check-bolt: handle references to early-drafts too.

Particularly for the onchain.md draft.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Makefile

@@ -8,7 +8,7 @@ PROTOCC:=protoc-c
 CCANDIR := ccan
 
 # Where we keep the BOLT RFCs
-BOLTDIR := ../lightning-rfc/bolts/
+BOLTDIR := ../lightning-rfc/
 
 # Bitcoin uses DER for signatures (Add BIP68 & HAS_CSV if it's supported)
 BITCOIN_FEATURES :=				\

check-bolt.c

@@ -12,6 +12,11 @@
 
 static bool verbose = false;
 
+struct bolt_file {
+	const char *prefix;
+	const char *contents;
+};
+
 /* Turn any whitespace into a single space. */
 static char *canonicalize(char *str)
 {
@@ -36,71 +41,91 @@ static char *canonicalize(char *str)
 	return str;
 }
 
-static char **get_bolt_files(const char *dir)
+static void get_files(const char *dir, const char *subdir,
+		      struct bolt_file **files)
 {
+	char *path = path_join(NULL, dir, subdir);
+	DIR *d = opendir(path);
+	size_t n = tal_count(*files);
 	struct dirent *e;
-	char **bolts = tal_arr(NULL, char *, 0);
+
-	DIR *d = opendir(dir);
 	if (!d)
-		err(1, "Opening BOLT dir %s", dir);
+		err(1, "Opening BOLT dir %s", path);
 
 	while ((e = readdir(d)) != NULL) {
-		char *endp;
+		int preflen;
-		unsigned long l;
-
-		/* Must start with the bold number. */
-		l = strtoul(e->d_name, &endp, 10);
-		if (endp == e->d_name)
-			continue;
 
 		/* Must end in .md */
 		if (!strends(e->d_name, ".md"))
 			continue;
 
-		if (l >= tal_count(bolts))
+		/* Prefix is anything up to - */
-			tal_resizez(&bolts, l+1);
+		preflen = strspn(e->d_name,
+				 "0123456789"
+				 "abcdefghijklmnopqrstuvwxyz"
+				 "ABCDEFGHIJKLMNOPQRSTUVWXYZ");
+		if (!preflen)
+			continue;
+		if (preflen + strlen(".md") != strlen(e->d_name)
+		    && e->d_name[preflen] != '-')
+			continue;
 
 		if (verbose)
-			printf("Found bolt %s: #%lu\n", e->d_name, l);
+			printf("Found bolt %.*s\n", preflen, e->d_name);
 
-		bolts[l] = canonicalize(grab_file(NULL,
+		tal_resize(files, n+1);
-						  path_join(NULL, dir,
+		(*files)[n].prefix = tal_strndup(*files,
+						 e->d_name, preflen);
+		(*files)[n].contents
+			= canonicalize(grab_file(*files,
+						 path_join(path, path,
 							   e->d_name)));
+		n++;
 	}
+}
+
+static struct bolt_file *get_bolt_files(const char *dir)
+{
+	struct bolt_file *bolts = tal_arr(NULL, struct bolt_file, 0);
+
+	get_files(dir, "bolts", &bolts);
+	get_files(dir, "early-drafts", &bolts);
 	return bolts;
 }
 
-static char *find_bolt_ref(char *p, size_t *len, size_t *bolt)
+static char *find_bolt_ref(char **p, size_t *len)
 {
 	for (;;) {
-		char *end;
+		char *bolt, *end;
+		size_t preflen;
 
 		/* BOLT #X: */
-		p = strstr(p, "BOLT");
+		*p = strstr(*p, "BOLT");
-		if (!p)
+		if (!*p)
 			return NULL;
-		p += 4;
+		*p += 4;
-		while (cisspace(*p))
+		while (cisspace(**p))
-			p++;
+			(*p)++;
-		if (*p != '#')
+		if (**p != '#')
 			continue;
-		p++;
+		(*p)++;
-		*bolt = strtoul(p, &end, 10);
+
-		if (!*bolt || p == end)
+		preflen = strcspn(*p, " :");
-			continue;
+		bolt = tal_strndup(NULL, *p, preflen);
-		p = end;
+
-		while (cisspace(*p))
+		(*p) += preflen;
-			p++;
+		while (cisspace(**p))
-		if (*p != ':')
+			(*p)++;
+		if (**p != ':')
 			continue;
-		p++;
+		(*p)++;
 
-		end = strstr(p, "*/");
+		end = strstr(*p, "*/");
 		if (!end)
-			*len = strlen(p);
+			*len = strlen(*p);
 		else
-			*len = end - p;
+			*len = end - *p;
-		return p;
+		return bolt;
 	}
 }
 
@@ -108,7 +133,7 @@ static char *code_to_regex(const char *code, size_t len, bool escape)
 {
 	char *pattern = tal_arr(NULL, char, len*2 + 1), *p;
 	size_t i;
-	bool after_nl = false;
+	bool after_nl = true;
 
 	/* We swallow '*' if first in line: block comments */
 	p = pattern;
@@ -153,30 +178,36 @@ static char *code_to_regex(const char *code, size_t len, bool escape)
 	return canonicalize(pattern);
 }
 
-static void fail(const char *filename, const char *raw, const char *pos,
+/* Moves *pos to start of line. */
-		 size_t len, const char *bolt)
+static unsigned linenum(const char *raw, const char **pos)
 {
 	unsigned line = 0; /* Out-by-one below */
-	const char *l = raw;
+	const char *l = raw, *point = *pos;
 
-	while (l < pos) {
+	while (l < point) {
+		*pos = l;
 		l = strchr(l, '\n');
 		line++;
 		if (!l)
-			l = pos + strlen(pos);
+			break;
-		else
 		l++;
 	}
+	return line;
+}
 
-	if (bolt) {
+static void fail_mismatch(const char *filename,
+			  const char *raw, const char *pos,
+			  size_t len, struct bolt_file *bolt)
+{
+	unsigned line = linenum(raw, &pos);
 	char *try;
 
-		fprintf(stderr, "%s:%u:%.*s\n", filename, line,
+	fprintf(stderr, "%s:%u:mismatch:%.*s\n",
-			(int)(l - pos), pos);
+		filename, line, (int)strcspn(pos, "\n"), pos);
 	/* Try to find longest match, as a hint. */
-		try = code_to_regex(pos, len, false);
+	try = code_to_regex(pos + strcspn(pos, "\n"), len, false);
 	while (strlen(try)) {
-			const char *p = strstr(bolt, try);
+		const char *p = strstr(bolt->contents, try);
 		if (p) {
 			fprintf(stderr, "Closest match: %s...[%.20s]\n",
 				try, p + strlen(try));
@@ -184,15 +215,44 @@ static void fail(const char *filename, const char *raw, const char *pos,
 		}
 		try[strlen(try)-1] = '\0';
 	}
-	} else {
+	exit(1);
-		fprintf(stderr, "%s:%u:Unknown bolt\n", filename, line);
 }
+
+static void fail_nobolt(const char *filename,
+			const char *raw, const char *pos,
+			const char *bolt_prefix)
+{
+	unsigned line = linenum(raw, &pos);
+
+	fprintf(stderr, "%s:%u:unknown bolt %s\n",
+		filename, line, bolt_prefix);
 	exit(1);
 }
 	
+static struct bolt_file *find_bolt(const char *bolt_prefix,
+				   struct bolt_file *bolts)
+{
+	size_t i, n = tal_count(bolts);
+	size_t boltnum;
+
+	for (i = 0; i < n; i++)
+		if (streq(bolts[i].prefix, bolt_prefix))
+			return bolts+i;
+
+	/* Now search for numerical match. */
+	boltnum = atoi(bolt_prefix);
+	if (boltnum) {
+		for (i = 0; i < n; i++)
+			if (atoi(bolts[i].prefix) == boltnum)
+				return bolts+i;
+	}
+
+	return NULL;
+}
+
 int main(int argc, char *argv[])
 {
-	char **bolts;
+	struct bolt_file *bolts;
 	int i;
 
 	err_set_progname(argv[0]);
@@ -211,8 +271,8 @@ int main(int argc, char *argv[])
 	bolts = get_bolt_files(argv[1]);
 
 	for (i = 2; i < argc; i++) {
-		char *f = grab_file(NULL, argv[i]), *p;
+		char *f = grab_file(NULL, argv[i]), *p, *bolt;
-		size_t len, bolt;
+		size_t len;
 		if (!f)
 			err(1, "Loading %s", argv[i]);
 
@@ -220,16 +280,17 @@ int main(int argc, char *argv[])
 			printf("Checking %s...\n", argv[i]);
 
 		p = f;
-		while ((p = find_bolt_ref(p, &len, &bolt)) != NULL) {
+		while ((bolt = find_bolt_ref(&p, &len)) != NULL) {
 			char *pattern = code_to_regex(p, len, true);
-			if (bolt >= tal_count(bolts) || !bolts[bolt])
+			struct bolt_file *b = find_bolt(bolt, bolts);
-				fail(argv[i], f, p, len, NULL);
+			if (!b)
-			if (!tal_strreg(f, bolts[bolt], pattern, NULL))
+				fail_nobolt(argv[i], f, p, bolt);
-				fail(argv[i], f, p, len, bolts[bolt]);
+			if (!tal_strreg(f, b->contents, pattern, NULL))
+				fail_mismatch(argv[i], f, p, len, b);
 
 			if (verbose)
-				printf("  Found %.10s... in %zu\n",
+				printf("  Found %.10s... in %s\n",
-				       p, bolt);
+				       p, b->prefix);
 			p += len;
 		}
 		tal_free(f);