client_golang v1.22.0

     

       14
        
       	github.com/urfave/cli/v2 v2.25.7

     

       15
        
       	go.uber.org/ratelimit v0.3.1

     

       16
       -
       	golang.org/x/sync v0.15.0

     

       17
        
       )

     

       18
        
       

     

       19
        
       require (

     
···

       103
        
       	golang.org/x/crypto v0.39.0 // indirect

     

       104
        
       	golang.org/x/exp v0.0.0-20240719175910-8a7402abbf56 // indirect

     

       105
        
       	golang.org/x/net v0.41.0 // indirect

     

       0
        
       
     

       106
        
       	golang.org/x/sys v0.33.0 // indirect

     

       107
        
       	golang.org/x/text v0.26.0 // indirect

     

       108
        
       	golang.org/x/time v0.11.0 // indirect

···

       13
        
       	github.com/prometheus/client_golang v1.22.0

     

       14
        
       	github.com/urfave/cli/v2 v2.25.7

     

       15
        
       	go.uber.org/ratelimit v0.3.1

     

       0
        
       
     

       16
        
       )

     

       17
        
       

     

       18
        
       require (

     
···

       102
        
       	golang.org/x/crypto v0.39.0 // indirect

     

       103
        
       	golang.org/x/exp v0.0.0-20240719175910-8a7402abbf56 // indirect

     

       104
        
       	golang.org/x/net v0.41.0 // indirect

     

       105
       +
       	golang.org/x/sync v0.15.0 // indirect

     

       106
        
       	golang.org/x/sys v0.33.0 // indirect

     

       107
        
       	golang.org/x/text v0.26.0 // indirect

     

       108
        
       	golang.org/x/time v0.11.0 // indirect

+30

handle_create.go

···

       4
        
       	"bytes"

     

       5
        
       	"context"

     

       6
        
       	"fmt"

     

       0
        
       
     

       7
        
       	"strings"

     

       8
        
       	"time"

     

       9
        
       

     
···

       84
        
       		IndexedAt: indexedAt,

     

       85
        
       		Did:       did,

     

       86
        
       		Lang:      lang,

     

       0
        
       
     

       87
        
       	}

     

       88
        
       

     

       89
        
       	if rec.Reply != nil {

     
···

       127
        
       

     

       128
        
       	if err := p.inserters.postsInserter.Insert(ctx, post); err != nil {

     

       129
        
       		return err

     

       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
        
       
     

       130
        
       	}

     

       131
        
       

     

       132
        
       	return nil

···

       4
        
       	"bytes"

     

       5
        
       	"context"

     

       6
        
       	"fmt"

     

       7
       +
       	"slices"

     

       8
        
       	"strings"

     

       9
        
       	"time"

     

       10
        
       

     
···

       85
        
       		IndexedAt: indexedAt,

     

       86
        
       		Did:       did,

     

       87
        
       		Lang:      lang,

     

       88
       +
       		Text:      rec.Text,

     

       89
        
       	}

     

       90
        
       

     

       91
        
       	if rec.Reply != nil {

     
···

       129
        
       

     

       130
        
       	if err := p.inserters.postsInserter.Insert(ctx, post); err != nil {

     

       131
        
       		return err

     

       132
       +
       	}

     

       133
       +
       

     

       134
       +
       	isEn := slices.Contains(rec.Langs, "en")

     

       135
       +
       	if rec.Text != "" && rec.Reply == nil && isEn && p.nervanaClient != nil {

     

       136
       +
       		go func(ctx context.Context, rec bsky.FeedPost, did, rkey string) {

     

       137
       +
       			ctx, cancel := context.WithTimeout(ctx, 5*time.Second)

     

       138
       +
       			defer cancel()

     

       139
       +
       

     

       140
       +
       			nervanaItems, err := p.nervanaClient.MakeRequest(ctx, rec.Text)

     

       141
       +
       			if err != nil {

     

       142
       +
       				p.logger.Error("error making nervana items request", "error", err)

     

       143
       +
       				return

     

       144
       +
       			}

     

       145
       +
       

     

       146
       +
       			for _, ni := range nervanaItems {

     

       147
       +
       				postLabel := models.PostLabel{

     

       148
       +
       					Did:         did,

     

       149
       +
       					Rkey:        rkey,

     

       150
       +
       					Text:        ni.Text,

     

       151
       +
       					Label:       ni.Label,

     

       152
       +
       					EntityId:    ni.EntityId,

     

       153
       +
       					Description: ni.Description,

     

       154
       +
       					Topic:       "",

     

       155
       +
       					CreatedAt:   time.Now(),

     

       156
       +
       				}

     

       157
       +
       				p.inserters.labelsInserter.Insert(ctx, postLabel)

     

       158
       +
       			}

     

       159
       +
       		}(ctx, rec, did, rkey)

     

       160
        
       	}

     

       161
        
       

     

       162
        
       	return nil

models/post.go

···

       15
        
       	QuoteUri  string    `ch:"quote_uri"`

     

       16
        
       	QuoteDid  string    `ch:"quote_did"`

     

       17
        
       	Lang      string    `ch:"lang"`

     

       0
        
       
     

       18
        
       }

···

       15
        
       	QuoteUri  string    `ch:"quote_uri"`

     

       16
        
       	QuoteDid  string    `ch:"quote_did"`

     

       17
        
       	Lang      string    `ch:"lang"`

     

       18
       +
       	Text      string    `ch:"text"`

     

       19
        
       }

+14

models/post_label.go

···

       1
       +
       package models

     

       2
       +
       

     

       3
       +
       import "time"

     

       4
       +
       

     

       5
       +
       type PostLabel struct {

     

       6
       +
       	Did         string    `ch:"did"`

     

       7
       +
       	Rkey        string    `ch:"rkey"`

     

       8
       +
       	CreatedAt   time.Time `ch:"created_at"`

     

       9
       +
       	Text        string    `ch:"text"`

     

       10
       +
       	Label       string    `ch:"label"`

     

       11
       +
       	EntityId    string    `ch:"entity_id"`

     

       12
       +
       	Description string    `ch:"description"`

     

       13
       +
       	Topic       string    `ch:"topic"`

     

       14
       +
       }

+77

nervana/client.go

···

       1
       +
       package nervana

     

       2
       +
       

     

       3
       +
       import (

     

       4
       +
       	"bytes"

     

       5
       +
       	"context"

     

       6
       +
       	"encoding/json"

     

       7
       +
       	"fmt"

     

       8
       +
       	"io"

     

       9
       +
       	"net/http"

     

       10
       +
       	"time"

     

       11
       +
       )

     

       12
       +
       

     

       13
       +
       type Client struct {

     

       14
       +
       	cli      *http.Client

     

       15
       +
       	endpoint string

     

       16
       +
       	apiKey   string

     

       17
       +
       }

     

       18
       +
       

     

       19
       +
       func NewClient(endpoint string, apiKey string) *Client {

     

       20
       +
       	return &Client{

     

       21
       +
       		cli: &http.Client{

     

       22
       +
       			Timeout: 5 * time.Second,

     

       23
       +
       		},

     

       24
       +
       		endpoint: endpoint,

     

       25
       +
       		apiKey:   apiKey,

     

       26
       +
       	}

     

       27
       +
       }

     

       28
       +
       

     

       29
       +
       type NervanaItem struct {

     

       30
       +
       	Text        string `json:"text"`

     

       31
       +
       	Label       string `json:"label"`

     

       32
       +
       	EntityId    string `json:"entityId"`

     

       33
       +
       	Description string `json:"description"`

     

       34
       +
       }

     

       35
       +
       

     

       36
       +
       func (c *Client) newRequest(ctx context.Context, text string) (*http.Request, error) {

     

       37
       +
       	payload := map[string]string{

     

       38
       +
       		"text":     text,

     

       39
       +
       		"language": "en",

     

       40
       +
       	}

     

       41
       +
       

     

       42
       +
       	b, err := json.Marshal(payload)

     

       43
       +
       	if err != nil {

     

       44
       +
       		return nil, err

     

       45
       +
       	}

     

       46
       +
       

     

       47
       +
       	req, err := http.NewRequestWithContext(ctx, "GET", c.endpoint, bytes.NewReader(b))

     

       48
       +
       

     

       49
       +
       	req.Header.Set("Authorization", "Bearer "+c.apiKey)

     

       50
       +
       

     

       51
       +
       	return req, err

     

       52
       +
       }

     

       53
       +
       

     

       54
       +
       func (c *Client) MakeRequest(ctx context.Context, text string) ([]NervanaItem, error) {

     

       55
       +
       	req, err := c.newRequest(ctx, text)

     

       56
       +
       	if err != nil {

     

       57
       +
       		return nil, err

     

       58
       +
       	}

     

       59
       +
       

     

       60
       +
       	resp, err := c.cli.Do(req)

     

       61
       +
       	if err != nil {

     

       62
       +
       		return nil, err

     

       63
       +
       	}

     

       64
       +
       	defer resp.Body.Close()

     

       65
       +
       

     

       66
       +
       	if resp.StatusCode != 200 {

     

       67
       +
       		io.Copy(io.Discard, resp.Body)

     

       68
       +
       		return nil, fmt.Errorf("received non-200 response code: %d", resp.StatusCode)

     

       69
       +
       	}

     

       70
       +
       

     

       71
       +
       	var nervanaResp []NervanaItem

     

       72
       +
       	if err := json.NewDecoder(resp.Body).Decode(&nervanaResp); err != nil {

     

       73
       +
       		return nil, err

     

       74
       +
       	}

     

       75
       +
       

     

       76
       +
       	return nervanaResp, nil

     

       77
       +
       }

+31 -5

photocopy.go

···

       11
        
       	"github.com/ClickHouse/clickhouse-go/v2"

     

       12
        
       	"github.com/ClickHouse/clickhouse-go/v2/lib/driver"

     

       13
        
       	"github.com/haileyok/photocopy/clickhouse_inserter"

     

       0
        
       
     

       14
        
       	"github.com/prometheus/client_golang/prometheus"

     

       15
        
       	"github.com/prometheus/client_golang/prometheus/promauto"

     

       16
        
       	"github.com/prometheus/client_golang/prometheus/promhttp"

     
···

       32
        
       	ratelimitBypassKey string

     

       33
        
       

     

       34
        
       	conn driver.Conn

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       35
        
       }

     

       36
        
       

     

       37
        
       type Inserters struct {

     
···

       41
        
       	plcInserter          *clickhouse_inserter.Inserter

     

       42
        
       	recordsInserter      *clickhouse_inserter.Inserter

     

       43
        
       	deletesInserter      *clickhouse_inserter.Inserter

     

       0
        
       
     

       44
        
       }

     

       45
        
       

     

       46
        
       type Args struct {

     
···

       54
        
       	ClickhouseUser       string

     

       55
        
       	ClickhousePass       string

     

       56
        
       	RatelimitBypassKey   string

     

       0
        
       
     

       0
        
       
     

       57
        
       }

     

       58
        
       

     

       59
        
       func New(ctx context.Context, args *Args) (*Photocopy, error) {

     
···

       88
        
       	fi, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       89
        
       		PrometheusCounterPrefix: "photocopy_follows",

     

       90
        
       		Histogram:               insertionsHist,

     

       91
       -
       		BatchSize:               250_000,

     

       92
        
       		Logger:                  p.logger,

     

       93
        
       		Conn:                    conn,

     

       94
        
       		Query:                   "INSERT INTO follow (uri, did, rkey, created_at, indexed_at, subject)",

     
···

       101
        
       	pi, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       102
        
       		PrometheusCounterPrefix: "photocopy_posts",

     

       103
        
       		Histogram:               insertionsHist,

     

       104
       -
       		BatchSize:               250_000,

     

       105
        
       		Logger:                  p.logger,

     

       106
        
       		Conn:                    conn,

     

       107
        
       		Query:                   "INSERT INTO post (uri, did, rkey, created_at, indexed_at, root_uri, root_did, parent_uri, parent_did, quote_uri, quote_did, lang)",

     
···

       114
        
       	ii, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       115
        
       		PrometheusCounterPrefix: "photocopy_interactions",

     

       116
        
       		Histogram:               insertionsHist,

     

       117
       -
       		BatchSize:               250_000,

     

       118
        
       		Logger:                  p.logger,

     

       119
        
       		Conn:                    conn,

     

       120
        
       		Query:                   "INSERT INTO interaction (uri, did, rkey, kind, created_at, indexed_at, subject_uri, subject_did)",

     
···

       127
        
       	ri, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       128
        
       		PrometheusCounterPrefix: "photocopy_records",

     

       129
        
       		Histogram:               insertionsHist,

     

       130
       -
       		BatchSize:               250_000,

     

       131
        
       		Logger:                  p.logger,

     

       132
        
       		Conn:                    conn,

     

       133
        
       		Query:                   "INSERT INTO record (did, rkey, collection, cid, seq, raw, created_at)",

     
···

       140
        
       	di, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       141
        
       		PrometheusCounterPrefix: "photocopy_deletes",

     

       142
        
       		Histogram:               insertionsHist,

     

       143
       -
       		BatchSize:               250_000,

     

       144
        
       		Logger:                  p.logger,

     

       145
        
       		Conn:                    conn,

     

       146
        
       		Query:                   "INSERT INTO delete (did, rkey, created_at)",

     
···

       150
        
       		return nil, err

     

       151
        
       	}

     

       152
        
       

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       153
        
       	is := &Inserters{

     

       154
        
       		followsInserter:      fi,

     

       155
        
       		postsInserter:        pi,

     

       156
        
       		interactionsInserter: ii,

     

       157
        
       		recordsInserter:      ri,

     

       158
        
       		deletesInserter:      di,

     

       0
        
       
     

       159
        
       	}

     

       160
        
       

     

       161
        
       	p.inserters = is

     
···

       189
        
       

     

       190
        
       	p.inserters.plcInserter = plci

     

       191
        
       	p.plcScraper = plcs

     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       0
        
       
     

       192
        
       

     

       193
        
       	return p, nil

     

       194
        
       }

···

       11
        
       	"github.com/ClickHouse/clickhouse-go/v2"

     

       12
        
       	"github.com/ClickHouse/clickhouse-go/v2/lib/driver"

     

       13
        
       	"github.com/haileyok/photocopy/clickhouse_inserter"

     

       14
       +
       	"github.com/haileyok/photocopy/nervana"

     

       15
        
       	"github.com/prometheus/client_golang/prometheus"

     

       16
        
       	"github.com/prometheus/client_golang/prometheus/promauto"

     

       17
        
       	"github.com/prometheus/client_golang/prometheus/promhttp"

     
···

       33
        
       	ratelimitBypassKey string

     

       34
        
       

     

       35
        
       	conn driver.Conn

     

       36
       +
       

     

       37
       +
       	nervanaClient   *nervana.Client

     

       38
       +
       	nervanaEndpoint string

     

       39
       +
       	nervanaApiKey   string

     

       40
        
       }

     

       41
        
       

     

       42
        
       type Inserters struct {

     
···

       46
        
       	plcInserter          *clickhouse_inserter.Inserter

     

       47
        
       	recordsInserter      *clickhouse_inserter.Inserter

     

       48
        
       	deletesInserter      *clickhouse_inserter.Inserter

     

       49
       +
       	labelsInserter       *clickhouse_inserter.Inserter

     

       50
        
       }

     

       51
        
       

     

       52
        
       type Args struct {

     
···

       60
        
       	ClickhouseUser       string

     

       61
        
       	ClickhousePass       string

     

       62
        
       	RatelimitBypassKey   string

     

       63
       +
       	NervanaEndpoint      string

     

       64
       +
       	NervanaApiKey        string

     

       65
        
       }

     

       66
        
       

     

       67
        
       func New(ctx context.Context, args *Args) (*Photocopy, error) {

     
···

       96
        
       	fi, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       97
        
       		PrometheusCounterPrefix: "photocopy_follows",

     

       98
        
       		Histogram:               insertionsHist,

     

       99
       +
       		BatchSize:               500,

     

       100
        
       		Logger:                  p.logger,

     

       101
        
       		Conn:                    conn,

     

       102
        
       		Query:                   "INSERT INTO follow (uri, did, rkey, created_at, indexed_at, subject)",

     
···

       109
        
       	pi, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       110
        
       		PrometheusCounterPrefix: "photocopy_posts",

     

       111
        
       		Histogram:               insertionsHist,

     

       112
       +
       		BatchSize:               300,

     

       113
        
       		Logger:                  p.logger,

     

       114
        
       		Conn:                    conn,

     

       115
        
       		Query:                   "INSERT INTO post (uri, did, rkey, created_at, indexed_at, root_uri, root_did, parent_uri, parent_did, quote_uri, quote_did, lang)",

     
···

       122
        
       	ii, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       123
        
       		PrometheusCounterPrefix: "photocopy_interactions",

     

       124
        
       		Histogram:               insertionsHist,

     

       125
       +
       		BatchSize:               1000,

     

       126
        
       		Logger:                  p.logger,

     

       127
        
       		Conn:                    conn,

     

       128
        
       		Query:                   "INSERT INTO interaction (uri, did, rkey, kind, created_at, indexed_at, subject_uri, subject_did)",

     
···

       135
        
       	ri, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       136
        
       		PrometheusCounterPrefix: "photocopy_records",

     

       137
        
       		Histogram:               insertionsHist,

     

       138
       +
       		BatchSize:               2500,

     

       139
        
       		Logger:                  p.logger,

     

       140
        
       		Conn:                    conn,

     

       141
        
       		Query:                   "INSERT INTO record (did, rkey, collection, cid, seq, raw, created_at)",

     
···

       148
        
       	di, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       149
        
       		PrometheusCounterPrefix: "photocopy_deletes",

     

       150
        
       		Histogram:               insertionsHist,

     

       151
       +
       		BatchSize:               500,

     

       152
        
       		Logger:                  p.logger,

     

       153
        
       		Conn:                    conn,

     

       154
        
       		Query:                   "INSERT INTO delete (did, rkey, created_at)",

     
···

       158
        
       		return nil, err

     

       159
        
       	}

     

       160
        
       

     

       161
       +
       	li, err := clickhouse_inserter.New(ctx, &clickhouse_inserter.Args{

     

       162
       +
       		PrometheusCounterPrefix: "photocopy_labels",

     

       163
       +
       		Histogram:               insertionsHist,

     

       164
       +
       		BatchSize:               100,

     

       165
       +
       		Logger:                  p.logger,

     

       166
       +
       		Conn:                    conn,

     

       167
       +
       		Query:                   "INSERT INTO post_label (did, rkey, text, label, entity_id, description, topic, created_at)",

     

       168
       +
       		RateLimit:               3,

     

       169
       +
       	})

     

       170
       +
       	if err != nil {

     

       171
       +
       		return nil, err

     

       172
       +
       	}

     

       173
       +
       

     

       174
        
       	is := &Inserters{

     

       175
        
       		followsInserter:      fi,

     

       176
        
       		postsInserter:        pi,

     

       177
        
       		interactionsInserter: ii,

     

       178
        
       		recordsInserter:      ri,

     

       179
        
       		deletesInserter:      di,

     

       180
       +
       		labelsInserter:       li,

     

       181
        
       	}

     

       182
        
       

     

       183
        
       	p.inserters = is

     
···

       211
        
       

     

       212
        
       	p.inserters.plcInserter = plci

     

       213
        
       	p.plcScraper = plcs

     

       214
       +
       

     

       215
       +
       	if args.NervanaApiKey != "" && args.NervanaEndpoint != "" {

     

       216
       +
       		p.nervanaClient = nervana.NewClient(args.NervanaEndpoint, args.NervanaApiKey)

     

       217
       +
       	}

     

       218
        
       

     

       219
        
       	return p, nil

     

       220
        
       }

Compare changes