Fix ignoring of separator within quotes

Author: tariibaba

src/index.test.ts

@@ -130,36 +130,71 @@ it('accepts custom indentation', () => {
   );
 });
 
-it('accepts CSV with quotes', () => {
+describe('accepts CSV with quotes', () => {
   const csv = `color,maxSpeed,age
+    "red, yellow",'120',2`;
+
+  const csv2 = `color,maxSpeed,age
 "red",'120',2`;
-  const xml = csvToXml(csv, { quotes: 'double' });
-  expect(xml).toBe(
-    `<row>
-    <color>red</color>
+  it('double', () => {
+    const xml = csvToXml(csv, { quote: 'double' });
+    expect(xml).toBe(
+      `<row>
+    <color>red, yellow</color>
     <maxSpeed>'120'</maxSpeed>
     <age>2</age>
 </row>
 `
-  );
+    );
+  });
 
-  const xml2 = csvToXml(csv, { quotes: 'single' });
-  expect(xml2).toBe(
-    `<row>
+  it('single', () => {
+    const xml2 = csvToXml(csv2, { quote: 'single' });
+    expect(xml2).toBe(
+      `<row>
     <color>"red"</color>
     <maxSpeed>120</maxSpeed>
     <age>2</age>
 </row>
 `
-  );
+    );
+  });
 
-  const xml3 = csvToXml(csv, { quotes: 'none' });
-  expect(xml3).toBe(
-    `<row>
+  it('none', () => {
+    const xml3 = csvToXml(csv2, { quote: 'none' });
+    expect(xml3).toBe(
+      `<row>
     <color>"red"</color>
     <maxSpeed>'120'</maxSpeed>
     <age>2</age>
 </row>
 `
-  );
+    );
+  });
+
+  it('all', () => {
+    const xml4 = csvToXml(csv2, { quote: 'all' });
+    expect(xml4).toBe(
+      `<row>
+    <color>red</color>
+    <maxSpeed>120</maxSpeed>
+    <age>2</age>
+</row>
+`
+    );
+  });
+
+  it('custom regex', () => {
+    const csv5 = `color,maxSpeed,age
+    (red, yellow),'120',2`;
+    const xml5 = csvToXml(csv5, { quote: /[\(|\)]/ });
+    expect(xml5).toBe(
+      `<row>
+    <color>red, yellow</color>
+    <maxSpeed>'120'</maxSpeed>
+    <age>2</age>
+</row>
+`
+    );
+  });
 });

src/index.ts

@@ -1,4 +1,5 @@
 import detectEOL from 'detect-eol';
+import { tryRegexToString, splitIgnoringChar } from './utils';
 
 type CSVToXMLOptions = {
   eol?: string;
@@ -7,7 +8,7 @@ type CSVToXMLOptions = {
   headerList?: string[];
   header?: boolean;
   indentation?: number | string;
-  quotes?: 'single' | 'double' | 'none';
+  quote?: 'single' | 'double' | 'all' | 'none' | RegExp;
 };
 
 const defaultOptions: CSVToXMLOptions = {
@@ -17,7 +18,7 @@ const defaultOptions: CSVToXMLOptions = {
   header: true,
   headerList: [],
   indentation: 4,
-  quotes: 'none',
+  quote: 'none',
 };
 
 export default function csvToXml(
@@ -29,12 +30,35 @@ export default function csvToXml(
   const csvData = csvString.split(eol!).map((row) => row.trim());
   const separator = usedOptions.separator!;
 
-  const firstRow = csvData[0].split(separator);
+  let quote;
+  if (usedOptions.quote instanceof RegExp) {
+    quote = usedOptions.quote;
+  } else if (usedOptions.quote) {
+    quote = usedOptions.quote;
+    quote = { single: "'", double: '"', all: /(?:'|")/, none: undefined }[
+      usedOptions.quote
+    ];
+  }
+
+  const split = (str: string) => {
+    return splitIgnoringChar({ str, separator, charToIgnore: quote }).map(
+      (ch) =>
+        ch.replace(
+          new RegExp(
+            String.raw`${tryRegexToString(quote) ?? ''}(.*?)${
+              tryRegexToString(quote) ?? ''
+            }`
+          ),
+          '$1'
+        )
+    );
+  };
+
+  const firstRow = split(csvData[0]);
+
   const colCount = firstRow.length;
 
-  const foundHeaders = usedOptions.header
-    ? csvData[0].split(separator)
-    : [...Array(colCount)];
+  const foundHeaders = usedOptions.header ? firstRow : [...Array(colCount)];
   const specifiedHeaders: (string | undefined)[] = [...usedOptions.headerList!];
 
   const usedHeaders: string[] = [];
@@ -69,7 +93,7 @@ export default function csvToXml(
       : usedOptions.indentation;
 
   for (let i = rowStartLine; i < csvData.length; i++) {
-    const details = csvData[i].split(separator);
+    const details = split(csvData[i]);
 
     if (details.length < colCount) {
       warn('rows found without enough columns.');
@@ -79,13 +103,6 @@ export default function csvToXml(
     xml += `<${usedOptions.rowName}>\n`;
     for (let j = 0; j < colCount; j++) {
       let colValue = details[j];
-      if (!usedOptions.quotes || usedOptions.quotes !== 'none') {
-        const quoteRemovingRegex = {
-          double: /"(.*?)"/,
-          single: /'(.*?)'/,
-        }[usedOptions.quotes!];
-        colValue = colValue.replace(quoteRemovingRegex, '$1');
-      }
       xml += `${spaces}<${usedHeaders[j]}>${colValue}</${usedHeaders[j]}>\n`;
     }
     xml += `</${usedOptions.rowName}>\n`;

src/utils.ts

@@ -0,0 +1,47 @@
+export function splitIgnoringChar(params: {
+  str: string;
+  separator: string;
+  charToIgnore: string | RegExp;
+}) {
+  const { str, separator, charToIgnore } = params;
+
+  let result: string[] = [];
+  let awaitingEndQuote = false;
+  let accElement = '';
+
+  for (let i = 0; i < str.length; i++) {
+    const ch = str[i];
+
+    let isCharToIgnore = false;
+    if (charToIgnore instanceof RegExp) {
+      isCharToIgnore = charToIgnore.test(ch);
+    } else if (typeof charToIgnore === 'string') {
+      isCharToIgnore = charToIgnore === ch;
+    }
+
+    if (isCharToIgnore) {
+      awaitingEndQuote = !awaitingEndQuote;
+    }
+    if (ch === separator) {
+      if (!awaitingEndQuote) {
+        result.push(accElement);
+        accElement = '';
+        continue;
+      }
+    }
+    accElement += ch;
+  }
+  result.push(accElement);
+
+  return result;
+}
+
+export function tryRegexToString(
+  regexOrString?: string | RegExp
+): string | undefined {
+  if (typeof regexOrString === 'string') {
+    return regexOrString;
+  } else if (regexOrString instanceof RegExp) {
+    return regexOrString?.toString()?.slice(1, -1);
+  } else return undefined;
+}