jsdatachecker.min.js
7.26 KB
function ArrayUtils(){}function DataTypeConverter(){this._fields=[],this._numOfRows=0}function DataTypesUtils(){}ArrayUtils.TestAndSet=function(arr,key,object){return"undefined"==typeof arr?null:0==Array.isArray(arr)?null:("undefined"==typeof arr[key]&&(arr[key]=object),arr[key])},ArrayUtils.TestAndIncrement=function(arr,key){var exists=arr[key];return"undefined"==typeof exists&&(arr[key]=0),arr[key]++,arr},ArrayUtils.IteratorOverKeys=function(arr,callback){for(var property in arr)if(arr.hasOwnProperty(property)){var item=arr[property];callback(item,property)}},ArrayUtils.FindMinMax=function(arr,fncompare){var result=null;for(var key in arr)(null==result||fncompare(arr[key],result.value))&&(result={index:-1,key:key,value:arr[key]});return result},DataTypeConverter.TYPES={TEXT:{value:0,name:"TEXT"},CODE:{value:1,name:"CODE"},NUMBER:{value:2,name:"NUMBER"},OBJECT:{value:3,name:"OBJECT"},BOOL:{value:5,name:"BOOL"},CONST:{value:6,name:"CONST"},CATEGORY:{value:7,name:"CATEGORY"},DATETIME:{value:8,name:"DATETIME"},EMPTY:{value:101,name:"NULL"}},DataTypeConverter.SUBTYPES={PERCENTAGE:{value:1e3,name:"PERCENTAGE"},LATITUDE:{value:1001,name:"LATITUDE"},LONGITUDE:{value:1002,name:"LONGITUDE"}},DataTypeConverter.prototype=function(){var _analyseDataTypes=function(fields){ArrayUtils.IteratorOverKeys(fields,function(field){if(field._inferredTypes[DataTypeConverter.TYPES.CODE.name]){var confidence=field._inferredTypes[DataTypeConverter.TYPES.CODE.name]/field.numOfItems,_numericalInferredType=field._inferredTypes[DataTypeConverter.TYPES.NUMBER.name];return"undefined"!=typeof _numericalInferredType&&(confidence+=_numericalInferredType/field.numOfItems),field.type=DataTypeConverter.TYPES.CODE.name,void(field.typeConfidence=confidence)}var max=ArrayUtils.FindMinMax(field._inferredTypes,function(curval,lastval){return curval>lastval});field.type=max.key,field.typeConfidence=field._inferredTypes[max.key]/field.numOfItems})},_processInferType=function(value){if(null===value||"undefined"==typeof value)return DataTypeConverter.TYPES.EMPTY;if("object"==typeof value)return DataTypeConverter.TYPES.OBJECT;if(/^0[0-9]+$/.test(value))return DataTypeConverter.TYPES.CODE;var isnumber=DataTypesUtils.FilterFloat(value);if(isNaN(isnumber)!==!0)return DataTypeConverter.TYPES.NUMBER;var _date=DataTypesUtils.FilterDateTime(value);return 0==isNaN(_date)&&null!=_date?DataTypeConverter.TYPES.DATETIME:DataTypeConverter.TYPES.TEXT},jsonTraverse=function(json,fieldKeys,callback){var stack=[],numOfRows=0;for(stack.push({item:json,fieldKeyIndex:0});stack.length>0;){var stackTask=stack.pop(),item=stackTask.item,fieldKeyIndex=stackTask.fieldKeyIndex,fieldKey=fieldKeys[fieldKeyIndex];if("*"!=fieldKey){var jsonSubtree=item[fieldKey];if(Array.isArray(jsonSubtree))for(var j=0;j<jsonSubtree.length;j++){var jsonItem=jsonSubtree[j];stack.push({item:jsonItem,fieldKeyIndex:fieldKeyIndex+1})}else stack.push({item:jsonSubtree,fieldKeyIndex:fieldKeyIndex+1})}else{var sProcessedKeys=fieldKeys.slice(0,fieldKeyIndex).toString();ArrayUtils.IteratorOverKeys(item,function(value,key){var curKey=sProcessedKeys+","+key,_value=callback(value,key,curKey,numOfRows);item[key]=_value}),numOfRows++}}};return{constructor:DataTypeConverter,convert:function(metadata){var lastRowIndex=0,numOfRows=0,numOfValues=0,datasetErrors=0,datasetMissingValues=0;return jsonTraverse(metadata.dataset,metadata.fieldKeys,function(value,key,traversedKeys,rowIndex){var inferredType=metadata.types[traversedKeys];if(numOfValues++,lastRowIndex!=rowIndex&&(lastRowIndex=rowIndex,numOfRows++),null==value||"undefined"==typeof value||0==(value+"").length,inferredType.type==DataTypeConverter.TYPES.NUMBER.name){var number=parseFloat(value);return isNaN(number)?(datasetErrors++,value):number}return value}),metadata.qualityIndex.notNullValues=(numOfValues-datasetMissingValues)/numOfValues,metadata.qualityIndex.errors=(numOfValues-datasetErrors)/numOfValues,metadata},inferJsonDataType:function(json,fieldKeys){var stack=[],fieldsType=[],numOfRows=0;if("undefined"==typeof fieldKeys)throw"IllegalArgumentException: undefined json path to analyse.";for(stack.push({item:json,fieldKeyIndex:0});stack.length>0;){var stackTask=stack.pop(),item=stackTask.item,fieldKeyIndex=stackTask.fieldKeyIndex,fieldKey=fieldKeys[fieldKeyIndex];if("*"!=fieldKey){var jsonSubtree=item[fieldKey];if(Array.isArray(jsonSubtree))for(var j=0;j<jsonSubtree.length;j++){var jsonItem=jsonSubtree[j];stack.push({item:jsonItem,fieldKeyIndex:fieldKeyIndex+1})}else stack.push({item:jsonSubtree,fieldKeyIndex:fieldKeyIndex+1})}else{var sProcessedKeys=fieldKeys.slice(0,fieldKeyIndex).toString();ArrayUtils.IteratorOverKeys(item,function(item,key){var inferredType=_processInferType(item),curKey=sProcessedKeys+","+key,fieldType=ArrayUtils.TestAndSet(fieldsType,curKey,{name:curKey,_inferredTypes:[],_inferredValues:[],numOfItems:0});fieldType.numOfItems++,ArrayUtils.TestAndIncrement(fieldType._inferredTypes,inferredType.name),inferredType===DataTypeConverter.TYPES.TEXT&&ArrayUtils.TestAndIncrement(fieldType._inferredValues,item)}),numOfRows++}}_analyseDataTypes(fieldsType);var quality={homogeneity:1};return ArrayUtils.IteratorOverKeys(fieldsType,function(fieldType){quality.homogeneity*=fieldType.typeConfidence}),{dataset:json,fieldKeys:fieldKeys,types:fieldsType,qualityIndex:quality}},inferDataTypeOfValue:function(value){return _processInferType(value)}}}(),DataTypesUtils.FilterTime=function(value){var expTime=/^[0-9]{2}:[0-9]{2}(:[0-9]{2})?(\+[0-9]{2}:[0-9]{2})?$/;if(0==expTime.test(value))return null;var splitted=value.split(/[:|\+]/),expNumber=/^[0-9]{2}$/,HH=expNumber.test(splitted[0])?parseInt(splitted[0]):0,MM=expNumber.test(splitted[1])?parseInt(splitted[1]):0,SS=splitted.length>=3&&expNumber.test(splitted[2])?parseInt(splitted[2]):0,dt=new Date;return dt.setHours(HH),dt.setMinutes(MM),dt.setSeconds(SS),dt},DataTypesUtils.FilterDateTime=function(value){var _dtSplitted=value.split(/[T|\s]/);if(2==_dtSplitted.length){var dtTime=DataTypesUtils.FilterTime(_dtSplitted[1]);if(null==dtTime)return null;var dtDateTime=DataTypesUtils.FilterDate(_dtSplitted[0],dtTime);return dtDateTime}var dtDate=DataTypesUtils.FilterDate(value);if(null!=dtDate)return dtDate;var dtTime=DataTypesUtils.FilterTime(value);return dtTime},DataTypesUtils.FilterDate=function(value,dtDate){if(null==dtDate&&(dtDate=new Date),/^[0-9][0-9][0-9][0-9]\-[0-9][0-9]$/.test(value)){var year=parseInt(value.substring(0,4)),month=parseInt(value.substring(5));return dtDate.setYear(year),dtDate.setMonth(month),dtDate}var expDate=/^[0-9]{4}(\-|\/)[0-9]{2}((\-|\/)[0-9]{2})?$/;if(expDate.test(value)){var splitted=value.split(/[\-|\/]/),year=parseInt(splitted[0]),month=parseInt(splitted[1]),day=3==splitted.length?parseInt(splitted[2]):0;return dtDate.setYear(year),dtDate.setMonth(month),dtDate.setDate(day),dtDate}if(expDate=/^[0-9]{2}(\-|\/)[0-9]{2}(\-|\/)[0-9]{4}$/,expDate.test(value)){var splitted=value.split(/[\-|\/]/),year=parseInt(splitted[2]),month=parseInt(splitted[1]),day=parseInt(splitted[0]);return dtDate.setYear(year),dtDate.setMonth(month),dtDate.setDate(day),dtDate}return null},DataTypesUtils.FilterFloat=function(value){return/^(\-|\+)?((0|([1-9][0-9]*))(\.[0-9]+)?|Infinity)$/.test(value)?Number(value):NaN},DataTypesUtils.DecimalPlaces=function(num){var match=(""+num).match(/(?:\.(\d+))?(?:[eE]([+-]?\d+))?$/);return match?Math.max(0,(match[1]?match[1].length:0)-(match[2]?+match[2]:0)):0};