@cloudant/couchbackup 2.9.14 → 2.9.15-SNAPSHOT.162

package/.scannerwork/scanner-report/source-55.txt

@@ -13,117 +13,152 @@
 // limitations under the License.
 'use strict';
 
-const cliutils = require('./cliutils.js');
-const config = require('./config.js');
+const async = require('async');
+const stream = require('stream');
 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;
+const debug = require('debug')('couchbackup:writer');
+
+module.exports = function(db, bufferSize, parallelism, ee) {
+  const writer = new stream.Transform({ objectMode: true });
+  let buffer = [];
+  let written = 0;
+  let linenumber = 0;
+
+  // 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);
+
+        // and add the chunk to the queue
+        debug(`Adding ${toSend.length} to the write queue.`);
+        q.push({ docs: toSend }, taskCallback);
+      } while (buffer.length >= bufferSize);
+
+      // 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();
+    }
   }
 
-  // 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
+  // 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();
+    }
+  };
+
+  // called when we need to flush everything
+  writer._flush = function(done) {
+    processBuffer(true, done);
+  };
+  return writer;
 };

package/.scannerwork/scanner-report/source-56.txt

@@ -1,4 +1,4 @@
-// Copyright © 2017 IBM Corp. All rights reserved.
+// Copyright © 2017, 2022 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 +13,99 @@
 // 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');
+const fs = require('fs');
+const liner = require('./liner.js');
+const change = require('./change.js');
+const error = require('./error.js');
+const debug = require('debug')('couchbackup:spoolchanges');
 
-module.exports = function() {
-  const liner = new stream.Transform({ objectMode: true });
+/**
+ * Write log file for all changes from a database, ready for downloading
+ * in batches.
+ *
+ * @param {string} dbUrl - URL of database
+ * @param {string} log - path to log file to use
+ * @param {number} bufferSize - the number of changes per batch/log line
+ * @param {function(err)} callback - a callback to run on completion
+ */
+module.exports = function(db, log, bufferSize, ee, callback) {
+  // list of document ids to process
+  const buffer = [];
+  let batch = 0;
+  let lastSeq = null;
+  const logStream = fs.createWriteStream(log);
+  let pending = 0;
+  // The number of changes to fetch per request
+  const limit = 100000;
 
-  liner._transform = function(chunk, encoding, done) {
-    let data = chunk.toString();
-    if (this._lastLineData) {
-      data = this._lastLineData + data;
+  // send documents ids to the queue in batches of bufferSize + the last batch
+  const processBuffer = function(lastOne) {
+    if (buffer.length >= bufferSize || (lastOne && buffer.length > 0)) {
+      debug('writing', buffer.length, 'changes to the backup file');
+      const b = { docs: buffer.splice(0, bufferSize), batch: batch };
+      logStream.write(':t batch' + batch + ' ' + JSON.stringify(b.docs) + '\n');
+      ee.emit('changes', batch);
+      batch++;
     }
-
-    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);
+  // called once per received change
+  const onChange = function(c) {
+    if (c) {
+      if (c.error) {
+        ee.emit('error', new error.BackupError('InvalidChange', `Received invalid change: ${c}`));
+      } else if (c.changes) {
+        const obj = { id: c.id };
+        buffer.push(obj);
+        processBuffer(false);
+      } else if (c.last_seq) {
+        lastSeq = c.last_seq;
+        pending = c.pending;
+      }
     }
-    this._lastLineData = null;
-    done();
   };
 
-  return liner;
+  function getChanges(since = 0) {
+    debug('making changes request since ' + since);
+    return db.service.postChangesAsStream({ db: db.db, since: since, limit: limit, seqInterval: limit })
+      .then(response => {
+        response.result.pipe(liner())
+          .on('error', function(err) {
+            logStream.end();
+            callback(err);
+          })
+          .pipe(change(onChange))
+          .on('error', function(err) {
+            logStream.end();
+            callback(err);
+          })
+          .on('finish', function() {
+            processBuffer(true);
+            if (!lastSeq) {
+              logStream.end();
+              debug('changes request terminated before last_seq was sent');
+              callback(new error.BackupError('SpoolChangesError', 'Changes request terminated before last_seq was sent'));
+            } else {
+              debug(`changes request completed with last_seq: ${lastSeq} and ${pending} changes pending.`);
+              if (pending > 0) {
+                // Return the next promise
+                return getChanges(lastSeq);
+              } else {
+                debug('finished streaming database changes');
+                logStream.end(':changes_complete ' + lastSeq + '\n', 'utf8', callback);
+              }
+            }
+          });
+      })
+      .catch(err => {
+        logStream.end();
+        if (err.status && err.status >= 400) {
+          callback(error.convertResponseError(err));
+        } else if (err.name !== 'SpoolChangesError') {
+          callback(new error.BackupError('SpoolChangesError', `Failed changes request - ${err.message}`));
+        }
+      });
+  }
+
+  getChanges();
 };