Fuller seperation into mdocml/libmdocml.

4 files changed, 240 insertions, 73 deletions

diff --git a/Makefile b/Makefile
index 47bbb565..af8856f0 100644
--- a/Makefile
+++ b/Makefile
@@ -2,13 +2,13 @@ CFLAGS += -W -Wall -g
 
 LINTFLAGS += -c -e -f -u
 
-CLEAN = mdocml mdocml.o mdocml.ln libmdocml.o libmdocml.ln mdocml.tgz
+CLEAN = mdocml mdocml.o mdocml.ln libmdocml.o libmdocml.ln mdocml.tgz llib-lmdocml.ln
 
 INSTALL = Makefile libmdocml.h mdocml.c libmdocml.c mdocml.1
 
 all: mdocml
 
-lint: mdocml.ln libmdocml.ln
+lint: llib-lmdocml.ln
 
 dist: mdocml.tgz
 
@@ -24,6 +24,9 @@ mdocml.tgz: $(INSTALL)
 	( cd .dist/ && tar zcf ../mdocml.tgz mdocml/ )
 	rm -rf .dist/
 
+llib-lmdocml.ln: mdocml.ln libmdocml.ln
+	$(LINT) $(LINTFLAGS) -Cmdocml mdocml.ln libmdocml.ln
+
 mdocml.ln: mdocml.c 
 
 mdocml.o: mdocml.c 
diff --git a/libmdocml.c b/libmdocml.c
index 07513496..d851a962 100644
--- a/libmdocml.c
+++ b/libmdocml.c
@@ -1,4 +1,4 @@
-/* $Id: libmdocml.c,v 1.2 2008/11/22 18:34:06 kristaps Exp $ */
+/* $Id: libmdocml.c,v 1.3 2008/11/22 20:15:34 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -28,12 +28,48 @@
 
 #define	BUFFER_LINE	 BUFSIZ
 
-typedef int (*md_line)	(struct md_mbuf *, const struct md_rbuf *,
+struct	md_rbuf {
+	int		 fd;
+	char		*name;
+	char		*buf;
+	size_t		 bufsz;
+	size_t		 line;
+};
+
+struct	md_mbuf {
+	int		 fd;
+	char		*name;
+	char		*buf;
+	size_t		 bufsz;
+	size_t		 pos;
+};
+
+typedef int (*md_line)	(const struct md_args *, struct md_mbuf *, 
+				const struct md_rbuf *,
+				const char *, size_t);
+typedef int (*md_init)	(const struct md_args *, struct md_mbuf *);
+typedef int (*md_exit)	(const struct md_args *, struct md_mbuf *);
+
+static int		 md_line_dummy(const struct md_args *,
+				struct md_mbuf *, 
+				const struct md_rbuf *, 
 				const char *, size_t);
 
-static int		 md_line_dummy(struct md_mbuf *, 
+static int		 md_line_html4_strict(const struct md_args *,
+				struct md_mbuf *, 
 				const struct md_rbuf *,
 				const char *, size_t);
+static int		 md_init_html4_strict(const struct md_args *,
+				struct md_mbuf *);
+static int		 md_exit_html4_strict(const struct md_args *,
+				struct md_mbuf *);
+
+static int		 md_run_enter(const struct md_args *, 
+				struct md_mbuf *, struct md_rbuf *);
+static int		 md_run_leave(const struct md_args *, 
+				struct md_mbuf *, 
+				struct md_rbuf *, int);
+
 static ssize_t		 md_buf_fill(struct md_rbuf *);
 static int		 md_buf_flush(struct md_mbuf *);
 static int		 md_buf_putchar(struct md_mbuf *, char);
@@ -41,7 +77,7 @@ static int		 md_buf_puts(struct md_mbuf *,
 				const char *, size_t);
 
 
-ssize_t
+static ssize_t
 md_buf_fill(struct md_rbuf *in)
 {
 	ssize_t		 ssz;
@@ -58,7 +94,7 @@ md_buf_fill(struct md_rbuf *in)
 }
 
 
-int
+static int
 md_buf_flush(struct md_mbuf *buf)
 {
 	ssize_t		 sz;
@@ -85,14 +121,14 @@ md_buf_flush(struct md_mbuf *buf)
 }
 
 
-int
+static int
 md_buf_putchar(struct md_mbuf *buf, char c)
 {
 	return(md_buf_puts(buf, &c, 1));
 }
 
 
-int
+static int
 md_buf_puts(struct md_mbuf *buf, const char *p, size_t sz)
 {
 	size_t		 ssz;
@@ -121,73 +157,177 @@ md_buf_puts(struct md_mbuf *buf, const char *p, size_t sz)
 }
 
 
-int
-md_run(enum md_type type, struct md_mbuf *out, struct md_rbuf *in)
+static int
+md_run_leave(const struct md_args *args, 
+		struct md_mbuf *mbuf, struct md_rbuf *rbuf, int c)
 {
-	ssize_t		 sz, i;
-	char		 line[BUFFER_LINE];
-	size_t		 pos;
-	md_line		 func;
+	assert(args);
+	assert(mbuf);
+	assert(rbuf);
+
+	/* Run exiters. */
+	switch (args->type) {
+	case (MD_HTML4_STRICT):
+		if ( ! md_exit_html4_strict(args, mbuf))
+			return(-1);
+		break;
+	case (MD_DUMMY):
+		break;
+	default:
+		abort();
+	}
 
-	assert(in);
-	assert(out); 
+	/* Make final flush of buffer. */
+	if ( ! md_buf_flush(mbuf))
+		return(-1);
+
+	return(c);
+}
 
-	out->pos = 0;
-	in->line = 1;
 
-	assert(MD_DUMMY == type);
-	func = md_line_dummy;
+static int
+md_run_enter(const struct md_args *args, 
+		struct md_mbuf *mbuf, struct md_rbuf *rbuf)
+{
+	ssize_t		 sz, i;
+	char		 line[BUFFER_LINE];
+	size_t		 pos;
+	md_line		 fp;
+
+	assert(args);
+	assert(mbuf);
+	assert(rbuf); 
+
+	/* Function ptrs to line-parsers. */
+	switch (args->type) {
+	case (MD_HTML4_STRICT):
+		fp = md_line_html4_strict;
+		break;
+	case (MD_DUMMY):
+		fp = md_line_dummy;
+		break;
+	default:
+		abort();
+	}
 
 	/* LINTED */
 	for (pos = 0; ; ) {
-		if (-1 == (sz = md_buf_fill(in)))
-			return(1);
+		if (-1 == (sz = md_buf_fill(rbuf)))
+			return(-1);
 		else if (0 == sz)
 			break;
 
 		for (i = 0; i < sz; i++) {
-			if ('\n' == in->buf[i]) {
-				if ((*func)(out, in, line, pos))
-					return(1);
-				in->line++;
+			if ('\n' == rbuf->buf[i]) {
+				if ( ! (*fp)(args, mbuf, rbuf, line, pos))
+					return(-1);
+				rbuf->line++;
 				pos = 0;
 				continue;
 			}
 
 			if (pos < BUFFER_LINE) {
 				/* LINTED */
-				line[pos++] = in->buf[i];
+				line[pos++] = rbuf->buf[i];
 				continue;
 			}
 
 			warnx("%s: line %zu too long",
-					in->name, in->line);
-			return(1);
+					rbuf->name, rbuf->line);
+			return(-1);
 		}
 	}
 
-	if (0 != pos && (*func)(out, in, line, pos))
-		return(1);
+	if (0 != pos && ! (*fp)(args, mbuf, rbuf, line, pos))
+		return(-1);
 
-	return(md_buf_flush(out) ? 0 : 1);
+	return(md_run_leave(args, mbuf, rbuf, 0));
+}
+
+
+int
+md_run(const struct md_args *args,
+		const struct md_buf *out, const struct md_buf *in)
+{
+	struct md_mbuf	 mbuf;
+	struct md_rbuf	 rbuf;
+
+	assert(args);
+	assert(in);
+	assert(out); 
+
+	(void)memcpy(&mbuf, out, sizeof(struct md_buf));
+	(void)memcpy(&rbuf, in, sizeof(struct md_buf));
+
+	mbuf.pos = 0;
+	rbuf.line = 1;
+
+	/* Run initialisers. */
+	switch (args->type) {
+	case (MD_HTML4_STRICT):
+		if ( ! md_init_html4_strict(args, &mbuf))
+			return(-1);
+		break;
+	case (MD_DUMMY):
+		break;
+	default:
+		abort();
+	}
+
+	/* Go into mainline. */
+	return(md_run_enter(args, &mbuf, &rbuf));
 }
 
 
 static int
-md_line_dummy(struct md_mbuf *out, const struct md_rbuf *in,
-		const char *buf, size_t sz)
+md_line_dummy(const struct md_args *args, struct md_mbuf *out, 
+		const struct md_rbuf *in, const char *buf, size_t sz)
 {
 
 	assert(buf);
 	assert(out);
 	assert(in);
+	assert(args);
 
 	if ( ! md_buf_puts(out, buf, sz))
-		return(1);
+		return(0);
 	if ( ! md_buf_putchar(out, '\n'))
-		return(1);
+		return(0);
 
-	return(0);
+	return(1);
+}
+
+
+static int
+md_exit_html4_strict(const struct md_args *args, struct md_mbuf *p) 
+{
+
+	assert(p);
+	assert(args);
+	return(1);
 }
 
 
+static int
+md_init_html4_strict(const struct md_args *args, struct md_mbuf *p) 
+{
+
+	assert(p);
+	assert(args);
+	return(1);
+}
+
+
+static int
+md_line_html4_strict(const struct md_args *args, struct md_mbuf *out, 
+		const struct md_rbuf *in, const char *buf, size_t sz)
+{
+
+	assert(args);
+	assert(buf);
+	assert(out);
+	assert(in);
+	(void)sz;
+
+	return(1);
+}
diff --git a/libmdocml.h b/libmdocml.h
index 63b8c3f4..60ea7f8f 100644
--- a/libmdocml.h
+++ b/libmdocml.h
@@ -1,4 +1,4 @@
-/* $Id: libmdocml.h,v 1.2 2008/11/22 18:34:06 kristaps Exp $ */
+/* $Id: libmdocml.h,v 1.3 2008/11/22 20:15:34 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -21,29 +21,43 @@
 
 #include <sys/types.h>
 
-struct	md_rbuf {
-	int		 fd;
-	const char	*name;
-	char		*buf;
-	size_t		 bufsz;
-	size_t		 line;
+struct 	md_params_dummy {
+	int		 dummy;
 };
 
-struct	md_mbuf {
-	int		 fd;
-	const char	*name;
-	char		*buf;
-	size_t		 bufsz;
-	size_t		 pos;
+struct 	md_params_html4_strict {
+	int		 dummy;
+};
+
+union	md_params {
+	struct md_params_dummy dummy;
+	struct md_params_html4_strict html4_strict;
 };
 
 enum	md_type {
-	MD_DUMMY
+	MD_DUMMY,		/* Dummy type echoes input. */
+	MD_HTML4_STRICT		/* HTML4.01-strict. */
+};
+
+struct	md_args {
+	union md_params	 params;/* Parameters for parser. */
+	enum md_type	 type;	/* Type of parser. */
+};
+
+struct	md_buf {
+	int		 fd;	/* Open file descriptor. */
+	char		*name;	/* Name of file/socket/whatever. */
+	char		*buf;	/* Buffer for storing data. */
+	size_t		 bufsz;	/* Size of buf. */
 };
 
 __BEGIN_DECLS
 
-int	md_run(enum md_type, struct md_mbuf *, struct md_rbuf *);
+/* Run the parser over prepared input and output buffers.  Returns -1 on
+ * failure and 0 on success.
+ */
+int	md_run(const struct md_args *,
+		const struct md_buf *, const struct md_buf *);
 
 __END_DECLS
 
diff --git a/mdocml.c b/mdocml.c
index 7f914211..2f3e95d8 100644
--- a/mdocml.c
+++ b/mdocml.c
@@ -1,4 +1,4 @@
-/* $Id: mdocml.c,v 1.4 2008/11/22 18:34:06 kristaps Exp $ */
+/* $Id: mdocml.c,v 1.5 2008/11/22 20:15:34 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -34,18 +34,21 @@
 #define	BUFFER_OUT_DEF	BUFSIZ
 
 static void		 usage(void);
-static int		 begin_io(const char *, const char *);
-static int		 leave_io(const struct md_mbuf *, 
-				const struct md_rbuf *, int);
-static int		 begin_bufs(struct md_mbuf *, struct md_rbuf *);
-static int		 leave_bufs(const struct md_mbuf *, 
-				const struct md_rbuf *, int);
+static int		 begin_io(const struct md_args *, 
+				char *, char *);
+static int		 leave_io(const struct md_buf *, 
+				const struct md_buf *, int);
+static int		 begin_bufs(const struct md_args *,
+				struct md_buf *, struct md_buf *);
+static int		 leave_bufs(const struct md_buf *, 
+				const struct md_buf *, int);
 
 int
 main(int argc, char *argv[])
 {
 	int		 c;
 	char		*out, *in;
+	struct md_args	 args;
 
 	extern char	*optarg;
 	extern int	 optind;
@@ -68,13 +71,15 @@ main(int argc, char *argv[])
 	if (1 == argc)
 		in = *argv++;
 
-	return(begin_io(out ? out : "-", in ? in : "-"));
+	args.type = MD_DUMMY;
+
+	return(begin_io(&args, out ? out : "-", in ? in : "-"));
 }
 
 
 static int
-leave_io(const struct md_mbuf *out, 
-		const struct md_rbuf *in, int c)
+leave_io(const struct md_buf *out, 
+		const struct md_buf *in, int c)
 {
 	assert(out);
 	assert(in);
@@ -96,19 +101,20 @@ leave_io(const struct md_mbuf *out,
 
 
 static int
-begin_io(const char *out, const char *in)
+begin_io(const struct md_args *args, char *out, char *in)
 {
-	struct md_rbuf	 fi;
-	struct md_mbuf	 fo;
+	struct md_buf	 fi;
+	struct md_buf	 fo;
 
 #define	FI_FL	O_RDONLY
 #define	FO_FL	O_WRONLY|O_CREAT|O_TRUNC
 
+	assert(args);
 	assert(out);
 	assert(in);
 
-	bzero(&fi, sizeof(struct md_rbuf));
-	bzero(&fo, sizeof(struct md_mbuf));
+	bzero(&fi, sizeof(struct md_buf));
+	bzero(&fo, sizeof(struct md_buf));
 
 	fi.fd = STDIN_FILENO;
 	fo.fd = STDOUT_FILENO;
@@ -128,13 +134,13 @@ begin_io(const char *out, const char *in)
 			return(leave_io(&fo, &fi, 1));
 		}
 
-	return(leave_io(&fo, &fi, begin_bufs(&fo, &fi)));
+	return(leave_io(&fo, &fi, begin_bufs(args, &fo, &fi)));
 }
 
 
 static int
-leave_bufs(const struct md_mbuf *out, 
-		const struct md_rbuf *in, int c)
+leave_bufs(const struct md_buf *out, 
+		const struct md_buf *in, int c)
 {
 	assert(out);
 	assert(in);
@@ -147,10 +153,13 @@ leave_bufs(const struct md_mbuf *out,
 
 
 static int
-begin_bufs(struct md_mbuf *out, struct md_rbuf *in)
+begin_bufs(const struct md_args *args, 
+		struct md_buf *out, struct md_buf *in)
 {
 	struct stat	 stin, stout;
+	int		 c;
 
+	assert(args);
 	assert(in);
 	assert(out);
 
@@ -173,7 +182,8 @@ begin_bufs(struct md_mbuf *out, struct md_rbuf *in)
 		return(leave_bufs(out, in, 1));
 	}
 
-	return(leave_bufs(out, in, md_run(MD_DUMMY, out, in)));
+	c = md_run(args, out, in);
+	return(leave_bufs(out, in, -1 == c ? 1 : 0));
 }