fix output
This commit is contained in:
+14
@@ -0,0 +1,14 @@
|
||||
'use strict';
|
||||
|
||||
import {validate} from './validator.js';
|
||||
import XMLParser from './xmlparser/XMLParser.js';
|
||||
import XMLBuilder from './xmlbuilder/json2xml.js';
|
||||
|
||||
const XMLValidator = {
|
||||
validate: validate
|
||||
}
|
||||
export {
|
||||
XMLParser,
|
||||
XMLValidator,
|
||||
XMLBuilder
|
||||
};
|
||||
+18
@@ -0,0 +1,18 @@
|
||||
export default function getIgnoreAttributesFn(ignoreAttributes) {
|
||||
if (typeof ignoreAttributes === 'function') {
|
||||
return ignoreAttributes
|
||||
}
|
||||
if (Array.isArray(ignoreAttributes)) {
|
||||
return (attrName) => {
|
||||
for (const pattern of ignoreAttributes) {
|
||||
if (typeof pattern === 'string' && attrName === pattern) {
|
||||
return true
|
||||
}
|
||||
if (pattern instanceof RegExp && pattern.test(attrName)) {
|
||||
return true
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
return () => false
|
||||
}
|
||||
+68
@@ -0,0 +1,68 @@
|
||||
'use strict';
|
||||
|
||||
const nameStartChar = ':A-Za-z_\\u00C0-\\u00D6\\u00D8-\\u00F6\\u00F8-\\u02FF\\u0370-\\u037D\\u037F-\\u1FFF\\u200C-\\u200D\\u2070-\\u218F\\u2C00-\\u2FEF\\u3001-\\uD7FF\\uF900-\\uFDCF\\uFDF0-\\uFFFD';
|
||||
const nameChar = nameStartChar + '\\-.\\d\\u00B7\\u0300-\\u036F\\u203F-\\u2040';
|
||||
export const nameRegexp = '[' + nameStartChar + '][' + nameChar + ']*';
|
||||
const regexName = new RegExp('^' + nameRegexp + '$');
|
||||
|
||||
export function getAllMatches(string, regex) {
|
||||
const matches = [];
|
||||
let match = regex.exec(string);
|
||||
while (match) {
|
||||
const allmatches = [];
|
||||
allmatches.startIndex = regex.lastIndex - match[0].length;
|
||||
const len = match.length;
|
||||
for (let index = 0; index < len; index++) {
|
||||
allmatches.push(match[index]);
|
||||
}
|
||||
matches.push(allmatches);
|
||||
match = regex.exec(string);
|
||||
}
|
||||
return matches;
|
||||
}
|
||||
|
||||
export const isName = function(string) {
|
||||
const match = regexName.exec(string);
|
||||
return !(match === null || typeof match === 'undefined');
|
||||
}
|
||||
|
||||
export function isExist(v) {
|
||||
return typeof v !== 'undefined';
|
||||
}
|
||||
|
||||
export function isEmptyObject(obj) {
|
||||
return Object.keys(obj).length === 0;
|
||||
}
|
||||
|
||||
/**
|
||||
* Copy all the properties of a into b.
|
||||
* @param {*} target
|
||||
* @param {*} a
|
||||
*/
|
||||
export function merge(target, a, arrayMode) {
|
||||
if (a) {
|
||||
const keys = Object.keys(a); // will return an array of own properties
|
||||
const len = keys.length; //don't make it inline
|
||||
for (let i = 0; i < len; i++) {
|
||||
if (arrayMode === 'strict') {
|
||||
target[keys[i]] = [ a[keys[i]] ];
|
||||
} else {
|
||||
target[keys[i]] = a[keys[i]];
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
/* exports.merge =function (b,a){
|
||||
return Object.assign(b,a);
|
||||
} */
|
||||
|
||||
export function getValue(v) {
|
||||
if (exports.isExist(v)) {
|
||||
return v;
|
||||
} else {
|
||||
return '';
|
||||
}
|
||||
}
|
||||
|
||||
// const fakeCall = function(a) {return a;};
|
||||
// const fakeCallNoReturn = function() {};
|
||||
+425
@@ -0,0 +1,425 @@
|
||||
'use strict';
|
||||
|
||||
import {getAllMatches, isName} from './util.js';
|
||||
|
||||
const defaultOptions = {
|
||||
allowBooleanAttributes: false, //A tag can have attributes without any value
|
||||
unpairedTags: []
|
||||
};
|
||||
|
||||
//const tagsPattern = new RegExp("<\\/?([\\w:\\-_\.]+)\\s*\/?>","g");
|
||||
export function validate(xmlData, options) {
|
||||
options = Object.assign({}, defaultOptions, options);
|
||||
|
||||
//xmlData = xmlData.replace(/(\r\n|\n|\r)/gm,"");//make it single line
|
||||
//xmlData = xmlData.replace(/(^\s*<\?xml.*?\?>)/g,"");//Remove XML starting tag
|
||||
//xmlData = xmlData.replace(/(<!DOCTYPE[\s\w\"\.\/\-\:]+(\[.*\])*\s*>)/g,"");//Remove DOCTYPE
|
||||
const tags = [];
|
||||
let tagFound = false;
|
||||
|
||||
//indicates that the root tag has been closed (aka. depth 0 has been reached)
|
||||
let reachedRoot = false;
|
||||
|
||||
if (xmlData[0] === '\ufeff') {
|
||||
// check for byte order mark (BOM)
|
||||
xmlData = xmlData.substr(1);
|
||||
}
|
||||
|
||||
for (let i = 0; i < xmlData.length; i++) {
|
||||
|
||||
if (xmlData[i] === '<' && xmlData[i+1] === '?') {
|
||||
i+=2;
|
||||
i = readPI(xmlData,i);
|
||||
if (i.err) return i;
|
||||
}else if (xmlData[i] === '<') {
|
||||
//starting of tag
|
||||
//read until you reach to '>' avoiding any '>' in attribute value
|
||||
let tagStartPos = i;
|
||||
i++;
|
||||
|
||||
if (xmlData[i] === '!') {
|
||||
i = readCommentAndCDATA(xmlData, i);
|
||||
continue;
|
||||
} else {
|
||||
let closingTag = false;
|
||||
if (xmlData[i] === '/') {
|
||||
//closing tag
|
||||
closingTag = true;
|
||||
i++;
|
||||
}
|
||||
//read tagname
|
||||
let tagName = '';
|
||||
for (; i < xmlData.length &&
|
||||
xmlData[i] !== '>' &&
|
||||
xmlData[i] !== ' ' &&
|
||||
xmlData[i] !== '\t' &&
|
||||
xmlData[i] !== '\n' &&
|
||||
xmlData[i] !== '\r'; i++
|
||||
) {
|
||||
tagName += xmlData[i];
|
||||
}
|
||||
tagName = tagName.trim();
|
||||
//console.log(tagName);
|
||||
|
||||
if (tagName[tagName.length - 1] === '/') {
|
||||
//self closing tag without attributes
|
||||
tagName = tagName.substring(0, tagName.length - 1);
|
||||
//continue;
|
||||
i--;
|
||||
}
|
||||
if (!validateTagName(tagName)) {
|
||||
let msg;
|
||||
if (tagName.trim().length === 0) {
|
||||
msg = "Invalid space after '<'.";
|
||||
} else {
|
||||
msg = "Tag '"+tagName+"' is an invalid name.";
|
||||
}
|
||||
return getErrorObject('InvalidTag', msg, getLineNumberForPosition(xmlData, i));
|
||||
}
|
||||
|
||||
const result = readAttributeStr(xmlData, i);
|
||||
if (result === false) {
|
||||
return getErrorObject('InvalidAttr', "Attributes for '"+tagName+"' have open quote.", getLineNumberForPosition(xmlData, i));
|
||||
}
|
||||
let attrStr = result.value;
|
||||
i = result.index;
|
||||
|
||||
if (attrStr[attrStr.length - 1] === '/') {
|
||||
//self closing tag
|
||||
const attrStrStart = i - attrStr.length;
|
||||
attrStr = attrStr.substring(0, attrStr.length - 1);
|
||||
const isValid = validateAttributeString(attrStr, options);
|
||||
if (isValid === true) {
|
||||
tagFound = true;
|
||||
//continue; //text may presents after self closing tag
|
||||
} else {
|
||||
//the result from the nested function returns the position of the error within the attribute
|
||||
//in order to get the 'true' error line, we need to calculate the position where the attribute begins (i - attrStr.length) and then add the position within the attribute
|
||||
//this gives us the absolute index in the entire xml, which we can use to find the line at last
|
||||
return getErrorObject(isValid.err.code, isValid.err.msg, getLineNumberForPosition(xmlData, attrStrStart + isValid.err.line));
|
||||
}
|
||||
} else if (closingTag) {
|
||||
if (!result.tagClosed) {
|
||||
return getErrorObject('InvalidTag', "Closing tag '"+tagName+"' doesn't have proper closing.", getLineNumberForPosition(xmlData, i));
|
||||
} else if (attrStr.trim().length > 0) {
|
||||
return getErrorObject('InvalidTag', "Closing tag '"+tagName+"' can't have attributes or invalid starting.", getLineNumberForPosition(xmlData, tagStartPos));
|
||||
} else if (tags.length === 0) {
|
||||
return getErrorObject('InvalidTag', "Closing tag '"+tagName+"' has not been opened.", getLineNumberForPosition(xmlData, tagStartPos));
|
||||
} else {
|
||||
const otg = tags.pop();
|
||||
if (tagName !== otg.tagName) {
|
||||
let openPos = getLineNumberForPosition(xmlData, otg.tagStartPos);
|
||||
return getErrorObject('InvalidTag',
|
||||
"Expected closing tag '"+otg.tagName+"' (opened in line "+openPos.line+", col "+openPos.col+") instead of closing tag '"+tagName+"'.",
|
||||
getLineNumberForPosition(xmlData, tagStartPos));
|
||||
}
|
||||
|
||||
//when there are no more tags, we reached the root level.
|
||||
if (tags.length == 0) {
|
||||
reachedRoot = true;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
const isValid = validateAttributeString(attrStr, options);
|
||||
if (isValid !== true) {
|
||||
//the result from the nested function returns the position of the error within the attribute
|
||||
//in order to get the 'true' error line, we need to calculate the position where the attribute begins (i - attrStr.length) and then add the position within the attribute
|
||||
//this gives us the absolute index in the entire xml, which we can use to find the line at last
|
||||
return getErrorObject(isValid.err.code, isValid.err.msg, getLineNumberForPosition(xmlData, i - attrStr.length + isValid.err.line));
|
||||
}
|
||||
|
||||
//if the root level has been reached before ...
|
||||
if (reachedRoot === true) {
|
||||
return getErrorObject('InvalidXml', 'Multiple possible root nodes found.', getLineNumberForPosition(xmlData, i));
|
||||
} else if(options.unpairedTags.indexOf(tagName) !== -1){
|
||||
//don't push into stack
|
||||
} else {
|
||||
tags.push({tagName, tagStartPos});
|
||||
}
|
||||
tagFound = true;
|
||||
}
|
||||
|
||||
//skip tag text value
|
||||
//It may include comments and CDATA value
|
||||
for (i++; i < xmlData.length; i++) {
|
||||
if (xmlData[i] === '<') {
|
||||
if (xmlData[i + 1] === '!') {
|
||||
//comment or CADATA
|
||||
i++;
|
||||
i = readCommentAndCDATA(xmlData, i);
|
||||
continue;
|
||||
} else if (xmlData[i+1] === '?') {
|
||||
i = readPI(xmlData, ++i);
|
||||
if (i.err) return i;
|
||||
} else{
|
||||
break;
|
||||
}
|
||||
} else if (xmlData[i] === '&') {
|
||||
const afterAmp = validateAmpersand(xmlData, i);
|
||||
if (afterAmp == -1)
|
||||
return getErrorObject('InvalidChar', "char '&' is not expected.", getLineNumberForPosition(xmlData, i));
|
||||
i = afterAmp;
|
||||
}else{
|
||||
if (reachedRoot === true && !isWhiteSpace(xmlData[i])) {
|
||||
return getErrorObject('InvalidXml', "Extra text at the end", getLineNumberForPosition(xmlData, i));
|
||||
}
|
||||
}
|
||||
} //end of reading tag text value
|
||||
if (xmlData[i] === '<') {
|
||||
i--;
|
||||
}
|
||||
}
|
||||
} else {
|
||||
if ( isWhiteSpace(xmlData[i])) {
|
||||
continue;
|
||||
}
|
||||
return getErrorObject('InvalidChar', "char '"+xmlData[i]+"' is not expected.", getLineNumberForPosition(xmlData, i));
|
||||
}
|
||||
}
|
||||
|
||||
if (!tagFound) {
|
||||
return getErrorObject('InvalidXml', 'Start tag expected.', 1);
|
||||
}else if (tags.length == 1) {
|
||||
return getErrorObject('InvalidTag', "Unclosed tag '"+tags[0].tagName+"'.", getLineNumberForPosition(xmlData, tags[0].tagStartPos));
|
||||
}else if (tags.length > 0) {
|
||||
return getErrorObject('InvalidXml', "Invalid '"+
|
||||
JSON.stringify(tags.map(t => t.tagName), null, 4).replace(/\r?\n/g, '')+
|
||||
"' found.", {line: 1, col: 1});
|
||||
}
|
||||
|
||||
return true;
|
||||
};
|
||||
|
||||
function isWhiteSpace(char){
|
||||
return char === ' ' || char === '\t' || char === '\n' || char === '\r';
|
||||
}
|
||||
/**
|
||||
* Read Processing insstructions and skip
|
||||
* @param {*} xmlData
|
||||
* @param {*} i
|
||||
*/
|
||||
function readPI(xmlData, i) {
|
||||
const start = i;
|
||||
for (; i < xmlData.length; i++) {
|
||||
if (xmlData[i] == '?' || xmlData[i] == ' ') {
|
||||
//tagname
|
||||
const tagname = xmlData.substr(start, i - start);
|
||||
if (i > 5 && tagname === 'xml') {
|
||||
return getErrorObject('InvalidXml', 'XML declaration allowed only at the start of the document.', getLineNumberForPosition(xmlData, i));
|
||||
} else if (xmlData[i] == '?' && xmlData[i + 1] == '>') {
|
||||
//check if valid attribut string
|
||||
i++;
|
||||
break;
|
||||
} else {
|
||||
continue;
|
||||
}
|
||||
}
|
||||
}
|
||||
return i;
|
||||
}
|
||||
|
||||
function readCommentAndCDATA(xmlData, i) {
|
||||
if (xmlData.length > i + 5 && xmlData[i + 1] === '-' && xmlData[i + 2] === '-') {
|
||||
//comment
|
||||
for (i += 3; i < xmlData.length; i++) {
|
||||
if (xmlData[i] === '-' && xmlData[i + 1] === '-' && xmlData[i + 2] === '>') {
|
||||
i += 2;
|
||||
break;
|
||||
}
|
||||
}
|
||||
} else if (
|
||||
xmlData.length > i + 8 &&
|
||||
xmlData[i + 1] === 'D' &&
|
||||
xmlData[i + 2] === 'O' &&
|
||||
xmlData[i + 3] === 'C' &&
|
||||
xmlData[i + 4] === 'T' &&
|
||||
xmlData[i + 5] === 'Y' &&
|
||||
xmlData[i + 6] === 'P' &&
|
||||
xmlData[i + 7] === 'E'
|
||||
) {
|
||||
let angleBracketsCount = 1;
|
||||
for (i += 8; i < xmlData.length; i++) {
|
||||
if (xmlData[i] === '<') {
|
||||
angleBracketsCount++;
|
||||
} else if (xmlData[i] === '>') {
|
||||
angleBracketsCount--;
|
||||
if (angleBracketsCount === 0) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
} else if (
|
||||
xmlData.length > i + 9 &&
|
||||
xmlData[i + 1] === '[' &&
|
||||
xmlData[i + 2] === 'C' &&
|
||||
xmlData[i + 3] === 'D' &&
|
||||
xmlData[i + 4] === 'A' &&
|
||||
xmlData[i + 5] === 'T' &&
|
||||
xmlData[i + 6] === 'A' &&
|
||||
xmlData[i + 7] === '['
|
||||
) {
|
||||
for (i += 8; i < xmlData.length; i++) {
|
||||
if (xmlData[i] === ']' && xmlData[i + 1] === ']' && xmlData[i + 2] === '>') {
|
||||
i += 2;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return i;
|
||||
}
|
||||
|
||||
const doubleQuote = '"';
|
||||
const singleQuote = "'";
|
||||
|
||||
/**
|
||||
* Keep reading xmlData until '<' is found outside the attribute value.
|
||||
* @param {string} xmlData
|
||||
* @param {number} i
|
||||
*/
|
||||
function readAttributeStr(xmlData, i) {
|
||||
let attrStr = '';
|
||||
let startChar = '';
|
||||
let tagClosed = false;
|
||||
for (; i < xmlData.length; i++) {
|
||||
if (xmlData[i] === doubleQuote || xmlData[i] === singleQuote) {
|
||||
if (startChar === '') {
|
||||
startChar = xmlData[i];
|
||||
} else if (startChar !== xmlData[i]) {
|
||||
//if vaue is enclosed with double quote then single quotes are allowed inside the value and vice versa
|
||||
} else {
|
||||
startChar = '';
|
||||
}
|
||||
} else if (xmlData[i] === '>') {
|
||||
if (startChar === '') {
|
||||
tagClosed = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
attrStr += xmlData[i];
|
||||
}
|
||||
if (startChar !== '') {
|
||||
return false;
|
||||
}
|
||||
|
||||
return {
|
||||
value: attrStr,
|
||||
index: i,
|
||||
tagClosed: tagClosed
|
||||
};
|
||||
}
|
||||
|
||||
/**
|
||||
* Select all the attributes whether valid or invalid.
|
||||
*/
|
||||
const validAttrStrRegxp = new RegExp('(\\s*)([^\\s=]+)(\\s*=)?(\\s*([\'"])(([\\s\\S])*?)\\5)?', 'g');
|
||||
|
||||
//attr, ="sd", a="amit's", a="sd"b="saf", ab cd=""
|
||||
|
||||
function validateAttributeString(attrStr, options) {
|
||||
//console.log("start:"+attrStr+":end");
|
||||
|
||||
//if(attrStr.trim().length === 0) return true; //empty string
|
||||
|
||||
const matches = getAllMatches(attrStr, validAttrStrRegxp);
|
||||
const attrNames = {};
|
||||
|
||||
for (let i = 0; i < matches.length; i++) {
|
||||
if (matches[i][1].length === 0) {
|
||||
//nospace before attribute name: a="sd"b="saf"
|
||||
return getErrorObject('InvalidAttr', "Attribute '"+matches[i][2]+"' has no space in starting.", getPositionFromMatch(matches[i]))
|
||||
} else if (matches[i][3] !== undefined && matches[i][4] === undefined) {
|
||||
return getErrorObject('InvalidAttr', "Attribute '"+matches[i][2]+"' is without value.", getPositionFromMatch(matches[i]));
|
||||
} else if (matches[i][3] === undefined && !options.allowBooleanAttributes) {
|
||||
//independent attribute: ab
|
||||
return getErrorObject('InvalidAttr', "boolean attribute '"+matches[i][2]+"' is not allowed.", getPositionFromMatch(matches[i]));
|
||||
}
|
||||
/* else if(matches[i][6] === undefined){//attribute without value: ab=
|
||||
return { err: { code:"InvalidAttr",msg:"attribute " + matches[i][2] + " has no value assigned."}};
|
||||
} */
|
||||
const attrName = matches[i][2];
|
||||
if (!validateAttrName(attrName)) {
|
||||
return getErrorObject('InvalidAttr', "Attribute '"+attrName+"' is an invalid name.", getPositionFromMatch(matches[i]));
|
||||
}
|
||||
if (!attrNames.hasOwnProperty(attrName)) {
|
||||
//check for duplicate attribute.
|
||||
attrNames[attrName] = 1;
|
||||
} else {
|
||||
return getErrorObject('InvalidAttr', "Attribute '"+attrName+"' is repeated.", getPositionFromMatch(matches[i]));
|
||||
}
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
function validateNumberAmpersand(xmlData, i) {
|
||||
let re = /\d/;
|
||||
if (xmlData[i] === 'x') {
|
||||
i++;
|
||||
re = /[\da-fA-F]/;
|
||||
}
|
||||
for (; i < xmlData.length; i++) {
|
||||
if (xmlData[i] === ';')
|
||||
return i;
|
||||
if (!xmlData[i].match(re))
|
||||
break;
|
||||
}
|
||||
return -1;
|
||||
}
|
||||
|
||||
function validateAmpersand(xmlData, i) {
|
||||
// https://www.w3.org/TR/xml/#dt-charref
|
||||
i++;
|
||||
if (xmlData[i] === ';')
|
||||
return -1;
|
||||
if (xmlData[i] === '#') {
|
||||
i++;
|
||||
return validateNumberAmpersand(xmlData, i);
|
||||
}
|
||||
let count = 0;
|
||||
for (; i < xmlData.length; i++, count++) {
|
||||
if (xmlData[i].match(/\w/) && count < 20)
|
||||
continue;
|
||||
if (xmlData[i] === ';')
|
||||
break;
|
||||
return -1;
|
||||
}
|
||||
return i;
|
||||
}
|
||||
|
||||
function getErrorObject(code, message, lineNumber) {
|
||||
return {
|
||||
err: {
|
||||
code: code,
|
||||
msg: message,
|
||||
line: lineNumber.line || lineNumber,
|
||||
col: lineNumber.col,
|
||||
},
|
||||
};
|
||||
}
|
||||
|
||||
function validateAttrName(attrName) {
|
||||
return isName(attrName);
|
||||
}
|
||||
|
||||
// const startsWithXML = /^xml/i;
|
||||
|
||||
function validateTagName(tagname) {
|
||||
return isName(tagname) /* && !tagname.match(startsWithXML) */;
|
||||
}
|
||||
|
||||
//this function returns the line number for the character at the given index
|
||||
function getLineNumberForPosition(xmlData, index) {
|
||||
const lines = xmlData.substring(0, index).split(/\r?\n/);
|
||||
return {
|
||||
line: lines.length,
|
||||
|
||||
// column number is last line's length + 1, because column numbering starts at 1:
|
||||
col: lines[lines.length - 1].length + 1
|
||||
};
|
||||
}
|
||||
|
||||
//this function returns the position of the first character of match within attrStr
|
||||
function getPositionFromMatch(match) {
|
||||
return match.startIndex + match[1].length;
|
||||
}
|
||||
+285
@@ -0,0 +1,285 @@
|
||||
'use strict';
|
||||
//parse Empty Node as self closing node
|
||||
import buildFromOrderedJs from './orderedJs2Xml.js';
|
||||
import getIgnoreAttributesFn from "../ignoreAttributes.js";
|
||||
|
||||
const defaultOptions = {
|
||||
attributeNamePrefix: '@_',
|
||||
attributesGroupName: false,
|
||||
textNodeName: '#text',
|
||||
ignoreAttributes: true,
|
||||
cdataPropName: false,
|
||||
format: false,
|
||||
indentBy: ' ',
|
||||
suppressEmptyNode: false,
|
||||
suppressUnpairedNode: true,
|
||||
suppressBooleanAttributes: true,
|
||||
tagValueProcessor: function(key, a) {
|
||||
return a;
|
||||
},
|
||||
attributeValueProcessor: function(attrName, a) {
|
||||
return a;
|
||||
},
|
||||
preserveOrder: false,
|
||||
commentPropName: false,
|
||||
unpairedTags: [],
|
||||
entities: [
|
||||
{ regex: new RegExp("&", "g"), val: "&" },//it must be on top
|
||||
{ regex: new RegExp(">", "g"), val: ">" },
|
||||
{ regex: new RegExp("<", "g"), val: "<" },
|
||||
{ regex: new RegExp("\'", "g"), val: "'" },
|
||||
{ regex: new RegExp("\"", "g"), val: """ }
|
||||
],
|
||||
processEntities: true,
|
||||
stopNodes: [],
|
||||
// transformTagName: false,
|
||||
// transformAttributeName: false,
|
||||
oneListGroup: false
|
||||
};
|
||||
|
||||
export default function Builder(options) {
|
||||
this.options = Object.assign({}, defaultOptions, options);
|
||||
if (this.options.ignoreAttributes === true || this.options.attributesGroupName) {
|
||||
this.isAttribute = function(/*a*/) {
|
||||
return false;
|
||||
};
|
||||
} else {
|
||||
this.ignoreAttributesFn = getIgnoreAttributesFn(this.options.ignoreAttributes)
|
||||
this.attrPrefixLen = this.options.attributeNamePrefix.length;
|
||||
this.isAttribute = isAttribute;
|
||||
}
|
||||
|
||||
this.processTextOrObjNode = processTextOrObjNode
|
||||
|
||||
if (this.options.format) {
|
||||
this.indentate = indentate;
|
||||
this.tagEndChar = '>\n';
|
||||
this.newLine = '\n';
|
||||
} else {
|
||||
this.indentate = function() {
|
||||
return '';
|
||||
};
|
||||
this.tagEndChar = '>';
|
||||
this.newLine = '';
|
||||
}
|
||||
}
|
||||
|
||||
Builder.prototype.build = function(jObj) {
|
||||
if(this.options.preserveOrder){
|
||||
return buildFromOrderedJs(jObj, this.options);
|
||||
}else {
|
||||
if(Array.isArray(jObj) && this.options.arrayNodeName && this.options.arrayNodeName.length > 1){
|
||||
jObj = {
|
||||
[this.options.arrayNodeName] : jObj
|
||||
}
|
||||
}
|
||||
return this.j2x(jObj, 0, []).val;
|
||||
}
|
||||
};
|
||||
|
||||
Builder.prototype.j2x = function(jObj, level, ajPath) {
|
||||
let attrStr = '';
|
||||
let val = '';
|
||||
const jPath = ajPath.join('.')
|
||||
for (let key in jObj) {
|
||||
if(!Object.prototype.hasOwnProperty.call(jObj, key)) continue;
|
||||
if (typeof jObj[key] === 'undefined') {
|
||||
// supress undefined node only if it is not an attribute
|
||||
if (this.isAttribute(key)) {
|
||||
val += '';
|
||||
}
|
||||
} else if (jObj[key] === null) {
|
||||
// null attribute should be ignored by the attribute list, but should not cause the tag closing
|
||||
if (this.isAttribute(key)) {
|
||||
val += '';
|
||||
} else if (key === this.options.cdataPropName) {
|
||||
val += '';
|
||||
} else if (key[0] === '?') {
|
||||
val += this.indentate(level) + '<' + key + '?' + this.tagEndChar;
|
||||
} else {
|
||||
val += this.indentate(level) + '<' + key + '/' + this.tagEndChar;
|
||||
}
|
||||
// val += this.indentate(level) + '<' + key + '/' + this.tagEndChar;
|
||||
} else if (jObj[key] instanceof Date) {
|
||||
val += this.buildTextValNode(jObj[key], key, '', level);
|
||||
} else if (typeof jObj[key] !== 'object') {
|
||||
//premitive type
|
||||
const attr = this.isAttribute(key);
|
||||
if (attr && !this.ignoreAttributesFn(attr, jPath)) {
|
||||
attrStr += this.buildAttrPairStr(attr, '' + jObj[key]);
|
||||
} else if (!attr) {
|
||||
//tag value
|
||||
if (key === this.options.textNodeName) {
|
||||
let newval = this.options.tagValueProcessor(key, '' + jObj[key]);
|
||||
val += this.replaceEntitiesValue(newval);
|
||||
} else {
|
||||
val += this.buildTextValNode(jObj[key], key, '', level);
|
||||
}
|
||||
}
|
||||
} else if (Array.isArray(jObj[key])) {
|
||||
//repeated nodes
|
||||
const arrLen = jObj[key].length;
|
||||
let listTagVal = "";
|
||||
let listTagAttr = "";
|
||||
for (let j = 0; j < arrLen; j++) {
|
||||
const item = jObj[key][j];
|
||||
if (typeof item === 'undefined') {
|
||||
// supress undefined node
|
||||
} else if (item === null) {
|
||||
if(key[0] === "?") val += this.indentate(level) + '<' + key + '?' + this.tagEndChar;
|
||||
else val += this.indentate(level) + '<' + key + '/' + this.tagEndChar;
|
||||
// val += this.indentate(level) + '<' + key + '/' + this.tagEndChar;
|
||||
} else if (typeof item === 'object') {
|
||||
if(this.options.oneListGroup){
|
||||
const result = this.j2x(item, level + 1, ajPath.concat(key));
|
||||
listTagVal += result.val;
|
||||
if (this.options.attributesGroupName && item.hasOwnProperty(this.options.attributesGroupName)) {
|
||||
listTagAttr += result.attrStr
|
||||
}
|
||||
}else{
|
||||
listTagVal += this.processTextOrObjNode(item, key, level, ajPath)
|
||||
}
|
||||
} else {
|
||||
if (this.options.oneListGroup) {
|
||||
let textValue = this.options.tagValueProcessor(key, item);
|
||||
textValue = this.replaceEntitiesValue(textValue);
|
||||
listTagVal += textValue;
|
||||
} else {
|
||||
listTagVal += this.buildTextValNode(item, key, '', level);
|
||||
}
|
||||
}
|
||||
}
|
||||
if(this.options.oneListGroup){
|
||||
listTagVal = this.buildObjectNode(listTagVal, key, listTagAttr, level);
|
||||
}
|
||||
val += listTagVal;
|
||||
} else {
|
||||
//nested node
|
||||
if (this.options.attributesGroupName && key === this.options.attributesGroupName) {
|
||||
const Ks = Object.keys(jObj[key]);
|
||||
const L = Ks.length;
|
||||
for (let j = 0; j < L; j++) {
|
||||
attrStr += this.buildAttrPairStr(Ks[j], '' + jObj[key][Ks[j]]);
|
||||
}
|
||||
} else {
|
||||
val += this.processTextOrObjNode(jObj[key], key, level, ajPath)
|
||||
}
|
||||
}
|
||||
}
|
||||
return {attrStr: attrStr, val: val};
|
||||
};
|
||||
|
||||
Builder.prototype.buildAttrPairStr = function(attrName, val){
|
||||
val = this.options.attributeValueProcessor(attrName, '' + val);
|
||||
val = this.replaceEntitiesValue(val);
|
||||
if (this.options.suppressBooleanAttributes && val === "true") {
|
||||
return ' ' + attrName;
|
||||
} else return ' ' + attrName + '="' + val + '"';
|
||||
}
|
||||
|
||||
function processTextOrObjNode (object, key, level, ajPath) {
|
||||
const result = this.j2x(object, level + 1, ajPath.concat(key));
|
||||
if (object[this.options.textNodeName] !== undefined && Object.keys(object).length === 1) {
|
||||
return this.buildTextValNode(object[this.options.textNodeName], key, result.attrStr, level);
|
||||
} else {
|
||||
return this.buildObjectNode(result.val, key, result.attrStr, level);
|
||||
}
|
||||
}
|
||||
|
||||
Builder.prototype.buildObjectNode = function(val, key, attrStr, level) {
|
||||
if(val === ""){
|
||||
if(key[0] === "?") return this.indentate(level) + '<' + key + attrStr+ '?' + this.tagEndChar;
|
||||
else {
|
||||
return this.indentate(level) + '<' + key + attrStr + this.closeTag(key) + this.tagEndChar;
|
||||
}
|
||||
}else{
|
||||
|
||||
let tagEndExp = '</' + key + this.tagEndChar;
|
||||
let piClosingChar = "";
|
||||
|
||||
if(key[0] === "?") {
|
||||
piClosingChar = "?";
|
||||
tagEndExp = "";
|
||||
}
|
||||
|
||||
// attrStr is an empty string in case the attribute came as undefined or null
|
||||
if ((attrStr || attrStr === '') && val.indexOf('<') === -1) {
|
||||
return ( this.indentate(level) + '<' + key + attrStr + piClosingChar + '>' + val + tagEndExp );
|
||||
} else if (this.options.commentPropName !== false && key === this.options.commentPropName && piClosingChar.length === 0) {
|
||||
return this.indentate(level) + `<!--${val}-->` + this.newLine;
|
||||
}else {
|
||||
return (
|
||||
this.indentate(level) + '<' + key + attrStr + piClosingChar + this.tagEndChar +
|
||||
val +
|
||||
this.indentate(level) + tagEndExp );
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Builder.prototype.closeTag = function(key){
|
||||
let closeTag = "";
|
||||
if(this.options.unpairedTags.indexOf(key) !== -1){ //unpaired
|
||||
if(!this.options.suppressUnpairedNode) closeTag = "/"
|
||||
}else if(this.options.suppressEmptyNode){ //empty
|
||||
closeTag = "/";
|
||||
}else{
|
||||
closeTag = `></${key}`
|
||||
}
|
||||
return closeTag;
|
||||
}
|
||||
|
||||
function buildEmptyObjNode(val, key, attrStr, level) {
|
||||
if (val !== '') {
|
||||
return this.buildObjectNode(val, key, attrStr, level);
|
||||
} else {
|
||||
if(key[0] === "?") return this.indentate(level) + '<' + key + attrStr+ '?' + this.tagEndChar;
|
||||
else {
|
||||
return this.indentate(level) + '<' + key + attrStr + '/' + this.tagEndChar;
|
||||
// return this.buildTagStr(level,key, attrStr);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Builder.prototype.buildTextValNode = function(val, key, attrStr, level) {
|
||||
if (this.options.cdataPropName !== false && key === this.options.cdataPropName) {
|
||||
return this.indentate(level) + `<![CDATA[${val}]]>` + this.newLine;
|
||||
}else if (this.options.commentPropName !== false && key === this.options.commentPropName) {
|
||||
return this.indentate(level) + `<!--${val}-->` + this.newLine;
|
||||
}else if(key[0] === "?") {//PI tag
|
||||
return this.indentate(level) + '<' + key + attrStr+ '?' + this.tagEndChar;
|
||||
}else{
|
||||
let textValue = this.options.tagValueProcessor(key, val);
|
||||
textValue = this.replaceEntitiesValue(textValue);
|
||||
|
||||
if( textValue === ''){
|
||||
return this.indentate(level) + '<' + key + attrStr + this.closeTag(key) + this.tagEndChar;
|
||||
}else{
|
||||
return this.indentate(level) + '<' + key + attrStr + '>' +
|
||||
textValue +
|
||||
'</' + key + this.tagEndChar;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Builder.prototype.replaceEntitiesValue = function(textValue){
|
||||
if(textValue && textValue.length > 0 && this.options.processEntities){
|
||||
for (let i=0; i<this.options.entities.length; i++) {
|
||||
const entity = this.options.entities[i];
|
||||
textValue = textValue.replace(entity.regex, entity.val);
|
||||
}
|
||||
}
|
||||
return textValue;
|
||||
}
|
||||
|
||||
function indentate(level) {
|
||||
return this.options.indentBy.repeat(level);
|
||||
}
|
||||
|
||||
function isAttribute(name /*, options*/) {
|
||||
if (name.startsWith(this.options.attributeNamePrefix) && name !== this.options.textNodeName) {
|
||||
return name.substr(this.attrPrefixLen);
|
||||
} else {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
+134
@@ -0,0 +1,134 @@
|
||||
const EOL = "\n";
|
||||
|
||||
/**
|
||||
*
|
||||
* @param {array} jArray
|
||||
* @param {any} options
|
||||
* @returns
|
||||
*/
|
||||
export default function toXml(jArray, options) {
|
||||
let indentation = "";
|
||||
if (options.format && options.indentBy.length > 0) {
|
||||
indentation = EOL;
|
||||
}
|
||||
return arrToStr(jArray, options, "", indentation);
|
||||
}
|
||||
|
||||
function arrToStr(arr, options, jPath, indentation) {
|
||||
let xmlStr = "";
|
||||
let isPreviousElementTag = false;
|
||||
|
||||
for (let i = 0; i < arr.length; i++) {
|
||||
const tagObj = arr[i];
|
||||
const tagName = propName(tagObj);
|
||||
if(tagName === undefined) continue;
|
||||
|
||||
let newJPath = "";
|
||||
if (jPath.length === 0) newJPath = tagName
|
||||
else newJPath = `${jPath}.${tagName}`;
|
||||
|
||||
if (tagName === options.textNodeName) {
|
||||
let tagText = tagObj[tagName];
|
||||
if (!isStopNode(newJPath, options)) {
|
||||
tagText = options.tagValueProcessor(tagName, tagText);
|
||||
tagText = replaceEntitiesValue(tagText, options);
|
||||
}
|
||||
if (isPreviousElementTag) {
|
||||
xmlStr += indentation;
|
||||
}
|
||||
xmlStr += tagText;
|
||||
isPreviousElementTag = false;
|
||||
continue;
|
||||
} else if (tagName === options.cdataPropName) {
|
||||
if (isPreviousElementTag) {
|
||||
xmlStr += indentation;
|
||||
}
|
||||
xmlStr += `<![CDATA[${tagObj[tagName][0][options.textNodeName]}]]>`;
|
||||
isPreviousElementTag = false;
|
||||
continue;
|
||||
} else if (tagName === options.commentPropName) {
|
||||
xmlStr += indentation + `<!--${tagObj[tagName][0][options.textNodeName]}-->`;
|
||||
isPreviousElementTag = true;
|
||||
continue;
|
||||
} else if (tagName[0] === "?") {
|
||||
const attStr = attr_to_str(tagObj[":@"], options);
|
||||
const tempInd = tagName === "?xml" ? "" : indentation;
|
||||
let piTextNodeName = tagObj[tagName][0][options.textNodeName];
|
||||
piTextNodeName = piTextNodeName.length !== 0 ? " " + piTextNodeName : ""; //remove extra spacing
|
||||
xmlStr += tempInd + `<${tagName}${piTextNodeName}${attStr}?>`;
|
||||
isPreviousElementTag = true;
|
||||
continue;
|
||||
}
|
||||
let newIdentation = indentation;
|
||||
if (newIdentation !== "") {
|
||||
newIdentation += options.indentBy;
|
||||
}
|
||||
const attStr = attr_to_str(tagObj[":@"], options);
|
||||
const tagStart = indentation + `<${tagName}${attStr}`;
|
||||
const tagValue = arrToStr(tagObj[tagName], options, newJPath, newIdentation);
|
||||
if (options.unpairedTags.indexOf(tagName) !== -1) {
|
||||
if (options.suppressUnpairedNode) xmlStr += tagStart + ">";
|
||||
else xmlStr += tagStart + "/>";
|
||||
} else if ((!tagValue || tagValue.length === 0) && options.suppressEmptyNode) {
|
||||
xmlStr += tagStart + "/>";
|
||||
} else if (tagValue && tagValue.endsWith(">")) {
|
||||
xmlStr += tagStart + `>${tagValue}${indentation}</${tagName}>`;
|
||||
} else {
|
||||
xmlStr += tagStart + ">";
|
||||
if (tagValue && indentation !== "" && (tagValue.includes("/>") || tagValue.includes("</"))) {
|
||||
xmlStr += indentation + options.indentBy + tagValue + indentation;
|
||||
} else {
|
||||
xmlStr += tagValue;
|
||||
}
|
||||
xmlStr += `</${tagName}>`;
|
||||
}
|
||||
isPreviousElementTag = true;
|
||||
}
|
||||
|
||||
return xmlStr;
|
||||
}
|
||||
|
||||
function propName(obj) {
|
||||
const keys = Object.keys(obj);
|
||||
for (let i = 0; i < keys.length; i++) {
|
||||
const key = keys[i];
|
||||
if(!obj.hasOwnProperty(key)) continue;
|
||||
if (key !== ":@") return key;
|
||||
}
|
||||
}
|
||||
|
||||
function attr_to_str(attrMap, options) {
|
||||
let attrStr = "";
|
||||
if (attrMap && !options.ignoreAttributes) {
|
||||
for (let attr in attrMap) {
|
||||
if(!attrMap.hasOwnProperty(attr)) continue;
|
||||
let attrVal = options.attributeValueProcessor(attr, attrMap[attr]);
|
||||
attrVal = replaceEntitiesValue(attrVal, options);
|
||||
if (attrVal === true && options.suppressBooleanAttributes) {
|
||||
attrStr += ` ${attr.substr(options.attributeNamePrefix.length)}`;
|
||||
} else {
|
||||
attrStr += ` ${attr.substr(options.attributeNamePrefix.length)}="${attrVal}"`;
|
||||
}
|
||||
}
|
||||
}
|
||||
return attrStr;
|
||||
}
|
||||
|
||||
function isStopNode(jPath, options) {
|
||||
jPath = jPath.substr(0, jPath.length - options.textNodeName.length - 1);
|
||||
let tagName = jPath.substr(jPath.lastIndexOf(".") + 1);
|
||||
for (let index in options.stopNodes) {
|
||||
if (options.stopNodes[index] === jPath || options.stopNodes[index] === "*." + tagName) return true;
|
||||
}
|
||||
return false;
|
||||
}
|
||||
|
||||
function replaceEntitiesValue(textValue, options) {
|
||||
if (textValue && textValue.length > 0 && options.processEntities) {
|
||||
for (let i = 0; i < options.entities.length; i++) {
|
||||
const entity = options.entities[i];
|
||||
textValue = textValue.replace(entity.regex, entity.val);
|
||||
}
|
||||
}
|
||||
return textValue;
|
||||
}
|
||||
+379
@@ -0,0 +1,379 @@
|
||||
import {isName} from '../util.js';
|
||||
|
||||
export default class DocTypeReader{
|
||||
constructor(processEntities){
|
||||
this.suppressValidationErr = !processEntities;
|
||||
}
|
||||
|
||||
readDocType(xmlData, i){
|
||||
|
||||
const entities = {};
|
||||
if( xmlData[i + 3] === 'O' &&
|
||||
xmlData[i + 4] === 'C' &&
|
||||
xmlData[i + 5] === 'T' &&
|
||||
xmlData[i + 6] === 'Y' &&
|
||||
xmlData[i + 7] === 'P' &&
|
||||
xmlData[i + 8] === 'E')
|
||||
{
|
||||
i = i+9;
|
||||
let angleBracketsCount = 1;
|
||||
let hasBody = false, comment = false;
|
||||
let exp = "";
|
||||
for(;i<xmlData.length;i++){
|
||||
if (xmlData[i] === '<' && !comment) { //Determine the tag type
|
||||
if( hasBody && hasSeq(xmlData, "!ENTITY",i)){
|
||||
i += 7;
|
||||
let entityName, val;
|
||||
[entityName, val,i] = this.readEntityExp(xmlData,i+1,this.suppressValidationErr);
|
||||
if(val.indexOf("&") === -1) //Parameter entities are not supported
|
||||
entities[ entityName ] = {
|
||||
regx : RegExp( `&${entityName};`,"g"),
|
||||
val: val
|
||||
};
|
||||
}
|
||||
else if( hasBody && hasSeq(xmlData, "!ELEMENT",i)) {
|
||||
i += 8;//Not supported
|
||||
const {index} = this.readElementExp(xmlData,i+1);
|
||||
i = index;
|
||||
}else if( hasBody && hasSeq(xmlData, "!ATTLIST",i)){
|
||||
i += 8;//Not supported
|
||||
// const {index} = this.readAttlistExp(xmlData,i+1);
|
||||
// i = index;
|
||||
}else if( hasBody && hasSeq(xmlData, "!NOTATION",i)) {
|
||||
i += 9;//Not supported
|
||||
const {index} = this.readNotationExp(xmlData,i+1,this.suppressValidationErr);
|
||||
i = index;
|
||||
}else if( hasSeq(xmlData, "!--",i) ) comment = true;
|
||||
else throw new Error(`Invalid DOCTYPE`);
|
||||
|
||||
angleBracketsCount++;
|
||||
exp = "";
|
||||
} else if (xmlData[i] === '>') { //Read tag content
|
||||
if(comment){
|
||||
if( xmlData[i - 1] === "-" && xmlData[i - 2] === "-"){
|
||||
comment = false;
|
||||
angleBracketsCount--;
|
||||
}
|
||||
}else{
|
||||
angleBracketsCount--;
|
||||
}
|
||||
if (angleBracketsCount === 0) {
|
||||
break;
|
||||
}
|
||||
}else if( xmlData[i] === '['){
|
||||
hasBody = true;
|
||||
}else{
|
||||
exp += xmlData[i];
|
||||
}
|
||||
}
|
||||
if(angleBracketsCount !== 0){
|
||||
throw new Error(`Unclosed DOCTYPE`);
|
||||
}
|
||||
}else{
|
||||
throw new Error(`Invalid Tag instead of DOCTYPE`);
|
||||
}
|
||||
return {entities, i};
|
||||
}
|
||||
readEntityExp(xmlData, i) {
|
||||
//External entities are not supported
|
||||
// <!ENTITY ext SYSTEM "http://normal-website.com" >
|
||||
|
||||
//Parameter entities are not supported
|
||||
// <!ENTITY entityname "&anotherElement;">
|
||||
|
||||
//Internal entities are supported
|
||||
// <!ENTITY entityname "replacement text">
|
||||
|
||||
// Skip leading whitespace after <!ENTITY
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Read entity name
|
||||
let entityName = "";
|
||||
while (i < xmlData.length && !/\s/.test(xmlData[i]) && xmlData[i] !== '"' && xmlData[i] !== "'") {
|
||||
entityName += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
validateEntityName(entityName);
|
||||
|
||||
// Skip whitespace after entity name
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Check for unsupported constructs (external entities or parameter entities)
|
||||
if(!this.suppressValidationErr){
|
||||
if (xmlData.substring(i, i + 6).toUpperCase() === "SYSTEM") {
|
||||
throw new Error("External entities are not supported");
|
||||
}else if (xmlData[i] === "%") {
|
||||
throw new Error("Parameter entities are not supported");
|
||||
}
|
||||
}
|
||||
|
||||
// Read entity value (internal entity)
|
||||
let entityValue = "";
|
||||
[i, entityValue] = this.readIdentifierVal(xmlData, i, "entity");
|
||||
i--;
|
||||
return [entityName, entityValue, i ];
|
||||
}
|
||||
|
||||
readNotationExp(xmlData, i) {
|
||||
// Skip leading whitespace after <!NOTATION
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Read notation name
|
||||
let notationName = "";
|
||||
while (i < xmlData.length && !/\s/.test(xmlData[i])) {
|
||||
notationName += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
!this.suppressValidationErr && validateEntityName(notationName);
|
||||
|
||||
// Skip whitespace after notation name
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Check identifier type (SYSTEM or PUBLIC)
|
||||
const identifierType = xmlData.substring(i, i + 6).toUpperCase();
|
||||
if (!this.suppressValidationErr && identifierType !== "SYSTEM" && identifierType !== "PUBLIC") {
|
||||
throw new Error(`Expected SYSTEM or PUBLIC, found "${identifierType}"`);
|
||||
}
|
||||
i += identifierType.length;
|
||||
|
||||
// Skip whitespace after identifier type
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Read public identifier (if PUBLIC)
|
||||
let publicIdentifier = null;
|
||||
let systemIdentifier = null;
|
||||
|
||||
if (identifierType === "PUBLIC") {
|
||||
[i, publicIdentifier ] = this.readIdentifierVal(xmlData, i, "publicIdentifier");
|
||||
|
||||
// Skip whitespace after public identifier
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Optionally read system identifier
|
||||
if (xmlData[i] === '"' || xmlData[i] === "'") {
|
||||
[i, systemIdentifier ] = this.readIdentifierVal(xmlData, i,"systemIdentifier");
|
||||
}
|
||||
} else if (identifierType === "SYSTEM") {
|
||||
// Read system identifier (mandatory for SYSTEM)
|
||||
[i, systemIdentifier ] = this.readIdentifierVal(xmlData, i, "systemIdentifier");
|
||||
|
||||
if (!this.suppressValidationErr && !systemIdentifier) {
|
||||
throw new Error("Missing mandatory system identifier for SYSTEM notation");
|
||||
}
|
||||
}
|
||||
|
||||
return {notationName, publicIdentifier, systemIdentifier, index: --i};
|
||||
}
|
||||
|
||||
readIdentifierVal(xmlData, i, type) {
|
||||
let identifierVal = "";
|
||||
const startChar = xmlData[i];
|
||||
if (startChar !== '"' && startChar !== "'") {
|
||||
throw new Error(`Expected quoted string, found "${startChar}"`);
|
||||
}
|
||||
i++;
|
||||
|
||||
while (i < xmlData.length && xmlData[i] !== startChar) {
|
||||
identifierVal += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
|
||||
if (xmlData[i] !== startChar) {
|
||||
throw new Error(`Unterminated ${type} value`);
|
||||
}
|
||||
i++;
|
||||
return [i, identifierVal];
|
||||
}
|
||||
|
||||
readElementExp(xmlData, i) {
|
||||
// <!ELEMENT br EMPTY>
|
||||
// <!ELEMENT div ANY>
|
||||
// <!ELEMENT title (#PCDATA)>
|
||||
// <!ELEMENT book (title, author+)>
|
||||
// <!ELEMENT name (content-model)>
|
||||
|
||||
// Skip leading whitespace after <!ELEMENT
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Read element name
|
||||
let elementName = "";
|
||||
while (i < xmlData.length && !/\s/.test(xmlData[i])) {
|
||||
elementName += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
|
||||
// Validate element name
|
||||
if (!this.suppressValidationErr && !isName(elementName)) {
|
||||
throw new Error(`Invalid element name: "${elementName}"`);
|
||||
}
|
||||
|
||||
// Skip whitespace after element name
|
||||
i = skipWhitespace(xmlData, i);
|
||||
let contentModel = "";
|
||||
// Expect '(' to start content model
|
||||
if(xmlData[i] === "E" && hasSeq(xmlData, "MPTY",i)) i+=4;
|
||||
else if(xmlData[i] === "A" && hasSeq(xmlData, "NY",i)) i+=2;
|
||||
else if (xmlData[i] === "(") {
|
||||
i++; // Move past '('
|
||||
|
||||
// Read content model
|
||||
while (i < xmlData.length && xmlData[i] !== ")") {
|
||||
contentModel += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
if (xmlData[i] !== ")") {
|
||||
throw new Error("Unterminated content model");
|
||||
}
|
||||
|
||||
}else if(!this.suppressValidationErr){
|
||||
throw new Error(`Invalid Element Expression, found "${xmlData[i]}"`);
|
||||
}
|
||||
|
||||
return {
|
||||
elementName,
|
||||
contentModel: contentModel.trim(),
|
||||
index: i
|
||||
};
|
||||
}
|
||||
|
||||
readAttlistExp(xmlData, i) {
|
||||
// Skip leading whitespace after <!ATTLIST
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Read element name
|
||||
let elementName = "";
|
||||
while (i < xmlData.length && !/\s/.test(xmlData[i])) {
|
||||
elementName += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
|
||||
// Validate element name
|
||||
validateEntityName(elementName)
|
||||
|
||||
// Skip whitespace after element name
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Read attribute name
|
||||
let attributeName = "";
|
||||
while (i < xmlData.length && !/\s/.test(xmlData[i])) {
|
||||
attributeName += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
|
||||
// Validate attribute name
|
||||
if (!validateEntityName(attributeName)) {
|
||||
throw new Error(`Invalid attribute name: "${attributeName}"`);
|
||||
}
|
||||
|
||||
// Skip whitespace after attribute name
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Read attribute type
|
||||
let attributeType = "";
|
||||
if (xmlData.substring(i, i + 8).toUpperCase() === "NOTATION") {
|
||||
attributeType = "NOTATION";
|
||||
i += 8; // Move past "NOTATION"
|
||||
|
||||
// Skip whitespace after "NOTATION"
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Expect '(' to start the list of notations
|
||||
if (xmlData[i] !== "(") {
|
||||
throw new Error(`Expected '(', found "${xmlData[i]}"`);
|
||||
}
|
||||
i++; // Move past '('
|
||||
|
||||
// Read the list of allowed notations
|
||||
let allowedNotations = [];
|
||||
while (i < xmlData.length && xmlData[i] !== ")") {
|
||||
let notation = "";
|
||||
while (i < xmlData.length && xmlData[i] !== "|" && xmlData[i] !== ")") {
|
||||
notation += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
|
||||
// Validate notation name
|
||||
notation = notation.trim();
|
||||
if (!validateEntityName(notation)) {
|
||||
throw new Error(`Invalid notation name: "${notation}"`);
|
||||
}
|
||||
|
||||
allowedNotations.push(notation);
|
||||
|
||||
// Skip '|' separator or exit loop
|
||||
if (xmlData[i] === "|") {
|
||||
i++; // Move past '|'
|
||||
i = skipWhitespace(xmlData, i); // Skip optional whitespace after '|'
|
||||
}
|
||||
}
|
||||
|
||||
if (xmlData[i] !== ")") {
|
||||
throw new Error("Unterminated list of notations");
|
||||
}
|
||||
i++; // Move past ')'
|
||||
|
||||
// Store the allowed notations as part of the attribute type
|
||||
attributeType += " (" + allowedNotations.join("|") + ")";
|
||||
} else {
|
||||
// Handle simple types (e.g., CDATA, ID, IDREF, etc.)
|
||||
while (i < xmlData.length && !/\s/.test(xmlData[i])) {
|
||||
attributeType += xmlData[i];
|
||||
i++;
|
||||
}
|
||||
|
||||
// Validate simple attribute type
|
||||
const validTypes = ["CDATA", "ID", "IDREF", "IDREFS", "ENTITY", "ENTITIES", "NMTOKEN", "NMTOKENS"];
|
||||
if (!this.suppressValidationErr && !validTypes.includes(attributeType.toUpperCase())) {
|
||||
throw new Error(`Invalid attribute type: "${attributeType}"`);
|
||||
}
|
||||
}
|
||||
|
||||
// Skip whitespace after attribute type
|
||||
i = skipWhitespace(xmlData, i);
|
||||
|
||||
// Read default value
|
||||
let defaultValue = "";
|
||||
if (xmlData.substring(i, i + 8).toUpperCase() === "#REQUIRED") {
|
||||
defaultValue = "#REQUIRED";
|
||||
i += 8;
|
||||
} else if (xmlData.substring(i, i + 7).toUpperCase() === "#IMPLIED") {
|
||||
defaultValue = "#IMPLIED";
|
||||
i += 7;
|
||||
} else {
|
||||
[i, defaultValue] = this.readIdentifierVal(xmlData, i, "ATTLIST");
|
||||
}
|
||||
|
||||
return {
|
||||
elementName,
|
||||
attributeName,
|
||||
attributeType,
|
||||
defaultValue,
|
||||
index: i
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
|
||||
const skipWhitespace = (data, index) => {
|
||||
while (index < data.length && /\s/.test(data[index])) {
|
||||
index++;
|
||||
}
|
||||
return index;
|
||||
};
|
||||
|
||||
|
||||
|
||||
function hasSeq(data, seq,i){
|
||||
for(let j=0;j<seq.length;j++){
|
||||
if(seq[j]!==data[i+j+1]) return false;
|
||||
}
|
||||
return true;
|
||||
}
|
||||
|
||||
function validateEntityName(name){
|
||||
if (isName(name))
|
||||
return name;
|
||||
else
|
||||
throw new Error(`Invalid entity name ${name}`);
|
||||
}
|
||||
+46
@@ -0,0 +1,46 @@
|
||||
|
||||
export const defaultOptions = {
|
||||
preserveOrder: false,
|
||||
attributeNamePrefix: '@_',
|
||||
attributesGroupName: false,
|
||||
textNodeName: '#text',
|
||||
ignoreAttributes: true,
|
||||
removeNSPrefix: false, // remove NS from tag name or attribute name if true
|
||||
allowBooleanAttributes: false, //a tag can have attributes without any value
|
||||
//ignoreRootElement : false,
|
||||
parseTagValue: true,
|
||||
parseAttributeValue: false,
|
||||
trimValues: true, //Trim string values of tag and attributes
|
||||
cdataPropName: false,
|
||||
numberParseOptions: {
|
||||
hex: true,
|
||||
leadingZeros: true,
|
||||
eNotation: true
|
||||
},
|
||||
tagValueProcessor: function(tagName, val) {
|
||||
return val;
|
||||
},
|
||||
attributeValueProcessor: function(attrName, val) {
|
||||
return val;
|
||||
},
|
||||
stopNodes: [], //nested tags will not be parsed even for errors
|
||||
alwaysCreateTextNode: false,
|
||||
isArray: () => false,
|
||||
commentPropName: false,
|
||||
unpairedTags: [],
|
||||
processEntities: true,
|
||||
htmlEntities: false,
|
||||
ignoreDeclaration: false,
|
||||
ignorePiTags: false,
|
||||
transformTagName: false,
|
||||
transformAttributeName: false,
|
||||
updateTag: function(tagName, jPath, attrs){
|
||||
return tagName
|
||||
},
|
||||
// skipEmptyListItem: false
|
||||
captureMetaData: false,
|
||||
};
|
||||
|
||||
export const buildOptions = function(options) {
|
||||
return Object.assign({}, defaultOptions, options);
|
||||
};
|
||||
+629
@@ -0,0 +1,629 @@
|
||||
'use strict';
|
||||
///@ts-check
|
||||
|
||||
import {getAllMatches, isExist} from '../util.js';
|
||||
import xmlNode from './xmlNode.js';
|
||||
import DocTypeReader from './DocTypeReader.js';
|
||||
import toNumber from "strnum";
|
||||
import getIgnoreAttributesFn from "../ignoreAttributes.js";
|
||||
|
||||
// const regx =
|
||||
// '<((!\\[CDATA\\[([\\s\\S]*?)(]]>))|((NAME:)?(NAME))([^>]*)>|((\\/)(NAME)\\s*>))([^<]*)'
|
||||
// .replace(/NAME/g, util.nameRegexp);
|
||||
|
||||
//const tagsRegx = new RegExp("<(\\/?[\\w:\\-\._]+)([^>]*)>(\\s*"+cdataRegx+")*([^<]+)?","g");
|
||||
//const tagsRegx = new RegExp("<(\\/?)((\\w*:)?([\\w:\\-\._]+))([^>]*)>([^<]*)("+cdataRegx+"([^<]*))*([^<]+)?","g");
|
||||
|
||||
export default class OrderedObjParser{
|
||||
constructor(options){
|
||||
this.options = options;
|
||||
this.currentNode = null;
|
||||
this.tagsNodeStack = [];
|
||||
this.docTypeEntities = {};
|
||||
this.lastEntities = {
|
||||
"apos" : { regex: /&(apos|#39|#x27);/g, val : "'"},
|
||||
"gt" : { regex: /&(gt|#62|#x3E);/g, val : ">"},
|
||||
"lt" : { regex: /&(lt|#60|#x3C);/g, val : "<"},
|
||||
"quot" : { regex: /&(quot|#34|#x22);/g, val : "\""},
|
||||
};
|
||||
this.ampEntity = { regex: /&(amp|#38|#x26);/g, val : "&"};
|
||||
this.htmlEntities = {
|
||||
"space": { regex: /&(nbsp|#160);/g, val: " " },
|
||||
// "lt" : { regex: /&(lt|#60);/g, val: "<" },
|
||||
// "gt" : { regex: /&(gt|#62);/g, val: ">" },
|
||||
// "amp" : { regex: /&(amp|#38);/g, val: "&" },
|
||||
// "quot" : { regex: /&(quot|#34);/g, val: "\"" },
|
||||
// "apos" : { regex: /&(apos|#39);/g, val: "'" },
|
||||
"cent" : { regex: /&(cent|#162);/g, val: "¢" },
|
||||
"pound" : { regex: /&(pound|#163);/g, val: "£" },
|
||||
"yen" : { regex: /&(yen|#165);/g, val: "¥" },
|
||||
"euro" : { regex: /&(euro|#8364);/g, val: "€" },
|
||||
"copyright" : { regex: /&(copy|#169);/g, val: "©" },
|
||||
"reg" : { regex: /&(reg|#174);/g, val: "®" },
|
||||
"inr" : { regex: /&(inr|#8377);/g, val: "₹" },
|
||||
"num_dec": { regex: /&#([0-9]{1,7});/g, val : (_, str) => String.fromCodePoint(Number.parseInt(str, 10)) },
|
||||
"num_hex": { regex: /&#x([0-9a-fA-F]{1,6});/g, val : (_, str) => String.fromCodePoint(Number.parseInt(str, 16)) },
|
||||
};
|
||||
this.addExternalEntities = addExternalEntities;
|
||||
this.parseXml = parseXml;
|
||||
this.parseTextData = parseTextData;
|
||||
this.resolveNameSpace = resolveNameSpace;
|
||||
this.buildAttributesMap = buildAttributesMap;
|
||||
this.isItStopNode = isItStopNode;
|
||||
this.replaceEntitiesValue = replaceEntitiesValue;
|
||||
this.readStopNodeData = readStopNodeData;
|
||||
this.saveTextToParentTag = saveTextToParentTag;
|
||||
this.addChild = addChild;
|
||||
this.ignoreAttributesFn = getIgnoreAttributesFn(this.options.ignoreAttributes)
|
||||
|
||||
if(this.options.stopNodes && this.options.stopNodes.length > 0){
|
||||
this.stopNodesExact = new Set();
|
||||
this.stopNodesWildcard = new Set();
|
||||
for(let i = 0; i < this.options.stopNodes.length; i++){
|
||||
const stopNodeExp = this.options.stopNodes[i];
|
||||
if(typeof stopNodeExp !== 'string') continue;
|
||||
if(stopNodeExp.startsWith("*.")){
|
||||
this.stopNodesWildcard.add(stopNodeExp.substring(2));
|
||||
}else{
|
||||
this.stopNodesExact.add(stopNodeExp);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
function addExternalEntities(externalEntities){
|
||||
const entKeys = Object.keys(externalEntities);
|
||||
for (let i = 0; i < entKeys.length; i++) {
|
||||
const ent = entKeys[i];
|
||||
this.lastEntities[ent] = {
|
||||
regex: new RegExp("&"+ent+";","g"),
|
||||
val : externalEntities[ent]
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* @param {string} val
|
||||
* @param {string} tagName
|
||||
* @param {string} jPath
|
||||
* @param {boolean} dontTrim
|
||||
* @param {boolean} hasAttributes
|
||||
* @param {boolean} isLeafNode
|
||||
* @param {boolean} escapeEntities
|
||||
*/
|
||||
function parseTextData(val, tagName, jPath, dontTrim, hasAttributes, isLeafNode, escapeEntities) {
|
||||
if (val !== undefined) {
|
||||
if (this.options.trimValues && !dontTrim) {
|
||||
val = val.trim();
|
||||
}
|
||||
if(val.length > 0){
|
||||
if(!escapeEntities) val = this.replaceEntitiesValue(val);
|
||||
|
||||
const newval = this.options.tagValueProcessor(tagName, val, jPath, hasAttributes, isLeafNode);
|
||||
if(newval === null || newval === undefined){
|
||||
//don't parse
|
||||
return val;
|
||||
}else if(typeof newval !== typeof val || newval !== val){
|
||||
//overwrite
|
||||
return newval;
|
||||
}else if(this.options.trimValues){
|
||||
return parseValue(val, this.options.parseTagValue, this.options.numberParseOptions);
|
||||
}else{
|
||||
const trimmedVal = val.trim();
|
||||
if(trimmedVal === val){
|
||||
return parseValue(val, this.options.parseTagValue, this.options.numberParseOptions);
|
||||
}else{
|
||||
return val;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
function resolveNameSpace(tagname) {
|
||||
if (this.options.removeNSPrefix) {
|
||||
const tags = tagname.split(':');
|
||||
const prefix = tagname.charAt(0) === '/' ? '/' : '';
|
||||
if (tags[0] === 'xmlns') {
|
||||
return '';
|
||||
}
|
||||
if (tags.length === 2) {
|
||||
tagname = prefix + tags[1];
|
||||
}
|
||||
}
|
||||
return tagname;
|
||||
}
|
||||
|
||||
//TODO: change regex to capture NS
|
||||
//const attrsRegx = new RegExp("([\\w\\-\\.\\:]+)\\s*=\\s*(['\"])((.|\n)*?)\\2","gm");
|
||||
const attrsRegx = new RegExp('([^\\s=]+)\\s*(=\\s*([\'"])([\\s\\S]*?)\\3)?', 'gm');
|
||||
|
||||
function buildAttributesMap(attrStr, jPath) {
|
||||
if (this.options.ignoreAttributes !== true && typeof attrStr === 'string') {
|
||||
// attrStr = attrStr.replace(/\r?\n/g, ' ');
|
||||
//attrStr = attrStr || attrStr.trim();
|
||||
|
||||
const matches = getAllMatches(attrStr, attrsRegx);
|
||||
const len = matches.length; //don't make it inline
|
||||
const attrs = {};
|
||||
for (let i = 0; i < len; i++) {
|
||||
const attrName = this.resolveNameSpace(matches[i][1]);
|
||||
if (this.ignoreAttributesFn(attrName, jPath)) {
|
||||
continue
|
||||
}
|
||||
let oldVal = matches[i][4];
|
||||
let aName = this.options.attributeNamePrefix + attrName;
|
||||
if (attrName.length) {
|
||||
if (this.options.transformAttributeName) {
|
||||
aName = this.options.transformAttributeName(aName);
|
||||
}
|
||||
if(aName === "__proto__") aName = "#__proto__";
|
||||
if (oldVal !== undefined) {
|
||||
if (this.options.trimValues) {
|
||||
oldVal = oldVal.trim();
|
||||
}
|
||||
oldVal = this.replaceEntitiesValue(oldVal);
|
||||
const newVal = this.options.attributeValueProcessor(attrName, oldVal, jPath);
|
||||
if(newVal === null || newVal === undefined){
|
||||
//don't parse
|
||||
attrs[aName] = oldVal;
|
||||
}else if(typeof newVal !== typeof oldVal || newVal !== oldVal){
|
||||
//overwrite
|
||||
attrs[aName] = newVal;
|
||||
}else{
|
||||
//parse
|
||||
attrs[aName] = parseValue(
|
||||
oldVal,
|
||||
this.options.parseAttributeValue,
|
||||
this.options.numberParseOptions
|
||||
);
|
||||
}
|
||||
} else if (this.options.allowBooleanAttributes) {
|
||||
attrs[aName] = true;
|
||||
}
|
||||
}
|
||||
}
|
||||
if (!Object.keys(attrs).length) {
|
||||
return;
|
||||
}
|
||||
if (this.options.attributesGroupName) {
|
||||
const attrCollection = {};
|
||||
attrCollection[this.options.attributesGroupName] = attrs;
|
||||
return attrCollection;
|
||||
}
|
||||
return attrs
|
||||
}
|
||||
}
|
||||
|
||||
const parseXml = function(xmlData) {
|
||||
xmlData = xmlData.replace(/\r\n?/g, "\n"); //TODO: remove this line
|
||||
const xmlObj = new xmlNode('!xml');
|
||||
let currentNode = xmlObj;
|
||||
let textData = "";
|
||||
let jPath = "";
|
||||
const docTypeReader = new DocTypeReader(this.options.processEntities);
|
||||
for(let i=0; i< xmlData.length; i++){//for each char in XML data
|
||||
const ch = xmlData[i];
|
||||
if(ch === '<'){
|
||||
// const nextIndex = i+1;
|
||||
// const _2ndChar = xmlData[nextIndex];
|
||||
if( xmlData[i+1] === '/') {//Closing Tag
|
||||
const closeIndex = findClosingIndex(xmlData, ">", i, "Closing Tag is not closed.")
|
||||
let tagName = xmlData.substring(i+2,closeIndex).trim();
|
||||
|
||||
if(this.options.removeNSPrefix){
|
||||
const colonIndex = tagName.indexOf(":");
|
||||
if(colonIndex !== -1){
|
||||
tagName = tagName.substr(colonIndex+1);
|
||||
}
|
||||
}
|
||||
|
||||
if(this.options.transformTagName) {
|
||||
tagName = this.options.transformTagName(tagName);
|
||||
}
|
||||
|
||||
if(currentNode){
|
||||
textData = this.saveTextToParentTag(textData, currentNode, jPath);
|
||||
}
|
||||
|
||||
//check if last tag of nested tag was unpaired tag
|
||||
const lastTagName = jPath.substring(jPath.lastIndexOf(".")+1);
|
||||
if(tagName && this.options.unpairedTags.indexOf(tagName) !== -1 ){
|
||||
throw new Error(`Unpaired tag can not be used as closing tag: </${tagName}>`);
|
||||
}
|
||||
let propIndex = 0
|
||||
if(lastTagName && this.options.unpairedTags.indexOf(lastTagName) !== -1 ){
|
||||
propIndex = jPath.lastIndexOf('.', jPath.lastIndexOf('.')-1)
|
||||
this.tagsNodeStack.pop();
|
||||
}else{
|
||||
propIndex = jPath.lastIndexOf(".");
|
||||
}
|
||||
jPath = jPath.substring(0, propIndex);
|
||||
|
||||
currentNode = this.tagsNodeStack.pop();//avoid recursion, set the parent tag scope
|
||||
textData = "";
|
||||
i = closeIndex;
|
||||
} else if( xmlData[i+1] === '?') {
|
||||
|
||||
let tagData = readTagExp(xmlData,i, false, "?>");
|
||||
if(!tagData) throw new Error("Pi Tag is not closed.");
|
||||
|
||||
textData = this.saveTextToParentTag(textData, currentNode, jPath);
|
||||
if( (this.options.ignoreDeclaration && tagData.tagName === "?xml") || this.options.ignorePiTags){
|
||||
//do nothing
|
||||
}else{
|
||||
|
||||
const childNode = new xmlNode(tagData.tagName);
|
||||
childNode.add(this.options.textNodeName, "");
|
||||
|
||||
if(tagData.tagName !== tagData.tagExp && tagData.attrExpPresent){
|
||||
childNode[":@"] = this.buildAttributesMap(tagData.tagExp, jPath);
|
||||
}
|
||||
this.addChild(currentNode, childNode, jPath, i);
|
||||
}
|
||||
|
||||
|
||||
i = tagData.closeIndex + 1;
|
||||
} else if(xmlData.substr(i + 1, 3) === '!--') {
|
||||
const endIndex = findClosingIndex(xmlData, "-->", i+4, "Comment is not closed.")
|
||||
if(this.options.commentPropName){
|
||||
const comment = xmlData.substring(i + 4, endIndex - 2);
|
||||
|
||||
textData = this.saveTextToParentTag(textData, currentNode, jPath);
|
||||
|
||||
currentNode.add(this.options.commentPropName, [ { [this.options.textNodeName] : comment } ]);
|
||||
}
|
||||
i = endIndex;
|
||||
} else if( xmlData.substr(i + 1, 2) === '!D') {
|
||||
const result = docTypeReader.readDocType(xmlData, i);
|
||||
this.docTypeEntities = result.entities;
|
||||
i = result.i;
|
||||
}else if(xmlData.substr(i + 1, 2) === '![') {
|
||||
const closeIndex = findClosingIndex(xmlData, "]]>", i, "CDATA is not closed.") - 2;
|
||||
const tagExp = xmlData.substring(i + 9,closeIndex);
|
||||
|
||||
textData = this.saveTextToParentTag(textData, currentNode, jPath);
|
||||
|
||||
let val = this.parseTextData(tagExp, currentNode.tagname, jPath, true, false, true, true);
|
||||
if(val == undefined) val = "";
|
||||
|
||||
//cdata should be set even if it is 0 length string
|
||||
if(this.options.cdataPropName){
|
||||
currentNode.add(this.options.cdataPropName, [ { [this.options.textNodeName] : tagExp } ]);
|
||||
}else{
|
||||
currentNode.add(this.options.textNodeName, val);
|
||||
}
|
||||
|
||||
i = closeIndex + 2;
|
||||
}else {//Opening tag
|
||||
let result = readTagExp(xmlData,i, this.options.removeNSPrefix);
|
||||
let tagName= result.tagName;
|
||||
const rawTagName = result.rawTagName;
|
||||
let tagExp = result.tagExp;
|
||||
let attrExpPresent = result.attrExpPresent;
|
||||
let closeIndex = result.closeIndex;
|
||||
|
||||
if (this.options.transformTagName) {
|
||||
//console.log(tagExp, tagName)
|
||||
const newTagName = this.options.transformTagName(tagName);
|
||||
if(tagExp === tagName) {
|
||||
tagExp = newTagName
|
||||
}
|
||||
tagName = newTagName;
|
||||
}
|
||||
|
||||
//save text as child node
|
||||
if (currentNode && textData) {
|
||||
if(currentNode.tagname !== '!xml'){
|
||||
//when nested tag is found
|
||||
textData = this.saveTextToParentTag(textData, currentNode, jPath, false);
|
||||
}
|
||||
}
|
||||
|
||||
//check if last tag was unpaired tag
|
||||
const lastTag = currentNode;
|
||||
if(lastTag && this.options.unpairedTags.indexOf(lastTag.tagname) !== -1 ){
|
||||
currentNode = this.tagsNodeStack.pop();
|
||||
jPath = jPath.substring(0, jPath.lastIndexOf("."));
|
||||
}
|
||||
if(tagName !== xmlObj.tagname){
|
||||
jPath += jPath ? "." + tagName : tagName;
|
||||
}
|
||||
const startIndex = i;
|
||||
if (this.isItStopNode(this.stopNodesExact, this.stopNodesWildcard, jPath, tagName)) {
|
||||
let tagContent = "";
|
||||
//self-closing tag
|
||||
if(tagExp.length > 0 && tagExp.lastIndexOf("/") === tagExp.length - 1){
|
||||
if(tagName[tagName.length - 1] === "/"){ //remove trailing '/'
|
||||
tagName = tagName.substr(0, tagName.length - 1);
|
||||
jPath = jPath.substr(0, jPath.length - 1);
|
||||
tagExp = tagName;
|
||||
}else{
|
||||
tagExp = tagExp.substr(0, tagExp.length - 1);
|
||||
}
|
||||
i = result.closeIndex;
|
||||
}
|
||||
//unpaired tag
|
||||
else if(this.options.unpairedTags.indexOf(tagName) !== -1){
|
||||
|
||||
i = result.closeIndex;
|
||||
}
|
||||
//normal tag
|
||||
else{
|
||||
//read until closing tag is found
|
||||
const result = this.readStopNodeData(xmlData, rawTagName, closeIndex + 1);
|
||||
if(!result) throw new Error(`Unexpected end of ${rawTagName}`);
|
||||
i = result.i;
|
||||
tagContent = result.tagContent;
|
||||
}
|
||||
|
||||
const childNode = new xmlNode(tagName);
|
||||
|
||||
if(tagName !== tagExp && attrExpPresent){
|
||||
childNode[":@"] = this.buildAttributesMap(tagExp, jPath
|
||||
);
|
||||
}
|
||||
if(tagContent) {
|
||||
tagContent = this.parseTextData(tagContent, tagName, jPath, true, attrExpPresent, true, true);
|
||||
}
|
||||
|
||||
jPath = jPath.substr(0, jPath.lastIndexOf("."));
|
||||
childNode.add(this.options.textNodeName, tagContent);
|
||||
|
||||
this.addChild(currentNode, childNode, jPath, startIndex);
|
||||
}else{
|
||||
//selfClosing tag
|
||||
if(tagExp.length > 0 && tagExp.lastIndexOf("/") === tagExp.length - 1){
|
||||
if(tagName[tagName.length - 1] === "/"){ //remove trailing '/'
|
||||
tagName = tagName.substr(0, tagName.length - 1);
|
||||
jPath = jPath.substr(0, jPath.length - 1);
|
||||
tagExp = tagName;
|
||||
}else{
|
||||
tagExp = tagExp.substr(0, tagExp.length - 1);
|
||||
}
|
||||
|
||||
if(this.options.transformTagName) {
|
||||
const newTagName = this.options.transformTagName(tagName);
|
||||
if(tagExp === tagName) {
|
||||
tagExp = newTagName
|
||||
}
|
||||
tagName = newTagName;
|
||||
}
|
||||
|
||||
const childNode = new xmlNode(tagName);
|
||||
if(tagName !== tagExp && attrExpPresent){
|
||||
childNode[":@"] = this.buildAttributesMap(tagExp, jPath);
|
||||
}
|
||||
this.addChild(currentNode, childNode, jPath, startIndex);
|
||||
jPath = jPath.substr(0, jPath.lastIndexOf("."));
|
||||
}
|
||||
//opening tag
|
||||
else{
|
||||
const childNode = new xmlNode( tagName);
|
||||
this.tagsNodeStack.push(currentNode);
|
||||
|
||||
if(tagName !== tagExp && attrExpPresent){
|
||||
childNode[":@"] = this.buildAttributesMap(tagExp, jPath);
|
||||
}
|
||||
this.addChild(currentNode, childNode, jPath, startIndex);
|
||||
currentNode = childNode;
|
||||
}
|
||||
textData = "";
|
||||
i = closeIndex;
|
||||
}
|
||||
}
|
||||
}else{
|
||||
textData += xmlData[i];
|
||||
}
|
||||
}
|
||||
return xmlObj.child;
|
||||
}
|
||||
|
||||
function addChild(currentNode, childNode, jPath, startIndex){
|
||||
// unset startIndex if not requested
|
||||
if (!this.options.captureMetaData) startIndex = undefined;
|
||||
const result = this.options.updateTag(childNode.tagname, jPath, childNode[":@"])
|
||||
if(result === false){
|
||||
//do nothing
|
||||
} else if(typeof result === "string"){
|
||||
childNode.tagname = result
|
||||
currentNode.addChild(childNode, startIndex);
|
||||
}else{
|
||||
currentNode.addChild(childNode, startIndex);
|
||||
}
|
||||
}
|
||||
|
||||
const replaceEntitiesValue = function(val){
|
||||
|
||||
if(this.options.processEntities){
|
||||
for(let entityName in this.docTypeEntities){
|
||||
const entity = this.docTypeEntities[entityName];
|
||||
val = val.replace( entity.regx, entity.val);
|
||||
}
|
||||
for(let entityName in this.lastEntities){
|
||||
const entity = this.lastEntities[entityName];
|
||||
val = val.replace( entity.regex, entity.val);
|
||||
}
|
||||
if(this.options.htmlEntities){
|
||||
for(let entityName in this.htmlEntities){
|
||||
const entity = this.htmlEntities[entityName];
|
||||
val = val.replace( entity.regex, entity.val);
|
||||
}
|
||||
}
|
||||
val = val.replace( this.ampEntity.regex, this.ampEntity.val);
|
||||
}
|
||||
return val;
|
||||
}
|
||||
function saveTextToParentTag(textData, currentNode, jPath, isLeafNode) {
|
||||
if (textData) { //store previously collected data as textNode
|
||||
if(isLeafNode === undefined) isLeafNode = currentNode.child.length === 0
|
||||
|
||||
textData = this.parseTextData(textData,
|
||||
currentNode.tagname,
|
||||
jPath,
|
||||
false,
|
||||
currentNode[":@"] ? Object.keys(currentNode[":@"]).length !== 0 : false,
|
||||
isLeafNode);
|
||||
|
||||
if (textData !== undefined && textData !== "")
|
||||
currentNode.add(this.options.textNodeName, textData);
|
||||
textData = "";
|
||||
}
|
||||
return textData;
|
||||
}
|
||||
|
||||
//TODO: use jPath to simplify the logic
|
||||
/**
|
||||
* @param {Set} stopNodesExact
|
||||
* @param {Set} stopNodesWildcard
|
||||
* @param {string} jPath
|
||||
* @param {string} currentTagName
|
||||
*/
|
||||
function isItStopNode(stopNodesExact, stopNodesWildcard, jPath, currentTagName){
|
||||
if(stopNodesWildcard && stopNodesWildcard.has(currentTagName)) return true;
|
||||
if(stopNodesExact && stopNodesExact.has(jPath)) return true;
|
||||
return false;
|
||||
}
|
||||
|
||||
/**
|
||||
* Returns the tag Expression and where it is ending handling single-double quotes situation
|
||||
* @param {string} xmlData
|
||||
* @param {number} i starting index
|
||||
* @returns
|
||||
*/
|
||||
function tagExpWithClosingIndex(xmlData, i, closingChar = ">"){
|
||||
let attrBoundary;
|
||||
let tagExp = "";
|
||||
for (let index = i; index < xmlData.length; index++) {
|
||||
let ch = xmlData[index];
|
||||
if (attrBoundary) {
|
||||
if (ch === attrBoundary) attrBoundary = "";//reset
|
||||
} else if (ch === '"' || ch === "'") {
|
||||
attrBoundary = ch;
|
||||
} else if (ch === closingChar[0]) {
|
||||
if(closingChar[1]){
|
||||
if(xmlData[index + 1] === closingChar[1]){
|
||||
return {
|
||||
data: tagExp,
|
||||
index: index
|
||||
}
|
||||
}
|
||||
}else{
|
||||
return {
|
||||
data: tagExp,
|
||||
index: index
|
||||
}
|
||||
}
|
||||
} else if (ch === '\t') {
|
||||
ch = " "
|
||||
}
|
||||
tagExp += ch;
|
||||
}
|
||||
}
|
||||
|
||||
function findClosingIndex(xmlData, str, i, errMsg){
|
||||
const closingIndex = xmlData.indexOf(str, i);
|
||||
if(closingIndex === -1){
|
||||
throw new Error(errMsg)
|
||||
}else{
|
||||
return closingIndex + str.length - 1;
|
||||
}
|
||||
}
|
||||
|
||||
function readTagExp(xmlData,i, removeNSPrefix, closingChar = ">"){
|
||||
const result = tagExpWithClosingIndex(xmlData, i+1, closingChar);
|
||||
if(!result) return;
|
||||
let tagExp = result.data;
|
||||
const closeIndex = result.index;
|
||||
const separatorIndex = tagExp.search(/\s/);
|
||||
let tagName = tagExp;
|
||||
let attrExpPresent = true;
|
||||
if(separatorIndex !== -1){//separate tag name and attributes expression
|
||||
tagName = tagExp.substring(0, separatorIndex);
|
||||
tagExp = tagExp.substring(separatorIndex + 1).trimStart();
|
||||
}
|
||||
|
||||
const rawTagName = tagName;
|
||||
if(removeNSPrefix){
|
||||
const colonIndex = tagName.indexOf(":");
|
||||
if(colonIndex !== -1){
|
||||
tagName = tagName.substr(colonIndex+1);
|
||||
attrExpPresent = tagName !== result.data.substr(colonIndex + 1);
|
||||
}
|
||||
}
|
||||
|
||||
return {
|
||||
tagName: tagName,
|
||||
tagExp: tagExp,
|
||||
closeIndex: closeIndex,
|
||||
attrExpPresent: attrExpPresent,
|
||||
rawTagName: rawTagName,
|
||||
}
|
||||
}
|
||||
/**
|
||||
* find paired tag for a stop node
|
||||
* @param {string} xmlData
|
||||
* @param {string} tagName
|
||||
* @param {number} i
|
||||
*/
|
||||
function readStopNodeData(xmlData, tagName, i){
|
||||
const startIndex = i;
|
||||
// Starting at 1 since we already have an open tag
|
||||
let openTagCount = 1;
|
||||
|
||||
for (; i < xmlData.length; i++) {
|
||||
if( xmlData[i] === "<"){
|
||||
if (xmlData[i+1] === "/") {//close tag
|
||||
const closeIndex = findClosingIndex(xmlData, ">", i, `${tagName} is not closed`);
|
||||
let closeTagName = xmlData.substring(i+2,closeIndex).trim();
|
||||
if(closeTagName === tagName){
|
||||
openTagCount--;
|
||||
if (openTagCount === 0) {
|
||||
return {
|
||||
tagContent: xmlData.substring(startIndex, i),
|
||||
i : closeIndex
|
||||
}
|
||||
}
|
||||
}
|
||||
i=closeIndex;
|
||||
} else if(xmlData[i+1] === '?') {
|
||||
const closeIndex = findClosingIndex(xmlData, "?>", i+1, "StopNode is not closed.")
|
||||
i=closeIndex;
|
||||
} else if(xmlData.substr(i + 1, 3) === '!--') {
|
||||
const closeIndex = findClosingIndex(xmlData, "-->", i+3, "StopNode is not closed.")
|
||||
i=closeIndex;
|
||||
} else if(xmlData.substr(i + 1, 2) === '![') {
|
||||
const closeIndex = findClosingIndex(xmlData, "]]>", i, "StopNode is not closed.") - 2;
|
||||
i=closeIndex;
|
||||
} else {
|
||||
const tagData = readTagExp(xmlData, i, '>')
|
||||
|
||||
if (tagData) {
|
||||
const openTagName = tagData && tagData.tagName;
|
||||
if (openTagName === tagName && tagData.tagExp[tagData.tagExp.length-1] !== "/") {
|
||||
openTagCount++;
|
||||
}
|
||||
i=tagData.closeIndex;
|
||||
}
|
||||
}
|
||||
}
|
||||
}//end for loop
|
||||
}
|
||||
|
||||
function parseValue(val, shouldParse, options) {
|
||||
if (shouldParse && typeof val === 'string') {
|
||||
//console.log(options)
|
||||
const newval = val.trim();
|
||||
if(newval === 'true' ) return true;
|
||||
else if(newval === 'false' ) return false;
|
||||
else return toNumber(val, options);
|
||||
} else {
|
||||
if (isExist(val)) {
|
||||
return val;
|
||||
} else {
|
||||
return '';
|
||||
}
|
||||
}
|
||||
}
|
||||
+71
@@ -0,0 +1,71 @@
|
||||
import { buildOptions} from './OptionsBuilder.js';
|
||||
import OrderedObjParser from './OrderedObjParser.js';
|
||||
import prettify from './node2json.js';
|
||||
import {validate} from "../validator.js";
|
||||
import XmlNode from './xmlNode.js';
|
||||
|
||||
export default class XMLParser{
|
||||
|
||||
constructor(options){
|
||||
this.externalEntities = {};
|
||||
this.options = buildOptions(options);
|
||||
|
||||
}
|
||||
/**
|
||||
* Parse XML dats to JS object
|
||||
* @param {string|Uint8Array} xmlData
|
||||
* @param {boolean|Object} validationOption
|
||||
*/
|
||||
parse(xmlData,validationOption){
|
||||
if(typeof xmlData !== "string" && xmlData.toString){
|
||||
xmlData = xmlData.toString();
|
||||
}else if(typeof xmlData !== "string"){
|
||||
throw new Error("XML data is accepted in String or Bytes[] form.")
|
||||
}
|
||||
|
||||
if( validationOption){
|
||||
if(validationOption === true) validationOption = {}; //validate with default options
|
||||
|
||||
const result = validate(xmlData, validationOption);
|
||||
if (result !== true) {
|
||||
throw Error( `${result.err.msg}:${result.err.line}:${result.err.col}` )
|
||||
}
|
||||
}
|
||||
const orderedObjParser = new OrderedObjParser(this.options);
|
||||
orderedObjParser.addExternalEntities(this.externalEntities);
|
||||
const orderedResult = orderedObjParser.parseXml(xmlData);
|
||||
if(this.options.preserveOrder || orderedResult === undefined) return orderedResult;
|
||||
else return prettify(orderedResult, this.options);
|
||||
}
|
||||
|
||||
/**
|
||||
* Add Entity which is not by default supported by this library
|
||||
* @param {string} key
|
||||
* @param {string} value
|
||||
*/
|
||||
addEntity(key, value){
|
||||
if(value.indexOf("&") !== -1){
|
||||
throw new Error("Entity value can't have '&'")
|
||||
}else if(key.indexOf("&") !== -1 || key.indexOf(";") !== -1){
|
||||
throw new Error("An entity must be set without '&' and ';'. Eg. use '#xD' for '
'")
|
||||
}else if(value === "&"){
|
||||
throw new Error("An entity with value '&' is not permitted");
|
||||
}else{
|
||||
this.externalEntities[key] = value;
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Returns a Symbol that can be used to access the metadata
|
||||
* property on a node.
|
||||
*
|
||||
* If Symbol is not available in the environment, an ordinary property is used
|
||||
* and the name of the property is here returned.
|
||||
*
|
||||
* The XMLMetaData property is only present when `captureMetaData`
|
||||
* is true in the options.
|
||||
*/
|
||||
static getMetaDataSymbol() {
|
||||
return XmlNode.getMetaDataSymbol();
|
||||
}
|
||||
}
|
||||
+119
@@ -0,0 +1,119 @@
|
||||
'use strict';
|
||||
|
||||
import XmlNode from './xmlNode.js';
|
||||
|
||||
const METADATA_SYMBOL = XmlNode.getMetaDataSymbol();
|
||||
|
||||
/**
|
||||
*
|
||||
* @param {array} node
|
||||
* @param {any} options
|
||||
* @returns
|
||||
*/
|
||||
export default function prettify(node, options){
|
||||
return compress( node, options);
|
||||
}
|
||||
|
||||
/**
|
||||
*
|
||||
* @param {array} arr
|
||||
* @param {object} options
|
||||
* @param {string} jPath
|
||||
* @returns object
|
||||
*/
|
||||
function compress(arr, options, jPath){
|
||||
let text;
|
||||
const compressedObj = {};
|
||||
for (let i = 0; i < arr.length; i++) {
|
||||
const tagObj = arr[i];
|
||||
const property = propName(tagObj);
|
||||
let newJpath = "";
|
||||
if(jPath === undefined) newJpath = property;
|
||||
else newJpath = jPath + "." + property;
|
||||
|
||||
if(property === options.textNodeName){
|
||||
if(text === undefined) text = tagObj[property];
|
||||
else text += "" + tagObj[property];
|
||||
}else if(property === undefined){
|
||||
continue;
|
||||
}else if(tagObj[property]){
|
||||
|
||||
let val = compress(tagObj[property], options, newJpath);
|
||||
const isLeaf = isLeafTag(val, options);
|
||||
if (tagObj[METADATA_SYMBOL] !== undefined) {
|
||||
val[METADATA_SYMBOL] = tagObj[METADATA_SYMBOL]; // copy over metadata
|
||||
}
|
||||
|
||||
if(tagObj[":@"]){
|
||||
assignAttributes( val, tagObj[":@"], newJpath, options);
|
||||
}else if(Object.keys(val).length === 1 && val[options.textNodeName] !== undefined && !options.alwaysCreateTextNode){
|
||||
val = val[options.textNodeName];
|
||||
}else if(Object.keys(val).length === 0){
|
||||
if(options.alwaysCreateTextNode) val[options.textNodeName] = "";
|
||||
else val = "";
|
||||
}
|
||||
|
||||
if(compressedObj[property] !== undefined && compressedObj.hasOwnProperty(property)) {
|
||||
if(!Array.isArray(compressedObj[property])) {
|
||||
compressedObj[property] = [ compressedObj[property] ];
|
||||
}
|
||||
compressedObj[property].push(val);
|
||||
}else{
|
||||
//TODO: if a node is not an array, then check if it should be an array
|
||||
//also determine if it is a leaf node
|
||||
if (options.isArray(property, newJpath, isLeaf )) {
|
||||
compressedObj[property] = [val];
|
||||
}else{
|
||||
compressedObj[property] = val;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
// if(text && text.length > 0) compressedObj[options.textNodeName] = text;
|
||||
if(typeof text === "string"){
|
||||
if(text.length > 0) compressedObj[options.textNodeName] = text;
|
||||
}else if(text !== undefined) compressedObj[options.textNodeName] = text;
|
||||
return compressedObj;
|
||||
}
|
||||
|
||||
function propName(obj){
|
||||
const keys = Object.keys(obj);
|
||||
for (let i = 0; i < keys.length; i++) {
|
||||
const key = keys[i];
|
||||
if(key !== ":@") return key;
|
||||
}
|
||||
}
|
||||
|
||||
function assignAttributes(obj, attrMap, jpath, options){
|
||||
if (attrMap) {
|
||||
const keys = Object.keys(attrMap);
|
||||
const len = keys.length; //don't make it inline
|
||||
for (let i = 0; i < len; i++) {
|
||||
const atrrName = keys[i];
|
||||
if (options.isArray(atrrName, jpath + "." + atrrName, true, true)) {
|
||||
obj[atrrName] = [ attrMap[atrrName] ];
|
||||
} else {
|
||||
obj[atrrName] = attrMap[atrrName];
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
function isLeafTag(obj, options){
|
||||
const { textNodeName } = options;
|
||||
const propCount = Object.keys(obj).length;
|
||||
|
||||
if (propCount === 0) {
|
||||
return true;
|
||||
}
|
||||
|
||||
if (
|
||||
propCount === 1 &&
|
||||
(obj[textNodeName] || typeof obj[textNodeName] === "boolean" || obj[textNodeName] === 0)
|
||||
) {
|
||||
return true;
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
+40
@@ -0,0 +1,40 @@
|
||||
'use strict';
|
||||
|
||||
let METADATA_SYMBOL;
|
||||
|
||||
if (typeof Symbol !== "function") {
|
||||
METADATA_SYMBOL = "@@xmlMetadata";
|
||||
} else {
|
||||
METADATA_SYMBOL = Symbol("XML Node Metadata");
|
||||
}
|
||||
|
||||
export default class XmlNode{
|
||||
constructor(tagname) {
|
||||
this.tagname = tagname;
|
||||
this.child = []; //nested tags, text, cdata, comments in order
|
||||
this[":@"] = {}; //attributes map
|
||||
}
|
||||
add(key,val){
|
||||
// this.child.push( {name : key, val: val, isCdata: isCdata });
|
||||
if(key === "__proto__") key = "#__proto__";
|
||||
this.child.push( {[key]: val });
|
||||
}
|
||||
addChild(node, startIndex) {
|
||||
if(node.tagname === "__proto__") node.tagname = "#__proto__";
|
||||
if(node[":@"] && Object.keys(node[":@"]).length > 0){
|
||||
this.child.push( { [node.tagname]: node.child, [":@"]: node[":@"] });
|
||||
}else{
|
||||
this.child.push( { [node.tagname]: node.child });
|
||||
}
|
||||
// if requested, add the startIndex
|
||||
if (startIndex !== undefined) {
|
||||
// Note: for now we just overwrite the metadata. If we had more complex metadata,
|
||||
// we might need to do an object append here: metadata = { ...metadata, startIndex }
|
||||
this.child[this.child.length - 1][METADATA_SYMBOL] = { startIndex };
|
||||
}
|
||||
}
|
||||
/** symbol used for metadata */
|
||||
static getMetaDataSymbol() {
|
||||
return METADATA_SYMBOL;
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user