Convert hyphens to en dashes, warn if field unused

3 files changed, 83 insertions, 75 deletions

diff --git a/Makefile b/Makefile
index 92211b0..a84614a 100644
--- a/Makefile
+++ b/Makefile
@@ -1,3 +1,3 @@
 fref: fref.lex
 	lex fref.lex
-	cc -O2 -o fref lex.yy.c -lfl
+	cc -std=c89 -O2 -o fref lex.yy.c -lfl
diff --git a/example.t b/example.t
index dd84a75..26775e4 100644
--- a/example.t
+++ b/example.t
@@ -1,3 +1,5 @@
+.\" $ export LC_ALL=en_US.UTF-8
+.\" $ ./fref <example.t | troff -ms | dpost | ps2pdf - >example.pdf
 .SH
 References
 .XP
@@ -10,7 +12,7 @@ References
 %pp	110-123
 %ad	http://www.example.com/b1887.pdf
 .XP
-%au	de Courtenay, J. Baudouin
+%au	Baudouin de Courtenay, J.
 %au	Author, Fake Extra
 %da	1972
 %ti	The Difference between Phonetics and Psychophonetics
diff --git a/fref.lex b/fref.lex
index bc26dc7..91ea0e0 100644
--- a/fref.lex
+++ b/fref.lex
@@ -5,6 +5,8 @@
  * one (like troff itself). This allows for a larger number of fields.
  */
 
+%x ENTRY
+
 	#include <err.h>
 	#include <stdarg.h>
 
@@ -13,49 +15,62 @@
 	#define AND "och"
 	#define AVAIL "Tillgänglig: "
 	#define ED "red."
+	#define P "s."
 	#define PP "ss."
 	#define TR "Övers."
 	#define VOL "vol."
 
+	#define DO_ENTRIES						\
+		DO("ad",ad,e.ad)	/* internet address */		\
+		DO("bo",bo,e.bo)	/* book */			\
+		DO("ci",ci,e.ci)	/* city */			\
+		DO("da",da,e.da)	/* date (year) */		\
+		DO("ed",ed,e.ed)	/* editor */			\
+		DO("jo",jo,e.jo)	/* journal */			\
+		DO("la",la,e.la)	/* label */			\
+		DO("no",no,e.no)	/* TODO: issue number */	\
+		DO("pp",pp,e.pp)	/* pages */			\
+		DO("pu",pu,e.pu)	/* publisher */			\
+		DO("se",se,e.se)	/* TODO: series */		\
+		DO("ti",ti,e.ti)	/* title */			\
+		DO("tr",tr,e.tr)	/* translator */		\
+		DO("vo",vo,e.vo)	/* volume */			\
+		DO("xx",xx,e.xx)	/* extra information */		\
+	/* END */
+
 	void field(char *);
 	void entry(void);
 	void harvard(void);
 	int pf(char *, ...);
 
 	char *name;
+	int line;	/* line at which entry begins */
+	int lines = 0;	/* total number of lines */
 
 	struct entry {
-		int a; /* Number of authors. */
+	#define DO(s,n,m) char n[MAX];
+		DO_ENTRIES
+	#undef DO
+		int a;			/* number of authors */
 		char au[MAX][MAX];	/* author */
-		char ad[MAX];		/* internet address */
-		char bo[MAX];		/* book */
-		char ci[MAX];		/* city */
-		char da[MAX];		/* date (year) */
-		char ed[MAX];		/* editor */
-		char jo[MAX];		/* journal */
-		char la[MAX];		/* label */
-		char no[MAX];		/* TODO: issue number */
-		char pp[MAX];		/* pages */
-		char pu[MAX];		/* publisher */
-		char se[MAX];		/* TODO: series */
-		char ti[MAX];		/* title */
-		char tr[MAX];		/* translator */
-		char vo[MAX];		/* volume */
-		char xx[MAX];		/* extra information */
 	} e;
 
-%s ENTRY
 %%
-^%.*\n		field(yytext); BEGIN(ENTRY);
-<ENTRY>^[^%]	entry(); ECHO; BEGIN(0);
-<ENTRY><<EOF>>	entry(); yyterminate();
+\n		lines++;	ECHO;
+^%.*\n		line = ++lines;	field(yytext); BEGIN(ENTRY);
+<ENTRY>^%.*\n	lines++;	field(yytext);
+<ENTRY>^[^%]			entry(); ECHO; BEGIN(0);
+<ENTRY><<EOF>>	entry();	yyterminate();
 %%
 
 void
 main(int argc, char *argv[0])
 {
-	name = argv[0];
-	e.a = 0;
+	name = strrchr(argv[0], '/');
+	if(name) name++;
+	else name = argv[0];
+
+	memset(&e, 0, sizeof(e));
 	yylex();
 }
 
@@ -63,28 +78,15 @@ main(int argc, char *argv[0])
 void
 field(char *t)
 {
-
-#define AS(a,b) if(strncmp(t, a, 2) == 0 && (t += 3)){ strcpy(b, t); return; }
-
 	t++;
 	t[strcspn(t, "\n")] = 0;
 
+	/* Fill corresponding field. */
+#define AS(a,b) if(strncmp(t, a, 2) == 0 && (t += 3)){ strcpy(b, t); return; }
 	AS("au",e.au[e.a++]);
-	AS("ad",e.ad);
-	AS("bo",e.bo);
-	AS("ci",e.ci);
-	AS("da",e.da);
-	AS("ed",e.ed);
-	AS("jo",e.jo);
-	AS("la",e.la);
-	AS("no",e.no);
-	AS("pp",e.pp);
-	AS("pu",e.pu);
-	AS("se",e.se);
-	AS("ti",e.ti);
-	AS("tr",e.tr);
-	AS("vo",e.vo);
-	AS("xx",e.xx);
+#define DO(s,n,m) AS(s,m)
+	DO_ENTRIES
+#undef DO
 
 	fprintf(stderr, "%s: unrecognized field %%%.2s\n", name, t);
 	exit(1);
@@ -127,9 +129,18 @@ harvard()
 	}else
 		pf("\\fI%\\fP", e.ti);
 
-	/* Print volume, pages. */
+	/* Print volume. */
 	pf(", "VOL" %", e.vo);
-	pf(", "PP" %", e.pp);
+
+	/* Print pages, converting hyphens to en dashes. */
+	if(*e.pp){
+		printf(", %s ", strpbrk(e.pp, ",-") ? PP : P);
+		for(i = 0; i < strlen(e.pp); i++){
+			if(e.pp[i] == '-') putchar('\\');
+			putchar(e.pp[i]);
+		}
+		*e.pp = 0;
+	}
 	pf(".\n");
 
 	/* Print translator. */
@@ -144,32 +155,22 @@ harvard()
 	/* Print internet address. */
 	pf(AVAIL"%\n", e.ad);
 
-#define CL(a) *a = 0
-
+	/* Clear entries. */
 	e.a = 0;
-	CL(e.ad);
-	CL(e.bo);
-	CL(e.ci);
-	CL(e.da);
-	CL(e.ed);
-	CL(e.jo);
-	CL(e.la);
-	CL(e.no);
-	CL(e.pp);
-	CL(e.pu);
-	CL(e.se);
-	CL(e.ti);
-	CL(e.tr);
-	CL(e.vo);
-	CL(e.xx);
+#define DO(s,n,m)							\
+	if(*m) fprintf(stderr, "%s: unused field %%"s" at line %d\n",	\
+		name, line);					\
+	*m = 0;
+	DO_ENTRIES
+#undef DO
 }
 
-/* Print formatted text if given fields are non-empty. */
+/* Print formatted text if given fields are non-empty. Clear fields. */
 int
 pf(char *fmt, ...)
 {
-	char *buf, *p;
-	int n, sz;
+	char *buf, *p, **fs;
+	int i, n, sz;
 	va_list ap;
 
 	/* Count given fields. */
@@ -177,24 +178,29 @@ pf(char *fmt, ...)
 		if(*p == '%')
 			n++;
 
-	/* Allocate enough memory to fit the given fields. */
-	sz = strlen(fmt)+n*MAX+1;
-	if(!(buf = malloc(sz)))
+	/* Allocate memory. */
+	if(!(fs = malloc(n)))
 		err(1, "malloc");
 
+	/* Verify that no field is empty. */
 	va_start(ap, fmt);
-	for(; *fmt; fmt++)
+	for(i = 0; i < n; i++){
+		p = va_arg(ap, char *);
+		if(!p || !*p)
+			return 0;
+		fs[i] = p;
+	}
+	va_end(ap);
+
+	/* Print fields according to format. */
+	for(i = 0; *fmt; fmt++)
 		switch(*fmt){
 		case '%':
-			p = va_arg(ap, char *);
-			if(!p || !*p)
-				return 0;
-			strncat(buf, p, sz-1);
+			printf("%s", fs[i]);
+			fs[i++][0] = 0;
 			break;
 		default:
-			strncat(buf, fmt, 1);
+			putchar(*fmt);
 		}
-	printf("%s", buf);
-	va_end(ap);
 	return 1;
 }