Optimize skipToLeadingKeyword for the common ViewedContent case

src/asm/lexer.cpp

@@ -2147,47 +2147,92 @@ static Token yylex_RAW() {
 	return Token(T_(YYEOF));
 }
 
-// This function is called when capturing `REPT`/`FOR` loops and `MACRO` bodies,
-// and when skipping unexecuted `IF`/`ELIF`/`ELSE` blocks and `REPT`/`FOR` loops.
-// It expects that these constructs' `ENDC`/`ENDR`/`ENDM` closing tokens are only
-// valid at the start of their lines, which enables ignoring everything except
-// the leading keyword in lines that have one (as well as line continuations).
-//
-// The only keywords it needs to recognize are case-insensitive `IF`, `ELIF`,
-// `ELSE`, `ENDC`, `REPT`, `FOR`, `ENDR`, and `ENDM` (not `MACRO`).
-//
-// Note that when these constructs are *evaluated*, they can perform expansions
-// (for macro args, interpolations, and macro invocations) which may produce
-// tokens that would change how these constructs were captured or skipped, if
-// they had been produced during the capture/skip non-evaluating phase.
-static Token skipToLeadingKeyword() {
-	assume(!lexerState->enableExpansions);
-
+template<bool Quick, typename PeekFnT, typename ShiftFnT, typename NextLineFnT>
+static Token skipToLeadingKeyword(PeekFnT peekFn, ShiftFnT shiftFn, NextLineFnT nextLineFn) {
 	for (;;) {
+		int c = peekFn();
 		if (lexerState->atLineStart) {
 			lexerState->atLineStart = false;
-			if (int c = skipChars(isBlankSpace); c == EOF) {
+			while (isBlankSpace(c)) {
+				shiftFn();
+				c = peekFn();
+			}
+			if (c == EOF) {
 				return Token(T_(YYEOF));
 			} else if (isLetter(c)) {
 				std::string keyword(1, c);
-				for (c = nextChar(); continuesIdentifier(c); c = nextChar()) {
+				shiftFn();
+				for (c = peekFn(); continuesIdentifier(c); c = peekFn()) {
 					keyword += c;
+					shiftFn();
 				}
 				if (auto search = keywords.find(keyword); search != keywords.end()) {
+					// There has been one more call to `peekFn` than to `shiftFn`.
+					// If they are optimized "quick" functions for `ViewedContent`,
+					// instead of `peek` and `shiftChar`, they have not updated
+					// `lexerState->expansionScanDistance`, so it must be incremented
+					// if it was previously zero.
+					if (Quick && lexerState->expansionScanDistance == 0) {
+						++lexerState->expansionScanDistance;
+					}
 					return Token(search->second);
 				}
 			}
 		}
-		if (int c = bumpChar(); c == EOF) {
+		shiftFn();
+		if (c == EOF) {
 			return Token(T_(YYEOF));
 		} else if (isNewline(c)) {
-			handleCRLF(c);
-			nextLine();
+			// Like `handleCRLF` but calling generic `shiftFn`
+			if (c == '\r' && peekFn() == '\n') {
+				shiftFn();
+			}
+			nextLineFn();
 			lexerState->atLineStart = true;
 		}
 	}
 }
 
+// This function is called when capturing `REPT`/`FOR` loops and `MACRO` bodies,
+// and when skipping unexecuted `IF`/`ELIF`/`ELSE` blocks and `REPT`/`FOR` loops.
+// It expects that these constructs' `ENDC`/`ENDR`/`ENDM` closing tokens are only
+// valid at the start of their lines, which enables ignoring everything except
+// the leading keyword in lines that have one (as well as line continuations).
+//
+// The only keywords it needs to recognize are case-insensitive `IF`, `ELIF`,
+// `ELSE`, `ENDC`, `REPT`, `FOR`, `ENDR`, and `ENDM` (not `MACRO`).
+//
+// Note that when these constructs are *evaluated*, they can perform expansions
+// (for macro args, interpolations, and macro invocations) which may produce
+// tokens that would change how these constructs were captured or skipped, if
+// they had been produced during the capture/skip non-evaluating phase.
+static Token skipToLeadingKeyword() {
+	assume(!lexerState->enableExpansions);
+
+	if (std::holds_alternative<ViewedContent>(lexerState->content)
+	    && lexerState->expansionStack.empty()) {
+		// Optimize the common case (a fully-read assembly file without ongoing
+		// expansions) to avoid the bookkeeping of `peek` and `shiftChar`.
+		auto &view = std::get<ViewedContent>(lexerState->content);
+		char const *ptr = view.span.ptr.get();
+		auto quickPeek = [&]() { return view.offset < view.span.size ? ptr[view.offset] : EOF; };
+		auto quickNextLine = []() { ++lexerState->lineNo; };
+		if (lexerState->capturing) {
+			assume(lexerState->captureBuf == nullptr);
+			auto quickCaptureShiftChar = [&]() {
+				++view.offset;
+				++lexerState->captureSize;
+			};
+			return skipToLeadingKeyword<true>(quickPeek, quickCaptureShiftChar, quickNextLine);
+		} else {
+			auto quickShiftChar = [&]() { ++view.offset; };
+			return skipToLeadingKeyword<true>(quickPeek, quickShiftChar, quickNextLine);
+		}
+	} else {
+		return skipToLeadingKeyword<false>(peek, shiftChar, nextLine);
+	}
+}
+
 static Token skipIfBlock(bool toEndc) {
 	lexer_SetMode(LEXER_NORMAL);