Reduce the usage of lodash (round 2) (#145)

* Removed lodash from index.js

* Removed lodash from stats.js

* Removed lodash from stream.js

Only function I could not re-write was defaults-deep.
I moved the function to a utils folder, so we can remove it in the
near future

* Removed lodash from tests

* fixed linting

* Updated lodash 🚀

I could not completly remove lodash, need help with 2 function calls:

    - stream.js
        --> _.get
        --> _.defaultsDeep

* Attempt to fix travic ci failing build

* Updated mocha - istanbul to modern API

* Moving travis "back" to use npm ci

Maybe this can solve the CI issue?

* Fix CI error by using "normal" install

* removed and re-installed node_modules

* upgraded mocha

* updating out-dated dependencies

* Removed zuul

It was depreceated and not maintained anymore:
https://github.com/defunctzombie/zuul#readme

* Fix pre-commit to match setup in compas

This shoul help align the style used in:
https://github.com/mongodb-js/compass

* updated github workflow to look like compass

* updated mognodb-js-precommit to latest

* very small change from ssh --> https

* fixed failing tests from upgrading BSON

.eslintrc

@@ -1,6 +1,6 @@
 {
   "parserOptions": {
-    "ecmaVersion": 2017
+    "ecmaVersion": 2019
   },
   "env": {
     "mocha": true,
@@ -18,6 +18,6 @@
   },
   "extends": [
     "mongodb-js/node",
-    "mongodb-js/browser",
+    "mongodb-js/browser"
   ]
 }

.github/workflows/unit-tests.yml

@@ -16,6 +16,7 @@ jobs:
         uses: actions/setup-node@v2
         with:
           node-version: ${{ matrix.node-version }}
-      - uses: bahmutov/npm-install@v1
+      - run: npm ci
+      - run: npm run check
       - run: npm test
       - run: npm run coverage

.gitignore

@@ -33,4 +33,4 @@ dist/
 # Test output
 .nyc_output
 # Mac OS files on folder create/delete
-.DS_Store 
+.DS_Store 

bin/mongodb-schema

@@ -141,7 +141,7 @@ var bar = new ProgressBar('analyzing [:bar] :percent :etas ', {
   clear: true
 });
 
-mongodb.connect(uri, function(err, conn) {
+mongodb.connect(uri, {useUnifiedTopology: true}, function(err, conn) {
   if (err) {
     console.error('Failed to connect to MongoDB: ', err);
     process.exit(1);

lib/index.js

@@ -1,6 +1,5 @@
 var stream = require('./stream');
 var es = require('event-stream');
-var _ = require('lodash');
 
 // var debug = require('debug')('mongodb-schema:wrapper');
 
@@ -20,7 +19,7 @@ var _ = require('lodash');
 module.exports = function(docs, options, callback) {
   const promise = new Promise((resolve, reject) => {
     // shift parameters if no options are specified
-    if (_.isUndefined(options) || (_.isFunction(options) && _.isUndefined(callback))) {
+    if (typeof options === 'undefined' || (typeof options === 'function' && typeof callback === 'undefined')) {
       callback = options;
       options = {};
     }
@@ -33,7 +32,7 @@ module.exports = function(docs, options, callback) {
     } else if (docs.pipe && typeof docs.pipe === 'function') {
       src = docs;
       // Arrays
-    } else if (_.isArray(docs)) {
+    } else if (Array.isArray(docs)) {
       src = es.readArray(docs);
     } else {
       reject(new Error(

lib/stats.js

@@ -1,4 +1,3 @@
-var _ = require('lodash');
 // var debug = require('debug')('mongodb-schema:stats');
 
 var widthRecursive = function(schema) {
@@ -8,17 +7,21 @@ var widthRecursive = function(schema) {
   }
   if (schema.fields !== undefined) {
     width += schema.fields.length;
-    width += _.sum(schema.fields.map(function(field) {
-      var doc = _.find(field.types, 'name', 'Document');
+
+    width += schema.fields.map(field => {
+      var doc = field.types.find(v => v.name === 'Document');
       return widthRecursive(doc);
-    }));
-    width += _.sum(schema.fields.map(function(field) {
-      var arr = _.find(field.types, 'name', 'Array');
+    }).reduce((p, c) => p + c || 0, 0);
+
+
+    width += schema.fields.map(field => {
+      var arr = field.types.find(v => v.name === 'Array');
       if (arr) {
-        var doc = _.find(arr.types, 'name', 'Document');
+        var doc = arr.types.find(v => v.name === 'Document');
         return widthRecursive(doc);
       }
-    }));
+    })
+      .reduce((p, c) => p + c || 0, 0);
   }
   return width;
 };
@@ -30,18 +33,19 @@ var depthRecursive = function(schema) {
   var maxChildDepth = 0;
   if (schema.fields !== undefined && schema.fields.length > 0) {
     maxChildDepth = 1 + Math.max(
-      _.max(schema.fields.map(function(field) {
-        var doc = _.find(field.types, 'name', 'Document');
+      Math.max(...schema.fields.map(field => {
+        var doc = field.types.find(v => v.name === 'Document');
         return depthRecursive(doc);
       })),
-      _.max(schema.fields.map(function(field) {
-        var arr = _.find(field.types, 'name', 'Array');
+      Math.max(...schema.fields.map(field => {
+        var arr = field.types.find(v => v.name === 'Array');
         if (arr) {
-          var doc = _.find(arr.types, 'name', 'Document');
+          var doc = arr.types.find(v => v.name === 'Document');
           return depthRecursive(doc);
         }
         return 0;
-      })));
+      }))
+    );
   }
   return maxChildDepth;
 };
@@ -55,21 +59,21 @@ var branchingFactors = function(schema) {
   if (schema.fields !== undefined && schema.fields.length > 0) {
     branchArray.push(schema.fields.length);
     res = schema.fields.map(function(field) {
-      var doc = _.find(field.types, 'name', 'Document');
+      var doc = field.types.find(v => v.name === 'Document');
       return branchingFactors(doc);
     });
-    branchArray.push.apply(branchArray, _.flatten(res, true));
+    branchArray.push(...res.flat(Infinity));
     res = schema.fields.map(function(field) {
-      var arr = _.find(field.types, 'name', 'Array');
+      var arr = field.types.find(v => v.name === 'Array');
       if (arr) {
-        var doc = _.find(arr.types, 'name', 'Document');
+        var doc = arr.types.find(v => v.name === 'Document');
         return branchingFactors(doc);
       }
       return [];
     });
-    branchArray.push.apply(branchArray, _.flatten(res, true));
+    branchArray.push(...res.flat(Infinity));
   }
-  return _.sortBy(branchArray).reverse();
+  return branchArray.sort().reverse();
 };
 
 module.exports = {

lib/stream.js

@@ -1,6 +1,6 @@
 var es = require('event-stream');
-var _ = require('lodash');
 var Reservoir = require('reservoir');
+var _ = require('lodash');
 
 // var debug = require('debug')('mongodb-schema:stream');
 
@@ -13,14 +13,14 @@ var Reservoir = require('reservoir');
 */
 var extractStringValueFromBSON = function(value) {
   if (value && value._bsontype) {
-    if (_.includes([ 'Decimal128', 'Long' ], value._bsontype)) {
+    if (['Decimal128', 'Long'].includes(value._bsontype)) {
       return value.toString();
     }
-    if (_.includes([ 'Double', 'Int32' ], value._bsontype)) {
+    if ([ 'Double', 'Int32' ].includes(value._bsontype)) {
       return String(value.value);
     }
   }
-  if (_.isString(value)) {
+  if (typeof value === 'string') {
     return value;
   }
   return String(value);
@@ -68,7 +68,7 @@ var finalizeSchema = function(schema, parent, tag) {
     finalizeSchema(schema.fields, schema, 'fields');
   }
   if (tag === 'fields') {
-    _.each(schema, function(field) {
+    Object.values(schema).forEach((field) => {
       // create `Undefined` pseudo-type
       var missing = parent.count - field.count;
       if (missing > 0) {
@@ -79,25 +79,27 @@ var finalizeSchema = function(schema, parent, tag) {
           count: missing
         };
       }
-      field.total_count = _.sum(field.types, 'count');
+      field.total_count = Object.values(field.types)
+        .map(v => v.count)
+        .reduce((p, c) => p + c, 0);
 
       // recursively finalize types
       finalizeSchema(field.types, field, 'types');
-      field.type = _.pluck(field.types, 'name');
+      field.type = field.types.map(v => v.name);
       if (field.type.length === 1) {
         field.type = field.type[0];
       }
       // a field has duplicates when any of its types have duplicates
-      field.has_duplicates = _.any(field.types, 'has_duplicates');
+      field.has_duplicates = !!field.types.find(v => v.has_duplicates);
       // compute probability
       field.probability = field.count / parent.count;
     });
     // turn object into array
-    parent.fields = _.values(parent.fields).sort(fieldComparator);
+    parent.fields = Object.values(parent.fields).sort(fieldComparator);
   }
   if (tag === 'types') {
-    _.each(schema, function(type) {
-      type.total_count = _.sum(type.lengths);
+    Object.values(schema).forEach(type => {
+      type.total_count = (type.lengths || []).reduce((p, c) => p + c || 0, 0);
       // debug('recursively calling schema.fields');
       finalizeSchema(type.fields, type, 'fields');
       // debug('recursively calling schema.types');
@@ -110,7 +112,7 @@ var finalizeSchema = function(schema, parent, tag) {
         type.unique = type.count === 0 ? 0 : 1;
         type.has_duplicates = type.count > 1;
       } else if (type.values) {
-        type.unique = _.uniq(type.values, false, extractStringValueFromBSON).length;
+        type.unique = new Set(type.values.map(extractStringValueFromBSON)).size;
         type.has_duplicates = type.unique !== type.values.length;
       }
       // compute `average_length` for array types
@@ -119,7 +121,7 @@ var finalizeSchema = function(schema, parent, tag) {
       }
       // recursively finalize fields and types
     });
-    parent.types = _.sortByOrder(_.values(parent.types), 'probability', 'desc');
+    parent.types = Object.values(parent.types).sort((a, b) => b.probability - a.probability);
   }
   return schema;
 };
@@ -146,31 +148,25 @@ module.exports = function parse(options) {
   /* eslint no-sync: 0 */
 
   // set default options
-  options = _.defaults({}, options, {
-    semanticTypes: false,
-    storeValues: true
-  });
+  options = { semanticTypes: false, storeValues: true, ...options};
 
   var semanticTypes = require('./semantic-types');
 
-  if (_.isObject(options.semanticTypes)) {
+  if (typeof options.semanticTypes === 'object') {
     // enable existing types that evaluate to true
-    var enabledTypes = _(options.semanticTypes)
-      .pick(function(val) {
-        return _.isBoolean(val) && val;
-      })
-      .keys()
-      .map(function(val) {
-        return val.toLowerCase();
-      })
-      .value();
-    semanticTypes = _.pick(semanticTypes, function(val, key) {
-      return _.includes(enabledTypes, key.toLowerCase());
-    });
-    // merge with custom types that are functions
-    semanticTypes = _.assign(semanticTypes,
-      _.pick(options.semanticTypes, _.isFunction)
-    );
+    var enabledTypes = Object.entries(options.semanticTypes)
+      .filter(([, v]) => typeof v === 'boolean' && v)
+      .map(([k]) => k.toLowerCase());
+
+    semanticTypes = {...
+    Object.entries(semanticTypes)
+      .filter(([k]) => enabledTypes.includes(k.toLowerCase()))
+      .reduce((p, [k, v]) => ({...p, [k]: v}), {}),
+    };
+
+    Object.entries(options.semanticTypes)
+      .filter(([, v]) => typeof v === 'function')
+      .forEach(([k, v]) => {semanticTypes[k] = v;});
   }
 
   var rootSchema = {
@@ -205,9 +201,13 @@ module.exports = function parse(options) {
 
   var getSemanticType = function(value, path) {
     // pass value to semantic type detectors, return first match or undefined
-    return _.findKey(semanticTypes, function(fn) {
-      return fn(value, path);
-    });
+
+    const returnValue = Object.entries(semanticTypes)
+      .filter(([, v]) => {
+        return v(value, path);
+      })
+      .map(([k]) => k)[0];
+    return returnValue;
   };
 
   /**
@@ -236,13 +236,13 @@ module.exports = function parse(options) {
    * @param {Object}  schema   the updated schema object
    */
 
+
   var addToType = function(path, value, schema) {
     var bsonType = getBSONType(value);
     // if semantic type detection is enabled, the type is the semantic type
     // or the original bson type if no semantic type was detected. If disabled,
     // it is always the bson type.
-    var typeName = (options.semanticTypes) ?
-      getSemanticType(value, path) || bsonType : bsonType;
+    var typeName = (options.semanticTypes) ? getSemanticType(value, path) || bsonType : bsonType;
     var type = schema[typeName] = _.get(schema, typeName, {
       name: typeName,
       bsonType: bsonType,
@@ -252,24 +252,22 @@ module.exports = function parse(options) {
     type.count++;
     // recurse into arrays by calling `addToType` for each element
     if (typeName === 'Array') {
-      type.types = _.get(type, 'types', {});
-      type.lengths = _.get(type, 'lengths', []);
+      type.types = type.types || {};
+      type.lengths = type.lengths || [];
       type.lengths.push(value.length);
-      _.each(value, function(v) {
-        addToType(path, v, type.types);
-      });
+      value.forEach(v => addToType(path, v, type.types));
 
     // recurse into nested documents by calling `addToField` for all sub-fields
     } else if (typeName === 'Document') {
       type.fields = _.get(type, 'fields', {});
-      _.forOwn(value, function(v, k) {
-        addToField(path + '.' + k, v, type.fields);
-      });
+      Object.entries(value).forEach(([k, v]) => addToField(path + '.' + k, v, type.fields));
 
     // if the `storeValues` option is enabled, store some example values
     } else if (options.storeValues) {
-      type.values = _.get(type, 'values', bsonType === 'String' ?
-        new Reservoir(100) : new Reservoir(10000));
+      var defaultValue = bsonType === 'String' ?
+        new Reservoir(100) : new Reservoir(10000);
+      type.values = type.values || defaultValue;
+
       addToValue(type, value);
     }
   };
@@ -284,8 +282,9 @@ module.exports = function parse(options) {
   addToField = function(path, value, schema) {
     var defaults = {};
 
+    var pathSplitOnDot = path.split('.');
     defaults[path] = {
-      name: _.last(path.split('.')),
+      name: pathSplitOnDot[pathSplitOnDot.length - 1],
       path: path,
       count: 0,
       types: {}
@@ -306,9 +305,7 @@ module.exports = function parse(options) {
   }
 
   var parser = es.through(function write(obj) {
-    _.each(_.keys(obj), function(key) {
-      addToField(key, obj[key], rootSchema.fields);
-    });
+    Object.keys(obj).forEach(key => addToField(key, obj[key], rootSchema.fields));
     rootSchema.count += 1;
     this.emit('progress', obj);
   }, function end() {