node-html-parser 5.2.2 → 5.3.0

package/README.md

@@ -203,6 +203,10 @@ Get class names.
 
 Clone a node.
 
+#### Node#getElementById(id: string): HTMLElement;
+
+Get element by it's ID.
+
 ## HTMLElement Properties
 
 ### HTMLElement#text

package/dist/main.js

@@ -374,7 +374,7 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
         }
         DOMTokenList.prototype._validate = function (c) {
             if (/\s/.test(c)) {
-                throw new Error("DOMException in DOMTokenList.add: The token '" + c + "' contains HTML space characters, which are not valid in tokens.");
+                throw new Error("DOMException in DOMTokenList.add: The token '".concat(c, "' contains HTML space characters, which are not valid in tokens."));
             }
         };
         DOMTokenList.prototype.add = function (c) {
@@ -458,14 +458,14 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
             );
             if (keyAttrs.id) {
                 if (!rawAttrs) {
-                    _this.rawAttrs = "id=\"" + keyAttrs.id + "\"";
+                    _this.rawAttrs = "id=\"".concat(keyAttrs.id, "\"");
                 }
             }
             if (keyAttrs.class) {
                 if (!rawAttrs) {
-                    var cls = "class=\"" + _this.classList.toString() + "\"";
+                    var cls = "class=\"".concat(_this.classList.toString(), "\"");
                     if (_this.rawAttrs) {
-                        _this.rawAttrs += " " + cls;
+                        _this.rawAttrs += " ".concat(cls);
                     }
                     else {
                         _this.rawAttrs = cls;
@@ -610,7 +610,7 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
                         else {
                             var text = node.trimmedText;
                             if (currentBlock.prependWhitespace) {
-                                text = " " + text;
+                                text = " ".concat(text);
                                 currentBlock.prependWhitespace = false;
                             }
                             currentBlock.push(text);
@@ -631,8 +631,8 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
         HTMLElement.prototype.toString = function () {
             var tag = this.rawTagName;
             if (tag) {
-                var attrs = this.rawAttrs ? " " + this.rawAttrs : '';
-                return this.isVoidElement ? "<" + tag + attrs + ">" : "<" + tag + attrs + ">" + this.innerHTML + "</" + tag + ">";
+                var attrs = this.rawAttrs ? " ".concat(this.rawAttrs) : '';
+                return this.isVoidElement ? "<".concat(tag).concat(attrs, ">") : "<".concat(tag).concat(attrs, ">").concat(this.innerHTML, "</").concat(tag, ">");
             }
             return this.innerHTML;
         };
@@ -728,9 +728,9 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
                     res.push('  '.repeat(indention) + str);
                 }
                 function dfs(node) {
-                    var idStr = node.id ? "#" + node.id : '';
-                    var classStr = node.classList.length ? "." + node.classList.value.join('.') : ''; // eslint-disable-line @typescript-eslint/no-unsafe-member-access, @typescript-eslint/no-unsafe-member-access, @typescript-eslint/restrict-template-expressions, @typescript-eslint/no-unsafe-call
-                    write("" + node.rawTagName + idStr + classStr);
+                    var idStr = node.id ? "#".concat(node.id) : '';
+                    var classStr = node.classList.length ? ".".concat(node.classList.value.join('.')) : ''; // eslint-disable-line @typescript-eslint/no-unsafe-member-access, @typescript-eslint/no-unsafe-member-access, @typescript-eslint/restrict-template-expressions, @typescript-eslint/no-unsafe-call
+                    write("".concat(node.rawTagName).concat(idStr).concat(classStr));
                     indention++;
                     node.childNodes.forEach(function (childNode) {
                         if (childNode.nodeType === type_3.default.ELEMENT_NODE) {
@@ -832,6 +832,43 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
             }
             return re;
         };
+        /**
+         * find element by it's id
+         * @param {string} id the id of the element to select
+         */
+        HTMLElement.prototype.getElementById = function (id) {
+            var stack = [];
+            var currentNodeReference = this;
+            var index = 0;
+            // index turns to undefined once the stack is empty and the first condition occurs
+            // which happens once all relevant children are searched through
+            while (index !== undefined) {
+                var child = void 0;
+                // make it work with sparse arrays
+                do {
+                    child = currentNodeReference.childNodes[index++];
+                } while (index < currentNodeReference.childNodes.length && child === undefined);
+                // if the child does not exist we move on with the last provided index (which belongs to the parentNode)
+                if (child === undefined) {
+                    currentNodeReference = currentNodeReference.parentNode;
+                    index = stack.pop();
+                    continue;
+                }
+                if (child.nodeType === type_3.default.ELEMENT_NODE) {
+                    if (child.id === id) {
+                        return child;
+                    }
+                    ;
+                    // if children are existing push the current status to the stack and keep searching for elements in the level below
+                    if (child.childNodes.length > 0) {
+                        stack.push(index);
+                        currentNodeReference = child;
+                        index = 0;
+                    }
+                }
+            }
+            return null;
+        };
         /**
          * traverses the Element and its parents (heading toward the document root) until it finds a node that matches the provided selector string. Will return itself or the matching ancestor. If no such element exists, it returns null.
          * @param selector a DOMString containing a selector list
@@ -989,7 +1026,7 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
                 if (val === undefined || val === 'null') {
                     return name;
                 }
-                return name + "=" + val;
+                return "".concat(name, "=").concat(val);
             })
                 .join(' ');
             // Update this.id
@@ -1036,7 +1073,7 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
                 var val = _this.quoteAttribute(attrs[name]);
                 if (val === 'null' || val === '""')
                     return name;
-                return name + "=" + val;
+                return "".concat(name, "=").concat(val);
             })
                 .join(' ');
             // Update this.id
@@ -1064,7 +1101,7 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
                 var val = attributes[name];
                 if (val === 'null' || val === '""')
                     return name;
-                return name + "=" + _this.quoteAttribute(String(val));
+                return "".concat(name, "=").concat(_this.quoteAttribute(String(val)));
             })
                 .join(' ');
         };
@@ -1106,7 +1143,7 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
                 });
             }
             else {
-                throw new Error("The value provided ('" + where + "') is not one of 'beforebegin', 'afterbegin', 'beforeend', or 'afterend'");
+                throw new Error("The value provided ('".concat(where, "') is not one of 'beforebegin', 'afterbegin', 'beforeend', or 'afterend'"));
             }
             // if (!where || html === undefined || html === null) {
             // 	return;
@@ -1295,8 +1332,8 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
             pre: true,
         };
         var element_names = Object.keys(elements);
-        var kBlockTextElements = element_names.map(function (it) { return new RegExp("^" + it + "$", 'i'); });
-        var kIgnoreElements = element_names.filter(function (it) { return elements[it]; }).map(function (it) { return new RegExp("^" + it + "$", 'i'); });
+        var kBlockTextElements = element_names.map(function (it) { return new RegExp("^".concat(it, "$"), 'i'); });
+        var kIgnoreElements = element_names.filter(function (it) { return elements[it]; }).map(function (it) { return new RegExp("^".concat(it, "$"), 'i'); });
         function element_should_be_ignore(tag) {
             return kIgnoreElements.some(function (it) { return it.test(tag); });
         }
@@ -1311,7 +1348,7 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
         var noNestedTagIndex = undefined;
         var match;
         // https://github.com/taoqf/node-html-parser/issues/38
-        data = "<" + frameflag + ">" + data + "</" + frameflag + ">";
+        data = "<".concat(frameflag, ">").concat(data, "</").concat(frameflag, ">");
         var lowerCaseTagName = options.lowerCaseTagName;
         var dataEndPos = data.length - (frameflag.length + 2);
         var frameFlagOffset = frameflag.length + 2;
@@ -1379,7 +1416,7 @@ define("nodes/html", ["require", "exports", "css-select", "he", "back", "matcher
                 stack.push(currentParent);
                 if (is_block_text_element(tagName)) {
                     // Find closing tag
-                    var closeMarkup = "</" + tagName + ">";
+                    var closeMarkup = "</".concat(tagName, ">");
                     var closeIndex = lowerCaseTagName
                         ? data.toLocaleLowerCase().indexOf(closeMarkup, kMarkupPattern.lastIndex)
                         : data.indexOf(closeMarkup, kMarkupPattern.lastIndex);
@@ -1515,7 +1552,7 @@ define("nodes/comment", ["require", "exports", "nodes/node", "nodes/type"], func
             configurable: true
         });
         CommentNode.prototype.toString = function () {
-            return "<!--" + this.rawText + "-->";
+            return "<!--".concat(this.rawText, "-->");
         };
         return CommentNode;
     }(node_3.default));

package/dist/nodes/comment.js

@@ -47,7 +47,7 @@ var CommentNode = /** @class */ (function (_super) {
         configurable: true
     });
     CommentNode.prototype.toString = function () {
-        return "<!--" + this.rawText + "-->";
+        return "<!--".concat(this.rawText, "-->");
     };
     return CommentNode;
 }(node_1.default));

package/dist/nodes/html.d.ts

@@ -135,6 +135,11 @@ export default class HTMLElement extends Node {
      * @param {string} tagName the tagName of the elements to select
      */
     getElementsByTagName(tagName: string): Array<HTMLElement>;
+    /**
+     * find element by it's id
+     * @param {string} id the id of the element to select
+     */
+    getElementById(id: string): HTMLElement;
     /**
      * traverses the Element and its parents (heading toward the document root) until it finds a node that matches the provided selector string. Will return itself or the matching ancestor. If no such element exists, it returns null.
      * @param selector a DOMString containing a selector list

package/dist/nodes/html.js

@@ -86,7 +86,7 @@ var DOMTokenList = /** @class */ (function () {
     }
     DOMTokenList.prototype._validate = function (c) {
         if (/\s/.test(c)) {
-            throw new Error("DOMException in DOMTokenList.add: The token '" + c + "' contains HTML space characters, which are not valid in tokens.");
+            throw new Error("DOMException in DOMTokenList.add: The token '".concat(c, "' contains HTML space characters, which are not valid in tokens."));
         }
     };
     DOMTokenList.prototype.add = function (c) {
@@ -170,14 +170,14 @@ var HTMLElement = /** @class */ (function (_super) {
         );
         if (keyAttrs.id) {
             if (!rawAttrs) {
-                _this.rawAttrs = "id=\"" + keyAttrs.id + "\"";
+                _this.rawAttrs = "id=\"".concat(keyAttrs.id, "\"");
             }
         }
         if (keyAttrs.class) {
             if (!rawAttrs) {
-                var cls = "class=\"" + _this.classList.toString() + "\"";
+                var cls = "class=\"".concat(_this.classList.toString(), "\"");
                 if (_this.rawAttrs) {
-                    _this.rawAttrs += " " + cls;
+                    _this.rawAttrs += " ".concat(cls);
                 }
                 else {
                     _this.rawAttrs = cls;
@@ -322,7 +322,7 @@ var HTMLElement = /** @class */ (function (_super) {
                     else {
                         var text = node.trimmedText;
                         if (currentBlock.prependWhitespace) {
-                            text = " " + text;
+                            text = " ".concat(text);
                             currentBlock.prependWhitespace = false;
                         }
                         currentBlock.push(text);
@@ -343,8 +343,8 @@ var HTMLElement = /** @class */ (function (_super) {
     HTMLElement.prototype.toString = function () {
         var tag = this.rawTagName;
         if (tag) {
-            var attrs = this.rawAttrs ? " " + this.rawAttrs : '';
-            return this.isVoidElement ? "<" + tag + attrs + ">" : "<" + tag + attrs + ">" + this.innerHTML + "</" + tag + ">";
+            var attrs = this.rawAttrs ? " ".concat(this.rawAttrs) : '';
+            return this.isVoidElement ? "<".concat(tag).concat(attrs, ">") : "<".concat(tag).concat(attrs, ">").concat(this.innerHTML, "</").concat(tag, ">");
         }
         return this.innerHTML;
     };
@@ -440,9 +440,9 @@ var HTMLElement = /** @class */ (function (_super) {
                 res.push('  '.repeat(indention) + str);
             }
             function dfs(node) {
-                var idStr = node.id ? "#" + node.id : '';
-                var classStr = node.classList.length ? "." + node.classList.value.join('.') : ''; // eslint-disable-line @typescript-eslint/no-unsafe-member-access, @typescript-eslint/no-unsafe-member-access, @typescript-eslint/restrict-template-expressions, @typescript-eslint/no-unsafe-call
-                write("" + node.rawTagName + idStr + classStr);
+                var idStr = node.id ? "#".concat(node.id) : '';
+                var classStr = node.classList.length ? ".".concat(node.classList.value.join('.')) : ''; // eslint-disable-line @typescript-eslint/no-unsafe-member-access, @typescript-eslint/no-unsafe-member-access, @typescript-eslint/restrict-template-expressions, @typescript-eslint/no-unsafe-call
+                write("".concat(node.rawTagName).concat(idStr).concat(classStr));
                 indention++;
                 node.childNodes.forEach(function (childNode) {
                     if (childNode.nodeType === type_1.default.ELEMENT_NODE) {
@@ -544,6 +544,43 @@ var HTMLElement = /** @class */ (function (_super) {
         }
         return re;
     };
+    /**
+     * find element by it's id
+     * @param {string} id the id of the element to select
+     */
+    HTMLElement.prototype.getElementById = function (id) {
+        var stack = [];
+        var currentNodeReference = this;
+        var index = 0;
+        // index turns to undefined once the stack is empty and the first condition occurs
+        // which happens once all relevant children are searched through
+        while (index !== undefined) {
+            var child = void 0;
+            // make it work with sparse arrays
+            do {
+                child = currentNodeReference.childNodes[index++];
+            } while (index < currentNodeReference.childNodes.length && child === undefined);
+            // if the child does not exist we move on with the last provided index (which belongs to the parentNode)
+            if (child === undefined) {
+                currentNodeReference = currentNodeReference.parentNode;
+                index = stack.pop();
+                continue;
+            }
+            if (child.nodeType === type_1.default.ELEMENT_NODE) {
+                if (child.id === id) {
+                    return child;
+                }
+                ;
+                // if children are existing push the current status to the stack and keep searching for elements in the level below
+                if (child.childNodes.length > 0) {
+                    stack.push(index);
+                    currentNodeReference = child;
+                    index = 0;
+                }
+            }
+        }
+        return null;
+    };
     /**
      * traverses the Element and its parents (heading toward the document root) until it finds a node that matches the provided selector string. Will return itself or the matching ancestor. If no such element exists, it returns null.
      * @param selector a DOMString containing a selector list
@@ -701,7 +738,7 @@ var HTMLElement = /** @class */ (function (_super) {
             if (val === undefined || val === 'null') {
                 return name;
             }
-            return name + "=" + val;
+            return "".concat(name, "=").concat(val);
         })
             .join(' ');
         // Update this.id
@@ -748,7 +785,7 @@ var HTMLElement = /** @class */ (function (_super) {
             var val = _this.quoteAttribute(attrs[name]);
             if (val === 'null' || val === '""')
                 return name;
-            return name + "=" + val;
+            return "".concat(name, "=").concat(val);
         })
             .join(' ');
         // Update this.id
@@ -776,7 +813,7 @@ var HTMLElement = /** @class */ (function (_super) {
             var val = attributes[name];
             if (val === 'null' || val === '""')
                 return name;
-            return name + "=" + _this.quoteAttribute(String(val));
+            return "".concat(name, "=").concat(_this.quoteAttribute(String(val)));
         })
             .join(' ');
     };
@@ -818,7 +855,7 @@ var HTMLElement = /** @class */ (function (_super) {
             });
         }
         else {
-            throw new Error("The value provided ('" + where + "') is not one of 'beforebegin', 'afterbegin', 'beforeend', or 'afterend'");
+            throw new Error("The value provided ('".concat(where, "') is not one of 'beforebegin', 'afterbegin', 'beforeend', or 'afterend'"));
         }
         // if (!where || html === undefined || html === null) {
         // 	return;
@@ -1007,8 +1044,8 @@ function base_parse(data, options) {
         pre: true,
     };
     var element_names = Object.keys(elements);
-    var kBlockTextElements = element_names.map(function (it) { return new RegExp("^" + it + "$", 'i'); });
-    var kIgnoreElements = element_names.filter(function (it) { return elements[it]; }).map(function (it) { return new RegExp("^" + it + "$", 'i'); });
+    var kBlockTextElements = element_names.map(function (it) { return new RegExp("^".concat(it, "$"), 'i'); });
+    var kIgnoreElements = element_names.filter(function (it) { return elements[it]; }).map(function (it) { return new RegExp("^".concat(it, "$"), 'i'); });
     function element_should_be_ignore(tag) {
         return kIgnoreElements.some(function (it) { return it.test(tag); });
     }
@@ -1023,7 +1060,7 @@ function base_parse(data, options) {
     var noNestedTagIndex = undefined;
     var match;
     // https://github.com/taoqf/node-html-parser/issues/38
-    data = "<" + frameflag + ">" + data + "</" + frameflag + ">";
+    data = "<".concat(frameflag, ">").concat(data, "</").concat(frameflag, ">");
     var lowerCaseTagName = options.lowerCaseTagName;
     var dataEndPos = data.length - (frameflag.length + 2);
     var frameFlagOffset = frameflag.length + 2;
@@ -1091,7 +1128,7 @@ function base_parse(data, options) {
             stack.push(currentParent);
             if (is_block_text_element(tagName)) {
                 // Find closing tag
-                var closeMarkup = "</" + tagName + ">";
+                var closeMarkup = "</".concat(tagName, ">");
                 var closeIndex = lowerCaseTagName
                     ? data.toLocaleLowerCase().indexOf(closeMarkup, kMarkupPattern.lastIndex)
                     : data.indexOf(closeMarkup, kMarkupPattern.lastIndex);

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "node-html-parser",
-	"version": "5.2.2",
+	"version": "5.3.0",
 	"description": "A very fast HTML parser, generating a simplified DOM, with basic element query support.",
 	"main": "dist/index.js",
 	"types": "dist/index.d.ts",
@@ -72,7 +72,6 @@
 		"high5": "^1.0.0",
 		"html-dom-parser": "^1.0.4",
 		"html-parser": "^0.11.0",
-		"html5": "^1.0.5",
 		"html5parser": "^2.0.2",
 		"htmljs-parser": "^2.11.1",
 		"htmlparser": "^1.7.7",