# HG changeset patch
# User ali <ali@juiblex.co.uk>
# Date 1374314822 -3600
# Node ID 4e6e7cc6b50d5ca0efdfa91d3e2f9e03234ef292
# Parent  dab0af6882ee476932ff6fcc79e8e46bc8461f6a
Give a useful error if input is neither UTF-8 nor Windows-1252

diff -r dab0af6882ee -r 4e6e7cc6b50d bookloupe/bookloupe.c
--- a/bookloupe/bookloupe.c	Thu May 30 20:27:48 2013 +0100
+++ b/bookloupe/bookloupe.c	Sat Jul 20 11:07:02 2013 +0100
@@ -404,8 +404,10 @@
  */
 gchar *read_etext(const char *filename,GError **err)
 {
+    GError *tmp_err=NULL;
     gchar *contents,*utf8;
-    gsize len,nb;
+    gsize len,bytes_read,bytes_written;
+    int i,line,col;
     if (!g_file_get_contents(filename,&contents,&len,err))
 	return NULL;
     if (g_utf8_validate(contents,len,NULL))
@@ -418,7 +420,27 @@
     }
     else
     {
-	utf8=g_convert(contents,len,"UTF-8","WINDOWS-1252",NULL,&nb,NULL);
+	utf8=g_convert(contents,len,"UTF-8","WINDOWS-1252",&bytes_read,
+	  &bytes_written,&tmp_err);
+	if (g_error_matches(tmp_err,G_CONVERT_ERROR,
+	  G_CONVERT_ERROR_ILLEGAL_SEQUENCE))
+	{
+	    line=col=1;
+	    for(i=0;i<bytes_read;i++)
+		if (contents[i]=='\n')
+		{
+		    line++;
+		    col=1;
+		}
+		else if (contents[i]!='\r')
+		    col++;
+	    g_set_error(err,G_CONVERT_ERROR,G_CONVERT_ERROR_ILLEGAL_SEQUENCE,
+	      "Input conversion failed. Byte %d at line %d, column %d is not a "
+	      "valid Windows-1252 character",
+	      ((unsigned char *)contents)[bytes_read],line,col);
+	}
+	else if (tmp_err)
+	    g_propagate_error(err,tmp_err);
 	g_set_print_handler(print_as_windows_1252);
 #ifdef __WIN32__
 	SetConsoleOutputCP(1252);