Add support for collating sequences on indexed files

cobc/ChangeLog

@@ -1,4 +1,15 @@
 
+2024-01-25  David Declerck <[email protected]>
+
+	FR #459: support COLLATING SEQUENCE clause on SELECT / INDEXED files
+	* codegen.c (output_file_initialization): output the indexed
+	  file collating sequence (was already present in the AST)
+	* parser.y (collating_sequence_clause): remove the
+	  CB_PENDING warning on file collating sequence
+	* flag.def, tree.h, cobc.c, parser.y: add and handle a new
+	  -fdefault-file-colseq flag to specify the default collating
+	  sequence to use for files without a collating sequence clause
+
 2023-10-17  David Declerck <[email protected]>
 
 	BUG #923: generated modules init/clear unused decimal constants

cobc/cobc.c

@@ -90,22 +90,23 @@ enum compile_level {
 	CB_LEVEL_EXECUTABLE	= 7
 };
 
-#define	CB_FLAG_GETOPT_STACK_SIZE       1
-#define	CB_FLAG_GETOPT_IF_CUTOFF        2
-#define	CB_FLAG_GETOPT_SIGN             3
-#define	CB_FLAG_GETOPT_FOLD_COPY        4
-#define	CB_FLAG_GETOPT_FOLD_CALL        5
-#define	CB_FLAG_GETOPT_TTITLE           6
-#define	CB_FLAG_GETOPT_MAX_ERRORS       7
-#define	CB_FLAG_GETOPT_DUMP             8
-#define	CB_FLAG_GETOPT_CALLFH           9
-#define	CB_FLAG_GETOPT_INTRINSICS      10
-#define	CB_FLAG_GETOPT_EC              11
-#define	CB_FLAG_GETOPT_NO_EC           12
-#define	CB_FLAG_GETOPT_NO_DUMP         13
-#define	CB_FLAG_GETOPT_EBCDIC_TABLE    14
-#define	CB_FLAG_GETOPT_DEFAULT_COLSEQ  15
-#define	CB_FLAG_MEMORY_CHECK           16
+#define	CB_FLAG_GETOPT_STACK_SIZE            1
+#define	CB_FLAG_GETOPT_IF_CUTOFF             2
+#define	CB_FLAG_GETOPT_SIGN                  3
+#define	CB_FLAG_GETOPT_FOLD_COPY             4
+#define	CB_FLAG_GETOPT_FOLD_CALL             5
+#define	CB_FLAG_GETOPT_TTITLE                6
+#define	CB_FLAG_GETOPT_MAX_ERRORS            7
+#define	CB_FLAG_GETOPT_DUMP                  8
+#define	CB_FLAG_GETOPT_CALLFH                9
+#define	CB_FLAG_GETOPT_INTRINSICS           10
+#define	CB_FLAG_GETOPT_EC                   11
+#define	CB_FLAG_GETOPT_NO_EC                12
+#define	CB_FLAG_GETOPT_NO_DUMP              13
+#define	CB_FLAG_GETOPT_EBCDIC_TABLE         14
+#define	CB_FLAG_GETOPT_DEFAULT_COLSEQ       15
+#define	CB_FLAG_GETOPT_DEFAULT_FILE_COLSEQ  16
+#define	CB_FLAG_MEMORY_CHECK                17
 
 
 /* Info display limits */
@@ -3809,6 +3810,13 @@ process_command_line (const int argc, char **argv)
 			}
 			break;
 
+		case CB_FLAG_GETOPT_DEFAULT_FILE_COLSEQ: /* 16 */
+			/* -fdefault-file-colseq=<ASCII/EBCDIC/NATIVE> */
+			if (cb_deciph_default_file_colseq_name (cob_optarg)) {
+				cobc_err_exit (COBC_INV_PAR, "-fdefault-file-colseq");
+			}
+			break;
+
 		case CB_FLAG_GETOPT_FOLD_COPY: /* 4 */
 			/* -ffold-copy=<UPPER/LOWER> : COPY fold case */
 			if (!cb_strcasecmp (cob_optarg, "UPPER")) {
@@ -3888,7 +3896,7 @@ process_command_line (const int argc, char **argv)
 			}
 			break;
 
-		case CB_FLAG_MEMORY_CHECK: /* 16 */
+		case CB_FLAG_MEMORY_CHECK: /* 17 */
 			/* -fmemory-check=<scope> :  */
 			if (!cob_optarg) {
 				cb_flag_memory_check = CB_MEMCHK_ALL;

cobc/codegen.c

@@ -9434,6 +9434,19 @@ output_file_initialization (struct cb_file *f)
 		output_line ("lingptr->lin_bot = 0;");
 	}
 
+	if (f->organization == COB_ORG_INDEXED) {
+		cb_tree col = f->collating_sequence;
+		output_prefix ();
+		output ("%s%s->sort_collating = ", CB_PREFIX_FILE, f->cname);
+		if ((col != NULL) && CB_REFERENCE_P (col)) {
+			output_param (cb_ref(col), -1);
+			output (";");
+		} else {
+			output ("NULL;");
+		}
+		output_newline ();
+	}
+
 	if (f->organization != COB_ORG_SORT
 	 && f->code_set) {
 		/* pass CODE-SET as collation */
@@ -9457,7 +9470,7 @@ output_file_initialization (struct cb_file *f)
 			break;
 		}
 
-		output_line ("%s%s->sort_collating = %s;", CB_PREFIX_FILE, f->cname, alph_write);
+		output_line ("%s%s->code_set_write = %s;", CB_PREFIX_FILE, f->cname, alph_write);
 		output_line ("%s%s->code_set_read = %s;", CB_PREFIX_FILE, f->cname, alph_read);
 		if (f->code_set_items) {
 			const unsigned int items = cb_list_length (CB_TREE (f->code_set_items));

cobc/flag.def

@@ -103,6 +103,10 @@ CB_FLAG_NQ (1, "default-colseq", CB_FLAG_GETOPT_DEFAULT_COLSEQ,
 	_("  -fdefault-colseq=[ASCII|EBCDIC|NATIVE]\tdefine default collating sequence\n"
 	  "                        * default: NATIVE"))
 
+CB_FLAG_NQ (1, "default-file-colseq", CB_FLAG_GETOPT_DEFAULT_FILE_COLSEQ,
+	_("  -fdefault-file-colseq=[ASCII|EBCDIC|NATIVE]\tdefine default file collating sequence\n"
+	  "                        * default: NATIVE"))
+
 /* Binary flags */
 
 /* Flags with suppressed help */

cobc/parser.y

@@ -342,23 +342,35 @@ enum cb_colseq {
 	CB_COLSEQ_ASCII,
 	CB_COLSEQ_EBCDIC,
 };
+
 enum cb_colseq cb_default_colseq = CB_COLSEQ_NATIVE;
+enum cb_colseq cb_default_file_colseq = CB_COLSEQ_NATIVE;
 
 /* Decipher character conversion table names */
-int cb_deciph_default_colseq_name (const char * const name)
+int cb_deciph_colseq_name (const char * const name, enum cb_colseq *colseq)
 {
 	if (!cb_strcasecmp (name, "ASCII")) {
-		cb_default_colseq = CB_COLSEQ_ASCII;
+		*colseq = CB_COLSEQ_ASCII;
 	} else if (!cb_strcasecmp (name, "EBCDIC")) {
-		cb_default_colseq = CB_COLSEQ_EBCDIC;
+		*colseq = CB_COLSEQ_EBCDIC;
 	} else if (!cb_strcasecmp (name, "NATIVE")) {
-		cb_default_colseq = CB_COLSEQ_NATIVE;
+		*colseq = CB_COLSEQ_NATIVE;
 	} else {
 		return 1;
 	}
 	return 0;
 }
 
+int cb_deciph_default_colseq_name (const char * const name)
+{
+	return cb_deciph_colseq_name (name, &cb_default_colseq);
+}
+
+int cb_deciph_default_file_colseq_name (const char * const name)
+{
+	return cb_deciph_colseq_name (name, &cb_default_file_colseq);
+}
+
 static cb_tree
 build_colseq_tree (const char *alphabet_name,
 		      int alphabet_type,
@@ -901,23 +913,34 @@ check_relaxed_syntax (const cob_flags_t lev)
 }
 
 static void
-setup_default_collation (struct cb_program *program) {
-	switch (cb_default_colseq) {
+prepare_default_collation (enum cb_colseq colseq) {
+	switch (colseq) {
 #ifdef COB_EBCDIC_MACHINE
 	case CB_COLSEQ_ASCII:
 #else
 	case CB_COLSEQ_EBCDIC:
 #endif
-		alphanumeric_collation = build_colseq (cb_default_colseq);
+		alphanumeric_collation = build_colseq (colseq);
 		break;
 	default:
 		alphanumeric_collation = NULL;
 	}
 	national_collation = NULL; /* TODO: default national collation */
+}
+
+static void
+setup_default_collation (struct cb_program *program) {
+	prepare_default_collation (cb_default_colseq);
 	program->collating_sequence = alphanumeric_collation;
 	program->collating_sequence_n = national_collation;
 }
 
+static void
+setup_default_file_collation (struct cb_file *file) {
+	prepare_default_collation (cb_default_file_colseq);
+	file->collating_sequence = alphanumeric_collation;
+}
+
 static void
 program_init_without_program_id (void)
 {
@@ -5365,6 +5388,7 @@ file_control_entry:
 
 	}
 	key_type = NO_KEY;
+        setup_default_file_collation (current_file);
   }
   _select_clauses_or_error
   {
@@ -5752,7 +5776,6 @@ collating_sequence_clause:
 	check_repeated ("COLLATING", SYN_CLAUSE_3, &check_duplicate);
 	current_file->collating_sequence = alphanumeric_collation;
 	current_file->collating_sequence_n = national_collation;
-	CB_PENDING ("FILE COLLATING SEQUENCE");
   }
 ;
 

cobc/tree.h

@@ -2343,6 +2343,7 @@ extern cb_tree		cb_debug_sub_3;
 extern cb_tree		cb_debug_contents;
 
 extern int		cb_deciph_default_colseq_name (const char *const);
+extern int		cb_deciph_default_file_colseq_name (const char *const);
 
 extern struct cb_program	*cb_build_program (struct cb_program *,
 						   const int);

libcob/ChangeLog

@@ -1,4 +1,14 @@
 
+2024-01-25  David Declerck <[email protected]>
+
+	FR #459: support COLLATING SEQUENCE clause on SELECT / INDEXED files
+	* fileio.c (bdb_setkeycol, bdb_bt_compare, indexed_open, ...):
+	  take the file collating sequence into account when comparing keys
+	* common.h, common.c, coblocal.h: make common_cmps available locally
+	* common.h, common.c, fileio.c: split the sort_collating field to
+	  distinguish actual collating sequence and code-set for writing
+	  (new field for the latter: code_set_write, to match code_set_read)
+
 2023-12-14  David Declerck <[email protected]>
 
 	* common.c (cob_terminate_routines, cob_call_with_exception_check):

libcob/coblocal.h

@@ -569,6 +569,9 @@ cob_max_int (const int x, const int y)
 	return y;
 }
 
+COB_HIDDEN int		common_cmps	(const unsigned char *, const unsigned char *,
+					 const size_t, const unsigned char *);
+
 #undef	COB_HIDDEN
 
 #endif	/* COB_LOCAL_H */

libcob/common.c

@@ -1822,7 +1822,7 @@ common_cmpc (const unsigned char *p, const unsigned int c,
 
 /* compare up to 'size' characters in 's1' to 's2'
    using collation 'col' */
-static int
+int
 common_cmps (const unsigned char *s1, const unsigned char *s2,
 	     const size_t size, const unsigned char *col)
 {

libcob/common.h

@@ -1395,7 +1395,7 @@ typedef struct __cob_file {
 	cob_file_key		*keys;			/* ISAM/RANDOM/SORT keys */
 	void			*file;			/* File specific pointer */
 	void			*linorkeyptr;		/* LINAGE or SPLIT KEY */
-	const unsigned char	*sort_collating;	/* SORT collating / CODE-SET (used for RE-/WRITE) */
+	const unsigned char	*sort_collating;	/* SORT collating */
 	void			*extfh_ptr;		/* For EXTFH usage */
 	size_t			record_min;		/* Record min size */
 	size_t			record_max;		/* Record max size */
@@ -1432,6 +1432,7 @@ typedef struct __cob_file {
 	char				*nxt_filename;	/* Next position in org_filename */
 	unsigned int		flag_is_concat:1;	/* SEQUENTIAL concatenated file names */
 
+	const unsigned char* code_set_write;	/* CODE-SET conversion for RE-/WRITE) */
 	const unsigned char* code_set_read;	/* CODE-SET conversion for READs */
 	size_t			nconvert_fields;	/* Number of logical fields to convert */
 	cob_field	*convert_field;		/* logical fields to convert for CODE-SET */