Fix bug #948: make HIGH/LOW-VALUE sensitive to program collating sequence

.github/workflows/windows-msvc.yml

@@ -178,18 +178,14 @@ jobs:
           # sed -i '/AT_SETUP(\[runtime check: write to internal storage (1)\])/a AT_SKIP_IF(\[true\])' tests/testsuite.src/run_misc.at
           sed -i '/run_misc/{N;/write to internal storage (1)/{N;N;N;N;s/traceon/traceon; echo "workflow:1">"$at_check_line_file"; at_fn_check_skip 77/;}}' tests/testsuite
 
-        # Fail two tests that behave differently under MSVC Debug
+        # Fail tests that behave differently under MSVC Debug
         # - System routine CBL_GC_HOSTED: fails because libcob is linked with the debug version
         #   of the C runtime while the generated module is linked with the release version
-        # - PROGRAM COLLATING SEQUENCE: fails because of a data loss in a cast, due
-        #   to lack of specific handling of LOW/HIGH-VALUE for NATIONAL alphabets
-        #   (see typeck.c:cb_validate_collating)
       - name: Adjust testsuite for Debug target
         if: ${{ matrix.target == 'Debug' }}
         shell: C:\shells\msys2bash.cmd {0}
         run: |
           sed -i '/run_extensions/{N;/System routine CBL_GC_HOSTED/{N;s/at_xfail=no/at_xfail=yes/;}}' tests/testsuite
-          sed -i '/syn_definition/{N;/PROGRAM COLLATING SEQUENCE/{N;s/at_xfail=no/at_xfail=yes/;}}' tests/testsuite
 
       - name: Run testsuite
         run: |

cobc/ChangeLog

@@ -177,6 +177,25 @@
 	  also forces -E, -foneline-deps, -MT=copybooks, disables errors on
 	  missing copybooks and removes output on stdout
 
+2024-02-26  David Declerck <[email protected]>
+
+	BUG #948: comparison with HIGH-VALUE in presence of collating sequences
+	* tree.h: add low_value and high_value fields to hold the low
+	  and high values used by the program collating sequence
+	* tree.c: initialize the low_value and high_value fields
+	  to reasonable default values
+	* typeck.c: replace cob_refer_ascii and cob_refer_ebcdic by
+	  ebcdic_to_ascii and ascii_to_ebcdic; add load_collating_table
+	  to load the tables; modify cb_validate_collating to call
+	  load_collating_table and set low_value and high_value
+	  fields modify validate_alphabet to use the new tables
+	* cobc.h: export the new symbols defined in typeck.c
+	* codegen.c: replace hard-coded 0 and 255 / 0xff contants with
+	  the low_value and high_value fields where appropriate; move
+	  the cob_all_low and cob_all_high fields from global to local;
+	  adjust the output_collating_tables function to use the tables and
+	  functions defined in typeck.c; set the new module field low_value
+
 2024-02-19  Boris Eng <[email protected]>
 
 	* parser.y (screen_value_clause): replaced basic literals by literals

cobc/cobc.c

@@ -4230,6 +4230,10 @@ process_command_line (const int argc, char **argv)
 			cb_flag_alt_ebcdic ? "alternate" : "default");
 	}
 
+	if (cob_load_collation (cb_ebcdic_table, ebcdic_to_ascii, ascii_to_ebcdic) < 0) {
+		cobc_err_exit (_("invalid parameter: %s"), "-febcdic-table");
+	}
+
 	/* Exit on missing options */
 #ifdef COB_INTERNAL_XREF
 	if (cb_listing_xref && !cb_listing_outputfile) {

cobc/cobc.h

@@ -624,6 +624,11 @@ extern int		yyparse (void);
 /* typeck.c */
 extern size_t		suppress_warn;	/* no warnings for internal generated stuff */
 
+extern cob_u8_t ebcdic_to_ascii[256];
+extern cob_u8_t ascii_to_ebcdic[256];
+
+void load_collating_tables (void);
+
 /* error.c */
 #define CB_MSG_STYLE_GCC	0
 #define CB_MSG_STYLE_MSC	1U

cobc/codegen.c

@@ -2515,23 +2515,23 @@ static void
 output_low_value (void)
 {
 	if (gen_figurative & CB_NEED_LOW) {
-		output ("static cob_field cob_all_low\t= ");
-		output ("{1, ");
-		output ("(cob_u8_ptr)\"\\0\", ");
-		output ("&cob_all_attr};");
-		output_newline ();
+		output_local ("static cob_field cob_all_low\t= ");
+		output_local ("{1, ");
+		output_local ("(cob_u8_ptr)\"\\x%02x\", ", current_prog->low_value);
+		output_local ("&cob_all_attr};");
+		output_local ("\n");
 	}
 }
 
 static void
 output_high_value (void)
 {
 	if (gen_figurative & CB_NEED_HIGH) {
-		output ("static cob_field cob_all_high\t= ");
-		output ("{1, ");
-		output ("(cob_u8_ptr)\"\\xff\", ");
-		output ("&cob_all_attr};");
-		output_newline ();
+		output_local ("static cob_field cob_all_high\t= ");
+		output_local ("{1, ");
+		output_local ("(cob_u8_ptr)\"\\x%02x\", ", current_prog->high_value);
+		output_local ("&cob_all_attr};");
+		output_local ("\n");
 	}
 }
 
@@ -2615,8 +2615,6 @@ output_literals_figuratives_and_constants (void)
 
 	if (gen_figurative) {
 		output_newline ();
-		output_low_value ();
-		output_high_value ();
 		output_quote ();
 		output_space ();
 		output_zero ();
@@ -2654,18 +2652,6 @@ output_colseq_table_field (const char * field_name, const char * table_name)
 static void
 output_collating_tables (void)
 {
-	cob_u8_t ebcdic_to_ascii[256];
-	cob_u8_t ascii_to_ebcdic[256];
-
-	/* Load the collating tables if needed */
-	if (gen_ascii_ebcdic || gen_ebcdic_ascii) {
-		if (cob_load_collation (cb_ebcdic_table,
-					gen_ebcdic_ascii ? ebcdic_to_ascii : NULL,
-					gen_ascii_ebcdic ? ascii_to_ebcdic : NULL) < 0) {
-			cobc_err_exit (_("invalid parameter: %s"), "-febcdic-table");
-		}
-	}
-
 	if (gen_native) {
 		output_storage ("\n/* NATIVE table */\n");
 		output_colseq_table ("cob_native", NULL);
@@ -4277,9 +4263,9 @@ output_funcall_typed (struct cb_funcall *p, const char type)
 		} else if (p->argv[1] == cb_zero) {
 			output (") - '0')");
 		} else if (p->argv[1] == cb_low) {
-			output ("))");
+			output (") - %d)", current_prog->low_value);
 		} else if (p->argv[1] == cb_high) {
-			output (") - 255)");
+			output (") - %d)", current_prog->high_value);
 		} else if (CB_LITERAL_P (p->argv[1])) {
 			output_char (") - ", CB_LITERAL (p->argv[1])->data[0], ")");
 		} else {
@@ -5064,10 +5050,10 @@ output_initialize_to_value (struct cb_field *f, cb_tree x,
 			output_figurative (x, f, ' ', init_occurs);
 			return;
 		} else if (value == cb_low) {
-			output_figurative (x, f, 0, init_occurs);
+			output_figurative (x, f, current_prog->low_value, init_occurs);
 			return;
 		} else if (value == cb_high) {
-			output_figurative (x, f, 255, init_occurs);
+			output_figurative (x, f, current_prog->high_value, init_occurs);
 			return;
 		} else if (value == cb_quote) {
 			if (cb_flag_apostrophe) {
@@ -10810,9 +10796,9 @@ output_class_name_definition (struct cb_class_name *p)
 			} else if (x == cb_null) {
 				vals[0] = 1;
 			} else if (x == cb_low) {
-				vals[0] = 1;
+				vals[current_prog->low_value] = 1;
 			} else if (x == cb_high) {
-				vals[255] = 1;
+				vals[current_prog->high_value] = 1;
 			} else {
 				size = CB_LITERAL (x)->size;
 				data = CB_LITERAL (x)->data;
@@ -14157,6 +14143,12 @@ codegen_internal (struct cb_program *prog, const int subsequent_call)
 			output_local ("\n");
 		}
 	}
+
+	/* Low and high values */
+	if (gen_figurative) {
+		output_low_value ();
+		output_high_value ();
+	}
 }
 
 void

cobc/scanner.l

@@ -1347,10 +1347,6 @@ static cob_u8_t
 scan_ebcdic_char (int c)
 {
 	char buff[10];	       /* Arbitrary limit, mostly for error-reporting */
-#ifndef COB_EBCDIC_MACHINE
-	static  cob_u8_t ebcdic_to_ascii[256] ;
-	static  int ebcdic_to_ascii_initialized = 0 ;
-#endif
 	unsigned int j = 0;
 	do {
 		buff[j++] = (char)c;
@@ -1367,20 +1363,7 @@ scan_ebcdic_char (int c)
 #ifdef COB_EBCDIC_MACHINE
 	return (cob_u8_t) c;
 #else
-	if (!ebcdic_to_ascii_initialized ) {
-		if (cob_load_collation (cb_ebcdic_table, ebcdic_to_ascii, NULL) < 0) {
-			cb_error (_("invalid parameter: %s"), "-febcdic-table");
-			ebcdic_to_ascii_initialized = -1;
-		} else {
-			ebcdic_to_ascii_initialized = 1;
-		}
-	}
-
-	if (ebcdic_to_ascii_initialized > 0) {
-		return ebcdic_to_ascii[c];
-	} else {
-		return '?';
-	}
+	return ebcdic_to_ascii[c];
 #endif
 }
 

cobc/tree.c

@@ -2193,6 +2193,10 @@ cb_build_program (struct cb_program *last_program, const int nest_level)
 	p->decimal_point = '.';
 	p->currency_symbol = '$';
 	p->numeric_separator = ',';
+	p->low_value = 0;
+	p->high_value = 255;
+	p->low_value_n = 0;
+	p->high_value_n = 65535;
 	if (cb_call_extfh) {
 		p->extfh = cobc_parse_strdup (cb_call_extfh);
 	}

cobc/tree.h

@@ -1916,6 +1916,10 @@ struct cb_program {
 	unsigned char	decimal_point;			/* '.' or ',' */
 	unsigned char	currency_symbol;		/* '$' or user-specified */
 	unsigned char	numeric_separator;		/* ',' or '.' */
+	cob_u8_t	low_value;			/* Low-value for this program */
+	cob_u8_t	high_value;			/* High-value for this program */
+	cob_u16_t	low_value_n;			/* National Low-value */
+	cob_u16_t	high_value_n;			/* National High-value  */
 	enum cob_module_type	prog_type;			/* Program type (program = 0, function = 1) */
 	cb_tree			entry_convention;	/* ENTRY convention / PROCEDURE convention */
 	struct literal_list	*decimal_constants;