libman using enum mant instead of #defines for macros.

Clean-ups, better documentation in man_hash.c.
Added extra space for "." in man_hash.c (unused for the time being).

5 files changed, 106 insertions, 77 deletions

diff --git a/libman.h b/libman.h
index 2efafe78..a4d5cbb5 100644
--- a/libman.h
+++ b/libman.h
@@ -63,7 +63,7 @@ enum	merr {
 	WERRMAX
 };
 
-#define	MACRO_PROT_ARGS	  struct man *m, int tok, int line, \
+#define	MACRO_PROT_ARGS	  struct man *m, enum mant tok, int line, \
 			  int ppos, int *pos, char *buf
 
 struct	man_macro {
@@ -89,15 +89,15 @@ __BEGIN_DECLS
 		  man_err((m), (n)->line, (n)->pos, 0, (t))
 
 int		  man_word_alloc(struct man *, int, int, const char *);
-int		  man_block_alloc(struct man *, int, int, int);
-int		  man_head_alloc(struct man *, int, int, int);
-int		  man_body_alloc(struct man *, int, int, int);
-int		  man_elem_alloc(struct man *, int, int, int);
+int		  man_block_alloc(struct man *, int, int, enum mant);
+int		  man_head_alloc(struct man *, int, int, enum mant);
+int		  man_body_alloc(struct man *, int, int, enum mant);
+int		  man_elem_alloc(struct man *, int, int, enum mant);
 void		  man_node_free(struct man_node *);
 void		  man_node_freelist(struct man_node *);
 void		  man_node_unlink(struct man *, struct man_node *);
 void		  man_hash_init(void);
-int		  man_hash_find(const char *);
+enum	mant	  man_hash_find(const char *);
 int		  man_macroend(struct man *);
 int		  man_args(struct man *, int, int *, char *, char **);
 #define	ARGS_ERROR	(-1)
diff --git a/man.c b/man.c
index 88035875..986efec6 100644
--- a/man.c
+++ b/man.c
@@ -67,7 +67,7 @@ const	char *const __man_macronames[MAN_MAX] = {
 const	char * const *man_macronames = __man_macronames;
 
 static	struct man_node	*man_node_alloc(int, int, 
-				enum man_type, int);
+				enum man_type, enum mant);
 static	int		 man_node_append(struct man *, 
 				struct man_node *);
 static	int		 man_ptext(struct man *, int, char *);
@@ -242,7 +242,7 @@ man_node_append(struct man *man, struct man_node *p)
 
 
 static struct man_node *
-man_node_alloc(int line, int pos, enum man_type type, int tok)
+man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 {
 	struct man_node *p;
 
@@ -256,7 +256,7 @@ man_node_alloc(int line, int pos, enum man_type type, int tok)
 
 
 int
-man_elem_alloc(struct man *m, int line, int pos, int tok)
+man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 {
 	struct man_node *p;
 
@@ -269,7 +269,7 @@ man_elem_alloc(struct man *m, int line, int pos, int tok)
 
 
 int
-man_head_alloc(struct man *m, int line, int pos, int tok)
+man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 {
 	struct man_node *p;
 
@@ -282,7 +282,7 @@ man_head_alloc(struct man *m, int line, int pos, int tok)
 
 
 int
-man_body_alloc(struct man *m, int line, int pos, int tok)
+man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 {
 	struct man_node *p;
 
@@ -295,7 +295,7 @@ man_body_alloc(struct man *m, int line, int pos, int tok)
 
 
 int
-man_block_alloc(struct man *m, int line, int pos, int tok)
+man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 {
 	struct man_node *p;
 
@@ -314,7 +314,7 @@ pstring(struct man *m, int line, int pos,
 	struct man_node	*n;
 	size_t		 sv;
 
-	n = man_node_alloc(line, pos, MAN_TEXT, -1);
+	n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
 	n->string = mandoc_malloc(len + 1);
 	sv = strlcpy(n->string, p, len + 1);
 
@@ -469,7 +469,8 @@ macrowarn(struct man *m, int ln, const char *buf)
 int
 man_pmacro(struct man *m, int ln, char *buf)
 {
-	int		 i, j, c, ppos, fl;
+	int		 i, j, ppos, fl;
+	enum mant	 tok;
 	char		 mac[5];
 	struct man_node	*n;
 
@@ -519,7 +520,7 @@ man_pmacro(struct man *m, int ln, char *buf)
 		return(1);
 	}
 	
-	if (MAN_MAX == (c = man_hash_find(mac))) {
+	if (MAN_MAX == (tok = man_hash_find(mac))) {
 		if ( ! macrowarn(m, ln, mac))
 			goto err;
 		return(1);
@@ -542,7 +543,7 @@ man_pmacro(struct man *m, int ln, char *buf)
 	 * macros---they don't print text---so we let those slip by.
 	 */
 
-	if ( ! (MAN_NSCOPED & man_macros[c].flags) &&
+	if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 			m->flags & MAN_ELINE) {
 		assert(MAN_TEXT != m->last->type);
 
@@ -574,9 +575,9 @@ man_pmacro(struct man *m, int ln, char *buf)
 
 	/* Begin recursive parse sequence. */
 
-	assert(man_macros[c].fp);
+	assert(man_macros[tok].fp);
 
-	if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))
+	if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 		goto err;
 
 out:
diff --git a/man.h b/man.h
index 524a5dd0..872fbf48 100644
--- a/man.h
+++ b/man.h
@@ -19,42 +19,44 @@
 
 #include <time.h>
 
-#define	MAN_br	 	 0
-#define	MAN_TH		 1
-#define	MAN_SH		 2
-#define	MAN_SS		 3
-#define	MAN_TP		 4
-#define	MAN_LP		 5
-#define	MAN_PP		 6
-#define	MAN_P		 7
-#define	MAN_IP		 8
-#define	MAN_HP		 9
-#define	MAN_SM		 10
-#define	MAN_SB		 11
-#define	MAN_BI		 12
-#define	MAN_IB		 13
-#define	MAN_BR		 14
-#define	MAN_RB		 15
-#define	MAN_R		 16
-#define	MAN_B		 17
-#define	MAN_I		 18
-#define	MAN_IR		 19
-#define	MAN_RI		 20
-#define	MAN_na		 21
-#define	MAN_i		 22
-#define	MAN_sp		 23
-#define	MAN_nf		 24
-#define	MAN_fi		 25
-#define	MAN_r		 26
-#define	MAN_RE		 27
-#define	MAN_RS		 28
-#define	MAN_DT		 29
-#define	MAN_UC		 30
-#define	MAN_PD		 31
-#define	MAN_Sp		 32
-#define	MAN_Vb		 33
-#define	MAN_Ve		 34
-#define	MAN_MAX	 	 35
+enum	mant {
+	MAN_br = 0,
+	MAN_TH,
+	MAN_SH,
+	MAN_SS,
+	MAN_TP,
+	MAN_LP,
+	MAN_PP,
+	MAN_P,
+	MAN_IP,
+	MAN_HP,
+	MAN_SM,
+	MAN_SB,
+	MAN_BI,
+	MAN_IB,
+	MAN_BR,
+	MAN_RB,
+	MAN_R,
+	MAN_B,
+	MAN_I,
+	MAN_IR,
+	MAN_RI,
+	MAN_na,
+	MAN_i,
+	MAN_sp,
+	MAN_nf,
+	MAN_fi,
+	MAN_r,
+	MAN_RE,
+	MAN_RS,
+	MAN_DT,
+	MAN_UC,
+	MAN_PD,
+	MAN_Sp,
+	MAN_Vb,
+	MAN_Ve,
+	MAN_MAX,
+};
 
 enum	man_type {
 	MAN_TEXT,
@@ -81,7 +83,7 @@ struct	man_node {
 	int		 nchild;
 	int		 line;
 	int		 pos;
-	int		 tok;
+	enum mant	 tok;
 	int		 flags;
 #define	MAN_VALID	(1 << 0)
 #define	MAN_ACTED	(1 << 1)
diff --git a/man_hash.c b/man_hash.c
index 7ca0da44..44b5c9f1 100644
--- a/man_hash.c
+++ b/man_hash.c
@@ -21,13 +21,33 @@
 #include <sys/types.h>
 
 #include <assert.h>
+#include <ctype.h>
 #include <limits.h>
 #include <stdlib.h>
 #include <string.h>
 
 #include "libman.h"
 
-static	u_char		table[26 * 6];
+#define	HASH_DEPTH	 6
+
+#define	HASH_ROW(x) do { \
+		if ('.' == (x)) \
+			(x) = 26; \
+		else if (isupper((u_char)(x))) \
+			(x) -= 65; \
+		else \
+			(x) -= 97; \
+		(x) *= HASH_DEPTH; \
+	} while (/* CONSTCOND */ 0)
+
+/*
+ * Lookup table is indexed first by lower-case first letter (plus one
+ * for the period, which is stored in the last row), then by lower or
+ * uppercase second letter.  Buckets correspond to the index of the
+ * macro (the integer value of the enum stored as a char to save a bit
+ * of space).
+ */
+static	u_char		 table[27 * HASH_DEPTH];
 
 /*
  * XXX - this hash has global scope, so if intended for use as a library
@@ -40,39 +60,44 @@ man_hash_init(void)
 
 	memset(table, UCHAR_MAX, sizeof(table));
 
+	assert(/* CONSTCOND */ MAN_MAX < UCHAR_MAX);
+
 	for (i = 0; i < MAN_MAX; i++) {
 		x = man_macronames[i][0];
-		assert((x >= 65 && x <= 90) ||
-				(x >= 97 && x <= 122));
 
-		x -= (x <= 90) ? 65 : 97;
-		x *= 6;
+		assert(isalpha((u_char)x) || '.' == x);
 
-		for (j = 0; j < 6; j++)
+		HASH_ROW(x);
+
+		for (j = 0; j < HASH_DEPTH; j++)
 			if (UCHAR_MAX == table[x + j]) {
 				table[x + j] = (u_char)i;
 				break;
 			}
-		assert(j < 6);
+
+		assert(j < HASH_DEPTH);
 	}
 }
 
-int
+
+enum mant
 man_hash_find(const char *tmp)
 {
-	int		 x, i, tok;
+	int		 x, y, i;
+	enum mant	 tok;
 
-	if (0 == (x = tmp[0]))
+	if ('\0' == (x = tmp[0]))
 		return(MAN_MAX);
-	if ( ! ((x >= 65 && x <= 90) || (x >= 97 && x <= 122)))
+	if ( ! (isalpha((u_char)x) || '.' == x))
 		return(MAN_MAX);
 
-	x -= (x <= 90) ? 65 : 97;
-	x *= 6;
+	HASH_ROW(x);
 
-	for (i = 0; i < 6; i++) {
-		if (UCHAR_MAX == (tok = table[x + i]))
+	for (i = 0; i < HASH_DEPTH; i++) {
+		if (UCHAR_MAX == (y = table[x + i]))
 			return(MAN_MAX);
+
+		tok = (enum mant)y;
 		if (0 == strcmp(tmp, man_macronames[tok]))
 			return(tok);
 	}
diff --git a/man_macro.c b/man_macro.c
index 1a7ed50c..3137e375 100644
--- a/man_macro.c
+++ b/man_macro.c
@@ -33,10 +33,11 @@ static	int		 in_line_eoln(MACRO_PROT_ARGS);
 static	int		 blk_imp(MACRO_PROT_ARGS);
 static	int		 blk_close(MACRO_PROT_ARGS);
 
-static	int		 rew_scope(enum man_type, struct man *, int);
-static	int 		 rew_dohalt(int, enum man_type, 
+static	int		 rew_scope(enum man_type, 
+				struct man *, enum mant);
+static	int 		 rew_dohalt(enum mant, enum man_type, 
 				const struct man_node *);
-static	int		 rew_block(int, enum man_type, 
+static	int		 rew_block(enum mant, enum man_type, 
 				const struct man_node *);
 
 const	struct man_macro __man_macros[MAN_MAX] = {
@@ -104,7 +105,7 @@ man_unscope(struct man *m, const struct man_node *n)
 
 
 static int
-rew_block(int ntok, enum man_type type, const struct man_node *n)
+rew_block(enum mant ntok, enum man_type type, const struct man_node *n)
 {
 
 	if (MAN_BLOCK == type && ntok == n->parent->tok && 
@@ -120,7 +121,7 @@ rew_block(int ntok, enum man_type type, const struct man_node *n)
  * sections and subsections).
  */
 static int 
-rew_dohalt(int tok, enum man_type type, const struct man_node *n)
+rew_dohalt(enum mant tok, enum man_type type, const struct man_node *n)
 {
 	int		 c;
 
@@ -175,7 +176,7 @@ rew_dohalt(int tok, enum man_type type, const struct man_node *n)
  * scopes.  When a scope is closed, it must be validated and actioned.
  */
 static int
-rew_scope(enum man_type type, struct man *m, int tok)
+rew_scope(enum man_type type, struct man *m, enum mant tok)
 {
 	struct man_node	*n;
 	int		 c;
@@ -205,7 +206,7 @@ rew_scope(enum man_type type, struct man *m, int tok)
 int
 blk_close(MACRO_PROT_ARGS)
 {
-	int 		 	 ntok;
+	enum mant	 	 ntok;
 	const struct man_node	*nn;
 
 	switch (tok) {