More work on the genunicode program

FossilOrigin-Name: edfc2aebb5a66067934b9a05bb88b9d7e3aab4d2070d710e34c4339a0d303a5c

2 files changed, 206 insertions, 2 deletions

diff --git a/src/lib/unicode.c b/src/lib/unicode.c
index a913b6d..b5c6471 100644
--- a/src/lib/unicode.c
+++ b/src/lib/unicode.c
@@ -19,3 +19,34 @@
 
 #include "aplwc_internal.h"
 #include "unicode.h"
+
+static int
+is(uint32_t c, enum unicode_property p)
+{
+	int n;
+	struct unicode_range *r;
+
+	UNUSED_PARAMETER(c);
+
+	unicode_get(p, &n, &r);
+
+	return 0;
+}
+
+int
+is_id_start(uint32_t c)
+{
+	return is(c, ID_Start);
+}
+
+int
+is_id_continue(uint32_t c)
+{
+	return is(c, ID_Continue);
+}
+
+int
+is_pattern_syntax(uint32_t c)
+{
+	return is(c, Pattern_Syntax);
+}
diff --git a/src/tools/genunicode.c b/src/tools/genunicode.c
index 0329ace..0d5d428 100644
--- a/src/tools/genunicode.c
+++ b/src/tools/genunicode.c
@@ -21,18 +21,42 @@
 #include <string.h>
 #include <stdlib.h>
 #include <assert.h>
+#include <setjmp.h>
 
 static char *files[] = {
 	"DerivedCoreProperties.txt",
 	"PropList.txt"
 };
 
+static char *properties[] = {
+	"ID_Start",
+	"ID_Continue",
+	"Pattern_Syntax"
+};
+
+struct line {
+	int from;
+	int to;
+	int valid;
+	char property[64];
+};
+
+static int peek(FILE *);
+static int parse_line(FILE *, struct line *);
+static int parse_int(FILE *);
+static void parse_dots(FILE *);
+static void parse_sep(FILE *);
+
+static jmp_buf env;
+
 int
 main(int argc, char *argv[])
 {
 	FILE *f;
 	char *dir, *buf;
-	size_t dir_size, tmp_size, max_size, i, nfiles;
+	size_t dir_size, tmp_size, max_size, i, j, nfiles;
+	struct line line;
+	int current, n;
 
 	assert(argc == 2);
 	dir = argv[1];
@@ -54,15 +78,164 @@ main(int argc, char *argv[])
 	assert(buf != NULL);
 	memcpy(buf, dir, dir_size);
 	buf[dir_size] = '/';
+	current = -1;
+	n = 0;
+
+	printf("enum unicode_property {");
+	for(i = 0; i < sizeof(properties)/sizeof(properties[0]); i++)
+		printf("%s%s", (i == 0) ? "\n\t" : ",\n\t", properties[i]);
+	printf("\n};\n\n");
+
+	printf("struct unicode_range {\n\tint from;\n\tint to;\n};\n\n");
 
 	for(i = 0; i < sizeof(files)/sizeof(files[0]); i++){
 		memcpy(buf+dir_size+1, files[i], strlen(files[i])+1);
 		f = fopen(buf, "r");
 		assert(f != NULL);
+
+		while(parse_line(f, &line)){
+			if(!line.valid)
+				continue;
+
+			if((current == -1) || (strcmp(line.property, properties[current]) != 0)){
+				if(current != -1){
+					printf("\n};\n");
+					printf("#define %s_count %d\n\n", properties[current], n);
+				}
+
+				current = -1;
+				for(j = 0; j < sizeof(properties)/sizeof(properties[0]); j++){
+					if(strcmp(line.property, properties[j]) == 0){
+						current = j;
+						n = 0;
+						printf("static struct unicode_range %s_ranges[] = {", properties[j]);
+						break;
+					}
+				}
+			}
+
+			if(current != -1){
+				printf("%s{%d, %d}", (n == 0) ? "\n\t" : ",\n\t", line.from, line.to);
+				n++;
+			}
+		}
 		fclose(f);
+	}
 
-		printf("/* Unicode file %s read from %s */\n", files[i], buf);
+	printf("static void\n");
+	printf("unicode_get(enum unicode_property p, int *n, struct unicode_range **r)\n{\n");
+	printf("\tswitch(p){\n");
+	for(i = 0; i < sizeof(properties)/sizeof(properties[0]); i++){
+		printf("\tcase %s:\n", properties[i]);
+		printf("\t\t*n = %s_count;\n", properties[i]);
+		printf("\t\t*r = %s_ranges;\n", properties[i]);
+		printf("\t\tbreak;\n");
 	}
+	printf("\t}\n}\n\n");
 
+	free(buf);
 	return 0;
 }
+
+static int
+peek(FILE *f)
+{
+	int c;
+
+	c = fgetc(f);
+	ungetc(c, f);
+	return c;
+}
+
+static int
+parse_line(FILE *f, struct line *l)
+{
+	int c, i;
+
+	if(setjmp(env)){
+		l->valid = 0;
+		do
+			c = fgetc(f);
+		while(c != EOF && c != '\n');
+		return c != EOF;
+	}else{
+		l->valid = 1;
+	}
+
+	l->from = parse_int(f);
+	if(peek(f) == '.'){
+		parse_dots(f);
+		l->to = parse_int(f);
+	}else{
+		l->to = l->from;
+	}
+	parse_sep(f);
+
+	for(i = 0; ; i++){
+		c = fgetc(f);
+		l->property[i] = c;
+		if(c == ' ' || c == EOF){
+			if(i == 0)
+				longjmp(env, 1);
+			break;
+		}
+	}
+	l->property[i] = 0;
+
+	for(;;){
+		c = fgetc(f);
+		if(c == '\n' || c == EOF)
+			break;
+	}
+	return 1;
+}
+
+static int
+parse_int(FILE *f)
+{
+	int n, m, c;
+
+	n = 0;
+
+	for(;;){
+		c = peek(f);
+		if((c >= '0') && (c <= '9'))
+			m = c - '0';
+		else if((c >= 'A') && (c <= 'F'))
+			m = 10 + (c - 'A');
+		else if(c == '.' || c == ' ')
+			return n;
+		else
+			longjmp(env, 1);
+		fgetc(f);
+		n <<= 4;
+		n += m;
+	}
+}
+
+static void
+parse_dots(FILE *f)
+{
+	int i;
+
+	for(i = 0; i < 2; i++){
+		if(peek(f) != '.')
+			longjmp(env, 1);
+		fgetc(f);
+	}
+}
+
+static void
+parse_sep(FILE *f)
+{
+	while(peek(f) == ' ')
+		fgetc(f);
+
+	if(peek(f) == ';')
+		fgetc(f);
+	else
+		longjmp(env, 1);
+
+	while(peek(f) == ' ')
+		fgetc(f);
+}