diff options
Diffstat (limited to 'internal/federation/dereferencing')
-rw-r--r-- | internal/federation/dereferencing/account.go | 243 | ||||
-rw-r--r-- | internal/federation/dereferencing/announce.go | 65 | ||||
-rw-r--r-- | internal/federation/dereferencing/blocked.go | 41 | ||||
-rw-r--r-- | internal/federation/dereferencing/collectionpage.go | 70 | ||||
-rw-r--r-- | internal/federation/dereferencing/dereferencer.go | 73 | ||||
-rw-r--r-- | internal/federation/dereferencing/handshake.go | 98 | ||||
-rw-r--r-- | internal/federation/dereferencing/instance.go | 40 | ||||
-rw-r--r-- | internal/federation/dereferencing/status.go | 369 | ||||
-rw-r--r-- | internal/federation/dereferencing/thread.go | 250 |
9 files changed, 1249 insertions, 0 deletions
diff --git a/internal/federation/dereferencing/account.go b/internal/federation/dereferencing/account.go new file mode 100644 index 000000000..c403ec66f --- /dev/null +++ b/internal/federation/dereferencing/account.go @@ -0,0 +1,243 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import ( + "context" + "encoding/json" + "errors" + "fmt" + "net/url" + + "github.com/go-fed/activity/streams" + "github.com/go-fed/activity/streams/vocab" + "github.com/sirupsen/logrus" + "github.com/superseriousbusiness/gotosocial/internal/ap" + "github.com/superseriousbusiness/gotosocial/internal/db" + "github.com/superseriousbusiness/gotosocial/internal/gtsmodel" + "github.com/superseriousbusiness/gotosocial/internal/id" + "github.com/superseriousbusiness/gotosocial/internal/transport" +) + +// EnrichRemoteAccount takes an account that's already been inserted into the database in a minimal form, +// and populates it with additional fields, media, etc. +// +// EnrichRemoteAccount is mostly useful for calling after an account has been initially created by +// the federatingDB's Create function, or during the federated authorization flow. +func (d *deref) EnrichRemoteAccount(username string, account *gtsmodel.Account) (*gtsmodel.Account, error) { + if err := d.populateAccountFields(account, username, false); err != nil { + return nil, err + } + + if err := d.db.UpdateByID(account.ID, account); err != nil { + return nil, fmt.Errorf("EnrichRemoteAccount: error updating account: %s", err) + } + + return account, nil +} + +// GetRemoteAccount completely dereferences a remote account, converts it to a GtS model account, +// puts it in the database, and returns it to a caller. The boolean indicates whether the account is new +// to us or not. If we haven't seen the account before, bool will be true. If we have seen the account before, +// it will be false. +// +// Refresh indicates whether--if the account exists in our db already--it should be refreshed by calling +// the remote instance again. +// +// SIDE EFFECTS: remote account will be stored in the database, or updated if it already exists (and refresh is true). +func (d *deref) GetRemoteAccount(username string, remoteAccountID *url.URL, refresh bool) (*gtsmodel.Account, bool, error) { + new := true + + // check if we already have the account in our db + maybeAccount := >smodel.Account{} + if err := d.db.GetWhere([]db.Where{{Key: "uri", Value: remoteAccountID.String()}}, maybeAccount); err == nil { + // we've seen this account before so it's not new + new = false + + // if we're not being asked to refresh, we can just return the maybeAccount as-is and avoid doing any external calls + if !refresh { + return maybeAccount, new, nil + } + } + + accountable, err := d.dereferenceAccountable(username, remoteAccountID) + if err != nil { + return nil, new, fmt.Errorf("FullyDereferenceAccount: error dereferencing accountable: %s", err) + } + + gtsAccount, err := d.typeConverter.ASRepresentationToAccount(accountable, false) + if err != nil { + return nil, new, fmt.Errorf("FullyDereferenceAccount: error converting accountable to account: %s", err) + } + + if new { + // generate a new id since we haven't seen this account before, and do a put + ulid, err := id.NewRandomULID() + if err != nil { + return nil, new, fmt.Errorf("FullyDereferenceAccount: error generating new id for account: %s", err) + } + gtsAccount.ID = ulid + + if err := d.populateAccountFields(gtsAccount, username, refresh); err != nil { + return nil, new, fmt.Errorf("FullyDereferenceAccount: error populating further account fields: %s", err) + } + + if err := d.db.Put(gtsAccount); err != nil { + return nil, new, fmt.Errorf("FullyDereferenceAccount: error putting new account: %s", err) + } + } else { + // take the id we already have and do an update + gtsAccount.ID = maybeAccount.ID + + if err := d.populateAccountFields(gtsAccount, username, refresh); err != nil { + return nil, new, fmt.Errorf("FullyDereferenceAccount: error populating further account fields: %s", err) + } + + if err := d.db.UpdateByID(gtsAccount.ID, gtsAccount); err != nil { + return nil, new, fmt.Errorf("FullyDereferenceAccount: error updating existing account: %s", err) + } + } + + return gtsAccount, new, nil +} + +// dereferenceAccountable calls remoteAccountID with a GET request, and tries to parse whatever +// it finds as something that an account model can be constructed out of. +// +// Will work for Person, Application, or Service models. +func (d *deref) dereferenceAccountable(username string, remoteAccountID *url.URL) (ap.Accountable, error) { + d.startHandshake(username, remoteAccountID) + defer d.stopHandshake(username, remoteAccountID) + + if blocked, err := d.blockedDomain(remoteAccountID.Host); blocked || err != nil { + return nil, fmt.Errorf("DereferenceAccountable: domain %s is blocked", remoteAccountID.Host) + } + + transport, err := d.transportController.NewTransportForUsername(username) + if err != nil { + return nil, fmt.Errorf("DereferenceAccountable: transport err: %s", err) + } + + b, err := transport.Dereference(context.Background(), remoteAccountID) + if err != nil { + return nil, fmt.Errorf("DereferenceAccountable: error deferencing %s: %s", remoteAccountID.String(), err) + } + + m := make(map[string]interface{}) + if err := json.Unmarshal(b, &m); err != nil { + return nil, fmt.Errorf("DereferenceAccountable: error unmarshalling bytes into json: %s", err) + } + + t, err := streams.ToType(context.Background(), m) + if err != nil { + return nil, fmt.Errorf("DereferenceAccountable: error resolving json into ap vocab type: %s", err) + } + + switch t.GetTypeName() { + case string(gtsmodel.ActivityStreamsPerson): + p, ok := t.(vocab.ActivityStreamsPerson) + if !ok { + return nil, errors.New("DereferenceAccountable: error resolving type as activitystreams person") + } + return p, nil + case string(gtsmodel.ActivityStreamsApplication): + p, ok := t.(vocab.ActivityStreamsApplication) + if !ok { + return nil, errors.New("DereferenceAccountable: error resolving type as activitystreams application") + } + return p, nil + case string(gtsmodel.ActivityStreamsService): + p, ok := t.(vocab.ActivityStreamsService) + if !ok { + return nil, errors.New("DereferenceAccountable: error resolving type as activitystreams service") + } + return p, nil + } + + return nil, fmt.Errorf("DereferenceAccountable: type name %s not supported", t.GetTypeName()) +} + +// populateAccountFields populates any fields on the given account that weren't populated by the initial +// dereferencing. This includes things like header and avatar etc. +func (d *deref) populateAccountFields(account *gtsmodel.Account, requestingUsername string, refresh bool) error { + l := d.log.WithFields(logrus.Fields{ + "func": "PopulateAccountFields", + "requestingUsername": requestingUsername, + }) + + accountURI, err := url.Parse(account.URI) + if err != nil { + return fmt.Errorf("PopulateAccountFields: couldn't parse account URI %s: %s", account.URI, err) + } + if blocked, err := d.blockedDomain(accountURI.Host); blocked || err != nil { + return fmt.Errorf("PopulateAccountFields: domain %s is blocked", accountURI.Host) + } + + t, err := d.transportController.NewTransportForUsername(requestingUsername) + if err != nil { + return fmt.Errorf("PopulateAccountFields: error getting transport for user: %s", err) + } + + // fetch the header and avatar + if err := d.fetchHeaderAndAviForAccount(account, t, refresh); err != nil { + // if this doesn't work, just skip it -- we can do it later + l.Debugf("error fetching header/avi for account: %s", err) + } + + return nil +} + +// fetchHeaderAndAviForAccount fetches the header and avatar for a remote account, using a transport +// on behalf of requestingUsername. +// +// targetAccount's AvatarMediaAttachmentID and HeaderMediaAttachmentID will be updated as necessary. +// +// SIDE EFFECTS: remote header and avatar will be stored in local storage. +func (d *deref) fetchHeaderAndAviForAccount(targetAccount *gtsmodel.Account, t transport.Transport, refresh bool) error { + accountURI, err := url.Parse(targetAccount.URI) + if err != nil { + return fmt.Errorf("fetchHeaderAndAviForAccount: couldn't parse account URI %s: %s", targetAccount.URI, err) + } + if blocked, err := d.blockedDomain(accountURI.Host); blocked || err != nil { + return fmt.Errorf("fetchHeaderAndAviForAccount: domain %s is blocked", accountURI.Host) + } + + if targetAccount.AvatarRemoteURL != "" && (targetAccount.AvatarMediaAttachmentID == "" || refresh) { + a, err := d.mediaHandler.ProcessRemoteHeaderOrAvatar(t, >smodel.MediaAttachment{ + RemoteURL: targetAccount.AvatarRemoteURL, + Avatar: true, + }, targetAccount.ID) + if err != nil { + return fmt.Errorf("error processing avatar for user: %s", err) + } + targetAccount.AvatarMediaAttachmentID = a.ID + } + + if targetAccount.HeaderRemoteURL != "" && (targetAccount.HeaderMediaAttachmentID == "" || refresh) { + a, err := d.mediaHandler.ProcessRemoteHeaderOrAvatar(t, >smodel.MediaAttachment{ + RemoteURL: targetAccount.HeaderRemoteURL, + Header: true, + }, targetAccount.ID) + if err != nil { + return fmt.Errorf("error processing header for user: %s", err) + } + targetAccount.HeaderMediaAttachmentID = a.ID + } + return nil +} diff --git a/internal/federation/dereferencing/announce.go b/internal/federation/dereferencing/announce.go new file mode 100644 index 000000000..2522a4034 --- /dev/null +++ b/internal/federation/dereferencing/announce.go @@ -0,0 +1,65 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import ( + "errors" + "fmt" + "net/url" + + "github.com/superseriousbusiness/gotosocial/internal/gtsmodel" +) + +func (d *deref) DereferenceAnnounce(announce *gtsmodel.Status, requestingUsername string) error { + if announce.GTSBoostedStatus == nil || announce.GTSBoostedStatus.URI == "" { + // we can't do anything unfortunately + return errors.New("DereferenceAnnounce: no URI to dereference") + } + + boostedStatusURI, err := url.Parse(announce.GTSBoostedStatus.URI) + if err != nil { + return fmt.Errorf("DereferenceAnnounce: couldn't parse boosted status URI %s: %s", announce.GTSBoostedStatus.URI, err) + } + if blocked, err := d.blockedDomain(boostedStatusURI.Host); blocked || err != nil { + return fmt.Errorf("DereferenceAnnounce: domain %s is blocked", boostedStatusURI.Host) + } + + // dereference statuses in the thread of the boosted status + if err := d.DereferenceThread(requestingUsername, boostedStatusURI); err != nil { + return fmt.Errorf("DereferenceAnnounce: error dereferencing thread of boosted status: %s", err) + } + + boostedStatus, _, _, err := d.GetRemoteStatus(requestingUsername, boostedStatusURI, false) + if err != nil { + return fmt.Errorf("DereferenceAnnounce: error dereferencing remote status with id %s: %s", announce.GTSBoostedStatus.URI, err) + } + + announce.Content = boostedStatus.Content + announce.ContentWarning = boostedStatus.ContentWarning + announce.ActivityStreamsType = boostedStatus.ActivityStreamsType + announce.Sensitive = boostedStatus.Sensitive + announce.Language = boostedStatus.Language + announce.Text = boostedStatus.Text + announce.BoostOfID = boostedStatus.ID + announce.BoostOfAccountID = boostedStatus.AccountID + announce.Visibility = boostedStatus.Visibility + announce.VisibilityAdvanced = boostedStatus.VisibilityAdvanced + announce.GTSBoostedStatus = boostedStatus + return nil +} diff --git a/internal/federation/dereferencing/blocked.go b/internal/federation/dereferencing/blocked.go new file mode 100644 index 000000000..a66afbb60 --- /dev/null +++ b/internal/federation/dereferencing/blocked.go @@ -0,0 +1,41 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import ( + "github.com/superseriousbusiness/gotosocial/internal/db" + "github.com/superseriousbusiness/gotosocial/internal/gtsmodel" +) + +func (d *deref) blockedDomain(host string) (bool, error) { + b := >smodel.DomainBlock{} + err := d.db.GetWhere([]db.Where{{Key: "domain", Value: host, CaseInsensitive: true}}, b) + if err == nil { + // block exists + return true, nil + } + + if _, ok := err.(db.ErrNoEntries); ok { + // there are no entries so there's no block + return false, nil + } + + // there's an actual error + return false, err +} diff --git a/internal/federation/dereferencing/collectionpage.go b/internal/federation/dereferencing/collectionpage.go new file mode 100644 index 000000000..5feadc1ad --- /dev/null +++ b/internal/federation/dereferencing/collectionpage.go @@ -0,0 +1,70 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import ( + "context" + "encoding/json" + "errors" + "fmt" + "net/url" + + "github.com/go-fed/activity/streams" + "github.com/go-fed/activity/streams/vocab" + "github.com/superseriousbusiness/gotosocial/internal/ap" + "github.com/superseriousbusiness/gotosocial/internal/gtsmodel" +) + +// DereferenceCollectionPage returns the activitystreams CollectionPage at the specified IRI, or an error if something goes wrong. +func (d *deref) DereferenceCollectionPage(username string, pageIRI *url.URL) (ap.CollectionPageable, error) { + if blocked, err := d.blockedDomain(pageIRI.Host); blocked || err != nil { + return nil, fmt.Errorf("DereferenceCollectionPage: domain %s is blocked", pageIRI.Host) + } + + transport, err := d.transportController.NewTransportForUsername(username) + if err != nil { + return nil, fmt.Errorf("DereferenceCollectionPage: error creating transport: %s", err) + } + + b, err := transport.Dereference(context.Background(), pageIRI) + if err != nil { + return nil, fmt.Errorf("DereferenceCollectionPage: error deferencing %s: %s", pageIRI.String(), err) + } + + m := make(map[string]interface{}) + if err := json.Unmarshal(b, &m); err != nil { + return nil, fmt.Errorf("DereferenceCollectionPage: error unmarshalling bytes into json: %s", err) + } + + t, err := streams.ToType(context.Background(), m) + if err != nil { + return nil, fmt.Errorf("DereferenceCollectionPage: error resolving json into ap vocab type: %s", err) + } + + if t.GetTypeName() != gtsmodel.ActivityStreamsCollectionPage { + return nil, fmt.Errorf("DereferenceCollectionPage: type name %s not supported", t.GetTypeName()) + } + + p, ok := t.(vocab.ActivityStreamsCollectionPage) + if !ok { + return nil, errors.New("DereferenceCollectionPage: error resolving type as activitystreams collection page") + } + + return p, nil +} diff --git a/internal/federation/dereferencing/dereferencer.go b/internal/federation/dereferencing/dereferencer.go new file mode 100644 index 000000000..03b90569a --- /dev/null +++ b/internal/federation/dereferencing/dereferencer.go @@ -0,0 +1,73 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import ( + "net/url" + "sync" + + "github.com/sirupsen/logrus" + "github.com/superseriousbusiness/gotosocial/internal/ap" + "github.com/superseriousbusiness/gotosocial/internal/config" + "github.com/superseriousbusiness/gotosocial/internal/db" + "github.com/superseriousbusiness/gotosocial/internal/gtsmodel" + "github.com/superseriousbusiness/gotosocial/internal/media" + "github.com/superseriousbusiness/gotosocial/internal/transport" + "github.com/superseriousbusiness/gotosocial/internal/typeutils" +) + +// Dereferencer wraps logic and functionality for doing dereferencing of remote accounts, statuses, etc, from federated instances. +type Dereferencer interface { + GetRemoteAccount(username string, remoteAccountID *url.URL, refresh bool) (*gtsmodel.Account, bool, error) + EnrichRemoteAccount(username string, account *gtsmodel.Account) (*gtsmodel.Account, error) + + GetRemoteStatus(username string, remoteStatusID *url.URL, refresh bool) (*gtsmodel.Status, ap.Statusable, bool, error) + EnrichRemoteStatus(username string, status *gtsmodel.Status) (*gtsmodel.Status, error) + + GetRemoteInstance(username string, remoteInstanceURI *url.URL) (*gtsmodel.Instance, error) + + DereferenceAnnounce(announce *gtsmodel.Status, requestingUsername string) error + DereferenceThread(username string, statusIRI *url.URL) error + + Handshaking(username string, remoteAccountID *url.URL) bool +} + +type deref struct { + log *logrus.Logger + db db.DB + typeConverter typeutils.TypeConverter + transportController transport.Controller + mediaHandler media.Handler + config *config.Config + handshakes map[string][]*url.URL + handshakeSync *sync.Mutex // mutex to lock/unlock when checking or updating the handshakes map +} + +// NewDereferencer returns a Dereferencer initialized with the given parameters. +func NewDereferencer(config *config.Config, db db.DB, typeConverter typeutils.TypeConverter, transportController transport.Controller, mediaHandler media.Handler, log *logrus.Logger) Dereferencer { + return &deref{ + log: log, + db: db, + typeConverter: typeConverter, + transportController: transportController, + mediaHandler: mediaHandler, + config: config, + handshakeSync: &sync.Mutex{}, + } +} diff --git a/internal/federation/dereferencing/handshake.go b/internal/federation/dereferencing/handshake.go new file mode 100644 index 000000000..cda8eafd0 --- /dev/null +++ b/internal/federation/dereferencing/handshake.go @@ -0,0 +1,98 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import "net/url" + +func (d *deref) Handshaking(username string, remoteAccountID *url.URL) bool { + d.handshakeSync.Lock() + defer d.handshakeSync.Unlock() + + if d.handshakes == nil { + // handshakes isn't even initialized yet so we can't be handshaking with anyone + return false + } + + remoteIDs, ok := d.handshakes[username] + if !ok { + // user isn't handshaking with anyone, bail + return false + } + + for _, id := range remoteIDs { + if id.String() == remoteAccountID.String() { + // we are currently handshaking with the remote account, yep + return true + } + } + + // didn't find it which means we're not handshaking + return false +} + +func (d *deref) startHandshake(username string, remoteAccountID *url.URL) { + d.handshakeSync.Lock() + defer d.handshakeSync.Unlock() + + // lazily initialize handshakes + if d.handshakes == nil { + d.handshakes = make(map[string][]*url.URL) + } + + remoteIDs, ok := d.handshakes[username] + if !ok { + // there was nothing in there yet, so just add this entry and return + d.handshakes[username] = []*url.URL{remoteAccountID} + return + } + + // add the remote ID to the slice + remoteIDs = append(remoteIDs, remoteAccountID) + d.handshakes[username] = remoteIDs +} + +func (d *deref) stopHandshake(username string, remoteAccountID *url.URL) { + d.handshakeSync.Lock() + defer d.handshakeSync.Unlock() + + if d.handshakes == nil { + return + } + + remoteIDs, ok := d.handshakes[username] + if !ok { + // there was nothing in there yet anyway so just bail + return + } + + newRemoteIDs := []*url.URL{} + for _, id := range remoteIDs { + if id.String() != remoteAccountID.String() { + newRemoteIDs = append(newRemoteIDs, id) + } + } + + if len(newRemoteIDs) == 0 { + // there are no handshakes so just remove this user entry from the map and save a few bytes + delete(d.handshakes, username) + } else { + // there are still other handshakes ongoing + d.handshakes[username] = newRemoteIDs + } +} diff --git a/internal/federation/dereferencing/instance.go b/internal/federation/dereferencing/instance.go new file mode 100644 index 000000000..80f626662 --- /dev/null +++ b/internal/federation/dereferencing/instance.go @@ -0,0 +1,40 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import ( + "context" + "fmt" + "net/url" + + "github.com/superseriousbusiness/gotosocial/internal/gtsmodel" +) + +func (d *deref) GetRemoteInstance(username string, remoteInstanceURI *url.URL) (*gtsmodel.Instance, error) { + if blocked, err := d.blockedDomain(remoteInstanceURI.Host); blocked || err != nil { + return nil, fmt.Errorf("GetRemoteInstance: domain %s is blocked", remoteInstanceURI.Host) + } + + transport, err := d.transportController.NewTransportForUsername(username) + if err != nil { + return nil, fmt.Errorf("transport err: %s", err) + } + + return transport.DereferenceInstance(context.Background(), remoteInstanceURI) +} diff --git a/internal/federation/dereferencing/status.go b/internal/federation/dereferencing/status.go new file mode 100644 index 000000000..b05f6e72c --- /dev/null +++ b/internal/federation/dereferencing/status.go @@ -0,0 +1,369 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import ( + "context" + "encoding/json" + "errors" + "fmt" + "net/url" + + "github.com/go-fed/activity/streams" + "github.com/go-fed/activity/streams/vocab" + "github.com/sirupsen/logrus" + "github.com/superseriousbusiness/gotosocial/internal/ap" + "github.com/superseriousbusiness/gotosocial/internal/db" + "github.com/superseriousbusiness/gotosocial/internal/gtsmodel" + "github.com/superseriousbusiness/gotosocial/internal/id" +) + +// EnrichRemoteStatus takes a status that's already been inserted into the database in a minimal form, +// and populates it with additional fields, media, etc. +// +// EnrichRemoteStatus is mostly useful for calling after a status has been initially created by +// the federatingDB's Create function, but additional dereferencing is needed on it. +func (d *deref) EnrichRemoteStatus(username string, status *gtsmodel.Status) (*gtsmodel.Status, error) { + if err := d.populateStatusFields(status, username); err != nil { + return nil, err + } + + if err := d.db.UpdateByID(status.ID, status); err != nil { + return nil, fmt.Errorf("EnrichRemoteStatus: error updating status: %s", err) + } + + return status, nil +} + +// GetRemoteStatus completely dereferences a remote status, converts it to a GtS model status, +// puts it in the database, and returns it to a caller. The boolean indicates whether the status is new +// to us or not. If we haven't seen the status before, bool will be true. If we have seen the status before, +// it will be false. +// +// If refresh is true, then even if we have the status in our database already, it will be dereferenced from its +// remote representation, as will its owner. +// +// If a dereference was performed, then the function also returns the ap.Statusable representation for further processing. +// +// SIDE EFFECTS: remote status will be stored in the database, and the remote status owner will also be stored. +func (d *deref) GetRemoteStatus(username string, remoteStatusID *url.URL, refresh bool) (*gtsmodel.Status, ap.Statusable, bool, error) { + new := true + + // check if we already have the status in our db + maybeStatus := >smodel.Status{} + if err := d.db.GetWhere([]db.Where{{Key: "uri", Value: remoteStatusID.String()}}, maybeStatus); err == nil { + // we've seen this status before so it's not new + new = false + + // if we're not being asked to refresh, we can just return the maybeStatus as-is and avoid doing any external calls + if !refresh { + return maybeStatus, nil, new, nil + } + } + + statusable, err := d.dereferenceStatusable(username, remoteStatusID) + if err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: error dereferencing statusable: %s", err) + } + + accountURI, err := ap.ExtractAttributedTo(statusable) + if err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: error extracting attributedTo: %s", err) + } + + // do this so we know we have the remote account of the status in the db + _, _, err = d.GetRemoteAccount(username, accountURI, false) + if err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: couldn't derive status author: %s", err) + } + + gtsStatus, err := d.typeConverter.ASStatusToStatus(statusable) + if err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: error converting statusable to status: %s", err) + } + + if new { + ulid, err := id.NewULIDFromTime(gtsStatus.CreatedAt) + if err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: error generating new id for status: %s", err) + } + gtsStatus.ID = ulid + + if err := d.populateStatusFields(gtsStatus, username); err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: error populating status fields: %s", err) + } + + if err := d.db.Put(gtsStatus); err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: error putting new status: %s", err) + } + } else { + gtsStatus.ID = maybeStatus.ID + + if err := d.populateStatusFields(gtsStatus, username); err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: error populating status fields: %s", err) + } + + if err := d.db.UpdateByID(gtsStatus.ID, gtsStatus); err != nil { + return nil, statusable, new, fmt.Errorf("GetRemoteStatus: error updating status: %s", err) + } + } + + return gtsStatus, statusable, new, nil +} + +func (d *deref) dereferenceStatusable(username string, remoteStatusID *url.URL) (ap.Statusable, error) { + if blocked, err := d.blockedDomain(remoteStatusID.Host); blocked || err != nil { + return nil, fmt.Errorf("DereferenceStatusable: domain %s is blocked", remoteStatusID.Host) + } + + transport, err := d.transportController.NewTransportForUsername(username) + if err != nil { + return nil, fmt.Errorf("DereferenceStatusable: transport err: %s", err) + } + + b, err := transport.Dereference(context.Background(), remoteStatusID) + if err != nil { + return nil, fmt.Errorf("DereferenceStatusable: error deferencing %s: %s", remoteStatusID.String(), err) + } + + m := make(map[string]interface{}) + if err := json.Unmarshal(b, &m); err != nil { + return nil, fmt.Errorf("DereferenceStatusable: error unmarshalling bytes into json: %s", err) + } + + t, err := streams.ToType(context.Background(), m) + if err != nil { + return nil, fmt.Errorf("DereferenceStatusable: error resolving json into ap vocab type: %s", err) + } + + // Article, Document, Image, Video, Note, Page, Event, Place, Mention, Profile + switch t.GetTypeName() { + case gtsmodel.ActivityStreamsArticle: + p, ok := t.(vocab.ActivityStreamsArticle) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsArticle") + } + return p, nil + case gtsmodel.ActivityStreamsDocument: + p, ok := t.(vocab.ActivityStreamsDocument) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsDocument") + } + return p, nil + case gtsmodel.ActivityStreamsImage: + p, ok := t.(vocab.ActivityStreamsImage) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsImage") + } + return p, nil + case gtsmodel.ActivityStreamsVideo: + p, ok := t.(vocab.ActivityStreamsVideo) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsVideo") + } + return p, nil + case gtsmodel.ActivityStreamsNote: + p, ok := t.(vocab.ActivityStreamsNote) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsNote") + } + return p, nil + case gtsmodel.ActivityStreamsPage: + p, ok := t.(vocab.ActivityStreamsPage) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsPage") + } + return p, nil + case gtsmodel.ActivityStreamsEvent: + p, ok := t.(vocab.ActivityStreamsEvent) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsEvent") + } + return p, nil + case gtsmodel.ActivityStreamsPlace: + p, ok := t.(vocab.ActivityStreamsPlace) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsPlace") + } + return p, nil + case gtsmodel.ActivityStreamsProfile: + p, ok := t.(vocab.ActivityStreamsProfile) + if !ok { + return nil, errors.New("DereferenceStatusable: error resolving type as ActivityStreamsProfile") + } + return p, nil + } + + return nil, fmt.Errorf("DereferenceStatusable: type name %s not supported", t.GetTypeName()) +} + +// populateStatusFields fetches all the information we temporarily pinned to an incoming +// federated status, back in the federating db's Create function. +// +// When a status comes in from the federation API, there are certain fields that +// haven't been dereferenced yet, because we needed to provide a snappy synchronous +// response to the caller. By the time it reaches this function though, it's being +// processed asynchronously, so we have all the time in the world to fetch the various +// bits and bobs that are attached to the status, and properly flesh it out, before we +// send the status to any timelines and notify people. +// +// Things to dereference and fetch here: +// +// 1. Media attachments. +// 2. Hashtags. +// 3. Emojis. +// 4. Mentions. +// 5. Posting account. +// 6. Replied-to-status. +// +// SIDE EFFECTS: +// This function will deference all of the above, insert them in the database as necessary, +// and attach them to the status. The status itself will not be added to the database yet, +// that's up the caller to do. +func (d *deref) populateStatusFields(status *gtsmodel.Status, requestingUsername string) error { + l := d.log.WithFields(logrus.Fields{ + "func": "dereferenceStatusFields", + "status": fmt.Sprintf("%+v", status), + }) + l.Debug("entering function") + + // make sure we have a status URI and that the domain in question isn't blocked + statusURI, err := url.Parse(status.URI) + if err != nil { + return fmt.Errorf("DereferenceStatusFields: couldn't parse status URI %s: %s", status.URI, err) + } + if blocked, err := d.blockedDomain(statusURI.Host); blocked || err != nil { + return fmt.Errorf("DereferenceStatusFields: domain %s is blocked", statusURI.Host) + } + + // we can continue -- create a new transport here because we'll probably need it + t, err := d.transportController.NewTransportForUsername(requestingUsername) + if err != nil { + return fmt.Errorf("error creating transport: %s", err) + } + + // in case the status doesn't have an id yet (ie., it hasn't entered the database yet), then create one + if status.ID == "" { + newID, err := id.NewULIDFromTime(status.CreatedAt) + if err != nil { + return err + } + status.ID = newID + } + + // 1. Media attachments. + // + // At this point we should know: + // * the media type of the file we're looking for (a.File.ContentType) + // * the blurhash (a.Blurhash) + // * the file type (a.Type) + // * the remote URL (a.RemoteURL) + // This should be enough to pass along to the media processor. + attachmentIDs := []string{} + for _, a := range status.GTSMediaAttachments { + l.Tracef("dereferencing attachment: %+v", a) + + // it might have been processed elsewhere so check first if it's already in the database or not + maybeAttachment := >smodel.MediaAttachment{} + err := d.db.GetWhere([]db.Where{{Key: "remote_url", Value: a.RemoteURL}}, maybeAttachment) + if err == nil { + // we already have it in the db, dereferenced, no need to do it again + l.Tracef("attachment already exists with id %s", maybeAttachment.ID) + attachmentIDs = append(attachmentIDs, maybeAttachment.ID) + continue + } + if _, ok := err.(db.ErrNoEntries); !ok { + // we have a real error + return fmt.Errorf("error checking db for existence of attachment with remote url %s: %s", a.RemoteURL, err) + } + // it just doesn't exist yet so carry on + l.Debug("attachment doesn't exist yet, calling ProcessRemoteAttachment", a) + deferencedAttachment, err := d.mediaHandler.ProcessRemoteAttachment(t, a, status.AccountID) + if err != nil { + l.Errorf("error dereferencing status attachment: %s", err) + continue + } + l.Debugf("dereferenced attachment: %+v", deferencedAttachment) + deferencedAttachment.StatusID = status.ID + deferencedAttachment.Description = a.Description + if err := d.db.Put(deferencedAttachment); err != nil { + return fmt.Errorf("error inserting dereferenced attachment with remote url %s: %s", a.RemoteURL, err) + } + attachmentIDs = append(attachmentIDs, deferencedAttachment.ID) + } + status.Attachments = attachmentIDs + + // 2. Hashtags + + // 3. Emojis + + // 4. Mentions + // At this point, mentions should have the namestring and mentionedAccountURI set on them. + // + // We should dereference any accounts mentioned here which we don't have in our db yet, by their URI. + mentions := []string{} + for _, m := range status.GTSMentions { + + if m.ID != "" { + continue + // we've already populated this mention, since it has an ID + } + + mID, err := id.NewRandomULID() + if err != nil { + return err + } + m.ID = mID + + uri, err := url.Parse(m.MentionedAccountURI) + if err != nil { + l.Debugf("error parsing mentioned account uri %s: %s", m.MentionedAccountURI, err) + continue + } + + m.StatusID = status.ID + m.OriginAccountID = status.GTSAuthorAccount.ID + m.OriginAccountURI = status.GTSAuthorAccount.URI + + targetAccount, _, err := d.GetRemoteAccount(requestingUsername, uri, false) + if err != nil { + continue + } + + // by this point, we know the targetAccount exists in our database with an ID :) + m.TargetAccountID = targetAccount.ID + if err := d.db.Put(m); err != nil { + return fmt.Errorf("error creating mention: %s", err) + } + mentions = append(mentions, m.ID) + } + status.Mentions = mentions + + // status has replyToURI but we don't have an ID yet for the status it replies to + if status.InReplyToURI != "" && status.InReplyToID == "" { + replyToStatus := >smodel.Status{} + if err := d.db.GetWhere([]db.Where{{Key: "uri", Value: status.InReplyToURI}}, replyToStatus); err == nil { + // we have the status + status.InReplyToID = replyToStatus.ID + status.InReplyToAccountID = replyToStatus.AccountID + } + } + + return nil +} diff --git a/internal/federation/dereferencing/thread.go b/internal/federation/dereferencing/thread.go new file mode 100644 index 000000000..2a407f923 --- /dev/null +++ b/internal/federation/dereferencing/thread.go @@ -0,0 +1,250 @@ +/* + GoToSocial + Copyright (C) 2021 GoToSocial Authors admin@gotosocial.org + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Affero General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Affero General Public License for more details. + + You should have received a copy of the GNU Affero General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. +*/ + +package dereferencing + +import ( + "fmt" + "net/url" + + "github.com/sirupsen/logrus" + "github.com/superseriousbusiness/gotosocial/internal/ap" + "github.com/superseriousbusiness/gotosocial/internal/gtsmodel" + "github.com/superseriousbusiness/gotosocial/internal/util" +) + +// DereferenceThread takes a statusable (something that has withReplies and withInReplyTo), +// and dereferences statusables in the conversation. +// +// This process involves working up and down the chain of replies, and parsing through the collections of IDs +// presented by remote instances as part of their replies collections, and will likely involve making several calls to +// multiple different hosts. +func (d *deref) DereferenceThread(username string, statusIRI *url.URL) error { + l := d.log.WithFields(logrus.Fields{ + "func": "DereferenceThread", + "username": username, + "statusIRI": statusIRI.String(), + }) + l.Debug("entering DereferenceThread") + + // if it's our status we already have everything stashed so we can bail early + if statusIRI.Host == d.config.Host { + l.Debug("iri belongs to us, bailing") + return nil + } + + // first make sure we have this status in our db + _, statusable, _, err := d.GetRemoteStatus(username, statusIRI, true) + if err != nil { + return fmt.Errorf("DereferenceThread: error getting status with id %s: %s", statusIRI.String(), err) + } + + // first iterate up through ancestors, dereferencing if necessary as we go + if err := d.iterateAncestors(username, *statusIRI); err != nil { + return fmt.Errorf("error iterating ancestors of status %s: %s", statusIRI.String(), err) + } + + // now iterate down through descendants, again dereferencing as we go + if err := d.iterateDescendants(username, *statusIRI, statusable); err != nil { + return fmt.Errorf("error iterating descendants of status %s: %s", statusIRI.String(), err) + } + + return nil +} + +// iterateAncestors has the goal of reaching the oldest ancestor of a given status, and stashing all statuses along the way. +func (d *deref) iterateAncestors(username string, statusIRI url.URL) error { + l := d.log.WithFields(logrus.Fields{ + "func": "iterateAncestors", + "username": username, + "statusIRI": statusIRI.String(), + }) + l.Debug("entering iterateAncestors") + + // if it's our status we don't need to dereference anything so we can immediately move up the chain + if statusIRI.Host == d.config.Host { + l.Debug("iri belongs to us, moving up to next ancestor") + + // since this is our status, we know we can extract the id from the status path + _, id, err := util.ParseStatusesPath(&statusIRI) + if err != nil { + return err + } + + status := >smodel.Status{} + if err := d.db.GetByID(id, status); err != nil { + return err + } + + if status.InReplyToURI == "" { + // status doesn't reply to anything + return nil + } + nextIRI, err := url.Parse(status.URI) + if err != nil { + return err + } + return d.iterateAncestors(username, *nextIRI) + } + + // If we reach here, we're looking at a remote status -- make sure we have it in our db by calling GetRemoteStatus + // We call it with refresh to true because we want the statusable representation to parse inReplyTo from. + status, statusable, _, err := d.GetRemoteStatus(username, &statusIRI, true) + if err != nil { + l.Debugf("error getting remote status: %s", err) + return nil + } + + inReplyTo := ap.ExtractInReplyToURI(statusable) + if inReplyTo == nil || inReplyTo.String() == "" { + // status doesn't reply to anything + return nil + } + + // get the ancestor status into our database if we don't have it yet + if _, _, _, err := d.GetRemoteStatus(username, inReplyTo, false); err != nil { + l.Debugf("error getting remote status: %s", err) + return nil + } + + // now enrich the current status, since we should have the ancestor in the db + if _, err := d.EnrichRemoteStatus(username, status); err != nil { + l.Debugf("error enriching remote status: %s", err) + return nil + } + + // now move up to the next ancestor + return d.iterateAncestors(username, *inReplyTo) +} + +func (d *deref) iterateDescendants(username string, statusIRI url.URL, statusable ap.Statusable) error { + l := d.log.WithFields(logrus.Fields{ + "func": "iterateDescendants", + "username": username, + "statusIRI": statusIRI.String(), + }) + l.Debug("entering iterateDescendants") + + // if it's our status we already have descendants stashed so we can bail early + if statusIRI.Host == d.config.Host { + l.Debug("iri belongs to us, bailing") + return nil + } + + replies := statusable.GetActivityStreamsReplies() + if replies == nil || !replies.IsActivityStreamsCollection() { + l.Debug("no replies, bailing") + return nil + } + + repliesCollection := replies.GetActivityStreamsCollection() + if repliesCollection == nil { + l.Debug("replies collection is nil, bailing") + return nil + } + + first := repliesCollection.GetActivityStreamsFirst() + if first == nil { + l.Debug("replies collection has no first, bailing") + return nil + } + + firstPage := first.GetActivityStreamsCollectionPage() + if firstPage == nil { + l.Debug("first has no collection page, bailing") + return nil + } + + firstPageNext := firstPage.GetActivityStreamsNext() + if firstPageNext == nil || !firstPageNext.IsIRI() { + l.Debug("next is not an iri, bailing") + return nil + } + + var foundReplies int + currentPageIRI := firstPageNext.GetIRI() + +pageLoop: + for { + l.Debugf("dereferencing page %s", currentPageIRI) + nextPage, err := d.DereferenceCollectionPage(username, currentPageIRI) + if err != nil { + return nil + } + + // next items could be either a list of URLs or a list of statuses + + nextItems := nextPage.GetActivityStreamsItems() + if nextItems.Len() == 0 { + // no items on this page, which means we're done + break pageLoop + } + + // have a look through items and see what we can find + for iter := nextItems.Begin(); iter != nextItems.End(); iter = iter.Next() { + // We're looking for a url to feed to GetRemoteStatus. + // Items can be either an IRI, or a Note. + // If a note, we grab the ID from it and call it, rather than parsing the note. + + var itemURI *url.URL + if iter.IsIRI() { + // iri, easy + itemURI = iter.GetIRI() + } else if iter.IsActivityStreamsNote() { + // note, get the id from it to use as iri + n := iter.GetActivityStreamsNote() + id := n.GetJSONLDId() + if id != nil && id.IsIRI() { + itemURI = id.GetIRI() + } + } else { + // if it's not an iri or a note, we don't know how to process it + continue + } + + if itemURI.Host == d.config.Host { + // skip if the reply is from us -- we already have it then + continue + } + + // we can confidently say now that we found something + foundReplies = foundReplies + 1 + + // get the remote statusable and put it in the db + _, statusable, new, err := d.GetRemoteStatus(username, itemURI, false) + if new && err == nil && statusable != nil { + // now iterate descendants of *that* status + if err := d.iterateDescendants(username, *itemURI, statusable); err != nil { + continue + } + } + } + + next := nextPage.GetActivityStreamsNext() + if next != nil && next.IsIRI() { + l.Debug("setting next page") + currentPageIRI = next.GetIRI() + } else { + l.Debug("no next page, bailing") + break pageLoop + } + } + + l.Debugf("foundReplies %d", foundReplies) + return nil +} |