bake ragel.tmpl into re2r executable

this allows a systemwide install without having to look around
for the template file.
also add option -t which allows the user to specify a custom template
file.

Author: rofl0r

Makefile

@@ -7,12 +7,20 @@ OBJS = lexer.o y.tab.o sblist.o sblist_pop.o hsearch.o tokens.o
 
 -include config.mak
 
+HOSTCC ?= $(CC)
+
 all: re2r re2r_test
 
 lexer.c: y.tab.h
-main.c: y.tab.h
+main.c: y.tab.h template.h
 tokens.c: y.tab.h
 
+template.h: file2hdr ragel.tmpl
+	./file2hdr ragel.tmpl > template.h
+
+file2hdr: file2hdr.c
+	$(HOSTCC) $(HOSTCFLAGS) $< -o $@
+
 y.tab.h: y.tab.c
 
 y.tab.c: ere.y

README

@@ -13,6 +13,7 @@ EXAMPLES
 	ipv4 [0-9]+[.][0-9]+[.][0-9]+[.][0-9]+
 
 OPTIONS
+	-t template.txt: use template.txt instead of builtin (ragel.tmpl)
 	-o outfile: write output to outfile instead of stdout
 
 BUGS

file2hdr.c

@@ -0,0 +1,113 @@
+/* (C) 2019 rofl0r */
+/* released into the public domain, or at your choice 0BSD or WTFPL */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <ctype.h>
+#include <unistd.h>
+
+static int usage() {
+	printf( "file2hdr - converts a file into a C string\n"
+		"the output is written to stdout.\n"
+		"usage: file2hdr filename\n"
+		);
+	return 1;
+}
+
+static int isprintable(int ch) {
+#define PRINTABLE " ;,:.^-+=*/%|&[](){}<>#_"
+	return ch > 0 && (isdigit(ch) || isalpha(ch) || strchr(PRINTABLE, ch));
+}
+
+static int escape_char(int ch, char buf[5], int *dirty) {
+	int len = 0;
+	*dirty = 0;
+	buf[len++] = '\\';
+	switch(ch) {
+		case '\a': /* 0x07 */
+			buf[len++] = 'a'; break;
+		case '\b': /* 0x08 */
+			buf[len++] = 'b'; break;
+		case '\t': /* 0x09 */
+			buf[len++] = 't'; break;
+		case '\n': /* 0x0a */
+			buf[len++] = 'n'; break;
+		case '\v': /* 0x0b */
+			buf[len++] = 'v'; break;
+		case '\f': /* 0x0c */
+			buf[len++] = 'f'; break;
+		case '\r': /* 0x0d */
+			buf[len++] = 'r'; break;
+		case '\"': /* 0x22 */
+			buf[len++] = '\"'; break;
+		case '\'': /* 0x27 */
+			buf[len++] = '\''; break;
+		case '\?': /* 0x3f */
+			buf[len++] = '\?'; break;
+		case '\\': /* 0x5c */
+			buf[len++] = '\\'; break;
+		default:
+			if(isprintable(ch)) buf[0] = ch;
+			else {
+				if (ch <= 077) *dirty = 1;
+				len += sprintf(buf+len, "%o", ch);
+			}
+	}
+	buf[len] = 0;
+	return len;
+}
+
+int main(int argc, char** argv) {
+	int f_arg = 1;
+	unsigned cpl = 77;
+	if(argc-1 < f_arg) return usage();
+	FILE *f = fopen(argv[f_arg], "r");
+	if(!f) { perror("fopen"); return 1; }
+	{
+		off_t end, start = ftello(f);
+		size_t insize, outsize;
+		fseeko(f, 0, SEEK_END);
+		end = ftello(f);
+		fseeko(f, start, SEEK_SET);
+		insize = end - start;
+		outsize = insize;
+
+		char *p = strrchr(argv[f_arg], '/');
+		if(!p) p = argv[f_arg];
+		else p++;
+		char *fn = strdup(p);
+		p = fn;
+		while(*p) { if(*p == '.') *p = '_'; ++p; }
+		printf( "const struct { unsigned clen, ulen;\n"
+			"const unsigned char data[]; } %s = { %zu, %zu,\n",
+			fn, outsize, insize);
+		free(fn);
+	}
+
+	int ch, dirty;
+	unsigned cnt = 0;
+	while((ch = fgetc(f)) != EOF) {
+		if(cnt == 0) { printf("\""); dirty = 0; }
+		char buf[5];
+		int cdirty;
+		cnt += escape_char(ch, buf, &cdirty);
+		// dirty and cdirty: ok
+		// dirty and not cdirty: ok if buf[0] not in '0'..'7'
+		if(dirty && !cdirty && buf[0] >= '0' && buf[0] <= '7') {
+			printf("\"\"");
+			cnt += 2;
+		}
+		dirty = cdirty;
+		printf("%s", buf);
+		if(cnt >= cpl) {
+			printf("\"\n");
+			cnt = 0;
+		}
+	}
+	if(cnt) printf("\"\n");
+	printf("};\n");
+	fclose(f);
+	return 0;
+}
+

main.c

@@ -9,6 +9,7 @@
 #include "sblist.h"
 #include "hsearch.h"
 #include "tokens.h"
+#include "template.h"
 
 extern int yyerror(const char*);
 
@@ -94,8 +95,7 @@ static void dump_parents(char *buf, sblist *tokens, const char* org_regex) {
 	fprintf(yyout, "%s", replace(buf, "%PARENTARRAY%", parbuf));
 }
 
-static void dump_ragel_parser(const char *machinename, const char* org_regex, const char* org_regex_end, int *maxgroups) {
-	FILE *f = fopen("ragel.tmpl", "r");
+static void dump_ragel_parser(FILE *f, const char *machinename, const char* org_regex, const char* org_regex_end, int *maxgroups) {
 	char buf[4096];
 	int groups, cgroup = 0;
 	sblist *tokens = lex_and_transform(org_regex, org_regex_end);
@@ -142,7 +142,6 @@ static void dump_ragel_parser(const char *machinename, const char* org_regex, co
 			fprintf(yyout, "%s", buf);
 		}
 	}
-	fclose(f);
 	sblist_free(group_order);
 	sblist_free(tokens);
 }
@@ -160,7 +159,9 @@ static int usage() {
 	        "EXAMPLES\n"
 	        "\tipv4 [0-9]+[.][0-9]+[.][0-9]+[.][0-9]+\n\n"
 	        "OPTIONS\n"
-	        "\t-o outfile: write output to outfile instead of stdout\n\n"
+	        "\t-t template.txt: use template.txt instead of builtin (ragel.tmpl)\n"
+	        "\t-o outfile: write output to outfile instead of stdout\n"
+		"\n"
 	        "BUGS\n"
 	        "\tPOSIX collation, equivalence, and character classes support is not implemented\n"
 	        "\tyou can replace character classes like [[:digit:]] with [0-9] using some sort\n"
@@ -174,20 +175,23 @@ int main(int argc, char**argv) {
 	extern int yydebug;
 	yydebug = 1;
 #endif
-	char buf[4096];
+	char buf[4096], *template = 0;
 	size_t lineno = 0;
 	yyin = stdin;
 	yyout = stdout;
 	int maxgroups = 0, err = 0, c;
 	struct htab *remap = htab_create(32);
-	while((c = getopt(argc, argv, "o:")) != EOF) switch(c) {
+	while((c = getopt(argc, argv, "t:o:")) != EOF) switch(c) {
 	case 'o':
 		yyout = fopen(optarg, "w");
 		if(!yyout) {
 			perror("open");
 			return 1;
 		}
 		break;
+	case 't':
+		template = optarg;
+		break;
 	default:
 		return usage();
 	}
@@ -211,9 +215,13 @@ int main(int argc, char**argv) {
 		}
 		lex_init(p, pe, LEXFLAG_SILENT);
 		if(yyparse() == 0) {
-			htab_insert(remap, strdup(p), HTV_P(strdup(buf)));
 			/* syntax check OK */
-			dump_ragel_parser(buf, p, pe, &maxgroups);
+			htab_insert(remap, strdup(p), HTV_P(strdup(buf)));
+			FILE *f;
+			if(template) f = fopen(template, "r");
+			else f = fmemopen((void*)ragel_tmpl.data, ragel_tmpl.ulen, "r");
+			dump_ragel_parser(f, buf, p, pe, &maxgroups);
+			fclose(f);
 		} else {
 			++err;
 			size_t errpos = lex_errpos();