@loaders.gl/xml 3.4.0-alpha.1 → 3.4.0-alpha.3

package/dist/index.d.ts

@@ -1,5 +1,9 @@
 export type { XMLLoaderOptions } from './xml-loader';
 export { XMLLoader } from './xml-loader';
+export type { HTMLLoaderOptions } from './html-loader';
+export { HTMLLoader } from './html-loader';
 export type { SAXParserOptions as SAXParserOptions } from './sax-ts/sax';
 export { SAXParser as SAXParser } from './sax-ts/sax';
+export { convertXMLValueToArray, convertXMLFieldToArrayInPlace } from './lib/xml-utils/xml-utils';
+export { uncapitalize as _uncapitalize, uncapitalizeKeys as _uncapitalizeKeys } from './lib/xml-utils/uncapitalize';
 //# sourceMappingURL=index.d.ts.map

package/dist/index.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAIA,YAAY,EAAC,gBAAgB,EAAC,MAAM,cAAc,CAAC;AACnD,OAAO,EAAC,SAAS,EAAC,MAAM,cAAc,CAAC;AAIvC,YAAY,EAAC,gBAAgB,IAAI,gBAAgB,EAAC,MAAM,cAAc,CAAC;AACvE,OAAO,EAAC,SAAS,IAAI,SAAS,EAAC,MAAM,cAAc,CAAC"}
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAIA,YAAY,EAAC,gBAAgB,EAAC,MAAM,cAAc,CAAC;AACnD,OAAO,EAAC,SAAS,EAAC,MAAM,cAAc,CAAC;AAIvC,YAAY,EAAC,iBAAiB,EAAC,MAAM,eAAe,CAAC;AACrD,OAAO,EAAC,UAAU,EAAC,MAAM,eAAe,CAAC;AAIzC,YAAY,EAAC,gBAAgB,IAAI,gBAAgB,EAAC,MAAM,cAAc,CAAC;AACvE,OAAO,EAAC,SAAS,IAAI,SAAS,EAAC,MAAM,cAAc,CAAC;AAIpD,OAAO,EAAC,sBAAsB,EAAE,6BAA6B,EAAC,MAAM,2BAA2B,CAAC;AAIhG,OAAO,EACL,YAAY,IAAI,aAAa,EAC7B,gBAAgB,IAAI,iBAAiB,EACtC,MAAM,8BAA8B,CAAC"}

package/dist/index.js

@@ -1,8 +1,18 @@
 "use strict";
 // loaders.gl, MIT license
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.SAXParser = exports.XMLLoader = void 0;
+exports._uncapitalizeKeys = exports._uncapitalize = exports.convertXMLFieldToArrayInPlace = exports.convertXMLValueToArray = exports.SAXParser = exports.HTMLLoader = exports.XMLLoader = void 0;
 var xml_loader_1 = require("./xml-loader");
 Object.defineProperty(exports, "XMLLoader", { enumerable: true, get: function () { return xml_loader_1.XMLLoader; } });
+var html_loader_1 = require("./html-loader");
+Object.defineProperty(exports, "HTMLLoader", { enumerable: true, get: function () { return html_loader_1.HTMLLoader; } });
 var sax_1 = require("./sax-ts/sax");
 Object.defineProperty(exports, "SAXParser", { enumerable: true, get: function () { return sax_1.SAXParser; } });
+// Utilities
+var xml_utils_1 = require("./lib/xml-utils/xml-utils");
+Object.defineProperty(exports, "convertXMLValueToArray", { enumerable: true, get: function () { return xml_utils_1.convertXMLValueToArray; } });
+Object.defineProperty(exports, "convertXMLFieldToArrayInPlace", { enumerable: true, get: function () { return xml_utils_1.convertXMLFieldToArrayInPlace; } });
+// Experimental
+var uncapitalize_1 = require("./lib/xml-utils/uncapitalize");
+Object.defineProperty(exports, "_uncapitalize", { enumerable: true, get: function () { return uncapitalize_1.uncapitalize; } });
+Object.defineProperty(exports, "_uncapitalizeKeys", { enumerable: true, get: function () { return uncapitalize_1.uncapitalizeKeys; } });

package/dist/lib/parsers/parse-xml.d.ts

@@ -0,0 +1,12 @@
+import type { X2jOptions } from 'fast-xml-parser';
+/** Type for passing through fast-xml-parser options */
+export type FastXMLParserOptions = Partial<X2jOptions>;
+export declare function fastParseXML(text: string, options: FastXMLParserOptions): any;
+/**
+ * @todo Build a streaming XML parser based on sax-js
+ * @param text
+ * @param options
+ * @returns
+ */
+export declare function parseXMLInBatches(text: string, options?: {}): any;
+//# sourceMappingURL=parse-xml.d.ts.map

package/dist/lib/parsers/parse-xml.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"parse-xml.d.ts","sourceRoot":"","sources":["../../../src/lib/parsers/parse-xml.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,EAAC,UAAU,EAAC,MAAM,iBAAiB,CAAC;AAEhD,uDAAuD;AACvD,MAAM,MAAM,oBAAoB,GAAG,OAAO,CAAC,UAAU,CAAC,CAAC;AAEvD,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,EAAE,oBAAoB,GAAG,GAAG,CAU7E;AAED;;;;;GAKG;AACH,wBAAgB,iBAAiB,CAAC,IAAI,EAAE,MAAM,EAAE,OAAO,KAAK,GAAG,GAAG,CAUjE"}

package/dist/lib/{parse-xml.js → parsers/parse-xml.js}

@@ -1,28 +1,27 @@
 "use strict";
 // loaders.gl, MIT license
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.parseXMLInBatches = exports.parseXML = void 0;
-const xml_parser_1 = require("../lib/parser/xml-parser");
+exports.parseXMLInBatches = exports.fastParseXML = void 0;
+const streaming_xml_parser_1 = require("./streaming-xml-parser");
 const fast_xml_parser_1 = require("fast-xml-parser");
-function parseXML(text, options) {
+function fastParseXML(text, options) {
     const parser = new fast_xml_parser_1.XMLParser({
         ignoreAttributes: false,
         attributeNamePrefix: '',
-        // parseAttributeValue: true,
-        ...options?.xml
+        ...options
     });
     const parsedXML = parser.parse(text);
     return parsedXML;
 }
-exports.parseXML = parseXML;
+exports.fastParseXML = fastParseXML;
 /**
  * @todo Build a streaming XML parser based on sax-js
  * @param text
  * @param options
  * @returns
  */
-function parseXMLInBatches(text, options) {
-    const parser = new xml_parser_1.XMLParser({
+function parseXMLInBatches(text, options = {}) {
+    const parser = new streaming_xml_parser_1.StreamingXMLParser({
         ...options,
         strict: true
     });

package/dist/lib/{parser/xml-parser.d.ts → parsers/streaming-xml-parser.d.ts}

@@ -1,5 +1,9 @@
 import { SAXParser, SAXParserOptions } from '../../sax-ts/sax';
-export declare class XMLParser {
+export type StreamingXMLParserOptions = SAXParserOptions;
+/**
+ * StreamingXMLParser builds a JSON object using the events emitted by the SAX parser
+ */
+export declare class StreamingXMLParser {
     readonly parser: SAXParser;
     result: undefined;
     previousStates: never[];
@@ -17,4 +21,4 @@ export declare class XMLParser {
     _openObject(newContainer?: {}): void;
     _closeObject(): void;
 }
-//# sourceMappingURL=xml-parser.d.ts.map
+//# sourceMappingURL=streaming-xml-parser.d.ts.map

package/dist/lib/parsers/streaming-xml-parser.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"streaming-xml-parser.d.ts","sourceRoot":"","sources":["../../../src/lib/parsers/streaming-xml-parser.ts"],"names":[],"mappings":"AAGA,OAAO,EAAC,SAAS,EAAE,gBAAgB,EAAC,MAAM,kBAAkB,CAAC;AAG7D,MAAM,MAAM,yBAAyB,GAAG,gBAAgB,CAAC;AAEzD;;GAEG;AACH,qBAAa,kBAAkB;IAC7B,QAAQ,CAAC,MAAM,EAAE,SAAS,CAAC;IAC3B,MAAM,YAAa;IACnB,cAAc,UAAM;IACpB,YAAY;;;OAA6C;gBAG7C,OAAO,EAAE,gBAAgB;IA+CrC,KAAK,IAAI,IAAI;IAMb,KAAK,CAAC,KAAK,KAAA,GAAG,IAAI;IAIlB,KAAK,IAAI,IAAI;IAMb,UAAU,CAAC,KAAK,KAAA,GAAG,IAAI;IAYvB,UAAU,CAAC,YAAY,UAAK,GAAG,IAAI;IAOnC,WAAW,IAAI,IAAI;IAKnB,WAAW,CAAC,YAAY,KAAK,GAAG,IAAI;IAOpC,YAAY,IAAI,IAAI;CAIrB"}

package/dist/lib/{parser/xml-parser.js → parsers/streaming-xml-parser.js}

@@ -2,12 +2,12 @@
 // @ts-nocheck
 /* eslint-disable */
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.XMLParser = void 0;
-// import ClarinetParser, {ClarinetParserOptions} from '../clarinet/clarinet';
+exports.StreamingXMLParser = void 0;
 const sax_1 = require("../../sax-ts/sax");
-// import JSONPath from '../jsonpath/jsonpath';
-// XMLParser builds a JSON object using the events emitted by the Clarinet parser
-class XMLParser {
+/**
+ * StreamingXMLParser builds a JSON object using the events emitted by the SAX parser
+ */
+class StreamingXMLParser {
     // jsonpath: JSONPath = new JSONPath();
     constructor(options) {
         this.result = undefined;
@@ -95,4 +95,4 @@ class XMLParser {
         this.currentState = this.previousStates.pop();
     }
 }
-exports.XMLParser = XMLParser;
+exports.StreamingXMLParser = StreamingXMLParser;

package/dist/lib/xml-utils/uncapitalize.d.ts

@@ -0,0 +1,13 @@
+/**
+ * Uncapitalize first letter of a string
+ * @param str
+ * @returns
+ */
+export declare function uncapitalize(str: string): string;
+/**
+ * Recursively uncapitalize all keys in a nested object
+ * @param object
+ * @returns
+ */
+export declare function uncapitalizeKeys(object: any): any;
+//# sourceMappingURL=uncapitalize.d.ts.map

package/dist/lib/xml-utils/uncapitalize.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"uncapitalize.d.ts","sourceRoot":"","sources":["../../../src/lib/xml-utils/uncapitalize.ts"],"names":[],"mappings":"AAEA;;;;GAIG;AACH,wBAAgB,YAAY,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAEhD;AAED;;;;GAIG;AACH,wBAAgB,gBAAgB,CAAC,MAAM,EAAE,GAAG,GAAG,GAAG,CAcjD"}

package/dist/lib/xml-utils/uncapitalize.js

@@ -0,0 +1,32 @@
+"use strict";
+// loaders.gl, MIT license
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.uncapitalizeKeys = exports.uncapitalize = void 0;
+/**
+ * Uncapitalize first letter of a string
+ * @param str
+ * @returns
+ */
+function uncapitalize(str) {
+    return typeof str === 'string' ? str.charAt(0).toLowerCase() + str.slice(1) : str;
+}
+exports.uncapitalize = uncapitalize;
+/**
+ * Recursively uncapitalize all keys in a nested object
+ * @param object
+ * @returns
+ */
+function uncapitalizeKeys(object) {
+    if (Array.isArray(object)) {
+        return object.map((element) => uncapitalizeKeys(element));
+    }
+    if (object && typeof object === 'object') {
+        const newObject = {};
+        for (const [key, value] of Object.entries(object)) {
+            newObject[uncapitalize(key)] = uncapitalizeKeys(value);
+        }
+        return newObject;
+    }
+    return object;
+}
+exports.uncapitalizeKeys = uncapitalizeKeys;

package/dist/lib/xml-utils/xml-utils.d.ts

@@ -0,0 +1,11 @@
+/**
+ * Extracts a value or array and always return an array
+ * Useful since XML parses to object instead of array when only a single value is provided
+ */
+export declare function convertXMLValueToArray(xmlValue: unknown): unknown[];
+/**
+ * Mutates a field in place, converting it to array
+ * Useful since XML parses to object instead of array when only a single value is provided
+ */
+export declare function convertXMLFieldToArrayInPlace(xml: any, key: string): void;
+//# sourceMappingURL=xml-utils.d.ts.map

package/dist/lib/xml-utils/xml-utils.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"xml-utils.d.ts","sourceRoot":"","sources":["../../../src/lib/xml-utils/xml-utils.ts"],"names":[],"mappings":"AAGA;;;GAGG;AACH,wBAAgB,sBAAsB,CAAC,QAAQ,EAAE,OAAO,GAAG,OAAO,EAAE,CAWnE;AAED;;;GAGG;AACH,wBAAgB,6BAA6B,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,MAAM,GAAG,IAAI,CAEzE"}

package/dist/lib/xml-utils/xml-utils.js

@@ -0,0 +1,30 @@
+"use strict";
+// TODO - these utilities could be moved to the XML parser.
+// uncapitalizeKeys could be an XMLLoader option
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.convertXMLFieldToArrayInPlace = exports.convertXMLValueToArray = void 0;
+/**
+ * Extracts a value or array and always return an array
+ * Useful since XML parses to object instead of array when only a single value is provided
+ */
+function convertXMLValueToArray(xmlValue) {
+    if (Array.isArray(xmlValue)) {
+        return xmlValue;
+    }
+    if (xmlValue && typeof xmlValue === 'object' && xmlValue['0']) {
+        // Error this is an objectified array
+    }
+    if (xmlValue) {
+        return [xmlValue];
+    }
+    return [];
+}
+exports.convertXMLValueToArray = convertXMLValueToArray;
+/**
+ * Mutates a field in place, converting it to array
+ * Useful since XML parses to object instead of array when only a single value is provided
+ */
+function convertXMLFieldToArrayInPlace(xml, key) {
+    xml[key] = convertXMLValueToArray(xml[key]);
+}
+exports.convertXMLFieldToArrayInPlace = convertXMLFieldToArrayInPlace;

package/dist/sax-ts/sax.js

@@ -1436,5 +1436,5 @@ class SAXParser extends SAX {
         this.flushBuffers();
     }
 }
-exports.SAXParser = SAXParser;
 SAXParser.ENTITIES = exports.ENTITIES;
+exports.SAXParser = SAXParser;

package/dist/xml-loader.d.ts

@@ -1,10 +1,22 @@
 import type { LoaderWithParser, LoaderOptions } from '@loaders.gl/loader-utils';
 import type { SAXParserOptions } from './sax-ts/sax';
+import { FastXMLParserOptions } from './lib/parsers/parse-xml';
 export type XMLLoaderOptions = LoaderOptions & {
-    xml?: SAXParserOptions;
+    xml?: {
+        parser?: 'fast-xml-parser' | string;
+        /** XML is typically PascalCase, JavaScript prefects camelCase */
+        uncapitalizeKeys?: boolean;
+        removeNSPrefix?: boolean;
+        textNodeName?: string;
+        arrayPaths?: string[];
+    };
+    /** @deprecated Experimental, passes options to fast-xml-parser, IF it is being used */
+    _fastXML?: FastXMLParserOptions;
+    /** @deprecated Experimental, passes options to the SAX XML parser, IF it is being used. */
+    _sax?: SAXParserOptions;
 };
 /**
- * Worker loader for the OBJ geometry format
+ * Loader for XML files
  */
 export declare const XMLLoader: {
     name: string;
@@ -16,7 +28,13 @@ export declare const XMLLoader: {
     mimeTypes: string[];
     testText: typeof testXMLFile;
     options: {
-        xml: {};
+        xml: {
+            parser: string;
+            uncapitalizeKeys: boolean;
+            removeNSPrefix: boolean;
+            textNodeName: string;
+            arrayPaths: never[];
+        };
     };
     parse: (arrayBuffer: ArrayBuffer, options?: XMLLoaderOptions) => Promise<any>;
     parseTextSync: (text: string, options?: XMLLoaderOptions) => any;

package/dist/xml-loader.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"xml-loader.d.ts","sourceRoot":"","sources":["../src/xml-loader.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAC,gBAAgB,EAAE,aAAa,EAAC,MAAM,0BAA0B,CAAC;AAC9E,OAAO,KAAK,EAAC,gBAAgB,EAAC,MAAM,cAAc,CAAC;AAOnD,MAAM,MAAM,gBAAgB,GAAG,aAAa,GAAG;IAC7C,GAAG,CAAC,EAAE,gBAAgB,CAAC;CACxB,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,SAAS;;;;;;;;;;;;yBAYO,WAAW,YAAY,gBAAgB;0BAE5C,MAAM,YAAY,gBAAgB;CACzD,CAAC;AAEF,iBAAS,WAAW,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAG1C;AAED,eAAO,MAAM,mBAAmB,EAAE,gBAA4B,CAAC"}
+{"version":3,"file":"xml-loader.d.ts","sourceRoot":"","sources":["../src/xml-loader.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAC,gBAAgB,EAAE,aAAa,EAAC,MAAM,0BAA0B,CAAC;AAC9E,OAAO,KAAK,EAAC,gBAAgB,EAAC,MAAM,cAAc,CAAC;AACnD,OAAO,EAAe,oBAAoB,EAAC,MAAM,yBAAyB,CAAC;AAO3E,MAAM,MAAM,gBAAgB,GAAG,aAAa,GAAG;IAC7C,GAAG,CAAC,EAAE;QAEJ,MAAM,CAAC,EAAE,iBAAiB,GAAG,MAAM,CAAC;QACpC,iEAAiE;QACjE,gBAAgB,CAAC,EAAE,OAAO,CAAC;QAC3B,cAAc,CAAC,EAAE,OAAO,CAAC;QACzB,YAAY,CAAC,EAAE,MAAM,CAAC;QACtB,UAAU,CAAC,EAAE,MAAM,EAAE,CAAC;KACvB,CAAC;IACF,uFAAuF;IACvF,QAAQ,CAAC,EAAE,oBAAoB,CAAC;IAChC,2FAA2F;IAC3F,IAAI,CAAC,EAAE,gBAAgB,CAAC;CACzB,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,SAAS;;;;;;;;;;;;;;;;;;yBAkBO,WAAW,YAAY,gBAAgB;0BAE5C,MAAM,YAAY,gBAAgB;CACzD,CAAC;AAEF,iBAAS,WAAW,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAG1C;AAwCD,eAAO,MAAM,mBAAmB,EAAE,gBAA4B,CAAC"}

package/dist/xml-loader.js

@@ -1,12 +1,13 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports._typecheckXMLLoader = exports.XMLLoader = void 0;
-const parse_xml_1 = require("./lib/parse-xml");
+const parse_xml_1 = require("./lib/parsers/parse-xml");
+const uncapitalize_1 = require("./lib/xml-utils/uncapitalize");
 // __VERSION__ is injected by babel-plugin-version-inline
 // @ts-ignore TS2304: Cannot find name '__VERSION__'.
 const VERSION = typeof __VERSION__ !== 'undefined' ? __VERSION__ : 'latest';
 /**
- * Worker loader for the OBJ geometry format
+ * Loader for XML files
  */
 exports.XMLLoader = {
     name: 'XML',
@@ -18,13 +19,48 @@ exports.XMLLoader = {
     mimeTypes: ['application/xml', 'text/xml'],
     testText: testXMLFile,
     options: {
-        xml: {}
+        xml: {
+            parser: 'fast-xml-parser',
+            uncapitalizeKeys: false,
+            removeNSPrefix: false,
+            textNodeName: 'value',
+            arrayPaths: []
+        }
     },
-    parse: async (arrayBuffer, options) => (0, parse_xml_1.parseXML)(new TextDecoder().decode(arrayBuffer), options),
-    parseTextSync: (text, options) => (0, parse_xml_1.parseXML)(text, options)
+    parse: async (arrayBuffer, options) => parseTextSync(new TextDecoder().decode(arrayBuffer), options),
+    parseTextSync: (text, options) => parseTextSync(text, options)
 };
 function testXMLFile(text) {
     // TODO - There could be space first.
     return text.startsWith('<?xml');
 }
+function parseTextSync(text, options) {
+    const xmlOptions = { ...exports.XMLLoader.options.xml, ...options?.xml };
+    switch (xmlOptions.parser) {
+        case 'fast-xml-parser':
+            const fastXMLOptions = {
+                // Default FastXML options
+                // https://github.com/NaturalIntelligence/fast-xml-parser/blob/master/docs/v4/2.XMLparseOptions.md#allowbooleanattributes
+                allowBooleanAttributes: true,
+                // https://github.com/NaturalIntelligence/fast-xml-parser/blob/master/docs/v4/2.XMLparseOptions.md#ignoredeclaration
+                ignoreDeclaration: true,
+                // XMLLoader Options: Map to FastXMLOptions
+                // https://github.com/NaturalIntelligence/fast-xml-parser/blob/master/docs/v4/2.XMLparseOptions.md#removensprefix
+                removeNSPrefix: xmlOptions.removeNSPrefix,
+                // https://github.com/NaturalIntelligence/fast-xml-parser/blob/master/docs/v4/2.XMLparseOptions.md#textnodename
+                textNodeName: xmlOptions.textNodeName,
+                isArray: (name, jpath, isLeafNode, isAttribute) => {
+                    const array = Boolean(xmlOptions?.arrayPaths?.some((path) => jpath === path));
+                    return array;
+                },
+                // Application overrides
+                ...options?._fastXML
+            };
+            const xml = (0, parse_xml_1.fastParseXML)(text, fastXMLOptions);
+            // Note - could be done with FastXML tag processing
+            return xmlOptions.uncapitalizeKeys ? (0, uncapitalize_1.uncapitalizeKeys)(xml) : xml;
+        default:
+            throw new Error(options?.xml?.parser);
+    }
+}
 exports._typecheckXMLLoader = exports.XMLLoader;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@loaders.gl/xml",
-  "version": "3.4.0-alpha.1",
+  "version": "3.4.0-alpha.3",
   "description": "Framework-independent loaders for the XML (eXtensible Markup Language) format",
   "license": "MIT",
   "publishConfig": {
@@ -33,12 +33,12 @@
   },
   "dependencies": {
     "@babel/runtime": "^7.3.1",
-    "@loaders.gl/loader-utils": "3.4.0-alpha.1",
-    "@loaders.gl/schema": "3.4.0-alpha.1",
-    "fast-xml-parser": "^4.0.11"
+    "@loaders.gl/loader-utils": "3.4.0-alpha.3",
+    "@loaders.gl/schema": "3.4.0-alpha.3",
+    "fast-xml-parser": "^4.1.3"
   },
   "devDependencies": {
     "xmldom": "0.6.0"
   },
-  "gitHead": "4085b0323050e4361614471319a1fb4729547bbf"
+  "gitHead": "a954528dd1d78a1f128d8f6b07e4baeb7a296924"
 }

package/src/html-loader.ts

@@ -0,0 +1,46 @@
+// loaders.gl, MIT license
+
+import type {LoaderWithParser} from '@loaders.gl/loader-utils';
+import {mergeLoaderOptions} from '@loaders.gl/loader-utils';
+import {XMLLoader, XMLLoaderOptions} from './xml-loader';
+
+export type HTMLLoaderOptions = XMLLoaderOptions;
+
+/**
+ * Loader for HTML files
+ * Essentially a copy of the XMLLoader with different mime types, file extensions and content tests.
+ * This split enables applications can control whether they want HTML responses to be parsed by the XML loader or not.
+ * This loader does not have any additional understanding of the structure of HTML or the document.
+ */
+export const HTMLLoader: LoaderWithParser = {
+  ...XMLLoader,
+  name: 'HTML',
+  id: 'html',
+  extensions: ['html', 'htm'],
+  mimeTypes: ['text/html'],
+  testText: testHTMLFile,
+  parse: async (arrayBuffer: ArrayBuffer, options?: XMLLoaderOptions) =>
+    parseTextSync(new TextDecoder().decode(arrayBuffer), options),
+  parseTextSync: (text: string, options?: XMLLoaderOptions) => parseTextSync(text, options)
+};
+
+function testHTMLFile(text: string): boolean {
+  // TODO - There could be space first.
+  return text.startsWith('<html');
+}
+
+function parseTextSync(text: string, options?: XMLLoaderOptions): any {
+  // fast-xml-parser can recognize HTML entities
+  // https://github.com/NaturalIntelligence/fast-xml-parser/blob/master/docs/v4/2.XMLparseOptions.md#htmlentities
+  // https://github.com/NaturalIntelligence/fast-xml-parser/blob/master/docs/v4/5.Entities.md
+  options = mergeLoaderOptions(options, {
+    xml: {
+      parser: 'fast-xml-parser'
+    },
+    _fastXML: {
+      htmlEntities: true
+    }
+  });
+
+  return XMLLoader.parseTextSync(text, options);
+}

package/src/index.ts

@@ -5,7 +5,23 @@
 export type {XMLLoaderOptions} from './xml-loader';
 export {XMLLoader} from './xml-loader';
 
+// HTMLLoader
+
+export type {HTMLLoaderOptions} from './html-loader';
+export {HTMLLoader} from './html-loader';
+
 // SAX
 
 export type {SAXParserOptions as SAXParserOptions} from './sax-ts/sax';
 export {SAXParser as SAXParser} from './sax-ts/sax';
+
+// Utilities
+
+export {convertXMLValueToArray, convertXMLFieldToArrayInPlace} from './lib/xml-utils/xml-utils';
+
+// Experimental
+
+export {
+  uncapitalize as _uncapitalize,
+  uncapitalizeKeys as _uncapitalizeKeys
+} from './lib/xml-utils/uncapitalize';

package/src/lib/{parse-xml.ts → parsers/parse-xml.ts}

@@ -1,14 +1,17 @@
 // loaders.gl, MIT license
 
-import {XMLParser} from '../lib/parser/xml-parser';
+import {StreamingXMLParser} from './streaming-xml-parser';
 import {XMLParser as FastXMLParser} from 'fast-xml-parser';
+import type {X2jOptions} from 'fast-xml-parser';
 
-export function parseXML(text: string, options): any {
+/** Type for passing through fast-xml-parser options */
+export type FastXMLParserOptions = Partial<X2jOptions>;
+
+export function fastParseXML(text: string, options: FastXMLParserOptions): any {
   const parser = new FastXMLParser({
     ignoreAttributes: false,
     attributeNamePrefix: '',
-    // parseAttributeValue: true,
-    ...options?.xml
+    ...options
   });
 
   const parsedXML = parser.parse(text);
@@ -22,8 +25,8 @@ export function parseXML(text: string, options): any {
  * @param options
  * @returns
  */
-export function parseXMLInBatches(text: string, options): any {
-  const parser = new XMLParser({
+export function parseXMLInBatches(text: string, options = {}): any {
+  const parser = new StreamingXMLParser({
     ...options,
     strict: true
   });

package/src/lib/{parser/xml-parser.ts → parsers/streaming-xml-parser.ts}

@@ -1,13 +1,15 @@
 // @ts-nocheck
 /* eslint-disable */
 
-// import ClarinetParser, {ClarinetParserOptions} from '../clarinet/clarinet';
 import {SAXParser, SAXParserOptions} from '../../sax-ts/sax';
 // import JSONPath from '../jsonpath/jsonpath';
 
-// XMLParser builds a JSON object using the events emitted by the Clarinet parser
+export type StreamingXMLParserOptions = SAXParserOptions;
 
-export class XMLParser {
+/**
+ * StreamingXMLParser builds a JSON object using the events emitted by the SAX parser
+ */
+export class StreamingXMLParser {
   readonly parser: SAXParser;
   result = undefined;
   previousStates = [];

package/src/lib/xml-utils/uncapitalize.ts

@@ -0,0 +1,31 @@
+// loaders.gl, MIT license
+
+/**
+ * Uncapitalize first letter of a string
+ * @param str
+ * @returns
+ */
+export function uncapitalize(str: string): string {
+  return typeof str === 'string' ? str.charAt(0).toLowerCase() + str.slice(1) : str;
+}
+
+/**
+ * Recursively uncapitalize all keys in a nested object
+ * @param object
+ * @returns
+ */
+export function uncapitalizeKeys(object: any): any {
+  if (Array.isArray(object)) {
+    return object.map((element) => uncapitalizeKeys(element));
+  }
+
+  if (object && typeof object === 'object') {
+    const newObject = {};
+    for (const [key, value] of Object.entries(object)) {
+      newObject[uncapitalize(key)] = uncapitalizeKeys(value);
+    }
+    return newObject;
+  }
+
+  return object;
+}

package/src/lib/xml-utils/xml-utils.ts

@@ -0,0 +1,27 @@
+// TODO - these utilities could be moved to the XML parser.
+// uncapitalizeKeys could be an XMLLoader option
+
+/**
+ * Extracts a value or array and always return an array
+ * Useful since XML parses to object instead of array when only a single value is provided
+ */
+export function convertXMLValueToArray(xmlValue: unknown): unknown[] {
+  if (Array.isArray(xmlValue)) {
+    return xmlValue;
+  }
+  if (xmlValue && typeof xmlValue === 'object' && xmlValue['0']) {
+    // Error this is an objectified array
+  }
+  if (xmlValue) {
+    return [xmlValue];
+  }
+  return [];
+}
+
+/**
+ * Mutates a field in place, converting it to array
+ * Useful since XML parses to object instead of array when only a single value is provided
+ */
+export function convertXMLFieldToArrayInPlace(xml: any, key: string): void {
+  xml[key] = convertXMLValueToArray(xml[key]);
+}