@cloudant/couchbackup 2.9.13 → 2.9.14-SNAPSHOT.146

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 +13,92 @@
 // 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(onChange) {
-  const change = new stream.Transform({ objectMode: true });
-
-  change._transform = function(line, encoding, done) {
-    let obj = null;
+// fatal errors
+const codes = {
+  Error: 1,
+  InvalidOption: 2,
+  DatabaseNotFound: 10,
+  Unauthorized: 11,
+  Forbidden: 12,
+  DatabaseNotEmpty: 13,
+  NoLogFileName: 20,
+  LogDoesNotExist: 21,
+  IncompleteChangesInLogFile: 22,
+  SpoolChangesError: 30,
+  HTTPFatalError: 40,
+  BulkGetError: 50
+};
 
-    // one change per line - remove the trailing comma
-    line = line.trim().replace(/,$/, '');
+class BackupError extends Error {
+  constructor(name, message) {
+    super(message);
+    this.name = name;
+  }
+}
 
-    // extract thee last_seq at the end of the changes feed
-    if (line.match(/^"last_seq":/)) {
-      line = '{' + line;
+class HTTPError extends BackupError {
+  constructor(responseError, name) {
+    // Special case some names for more useful error messages
+    switch (responseError.status) {
+      case 401:
+        name = 'Unauthorized';
+        break;
+      case 403:
+        name = 'Forbidden';
+        break;
+      default:
+        name = name || 'HTTPFatalError';
     }
-    try {
-      obj = JSON.parse(line);
-    } catch (e) {
+    super(name, responseError.message);
+  }
+}
+
+// Default function to return an error for HTTP status codes
+// < 400 -> OK
+// 4XX (except 429) -> Fatal
+// 429 & >=500 -> Transient
+function checkResponse(err) {
+  if (err) {
+    // Construct an HTTPError if there is request information on the error
+    // Codes < 400 are considered OK
+    if (err.status >= 400) {
+      return new HTTPError(err);
+    } else {
+      // Send it back again if there was no status code, e.g. a cxn error
+      return augmentMessage(err);
     }
-    onChange(obj);
-    done();
-  };
+  }
+}
 
-  return change;
+function convertResponseError(responseError, errorFactory) {
+  if (!errorFactory) {
+    errorFactory = checkResponse;
+  }
+  return errorFactory(responseError);
+}
+
+function augmentMessage(err) {
+  // For errors that don't have a status code, we are likely looking at a cxn
+  // error.
+  // Try to augment the message with more detail (core puts the code in statusText)
+  if (err && err.statusText) {
+    err.message = `${err.message} ${err.statusText}`;
+  }
+  if (err && err.description) {
+    err.message = `${err.message} ${err.description}`;
+  }
+  return err;
+}
+
+module.exports = {
+  BackupError: BackupError,
+  HTTPError: HTTPError,
+  convertResponseError: convertResponseError,
+  terminationCallback: function terminationCallback(err, data) {
+    if (err) {
+      console.error(`ERROR: ${err.message}`);
+      process.exitCode = codes[err.name] || 1;
+      process.exit();
+    }
+  }
 };

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

@@ -1,4 +1,4 @@
-// Copyright © 2017, 2018 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.
@@ -13,19 +13,68 @@
 // 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);
+const async = require('async');
+const error = require('./error.js');
+const events = require('events');
 
-  // 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
+module.exports = function(db, options) {
+  const ee = new events.EventEmitter();
+  const start = new Date().getTime();
+  let batch = 0;
+  let hasErrored = false;
+  let startKey = null;
+  let total = 0;
 
-  callback(null, writer);
+  async.doUntil(
+    function(callback) {
+      // Note, include_docs: true is set automatically when using the
+      // fetch function.
+      const opts = { db: db.db, limit: options.bufferSize, includeDocs: true };
+
+      // To avoid double fetching a document solely for the purposes of getting
+      // the next ID to use as a startKey for the next page we instead use the
+      // last ID of the current page and append the lowest unicode sort
+      // character.
+      if (startKey) opts.startKey = `${startKey}\0`;
+      db.service.postAllDocs(opts).then(response => {
+        const body = response.result;
+        if (!body.rows) {
+          ee.emit('error', new error.BackupError(
+            'AllDocsError', 'ERROR: Invalid all docs response'));
+          callback();
+        } else {
+          if (body.rows.length < opts.limit) {
+            startKey = null; // last batch
+          } else {
+            startKey = body.rows[opts.limit - 1].id;
+          }
+
+          const docs = [];
+          body.rows.forEach(function(doc) {
+            docs.push(doc.doc);
+          });
+
+          if (docs.length > 0) {
+            ee.emit('received', {
+              batch: batch++,
+              data: docs,
+              length: docs.length,
+              time: (new Date().getTime() - start) / 1000,
+              total: total += docs.length
+            });
+          }
+          callback();
+        }
+      }).catch(err => {
+        err = error.convertResponseError(err);
+        ee.emit('error', err);
+        hasErrored = true;
+        callback();
+      });
+    },
+    function(callback) { callback(null, hasErrored || startKey == null); },
+    function() { ee.emit('finished', { total: total }); }
+  );
+
+  return ee;
 };

package/.scannerwork/scanner-report/source-54.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.
@@ -13,34 +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() {
-  const liner = 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;
 
-  liner._transform = function(chunk, encoding, done) {
-    let data = chunk.toString();
-    if (this._lastLineData) {
-      data = this._lastLineData + data;
+  // 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.');
     }
 
-    const lines = data.split('\n');
-    this._lastLineData = lines.splice(lines.length - 1, 1)[0];
+    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;
 
-    for (const i in lines) {
-      this.push(lines[i]);
+  // 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);
+      }
     }
-    done();
-  };
 
-  liner._flush = function(done) {
-    if (this._lastLineData) {
-      this.push(this._lastLineData);
+    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();
+    }
+  }
+
+  // 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();
     }
-    this._lastLineData = null;
-    done();
   };
 
-  return liner;
+  // called when we need to flush everything
+  writer._flush = function(done) {
+    processBuffer(true, done);
+  };
+  return writer;
 };

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

@@ -1,4 +1,4 @@
-// Copyright © 2017, 2021 IBM Corp. All rights reserved.
+// Copyright © 2017 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.
@@ -13,117 +13,80 @@
 // limitations under the License.
 'use strict';
 
-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;
-}
+const fs = require('fs');
+const stream = require('stream');
+const liner = require('./liner.js');
+
+const onLine = function(onCommand, getDocs) {
+  const change = new stream.Transform({ objectMode: true });
+
+  change._transform = function(line, encoding, done) {
+    if (line && line[0] === ':') {
+      const obj = {
+        command: null,
+        batch: null,
+        docs: []
+      };
+
+      let matches;
+
+      // extract command
+      matches = line.match(/^:([a-z_]+) ?/);
+      if (matches) {
+        obj.command = matches[1];
+      }
+
+      // extract batch
+      matches = line.match(/ batch([0-9]+)/);
+      if (matches) {
+        obj.batch = parseInt(matches[1]);
+      }
+
+      // extract doc ids
+      if (getDocs && obj.command === 't') {
+        const json = line.replace(/^.* batch[0-9]+ /, '').trim();
+        obj.docs = JSON.parse(json);
+      }
+      onCommand(obj);
+    }
+    done();
+  };
+  return change;
+};
 
-module.exports = {
-  parseBackupArgs: parseBackupArgs,
-  parseRestoreArgs: parseRestoreArgs
+/**
+ * Generate a list of remaining batches from a download file.
+ *
+ * @param {string} log - log file name
+ * @param {function} callback - callback with err, {changesComplete: N, batches: N}.
+ *  changesComplete signifies whether the log file appeared to
+ *  have completed reading the changes feed (contains :changes_complete).
+ *  batches are remaining batch IDs for download.
+ */
+module.exports = function(log, callback) {
+  // our sense of state
+  const state = {
+
+  };
+  let changesComplete = false;
+
+  // called with each line from the log file
+  const onCommand = function(obj) {
+    if (obj.command === 't') {
+      state[obj.batch] = true;
+    } else if (obj.command === 'd') {
+      delete state[obj.batch];
+    } else if (obj.command === 'changes_complete') {
+      changesComplete = true;
+    }
+  };
+
+  // stream through the previous log file
+  fs.createReadStream(log)
+    .pipe(liner())
+    .pipe(onLine(onCommand, false))
+    .on('finish', function() {
+      const obj = { changesComplete: changesComplete, batches: state };
+      callback(null, obj);
+    });
 };