This repository has no description
0

Configure Feed

Select the types of activity you want to include in your feed.

core / appview / state / knotstream.go
13 kB 455 lines
1package state 2 3import ( 4 "context" 5 "database/sql" 6 "encoding/json" 7 "errors" 8 "fmt" 9 "slices" 10 "strings" 11 "time" 12 13 "tangled.org/core/appview/cloudflare" 14 "tangled.org/core/appview/notify" 15 16 "tangled.org/core/api/tangled" 17 "tangled.org/core/appview/cache" 18 "tangled.org/core/appview/config" 19 "tangled.org/core/appview/db" 20 "tangled.org/core/appview/models" 21 "tangled.org/core/appview/sites" 22 ec "tangled.org/core/eventconsumer" 23 "tangled.org/core/eventconsumer/cursor" 24 knotdb "tangled.org/core/knotserver/db" 25 "tangled.org/core/log" 26 "tangled.org/core/orm" 27 "tangled.org/core/rbac" 28 "tangled.org/core/workflow" 29 30 "github.com/bluesky-social/indigo/atproto/syntax" 31 "github.com/go-git/go-git/v5/plumbing" 32 "github.com/posthog/posthog-go" 33) 34 35func Knotstream(ctx context.Context, c *config.Config, d *db.DB, enforcer *rbac.Enforcer, posthog posthog.Client, notifier notify.Notifier, cfClient *cloudflare.Client) (*ec.Consumer, error) { 36 logger := log.FromContext(ctx) 37 logger = log.SubLogger(logger, "knotstream") 38 39 knots, err := db.GetRegistrations( 40 d, 41 orm.FilterIsNot("registered", "null"), 42 ) 43 if err != nil { 44 return nil, err 45 } 46 47 srcs := make(map[ec.Source]struct{}) 48 for _, k := range knots { 49 s := ec.NewKnotSource(k.Domain) 50 srcs[s] = struct{}{} 51 } 52 53 cache := cache.New(c.Redis.Addr) 54 cursorStore := cursor.NewRedisCursorStore(cache) 55 56 cfg := ec.ConsumerConfig{ 57 Sources: srcs, 58 ProcessFunc: knotIngester(d, enforcer, posthog, notifier, c.Core.Dev, c, cfClient), 59 RetryInterval: c.Knotstream.RetryInterval, 60 MaxRetryInterval: c.Knotstream.MaxRetryInterval, 61 ConnectionTimeout: c.Knotstream.ConnectionTimeout, 62 WorkerCount: c.Knotstream.WorkerCount, 63 QueueSize: c.Knotstream.QueueSize, 64 Logger: logger, 65 Dev: c.Core.Dev, 66 CursorStore: &cursorStore, 67 } 68 69 return ec.NewConsumer(cfg), nil 70} 71 72func resolveRepo(d *db.DB, repoDid *string, ownerDid, repoName string) (*models.Repo, error) { 73 if repoDid != nil && *repoDid != "" { 74 return db.GetRepoByDid(d, *repoDid) 75 } 76 repos, err := db.GetRepos(d, orm.FilterEq("did", ownerDid), orm.FilterEq("rkey", strings.ToLower(repoName))) 77 if err != nil { 78 return nil, err 79 } 80 if len(repos) == 0 { 81 return nil, sql.ErrNoRows 82 } 83 return &repos[0], nil 84} 85 86func knotIngester(d *db.DB, enforcer *rbac.Enforcer, posthog posthog.Client, notifier notify.Notifier, dev bool, c *config.Config, cfClient *cloudflare.Client) ec.ProcessFunc { 87 return func(ctx context.Context, source ec.Source, msg ec.Message) error { 88 switch msg.Nsid { 89 case tangled.GitRefUpdateNSID: 90 return ingestRefUpdate(ctx, d, enforcer, posthog, notifier, dev, c, cfClient, source, msg) 91 case tangled.PipelineNSID: 92 return ingestPipeline(d, source, msg) 93 case knotdb.RepoDIDAssignNSID: 94 return ingestDIDAssign(d, enforcer, source, msg, ctx) 95 } 96 97 return nil 98 } 99} 100 101func ingestRefUpdate(ctx context.Context, d *db.DB, enforcer *rbac.Enforcer, pc posthog.Client, notifier notify.Notifier, dev bool, c *config.Config, cfClient *cloudflare.Client, source ec.Source, msg ec.Message) error { 102 logger := log.FromContext(ctx) 103 104 var record tangled.GitRefUpdate 105 err := json.Unmarshal(msg.EventJson, &record) 106 if err != nil { 107 return err 108 } 109 110 knownKnots, err := enforcer.GetKnotsForUser(record.CommitterDid) 111 if err != nil { 112 return err 113 } 114 if !slices.Contains(knownKnots, source.Key()) { 115 return fmt.Errorf("%s does not belong to %s, something is fishy", record.CommitterDid, source.Key()) 116 } 117 118 if record.Repo == "" { 119 return fmt.Errorf("gitRefUpdate from %s missing repo", source.Key()) 120 } 121 122 repo, lookupErr := db.GetRepoByDid(d, record.Repo) 123 if lookupErr != nil { 124 return fmt.Errorf("failed to look up repo: %w", lookupErr) 125 } 126 127 logger.Info("processing gitRefUpdate event", 128 "repo", repo.RepoIdentifier(), 129 "ref", record.Ref, 130 "old_sha", record.OldSha, 131 "new_sha", record.NewSha) 132 133 notifier.Push(ctx, repo, record.Ref, record.OldSha, record.NewSha, record.CommitterDid) 134 135 errPunchcard := populatePunchcard(d, record) 136 errLanguages := updateRepoLanguages(d, record) 137 138 var errPosthog error 139 if !dev && record.CommitterDid != "" { 140 errPosthog = pc.Enqueue(posthog.Capture{ 141 DistinctId: record.CommitterDid, 142 Event: "git_ref_update", 143 }) 144 } 145 146 // Trigger a sites redeploy if this push is to the configured sites branch. 147 if cfClient.Enabled() { 148 go triggerSitesDeployIfNeeded(ctx, d, cfClient, c, record, source) 149 } 150 151 return errors.Join(errPunchcard, errLanguages, errPosthog) 152} 153 154// triggerSitesDeployIfNeeded checks whether the pushed ref matches the sites 155// branch configured for this repo and, if so, syncs the site to R2 156func triggerSitesDeployIfNeeded(ctx context.Context, d *db.DB, cfClient *cloudflare.Client, cfg *config.Config, record tangled.GitRefUpdate, source ec.Source) { 157 logger := log.FromContext(ctx) 158 159 ref := plumbing.ReferenceName(record.Ref) 160 if !ref.IsBranch() { 161 return 162 } 163 pushedBranch := ref.Short() 164 165 repo, err := db.GetRepoByDid(d, record.Repo) 166 if err != nil { 167 return 168 } 169 170 siteConfig, err := db.GetRepoSiteConfig(d, repo.RepoDid) 171 if err != nil || siteConfig == nil { 172 return 173 } 174 if siteConfig.Branch != pushedBranch { 175 return 176 } 177 178 deploy := &models.SiteDeploy{ 179 RepoDid: syntax.DID(repo.RepoDid), 180 Branch: siteConfig.Branch, 181 Dir: siteConfig.Dir, 182 CommitSHA: record.NewSha, 183 Trigger: models.SiteDeployTriggerPush, 184 } 185 186 deployErr := sites.Deploy(ctx, cfClient, cfg, repo, siteConfig.Branch, siteConfig.Dir) 187 if deployErr != nil { 188 logger.Error("sites: R2 sync failed on push", "repo", repo.RepoIdentifier(), "err", deployErr) 189 deploy.Status = models.SiteDeployStatusFailure 190 deploy.Error = deployErr.Error() 191 } else { 192 deploy.Status = models.SiteDeployStatusSuccess 193 } 194 195 if err := db.AddSiteDeploy(d, deploy); err != nil { 196 logger.Error("sites: failed to record deploy", "repo", repo.RepoIdentifier(), "err", err) 197 } 198 199 if deployErr == nil { 200 logger.Info("site deployed to r2", "repo", repo.RepoIdentifier()) 201 } 202} 203 204func populatePunchcard(d *db.DB, record tangled.GitRefUpdate) error { 205 if record.CommitterDid == "" { 206 return nil 207 } 208 209 knownEmails, err := db.GetAllEmails(d, record.CommitterDid) 210 if err != nil { 211 return err 212 } 213 214 count := 0 215 for _, ke := range knownEmails { 216 if record.Meta == nil { 217 continue 218 } 219 if record.Meta.CommitCount == nil { 220 continue 221 } 222 for _, ce := range record.Meta.CommitCount.ByEmail { 223 if ce == nil { 224 continue 225 } 226 if ce.Email == ke.Address || ce.Email == record.CommitterDid { 227 count += int(ce.Count) 228 } 229 } 230 } 231 232 punch := models.Punch{ 233 Did: record.CommitterDid, 234 Date: time.Now(), 235 Count: count, 236 } 237 return db.AddPunch(d, punch) 238} 239 240func updateRepoLanguages(d *db.DB, record tangled.GitRefUpdate) error { 241 if record.Meta == nil || record.Meta.LangBreakdown == nil || record.Meta.LangBreakdown.Inputs == nil { 242 return fmt.Errorf("empty language data for repo: %s", record.Repo) 243 } 244 245 r, lookupErr := db.GetRepoByDid(d, record.Repo) 246 if lookupErr != nil { 247 return fmt.Errorf("failed to look up repo: %w", lookupErr) 248 } 249 repo := *r 250 251 ref := plumbing.ReferenceName(record.Ref) 252 if !ref.IsBranch() { 253 return fmt.Errorf("%s is not a valid reference name", ref) 254 } 255 256 var langs []models.RepoLanguage 257 for _, l := range record.Meta.LangBreakdown.Inputs { 258 if l == nil { 259 continue 260 } 261 262 langs = append(langs, models.RepoLanguage{ 263 RepoDid: syntax.DID(repo.RepoDid), 264 Ref: ref.Short(), 265 IsDefaultRef: record.Meta.IsDefaultRef, 266 Language: l.Lang, 267 Bytes: l.Size, 268 }) 269 } 270 271 tx, err := d.Begin() 272 if err != nil { 273 return err 274 } 275 defer tx.Rollback() 276 277 // update appview's cache 278 err = db.UpdateRepoLanguages(tx, syntax.DID(repo.RepoDid), ref.Short(), langs) 279 if err != nil { 280 fmt.Printf("failed; %s\n", err) 281 // non-fatal 282 } 283 284 return tx.Commit() 285} 286 287func ingestPipeline(d *db.DB, source ec.Source, msg ec.Message) error { 288 var record tangled.Pipeline 289 err := json.Unmarshal(msg.EventJson, &record) 290 if err != nil { 291 return err 292 } 293 294 if record.TriggerMetadata == nil { 295 return fmt.Errorf("empty trigger metadata: nsid %s, rkey %s", msg.Nsid, msg.Rkey) 296 } 297 298 if record.TriggerMetadata.Repo == nil { 299 return fmt.Errorf("empty repo: nsid %s, rkey %s", msg.Nsid, msg.Rkey) 300 } 301 302 repoName := "" 303 if record.TriggerMetadata.Repo.Repo != nil { 304 repoName = *record.TriggerMetadata.Repo.Repo 305 } 306 307 repo, lookupErr := resolveRepo(d, record.TriggerMetadata.Repo.RepoDid, record.TriggerMetadata.Repo.Did, repoName) 308 if lookupErr != nil { 309 return fmt.Errorf("failed to look up repo: %w", lookupErr) 310 } 311 if repo.Spindle == "" { 312 return fmt.Errorf("repo does not have a spindle configured yet: nsid %s, rkey %s", msg.Nsid, msg.Rkey) 313 } 314 315 // trigger info 316 var trigger models.Trigger 317 var sha string 318 trigger.Kind = workflow.TriggerKind(record.TriggerMetadata.Kind) 319 switch trigger.Kind { 320 case workflow.TriggerKindPush: 321 trigger.PushRef = &record.TriggerMetadata.Push.Ref 322 trigger.PushNewSha = &record.TriggerMetadata.Push.NewSha 323 trigger.PushOldSha = &record.TriggerMetadata.Push.OldSha 324 sha = *trigger.PushNewSha 325 case workflow.TriggerKindPullRequest: 326 trigger.PRSourceBranch = &record.TriggerMetadata.PullRequest.SourceBranch 327 trigger.PRTargetBranch = &record.TriggerMetadata.PullRequest.TargetBranch 328 trigger.PRSourceSha = &record.TriggerMetadata.PullRequest.SourceSha 329 trigger.PRAction = &record.TriggerMetadata.PullRequest.Action 330 sha = *trigger.PRSourceSha 331 } 332 333 tx, err := d.Begin() 334 if err != nil { 335 return fmt.Errorf("failed to start txn: %w", err) 336 } 337 338 triggerId, err := db.AddTrigger(tx, trigger) 339 if err != nil { 340 return fmt.Errorf("failed to add trigger entry: %w", err) 341 } 342 343 pipeline := models.Pipeline{ 344 Rkey: msg.Rkey, 345 Knot: source.Key(), 346 RepoOwner: syntax.DID(record.TriggerMetadata.Repo.Did), 347 RepoName: repoName, 348 RepoDid: repo.RepoDid, 349 TriggerId: int(triggerId), 350 Sha: sha, 351 } 352 353 err = db.AddPipeline(tx, pipeline) 354 if err != nil { 355 return fmt.Errorf("failed to add pipeline: %w", err) 356 } 357 358 err = tx.Commit() 359 if err != nil { 360 return fmt.Errorf("failed to commit txn: %w", err) 361 } 362 363 return nil 364} 365 366func ingestDIDAssign(d *db.DB, enforcer *rbac.Enforcer, source ec.Source, msg ec.Message, ctx context.Context) error { 367 logger := log.FromContext(ctx) 368 369 var record knotdb.RepoDIDAssign 370 if err := json.Unmarshal(msg.EventJson, &record); err != nil { 371 return fmt.Errorf("unmarshal didAssign: %w", err) 372 } 373 374 if record.RepoDid == "" || record.OwnerDid == "" || record.RepoName == "" { 375 return fmt.Errorf("didAssign missing required fields: repoDid=%q ownerDid=%q repoName=%q", 376 record.RepoDid, record.OwnerDid, record.RepoName) 377 } 378 379 logger.Info("processing didAssign event", 380 "repo_did", record.RepoDid, 381 "owner_did", record.OwnerDid, 382 "repo_name", record.RepoName) 383 384 repos, err := db.GetRepos(d, 385 orm.FilterEq("did", record.OwnerDid), 386 orm.FilterEq("rkey", strings.ToLower(record.RepoName)), 387 ) 388 if err != nil || len(repos) == 0 { 389 logger.Warn("didAssign for unknown repo, skipping", 390 "owner_did", record.OwnerDid, 391 "repo_name", record.RepoName) 392 return nil 393 } 394 repo := repos[0] 395 knot := source.Key() 396 397 if repo.Knot != knot { 398 return fmt.Errorf("didAssign from %s for repo hosted on %s, rejecting", knot, repo.Knot) 399 } 400 401 repoAtUri := repo.RepoAt().String() 402 legacyResource := record.OwnerDid + "/" + record.RepoName 403 404 if repo.RepoDid != record.RepoDid { 405 tx, err := d.Begin() 406 if err != nil { 407 return fmt.Errorf("begin didAssign txn: %w", err) 408 } 409 defer tx.Rollback() 410 411 if err := db.CascadeRepoDid(tx, repoAtUri, record.RepoDid); err != nil { 412 return fmt.Errorf("cascade repo_did: %w", err) 413 } 414 415 if err := db.EnqueuePdsRewritesForRepo(tx, record.RepoDid, repoAtUri); err != nil { 416 return fmt.Errorf("enqueue pds rewrites: %w", err) 417 } 418 419 if err := tx.Commit(); err != nil { 420 return fmt.Errorf("commit didAssign txn: %w", err) 421 } 422 } 423 424 if err := enforcer.RemoveRepo(record.OwnerDid, knot, legacyResource); err != nil { 425 return fmt.Errorf("remove legacy RBAC policies for %s: %w", legacyResource, err) 426 } 427 if err := enforcer.AddRepo(record.OwnerDid, knot, record.RepoDid); err != nil { 428 return fmt.Errorf("add RBAC policies for %s: %w", record.RepoDid, err) 429 } 430 431 collabs, collabErr := db.GetCollaborators(d, orm.FilterEq("repo_did", record.RepoDid)) 432 if collabErr != nil { 433 return fmt.Errorf("get collaborators for RBAC update: %w", collabErr) 434 } 435 for _, c := range collabs { 436 collabDid := c.SubjectDid.String() 437 if err := enforcer.RemoveCollaborator(collabDid, knot, legacyResource); err != nil { 438 return fmt.Errorf("remove collaborator RBAC for %s: %w", collabDid, err) 439 } 440 if err := enforcer.AddCollaborator(collabDid, knot, record.RepoDid); err != nil { 441 return fmt.Errorf("add collaborator RBAC for %s: %w", collabDid, err) 442 } 443 } 444 445 if err := enforcer.E.SavePolicy(); err != nil { 446 return fmt.Errorf("save RBAC policies after didAssign: %w", err) 447 } 448 449 logger.Info("didAssign processed successfully", 450 "repo_did", record.RepoDid, 451 "owner_did", record.OwnerDid, 452 "repo_name", record.RepoName) 453 454 return nil 455}