Split index.ts into multiple scripts in scripts/ and detect last known block when pulling events

3 files changed, 172 insertions, 0 deletions

diff --git a/packages/pipeline/src/scripts/merge_v2_events.ts b/packages/pipeline/src/scripts/merge_v2_events.ts
new file mode 100644
index 000000000..227ece121
--- /dev/null
+++ b/packages/pipeline/src/scripts/merge_v2_events.ts
@@ -0,0 +1,81 @@
+import { web3Factory } from '@0x/dev-utils';
+import 'reflect-metadata';
+import { Connection, createConnection } from 'typeorm';
+
+import { ExchangeEventsSource } from '../data_sources/contract-wrappers/exchange_events';
+import { ExchangeFillEvent } from '../entities/ExchangeFillEvent';
+import { deployConfig } from '../ormconfig';
+import { parseExchangeEvents } from '../parsers/events';
+
+let connection: Connection;
+
+(async () => {
+    connection = await createConnection(deployConfig);
+    await getExchangeEventsAsync();
+    await mergeExchangeEventsAsync();
+    console.log('Exiting process');
+    process.exit(0);
+})();
+
+// TODO(albrow): Separately: Errors do not appear to be handled correctly. If you use the
+// wrong rpcUrl it just returns early with no error.
+async function getExchangeEventsAsync(): Promise<void> {
+    console.log('Getting event logs...');
+    const provider = web3Factory.getRpcProvider({
+        rpcUrl: 'https://mainnet.infura.io',
+    });
+    const eventsRepository = connection.getRepository(ExchangeFillEvent);
+    const exchangeEvents = new ExchangeEventsSource(provider, 1);
+    const eventLogs = await exchangeEvents.getFillEventsAsync();
+    console.log('Parsing events...');
+    const events = parseExchangeEvents(eventLogs);
+    console.log(`Retrieved and parsed ${events.length} total events.`);
+    console.log('Saving events...');
+    for (const event of events) {
+        await eventsRepository.save(event);
+    }
+    await eventsRepository.save(events);
+    console.log('Saved events.');
+}
+
+const insertEventsRawQuery = `INSERT INTO events_raw (
+    event_type,
+    error_id,
+    order_hash,
+    maker,
+    maker_amount,
+    maker_fee,
+    maker_token,
+    taker,
+    taker_amount,
+    taker_fee,
+    taker_token,
+    txn_hash,
+    fee_recipient,
+    block_number,
+    log_index
+)
+(
+    SELECT
+        'LogFill',
+        null,
+        "orderHash",
+        "makerAddress",
+        "makerAssetFilledAmount"::numeric(78),
+        "makerFeePaid"::numeric(78),
+        "makerTokenAddress",
+        "takerAddress",
+        "takerAssetFilledAmount"::numeric(78),
+        "takerFeePaid"::numeric(78),
+        "takerTokenAddress",
+        "transactionHash",
+        "feeRecipientAddress",
+        "blockNumber",
+        "logIndex"
+    FROM exchange_fill_event
+) ON CONFLICT (order_hash, txn_hash, log_index) DO NOTHING`;
+
+async function mergeExchangeEventsAsync(): Promise<void> {
+    console.log('Merging results into events_raw...');
+    await connection.query(insertEventsRawQuery);
+}
diff --git a/packages/pipeline/src/scripts/pull_missing_events.ts b/packages/pipeline/src/scripts/pull_missing_events.ts
new file mode 100644
index 000000000..1f71722a3
--- /dev/null
+++ b/packages/pipeline/src/scripts/pull_missing_events.ts
@@ -0,0 +1,60 @@
+import { web3Factory } from '@0x/dev-utils';
+import { Web3ProviderEngine } from '@0x/subproviders';
+import R = require('ramda');
+import 'reflect-metadata';
+import { Connection, createConnection, Repository } from 'typeorm';
+
+import { ExchangeEventsSource } from '../data_sources/contract-wrappers/exchange_events';
+import { ExchangeFillEvent } from '../entities/ExchangeFillEvent';
+import { deployConfig } from '../ormconfig';
+import { parseExchangeEvents } from '../parsers/events';
+
+const EXCHANGE_START_BLOCK = 6271590; // Block number when the Exchange contract was deployed to mainnet.
+const START_BLOCK_OFFSET = 1000; // Number of blocks before the last known block to consider when updating fill events.
+const BATCH_SAVE_SIZE = 1000; // Number of events to save at once.
+
+let connection: Connection;
+
+(async () => {
+    connection = await createConnection(deployConfig);
+    const provider = web3Factory.getRpcProvider({
+        rpcUrl: 'https://mainnet.infura.io',
+    });
+    await getExchangeEventsAsync(provider);
+    process.exit(0);
+})();
+
+async function getExchangeEventsAsync(provider: Web3ProviderEngine): Promise<void> {
+    console.log('Checking existing event logs...');
+    const eventsRepository = connection.getRepository(ExchangeFillEvent);
+    const startBlock = await getStartBlockAsync(eventsRepository);
+    console.log(`Getting event logs starting at ${startBlock}...`);
+    const exchangeEvents = new ExchangeEventsSource(provider, 1);
+    const eventLogs = await exchangeEvents.getFillEventsAsync(startBlock);
+    console.log('Parsing events...');
+    const events = parseExchangeEvents(eventLogs);
+    console.log(`Retrieved and parsed ${events.length} total events.`);
+    console.log('Saving events...');
+    // Split the events into batches of size BATCH_SAVE_SIZE and save each batch
+    // in a single request. This reduces round-trip latency to the DB. We need
+    // to batch this way because saving an extremely large number of events in a
+    // single request causes problems.
+    for (const eventsBatch of R.splitEvery(BATCH_SAVE_SIZE, events)) {
+        await eventsRepository.save(eventsBatch);
+    }
+    const totalEvents = await eventsRepository.count();
+    console.log(`Done saving events. There are now ${totalEvents} total events.`);
+}
+
+async function getStartBlockAsync(eventsRepository: Repository<ExchangeFillEvent>): Promise<number> {
+    const fillEventCount = await eventsRepository.count();
+    if (fillEventCount === 0) {
+        console.log('No existing fill events found.');
+        return EXCHANGE_START_BLOCK;
+    }
+    const queryResult = await connection.query(
+        'SELECT "blockNumber" FROM exchange_fill_event ORDER BY "blockNumber" DESC LIMIT 1',
+    );
+    const lastKnownBlock = queryResult[0].blockNumber;
+    return lastKnownBlock - START_BLOCK_OFFSET;
+}
diff --git a/packages/pipeline/src/scripts/update_relayer_info.ts b/packages/pipeline/src/scripts/update_relayer_info.ts
new file mode 100644
index 000000000..05e045ff4
--- /dev/null
+++ b/packages/pipeline/src/scripts/update_relayer_info.ts
@@ -0,0 +1,31 @@
+import 'reflect-metadata';
+import { Connection, createConnection } from 'typeorm';
+
+import { RelayerRegistrySource } from '../data_sources/relayer-registry';
+import { Relayer } from '../entities/Relayer';
+import { deployConfig } from '../ormconfig';
+import { parseRelayers } from '../parsers/relayer_registry';
+
+// NOTE(albrow): We need to manually update this URL for now. Fix this when we
+// have the relayer-registry behind semantic versioning.
+const RELAYER_REGISTRY_URL =
+    'https://raw.githubusercontent.com/0xProject/0x-relayer-registry/4701c85677d161ea729a466aebbc1826c6aa2c0b/relayers.json';
+
+let connection: Connection;
+
+(async () => {
+    connection = await createConnection(deployConfig);
+    await getRelayers();
+    process.exit(0);
+})();
+
+async function getRelayers(): Promise<void> {
+    console.log('Getting latest relayer info...');
+    const relayerRepository = connection.getRepository(Relayer);
+    const relayerSource = new RelayerRegistrySource(RELAYER_REGISTRY_URL);
+    const relayersResp = await relayerSource.getRelayerInfoAsync();
+    const relayers = parseRelayers(relayersResp);
+    console.log('Saving relayer info...');
+    await relayerRepository.save(relayers);
+    console.log('Done saving relayer info.');
+}