@cloudant/couchbackup 2.9.13 → 2.9.14-SNAPSHOT.147

package/.scannerwork/scanner-report/source-51.txt

@@ -1,4 +1,4 @@
-// Copyright © 2017 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.
@@ -13,29 +13,152 @@
 // limitations under the License.
 'use strict';
 
-// stolen from http://strongloop.com/strongblog/practical-examples-of-the-new-node-js-streams-api/
+const async = require('async');
 const stream = require('stream');
+const error = require('./error.js');
+const debug = require('debug')('couchbackup:writer');
 
-module.exports = function(onChange) {
-  const change = new stream.Transform({ objectMode: true });
+module.exports = function(db, bufferSize, parallelism, ee) {
+  const writer = new stream.Transform({ objectMode: true });
+  let buffer = [];
+  let written = 0;
+  let linenumber = 0;
 
-  change._transform = function(line, encoding, done) {
-    let obj = null;
+  // this is the queue of chunks that are written to the database
+  // the queue's payload will be an array of documents to be written,
+  // the size of the array will be bufferSize. The variable parallelism
+  // determines how many HTTP requests will occur at any one time.
+  const q = async.queue(function(payload, cb) {
+    // if we are restoring known revisions, we need to supply new_edits=false
+    if (payload.docs && payload.docs[0] && payload.docs[0]._rev) {
+      payload.new_edits = false;
+      debug('Using new_edits false mode.');
+    }
+
+    if (!didError) {
+      db.service.postBulkDocs({
+        db: db.db,
+        bulkDocs: payload
+      }).then(response => {
+        if (!response.result || (payload.new_edits === false && response.result.length > 0)) {
+          throw new Error(`Error writing batch with new_edits:${payload.new_edits !== false}` +
+            ` and ${response.result ? response.result.length : 'unavailable'} items`);
+        }
+        written += payload.docs.length;
+        writer.emit('restored', { documents: payload.docs.length, total: written });
+        cb();
+      }).catch(err => {
+        err = error.convertResponseError(err);
+        debug(`Error writing docs ${err.name} ${err.message}`);
+        cb(err, payload);
+      });
+    }
+  }, parallelism);
+
+  let didError = false;
+
+  // write the contents of the buffer to CouchDB in blocks of bufferSize
+  function processBuffer(flush, callback) {
+    function taskCallback(err, payload) {
+      if (err && !didError) {
+        debug(`Queue task failed with error ${err.name}`);
+        didError = true;
+        q.kill();
+        writer.emit('error', err);
+      }
+    }
+
+    if (flush || buffer.length >= bufferSize) {
+      // work through the buffer to break off bufferSize chunks
+      // and feed the chunks to the queue
+      do {
+        // split the buffer into bufferSize chunks
+        const toSend = buffer.splice(0, bufferSize);
 
-    // one change per line - remove the trailing comma
-    line = line.trim().replace(/,$/, '');
+        // and add the chunk to the queue
+        debug(`Adding ${toSend.length} to the write queue.`);
+        q.push({ docs: toSend }, taskCallback);
+      } while (buffer.length >= bufferSize);
 
-    // extract thee last_seq at the end of the changes feed
-    if (line.match(/^"last_seq":/)) {
-      line = '{' + line;
+      // send any leftover documents to the queue
+      if (flush && buffer.length > 0) {
+        debug(`Adding remaining ${buffer.length} to the write queue.`);
+        q.push({ docs: buffer }, taskCallback);
+      }
+
+      // wait until the queue size falls to a reasonable level
+      async.until(
+        // wait until the queue length drops to twice the paralellism
+        // or until empty on the last write
+        function(callback) {
+          // if we encountered an error, stop this until loop
+          if (didError) {
+            return callback(null, true);
+          }
+          if (flush) {
+            callback(null, q.idle() && q.length() === 0);
+          } else {
+            callback(null, q.length() <= parallelism * 2);
+          }
+        },
+        function(cb) {
+          setTimeout(cb, 20);
+        },
+
+        function() {
+          if (flush && !didError) {
+            writer.emit('finished', { total: written });
+          }
+          // callback when we're happy with the queue size
+          callback();
+        });
+    } else {
+      callback();
     }
-    try {
-      obj = JSON.parse(line);
-    } catch (e) {
+  }
+
+  // take an object
+  writer._transform = function(obj, encoding, done) {
+    // each obj that arrives here is a line from the backup file
+    // it should contain an array of objects. The length of the array
+    // depends on the bufferSize at backup time.
+    linenumber++;
+    if (!didError && obj !== '') {
+      // see if it parses as JSON
+      try {
+        const arr = JSON.parse(obj);
+
+        // if it's an array with a length
+        if (typeof arr === 'object' && arr.length > 0) {
+          // push each document into a buffer
+          buffer = buffer.concat(arr);
+
+          // pause the stream
+          // it's likely that the speed with which data can be read from disk
+          // may exceed the rate it can be written to CouchDB. To prevent
+          // the whole file being buffered in memory, we pause the stream here.
+          // it is resumed, when processBuffer calls back and we call done()
+          this.pause();
+
+          // break the buffer in to bufferSize chunks to be written to the database
+          processBuffer(false, done);
+        } else {
+          ee.emit('error', new error.BackupError('BackupFileJsonError', `Error on line ${linenumber} of backup file - not an array`));
+          done();
+        }
+      } catch (e) {
+        ee.emit('error', new error.BackupError('BackupFileJsonError', `Error on line ${linenumber} of backup file - cannot parse as JSON`));
+        // Could be an incomplete write that was subsequently resumed
+        done();
+      }
+    } else {
+      done();
     }
-    onChange(obj);
-    done();
   };
 
-  return change;
+  // called when we need to flush everything
+  writer._flush = function(done) {
+    processBuffer(true, done);
+  };
+  return writer;
 };

package/.scannerwork/scanner-report/source-52.txt

@@ -1,4 +1,4 @@
-// Copyright © 2017, 2018 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.
@@ -13,19 +13,117 @@
 // limitations under the License.
 'use strict';
 
-module.exports = function(db, options, readstream, ee, callback) {
-  const liner = require('../includes/liner.js')();
-  const writer = require('../includes/writer.js')(db, options.bufferSize, options.parallelism, ee);
-
-  // pipe the input to the output, via transformation functions
-  readstream
-    .pipe(liner) // transform the input stream into per-line
-    .on('error', function(err) {
-      // Forward the error to the writer event emitter where we already have
-      // listeners on for handling errors
-      writer.emit('error', err);
-    })
-    .pipe(writer); // transform the data
-
-  callback(null, writer);
+const cliutils = require('./cliutils.js');
+const config = require('./config.js');
+const error = require('./error.js');
+const path = require('path');
+const pkg = require('../package.json');
+
+function parseBackupArgs() {
+  const program = require('commander');
+
+  // Option CLI defaults
+  const defaults = config.cliDefaults();
+
+  // Options set by environment variables
+  const envVarOptions = {};
+  config.applyEnvironmentVariables(envVarOptions);
+
+  program
+    .version(pkg.version)
+    .description('Backup a CouchDB/Cloudant database to a backup text file.')
+    .usage('[options...]')
+    .option('-b, --buffer-size <n>',
+      cliutils.getUsage('number of documents fetched at once', defaults.bufferSize),
+      Number)
+    .option('-d, --db <db>',
+      cliutils.getUsage('name of the database to backup', defaults.db))
+    .option('-k, --iam-api-key <API key>',
+      cliutils.getUsage('IAM API key to access the Cloudant server'))
+    .option('-l, --log <file>',
+      cliutils.getUsage('file to store logging information during backup; invalid in "shallow" mode', 'a temporary file'),
+      path.normalize)
+    .option('-m, --mode <mode>',
+      cliutils.getUsage('"shallow" if only a superficial backup is done (ignoring conflicts and revision tokens), else "full" for complete backup', defaults.mode),
+      (mode) => { return mode.toLowerCase(); })
+    .option('-o, --output <file>',
+      cliutils.getUsage('file name to store the backup data', 'stdout'),
+      path.normalize)
+    .option('-p, --parallelism <n>',
+      cliutils.getUsage('number of HTTP requests to perform in parallel when performing a backup; ignored in "shallow" mode', defaults.parallelism),
+      Number)
+    .option('-q, --quiet',
+      cliutils.getUsage('suppress batch messages', defaults.quiet))
+    .option('-r, --resume',
+      cliutils.getUsage('continue a previous backup from its last known position; invalid in "shallow" mode', defaults.resume))
+    .option('-t, --request-timeout <n>',
+      cliutils.getUsage('milliseconds to wait for a response to a HTTP request before retrying the request', defaults.requestTimeout),
+      Number)
+    .option('-u, --url <url>',
+      cliutils.getUsage('URL of the CouchDB/Cloudant server', defaults.url))
+    .parse(process.argv);
+
+  // Remove defaults that don't apply when using shallow mode
+  if (program.opts().mode === 'shallow' || envVarOptions.mode === 'shallow') {
+    delete defaults.parallelism;
+    delete defaults.log;
+    delete defaults.resume;
+  }
+
+  // Apply the options in order so that the CLI overrides env vars and env variables
+  // override defaults.
+  const opts = Object.assign({}, defaults, envVarOptions, program.opts());
+
+  if (opts.resume && (opts.log === defaults.log)) {
+    // If resuming and the log file arg is the newly generated tmp name from defaults then we know that --log wasn't specified.
+    // We have to do this check here for the CLI case because of the default.
+    error.terminationCallback(new error.BackupError('NoLogFileName', 'To resume a backup, a log file must be specified'));
+  }
+
+  return opts;
+}
+
+function parseRestoreArgs() {
+  const program = require('commander');
+
+  // Option CLI defaults
+  const defaults = config.cliDefaults();
+
+  // Options set by environment variables
+  const envVarOptions = {};
+  config.applyEnvironmentVariables(envVarOptions);
+
+  program
+    .version(pkg.version)
+    .description('Restore a CouchDB/Cloudant database from a backup text file.')
+    .usage('[options...]')
+    .option('-b, --buffer-size <n>',
+      cliutils.getUsage('number of documents restored at once', defaults.bufferSize),
+      Number)
+    .option('-d, --db <db>',
+      cliutils.getUsage('name of the new, existing database to restore to', defaults.db))
+    .option('-k, --iam-api-key <API key>',
+      cliutils.getUsage('IAM API key to access the Cloudant server'))
+    .option('-p, --parallelism <n>',
+      cliutils.getUsage('number of HTTP requests to perform in parallel when restoring a backup', defaults.parallelism),
+      Number)
+    .option('-q, --quiet',
+      cliutils.getUsage('suppress batch messages', defaults.quiet))
+    .option('-t, --request-timeout <n>',
+      cliutils.getUsage('milliseconds to wait for a response to a HTTP request before retrying the request', defaults.requestTimeout),
+      Number)
+    .option('-u, --url <url>',
+      cliutils.getUsage('URL of the CouchDB/Cloudant server', defaults.url))
+    .parse(process.argv);
+
+  // Apply the options in order so that the CLI overrides env vars and env variables
+  // override defaults.
+  const opts = Object.assign({}, defaults, envVarOptions, program.opts());
+
+  return opts;
+}
+
+module.exports = {
+  parseBackupArgs: parseBackupArgs,
+  parseRestoreArgs: parseRestoreArgs
 };

package/.scannerwork/scanner-report/source-53.txt

@@ -13,166 +13,106 @@
 // limitations under the License.
 'use strict';
 
-const pkg = require('../package.json');
-const stream = require('stream');
-const { CloudantV1, CouchdbSessionAuthenticator } = require('@ibm-cloud/cloudant');
-const { IamAuthenticator, NoAuthAuthenticator } = require('ibm-cloud-sdk-core');
-const retryPlugin = require('retry-axios');
-
-const userAgent = 'couchbackup-cloudant/' + pkg.version + ' (Node.js ' +
-      process.version + ')';
-
-// Class for streaming _changes error responses into
-// In general the response is a small error/reason JSON object
-// so it is OK to have this in memory.
-class ResponseWriteable extends stream.Writable {
-  constructor(options) {
-    super(options);
-    this.data = [];
+const path = require('path');
+const tmp = require('tmp');
+
+/**
+  Return API default settings.
+*/
+function apiDefaults() {
+  return {
+    parallelism: 5,
+    bufferSize: 500,
+    requestTimeout: 120000,
+    log: tmp.tmpNameSync(),
+    resume: false,
+    mode: 'full'
+  };
+}
+
+/**
+  Return CLI default settings.
+*/
+function cliDefaults() {
+  const defaults = apiDefaults();
+
+  // add additional legacy settings
+  defaults.db = 'test';
+  defaults.url = 'http://localhost:5984';
+
+  // add CLI only option
+  defaults.quiet = false;
+
+  return defaults;
+}
+
+/**
+  Override settings **in-place** with environment variables.
+*/
+function applyEnvironmentVariables(opts) {
+  // if we have a custom CouchDB url
+  if (typeof process.env.COUCH_URL !== 'undefined') {
+    opts.url = process.env.COUCH_URL;
   }
 
-  _write(chunk, encoding, callback) {
-    this.data.push(chunk);
-    callback();
+  // if we have a specified databases
+  if (typeof process.env.COUCH_DATABASE !== 'undefined') {
+    opts.db = process.env.COUCH_DATABASE;
   }
 
-  stringBody() {
-    return Buffer.concat(this.data).toString();
+  // if we have a specified buffer size
+  if (typeof process.env.COUCH_BUFFER_SIZE !== 'undefined') {
+    opts.bufferSize = parseInt(process.env.COUCH_BUFFER_SIZE);
   }
-}
 
-// An interceptor function to help augment error bodies with a little
-// extra information so we can continue to use consistent messaging
-// after the ugprade to @ibm-cloud/cloudant
-const errorHelper = async function(err) {
-  let method;
-  let requestUrl;
-  if (err.response) {
-    if (err.response.config.url) {
-      requestUrl = err.response.config.url;
-      method = err.response.config.method;
-    }
-    // Override the status text with an improved message
-    let errorMsg = `${err.response.status} ${err.response.statusText || ''}: ` +
-    `${method} ${requestUrl}`;
-    if (err.response.data) {
-      // Check if we have a JSON response and try to get the error/reason
-      if (err.response.headers['content-type'] === 'application/json') {
-        if (!err.response.data.error && err.response.data.pipe) {
-          // If we didn't find a JSON object with `error` then we might have a stream response.
-          // Detect the stream by the presence of `pipe` and use it to get the body and parse
-          // the error information.
-          const p = new Promise((resolve, reject) => {
-            const errorBody = new ResponseWriteable();
-            err.response.data.pipe(errorBody)
-              .on('finish', () => { resolve(JSON.parse(errorBody.stringBody())); })
-              .on('error', () => { reject(err); });
-          });
-          // Replace the stream on the response with the parsed object
-          err.response.data = await p;
-        }
-        // Append the error/reason if available
-        if (err.response.data.error) {
-          // Override the status text with our more complete message
-          errorMsg += ` - Error: ${err.response.data.error}`;
-          if (err.response.data.reason) {
-            errorMsg += `, Reason: ${err.response.data.reason}`;
-          }
-        }
-      } else {
-        errorMsg += err.response.data;
-      }
-      // Set a new message for use by the node-sdk-core
-      // We use the errors array because it gets processed
-      // ahead of all other service errors.
-      err.response.data.errors = [{ message: errorMsg }];
-    }
-  } else if (err.request) {
-    if (!err.message.includes(err.config.url)) {
-      // Augment the message with the URL and method
-      // but don't do it again if we already have the URL.
-      err.message = `${err.message}: ${err.config.method} ${err.config.url}`;
-    }
+  // if we have a specified parallelism
+  if (typeof process.env.COUCH_PARALLELISM !== 'undefined') {
+    opts.parallelism = parseInt(process.env.COUCH_PARALLELISM);
   }
-  return Promise.reject(err);
-};
 
-module.exports = {
-  client: function(rawUrl, opts) {
-    const url = new URL(rawUrl);
-    // Split the URL to separate service from database
-    // Use origin as the "base" to remove auth elements
-    const actUrl = new URL(url.pathname.substring(0, url.pathname.lastIndexOf('/')), url.origin);
-    const dbName = url.pathname.substring(url.pathname.lastIndexOf('/') + 1);
-    let authenticator;
-    // Default to cookieauth unless an IAM key is provided
-    if (opts.iamApiKey) {
-      const iamAuthOpts = { apikey: opts.iamApiKey };
-      if (opts.iamTokenUrl) {
-        iamAuthOpts.url = opts.iamTokenUrl;
-      }
-      authenticator = new IamAuthenticator(iamAuthOpts);
-    } else if (url.username) {
-      authenticator = new CouchdbSessionAuthenticator({
-        username: decodeURIComponent(url.username),
-        password: decodeURIComponent(url.password)
-      });
-    } else {
-      authenticator = new NoAuthAuthenticator();
-    }
-    const serviceOpts = {
-      authenticator: authenticator,
-      timeout: opts.requestTimeout,
-      // Axios performance options
-      maxContentLength: -1
-    };
-
-    const service = new CloudantV1(serviceOpts);
-    // Configure retries
-    const maxRetries = 2; // for 3 total attempts
-    service.getHttpClient().defaults.raxConfig = {
-      // retries for status codes
-      retry: maxRetries,
-      // retries for non-response e.g. ETIMEDOUT
-      noResponseRetries: maxRetries,
-      backoffType: 'exponential',
-      httpMethodsToRetry: ['GET', 'HEAD', 'POST'],
-      statusCodesToRetry: [
-        [429, 429],
-        [500, 599]
-      ],
-      shouldRetry: err => {
-        const cfg = retryPlugin.getConfig(err);
-        // cap at max retries regardless of response/non-response type
-        if (cfg.currentRetryAttempt >= maxRetries) {
-          return false;
-        } else {
-          return retryPlugin.shouldRetryRequest(err);
-        }
-      },
-      instance: service.getHttpClient()
-    };
-    retryPlugin.attach(service.getHttpClient());
-
-    service.setServiceUrl(actUrl.toString());
-    if (authenticator instanceof CouchdbSessionAuthenticator) {
-      // Awkward workaround for known Couch issue with compression on _session requests
-      // It is not feasible to disable compression on all requests with the amount of
-      // data this lib needs to move, so override the property in the tokenManager instance.
-      authenticator.tokenManager.requestWrapperInstance.compressRequestData = false;
-    }
-    if (authenticator.tokenManager && authenticator.tokenManager.requestWrapperInstance) {
-      authenticator.tokenManager.requestWrapperInstance.axiosInstance.interceptors.response.use(null, errorHelper);
-    }
-    // Add error interceptors to put URLs in error messages
-    service.getHttpClient().interceptors.response.use(null, errorHelper);
-
-    // Add request interceptor to add user-agent (adding it with custom request headers gets overwritten)
-    service.getHttpClient().interceptors.request.use(function(requestConfig) {
-      requestConfig.headers['User-Agent'] = userAgent;
-      return requestConfig;
-    }, null);
-
-    return { service: service, db: dbName, url: actUrl.toString() };
+  // if we have a specified request timeout
+  if (typeof process.env.COUCH_REQUEST_TIMEOUT !== 'undefined') {
+    opts.requestTimeout = parseInt(process.env.COUCH_REQUEST_TIMEOUT);
+  }
+
+  // if we have a specified log file
+  if (typeof process.env.COUCH_LOG !== 'undefined') {
+    opts.log = path.normalize(process.env.COUCH_LOG);
+  }
+
+  // if we are instructed to resume
+  if (typeof process.env.COUCH_RESUME !== 'undefined' && process.env.COUCH_RESUME === 'true') {
+    opts.resume = true;
+  }
+
+  // if we are given an output filename
+  if (typeof process.env.COUCH_OUTPUT !== 'undefined') {
+    opts.output = path.normalize(process.env.COUCH_OUTPUT);
+  }
+
+  // if we only want a shallow copy
+  if (typeof process.env.COUCH_MODE !== 'undefined' && process.env.COUCH_MODE === 'shallow') {
+    opts.mode = 'shallow';
+  }
+
+  // if we are instructed to be quiet
+  if (typeof process.env.COUCH_QUIET !== 'undefined' && process.env.COUCH_QUIET === 'true') {
+    opts.quiet = true;
   }
+
+  // if we have a specified API key
+  if (typeof process.env.CLOUDANT_IAM_API_KEY !== 'undefined') {
+    opts.iamApiKey = process.env.CLOUDANT_IAM_API_KEY;
+  }
+
+  // if we have a specified IAM token endpoint
+  if (typeof process.env.CLOUDANT_IAM_TOKEN_URL !== 'undefined') {
+    opts.iamTokenUrl = process.env.CLOUDANT_IAM_TOKEN_URL;
+  }
+}
+
+module.exports = {
+  apiDefaults: apiDefaults,
+  cliDefaults: cliDefaults,
+  applyEnvironmentVariables: applyEnvironmentVariables
 };

package/.scannerwork/scanner-report/source-54.txt

@@ -1,4 +1,4 @@
-// Copyright © 2017 IBM Corp. All rights reserved.
+// Copyright © 2017, 2018 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.
@@ -13,34 +13,19 @@
 // limitations under the License.
 'use strict';
 
-// stolen from http://strongloop.com/strongblog/practical-examples-of-the-new-node-js-streams-api/
-const stream = require('stream');
+module.exports = function(db, options, readstream, ee, callback) {
+  const liner = require('../includes/liner.js')();
+  const writer = require('../includes/writer.js')(db, options.bufferSize, options.parallelism, ee);
 
-module.exports = function() {
-  const liner = new stream.Transform({ objectMode: true });
+  // pipe the input to the output, via transformation functions
+  readstream
+    .pipe(liner) // transform the input stream into per-line
+    .on('error', function(err) {
+      // Forward the error to the writer event emitter where we already have
+      // listeners on for handling errors
+      writer.emit('error', err);
+    })
+    .pipe(writer); // transform the data
 
-  liner._transform = function(chunk, encoding, done) {
-    let data = chunk.toString();
-    if (this._lastLineData) {
-      data = this._lastLineData + data;
-    }
-
-    const lines = data.split('\n');
-    this._lastLineData = lines.splice(lines.length - 1, 1)[0];
-
-    for (const i in lines) {
-      this.push(lines[i]);
-    }
-    done();
-  };
-
-  liner._flush = function(done) {
-    if (this._lastLineData) {
-      this.push(this._lastLineData);
-    }
-    this._lastLineData = null;
-    done();
-  };
-
-  return liner;
+  callback(null, writer);
 };