fast-xml-parser 5.2.0 → 5.2.2

package/src/xmlparser/DocTypeReader.js

@@ -22,16 +22,25 @@ export default function readDocType(xmlData, i){
                     let entityName, val;
                     [entityName, val,i] = readEntityExp(xmlData,i+1);
                     if(val.indexOf("&") === -1) //Parameter entities are not supported
-                        entities[ validateEntityName(entityName) ] = {
+                        entities[ entityName ] = {
                             regx : RegExp( `&${entityName};`,"g"),
                             val: val
                         };
                 }
-                else if( hasBody && isElement(xmlData, i))  i += 8;//Not supported
-                else if( hasBody && isAttlist(xmlData, i))  i += 8;//Not supported
-                else if( hasBody && isNotation(xmlData, i)) i += 9;//Not supported
-                else if( isComment)                         comment = true;
-                else                                        throw new Error("Invalid DOCTYPE");
+                else if( hasBody && isElement(xmlData, i))  {
+                    i += 8;//Not supported
+                    const {index} = readElementExp(xmlData,i+1);
+                    i = index;
+                }else if( hasBody && isAttlist(xmlData, i)){
+                    i += 8;//Not supported
+                    // const {index} = readAttlistExp(xmlData,i+1);
+                    // i = index;
+                }else if( hasBody && isNotation(xmlData, i)) {
+                    i += 9;//Not supported
+                    const {index} = readNotationExp(xmlData,i+1);
+                    i = index;
+                }else if( isComment) comment = true;
+                else throw new Error("Invalid DOCTYPE");
 
                 angleBracketsCount++;
                 exp = "";
@@ -62,7 +71,14 @@ export default function readDocType(xmlData, i){
     return {entities, i};
 }
 
-function readEntityExp(xmlData,i){
+const skipWhitespace = (data, index) => {
+    while (index < data.length && /\s/.test(data[index])) {
+        index++;
+    }
+    return index;
+};
+
+function readEntityExp(xmlData, i) {    
     //External entities are not supported
     //    <!ENTITY ext SYSTEM "http://normal-website.com" >
 
@@ -71,24 +87,265 @@ function readEntityExp(xmlData,i){
 
     //Internal entities are supported
     //    <!ENTITY entityname "replacement text">
-    
-    //read EntityName
+
+    // Skip leading whitespace after <!ENTITY
+    i = skipWhitespace(xmlData, i);
+
+    // Read entity name
     let entityName = "";
-    for (; i < xmlData.length && (xmlData[i] !== "'" && xmlData[i] !== '"' ); i++) {
-        // if(xmlData[i] === " ") continue;
-        // else 
+    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 (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] = readIdentifierVal(xmlData, i, "entity");
+    i--;
+    return [entityName, entityValue, i ];
+}
+
+function 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++;
+    }
+    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 (identifierType !== "SYSTEM" && identifierType !== "PUBLIC") {
+        throw new Error(`Expected SYSTEM or PUBLIC, found "${identifierType}"`);
+    }
+    i += identifierType.length;
+
+    // Skip whitespace after identifier type
+    i = skipWhitespace(xmlData, i);
+
+    // Read public identifier (if PUBLIC)
+    let publicIdentifier = null;
+    let systemIdentifier = null;
+
+    if (identifierType === "PUBLIC") {
+        [i, publicIdentifier ] = readIdentifierVal(xmlData, i, "publicIdentifier");
+
+        // Skip whitespace after public identifier
+        i = skipWhitespace(xmlData, i);
+
+        // Optionally read system identifier
+        if (xmlData[i] === '"' || xmlData[i] === "'") {
+            [i, systemIdentifier ] = readIdentifierVal(xmlData, i,"systemIdentifier");
+        }
+    } else if (identifierType === "SYSTEM") {
+        // Read system identifier (mandatory for SYSTEM)
+        [i, systemIdentifier ] = readIdentifierVal(xmlData, i, "systemIdentifier");
+
+        if (!systemIdentifier) {
+            throw new Error("Missing mandatory system identifier for SYSTEM notation");
+        }
+    }
+    
+    return {notationName, publicIdentifier, systemIdentifier, index: --i};
+}
+
+function readIdentifierVal(xmlData, i, type) {
+    let identifierVal = "";
+    const startChar = xmlData[i];
+    if (startChar !== '"' && startChar !== "'") {
+        throw new Error(`Expected quoted string, found "${startChar}"`);
+    }
+    i++;
+
+    while (i < xmlData.length && xmlData[i] !== startChar) {
+        identifierVal += xmlData[i];
+        i++;
+    }
+
+    if (xmlData[i] !== startChar) {
+        throw new Error(`Unterminated ${type} value`);
+    }
+    i++;
+    return [i, identifierVal];
+}
+
+function readElementExp(xmlData, i) {
+    // <!ELEMENT name (content-model)>
+
+    // Skip leading whitespace after <!ELEMENT
+    i = skipWhitespace(xmlData, i);
+
+    // Read element name
+    let elementName = "";
+    while (i < xmlData.length && !/\s/.test(xmlData[i])) {
+        elementName += xmlData[i];
+        i++;
+    }
+
+    // Validate element name
+    if (!validateEntityName(elementName)) {
+        throw new Error(`Invalid element name: "${elementName}"`);
+    }
+
+    // Skip whitespace after element name
+    i = skipWhitespace(xmlData, i);
+
+    // Expect '(' to start content model
+    if (xmlData[i] !== "(") {
+        throw new Error(`Expected '(', found "${xmlData[i]}"`);
+    }
+    i++; // Move past '('
+
+    // Read content model
+    let contentModel = "";
+    while (i < xmlData.length && xmlData[i] !== ")") {
+        contentModel += xmlData[i];
+        i++;
+    }
+
+    if (xmlData[i] !== ")") {
+        throw new Error("Unterminated content model");
+    }
+
+    return {
+        elementName,
+        contentModel: contentModel.trim(),
+        index: i
+    };
+}
+
+function readAttlistExp(xmlData, i) {
+    // Skip leading whitespace after <!ATTLIST
+    i = skipWhitespace(xmlData, i);
+
+    // Read element name
+    let elementName = "";
+    while (i < xmlData.length && !/\s/.test(xmlData[i])) {
+        elementName += xmlData[i];
+        i++;
+    }
+
+    // Validate element name
+    validateEntityName(elementName)
+
+    // Skip whitespace after element name
+    i = skipWhitespace(xmlData, i);
+
+    // Read attribute name
+    let attributeName = "";
+    while (i < xmlData.length && !/\s/.test(xmlData[i])) {
+        attributeName += xmlData[i];
+        i++;
+    }
+
+    // Validate attribute name
+    if (!validateEntityName(attributeName)) {
+        throw new Error(`Invalid attribute name: "${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 (!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] = readIdentifierVal(xmlData, i, "ATTLIST");
     }
-    entityName = entityName.trim();
-    if(entityName.indexOf(" ") !== -1) throw new Error("External entites are not supported");
 
-    //read Entity Value
-    const startChar = xmlData[i++];
-    let val = ""
-    for (; i < xmlData.length && xmlData[i] !== startChar ; i++) {
-        val += xmlData[i];
+    return {
+        elementName,
+        attributeName,
+        attributeType,
+        defaultValue,
+        index: i
     }
-    return [entityName, val, i];
 }
 
 function isComment(xmlData, i){