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

4 files changed, 95 insertions, 92 deletions

diff --git a/packages/pipeline/src/index.ts b/packages/pipeline/src/index.ts
deleted file mode 100644
index 136d7a64b..000000000
--- a/packages/pipeline/src/index.ts
+++ /dev/null
@@ -1,88 +0,0 @@
-import { web3Factory } from '@0x/dev-utils';
-import { Web3ProviderEngine } from '@0x/subproviders';
-import 'reflect-metadata';
-import { Connection, createConnection } from 'typeorm';
-
-import { ExchangeEventsSource } from './data_sources/contract-wrappers/exchange_events';
-import { RelayerRegistrySource } from './data_sources/relayer-registry';
-import { Web3Source } from './data_sources/web3';
-import { Block } from './entities/Block';
-import { ExchangeFillEvent } from './entities/ExchangeFillEvent';
-import { Relayer } from './entities/Relayer';
-import { Transaction } from './entities/Transaction';
-import { deployConfig } from './ormconfig';
-import { parseExchangeEvents } from './parsers/events';
-import { parseRelayers } from './parsers/relayer_registry';
-import { parseBlock, parseTransaction } from './parsers/web3';
-
-const EXCHANGE_START_BLOCK = 6271590; // Block number when the Exchange contract was deployed to mainnet.
-// 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);
-    const provider = web3Factory.getRpcProvider({
-        rpcUrl: 'https://mainnet.infura.io',
-    });
-    await getExchangeEventsAsync(provider);
-    await getBlockAsync(provider);
-    await getTransactionAsync(provider);
-    await getRelayers();
-    process.exit(0);
-})();
-
-async function getExchangeEventsAsync(provider: Web3ProviderEngine): Promise<void> {
-    console.log('Getting event logs...');
-    const eventsRepository = connection.getRepository(ExchangeFillEvent);
-    const exchangeEvents = new ExchangeEventsSource(provider, 1);
-    const eventLogs = await exchangeEvents.getFillEventsAsync(EXCHANGE_START_BLOCK, EXCHANGE_START_BLOCK + 100000);
-    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);
-    }
-    console.log('Saved events.');
-}
-
-async function getBlockAsync(provider: Web3ProviderEngine): Promise<void> {
-    console.log('Getting block info...');
-    const blocksRepository = connection.getRepository(Block);
-    const web3Source = new Web3Source(provider);
-    const rawBlock = await web3Source.getBlockInfoAsync(EXCHANGE_START_BLOCK);
-    const block = parseBlock(rawBlock);
-    console.log('Saving block info...');
-    await blocksRepository.save(block);
-    console.log('Done saving block.');
-}
-
-async function getTransactionAsync(provider: Web3ProviderEngine): Promise<void> {
-    console.log('Getting tx info...');
-    const txsRepository = connection.getRepository(Transaction);
-    const web3Source = new Web3Source(provider);
-    const rawTx = await web3Source.getTransactionInfoAsync(
-        '0x6dd106d002873746072fc5e496dd0fb2541b68c77bcf9184ae19a42fd33657fe',
-    );
-    const tx = parseTransaction(rawTx);
-    console.log('Saving tx info...');
-    await txsRepository.save(tx);
-    console.log('Done saving tx.');
-}
-
-async function getRelayers(): Promise<void> {
-    console.log('Getting 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...');
-    // for (const relayer of relayers) {
-    await relayerRepository.save(relayers);
-    // }
-    console.log('Done saving relayers.');
-}
diff --git a/packages/pipeline/src/merge_v2_events.ts b/packages/pipeline/src/scripts/merge_v2_events.ts
index 24439423f..227ece121 100644
--- a/packages/pipeline/src/merge_v2_events.ts
+++ b/packages/pipeline/src/scripts/merge_v2_events.ts
@@ -2,10 +2,10 @@ 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';
+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;
 
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.');
+}