ref: 03d1fce55d597a3b06cc7c078f510791ab90a430
parent: acf6476228825e442f68bd4b36f4dcede001eab0
author: Ali Gholami Rudi <ali@rudi.ir>
date: Fri Jul 26 07:51:00 EDT 2013
wb: allow longer ligatures Now ligatures can be LIGLEN long; wb struct stores the last LIGLEN characters for wb_lig() and wb_eos(). Also after this change, kerning pairs do not prevent ligatures from being matched. Note however that this makes neatroff slower.
--- a/font.c
+++ b/font.c
@@ -99,13 +99,32 @@
return 1;
}
-/* return 1 if lig is a ligature */
-int font_lig(struct font *fn, char *lig)
+/*
+ * Given a list of characters in the reverse order, font_lig()
+ * returns the number of characters from the beginning of this
+ * list that form a ligature in this font. Zero naturally means
+ * no ligature was matched.
+ */
+int font_lig(struct font *fn, char **c, int n)
{int i;
- for (i = 0; i < fn->nlig; i++)
- if (!strcmp(lig, fn->lig[i]))
- return font_find(fn, lig) != NULL;
+ /* concatenated characters in c[], in the correct order */
+ char s[GNLEN * 2] = "";
+ /* b[i] is the number of character of c[] in s + i */
+ int b[GNLEN * 2] = {0};+ int len = 0;
+ for (i = 0; i < n; i++) {+ char *cur = c[n - i - 1];
+ b[len] = n - i;
+ strcpy(s + len, cur);
+ len += strlen(cur);
+ }
+ for (i = 0; i < fn->nlig; i++) {+ int l = strlen(fn->lig[i]);
+ if (b[len - l] && !strcmp(s + len - l, fn->lig[i]))
+ if (font_find(fn, fn->lig[i]))
+ return b[len - l];
+ }
return 0;
}
--- a/roff.h
+++ b/roff.h
@@ -26,6 +26,7 @@
#define NTABS 16 /* number of tab stops */
#define NFIELDS 32 /* number of fields */
#define MAXFRAC 100000 /* maximum value of the fractional part */
+#define LIGLEN 4 /* length of ligatures */
/* escape sequences */
#define ESC_Q "bCDhHlLNoSvwxX" /* \X'ccc' quoted escape sequences */
@@ -125,7 +126,7 @@
void font_close(struct font *fn);
struct glyph *font_glyph(struct font *fn, char *id);
struct glyph *font_find(struct font *fn, char *name);
-int font_lig(struct font *fn, char *c);
+int font_lig(struct font *fn, char **c, int n);
int font_kern(struct font *fn, char *c1, char *c2);
/* glyph handling functions */
@@ -164,7 +165,6 @@
char *s; /* allocated buffer */
int sz; /* buffer size */
int n; /* length of the string stored in s */
- int prev_n; /* n before the last sbuf_append() */
};
void sbuf_init(struct sbuf *sbuf);
@@ -174,7 +174,7 @@
void sbuf_append(struct sbuf *sbuf, char *s);
void sbuf_printf(struct sbuf *sbuf, char *s, ...);
void sbuf_putnl(struct sbuf *sbuf);
-void sbuf_pop(struct sbuf *sbuf);
+void sbuf_cut(struct sbuf *sbuf, int n);
int sbuf_len(struct sbuf *sbuf);
int sbuf_empty(struct sbuf *sbuf);
@@ -187,10 +187,12 @@
int els_neg, els_pos; /* extra line spacing */
int h, v; /* buffer vertical and horizontal positions */
int ct, sb, st; /* \w registers */
- char prev_c[GNLEN]; /* previous character added via wb_put() */
- int prev_h; /* wb->h after wb_put() calls */
- int prev_l; /* sbuf_len(&wb->sbuf) after wb_put() calls */
- int eos; /* nonzero if wb ends a sentence (.?!) */
+ /* saving previous characters added via wb_put() */
+ char prev_c[LIGLEN][GNLEN];
+ int prev_l[LIGLEN]; /* sbuf_len(&wb->sbuf) before wb_put() calls */
+ int prev_h[LIGLEN]; /* wb->h before wb_put() calls */
+ int prev_n; /* number of characters in prev_c[] */
+ int prev_ll; /* sbuf_len(&wb->sbuf) after the last wb_put() */
};
void wb_init(struct wb *wb);
--- a/sbuf.c
+++ b/sbuf.c
@@ -35,7 +35,6 @@
if (sbuf->n + len + 1 >= sbuf->sz)
sbuf_extend(sbuf, sbuf->n + len + 1);
memcpy(sbuf->s + sbuf->n, s, len);
- sbuf->prev_n = sbuf->n;
sbuf->n += len;
}
@@ -71,11 +70,11 @@
return sbuf->n;
}
-/* undo last sbuf_append() */
-void sbuf_pop(struct sbuf *sbuf)
+/* shorten the sbuf */
+void sbuf_cut(struct sbuf *sbuf, int n)
{- if (sbuf->prev_n < sbuf->n)
- sbuf->n = sbuf->prev_n;
+ if (sbuf->n > n)
+ sbuf->n = n;
}
void sbuf_done(struct sbuf *sbuf)
--- a/wb.c
+++ b/wb.c
@@ -76,9 +76,58 @@
sbuf_printf(&wb->sbuf, "%cX%s", c_ec, x);
}
+/* make sure nothing is appended to wb after the last wb_put() */
+static void wb_prevcheck(struct wb *wb)
+{+ if (wb->prev_ll != sbuf_len(&wb->sbuf))
+ wb->prev_n = 0;
+}
+
+/* mark wb->prev_c[] as valid */
+static void wb_prevok(struct wb *wb)
+{+ wb->prev_ll = sbuf_len(&wb->sbuf);
+}
+
+/* append c to wb->prev_c[] */
+static void wb_prevput(struct wb *wb, char *c, int ll)
+{+ if (wb->prev_n == LEN(wb->prev_c))
+ wb->prev_n--;
+ memmove(wb->prev_l + 1, wb->prev_l, wb->prev_n * sizeof(wb->prev_l[0]));
+ memmove(wb->prev_h + 1, wb->prev_h, wb->prev_n * sizeof(wb->prev_h[0]));
+ memmove(wb->prev_c + 1, wb->prev_c, wb->prev_n * sizeof(wb->prev_c[0]));
+ wb->prev_l[0] = ll;
+ wb->prev_h[0] = wb->h;
+ strcpy(wb->prev_c[0], c);
+ wb->prev_n++;
+ wb_prevok(wb);
+}
+
+/* strip the last i characters from wb */
+static void wb_prevpop(struct wb *wb, int i)
+{+ int n = wb->prev_n - i;
+ sbuf_cut(&wb->sbuf, wb->prev_l[i - 1]);
+ wb->h = wb->prev_h[i - 1];
+ memmove(wb->prev_l, wb->prev_l + i, n * sizeof(wb->prev_l[0]));
+ memmove(wb->prev_h, wb->prev_h + i, n * sizeof(wb->prev_h[0]));
+ memmove(wb->prev_c, wb->prev_c + i, n * sizeof(wb->prev_c[0]));
+ wb->prev_n = n;
+ wb->prev_ll = sbuf_len(&wb->sbuf);
+}
+
+/* return the i-th last character inserted via wb_put() */
+static char *wb_prev(struct wb *wb, int i)
+{+ wb_prevcheck(wb);
+ return i < wb->prev_n ? wb->prev_c[i] : NULL;
+}
+
void wb_put(struct wb *wb, char *c)
{struct glyph *g;
+ int ll;
if (c[0] == '\n') {wb->part = 0;
return;
@@ -94,6 +143,8 @@
}
g = dev_glyph(c, R_F(wb));
wb_font(wb);
+ wb_prevcheck(wb); /* make sure wb->prev_c[] is valid */
+ ll = sbuf_len(&wb->sbuf); /* sbuf length before inserting c */
if (!c[1] || c[0] == c_ec || c[0] == c_ni ||
utf8len((unsigned char) c[0]) == strlen(c)) {if (c[0] == c_ni && c[1] == c_ec)
@@ -107,15 +158,9 @@
sbuf_printf(&wb->sbuf, "%cC'%s'", c_ec, c);
}
if (strcmp(c_hc, c)) {- strcpy(wb->prev_c, c);
- wb->prev_l = sbuf_len(&wb->sbuf);
- wb->prev_h = wb->h;
+ wb_prevput(wb, c, ll);
wb->h += charwid(R_F(wb), R_S(wb), g ? g->wid : SC_DW);
wb->ct |= g ? g->type : 0;
- if (c[1])
- wb->eos = 0;
- else if (strchr("'\")]*", c[0]) == NULL)- wb->eos = strchr(".?!", c[0]) != NULL;wb_stsb(wb);
}
}
@@ -123,13 +168,19 @@
/* return zero if c formed a ligature with its previous character */
int wb_lig(struct wb *wb, char *c)
{- char lig[GNLEN * 2];
- if (wb->prev_l != sbuf_len(&wb->sbuf) || !wb->prev_c[0])
- return 1;
- sprintf(lig, "%s%s", wb->prev_c, c);
- if (font_lig(dev_font(R_F(wb)), lig)) {- wb->h = wb->prev_h;
- sbuf_pop(&wb->sbuf);
+ char lig[GNLEN] = "";
+ char *cs[LIGLEN + 2];
+ int i = -1;
+ int ligpos;
+ cs[0] = c;
+ while (wb_prev(wb, ++i))
+ cs[i + 1] = wb_prev(wb, i);
+ ligpos = font_lig(dev_font(R_F(wb)), cs, i + 1);
+ if (ligpos > 0) {+ for (i = 0; i < ligpos - 1; i++)
+ strcat(lig, wb_prev(wb, ligpos - i - 2));
+ strcat(lig, c);
+ wb_prevpop(wb, ligpos - 1);
wb_put(wb, lig);
return 0;
}
@@ -140,11 +191,12 @@
int wb_kern(struct wb *wb, char *c)
{int val;
- if (wb->prev_l != sbuf_len(&wb->sbuf) || !wb->prev_c[0])
+ if (!wb_prev(wb, 0))
return 1;
- val = font_kern(dev_font(R_F(wb)), wb->prev_c, c);
+ val = font_kern(dev_font(R_F(wb)), wb_prev(wb, 0), c);
if (val)
wb_hmov(wb, charwid(R_F(wb), R_S(wb), val));
+ wb_prevok(wb); /* kerning should not prevent ligatures */
return !val;
}
@@ -274,9 +326,13 @@
return sbuf_empty(&wb->sbuf);
}
+/* return 1 if wb ends a sentence (.?!) */
int wb_eos(struct wb *wb)
{- return wb->eos;
+ int i = 0;
+ while (wb_prev(wb, i) && strchr("'\")]*", wb_prev(wb, i)[0]))+ i++;
+ return wb_prev(wb, i) && strchr(".?!", wb_prev(wb, i)[0]);}
void wb_wconf(struct wb *wb, int *ct, int *st, int *sb)