@cloudant/couchbackup 2.9.14-SNAPSHOT.159 → 2.9.14

package/.scannerwork/scanner-report/source-57.txt

@@ -1,4 +1,4 @@
-// Copyright © 2017, 2023 IBM Corp. All rights reserved.
+// Copyright © 2017, 2021 IBM Corp. All rights reserved.
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -11,181 +11,271 @@
 // 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.
-
-/* global describe it */
 'use strict';
 
-const assert = require('assert');
-const nock = require('nock');
-const request = require('../includes/request.js');
-const changes = require('../includes/spoolchanges.js');
-
-const url = 'http://localhost:7777';
-const dbName = 'fakenockdb';
-const longTestTimeout = 3000;
-
-const db = request.client(`${url}/${dbName}`, { parallelism: 1 });
+const async = require('async');
+const events = require('events');
+const fs = require('fs');
+const error = require('./error.js');
+const spoolchanges = require('./spoolchanges.js');
+const logfilesummary = require('./logfilesummary.js');
+const logfilegetbatches = require('./logfilegetbatches.js');
 
-const seqSuffix = Buffer.alloc(124, 'abc123').toString('base64');
-function provideChanges(batchSize, totalChanges, fullResponse = false) {
-  let pending = totalChanges;
-  const sparseResultsArray = Array(batchSize).fill({
-    seq: null,
-    id: 'doc',
-    changes: [{ rev: '1-abcdef0123456789abcdef0123456789' }]
-  });
-  nock(url)
-    .post(`/${dbName}/_changes`)
-    .query(true)
-    .times(totalChanges / batchSize + (totalChanges % batchSize > 0 ? 1 : 0))
-    .reply(200, (uri, requestBody) => {
-      pending -= batchSize;
-      const lastSeq = (totalChanges - pending);
-      const seq = lastSeq - batchSize;
-      return {
-        results: fullResponse
-          ? Array.from(Array(batchSize), (_, i) => {
-            return {
-              seq: `${seq + i}-${seqSuffix}`,
-              id: `doc${seq + i}`,
-              changes: [{ rev: '1-abcdef0123456789abcdef0123456789' }]
-            };
-          })
-          : sparseResultsArray,
-        pending: pending,
-        last_seq: `${lastSeq}-abc`
-      };
-    });
-}
+/**
+ * Read documents from a database to be backed up.
+ *
+ * @param {string} db - `@cloudant/cloudant` DB object for source database.
+ * @param {number} blocksize - number of documents to download in single request
+ * @param {number} parallelism - number of concurrent downloads
+ * @param {string} log - path to log file to use
+ * @param {boolean} resume - whether to resume from an existing log file
+ * @returns EventEmitter with following events:
+ *  - `received` - called with a block of documents to write to backup
+ *  - `error` - on error
+ *  - `finished` - when backup process is finished (either complete or errored)
+ */
+module.exports = function(db, options) {
+  const ee = new events.EventEmitter();
+  const start = new Date().getTime(); // backup start time
+  const batchesPerDownloadSession = 50; // max batches to read from log file for download at a time (prevent OOM)
 
-describe('#unit Check spool changes', function() {
-  it('should terminate on request error', async function() {
-    nock(url)
-      .post(`/${dbName}/_changes`)
-      .query(true)
-      .times(3)
-      .replyWithError({ code: 'ECONNRESET', message: 'socket hang up' });
-
-    return new Promise((resolve, reject) => {
-      changes(db, '/dev/null', 500, null, function(err) {
-        try {
-          assert.strictEqual(err.name, 'SpoolChangesError');
-          assert.strictEqual(err.message, `Failed changes request - socket hang up: post ${url}/${dbName}/_changes`);
-          assert.ok(nock.isDone());
-          resolve();
-        } catch (err) {
-          reject(err);
-        }
-      });
-    });
-  }).timeout(longTestTimeout);
-
-  it('should terminate on bad HTTP status code response', async function() {
-    nock(url)
-      .post(`/${dbName}/_changes`)
-      .query(true)
-      .times(3)
-      .reply(500, function(uri, requestBody) {
-        this.req.response.statusMessage = 'Internal Server Error';
-        return { error: 'foo', reason: 'bar' };
-      });
-    return new Promise((resolve, reject) => {
-      changes(db, '/dev/null', 500, null, function(err) {
-        try {
-          assert.strictEqual(err.name, 'HTTPFatalError');
-          assert.strictEqual(err.message, `500 Internal Server Error: post ${url}/${dbName}/_changes - Error: foo, Reason: bar`);
-          assert.ok(nock.isDone());
-          resolve();
-        } catch (err) {
-          reject(err);
+  function proceedWithBackup() {
+    if (options.resume) {
+      // pick up from existing log file from previous run
+      downloadRemainingBatches(options.log, db, ee, start, batchesPerDownloadSession, options.parallelism);
+    } else {
+      // create new log file and process
+      spoolchanges(db, options.log, options.bufferSize, ee, function(err) {
+        if (err) {
+          ee.emit('error', err);
+        } else {
+          downloadRemainingBatches(options.log, db, ee, start, batchesPerDownloadSession, options.parallelism);
         }
       });
-    });
-  }).timeout(longTestTimeout);
-
-  it('should keep collecting changes', async function() {
-    // This test validates that spooling will correctly
-    // continue across multiple requests
-    // (4 batches of 100000 to be precise).
-    // This test might take up to 10 seconds
-    this.timeout(10 * 1000);
-
-    // Use full changes for this test
-    provideChanges(100000, 400000, true);
-    return new Promise((resolve, reject) => {
-      changes(db, '/dev/null', 500, null, function(err) {
-        try {
-          assert.ok(!err);
-          assert.ok(nock.isDone());
-          resolve();
-        } catch (err) {
-          reject(err);
-        }
-      });
-    });
+    }
+  }
+
+  validateBulkGetSupport(db, function(err) {
+    if (err) {
+      return ee.emit('error', err);
+    } else {
+      proceedWithBackup();
+    }
   });
 
-  it('should keep collecting sparse changes', async function() {
-    // This test checks that making thousands of requests doesn't
-    // make anything bad happen.
-    // This test might take up to 25 seconds
-    this.timeout(25 * 1000);
-    // Use sparse changes for this test and a batch size of 1
-    provideChanges(1, 2500);
-    return new Promise((resolve, reject) => {
-      changes(db, '/dev/null', 500, null, function(err) {
-        try {
-          assert.ok(!err);
-          assert.ok(nock.isDone());
-          resolve();
-        } catch (err) {
-          reject(err);
-        }
-      });
+  return ee;
+};
+
+/**
+ * Validate /_bulk_get support for a specified database.
+ *
+ * @param {string} db - nodejs-cloudant db
+ * @param {function} callback - called on completion with signature (err)
+ */
+function validateBulkGetSupport(db, callback) {
+  db.service.postBulkGet({ db: db.db, docs: [] }).then(() => { callback(); }).catch(err => {
+    err = error.convertResponseError(err, function(err) {
+      switch (err.status) {
+        case undefined:
+          // There was no status code on the error
+          return err;
+        case 404:
+          return new error.BackupError('BulkGetError', 'Database does not support /_bulk_get endpoint');
+        default:
+          return new error.HTTPError(err);
+      }
     });
+    callback(err);
   });
-});
-
-describe('Longer spool changes checks', function() {
-  it('#slow should keep collecting changes (25M)', async function() {
-    // This test might take up to 5 minutes
-    this.timeout(5 * 60 * 1000);
-    // Note changes spooling uses a constant batch size, we are setting
-    // a test value here and setting the buffer to match
-    const batch = 100000;
-    // Use sparse changes for this test
-    provideChanges(batch, 25000000);
-    return new Promise((resolve, reject) => {
-      changes(db, '/dev/null', batch, null, function(err) {
-        try {
-          assert.ok(!err);
-          assert.ok(nock.isDone());
-          resolve();
-        } catch (err) {
-          reject(err);
+}
+
+/**
+ * Download remaining batches in a log file, splitting batches into sets
+ * to avoid enqueueing too many in one go.
+ *
+ * @param {string} log - log file name to maintain download state
+ * @param {string} db - nodejs-cloudant db
+ * @param {events.EventEmitter} ee - event emitter to emit received events on
+ * @param {time} startTime - start time for backup process
+ * @param {number} batchesPerDownloadSession - max batches to enqueue for
+ *  download at a time. As batches contain many doc IDs, this helps avoid
+ *  exhausting memory.
+ * @param {number} parallelism - number of concurrent downloads
+ * @returns function to call do download remaining batches with signature
+ *  (err, {batches: batch, docs: doccount}) {@see spoolchanges}.
+ */
+function downloadRemainingBatches(log, db, ee, startTime, batchesPerDownloadSession, parallelism) {
+  let total = 0; // running total of documents downloaded so far
+  let noRemainingBatches = false;
+
+  // Generate a set of batches (up to batchesPerDownloadSession) to download from the
+  // log file and download them. Set noRemainingBatches to `true` for last batch.
+  function downloadSingleBatchSet(done) {
+    // Fetch the doc IDs for the batches in the current set to
+    // download them.
+    function batchSetComplete(err, data) {
+      if (!err) {
+        total = data.total;
+      }
+      done(err);
+    }
+    function processRetrievedBatches(err, batches) {
+      if (!err) {
+        // process them in parallelised queue
+        processBatchSet(db, parallelism, log, batches, ee, startTime, total, batchSetComplete);
+      } else {
+        batchSetComplete(err);
+      }
+    }
+
+    readBatchSetIdsFromLogFile(log, batchesPerDownloadSession, function(err, batchSetIds) {
+      if (err) {
+        ee.emit('error', err);
+        // Stop processing changes file for fatal errors
+        noRemainingBatches = true;
+        done();
+      } else {
+        if (batchSetIds.length === 0) {
+          noRemainingBatches = true;
+          return done();
         }
-      });
+        logfilegetbatches(log, batchSetIds, processRetrievedBatches);
+      }
     });
+  }
+
+  // Return true if all batches in log file have been downloaded
+  function isFinished(callback) { callback(null, noRemainingBatches); }
+
+  function onComplete() {
+    ee.emit('finished', { total: total });
+  }
+
+  async.doUntil(downloadSingleBatchSet, isFinished, onComplete);
+}
+
+/**
+ * Return a set of uncompleted download batch IDs from the log file.
+ *
+ * @param {string} log - log file path
+ * @param {number} batchesPerDownloadSession - maximum IDs to return
+ * @param {function} callback - sign (err, batchSetIds array)
+ */
+function readBatchSetIdsFromLogFile(log, batchesPerDownloadSession, callback) {
+  logfilesummary(log, function processSummary(err, summary) {
+    if (!err) {
+      if (!summary.changesComplete) {
+        callback(new error.BackupError('IncompleteChangesInLogFile',
+          'WARNING: Changes did not finish spooling'));
+        return;
+      }
+      if (Object.keys(summary.batches).length === 0) {
+        return callback(null, []);
+      }
+
+      // batch IDs are the property names of summary.batches
+      const batchSetIds = getPropertyNames(summary.batches, batchesPerDownloadSession);
+      callback(null, batchSetIds);
+    } else {
+      callback(err);
+    }
   });
+}
+
+/**
+ * Download a set of batches retrieved from a log file. When a download is
+ * complete, add a line to the logfile indicating such.
+ *
+ * @param {any} db - nodejs-cloudant database
+ * @param {any} parallelism - number of concurrent requests to make
+ * @param {any} log - log file to drive downloads from
+ * @param {any} batches - batches to download
+ * @param {any} ee - event emitter for progress. This funciton emits
+ *  received and error events.
+ * @param {any} start - time backup started, to report deltas
+ * @param {any} grandtotal - count of documents downloaded prior to this set
+ *  of batches
+ * @param {any} callback - completion callback, (err, {total: number}).
+ */
+function processBatchSet(db, parallelism, log, batches, ee, start, grandtotal, callback) {
+  let hasErrored = false;
+  let total = grandtotal;
+
+  // queue to process the fetch requests in an orderly fashion using _bulk_get
+  const q = async.queue(function(payload, done) {
+    const output = [];
+    const thisBatch = payload.batch;
+    delete payload.batch;
+    delete payload.command;
 
-  it('#slower should keep collecting changes (500M)', async function() {
-    // This test might take up to 90 minutes
-    this.timeout(90 * 60 * 1000);
-    // Note changes spooling uses a constant batch size, we are setting
-    // a test value here and setting the buffer to match
-    const batch = 1000000;
-    // Use full changes for this test to exercise load
-    provideChanges(batch, 500000000, true);
-    return new Promise((resolve, reject) => {
-      changes(db, '/dev/null', batch, null, function(err) {
-        try {
-          assert.ok(!err);
-          assert.ok(nock.isDone());
-          resolve();
-        } catch (err) {
-          reject(err);
+    function logCompletedBatch(batch) {
+      if (log) {
+        fs.appendFile(log, ':d batch' + thisBatch + '\n', done);
+      } else {
+        done();
+      }
+    }
+
+    // do the /db/_bulk_get request
+    db.service.postBulkGet({
+      db: db.db,
+      revs: true,
+      docs: payload.docs
+    }).then(response => {
+      // create an output array with the docs returned
+      response.result.results.forEach(function(d) {
+        if (d.docs) {
+          d.docs.forEach(function(doc) {
+            if (doc.ok) {
+              output.push(doc.ok);
+            }
+          });
         }
       });
+      total += output.length;
+      const t = (new Date().getTime() - start) / 1000;
+      ee.emit('received', {
+        batch: thisBatch,
+        data: output,
+        length: output.length,
+        time: t,
+        total: total
+      }, q, logCompletedBatch);
+    }).catch(err => {
+      if (!hasErrored) {
+        hasErrored = true;
+        err = error.convertResponseError(err);
+        // Kill the queue for fatal errors
+        q.kill();
+        ee.emit('error', err);
+      }
+      done();
     });
+  }, parallelism);
+
+  for (const i in batches) {
+    q.push(batches[i]);
+  }
+
+  q.drain(function() {
+    callback(null, { total: total });
   });
-});
+}
+
+/**
+ * Returns first N properties on an object.
+ *
+ * @param {object} obj - object with properties
+ * @param {number} count - number of properties to return
+ */
+function getPropertyNames(obj, count) {
+  // decide which batch numbers to deal with
+  const batchestofetch = [];
+  let j = 0;
+  for (const i in obj) {
+    batchestofetch.push(parseInt(i));
+    j++;
+    if (j >= count) break;
+  }
+  return batchestofetch;
+}