@cj-tech-master/excelts 7.5.0 → 7.6.0

package/dist/iife/excelts.iife.js

@@ -1,5 +1,5 @@
 /*!
-* @cj-tech-master/excelts v7.5.0
+* @cj-tech-master/excelts v7.6.0
 * TypeScript Excel Workbook Manager - Read and Write xlsx and csv Files.
 * (c) 2026 cjnoname
 * Released under the MIT License
@@ -11429,6 +11429,8 @@ var ExcelTS = (function(exports) {
 	const QUESTION = 63;
 	const OPEN_BRACKET = 91;
 	const CLOSE_BRACKET = 93;
+	const REPLACEMENT_CHAR = 65533;
+	const REPLACEMENT_STR = "�";
 	const ASCII_CHARS = /* @__PURE__ */ (() => {
 		const t = new Array(128);
 		for (let i = 0; i < 128; i++) t[i] = String.fromCharCode(i);
@@ -11557,6 +11559,7 @@ var ExcelTS = (function(exports) {
 			this.xmlns = options?.xmlns ?? false;
 			this.maxDepth = options?.maxDepth !== void 0 ? options.maxDepth : 256;
 			this.maxEntityExpansions = options?.maxEntityExpansions !== void 0 ? options.maxEntityExpansions : 1e4;
+			this.invalidCharHandling = options?.invalidCharHandling ?? "error";
 			this._init();
 		}
 		get closed() {
@@ -11650,55 +11653,150 @@ var ExcelTS = (function(exports) {
 		close() {
 			return this.write(null);
 		}
-		getCode() {
-			const { chunk, i } = this;
-			this.prevI = i;
-			this.i = i + 1;
-			if (i >= chunk.length) return -1;
-			const code = chunk.charCodeAt(i);
-			if (code >= 32 && code <= 126) {
-				if (this.trackPosition) this.column++;
-				return code;
-			}
-			if (code === TAB) {
-				if (this.trackPosition) this.column++;
-				return code;
+		/**
+		* Handle an invalid XML character according to the configured strategy.
+		*
+		* Used by `handleTextInRoot()` fast path which manages its own text accumulation
+		* and cannot use the `getCode()` loop approach.
+		*
+		* - `"error"`: call `fail()` and return the original code.
+		* - `"skip"`: return `REPLACEMENT_CHAR` as a sentinel (caller handles skip).
+		* - `"replace"`: return `REPLACEMENT_CHAR`.
+		*
+		* Note: For `getCode()`, invalid char handling is inlined to avoid recursion.
+		*
+		* @param code - The invalid character code point.
+		* @param kind - Optional description (e.g. "lone surrogate") for error messages.
+		* @returns The code point to use.
+		*/
+		handleInvalidChar(code, kind) {
+			switch (this.invalidCharHandling) {
+				case "replace": return REPLACEMENT_CHAR;
+				case "skip": return -2;
+				default: {
+					const label = kind ? `invalid XML character: ${kind} 0x${code.toString(16)}` : `invalid XML character: 0x${code.toString(16)}`;
+					this.fail(label);
+					return code;
+				}
 			}
-			if (code === CR) {
-				if (chunk.charCodeAt(i + 1) === NL) this.i = i + 2;
-				if (this.trackPosition) {
-					this.line++;
-					this.column = 0;
-					this.positionAtNewLine = this.position;
+		}
+		/**
+		* Handle an invalid character inside the `handleTextInRoot()` fast loop.
+		*
+		* Unlike `handleInvalidChar()` (which returns a code point for `getCode()`),
+		* this method manages the text accumulation state (`this.text`, `start`) that
+		* the fast text loop relies on.
+		*
+		* - `"error"`: call `fail()`, leave text accumulation unchanged (char stays in output).
+		* - `"skip"`: flush text up to the invalid char, skip it, return new `start`.
+		* - `"replace"`: flush text up to the invalid char, append U+FFFD, return new `start`.
+		*
+		* @returns The updated `start` index for the text accumulation loop.
+		*/
+		handleInvalidCharInText(code, handler, start, kind) {
+			switch (this.invalidCharHandling) {
+				case "skip":
+					if (handler && start < this.prevI) this.text += this.chunk.slice(start, this.prevI);
+					return this.i;
+				case "replace":
+					if (handler) {
+						if (start < this.prevI) this.text += this.chunk.slice(start, this.prevI);
+						this.text += REPLACEMENT_STR;
+					}
+					return this.i;
+				default: {
+					const label = kind ? `invalid XML character: ${kind} 0x${code.toString(16)}` : `invalid XML character: 0x${code.toString(16)}`;
+					this.fail(label);
+					return start;
 				}
-				return NL;
 			}
-			if (code === NL) {
-				if (this.trackPosition) {
-					this.line++;
-					this.column = 0;
-					this.positionAtNewLine = this.position;
+		}
+		/**
+		* Handle an invalid character inside `sAttribValueQuoted()`.
+		*
+		* Same pattern as `handleInvalidCharInText()` but for attribute value
+		* accumulation (always uses `this.text`, no conditional handler check).
+		*
+		* @returns The updated `start` index.
+		*/
+		handleInvalidCharInAttr(code, start, kind) {
+			switch (this.invalidCharHandling) {
+				case "skip":
+					if (start < this.prevI) this.text += this.chunk.slice(start, this.prevI);
+					return this.i;
+				case "replace":
+					if (start < this.prevI) this.text += this.chunk.slice(start, this.prevI);
+					this.text += REPLACEMENT_STR;
+					return this.i;
+				default: {
+					const label = kind ? `invalid XML character: ${kind} 0x${code.toString(16)}` : `invalid XML character: 0x${code.toString(16)}`;
+					this.fail(label);
+					return start;
 				}
-				return NL;
 			}
-			if (code >= 55296 && code <= 56319) {
-				const next = chunk.charCodeAt(i + 1);
-				if (next >= 56320 && next <= 57343) {
-					this.i = i + 2;
+		}
+		getCode() {
+			for (;;) {
+				const { chunk } = this;
+				const i = this.i;
+				this.prevI = i;
+				this.i = i + 1;
+				if (i >= chunk.length) return -1;
+				const code = chunk.charCodeAt(i);
+				if (code >= 32 && code <= 126) {
 					if (this.trackPosition) this.column++;
-					return 65536 + ((code - 55296) * 1024 + (next - 56320));
+					return code;
+				}
+				if (code === TAB) {
+					if (this.trackPosition) this.column++;
+					return code;
+				}
+				if (code === CR) {
+					if (chunk.charCodeAt(i + 1) === NL) this.i = i + 2;
+					if (this.trackPosition) {
+						this.line++;
+						this.column = 0;
+						this.positionAtNewLine = this.position;
+					}
+					return NL;
+				}
+				if (code === NL) {
+					if (this.trackPosition) {
+						this.line++;
+						this.column = 0;
+						this.positionAtNewLine = this.position;
+					}
+					return NL;
+				}
+				if (code >= 55296 && code <= 56319) {
+					const next = chunk.charCodeAt(i + 1);
+					if (next >= 56320 && next <= 57343) {
+						this.i = i + 2;
+						if (this.trackPosition) this.column++;
+						return 65536 + ((code - 55296) * 1024 + (next - 56320));
+					}
+					const result = this.handleInvalidChar(code, "lone surrogate");
+					if (result !== -2) return result;
+					continue;
+				}
+				if (code >= 56320 && code <= 57343) {
+					const result = this.handleInvalidChar(code, "lone surrogate");
+					if (result !== -2) return result;
+					continue;
+				}
+				if (code >= 128) {
+					if (this.trackPosition) this.column++;
+					if (code === 65534 || code === 65535) {
+						const result = this.handleInvalidChar(code);
+						if (result !== -2) return result;
+						continue;
+					}
+					return code;
 				}
-				this.fail("invalid XML character: lone surrogate 0x" + code.toString(16));
-			}
-			if (code >= 56320 && code <= 57343) this.fail("invalid XML character: lone surrogate 0x" + code.toString(16));
-			if (code >= 128) {
 				if (this.trackPosition) this.column++;
-				if (code === 65534 || code === 65535) this.fail("invalid XML character: 0x" + code.toString(16));
-				return code;
+				const result = this.handleInvalidChar(code);
+				if (result !== -2) return result;
 			}
-			if (this.trackPosition) this.column++;
-			this.fail("invalid XML character: 0x" + code.toString(16));
-			return code;
 		}
 		unget() {
 			this.i = this.prevI;
@@ -11863,14 +11961,14 @@ var ExcelTS = (function(exports) {
 						if (next >= 56320 && next <= 57343) this.i += 2;
 						else {
 							this.i++;
-							this.fail("invalid XML character: lone surrogate 0x" + code.toString(16));
+							start = this.handleInvalidCharInText(code, handler, start, "lone surrogate");
 						}
 					} else if (code >= 56320 && code <= 57343) {
 						this.i++;
-						this.fail("invalid XML character: lone surrogate 0x" + code.toString(16));
+						start = this.handleInvalidCharInText(code, handler, start, "lone surrogate");
 					} else if (code === 65534 || code === 65535) {
 						this.i++;
-						this.fail("invalid XML character: 0x" + code.toString(16));
+						start = this.handleInvalidCharInText(code, handler, start);
 					} else this.i++;
 					if (this.trackPosition) this.column++;
 					continue;
@@ -11878,7 +11976,7 @@ var ExcelTS = (function(exports) {
 				this.prevI = this.i;
 				this.i++;
 				if (this.trackPosition) this.column++;
-				this.fail("invalid XML character: 0x" + code.toString(16));
+				start = this.handleInvalidCharInText(code, handler, start);
 			}
 			if (handler && start < this.i) this.text += chunk.slice(start, this.i);
 		}
@@ -11887,12 +11985,28 @@ var ExcelTS = (function(exports) {
 			let { i: start } = this;
 			const handler = this._handlers.text;
 			let nonSpace = false;
+			const isSkip = this.invalidCharHandling === "skip";
+			const isReplace = this.invalidCharHandling === "replace";
 			while (true) {
+				const iBeforeGet = this.i;
 				const c = this.getCode();
 				if (c === -1) {
-					if (handler && start < this.i) this.text += chunk.slice(start, this.i);
+					if (handler && start < iBeforeGet) this.text += chunk.slice(start, iBeforeGet);
 					break;
 				}
+				if (isSkip && this.prevI > iBeforeGet) {
+					if (handler && start < iBeforeGet) this.text += chunk.slice(start, iBeforeGet);
+					start = this.prevI;
+				}
+				if (isReplace && c === REPLACEMENT_CHAR && chunk.charCodeAt(this.prevI) !== REPLACEMENT_CHAR) {
+					if (handler) {
+						if (start < this.prevI) this.text += chunk.slice(start, this.prevI);
+						this.text += REPLACEMENT_STR;
+					}
+					start = this.i;
+					nonSpace = true;
+					continue;
+				}
 				if (c === LESS) {
 					if (handler) {
 						const slice = chunk.slice(start, this.prevI);
@@ -12174,10 +12288,29 @@ var ExcelTS = (function(exports) {
 					start = this.i;
 					continue;
 				}
-				if (this.getCode() === -1) {
-					this.text += chunk.slice(start, this.i);
-					return;
+				if (code >= 128) {
+					this.prevI = this.i;
+					if (code >= 55296 && code <= 56319) {
+						const next = chunk.charCodeAt(this.i + 1);
+						if (next >= 56320 && next <= 57343) this.i += 2;
+						else {
+							this.i++;
+							start = this.handleInvalidCharInAttr(code, start, "lone surrogate");
+						}
+					} else if (code >= 56320 && code <= 57343) {
+						this.i++;
+						start = this.handleInvalidCharInAttr(code, start, "lone surrogate");
+					} else if (code === 65534 || code === 65535) {
+						this.i++;
+						start = this.handleInvalidCharInAttr(code, start);
+					} else this.i++;
+					if (this.trackPosition) this.column++;
+					continue;
 				}
+				this.prevI = this.i;
+				this.i++;
+				if (this.trackPosition) this.column++;
+				start = this.handleInvalidCharInAttr(code, start);
 			}
 			this.text += chunk.slice(start, this.i);
 		}
@@ -12688,7 +12821,7 @@ var ExcelTS = (function(exports) {
 		* Use this instead of parse(parseSax(stream)) for hot paths.
 		*/
 		async parseStreamDirect(stream) {
-			const parser = new SaxParser();
+			const parser = new SaxParser({ invalidCharHandling: "skip" });
 			const decoder = new TextDecoder("utf-8", { fatal: true });
 			let done = false;
 			let finalModel;
@@ -38161,7 +38294,10 @@ onmessage = async (ev) => {
 			let c = null;
 			let current = null;
 			let worksheetEvents = null;
-			const parser = new SaxParser({ position: false });
+			const parser = new SaxParser({
+				position: false,
+				invalidCharHandling: "skip"
+			});
 			parser.on("opentag", (node) => {
 				if (emitSheet) switch (node.name) {
 					case "cols":
@@ -38380,7 +38516,10 @@ onmessage = async (ev) => {
 				return;
 			}
 			try {
-				const parser = new SaxParser({ position: false });
+				const parser = new SaxParser({
+					position: false,
+					invalidCharHandling: "skip"
+				});
 				const decoder = new TextDecoder("utf-8", { fatal: true });
 				parser.on("opentag", (node) => {
 					if (node.name !== "Relationship") return;
@@ -38558,7 +38697,10 @@ onmessage = async (ev) => {
 			let inRichText = false;
 			if (this.options.sharedStrings === "cache") {
 				const sharedStrings = this.sharedStrings;
-				const parser = new SaxParser({ position: false });
+				const parser = new SaxParser({
+					position: false,
+					invalidCharHandling: "skip"
+				});
 				parser.on("opentag", (node) => {
 					switch (node.name) {
 						case "b":
@@ -38652,7 +38794,7 @@ onmessage = async (ev) => {
 				await saxStream(parser, iterateStream(entry));
 				return;
 			}
-			const emitParser = new SaxParser();
+			const emitParser = new SaxParser({ invalidCharHandling: "skip" });
 			const emitDecoder = new TextDecoder("utf-8", { fatal: true });
 			let pendingEmits = [];
 			emitParser.on("opentag", (node) => {