comparing ec8ecf508f0c3df525fb5c88aad6193748149e0b and hailey/nervana-topics on hailey.at/photocopy

+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)

+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