dendrite/appservice/appservice.go

131 lines
4.7 KiB
Go
Raw Normal View History

// Copyright 2018 Vector Creations Ltd
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package appservice
import (
"context"
"errors"
2018-07-17 09:36:04 -05:00
"net/http"
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 11:34:59 -05:00
"sync"
2018-07-17 09:36:04 -05:00
"time"
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 11:34:59 -05:00
"github.com/gorilla/mux"
2018-07-17 09:36:04 -05:00
appserviceAPI "github.com/matrix-org/dendrite/appservice/api"
"github.com/matrix-org/dendrite/appservice/consumers"
"github.com/matrix-org/dendrite/appservice/inthttp"
2018-07-17 09:36:04 -05:00
"github.com/matrix-org/dendrite/appservice/query"
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 11:34:59 -05:00
"github.com/matrix-org/dendrite/appservice/storage"
"github.com/matrix-org/dendrite/appservice/types"
"github.com/matrix-org/dendrite/appservice/workers"
"github.com/matrix-org/dendrite/clientapi/auth/storage/accounts"
"github.com/matrix-org/dendrite/clientapi/auth/storage/devices"
"github.com/matrix-org/dendrite/internal/config"
"github.com/matrix-org/dendrite/internal/setup"
"github.com/matrix-org/dendrite/internal/sqlutil"
2018-07-17 09:36:04 -05:00
roomserverAPI "github.com/matrix-org/dendrite/roomserver/api"
"github.com/sirupsen/logrus"
)
// AddInternalRoutes registers HTTP handlers for internal API calls
func AddInternalRoutes(router *mux.Router, queryAPI appserviceAPI.AppServiceQueryAPI) {
inthttp.AddRoutes(queryAPI, router)
}
// NewInternalAPI returns a concerete implementation of the internal API. Callers
// can call functions directly on the returned API or via an HTTP interface using AddInternalRoutes.
func NewInternalAPI(
base *setup.BaseDendrite,
Support sqlite in addition to postgres (#869) * Move current work into single branch * Initial massaging of clientapi etc (not working yet) * Interfaces for accounts/devices databases * Duplicate postgres package for sqlite3 (no changes made to it yet) * Some keydb, accountdb, devicedb, common partition fixes, some more syncapi tweaking * Fix accounts DB, device DB * Update naffka dependency for SQLite * Naffka SQLite * Update naffka to latest master * SQLite support for federationsender * Mostly not-bad support for SQLite in syncapi (although there are problems where lots of events get classed incorrectly as backward extremities, probably because of IN/ANY clauses that are badly supported) * Update Dockerfile -> Go 1.13.7, add build-base (as gcc and friends are needed for SQLite) * Implement GET endpoints for account_data in clientapi * Nuke filtering for now... * Revert "Implement GET endpoints for account_data in clientapi" This reverts commit 4d80dff4583d278620d9b3ed437e9fcd8d4674ee. * Implement GET endpoints for account_data in clientapi (#861) * Implement GET endpoints for account_data in clientapi * Fix accountDB parameter * Remove fmt.Println * Fix insertAccountData SQLite query * Fix accountDB storage interfaces * Add empty push rules into account data on account creation (#862) * Put SaveAccountData into the right function this time * Not sure if roomserver is better or worse now * sqlite work * Allow empty last sent ID for the first event * sqlite: room creation works * Support sending messages * Nuke fmt.println * Move QueryVariadic etc into common, other device fixes * Fix some linter issues * Fix bugs * Fix some linting errors * Fix errcheck lint errors * Make naffka use postgres as fallback, fix couple of compile errors * What on earth happened to the /rooms/{roomID}/send/{eventType} routing Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2020-02-13 11:27:33 -06:00
accountsDB accounts.Database,
deviceDB devices.Database,
rsAPI roomserverAPI.RoomserverInternalAPI,
2018-07-17 09:36:04 -05:00
) appserviceAPI.AppServiceQueryAPI {
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 11:34:59 -05:00
// Create a connection to the appservice postgres DB
appserviceDB, err := storage.NewDatabase(string(base.Cfg.Database.AppService), base.Cfg.DbProperties())
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 11:34:59 -05:00
if err != nil {
logrus.WithError(err).Panicf("failed to connect to appservice db")
}
// Wrap application services in a type that relates the application service and
// a sync.Cond object that can be used to notify workers when there are new
// events to be sent out.
workerStates := make([]types.ApplicationServiceWorkerState, len(base.Cfg.Derived.ApplicationServices))
for i, appservice := range base.Cfg.Derived.ApplicationServices {
m := sync.Mutex{}
ws := types.ApplicationServiceWorkerState{
AppService: appservice,
Cond: sync.NewCond(&m),
}
workerStates[i] = ws
// Create bot account for this AS if it doesn't already exist
if err = generateAppServiceAccount(accountsDB, deviceDB, appservice); err != nil {
logrus.WithFields(logrus.Fields{
"appservice": appservice.ID,
}).WithError(err).Panicf("failed to generate bot account for appservice")
}
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 11:34:59 -05:00
}
// Create appserivce query API with an HTTP client that will be used for all
// outbound and inbound requests (inbound only for the internal API)
appserviceQueryAPI := &query.AppServiceQueryAPI{
HTTPClient: &http.Client{
Timeout: time.Second * 30,
},
Cfg: base.Cfg,
2018-07-17 09:36:04 -05:00
}
// Only consume if we actually have ASes to track, else we'll just chew cycles needlessly.
// We can't add ASes at runtime so this is safe to do.
if len(workerStates) > 0 {
consumer := consumers.NewOutputRoomEventConsumer(
base.Cfg, base.KafkaConsumer, accountsDB, appserviceDB,
rsAPI, workerStates,
)
if err := consumer.Start(); err != nil {
logrus.WithError(err).Panicf("failed to start appservice roomserver consumer")
}
}
Send Application Service Events (#477) * Prevent sql scanning into nil value in accounts_table Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Remove uneccessary logging, null checking * Don't forget to set the localpart * Simplify error checking * Store And Send Application Service Events * Modify INSTALL.md and dendrite-config.yaml for the new appservice database * Correct all instances of casing on 'application service' to align with spec * Store incoming events that an app service is interested in in the database to be later read by transaction workers. * Retrieve these events from transaction workers, one per AS. * Minimal transaction ID data is stored as well to recover after server failure. * Send events to AS and exponentially backoff on failure. Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Finish my own sentences. * Fix up database interaction * Change to event-based AS sending * Reduce cyclomatic complexity * Appease the errcheck gods * Delete by int ID instead of string. This was causing some events to not be deleted, as < an eventID doesn't really make much sense. * Check if there are more events to send before sleeping * Send same transaction if last send attempt failed * Don't backoff on non-200s, tight send loop, 1 event query * Remove tight send loop. Fix events not being deleted * Additionally order by event id, track main.go * Return the last txnID, which our events are using * Remove old main.go file * Prevent duplicate events from being sent... * Strip event content if it doesn't contain anything Signed-off-by: Andrew Morgan <andrewm@matrix.org> * Update gomatrixserverlib and use Unsigned AS event prop * Fixes * Fix sync server comment * Remove unnecessary printlns * Use logrus Fields * Worker state methods * Remove sillyness * Fix up event filtering * Handle transaction event limit in loop * Switch to using a sequence for transaction IDs * Don't verify self-signed AS certificates * Fix logging * Use gmsl.Event instead of AS-only event in transactions Also clear up the logic on lookupStateEvents a little bit. * Change invalid_txn_id to global (for efficiency) * Use a bool for EventsReady instead of an int
2018-07-05 11:34:59 -05:00
// Create application service transaction workers
if err := workers.SetupTransactionWorkers(appserviceDB, workerStates); err != nil {
logrus.WithError(err).Panicf("failed to start app service transaction workers")
}
return appserviceQueryAPI
}
// generateAppServiceAccounts creates a dummy account based off the
// `sender_localpart` field of each application service if it doesn't
// exist already
func generateAppServiceAccount(
Support sqlite in addition to postgres (#869) * Move current work into single branch * Initial massaging of clientapi etc (not working yet) * Interfaces for accounts/devices databases * Duplicate postgres package for sqlite3 (no changes made to it yet) * Some keydb, accountdb, devicedb, common partition fixes, some more syncapi tweaking * Fix accounts DB, device DB * Update naffka dependency for SQLite * Naffka SQLite * Update naffka to latest master * SQLite support for federationsender * Mostly not-bad support for SQLite in syncapi (although there are problems where lots of events get classed incorrectly as backward extremities, probably because of IN/ANY clauses that are badly supported) * Update Dockerfile -> Go 1.13.7, add build-base (as gcc and friends are needed for SQLite) * Implement GET endpoints for account_data in clientapi * Nuke filtering for now... * Revert "Implement GET endpoints for account_data in clientapi" This reverts commit 4d80dff4583d278620d9b3ed437e9fcd8d4674ee. * Implement GET endpoints for account_data in clientapi (#861) * Implement GET endpoints for account_data in clientapi * Fix accountDB parameter * Remove fmt.Println * Fix insertAccountData SQLite query * Fix accountDB storage interfaces * Add empty push rules into account data on account creation (#862) * Put SaveAccountData into the right function this time * Not sure if roomserver is better or worse now * sqlite work * Allow empty last sent ID for the first event * sqlite: room creation works * Support sending messages * Nuke fmt.println * Move QueryVariadic etc into common, other device fixes * Fix some linter issues * Fix bugs * Fix some linting errors * Fix errcheck lint errors * Make naffka use postgres as fallback, fix couple of compile errors * What on earth happened to the /rooms/{roomID}/send/{eventType} routing Co-authored-by: Neil Alexander <neilalexander@users.noreply.github.com>
2020-02-13 11:27:33 -06:00
accountsDB accounts.Database,
deviceDB devices.Database,
as config.ApplicationService,
) error {
ctx := context.Background()
// Create an account for the application service
_, err := accountsDB.CreateAccount(ctx, as.SenderLocalpart, "", as.ID)
if err != nil {
if errors.Is(err, sqlutil.ErrUserExists) { // This account already exists
return nil
}
return err
}
// Create a dummy device with a dummy token for the application service
_, err = deviceDB.CreateDevice(ctx, as.SenderLocalpart, nil, as.ASToken, &as.SenderLocalpart)
return err
}