read-excel-file 9.3.3 → 9.3.5

package/modules/xlsx/parseSharedStrings.js

@@ -1,49 +1,103 @@
-import { onOpenTagInSharedString, onCloseTagInSharedString, onTextInSharedString, createInitialStateInSharedString } from './parseSharedString.js';
-
 /**
  * Parses `sharedStrings.xml` file.
  * @param {string} content
- * @param {function} parseXmlStream — SAX XML parser.
+ * @param {function} parseXml — SAX XML parser.
  * @returns {Promise<string[]>}
  */
-export default function parseSharedStrings(content, parseXmlStream) {
-  return parseXmlStream(content, {
-    createInitialState: createInitialStateInSharedStrings,
-    onOpenTag: onOpenTagInSharedStrings,
-    onCloseTag: onCloseTagInSharedStrings,
-    onText: onTextInSharedStrings
-  }).then(function (state) {
-    return state.strings;
+export default function parseSharedStrings(content, parseXml) {
+  var state = createInitialState();
+  return parseXml(content, state, onOpenTag, onCloseTag, onText).then(function () {
+    return getResultFromState(state);
   });
-}
-function createInitialStateInSharedStrings() {
-  return {
-    si: undefined,
-    strings: []
-  };
-}
+  function createInitialState() {
+    return {
+      si: undefined,
+      strings: []
+    };
+  }
+  function getResultFromState(state) {
+    return state.strings;
+  }
 
-// In an XLSX `sharedStrings.xml` file, a <si/> (string item) element can have
-// either a single child <t/> element (meaning that "rich formatting" is not used)
-// or multiple child <r/> elements (meaning that "rich formatting" is used).
-function onOpenTagInSharedStrings(tagName, attributes, state) {
-  if (tagName === 'si') {
-    state.si = createInitialStateInSharedString();
-  } else if (state.si) {
-    onOpenTagInSharedString(tagName, attributes, state.si);
+  // In an XLSX `sharedStrings.xml` file, a <si/> (string item) element can have
+  // either a single child <t/> element (meaning that "rich formatting" is not used)
+  // or multiple child <r/> elements (meaning that "rich formatting" is used).
+  function onOpenTag(tagName, attributes, state) {
+    if (tagName === 'si') {
+      state.si = createInitialStateInSharedString();
+    } else if (state.si) {
+      onOpenTagInSharedString(tagName, attributes, state.si);
+    }
   }
-}
-function onCloseTagInSharedStrings(tagName, state) {
-  if (tagName === 'si') {
-    state.strings.push(state.si.string);
-    state.si = undefined;
-  } else if (state.si) {
-    onCloseTagInSharedString(tagName, state.si);
+  function onCloseTag(tagName, state) {
+    if (tagName === 'si') {
+      state.strings.push(state.si.string);
+      state.si = undefined;
+    } else if (state.si) {
+      onCloseTagInSharedString(tagName, state.si);
+    }
   }
-}
-function onTextInSharedStrings(text, state) {
-  if (state.si) {
-    onTextInSharedString(text, state.si);
+  function onText(text, state) {
+    if (state.si) {
+      onTextInSharedString(text, state.si);
+    }
+  }
+
+  // Shared String
+
+  function createInitialStateInSharedString() {
+    return {
+      t: false,
+      r: false,
+      rPh: false,
+      string: ''
+    };
+  }
+
+  // In an XLSX `sharedStrings.xml` file, a <si/> (string item) element can have
+  // either a single child <t/> element (meaning that "rich formatting" is not used)
+  // or multiple child <r/> elements (meaning that "rich formatting" is used).
+  function onOpenTagInSharedString(tagName, attributes, state) {
+    // The possible children of <si> are:
+    // * <t>: Represents a simple text string. For plain text, this is the only child.
+    // * <r>: Represents a rich text run. When a cell contains text where only some words are bolded, colored, or styled differently, multiple <r> elements are used to break the text into formatted segments.
+    // * <rPh> (Phonetic Run): Used primarily for East Asian languages to provide phonetic reading/pronunciation data (e.g., furigana in Japanese). It associates a phonetic pronunciation run right alongside the base string text tag <t>.
+    // * <phoneticPr>: Defines the phonetic properties used for East Asian languages, specifically mapping how strings should be read for sorting and pronunciation.
+    if (tagName === 't') {
+      state.t = true;
+    } else if (tagName === 'r') {
+      // The possible children of <r/> are:
+      // * <rPr> (Run Properties): The formatting properties for the text (font, size, color, bold, italic, etc.).
+      // * <t> (Text): The actual text payload.
+      // * <rPh> (Phonetic Run): Phonetic pronunciation guidance (used for East Asian languages like Japanese). It associates a phonetic pronunciation run right alongside the base string text tag <t>.
+      state.r = true;
+    } else if (tagName === 'rPh') {
+      // The possible children of <rPh/> are:
+      // * <t> (Text):  Contains the actual phonetic text or reading (usually in Katakana for Japanese) that corresponds to the associated character string.
+      state.rPh = true;
+    }
+  }
+  function onCloseTagInSharedString(tagName, state) {
+    if (tagName === 't') {
+      state.t = false;
+    } else if (tagName === 'r') {
+      state.r = false;
+    } else if (tagName === 'rPh') {
+      state.rPh = false;
+    }
+  }
+  function onTextInSharedString(text, state) {
+    if (state.rPh) {
+      // Ignore anything inside `<rPh/>` tags
+    } else if (state.t) {
+      if (state.r) {
+        // An `<r/>` element could contain multiple `<t/>` elements,
+        // the text content from all of which should be concatenated.
+        state.string += text;
+      } else {
+        state.string = text;
+      }
+    }
   }
 }
 //# sourceMappingURL=parseSharedStrings.js.map

package/modules/xlsx/parseSharedStrings.js.map

@@ -1 +1 @@
-{"version":3,"file":"parseSharedStrings.js","names":["onOpenTagInSharedString","onCloseTagInSharedString","onTextInSharedString","createInitialStateInSharedString","parseSharedStrings","content","parseXmlStream","createInitialState","createInitialStateInSharedStrings","onOpenTag","onOpenTagInSharedStrings","onCloseTag","onCloseTagInSharedStrings","onText","onTextInSharedStrings","then","state","strings","si","undefined","tagName","attributes","push","string","text"],"sources":["../../source/xlsx/parseSharedStrings.js"],"sourcesContent":["import {\r\n  onOpenTagInSharedString,\r\n  onCloseTagInSharedString,\r\n  onTextInSharedString,\r\n  createInitialStateInSharedString\r\n} from './parseSharedString.js'\r\n\r\n/**\r\n * Parses `sharedStrings.xml` file.\r\n * @param {string} content\r\n * @param {function} parseXmlStream — SAX XML parser.\r\n * @returns {Promise<string[]>}\r\n */\r\nexport default function parseSharedStrings(content, parseXmlStream) {\r\n  return parseXmlStream(content, {\r\n    createInitialState: createInitialStateInSharedStrings,\r\n    onOpenTag: onOpenTagInSharedStrings,\r\n    onCloseTag: onCloseTagInSharedStrings,\r\n    onText: onTextInSharedStrings\r\n  }).then((state) => {\r\n    return state.strings\r\n  })\r\n}\r\n\r\nfunction createInitialStateInSharedStrings() {\r\n  return {\r\n    si: undefined,\r\n    strings: []\r\n  }\r\n}\r\n\r\n// In an XLSX `sharedStrings.xml` file, a <si/> (string item) element can have\r\n// either a single child <t/> element (meaning that \"rich formatting\" is not used)\r\n// or multiple child <r/> elements (meaning that \"rich formatting\" is used).\r\nfunction onOpenTagInSharedStrings(tagName, attributes, state) {\r\n  if (tagName === 'si') {\r\n    state.si = createInitialStateInSharedString()\r\n  } else if (state.si) {\r\n    onOpenTagInSharedString(tagName, attributes, state.si)\r\n  }\r\n}\r\n\r\nfunction onCloseTagInSharedStrings(tagName, state) {\r\n  if (tagName === 'si') {\r\n    state.strings.push(state.si.string)\r\n    state.si = undefined\r\n  } else if (state.si) {\r\n    onCloseTagInSharedString(tagName, state.si)\r\n  }\r\n}\r\n\r\nfunction onTextInSharedStrings(text, state) {\r\n  if (state.si) {\r\n    onTextInSharedString(text, state.si)\r\n  }\r\n}"],"mappings":"AAAA,SACEA,uBAAuB,EACvBC,wBAAwB,EACxBC,oBAAoB,EACpBC,gCAAgC,QAC3B,wBAAwB;;AAE/B;AACA;AACA;AACA;AACA;AACA;AACA,eAAe,SAASC,kBAAkBA,CAACC,OAAO,EAAEC,cAAc,EAAE;EAClE,OAAOA,cAAc,CAACD,OAAO,EAAE;IAC7BE,kBAAkB,EAAEC,iCAAiC;IACrDC,SAAS,EAAEC,wBAAwB;IACnCC,UAAU,EAAEC,yBAAyB;IACrCC,MAAM,EAAEC;EACV,CAAC,CAAC,CAACC,IAAI,CAAC,UAACC,KAAK,EAAK;IACjB,OAAOA,KAAK,CAACC,OAAO;EACtB,CAAC,CAAC;AACJ;AAEA,SAAST,iCAAiCA,CAAA,EAAG;EAC3C,OAAO;IACLU,EAAE,EAAEC,SAAS;IACbF,OAAO,EAAE;EACX,CAAC;AACH;;AAEA;AACA;AACA;AACA,SAASP,wBAAwBA,CAACU,OAAO,EAAEC,UAAU,EAAEL,KAAK,EAAE;EAC5D,IAAII,OAAO,KAAK,IAAI,EAAE;IACpBJ,KAAK,CAACE,EAAE,GAAGf,gCAAgC,CAAC,CAAC;EAC/C,CAAC,MAAM,IAAIa,KAAK,CAACE,EAAE,EAAE;IACnBlB,uBAAuB,CAACoB,OAAO,EAAEC,UAAU,EAAEL,KAAK,CAACE,EAAE,CAAC;EACxD;AACF;AAEA,SAASN,yBAAyBA,CAACQ,OAAO,EAAEJ,KAAK,EAAE;EACjD,IAAII,OAAO,KAAK,IAAI,EAAE;IACpBJ,KAAK,CAACC,OAAO,CAACK,IAAI,CAACN,KAAK,CAACE,EAAE,CAACK,MAAM,CAAC;IACnCP,KAAK,CAACE,EAAE,GAAGC,SAAS;EACtB,CAAC,MAAM,IAAIH,KAAK,CAACE,EAAE,EAAE;IACnBjB,wBAAwB,CAACmB,OAAO,EAAEJ,KAAK,CAACE,EAAE,CAAC;EAC7C;AACF;AAEA,SAASJ,qBAAqBA,CAACU,IAAI,EAAER,KAAK,EAAE;EAC1C,IAAIA,KAAK,CAACE,EAAE,EAAE;IACZhB,oBAAoB,CAACsB,IAAI,EAAER,KAAK,CAACE,EAAE,CAAC;EACtC;AACF"}
+{"version":3,"file":"parseSharedStrings.js","names":["parseSharedStrings","content","parseXml","state","createInitialState","onOpenTag","onCloseTag","onText","then","getResultFromState","si","undefined","strings","tagName","attributes","createInitialStateInSharedString","onOpenTagInSharedString","push","string","onCloseTagInSharedString","text","onTextInSharedString","t","r","rPh"],"sources":["../../source/xlsx/parseSharedStrings.js"],"sourcesContent":["/**\r\n * Parses `sharedStrings.xml` file.\r\n * @param {string} content\r\n * @param {function} parseXml — SAX XML parser.\r\n * @returns {Promise<string[]>}\r\n */\r\nexport default function parseSharedStrings(content, parseXml) {\r\n  const state = createInitialState()\r\n  return parseXml(\r\n    content,\r\n    state,\r\n    onOpenTag,\r\n    onCloseTag,\r\n    onText\r\n  ).then(() => getResultFromState(state))\r\n\r\n  function createInitialState() {\r\n    return {\r\n      si: undefined,\r\n      strings: []\r\n    }\r\n  }\r\n\r\n  function getResultFromState(state) {\r\n    return state.strings\r\n  }\r\n\r\n  // In an XLSX `sharedStrings.xml` file, a <si/> (string item) element can have\r\n  // either a single child <t/> element (meaning that \"rich formatting\" is not used)\r\n  // or multiple child <r/> elements (meaning that \"rich formatting\" is used).\r\n  function onOpenTag(tagName, attributes, state) {\r\n    if (tagName === 'si') {\r\n      state.si = createInitialStateInSharedString()\r\n    } else if (state.si) {\r\n      onOpenTagInSharedString(tagName, attributes, state.si)\r\n    }\r\n  }\r\n\r\n  function onCloseTag(tagName, state) {\r\n    if (tagName === 'si') {\r\n      state.strings.push(state.si.string)\r\n      state.si = undefined\r\n    } else if (state.si) {\r\n      onCloseTagInSharedString(tagName, state.si)\r\n    }\r\n  }\r\n\r\n  function onText(text, state) {\r\n    if (state.si) {\r\n      onTextInSharedString(text, state.si)\r\n    }\r\n  }\r\n\r\n  // Shared String\r\n\r\n  function createInitialStateInSharedString() {\r\n    return {\r\n      t: false,\r\n      r: false,\r\n      rPh: false,\r\n      string: ''\r\n    }\r\n  }\r\n\r\n  // In an XLSX `sharedStrings.xml` file, a <si/> (string item) element can have\r\n  // either a single child <t/> element (meaning that \"rich formatting\" is not used)\r\n  // or multiple child <r/> elements (meaning that \"rich formatting\" is used).\r\n  function onOpenTagInSharedString(tagName, attributes, state) {\r\n    // The possible children of <si> are:\r\n    // * <t>: Represents a simple text string. For plain text, this is the only child.\r\n    // * <r>: Represents a rich text run. When a cell contains text where only some words are bolded, colored, or styled differently, multiple <r> elements are used to break the text into formatted segments.\r\n    // * <rPh> (Phonetic Run): Used primarily for East Asian languages to provide phonetic reading/pronunciation data (e.g., furigana in Japanese). It associates a phonetic pronunciation run right alongside the base string text tag <t>.\r\n    // * <phoneticPr>: Defines the phonetic properties used for East Asian languages, specifically mapping how strings should be read for sorting and pronunciation.\r\n    if (tagName === 't') {\r\n      state.t = true\r\n    } else if (tagName === 'r') {\r\n      // The possible children of <r/> are:\r\n      // * <rPr> (Run Properties): The formatting properties for the text (font, size, color, bold, italic, etc.).\r\n      // * <t> (Text): The actual text payload.\r\n      // * <rPh> (Phonetic Run): Phonetic pronunciation guidance (used for East Asian languages like Japanese). It associates a phonetic pronunciation run right alongside the base string text tag <t>.\r\n      state.r = true\r\n    } else if (tagName === 'rPh') {\r\n      // The possible children of <rPh/> are:\r\n      // * <t> (Text):  Contains the actual phonetic text or reading (usually in Katakana for Japanese) that corresponds to the associated character string.\r\n      state.rPh = true\r\n    }\r\n  }\r\n\r\n  function onCloseTagInSharedString(tagName, state) {\r\n    if (tagName === 't') {\r\n      state.t = false\r\n    } else if (tagName === 'r') {\r\n      state.r = false\r\n    } else if (tagName === 'rPh') {\r\n      state.rPh = false\r\n    }\r\n  }\r\n\r\n  function onTextInSharedString(text, state) {\r\n    if (state.rPh) {\r\n      // Ignore anything inside `<rPh/>` tags\r\n    } else if (state.t) {\r\n      if (state.r) {\r\n        // An `<r/>` element could contain multiple `<t/>` elements,\r\n        // the text content from all of which should be concatenated.\r\n        state.string += text\r\n      } else {\r\n        state.string = text\r\n      }\r\n    }\r\n  }\r\n}"],"mappings":"AAAA;AACA;AACA;AACA;AACA;AACA;AACA,eAAe,SAASA,kBAAkBA,CAACC,OAAO,EAAEC,QAAQ,EAAE;EAC5D,IAAMC,KAAK,GAAGC,kBAAkB,CAAC,CAAC;EAClC,OAAOF,QAAQ,CACbD,OAAO,EACPE,KAAK,EACLE,SAAS,EACTC,UAAU,EACVC,MACF,CAAC,CAACC,IAAI,CAAC;IAAA,OAAMC,kBAAkB,CAACN,KAAK,CAAC;EAAA,EAAC;EAEvC,SAASC,kBAAkBA,CAAA,EAAG;IAC5B,OAAO;MACLM,EAAE,EAAEC,SAAS;MACbC,OAAO,EAAE;IACX,CAAC;EACH;EAEA,SAASH,kBAAkBA,CAACN,KAAK,EAAE;IACjC,OAAOA,KAAK,CAACS,OAAO;EACtB;;EAEA;EACA;EACA;EACA,SAASP,SAASA,CAACQ,OAAO,EAAEC,UAAU,EAAEX,KAAK,EAAE;IAC7C,IAAIU,OAAO,KAAK,IAAI,EAAE;MACpBV,KAAK,CAACO,EAAE,GAAGK,gCAAgC,CAAC,CAAC;IAC/C,CAAC,MAAM,IAAIZ,KAAK,CAACO,EAAE,EAAE;MACnBM,uBAAuB,CAACH,OAAO,EAAEC,UAAU,EAAEX,KAAK,CAACO,EAAE,CAAC;IACxD;EACF;EAEA,SAASJ,UAAUA,CAACO,OAAO,EAAEV,KAAK,EAAE;IAClC,IAAIU,OAAO,KAAK,IAAI,EAAE;MACpBV,KAAK,CAACS,OAAO,CAACK,IAAI,CAACd,KAAK,CAACO,EAAE,CAACQ,MAAM,CAAC;MACnCf,KAAK,CAACO,EAAE,GAAGC,SAAS;IACtB,CAAC,MAAM,IAAIR,KAAK,CAACO,EAAE,EAAE;MACnBS,wBAAwB,CAACN,OAAO,EAAEV,KAAK,CAACO,EAAE,CAAC;IAC7C;EACF;EAEA,SAASH,MAAMA,CAACa,IAAI,EAAEjB,KAAK,EAAE;IAC3B,IAAIA,KAAK,CAACO,EAAE,EAAE;MACZW,oBAAoB,CAACD,IAAI,EAAEjB,KAAK,CAACO,EAAE,CAAC;IACtC;EACF;;EAEA;;EAEA,SAASK,gCAAgCA,CAAA,EAAG;IAC1C,OAAO;MACLO,CAAC,EAAE,KAAK;MACRC,CAAC,EAAE,KAAK;MACRC,GAAG,EAAE,KAAK;MACVN,MAAM,EAAE;IACV,CAAC;EACH;;EAEA;EACA;EACA;EACA,SAASF,uBAAuBA,CAACH,OAAO,EAAEC,UAAU,EAAEX,KAAK,EAAE;IAC3D;IACA;IACA;IACA;IACA;IACA,IAAIU,OAAO,KAAK,GAAG,EAAE;MACnBV,KAAK,CAACmB,CAAC,GAAG,IAAI;IAChB,CAAC,MAAM,IAAIT,OAAO,KAAK,GAAG,EAAE;MAC1B;MACA;MACA;MACA;MACAV,KAAK,CAACoB,CAAC,GAAG,IAAI;IAChB,CAAC,MAAM,IAAIV,OAAO,KAAK,KAAK,EAAE;MAC5B;MACA;MACAV,KAAK,CAACqB,GAAG,GAAG,IAAI;IAClB;EACF;EAEA,SAASL,wBAAwBA,CAACN,OAAO,EAAEV,KAAK,EAAE;IAChD,IAAIU,OAAO,KAAK,GAAG,EAAE;MACnBV,KAAK,CAACmB,CAAC,GAAG,KAAK;IACjB,CAAC,MAAM,IAAIT,OAAO,KAAK,GAAG,EAAE;MAC1BV,KAAK,CAACoB,CAAC,GAAG,KAAK;IACjB,CAAC,MAAM,IAAIV,OAAO,KAAK,KAAK,EAAE;MAC5BV,KAAK,CAACqB,GAAG,GAAG,KAAK;IACnB;EACF;EAEA,SAASH,oBAAoBA,CAACD,IAAI,EAAEjB,KAAK,EAAE;IACzC,IAAIA,KAAK,CAACqB,GAAG,EAAE;MACb;IAAA,CACD,MAAM,IAAIrB,KAAK,CAACmB,CAAC,EAAE;MAClB,IAAInB,KAAK,CAACoB,CAAC,EAAE;QACX;QACA;QACApB,KAAK,CAACe,MAAM,IAAIE,IAAI;MACtB,CAAC,MAAM;QACLjB,KAAK,CAACe,MAAM,GAAGE,IAAI;MACrB;IACF;EACF;AACF"}