shithub: neatroff

Info  •  Files  •  Log  •  Branches

hyph: drop non-alpha characters for HY_FIRST2 and HY_FINAL2

This prevents "finally..." from being hyphenated as "final-ly...".

--- a/hyph.c

+++ b/hyph.c

@@ -21,7 +21,7 @@

 static void hyph_initpatterns(void);

 static void hyph_initexceptions(void);

-static void hyfind(char *hyph, char *word);

+static void hyfind(char *hyph, char *word, int flg);

 static void hyexcept_add(char *s)

 {

@@ -66,7 +66,7 @@

 	return NULL;

 }

-void hyphenate(char *hyph, char *word)

+void hyphenate(char *hyph, char *word, int flg)

 {

 	char *r;

 	if (!hyinit) {

@@ -78,7 +78,7 @@

 	if (r)

 		memcpy(hyph, r, strlen(word) + 1);

 	else

-		hyfind(hyph, word);

+		hyfind(hyph, word, flg);

 }

 void tr_hw(char **args)

@@ -98,7 +98,7 @@

 	return (HYC_MAP(a) << 5) | HYC_MAP(b);

 }

-static void hyfind(char *hyph, char *word)

+static void hyfind(char *hyph, char *word, int flg)

 {

 	char n[ILNLEN] = {0};

 	char w[ILNLEN];

@@ -124,8 +124,11 @@

 				break;

 		}

 	}

-	for (i = 1; i < wlen - 1; i++)

-		hyph[i - 1] = i > 2 && i < wlen - 3 && n[i] % 2;

+	memset(hyph, 0, wlen * sizeof(hyph[0]));

+	for (i = 3; i < wlen - 2; i++)

+		if (n[i] % 2 && w[i - 1] != '.' && w[i - 2] != '.' && w[i + 1] != '.')

+			hyph[i - 1] = (~flg & HY_FINAL2 || w[i + 2] != '.') &&

+				(~flg & HY_FIRST2 || w[i - 3] != '.');

 }

 static void hyins(char *s)

--- a/roff.h

+++ b/roff.h

@@ -244,7 +244,7 @@

 #define HY_FIRST2	0x08	/* do not hyphenate the first two characters */

 #define HY_ANY		0x10	/* break at any possible position */

-void hyphenate(char *hyphs, char *word);

+void hyphenate(char *hyphs, char *word, int flg);

 /* adjustment */

 #define AD_L		0

--- a/wb.c

+++ b/wb.c

@@ -399,35 +399,36 @@

 static char *hyphpos(char *s, int w, struct wb *w1, int flg)

 {

-	char *map[ILNLEN] = {NULL};	/* mapping from word to s */

-	int fits[ILNLEN] = {0};		/* fits[i] if word[0..i]- fits w */

-	char word[ILNLEN];

-	char hyph[ILNLEN];

+	char word[ILNLEN];	/* word to pass to hyphenate() */

+	char hyph[ILNLEN];	/* hyphenation points returned from hyphenate() */

+	char *iw[ILNLEN];	/* beginning of i-th char in word */

+	char *is[ILNLEN];	/* beginning of i-th char in s */

+	int fits[ILNLEN];	/* fits[i] is 1, if the first i chars fit w */

+	int n = 0;		/* the number of characters in word */

 	char d[ILNLEN];

 	char *prev_s = s;

 	char *r = NULL;

 	char *wp = word, *we = word + sizeof(word);

-	int beg, end;

 	int i, c;

 	skipreqs(&s, w1);

-	while ((c = escread(&s, d)) >= 0 && wp + strlen(d) + 1 < we) {

+	while ((c = escread(&s, d)) >= 0 && (c > 0 || strlen(d) + 1 < we - wp)) {

+		fits[n] = wb_wid(w1) + wb_dashwid(w1) <= w;

 		wb_putc(w1, c, d);

 		if (c == 0) {

+			iw[n] = wp;

+			is[n] = prev_s;

 			strcpy(wp, d);

-			map[wp - word] = prev_s;

 			wp = strchr(wp, '\0');

-			fits[wp - word] = wb_wid(w1) + wb_dashwid(w1) <= w;

+			n++;

 		}

 		prev_s = s;

 	}

-	if (strlen(word) < 4)

+	if (n < 4)

 		return NULL;

-	hyphenate(hyph, word);

-	beg = flg & HY_FIRST2 ? 3 : 2;

-	end = strlen(word) - (flg & HY_FINAL2 ? 2 : 1);

-	for (i = beg; i < end; i++)

-		if (map[i] && hyph[i] && (fits[i] || ((flg & HY_ANY) && !r)))

-			r = map[i];

+	hyphenate(hyph, word, flg);

+	for (i = 2; i < n - 1; i++)

+		if (hyph[iw[i] - word] && (fits[i] || ((flg & HY_ANY) && !r)))

+			r = is[i];

 	return r;

 }