rebuild package for new release

dist/node/browser-utils/utils.js

@@ -11,7 +11,7 @@ var _stream = require("stream");
 var _readableWebToNodeStream = require("readable-web-to-node-stream");
 
 async function webToNodeStream(stream, size) {
-  if (size == undefined || size == -1) {
+  if (size === undefined || size === -1) {
     return new _readableWebToNodeStream.ReadableWebToNodeStream(stream);
   } else {
     const nodeStream = new _stream.Readable({

dist/node/file-base.js

@@ -3,6 +3,7 @@
 Object.defineProperty(exports, "__esModule", {
   value: true
 });
+exports.computeHash = computeHash;
 exports.File = void 0;
 
 var _data = require("./data");
@@ -26,6 +27,15 @@ const {
 } = require('stream');
 
 class File {
+  constructor(descriptor, {
+    basePath
+  } = {}) {
+    this._descriptor = descriptor;
+    this._basePath = basePath;
+    this._descriptor.encoding = this.encoding || _data.DEFAULT_ENCODING;
+    this._computedHashes = {};
+  }
+
   static load(pathOrDescriptor, {
     basePath,
     format
@@ -37,14 +47,6 @@ class File {
     });
   }
 
-  constructor(descriptor, {
-    basePath
-  } = {}) {
-    this._descriptor = descriptor;
-    this._basePath = basePath;
-    this._descriptor.encoding = this.encoding || _data.DEFAULT_ENCODING;
-  }
-
   get descriptor() {
     return this._descriptor;
   }
@@ -99,8 +101,22 @@ class File {
     })();
   }
 
-  async hash(hashType = 'md5', progress) {
-    return _computeHash(await this.stream(), this.size, hashType, progress);
+  async hash(hashType = 'md5', progress, cache = true) {
+    if (cache && hashType in this._computedHashes) {
+      if (typeof progress === 'function') {
+        progress(100);
+      }
+
+      return this._computedHashes[hashType];
+    } else {
+      let hash = await computeHash(await this.stream(), this.size, hashType, progress);
+
+      if (cache && this != null) {
+        this._computedHashes[hashType] = hash;
+      }
+
+      return hash;
+    }
   }
 
   async hashSha256(progress) {
@@ -137,20 +153,31 @@ class File {
     throw new Error(`We do not have a parser for that format: ${this.descriptor.format}`);
   }
 
+  getSample() {
+    return new Promise(async (resolve, reject) => {
+      let smallStream = await this.rows({
+        size: 100
+      });
+      resolve(await (0, _streamToArray.default)(smallStream));
+    });
+  }
+
   async addSchema() {
     if (this.displayName === 'FileInline') {
       this.descriptor.schema = await (0, _tableschema.infer)(this.descriptor.data);
       return;
     }
 
-    if (this.descriptor.format === 'xlsx' && this.descriptor.sample) {
+    let sample = await this.getSample();
+
+    if (this.descriptor.format === 'xlsx' && sample) {
       let headers = 1;
 
-      if ((0, _lodash.isPlainObject)(this.descriptor.sample[0])) {
-        headers = Object.keys(this.descriptor.sample[0]);
+      if ((0, _lodash.isPlainObject)(sample[0])) {
+        headers = Object.keys(sample[0]);
       }
 
-      this.descriptor.schema = await (0, _tableschema.infer)(this.descriptor.sample, {
+      this.descriptor.schema = await (0, _tableschema.infer)(sample, {
         headers
       });
       return;
@@ -175,14 +202,18 @@ class File {
 
 exports.File = File;
 
-function _computeHash(fileStream, fileSize, algorithm, progress) {
+function computeHash(fileStream, fileSize, algorithm, progress, encoding = 'hex') {
   return new Promise((resolve, reject) => {
     let hash = _crypto.default.createHash(algorithm);
 
     let offset = 0;
     let totalChunkSize = 0;
     let chunkCount = 0;
 
+    if (!['hex', 'latin1', 'binary', 'base64'].includes(encoding)) {
+      throw new Error(`Invalid encoding value: ${encoding}; Expecting 'hex', 'latin1', 'binary' or 'base64'`);
+    }
+
     const _reportProgress = new Transform({
       transform(chunk, encoding, callback) {
         if (chunkCount % 20 == 0) {
@@ -206,7 +237,7 @@ function _computeHash(fileStream, fileSize, algorithm, progress) {
       chunkCount += 1;
       hash.update(chunk);
     }).on('end', function () {
-      hash = hash.digest('hex');
+      hash = hash.digest(encoding);
 
       if (typeof progress === 'function') {
         progress(100);

dist/node/file-interface.js

@@ -33,7 +33,9 @@ class FileInterface extends _fileBase.File {
     return this._encoding || _data.DEFAULT_ENCODING;
   }
 
-  async stream(size) {
+  async stream({
+    size
+  } = {}) {
     return (0, _index.webToNodeStream)(await this.descriptor.stream(), size);
   }
 

dist/node/file-local.js

@@ -25,8 +25,9 @@ class FileLocal extends _fileBase.File {
   }
 
   stream({
-    end
+    size
   } = {}) {
+    let end = size;
     return _fs.default.createReadStream(this.path, {
       start: 0,
       end

dist/node/index.js

@@ -39,6 +39,12 @@ Object.defineProperty(exports, "File", {
     return _fileBase.File;
   }
 });
+Object.defineProperty(exports, "computeHash", {
+  enumerable: true,
+  get: function () {
+    return _fileBase.computeHash;
+  }
+});
 Object.defineProperty(exports, "FileInterface", {
   enumerable: true,
   get: function () {

dist/node/parser/csv.js

@@ -23,7 +23,9 @@ async function csvParser(file, {
   size
 } = {}) {
   const parseOptions = await getParseOptions(file, keyed);
-  let stream = await file.stream(size);
+  let stream = await file.stream({
+    size
+  });
 
   if (file.descriptor.encoding.toLowerCase().replace('-', '') === 'utf8') {
     return stream.pipe((0, _csvParse.default)(parseOptions));
@@ -39,11 +41,13 @@ async function guessParseOptions(file) {
 
   if (file.displayName === 'FileLocal') {
     const stream = await file.stream({
-      end: 50000
+      size: 50000
     });
     text = await (0, _streamToString.default)(stream);
   } else if (file.displayName === 'FileInterface') {
-    let stream = await file.stream(10);
+    let stream = await file.stream({
+      size: 10
+    });
     text = await (0, _streamToString.default)(stream);
   } else if (file.displayName === 'FileRemote') {
     const stream = await file.stream({