From 69eadb67d2e5b7c2207de066c1a13aaa8614bf00 Mon Sep 17 00:00:00 2001
From: Lephenixnoir <sebastien.michelland@protonmail.com>
Date: Sun, 14 Jan 2024 20:07:24 +0100
Subject: [PATCH] stdio: deduplicate scanf cases and remove most gotos

---
 src/stdio/scanf/scan.c | 257 +++++++++++------------------------------
 1 file changed, 65 insertions(+), 192 deletions(-)

diff --git a/src/stdio/scanf/scan.c b/src/stdio/scanf/scan.c
index 9643852..f6684ca 100644
--- a/src/stdio/scanf/scan.c
+++ b/src/stdio/scanf/scan.c
@@ -161,9 +161,9 @@ void __define_set_range( char c1, char c2, bool value )
 }
 
 /* return true if the char is in the allowed set or false otherwise */
-bool __is_allowed( const unsigned char c )
+bool __is_allowed(int c)
 {
-	return __asciiallowed[ c ];
+	return (c != EOF) && __asciiallowed[c];
 }
 
 
@@ -200,37 +200,34 @@ int __scanset(char const * __restrict__ format, int *pos )
 		}
 	}
 
-	// start of format analysis loop
-	loopset:
-
-	// we find a '-' so need to check if we are considering a range or the char '-' only
-	if (format[*pos]=='-') {
-		// the char '-' is included in the allowed set
-		if (format[*pos+1]==']') {
-			__asciiallowed[ '-' ] = !__neg;	// if set in very final position before']', this is the char '-' only
-			(*pos)++;
-			// we have now finished the reading of the set cause the following char is ']'
-			return 0;
+	while(1) {
+		// we find a '-' so need to check if we are considering a range or the char '-' only
+		if (format[*pos]=='-') {
+			// the char '-' is included in the allowed set
+			if (format[*pos+1]==']') {
+				__asciiallowed[ '-' ] = !__neg;	// if set in very final position before']', this is the char '-' only
+				(*pos)++;
+				// we have now finished the reading of the set cause the following char is ']'
+				return 0;
+			}
+			// the char '-' indicates a range of char to be included into the set
+			else {
+				(*pos)++;
+				__eor = format[*pos];
+				__define_set_range( __sor, __eor, !__neg );
+			}
 		}
-		// the char '-' indicates a range of char to be included into the set
+		// we find the char ']' so it means we reach the end of this set
+		else if (format[*pos]==']') return 0;
+		// if we reach the '\0' we have a syntax problem
+		else if (format[*pos]=='\0') return -1;
+		// we are considering one particular char and prepare for a potential range if we find the char '-' later on
 		else {
+			__sor = format[*pos];
+			__asciiallowed[ __sor ] = !__neg;
 			(*pos)++;
-			__eor = format[*pos];
-			__define_set_range( __sor, __eor, !__neg );
-			goto loopset;
 		}
 	}
-	// we find the char ']' so it means we reach the end of this set
-	else if (format[*pos]==']') return 0;
-	// if we reach the '\0' we have a syntax problem
-	else if (format[*pos]=='\0') return -1;
-	// we are considering one particular char and prepare for a potential range if we find the char '-' later on
-	else {
-		__sor = format[*pos];
-		__asciiallowed[ __sor ] = !__neg;
-		(*pos)++;
-		goto loopset;
-	}
 }
 
 
@@ -255,6 +252,8 @@ int __scanf(
 
 	__scanf_start( in );
 
+	// TODO: No __scanf_end() in any of the "return validrets"!!
+
 	while( format[pos] != 0 ) {
 		in->readmaxlength = -1;
 		user_length = 0;
@@ -278,83 +277,26 @@ int __scanf(
 			switch( format[pos] ) {
 				// we need to decrypt the corresponding scanf set of character
 				case '[': {
-					char temp;
 					err = __scanset( format, &pos );
 					if (err!=0) return validrets;
 					int currentlength = 0;
 					// we need to assign the read char to the corresponding pointer 
-					if (!skip) {
-						char *c = (char *) va_arg( *args, char* );
-						if (in->readmaxlength==INT_MAX) {
-							for(;;) {
-								temp = __scanf_peek( in );
-								if (temp==EOF) return EOF;
-								else if (__is_allowed( temp )) {
-									*c++ = __scanf_in( in );
-									currentlength++;
-								}
-								else {
-									if (currentlength>0) {
-										*c = '\0';
-										validrets++;
-										goto exitset;
-									}
-									else return validrets;
-								}
-							}
-						}
-						else {
-							for(unsigned int u=0; u<in->readmaxlength; u++) {
-								temp = __scanf_peek( in );
-								if (temp==EOF) return EOF;
-								else if (__is_allowed( temp ))	{
-									*c++ = __scanf_in( in );
-									currentlength++;
-								}
-								else {
-									if (currentlength>0) {
-										*c = '\0';
-										validrets++;
-										goto exitset;
-									}
-									else return validrets;
-								}
-							}
-							if (currentlength>0) {
-								*c = '\0';
-								validrets++;
-								goto exitset;
-							}
-							else return validrets;
+					char *c = skip ? NULL : va_arg(*args, char *);
+					for(unsigned int u=0; u<in->readmaxlength; u++) {
+						int temp = __scanf_peek(in);
+						if(__is_allowed(temp)) {
+							__scanf_in(in);
+							if(c) *c++ = temp;
+							currentlength++;
 						}
+						else if(temp==EOF && !currentlength && !validrets)
+							return EOF;
+						else break;
 					}
-
-					else
-					{
-						if (in->readmaxlength==INT_MAX) {
-							for(;;)	{
-								temp = __scanf_peek( in );
-								if (temp==EOF) return EOF;
-								else if (__is_allowed( temp ))	{
-									__scanf_in( in );
-									currentlength++;
-								} 
-								else if (currentlength>0) goto exitset;
-								else return validrets;
-							}
-						}
-						else {
-							for(unsigned int u=0; u<in->readmaxlength; u++) {
-								temp = __scanf_peek( in );
-								if (temp==EOF) return EOF;
-								else if (__is_allowed( temp )) __scanf_in( in );
-								else if (currentlength>0) goto exitset;
-								else return validrets;
-							}
-						}
-					}
-
-					exitset:
+					if(!currentlength)
+						return validrets;
+					*c = '\0';
+					validrets += !skip;
 					break;
 				}
 				
@@ -499,111 +441,42 @@ int __scanf(
 				}
 
 				case 'c': {
-					int temp;
-					if (!skip) {
-						char *c = (char *) va_arg( *args, char* );
-						if (in->readmaxlength==INT_MAX) {
-							temp = __scanf_peek( in );
-							if (temp==EOF) return EOF;
-							else *c = __scanf_in( in );
-						}
-						else {
-							for( unsigned int u = 0; u < in->readmaxlength; u++ ) {
-								temp = __scanf_peek( in );
-								if (temp==EOF) return EOF;
-								else *c++ = __scanf_in( in );
-							}
-						}
-					}
-					else {
-						if (in->readmaxlength==INT_MAX) {
-							temp = __scanf_peek( in );
-							if (temp==EOF) return EOF;
-							else {
-								__scanf_in( in );
-								goto endcharskip;
-							} 
-						}
-						else {
-							for( unsigned int u = 0; u < in->readmaxlength; u++ ) {
-								temp = __scanf_peek( in );
-								if (temp==EOF) return EOF;
-								else __scanf_in( in );
-							}
-							goto endcharskip;
-						}
-					}
-					
-					validrets++;
+					if(in->readmaxlength == INT_MAX)
+						in->readmaxlength = 1;
+					char *c = skip ? NULL : va_arg(*args, char *);
 
-					endcharskip:
+					for(unsigned int u = 0; u < in->readmaxlength; u++) {
+						int temp = __scanf_in(in);
+						if(temp==EOF) return EOF;
+						else if(c) *c++ = temp;
+					}
+					validrets += !skip;
 					break;
 				}
 
 				case 's': {
 					char temp;
 					int curstrlength = 0;
-					__purge_space( in );
-					if (!skip)	{
-						char *c = (char *) va_arg( *args, char* );
-						if (in->readmaxlength==INT_MAX) {
-							loopstring:
-							temp = __scanf_peek( in );
-							if (temp==EOF && curstrlength==0) return validrets;
-							if (isspace(temp) || ( (temp==EOF && curstrlength!=0) )) {
-									*c = 0;
-									goto loopstringend;
-							}
-							else {
-								*c++ = __scanf_in( in );
-								curstrlength++;
-								goto loopstring;
+					__purge_space(in);
+
+					char *c = skip ? NULL : va_arg(*args, char *);
+					for(unsigned int u = 0; u < in->readmaxlength; u++) {
+						temp = __scanf_peek(in);
+						if(temp==EOF && curstrlength==0) return validrets;
+						if(isspace(temp) || ((temp==EOF && curstrlength!=0))) {
+							if(c) {
+								*c = 0;
+								validrets++;
 							}
+							break;
 						}
 						else {
-							for( unsigned int u = 0; u < in->readmaxlength; u++ ) {
-								temp = __scanf_peek( in );					
-								if (temp==EOF && curstrlength==0) return validrets;
-								if (isspace(temp) || ( (temp==EOF && curstrlength!=0) )) {
-									*c  = 0;
-									goto loopstringend;
-								}
-								else {
-									*c++ = __scanf_in( in );
-									curstrlength++;
-								}
-							}
+							int temp = __scanf_in( in );
+							if(c)
+								*c++ = temp;
+							curstrlength++;
 						}
 					}
-					else {
-						if (in->readmaxlength==INT_MAX) {
-							loopstringskip:
-							temp = __scanf_peek( in );
-							if (temp==EOF && curstrlength==0) return validrets;
-							if (isspace(temp) || ( (temp==EOF && curstrlength!=0) )) goto loopstringendskip;
-							else {
-								__scanf_in( in );
-								curstrlength++;
-								goto loopstringskip;
-							}
-						}
-						else {
-							for( unsigned int u = 0; u < in->readmaxlength; u++ ) {
-								temp = __scanf_peek( in );							
-								if (temp==EOF && curstrlength==0) return validrets;
-								if (isspace(temp) || ( (temp==EOF && curstrlength!=0) )) goto loopstringendskip;
-								else {
-									__scanf_in( in );
-									curstrlength++;
-								}
-							}
-						}
-					}
-
-					loopstringend:
-					validrets++;
-
-					loopstringendskip:
 					break;
 				}
 			}