From e1b370f6269a0bffac54fe20dfcb7ecc3d1a5fc1 Mon Sep 17 00:00:00 2001
From: Victor Wagner <vitus@wagner.pp.ru>
Date: Sat, 15 Nov 2025 11:35:00 +0300
Subject: [PATCH] Fix handling of dest charsets with chars > u7FFF and missing
 chasets

Prevoisly, if unexisting charset was specified in the command line,
catdoc/xls2csv/catppt silently falls back to default charset. Now it
exits with non-zero exit code and appropriate message

It was also found that in make_reverse_map unicode code was incorrectly
casted to (unsigned) instead of (unsigned short) which cause segfault
if charater with 16-bit unicode code highest bit set present in the
destination charset (i.e. in mac-roman).
---
 src/catdoc.h   |  2 +-
 src/charsets.c |  2 +-
 src/confutil.c | 23 ++++++++++-------------
 src/fileutil.c | 13 +++++++------
 4 files changed, 19 insertions(+), 21 deletions(-)

diff --git a/src/catdoc.h b/src/catdoc.h
index 3a4010f..4c189de 100644
--- a/src/catdoc.h
+++ b/src/catdoc.h
@@ -233,7 +233,7 @@ extern int longest_sequence;/* for checking which value of wrap_margin
                              can cause buffer overflow*/
 char *map_subst(SUBSTMAP map,int uc);
 
-int check_charset(char **filename,const char *charset);
+void check_charset(char **filename,const char *charset);
 int process_file(FILE *f,long stop);
 void copy_out(FILE *f, char *header);
 void output_paragraph(unsigned short int *buffer) ;
diff --git a/src/charsets.c b/src/charsets.c
index e64e24c..3f63e74 100644
--- a/src/charsets.c
+++ b/src/charsets.c
@@ -51,7 +51,7 @@ CHARSET make_reverse_map(short int *charset) {
 	}	
 	for (i=0;i<256;i++) {
 		k= charset[i];
-		j=  (unsigned)k>>8;
+		j=  (unsigned short int)k>>8;
 		if (!newmap[j]) {
 			newmap[j] = (short int *)malloc(sizeof(short int)*256);
 			if (!newmap[j]) {
diff --git a/src/confutil.c b/src/confutil.c
index 1bfe199..a10aa67 100644
--- a/src/confutil.c
+++ b/src/confutil.c
@@ -110,10 +110,9 @@ void get_locale_charset() {
 		if (!strncmp(codeset,"646",3)) {
 			/* ISO 646 is another name for us=ascii */
 			check_charset(&dest_csname,"us-ascii") ;
-		}	else {	 
-			if (check_charset(&dest_csname,codeset)) {
-				locale_charset = dest_csname;
-			}
+		} else {	 
+			check_charset(&dest_csname,codeset);
+			locale_charset = dest_csname;
 		}
 	} else if (!strcmp(codeset,"ANSI_X3.4-1968")) {
 		check_charset(&dest_csname,"us-ascii");
@@ -125,9 +124,8 @@ void get_locale_charset() {
 		newstr	= malloc(strlen(codeset)-4+2+1);
 		strcpy(newstr,"cp");
 		strcpy(newstr+2,codeset+4);
-		if (check_charset(&dest_csname,newstr)) {
-			locale_charset = dest_csname;
-		}
+		check_charset(&dest_csname,newstr);
+		locale_charset = dest_csname;
 		free(newstr);
 	} else if (!strncmp(codeset,"IBM",3)) {
 		char *newstr;
@@ -136,18 +134,17 @@ void get_locale_charset() {
 		newstr=malloc(strlen(codeset)+2+1);
 		strcpy(newstr,"cp");
 		strcpy(newstr+2,codeset);
-		if (check_charset(&dest_csname, newstr)) {
-			locale_charset=dest_csname;
-		}
+		check_charset(&dest_csname, newstr);
+		locale_charset=dest_csname;
 		free(newstr);
 	} else {
 		char *i,*newstr = strdup(codeset);
 		for (i=newstr;*i;i++) {
 			*i=tolower(*i);
 		}	
-		if (check_charset(&dest_csname,newstr)) {
-			locale_charset = dest_csname;
-		}	
+		check_charset(&dest_csname,newstr);
+		locale_charset = dest_csname;
+		
 	}	
 
 }	
diff --git a/src/fileutil.c b/src/fileutil.c
index fc8e7f7..201f259 100644
--- a/src/fileutil.c
+++ b/src/fileutil.c
@@ -95,24 +95,25 @@ char *find_file(char *name, const char *path)
 /************************************************************************/
 /* Searches for charset with given name and put pointer to malloced copy*/
 /* of its name into first arg if found. Otherwise leaves first arg      */
-/*  unchanged. Returns non-zero on success                              */ 
+/*  unchanged. Terminatnes program if charset not  found */ 
 /************************************************************************/
-int check_charset(char **filename,const char *charset) {
+void check_charset(char **filename,const char *charset) {
 	char *tmppath;
 	if (charset == NULL ) {
-		return 0;
+		exit(1);
 	}
 	if (!strncmp(charset,"utf-8",6)) {
 		*filename=strdup("utf-8");
-		return 1;
+		return;
 	}   
 	tmppath=find_file(stradd(charset,CHARSET_EXT),charset_path);
 	if (tmppath && *tmppath) {
 			*filename=strdup(charset);
 			free(tmppath);
-			return 1;
+			return;
 	} 
-	return 0;
+	fprintf(stderr, "charset %s not found\n", charset);
+	exit(1);
 }
 
 /**********************************************************************/
-- 
2.47.3