read-excel-file 9.2.0 → 9.3.1

package/modules/xlsx/parseStyles.js

@@ -1,56 +1,108 @@
 function _typeof(o) { "@babel/helpers - typeof"; return _typeof = "function" == typeof Symbol && "symbol" == typeof Symbol.iterator ? function (o) { return typeof o; } : function (o) { return o && "function" == typeof Symbol && o.constructor === Symbol && o !== Symbol.prototype ? "symbol" : typeof o; }, _typeof(o); }
+var _excluded = ["xfId"];
 function ownKeys(e, r) { var t = Object.keys(e); if (Object.getOwnPropertySymbols) { var o = Object.getOwnPropertySymbols(e); r && (o = o.filter(function (r) { return Object.getOwnPropertyDescriptor(e, r).enumerable; })), t.push.apply(t, o); } return t; }
 function _objectSpread(e) { for (var r = 1; r < arguments.length; r++) { var t = null != arguments[r] ? arguments[r] : {}; r % 2 ? ownKeys(Object(t), !0).forEach(function (r) { _defineProperty(e, r, t[r]); }) : Object.getOwnPropertyDescriptors ? Object.defineProperties(e, Object.getOwnPropertyDescriptors(t)) : ownKeys(Object(t)).forEach(function (r) { Object.defineProperty(e, r, Object.getOwnPropertyDescriptor(t, r)); }); } return e; }
 function _defineProperty(obj, key, value) { key = _toPropertyKey(key); if (key in obj) { Object.defineProperty(obj, key, { value: value, enumerable: true, configurable: true, writable: true }); } else { obj[key] = value; } return obj; }
 function _toPropertyKey(arg) { var key = _toPrimitive(arg, "string"); return _typeof(key) === "symbol" ? key : String(key); }
 function _toPrimitive(input, hint) { if (_typeof(input) !== "object" || input === null) return input; var prim = input[Symbol.toPrimitive]; if (prim !== undefined) { var res = prim.call(input, hint || "default"); if (_typeof(res) !== "object") return res; throw new TypeError("@@toPrimitive must return a primitive value."); } return (hint === "string" ? String : Number)(input); }
-import { getBaseStyles, getCellStyles, getNumberFormats } from '../xml/xlsx.js';
-
-// http://officeopenxml.com/SSstyles.php
-// Returns an array of cell styles.
-// A cell style index is its ID.
-export default function parseStyles(content, xml) {
-  if (!content) {
-    return {};
-  }
-
+function _objectWithoutProperties(source, excluded) { if (source == null) return {}; var target = _objectWithoutPropertiesLoose(source, excluded); var key, i; if (Object.getOwnPropertySymbols) { var sourceSymbolKeys = Object.getOwnPropertySymbols(source); for (i = 0; i < sourceSymbolKeys.length; i++) { key = sourceSymbolKeys[i]; if (excluded.indexOf(key) >= 0) continue; if (!Object.prototype.propertyIsEnumerable.call(source, key)) continue; target[key] = source[key]; } } return target; }
+function _objectWithoutPropertiesLoose(source, excluded) { if (source == null) return {}; var target = {}; var sourceKeys = Object.keys(source); var key, i; for (i = 0; i < sourceKeys.length; i++) { key = sourceKeys[i]; if (excluded.indexOf(key) >= 0) continue; target[key] = source[key]; } return target; }
+/**
+ * Parses `.xlsx` file styles.
+ * http://officeopenxml.com/SSstyles.php
+ * Returns an array of cell styles.
+ * A cell style index is the cell style ID.
+ * @param {string} content
+ * @param  {function} parseXmlStream — SAX XML parser.
+ * @returns {object} styles
+ */
+export default function parseStyles(content, parseXmlStream) {
   // https://social.msdn.microsoft.com/Forums/sqlserver/en-US/708978af-b598-45c4-a598-d3518a5a09f0/howwhen-is-cellstylexfs-vs-cellxfs-applied-to-a-cell?forum=os_binaryfile
   // https://www.office-forums.com/threads/cellxfs-cellstylexfs.2163519/
-  var doc = xml.createDocument(content);
-  var baseStyles = getBaseStyles(doc).map(parseCellStyle);
-  var numberFormats = getNumberFormats(doc).map(parseNumberFormatStyle).reduce(function (formats, format) {
-    // Format ID is a numeric index.
-    // There're some standard "built-in" formats (in Excel) up to about `100`.
-    formats[format.id] = format;
-    return formats;
-  }, []);
-  var getCellStyle = function getCellStyle(xf) {
-    if (xf.hasAttribute('xfId')) {
-      return _objectSpread(_objectSpread({}, baseStyles[xf.xfId]), parseCellStyle(xf, numberFormats));
-    }
-    return parseCellStyle(xf, numberFormats);
-  };
-  return getCellStyles(doc).map(getCellStyle);
+  return parseXmlStream(content, {
+    createInitialState: createInitialStateInStyles,
+    onOpenTag: onOpenTagInStyles,
+    onCloseTag: onCloseTagInStyles
+  }).then(function (state) {
+    return state.styles.map(function (style) {
+      if (style.xfId) {
+        var xfId = style.xfId,
+          styleProperties = _objectWithoutProperties(style, _excluded);
+        return _objectSpread(_objectSpread({}, state.baseStyles[xfId]), styleProperties);
+      } else {
+        return style;
+      }
+    });
+  });
 }
-function parseNumberFormatStyle(numFmt) {
+function createInitialStateInStyles() {
   return {
-    id: numFmt.getAttribute('numFmtId'),
-    template: numFmt.getAttribute('formatCode')
+    numberFormats: {},
+    baseStyles: [],
+    // The first `164` elements of the `styles` array are going to be `undefined`
+    // because those represent the built-in "default" styles in `.xlsx` specification.
+    // These "default" styles have IDs from `0` to `163`, i.e. according to their index.
+    styles: [],
+    cellStyleXfs: false,
+    cellXfs: false
   };
 }
-
-// http://www.datypic.com/sc/ooxml/e-ssml_xf-2.html
-function parseCellStyle(xf, numFmts) {
+function onOpenTagInStyles(tagName, attributes, state) {
+  if (tagName === 'numFmt') {
+    // `<numFmts/>` element contains `<numFmt/>` elements.
+    // `<numFmt/>` defines a "number format" template.
+    // The value of `numFmtId` attribute is a (numeric) ID of this "number format".
+    state.numberFormats[attributes.numFmtId] = {
+      id: Number(attributes.numFmtId),
+      template: attributes.formatCode
+    };
+  } else if (tagName === 'cellStyleXfs') {
+    // `<cellStyleXfs/>` contains "base styles" that could be extended by `<cellXfs/>`.
+    // These styles don't apply to individual cells directly. They can only be extended.
+    state.cellStyleXfs = true;
+  } else if (tagName === 'cellXfs') {
+    // `<cellXfs/>` contains individual cell styles. They can extend "base styles".
+    state.cellXfs = true;
+  } else if (tagName === 'xf') {
+    if (state.cellStyleXfs) {
+      // The index of a `<xf/>` element is by definition its numeric ID.
+      // Therefore, even empty `style` elements still must be put into the array.
+      state.baseStyles.push(parseCellStyle(attributes));
+    } else if (state.cellXfs) {
+      var style = parseCellStyle(attributes, state.numberFormats);
+      // A style might extend a "base style" by its numeric ID.
+      if (attributes.xfId) {
+        style.xfId = Number(attributes.xfId);
+      }
+      // The index of a `<xf/>` element is by definition its numeric ID.
+      // Therefore, even empty `style` elements still must be put into the array.
+      state.styles.push(style);
+    }
+  }
+}
+function onCloseTagInStyles(tagName, state) {
+  if (tagName === 'cellStyleXfs') {
+    state.cellStyleXfs = false;
+  } else if (tagName === 'cellXfs') {
+    state.cellXfs = false;
+  }
+}
+function parseCellStyle(attributes, numberFormats) {
   var style = {};
-  if (xf.hasAttribute('numFmtId')) {
-    var numberFormatId = xf.getAttribute('numFmtId');
+  // `numFmtId` in `<cellStyleXfs/>` and `<cellXfs/>` is a numerical ID
+  // referencing the cell's "number format". Values from `0` to `163` are built-in formats,
+  // while values `164` and above are custom formats defined in the `<numFmts/>` element.
+  var numFmtId = attributes.numFmtId;
+  if (numFmtId) {
     // Built-in number formats don't have a `<numFmt/>` element in `styles.xml`.
     // https://hexdocs.pm/xlsxir/number_styles.html
-    if (numFmts[numberFormatId]) {
-      style.numberFormat = numFmts[numberFormatId];
+    if (numberFormats && numberFormats[numFmtId]) {
+      // Non-built-in number format.
+      style.numberFormat = numberFormats[numFmtId];
     } else {
+      // Built-in number format.
       style.numberFormat = {
-        id: numberFormatId
+        id: Number(numFmtId)
       };
     }
   }

package/modules/xlsx/parseStyles.js.map

@@ -1 +1 @@
-{"version":3,"file":"parseStyles.js","names":["getBaseStyles","getCellStyles","getNumberFormats","parseStyles","content","xml","doc","createDocument","baseStyles","map","parseCellStyle","numberFormats","parseNumberFormatStyle","reduce","formats","format","id","getCellStyle","xf","hasAttribute","_objectSpread","xfId","numFmt","getAttribute","template","numFmts","style","numberFormatId","numberFormat"],"sources":["../../source/xlsx/parseStyles.js"],"sourcesContent":["import {\r\n  getBaseStyles,\r\n  getCellStyles,\r\n  getNumberFormats\r\n} from '../xml/xlsx.js'\r\n\r\n// http://officeopenxml.com/SSstyles.php\r\n// Returns an array of cell styles.\r\n// A cell style index is its ID.\r\nexport default function parseStyles(content, xml) {\r\n  if (!content) {\r\n    return {}\r\n  }\r\n\r\n  // https://social.msdn.microsoft.com/Forums/sqlserver/en-US/708978af-b598-45c4-a598-d3518a5a09f0/howwhen-is-cellstylexfs-vs-cellxfs-applied-to-a-cell?forum=os_binaryfile\r\n  // https://www.office-forums.com/threads/cellxfs-cellstylexfs.2163519/\r\n  const doc = xml.createDocument(content)\r\n\r\n  const baseStyles = getBaseStyles(doc)\r\n    .map(parseCellStyle)\r\n\r\n  const numberFormats = getNumberFormats(doc)\r\n    .map(parseNumberFormatStyle)\r\n    .reduce((formats, format) => {\r\n      // Format ID is a numeric index.\r\n      // There're some standard \"built-in\" formats (in Excel) up to about `100`.\r\n      formats[format.id] = format\r\n      return formats\r\n    }, [])\r\n\r\n  const getCellStyle = (xf) => {\r\n    if (xf.hasAttribute('xfId')) {\r\n      return {\r\n        ...baseStyles[xf.xfId],\r\n        ...parseCellStyle(xf, numberFormats)\r\n      }\r\n    }\r\n    return parseCellStyle(xf, numberFormats)\r\n  }\r\n\r\n  return getCellStyles(doc).map(getCellStyle)\r\n}\r\n\r\nfunction parseNumberFormatStyle(numFmt) {\r\n  return {\r\n    id: numFmt.getAttribute('numFmtId'),\r\n    template: numFmt.getAttribute('formatCode')\r\n  }\r\n}\r\n\r\n// http://www.datypic.com/sc/ooxml/e-ssml_xf-2.html\r\nfunction parseCellStyle(xf, numFmts) {\r\n  const style = {}\r\n  if (xf.hasAttribute('numFmtId')) {\r\n    const numberFormatId = xf.getAttribute('numFmtId')\r\n    // Built-in number formats don't have a `<numFmt/>` element in `styles.xml`.\r\n    // https://hexdocs.pm/xlsxir/number_styles.html\r\n    if (numFmts[numberFormatId]) {\r\n      style.numberFormat = numFmts[numberFormatId]\r\n    } else {\r\n      style.numberFormat = { id: numberFormatId }\r\n    }\r\n  }\r\n  return style\r\n}"],"mappings":";;;;;;AAAA,SACEA,aAAa,EACbC,aAAa,EACbC,gBAAgB,QACX,gBAAgB;;AAEvB;AACA;AACA;AACA,eAAe,SAASC,WAAWA,CAACC,OAAO,EAAEC,GAAG,EAAE;EAChD,IAAI,CAACD,OAAO,EAAE;IACZ,OAAO,CAAC,CAAC;EACX;;EAEA;EACA;EACA,IAAME,GAAG,GAAGD,GAAG,CAACE,cAAc,CAACH,OAAO,CAAC;EAEvC,IAAMI,UAAU,GAAGR,aAAa,CAACM,GAAG,CAAC,CAClCG,GAAG,CAACC,cAAc,CAAC;EAEtB,IAAMC,aAAa,GAAGT,gBAAgB,CAACI,GAAG,CAAC,CACxCG,GAAG,CAACG,sBAAsB,CAAC,CAC3BC,MAAM,CAAC,UAACC,OAAO,EAAEC,MAAM,EAAK;IAC3B;IACA;IACAD,OAAO,CAACC,MAAM,CAACC,EAAE,CAAC,GAAGD,MAAM;IAC3B,OAAOD,OAAO;EAChB,CAAC,EAAE,EAAE,CAAC;EAER,IAAMG,YAAY,GAAG,SAAfA,YAAYA,CAAIC,EAAE,EAAK;IAC3B,IAAIA,EAAE,CAACC,YAAY,CAAC,MAAM,CAAC,EAAE;MAC3B,OAAAC,aAAA,CAAAA,aAAA,KACKZ,UAAU,CAACU,EAAE,CAACG,IAAI,CAAC,GACnBX,cAAc,CAACQ,EAAE,EAAEP,aAAa,CAAC;IAExC;IACA,OAAOD,cAAc,CAACQ,EAAE,EAAEP,aAAa,CAAC;EAC1C,CAAC;EAED,OAAOV,aAAa,CAACK,GAAG,CAAC,CAACG,GAAG,CAACQ,YAAY,CAAC;AAC7C;AAEA,SAASL,sBAAsBA,CAACU,MAAM,EAAE;EACtC,OAAO;IACLN,EAAE,EAAEM,MAAM,CAACC,YAAY,CAAC,UAAU,CAAC;IACnCC,QAAQ,EAAEF,MAAM,CAACC,YAAY,CAAC,YAAY;EAC5C,CAAC;AACH;;AAEA;AACA,SAASb,cAAcA,CAACQ,EAAE,EAAEO,OAAO,EAAE;EACnC,IAAMC,KAAK,GAAG,CAAC,CAAC;EAChB,IAAIR,EAAE,CAACC,YAAY,CAAC,UAAU,CAAC,EAAE;IAC/B,IAAMQ,cAAc,GAAGT,EAAE,CAACK,YAAY,CAAC,UAAU,CAAC;IAClD;IACA;IACA,IAAIE,OAAO,CAACE,cAAc,CAAC,EAAE;MAC3BD,KAAK,CAACE,YAAY,GAAGH,OAAO,CAACE,cAAc,CAAC;IAC9C,CAAC,MAAM;MACLD,KAAK,CAACE,YAAY,GAAG;QAAEZ,EAAE,EAAEW;MAAe,CAAC;IAC7C;EACF;EACA,OAAOD,KAAK;AACd"}
+{"version":3,"file":"parseStyles.js","names":["parseStyles","content","parseXmlStream","createInitialState","createInitialStateInStyles","onOpenTag","onOpenTagInStyles","onCloseTag","onCloseTagInStyles","then","state","styles","map","style","xfId","styleProperties","_objectWithoutProperties","_excluded","_objectSpread","baseStyles","numberFormats","cellStyleXfs","cellXfs","tagName","attributes","numFmtId","id","Number","template","formatCode","push","parseCellStyle","numberFormat"],"sources":["../../source/xlsx/parseStyles.js"],"sourcesContent":["/**\r\n * Parses `.xlsx` file styles.\r\n * http://officeopenxml.com/SSstyles.php\r\n * Returns an array of cell styles.\r\n * A cell style index is the cell style ID.\r\n * @param {string} content\r\n * @param  {function} parseXmlStream — SAX XML parser.\r\n * @returns {object} styles\r\n */\r\nexport default function parseStyles(content, parseXmlStream) {\r\n  // https://social.msdn.microsoft.com/Forums/sqlserver/en-US/708978af-b598-45c4-a598-d3518a5a09f0/howwhen-is-cellstylexfs-vs-cellxfs-applied-to-a-cell?forum=os_binaryfile\r\n  // https://www.office-forums.com/threads/cellxfs-cellstylexfs.2163519/\r\n  return parseXmlStream(content, {\r\n    createInitialState: createInitialStateInStyles,\r\n    onOpenTag: onOpenTagInStyles,\r\n    onCloseTag: onCloseTagInStyles\r\n  }).then((state) => {\r\n    return state.styles.map((style) => {\r\n      if (style.xfId) {\r\n        const { xfId, ...styleProperties } = style\r\n        return {\r\n          ...state.baseStyles[xfId],\r\n          ...styleProperties\r\n        }\r\n      } else {\r\n        return style\r\n      }\r\n    })\r\n  })\r\n}\r\n\r\nfunction createInitialStateInStyles() {\r\n  return {\r\n    numberFormats: {},\r\n    baseStyles: [],\r\n    // The first `164` elements of the `styles` array are going to be `undefined`\r\n    // because those represent the built-in \"default\" styles in `.xlsx` specification.\r\n    // These \"default\" styles have IDs from `0` to `163`, i.e. according to their index.\r\n    styles: [],\r\n    cellStyleXfs: false,\r\n    cellXfs: false\r\n  }\r\n}\r\n\r\nfunction onOpenTagInStyles(tagName, attributes, state) {\r\n  if (tagName === 'numFmt') {\r\n    // `<numFmts/>` element contains `<numFmt/>` elements.\r\n    // `<numFmt/>` defines a \"number format\" template.\r\n    // The value of `numFmtId` attribute is a (numeric) ID of this \"number format\".\r\n    state.numberFormats[attributes.numFmtId] = {\r\n      id: Number(attributes.numFmtId),\r\n      template: attributes.formatCode\r\n    }\r\n  } else if (tagName === 'cellStyleXfs') {\r\n    // `<cellStyleXfs/>` contains \"base styles\" that could be extended by `<cellXfs/>`.\r\n    // These styles don't apply to individual cells directly. They can only be extended.\r\n    state.cellStyleXfs = true\r\n  } else if (tagName === 'cellXfs') {\r\n    // `<cellXfs/>` contains individual cell styles. They can extend \"base styles\".\r\n    state.cellXfs = true\r\n  } else if (tagName === 'xf') {\r\n    if (state.cellStyleXfs) {\r\n      // The index of a `<xf/>` element is by definition its numeric ID.\r\n      // Therefore, even empty `style` elements still must be put into the array.\r\n      state.baseStyles.push(parseCellStyle(attributes))\r\n    } else if (state.cellXfs) {\r\n      const style = parseCellStyle(attributes, state.numberFormats)\r\n      // A style might extend a \"base style\" by its numeric ID.\r\n      if (attributes.xfId) {\r\n        style.xfId = Number(attributes.xfId)\r\n      }\r\n      // The index of a `<xf/>` element is by definition its numeric ID.\r\n      // Therefore, even empty `style` elements still must be put into the array.\r\n      state.styles.push(style)\r\n    }\r\n  }\r\n}\r\n\r\nfunction onCloseTagInStyles(tagName, state) {\r\n  if (tagName === 'cellStyleXfs') {\r\n    state.cellStyleXfs = false\r\n  } else if (tagName === 'cellXfs') {\r\n    state.cellXfs = false\r\n  }\r\n}\r\n\r\nfunction parseCellStyle(attributes, numberFormats) {\r\n  const style = {}\r\n  // `numFmtId` in `<cellStyleXfs/>` and `<cellXfs/>` is a numerical ID\r\n  // referencing the cell's \"number format\". Values from `0` to `163` are built-in formats,\r\n  // while values `164` and above are custom formats defined in the `<numFmts/>` element.\r\n  const { numFmtId } = attributes\r\n  if (numFmtId) {\r\n    // Built-in number formats don't have a `<numFmt/>` element in `styles.xml`.\r\n    // https://hexdocs.pm/xlsxir/number_styles.html\r\n    if (numberFormats && numberFormats[numFmtId]) {\r\n      // Non-built-in number format.\r\n      style.numberFormat = numberFormats[numFmtId]\r\n    } else {\r\n      // Built-in number format.\r\n      style.numberFormat = { id: Number(numFmtId) }\r\n    }\r\n  }\r\n  return style\r\n}"],"mappings":";;;;;;;;;AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA,eAAe,SAASA,WAAWA,CAACC,OAAO,EAAEC,cAAc,EAAE;EAC3D;EACA;EACA,OAAOA,cAAc,CAACD,OAAO,EAAE;IAC7BE,kBAAkB,EAAEC,0BAA0B;IAC9CC,SAAS,EAAEC,iBAAiB;IAC5BC,UAAU,EAAEC;EACd,CAAC,CAAC,CAACC,IAAI,CAAC,UAACC,KAAK,EAAK;IACjB,OAAOA,KAAK,CAACC,MAAM,CAACC,GAAG,CAAC,UAACC,KAAK,EAAK;MACjC,IAAIA,KAAK,CAACC,IAAI,EAAE;QACd,IAAQA,IAAI,GAAyBD,KAAK,CAAlCC,IAAI;UAAKC,eAAe,GAAAC,wBAAA,CAAKH,KAAK,EAAAI,SAAA;QAC1C,OAAAC,aAAA,CAAAA,aAAA,KACKR,KAAK,CAACS,UAAU,CAACL,IAAI,CAAC,GACtBC,eAAe;MAEtB,CAAC,MAAM;QACL,OAAOF,KAAK;MACd;IACF,CAAC,CAAC;EACJ,CAAC,CAAC;AACJ;AAEA,SAAST,0BAA0BA,CAAA,EAAG;EACpC,OAAO;IACLgB,aAAa,EAAE,CAAC,CAAC;IACjBD,UAAU,EAAE,EAAE;IACd;IACA;IACA;IACAR,MAAM,EAAE,EAAE;IACVU,YAAY,EAAE,KAAK;IACnBC,OAAO,EAAE;EACX,CAAC;AACH;AAEA,SAAShB,iBAAiBA,CAACiB,OAAO,EAAEC,UAAU,EAAEd,KAAK,EAAE;EACrD,IAAIa,OAAO,KAAK,QAAQ,EAAE;IACxB;IACA;IACA;IACAb,KAAK,CAACU,aAAa,CAACI,UAAU,CAACC,QAAQ,CAAC,GAAG;MACzCC,EAAE,EAAEC,MAAM,CAACH,UAAU,CAACC,QAAQ,CAAC;MAC/BG,QAAQ,EAAEJ,UAAU,CAACK;IACvB,CAAC;EACH,CAAC,MAAM,IAAIN,OAAO,KAAK,cAAc,EAAE;IACrC;IACA;IACAb,KAAK,CAACW,YAAY,GAAG,IAAI;EAC3B,CAAC,MAAM,IAAIE,OAAO,KAAK,SAAS,EAAE;IAChC;IACAb,KAAK,CAACY,OAAO,GAAG,IAAI;EACtB,CAAC,MAAM,IAAIC,OAAO,KAAK,IAAI,EAAE;IAC3B,IAAIb,KAAK,CAACW,YAAY,EAAE;MACtB;MACA;MACAX,KAAK,CAACS,UAAU,CAACW,IAAI,CAACC,cAAc,CAACP,UAAU,CAAC,CAAC;IACnD,CAAC,MAAM,IAAId,KAAK,CAACY,OAAO,EAAE;MACxB,IAAMT,KAAK,GAAGkB,cAAc,CAACP,UAAU,EAAEd,KAAK,CAACU,aAAa,CAAC;MAC7D;MACA,IAAII,UAAU,CAACV,IAAI,EAAE;QACnBD,KAAK,CAACC,IAAI,GAAGa,MAAM,CAACH,UAAU,CAACV,IAAI,CAAC;MACtC;MACA;MACA;MACAJ,KAAK,CAACC,MAAM,CAACmB,IAAI,CAACjB,KAAK,CAAC;IAC1B;EACF;AACF;AAEA,SAASL,kBAAkBA,CAACe,OAAO,EAAEb,KAAK,EAAE;EAC1C,IAAIa,OAAO,KAAK,cAAc,EAAE;IAC9Bb,KAAK,CAACW,YAAY,GAAG,KAAK;EAC5B,CAAC,MAAM,IAAIE,OAAO,KAAK,SAAS,EAAE;IAChCb,KAAK,CAACY,OAAO,GAAG,KAAK;EACvB;AACF;AAEA,SAASS,cAAcA,CAACP,UAAU,EAAEJ,aAAa,EAAE;EACjD,IAAMP,KAAK,GAAG,CAAC,CAAC;EAChB;EACA;EACA;EACA,IAAQY,QAAQ,GAAKD,UAAU,CAAvBC,QAAQ;EAChB,IAAIA,QAAQ,EAAE;IACZ;IACA;IACA,IAAIL,aAAa,IAAIA,aAAa,CAACK,QAAQ,CAAC,EAAE;MAC5C;MACAZ,KAAK,CAACmB,YAAY,GAAGZ,aAAa,CAACK,QAAQ,CAAC;IAC9C,CAAC,MAAM;MACL;MACAZ,KAAK,CAACmB,YAAY,GAAG;QAAEN,EAAE,EAAEC,MAAM,CAACF,QAAQ;MAAE,CAAC;IAC/C;EACF;EACA,OAAOZ,KAAK;AACd"}

package/modules/xml/dom.js

@@ -77,7 +77,8 @@ export function getFirstElementChild(element) {
   }
 }
 
-// This function is only used for occasional debug messages.
+// This function was only used for occasional debug messages during development.
+// It's not used in the actual code.
 export function getOuterXml(node) {
   // `nodeType: 1` means "Element".
   // https://www.w3schools.com/xml/prop_element_nodetype.asp

package/modules/xml/dom.js.map

@@ -1 +1 @@
-{"version":3,"file":"dom.js","names":["findChild","node","tagName","i","childNodes","length","childNode","nodeType","getTagName","findChildren","results","push","forEach","func","map","NAMESPACE_REG_EXP","element","replace","isElement","getFirstElementChild","getOuterXml","textContent","xml","j","attributes","name","value"],"sources":["../../source/xml/dom.js"],"sourcesContent":["export function findChild(node, tagName) {\r\n\tlet i = 0\r\n\twhile (i < node.childNodes.length) {\r\n\t\tconst childNode = node.childNodes[i]\r\n\t\t// `nodeType: 1` means \"Element\".\r\n\t\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\t\tif (childNode.nodeType === 1 && getTagName(childNode) === tagName) {\r\n\t\t\treturn childNode\r\n\t\t}\r\n\t\ti++\r\n\t}\r\n}\r\n\r\nexport function findChildren(node, tagName) {\r\n\tconst results = []\r\n\tlet i = 0\r\n\twhile (i < node.childNodes.length) {\r\n\t\tconst childNode = node.childNodes[i]\r\n\t\t// `nodeType: 1` means \"Element\".\r\n\t\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\t\tif (childNode.nodeType === 1 && getTagName(childNode) === tagName) {\r\n\t\t\tresults.push(childNode)\r\n\t\t}\r\n\t\ti++\r\n\t}\r\n\treturn results\r\n}\r\n\r\nexport function forEach(node, tagName, func) {\r\n\t// if (typeof tagName === 'function') {\r\n\t// \tfunc = tagName\r\n\t// \ttagName = undefined\r\n\t// }\r\n\tlet i = 0\r\n\twhile (i < node.childNodes.length) {\r\n\t\tconst childNode = node.childNodes[i]\r\n\t\tif (tagName) {\r\n\t\t\t// `nodeType: 1` means \"Element\".\r\n\t\t\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\t\t\tif (childNode.nodeType === 1 && getTagName(childNode) === tagName) {\r\n\t\t\t\tfunc(childNode, i)\r\n\t\t\t}\r\n\t\t} else {\r\n\t\t\tfunc(childNode, i)\r\n\t\t}\r\n\t\ti++\r\n\t}\r\n}\r\n\r\nexport function map(node, tagName, func) {\r\n\tconst results = []\r\n\tforEach(node, tagName, (node, i) => {\r\n\t\tresults.push(func(node, i))\r\n\t})\r\n\treturn results\r\n}\r\n\r\nconst NAMESPACE_REG_EXP = /.+\\:/\r\nexport function getTagName(element) {\r\n\t// For some weird reason, if an element is declared as,\r\n\t// for example, `<x:sheets/>`, then its `.tagName` will be\r\n\t// \"x:sheets\" instead of just \"sheets\".\r\n\t// https://gitlab.com/catamphetamine/read-excel-file/-/issues/25\r\n\t// Its not clear how to tell it to ignore any namespaces\r\n\t// when getting `.tagName`, so just replacing anything\r\n\t// before a colon, if any.\r\n\treturn element.tagName.replace(NAMESPACE_REG_EXP, '')\r\n}\r\n\r\nexport function isElement(node) {\r\n\t// `nodeType: 1` means \"Element\".\r\n\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\treturn node.nodeType === 1\r\n}\r\n\r\nexport function getFirstElementChild(element) {\r\n  let i = 0\r\n  while (i < element.childNodes.length) {\r\n  \tif (isElement(element.childNodes[i])) {\r\n\t\t\treturn element.childNodes[i]\r\n\t\t}\r\n    i++\r\n  }\r\n}\r\n\r\n// This function is only used for occasional debug messages.\r\nexport function getOuterXml(node) {\r\n\t// `nodeType: 1` means \"Element\".\r\n\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\tif (node.nodeType !== 1) {\r\n\t\treturn node.textContent\r\n\t}\r\n\r\n  let xml = '<' + getTagName(node)\r\n\r\n  let j = 0\r\n  while (j < node.attributes.length) {\r\n    xml += ' ' + node.attributes[j].name + '=' + '\"' + node.attributes[j].value + '\"'\r\n    j++\r\n  }\r\n\r\n  xml += '>'\r\n\r\n  let i = 0\r\n  while (i < node.childNodes.length) {\r\n  \txml += getOuterXml(node.childNodes[i])\r\n    i++\r\n  }\r\n\r\n  xml += '</' + getTagName(node) + '>'\r\n\r\n  return xml\r\n}"],"mappings":"AAAA,OAAO,SAASA,SAASA,CAACC,IAAI,EAAEC,OAAO,EAAE;EACxC,IAAIC,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGF,IAAI,CAACG,UAAU,CAACC,MAAM,EAAE;IAClC,IAAMC,SAAS,GAAGL,IAAI,CAACG,UAAU,CAACD,CAAC,CAAC;IACpC;IACA;IACA,IAAIG,SAAS,CAACC,QAAQ,KAAK,CAAC,IAAIC,UAAU,CAACF,SAAS,CAAC,KAAKJ,OAAO,EAAE;MAClE,OAAOI,SAAS;IACjB;IACAH,CAAC,EAAE;EACJ;AACD;AAEA,OAAO,SAASM,YAAYA,CAACR,IAAI,EAAEC,OAAO,EAAE;EAC3C,IAAMQ,OAAO,GAAG,EAAE;EAClB,IAAIP,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGF,IAAI,CAACG,UAAU,CAACC,MAAM,EAAE;IAClC,IAAMC,SAAS,GAAGL,IAAI,CAACG,UAAU,CAACD,CAAC,CAAC;IACpC;IACA;IACA,IAAIG,SAAS,CAACC,QAAQ,KAAK,CAAC,IAAIC,UAAU,CAACF,SAAS,CAAC,KAAKJ,OAAO,EAAE;MAClEQ,OAAO,CAACC,IAAI,CAACL,SAAS,CAAC;IACxB;IACAH,CAAC,EAAE;EACJ;EACA,OAAOO,OAAO;AACf;AAEA,OAAO,SAASE,OAAOA,CAACX,IAAI,EAAEC,OAAO,EAAEW,IAAI,EAAE;EAC5C;EACA;EACA;EACA;EACA,IAAIV,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGF,IAAI,CAACG,UAAU,CAACC,MAAM,EAAE;IAClC,IAAMC,SAAS,GAAGL,IAAI,CAACG,UAAU,CAACD,CAAC,CAAC;IACpC,IAAID,OAAO,EAAE;MACZ;MACA;MACA,IAAII,SAAS,CAACC,QAAQ,KAAK,CAAC,IAAIC,UAAU,CAACF,SAAS,CAAC,KAAKJ,OAAO,EAAE;QAClEW,IAAI,CAACP,SAAS,EAAEH,CAAC,CAAC;MACnB;IACD,CAAC,MAAM;MACNU,IAAI,CAACP,SAAS,EAAEH,CAAC,CAAC;IACnB;IACAA,CAAC,EAAE;EACJ;AACD;AAEA,OAAO,SAASW,GAAGA,CAACb,IAAI,EAAEC,OAAO,EAAEW,IAAI,EAAE;EACxC,IAAMH,OAAO,GAAG,EAAE;EAClBE,OAAO,CAACX,IAAI,EAAEC,OAAO,EAAE,UAACD,IAAI,EAAEE,CAAC,EAAK;IACnCO,OAAO,CAACC,IAAI,CAACE,IAAI,CAACZ,IAAI,EAAEE,CAAC,CAAC,CAAC;EAC5B,CAAC,CAAC;EACF,OAAOO,OAAO;AACf;AAEA,IAAMK,iBAAiB,GAAG,MAAM;AAChC,OAAO,SAASP,UAAUA,CAACQ,OAAO,EAAE;EACnC;EACA;EACA;EACA;EACA;EACA;EACA;EACA,OAAOA,OAAO,CAACd,OAAO,CAACe,OAAO,CAACF,iBAAiB,EAAE,EAAE,CAAC;AACtD;AAEA,OAAO,SAASG,SAASA,CAACjB,IAAI,EAAE;EAC/B;EACA;EACA,OAAOA,IAAI,CAACM,QAAQ,KAAK,CAAC;AAC3B;AAEA,OAAO,SAASY,oBAAoBA,CAACH,OAAO,EAAE;EAC5C,IAAIb,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGa,OAAO,CAACZ,UAAU,CAACC,MAAM,EAAE;IACrC,IAAIa,SAAS,CAACF,OAAO,CAACZ,UAAU,CAACD,CAAC,CAAC,CAAC,EAAE;MACtC,OAAOa,OAAO,CAACZ,UAAU,CAACD,CAAC,CAAC;IAC7B;IACEA,CAAC,EAAE;EACL;AACF;;AAEA;AACA,OAAO,SAASiB,WAAWA,CAACnB,IAAI,EAAE;EACjC;EACA;EACA,IAAIA,IAAI,CAACM,QAAQ,KAAK,CAAC,EAAE;IACxB,OAAON,IAAI,CAACoB,WAAW;EACxB;EAEC,IAAIC,GAAG,GAAG,GAAG,GAAGd,UAAU,CAACP,IAAI,CAAC;EAEhC,IAAIsB,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGtB,IAAI,CAACuB,UAAU,CAACnB,MAAM,EAAE;IACjCiB,GAAG,IAAI,GAAG,GAAGrB,IAAI,CAACuB,UAAU,CAACD,CAAC,CAAC,CAACE,IAAI,GAAG,GAAG,GAAG,GAAG,GAAGxB,IAAI,CAACuB,UAAU,CAACD,CAAC,CAAC,CAACG,KAAK,GAAG,GAAG;IACjFH,CAAC,EAAE;EACL;EAEAD,GAAG,IAAI,GAAG;EAEV,IAAInB,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGF,IAAI,CAACG,UAAU,CAACC,MAAM,EAAE;IAClCiB,GAAG,IAAIF,WAAW,CAACnB,IAAI,CAACG,UAAU,CAACD,CAAC,CAAC,CAAC;IACrCA,CAAC,EAAE;EACL;EAEAmB,GAAG,IAAI,IAAI,GAAGd,UAAU,CAACP,IAAI,CAAC,GAAG,GAAG;EAEpC,OAAOqB,GAAG;AACZ"}
+{"version":3,"file":"dom.js","names":["findChild","node","tagName","i","childNodes","length","childNode","nodeType","getTagName","findChildren","results","push","forEach","func","map","NAMESPACE_REG_EXP","element","replace","isElement","getFirstElementChild","getOuterXml","textContent","xml","j","attributes","name","value"],"sources":["../../source/xml/dom.js"],"sourcesContent":["export function findChild(node, tagName) {\r\n\tlet i = 0\r\n\twhile (i < node.childNodes.length) {\r\n\t\tconst childNode = node.childNodes[i]\r\n\t\t// `nodeType: 1` means \"Element\".\r\n\t\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\t\tif (childNode.nodeType === 1 && getTagName(childNode) === tagName) {\r\n\t\t\treturn childNode\r\n\t\t}\r\n\t\ti++\r\n\t}\r\n}\r\n\r\nexport function findChildren(node, tagName) {\r\n\tconst results = []\r\n\tlet i = 0\r\n\twhile (i < node.childNodes.length) {\r\n\t\tconst childNode = node.childNodes[i]\r\n\t\t// `nodeType: 1` means \"Element\".\r\n\t\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\t\tif (childNode.nodeType === 1 && getTagName(childNode) === tagName) {\r\n\t\t\tresults.push(childNode)\r\n\t\t}\r\n\t\ti++\r\n\t}\r\n\treturn results\r\n}\r\n\r\nexport function forEach(node, tagName, func) {\r\n\t// if (typeof tagName === 'function') {\r\n\t// \tfunc = tagName\r\n\t// \ttagName = undefined\r\n\t// }\r\n\tlet i = 0\r\n\twhile (i < node.childNodes.length) {\r\n\t\tconst childNode = node.childNodes[i]\r\n\t\tif (tagName) {\r\n\t\t\t// `nodeType: 1` means \"Element\".\r\n\t\t\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\t\t\tif (childNode.nodeType === 1 && getTagName(childNode) === tagName) {\r\n\t\t\t\tfunc(childNode, i)\r\n\t\t\t}\r\n\t\t} else {\r\n\t\t\tfunc(childNode, i)\r\n\t\t}\r\n\t\ti++\r\n\t}\r\n}\r\n\r\nexport function map(node, tagName, func) {\r\n\tconst results = []\r\n\tforEach(node, tagName, (node, i) => {\r\n\t\tresults.push(func(node, i))\r\n\t})\r\n\treturn results\r\n}\r\n\r\nconst NAMESPACE_REG_EXP = /.+\\:/\r\nexport function getTagName(element) {\r\n\t// For some weird reason, if an element is declared as,\r\n\t// for example, `<x:sheets/>`, then its `.tagName` will be\r\n\t// \"x:sheets\" instead of just \"sheets\".\r\n\t// https://gitlab.com/catamphetamine/read-excel-file/-/issues/25\r\n\t// Its not clear how to tell it to ignore any namespaces\r\n\t// when getting `.tagName`, so just replacing anything\r\n\t// before a colon, if any.\r\n\treturn element.tagName.replace(NAMESPACE_REG_EXP, '')\r\n}\r\n\r\nexport function isElement(node) {\r\n\t// `nodeType: 1` means \"Element\".\r\n\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\treturn node.nodeType === 1\r\n}\r\n\r\nexport function getFirstElementChild(element) {\r\n  let i = 0\r\n  while (i < element.childNodes.length) {\r\n  \tif (isElement(element.childNodes[i])) {\r\n\t\t\treturn element.childNodes[i]\r\n\t\t}\r\n    i++\r\n  }\r\n}\r\n\r\n// This function was only used for occasional debug messages during development.\r\n// It's not used in the actual code.\r\nexport function getOuterXml(node) {\r\n\t// `nodeType: 1` means \"Element\".\r\n\t// https://www.w3schools.com/xml/prop_element_nodetype.asp\r\n\tif (node.nodeType !== 1) {\r\n\t\treturn node.textContent\r\n\t}\r\n\r\n  let xml = '<' + getTagName(node)\r\n\r\n  let j = 0\r\n  while (j < node.attributes.length) {\r\n    xml += ' ' + node.attributes[j].name + '=' + '\"' + node.attributes[j].value + '\"'\r\n    j++\r\n  }\r\n\r\n  xml += '>'\r\n\r\n  let i = 0\r\n  while (i < node.childNodes.length) {\r\n  \txml += getOuterXml(node.childNodes[i])\r\n    i++\r\n  }\r\n\r\n  xml += '</' + getTagName(node) + '>'\r\n\r\n  return xml\r\n}"],"mappings":"AAAA,OAAO,SAASA,SAASA,CAACC,IAAI,EAAEC,OAAO,EAAE;EACxC,IAAIC,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGF,IAAI,CAACG,UAAU,CAACC,MAAM,EAAE;IAClC,IAAMC,SAAS,GAAGL,IAAI,CAACG,UAAU,CAACD,CAAC,CAAC;IACpC;IACA;IACA,IAAIG,SAAS,CAACC,QAAQ,KAAK,CAAC,IAAIC,UAAU,CAACF,SAAS,CAAC,KAAKJ,OAAO,EAAE;MAClE,OAAOI,SAAS;IACjB;IACAH,CAAC,EAAE;EACJ;AACD;AAEA,OAAO,SAASM,YAAYA,CAACR,IAAI,EAAEC,OAAO,EAAE;EAC3C,IAAMQ,OAAO,GAAG,EAAE;EAClB,IAAIP,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGF,IAAI,CAACG,UAAU,CAACC,MAAM,EAAE;IAClC,IAAMC,SAAS,GAAGL,IAAI,CAACG,UAAU,CAACD,CAAC,CAAC;IACpC;IACA;IACA,IAAIG,SAAS,CAACC,QAAQ,KAAK,CAAC,IAAIC,UAAU,CAACF,SAAS,CAAC,KAAKJ,OAAO,EAAE;MAClEQ,OAAO,CAACC,IAAI,CAACL,SAAS,CAAC;IACxB;IACAH,CAAC,EAAE;EACJ;EACA,OAAOO,OAAO;AACf;AAEA,OAAO,SAASE,OAAOA,CAACX,IAAI,EAAEC,OAAO,EAAEW,IAAI,EAAE;EAC5C;EACA;EACA;EACA;EACA,IAAIV,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGF,IAAI,CAACG,UAAU,CAACC,MAAM,EAAE;IAClC,IAAMC,SAAS,GAAGL,IAAI,CAACG,UAAU,CAACD,CAAC,CAAC;IACpC,IAAID,OAAO,EAAE;MACZ;MACA;MACA,IAAII,SAAS,CAACC,QAAQ,KAAK,CAAC,IAAIC,UAAU,CAACF,SAAS,CAAC,KAAKJ,OAAO,EAAE;QAClEW,IAAI,CAACP,SAAS,EAAEH,CAAC,CAAC;MACnB;IACD,CAAC,MAAM;MACNU,IAAI,CAACP,SAAS,EAAEH,CAAC,CAAC;IACnB;IACAA,CAAC,EAAE;EACJ;AACD;AAEA,OAAO,SAASW,GAAGA,CAACb,IAAI,EAAEC,OAAO,EAAEW,IAAI,EAAE;EACxC,IAAMH,OAAO,GAAG,EAAE;EAClBE,OAAO,CAACX,IAAI,EAAEC,OAAO,EAAE,UAACD,IAAI,EAAEE,CAAC,EAAK;IACnCO,OAAO,CAACC,IAAI,CAACE,IAAI,CAACZ,IAAI,EAAEE,CAAC,CAAC,CAAC;EAC5B,CAAC,CAAC;EACF,OAAOO,OAAO;AACf;AAEA,IAAMK,iBAAiB,GAAG,MAAM;AAChC,OAAO,SAASP,UAAUA,CAACQ,OAAO,EAAE;EACnC;EACA;EACA;EACA;EACA;EACA;EACA;EACA,OAAOA,OAAO,CAACd,OAAO,CAACe,OAAO,CAACF,iBAAiB,EAAE,EAAE,CAAC;AACtD;AAEA,OAAO,SAASG,SAASA,CAACjB,IAAI,EAAE;EAC/B;EACA;EACA,OAAOA,IAAI,CAACM,QAAQ,KAAK,CAAC;AAC3B;AAEA,OAAO,SAASY,oBAAoBA,CAACH,OAAO,EAAE;EAC5C,IAAIb,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGa,OAAO,CAACZ,UAAU,CAACC,MAAM,EAAE;IACrC,IAAIa,SAAS,CAACF,OAAO,CAACZ,UAAU,CAACD,CAAC,CAAC,CAAC,EAAE;MACtC,OAAOa,OAAO,CAACZ,UAAU,CAACD,CAAC,CAAC;IAC7B;IACEA,CAAC,EAAE;EACL;AACF;;AAEA;AACA;AACA,OAAO,SAASiB,WAAWA,CAACnB,IAAI,EAAE;EACjC;EACA;EACA,IAAIA,IAAI,CAACM,QAAQ,KAAK,CAAC,EAAE;IACxB,OAAON,IAAI,CAACoB,WAAW;EACxB;EAEC,IAAIC,GAAG,GAAG,GAAG,GAAGd,UAAU,CAACP,IAAI,CAAC;EAEhC,IAAIsB,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGtB,IAAI,CAACuB,UAAU,CAACnB,MAAM,EAAE;IACjCiB,GAAG,IAAI,GAAG,GAAGrB,IAAI,CAACuB,UAAU,CAACD,CAAC,CAAC,CAACE,IAAI,GAAG,GAAG,GAAG,GAAG,GAAGxB,IAAI,CAACuB,UAAU,CAACD,CAAC,CAAC,CAACG,KAAK,GAAG,GAAG;IACjFH,CAAC,EAAE;EACL;EAEAD,GAAG,IAAI,GAAG;EAEV,IAAInB,CAAC,GAAG,CAAC;EACT,OAAOA,CAAC,GAAGF,IAAI,CAACG,UAAU,CAACC,MAAM,EAAE;IAClCiB,GAAG,IAAIF,WAAW,CAACnB,IAAI,CAACG,UAAU,CAACD,CAAC,CAAC,CAAC;IACrCA,CAAC,EAAE;EACL;EAEAmB,GAAG,IAAI,IAAI,GAAGd,UAAU,CAACP,IAAI,CAAC,GAAG,GAAG;EAEpC,OAAOqB,GAAG;AACZ"}

package/modules/xml/parseXmlStream.js

@@ -0,0 +1,18 @@
+// Currently there're 2 XML stream parser implementations:
+// * `sax` — The "classic" one. Supports Node.js streams.
+// * `saxen` — The small and fast one.
+//
+// `sax` has been around for a long time, and is written by a well-known developer,
+// but it performance is very subpar compared to `saxen`: reading a `10 MB` `.xlsx` file
+// with `sax` is about `1.4` secs while with `saxen` it's about `0.3` secs,
+// meaning that `saxen` is light-years ahead of `sax` in terms of performance.
+//
+// `sax` has an advantage of parsing XML from a Node.js stream rather than from just a `string`,
+// but the weird part is that when enabled, "streaming" mode results in significantly slower parsing:
+// about `3.1` secs. in "streaming" mode vs about `1.4` secs. in non-"streaming" mode.
+//
+// `sax` "bundled" size is `22.5 kB` (`8.1 kB` gzipped).
+// `saxen` "bundled" size is `6.15 kB` (`2.7 kB` gzipped).
+//
+export { default } from './parseXmlStream.saxen.js';
+//# sourceMappingURL=parseXmlStream.js.map

package/modules/xml/parseXmlStream.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"parseXmlStream.js","names":["default"],"sources":["../../source/xml/parseXmlStream.js"],"sourcesContent":["// Currently there're 2 XML stream parser implementations:\r\n// * `sax` — The \"classic\" one. Supports Node.js streams.\r\n// * `saxen` — The small and fast one.\r\n//\r\n// `sax` has been around for a long time, and is written by a well-known developer,\r\n// but it performance is very subpar compared to `saxen`: reading a `10 MB` `.xlsx` file\r\n// with `sax` is about `1.4` secs while with `saxen` it's about `0.3` secs,\r\n// meaning that `saxen` is light-years ahead of `sax` in terms of performance.\r\n//\r\n// `sax` has an advantage of parsing XML from a Node.js stream rather than from just a `string`,\r\n// but the weird part is that when enabled, \"streaming\" mode results in significantly slower parsing:\r\n// about `3.1` secs. in \"streaming\" mode vs about `1.4` secs. in non-\"streaming\" mode.\r\n//\r\n// `sax` \"bundled\" size is `22.5 kB` (`8.1 kB` gzipped).\r\n// `saxen` \"bundled\" size is `6.15 kB` (`2.7 kB` gzipped).\r\n//\r\nexport { default as default } from './parseXmlStream.saxen.js'"],"mappings":"AAAA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA;AACA,SAASA,OAAkB,QAAQ,2BAA2B"}

package/modules/xml/parseXmlStream.sax.js

@@ -0,0 +1,103 @@
+import { Readable } from 'node:stream';
+import sax from 'sax';
+
+// For some strange reason, using "streaming" mode results in
+// quite a dramatic drop in performance.
+// For example, reading a `10 MB` `.xlsx` file in "streaming" mode is about `3.1` secs.
+// while in non-"streaming" mode it's about `1.4` secs.
+// Because of such strange performance issue, "streaming" mode is turned off.
+var STREAMING_MODE = false;
+export default function parseXmlStream(xml, _ref) {
+  var createInitialState = _ref.createInitialState,
+    onOpenTag = _ref.onOpenTag,
+    onCloseTag = _ref.onCloseTag,
+    onText = _ref.onText;
+  return new Promise(function (resolve, reject) {
+    var errored = false;
+    var strict = true;
+    var xmlns = true;
+    var parserOptions = {
+      lowercase: true,
+      xmlns: xmlns
+    };
+    var prefixedTagNameToUnprefixedTagName = {};
+    var state = createInitialState();
+
+    // an error happened.
+    var onerror = function onerror(error) {
+      if (errored) {
+        return;
+      }
+      errored = true;
+      reject(error);
+    };
+
+    // got some text. `text` is the string of text.
+    var ontext = function ontext(text) {
+      if (onText) {
+        onText(text, state);
+      }
+    };
+
+    // opened a tag. `node` has "name" and "attributes"
+    var onopentag = function onopentag(node) {
+      if (onOpenTag) {
+        if (xmlns) {
+          prefixedTagNameToUnprefixedTagName[node.name] = node.local;
+        }
+        onOpenTag(xmlns ? node.local : node.name, xmlns ? getAttributesWithoutXmlnsPrefixes(node.attributes) : node.attributes, state);
+      }
+    };
+
+    // closed a tag.
+    var onclosetag = function onclosetag(name) {
+      if (onCloseTag) {
+        var tagName = xmlns ? prefixedTagNameToUnprefixedTagName[name] : name;
+        if (xmlns && !tagName) {
+          throw new Error("Unknown closing tag: ".concat(name));
+        }
+        onCloseTag(tagName, state);
+      }
+    };
+
+    // an attribute. `attribute` has "name" and "value"
+    var onattribute = function onattribute(attribute) {
+      // const name = xmlns ? attribute.local : attribute.name
+      // const value = attribute.value
+    };
+
+    // parser stream is done, and ready to have more stuff written to it.
+    var onend = function onend() {
+      if (errored) {
+        return;
+      }
+      resolve(state);
+    };
+    if (STREAMING_MODE) {
+      var parser = sax.createStream(strict, parserOptions);
+      parser.on('error', onerror);
+      parser.on('text', ontext);
+      parser.on('opentag', onopentag);
+      parser.on('closetag', onclosetag);
+      parser.on('attribute', onattribute);
+      parser.on('end', onend);
+      Readable.from(xml).pipe(parser);
+    } else {
+      var _parser = sax.parser(strict, parserOptions);
+      _parser.onerror = onerror;
+      _parser.ontext = ontext;
+      _parser.onopentag = onopentag;
+      _parser.onclosetag = onclosetag;
+      _parser.onattribute = onattribute;
+      _parser.onend = onend;
+      _parser.write(xml).end();
+    }
+  });
+}
+function getAttributesWithoutXmlnsPrefixes(attributes) {
+  return Object.keys(attributes).reduce(function (attributesWithoutPrefixes, nameWithPrefix) {
+    attributesWithoutPrefixes[attributes[nameWithPrefix].local] = attributes[nameWithPrefix].value;
+    return attributesWithoutPrefixes;
+  }, {});
+}
+//# sourceMappingURL=parseXmlStream.sax.js.map

package/modules/xml/parseXmlStream.sax.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"parseXmlStream.sax.js","names":["Readable","sax","STREAMING_MODE","parseXmlStream","xml","_ref","createInitialState","onOpenTag","onCloseTag","onText","Promise","resolve","reject","errored","strict","xmlns","parserOptions","lowercase","prefixedTagNameToUnprefixedTagName","state","onerror","error","ontext","text","onopentag","node","name","local","getAttributesWithoutXmlnsPrefixes","attributes","onclosetag","tagName","Error","concat","onattribute","attribute","onend","parser","createStream","on","from","pipe","write","end","Object","keys","reduce","attributesWithoutPrefixes","nameWithPrefix","value"],"sources":["../../source/xml/parseXmlStream.sax.js"],"sourcesContent":["import { Readable } from 'node:stream'\r\n\r\nimport sax from 'sax'\r\n\r\n// For some strange reason, using \"streaming\" mode results in\r\n// quite a dramatic drop in performance.\r\n// For example, reading a `10 MB` `.xlsx` file in \"streaming\" mode is about `3.1` secs.\r\n// while in non-\"streaming\" mode it's about `1.4` secs.\r\n// Because of such strange performance issue, \"streaming\" mode is turned off.\r\nconst STREAMING_MODE = false\r\n\r\nexport default function parseXmlStream(xml, {\r\n\tcreateInitialState,\r\n\tonOpenTag,\r\n\tonCloseTag,\r\n\tonText\r\n}) {\r\n\treturn new Promise((resolve, reject) => {\r\n\t\tlet errored = false\r\n\r\n\t\tconst strict = true\r\n\t\tconst xmlns = true\r\n\t\tconst parserOptions = { lowercase: true, xmlns }\r\n\r\n\t\tconst prefixedTagNameToUnprefixedTagName = {}\r\n\t\tconst state = createInitialState()\r\n\r\n\t\t// an error happened.\r\n\t\tconst onerror = (error) => {\r\n\t\t\tif (errored) {\r\n\t\t\t\treturn\r\n\t\t\t}\r\n\t\t\terrored = true\r\n\t\t\treject(error)\r\n\t\t}\r\n\r\n\t\t// got some text. `text` is the string of text.\r\n\t\tconst ontext = (text) => {\r\n\t\t\tif (onText) {\r\n\t\t\t\tonText(text, state)\r\n\t\t\t}\r\n\t\t}\r\n\r\n\t\t// opened a tag. `node` has \"name\" and \"attributes\"\r\n\t\tconst onopentag = (node) => {\r\n\t\t\tif (onOpenTag) {\r\n\t\t\t\tif (xmlns) {\r\n\t\t\t\t\tprefixedTagNameToUnprefixedTagName[node.name] = node.local\r\n\t\t\t\t}\r\n\t\t\t\tonOpenTag(\r\n\t\t\t\t\txmlns ? node.local : node.name,\r\n\t\t\t\t\txmlns ? getAttributesWithoutXmlnsPrefixes(node.attributes) : node.attributes,\r\n\t\t\t\t\tstate\r\n\t\t\t\t)\r\n\t\t\t}\r\n\t\t}\r\n\r\n\t\t// closed a tag.\r\n\t\tconst onclosetag = (name) => {\r\n\t\t\tif (onCloseTag) {\r\n\t\t\t\tconst tagName = xmlns ? prefixedTagNameToUnprefixedTagName[name] : name\r\n\t\t\t\tif (xmlns && !tagName) {\r\n\t\t\t\t\tthrow new Error(`Unknown closing tag: ${name}`)\r\n\t\t\t\t}\r\n\t\t\t\tonCloseTag(tagName, state)\r\n\t\t\t}\r\n\t\t}\r\n\r\n\t\t// an attribute. `attribute` has \"name\" and \"value\"\r\n\t\tconst onattribute = (attribute) => {\r\n\t\t\t// const name = xmlns ? attribute.local : attribute.name\r\n\t\t\t// const value = attribute.value\r\n\t\t}\r\n\r\n\t\t// parser stream is done, and ready to have more stuff written to it.\r\n\t\tconst onend = () => {\r\n\t\t\tif (errored) {\r\n\t\t\t\treturn\r\n\t\t\t}\r\n\t\t\tresolve(state)\r\n\t\t}\r\n\r\n\t\tif (STREAMING_MODE) {\r\n\t\t\tconst parser = sax.createStream(strict, parserOptions)\r\n\t\t\tparser.on('error', onerror)\r\n\t\t\tparser.on('text', ontext)\r\n\t\t\tparser.on('opentag', onopentag)\r\n\t\t\tparser.on('closetag', onclosetag)\r\n\t\t\tparser.on('attribute', onattribute)\r\n\t\t\tparser.on('end', onend)\r\n\t\t\tReadable.from(xml).pipe(parser)\r\n\t\t} else {\r\n\t\t\tconst parser = sax.parser(strict, parserOptions)\r\n\t\t\tparser.onerror = onerror\r\n\t\t\tparser.ontext = ontext\r\n\t\t\tparser.onopentag = onopentag\r\n\t\t\tparser.onclosetag = onclosetag\r\n\t\t\tparser.onattribute = onattribute\r\n\t\t\tparser.onend = onend\r\n\t\t\tparser.write(xml).end()\r\n\t\t}\r\n\t})\r\n}\r\n\r\nfunction getAttributesWithoutXmlnsPrefixes(attributes) {\r\n  return Object.keys(attributes).reduce((attributesWithoutPrefixes, nameWithPrefix) => {\r\n    attributesWithoutPrefixes[attributes[nameWithPrefix].local] = attributes[nameWithPrefix].value\r\n    return attributesWithoutPrefixes\r\n  }, {})\r\n}"],"mappings":"AAAA,SAASA,QAAQ,QAAQ,aAAa;AAEtC,OAAOC,GAAG,MAAM,KAAK;;AAErB;AACA;AACA;AACA;AACA;AACA,IAAMC,cAAc,GAAG,KAAK;AAE5B,eAAe,SAASC,cAAcA,CAACC,GAAG,EAAAC,IAAA,EAKvC;EAAA,IAJFC,kBAAkB,GAAAD,IAAA,CAAlBC,kBAAkB;IAClBC,SAAS,GAAAF,IAAA,CAATE,SAAS;IACTC,UAAU,GAAAH,IAAA,CAAVG,UAAU;IACVC,MAAM,GAAAJ,IAAA,CAANI,MAAM;EAEN,OAAO,IAAIC,OAAO,CAAC,UAACC,OAAO,EAAEC,MAAM,EAAK;IACvC,IAAIC,OAAO,GAAG,KAAK;IAEnB,IAAMC,MAAM,GAAG,IAAI;IACnB,IAAMC,KAAK,GAAG,IAAI;IAClB,IAAMC,aAAa,GAAG;MAAEC,SAAS,EAAE,IAAI;MAAEF,KAAK,EAALA;IAAM,CAAC;IAEhD,IAAMG,kCAAkC,GAAG,CAAC,CAAC;IAC7C,IAAMC,KAAK,GAAGb,kBAAkB,CAAC,CAAC;;IAElC;IACA,IAAMc,OAAO,GAAG,SAAVA,OAAOA,CAAIC,KAAK,EAAK;MAC1B,IAAIR,OAAO,EAAE;QACZ;MACD;MACAA,OAAO,GAAG,IAAI;MACdD,MAAM,CAACS,KAAK,CAAC;IACd,CAAC;;IAED;IACA,IAAMC,MAAM,GAAG,SAATA,MAAMA,CAAIC,IAAI,EAAK;MACxB,IAAId,MAAM,EAAE;QACXA,MAAM,CAACc,IAAI,EAAEJ,KAAK,CAAC;MACpB;IACD,CAAC;;IAED;IACA,IAAMK,SAAS,GAAG,SAAZA,SAASA,CAAIC,IAAI,EAAK;MAC3B,IAAIlB,SAAS,EAAE;QACd,IAAIQ,KAAK,EAAE;UACVG,kCAAkC,CAACO,IAAI,CAACC,IAAI,CAAC,GAAGD,IAAI,CAACE,KAAK;QAC3D;QACApB,SAAS,CACRQ,KAAK,GAAGU,IAAI,CAACE,KAAK,GAAGF,IAAI,CAACC,IAAI,EAC9BX,KAAK,GAAGa,iCAAiC,CAACH,IAAI,CAACI,UAAU,CAAC,GAAGJ,IAAI,CAACI,UAAU,EAC5EV,KACD,CAAC;MACF;IACD,CAAC;;IAED;IACA,IAAMW,UAAU,GAAG,SAAbA,UAAUA,CAAIJ,IAAI,EAAK;MAC5B,IAAIlB,UAAU,EAAE;QACf,IAAMuB,OAAO,GAAGhB,KAAK,GAAGG,kCAAkC,CAACQ,IAAI,CAAC,GAAGA,IAAI;QACvE,IAAIX,KAAK,IAAI,CAACgB,OAAO,EAAE;UACtB,MAAM,IAAIC,KAAK,yBAAAC,MAAA,CAAyBP,IAAI,CAAE,CAAC;QAChD;QACAlB,UAAU,CAACuB,OAAO,EAAEZ,KAAK,CAAC;MAC3B;IACD,CAAC;;IAED;IACA,IAAMe,WAAW,GAAG,SAAdA,WAAWA,CAAIC,SAAS,EAAK;MAClC;MACA;IAAA,CACA;;IAED;IACA,IAAMC,KAAK,GAAG,SAARA,KAAKA,CAAA,EAAS;MACnB,IAAIvB,OAAO,EAAE;QACZ;MACD;MACAF,OAAO,CAACQ,KAAK,CAAC;IACf,CAAC;IAED,IAAIjB,cAAc,EAAE;MACnB,IAAMmC,MAAM,GAAGpC,GAAG,CAACqC,YAAY,CAACxB,MAAM,EAAEE,aAAa,CAAC;MACtDqB,MAAM,CAACE,EAAE,CAAC,OAAO,EAAEnB,OAAO,CAAC;MAC3BiB,MAAM,CAACE,EAAE,CAAC,MAAM,EAAEjB,MAAM,CAAC;MACzBe,MAAM,CAACE,EAAE,CAAC,SAAS,EAAEf,SAAS,CAAC;MAC/Ba,MAAM,CAACE,EAAE,CAAC,UAAU,EAAET,UAAU,CAAC;MACjCO,MAAM,CAACE,EAAE,CAAC,WAAW,EAAEL,WAAW,CAAC;MACnCG,MAAM,CAACE,EAAE,CAAC,KAAK,EAAEH,KAAK,CAAC;MACvBpC,QAAQ,CAACwC,IAAI,CAACpC,GAAG,CAAC,CAACqC,IAAI,CAACJ,MAAM,CAAC;IAChC,CAAC,MAAM;MACN,IAAMA,OAAM,GAAGpC,GAAG,CAACoC,MAAM,CAACvB,MAAM,EAAEE,aAAa,CAAC;MAChDqB,OAAM,CAACjB,OAAO,GAAGA,OAAO;MACxBiB,OAAM,CAACf,MAAM,GAAGA,MAAM;MACtBe,OAAM,CAACb,SAAS,GAAGA,SAAS;MAC5Ba,OAAM,CAACP,UAAU,GAAGA,UAAU;MAC9BO,OAAM,CAACH,WAAW,GAAGA,WAAW;MAChCG,OAAM,CAACD,KAAK,GAAGA,KAAK;MACpBC,OAAM,CAACK,KAAK,CAACtC,GAAG,CAAC,CAACuC,GAAG,CAAC,CAAC;IACxB;EACD,CAAC,CAAC;AACH;AAEA,SAASf,iCAAiCA,CAACC,UAAU,EAAE;EACrD,OAAOe,MAAM,CAACC,IAAI,CAAChB,UAAU,CAAC,CAACiB,MAAM,CAAC,UAACC,yBAAyB,EAAEC,cAAc,EAAK;IACnFD,yBAAyB,CAAClB,UAAU,CAACmB,cAAc,CAAC,CAACrB,KAAK,CAAC,GAAGE,UAAU,CAACmB,cAAc,CAAC,CAACC,KAAK;IAC9F,OAAOF,yBAAyB;EAClC,CAAC,EAAE,CAAC,CAAC,CAAC;AACR"}