1*7c478bd9Sstevel@tonic-gate /* 2*7c478bd9Sstevel@tonic-gate * CDDL HEADER START 3*7c478bd9Sstevel@tonic-gate * 4*7c478bd9Sstevel@tonic-gate * The contents of this file are subject to the terms of the 5*7c478bd9Sstevel@tonic-gate * Common Development and Distribution License, Version 1.0 only 6*7c478bd9Sstevel@tonic-gate * (the "License"). You may not use this file except in compliance 7*7c478bd9Sstevel@tonic-gate * with the License. 8*7c478bd9Sstevel@tonic-gate * 9*7c478bd9Sstevel@tonic-gate * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 10*7c478bd9Sstevel@tonic-gate * or http://www.opensolaris.org/os/licensing. 11*7c478bd9Sstevel@tonic-gate * See the License for the specific language governing permissions 12*7c478bd9Sstevel@tonic-gate * and limitations under the License. 13*7c478bd9Sstevel@tonic-gate * 14*7c478bd9Sstevel@tonic-gate * When distributing Covered Code, include this CDDL HEADER in each 15*7c478bd9Sstevel@tonic-gate * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 16*7c478bd9Sstevel@tonic-gate * If applicable, add the following below this CDDL HEADER, with the 17*7c478bd9Sstevel@tonic-gate * fields enclosed by brackets "[]" replaced with your own identifying 18*7c478bd9Sstevel@tonic-gate * information: Portions Copyright [yyyy] [name of copyright owner] 19*7c478bd9Sstevel@tonic-gate * 20*7c478bd9Sstevel@tonic-gate * CDDL HEADER END 21*7c478bd9Sstevel@tonic-gate */ 22*7c478bd9Sstevel@tonic-gate /* 23*7c478bd9Sstevel@tonic-gate * Copyright 2003 Sun Microsystems, Inc. All rights reserved. 24*7c478bd9Sstevel@tonic-gate * Use is subject to license terms. 25*7c478bd9Sstevel@tonic-gate */ 26*7c478bd9Sstevel@tonic-gate 27*7c478bd9Sstevel@tonic-gate /* Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T */ 28*7c478bd9Sstevel@tonic-gate /* All Rights Reserved */ 29*7c478bd9Sstevel@tonic-gate 30*7c478bd9Sstevel@tonic-gate 31*7c478bd9Sstevel@tonic-gate #pragma ident "%Z%%M% %I% %E% SMI" 32*7c478bd9Sstevel@tonic-gate 33*7c478bd9Sstevel@tonic-gate /* 34*7c478bd9Sstevel@tonic-gate * University Copyright- Copyright (c) 1982, 1986, 1988 35*7c478bd9Sstevel@tonic-gate * The Regents of the University of California 36*7c478bd9Sstevel@tonic-gate * All Rights Reserved 37*7c478bd9Sstevel@tonic-gate * 38*7c478bd9Sstevel@tonic-gate * University Acknowledgment- Portions of this document are derived from 39*7c478bd9Sstevel@tonic-gate * software developed by the University of California, Berkeley, and its 40*7c478bd9Sstevel@tonic-gate * contributors. 41*7c478bd9Sstevel@tonic-gate */ 42*7c478bd9Sstevel@tonic-gate 43*7c478bd9Sstevel@tonic-gate #include <ctype.h> 44*7c478bd9Sstevel@tonic-gate #include "tdef.h" 45*7c478bd9Sstevel@tonic-gate #include "ext.h" 46*7c478bd9Sstevel@tonic-gate #define HY_BIT 0200 /* stuff in here only works for ascii */ 47*7c478bd9Sstevel@tonic-gate 48*7c478bd9Sstevel@tonic-gate /* 49*7c478bd9Sstevel@tonic-gate * troff8.c 50*7c478bd9Sstevel@tonic-gate * 51*7c478bd9Sstevel@tonic-gate * hyphenation 52*7c478bd9Sstevel@tonic-gate */ 53*7c478bd9Sstevel@tonic-gate 54*7c478bd9Sstevel@tonic-gate char hbuf[NHEX]; 55*7c478bd9Sstevel@tonic-gate char *nexth = hbuf; 56*7c478bd9Sstevel@tonic-gate tchar *hyend; 57*7c478bd9Sstevel@tonic-gate #define THRESH 160 /*digram goodness threshold*/ 58*7c478bd9Sstevel@tonic-gate int thresh = THRESH; 59*7c478bd9Sstevel@tonic-gate 60*7c478bd9Sstevel@tonic-gate hyphen(wp) 61*7c478bd9Sstevel@tonic-gate tchar *wp; 62*7c478bd9Sstevel@tonic-gate { 63*7c478bd9Sstevel@tonic-gate register j; 64*7c478bd9Sstevel@tonic-gate register tchar *i; 65*7c478bd9Sstevel@tonic-gate 66*7c478bd9Sstevel@tonic-gate i = wp; 67*7c478bd9Sstevel@tonic-gate while (punct(cbits(*i++))) 68*7c478bd9Sstevel@tonic-gate ; 69*7c478bd9Sstevel@tonic-gate if (!alph(cbits(*--i))) 70*7c478bd9Sstevel@tonic-gate return; 71*7c478bd9Sstevel@tonic-gate wdstart = i++; 72*7c478bd9Sstevel@tonic-gate while (alph(cbits(*i++))) 73*7c478bd9Sstevel@tonic-gate ; 74*7c478bd9Sstevel@tonic-gate hyend = wdend = --i - 1; 75*7c478bd9Sstevel@tonic-gate while (punct(cbits(*i++))) 76*7c478bd9Sstevel@tonic-gate ; 77*7c478bd9Sstevel@tonic-gate if (*--i) 78*7c478bd9Sstevel@tonic-gate return; 79*7c478bd9Sstevel@tonic-gate if ((wdend - wdstart - 4) < 0) 80*7c478bd9Sstevel@tonic-gate return; 81*7c478bd9Sstevel@tonic-gate hyp = hyptr; 82*7c478bd9Sstevel@tonic-gate *hyp = 0; 83*7c478bd9Sstevel@tonic-gate hyoff = 2; 84*7c478bd9Sstevel@tonic-gate if (!exword() && !suffix()) 85*7c478bd9Sstevel@tonic-gate digram(); 86*7c478bd9Sstevel@tonic-gate *hyp++ = 0; 87*7c478bd9Sstevel@tonic-gate if (*hyptr) 88*7c478bd9Sstevel@tonic-gate for (j = 1; j; ) { 89*7c478bd9Sstevel@tonic-gate j = 0; 90*7c478bd9Sstevel@tonic-gate for (hyp = hyptr + 1; *hyp != 0; hyp++) { 91*7c478bd9Sstevel@tonic-gate if (*(hyp - 1) > *hyp) { 92*7c478bd9Sstevel@tonic-gate j++; 93*7c478bd9Sstevel@tonic-gate i = *hyp; 94*7c478bd9Sstevel@tonic-gate *hyp = *(hyp - 1); 95*7c478bd9Sstevel@tonic-gate *(hyp - 1) = i; 96*7c478bd9Sstevel@tonic-gate } 97*7c478bd9Sstevel@tonic-gate } 98*7c478bd9Sstevel@tonic-gate } 99*7c478bd9Sstevel@tonic-gate } 100*7c478bd9Sstevel@tonic-gate 101*7c478bd9Sstevel@tonic-gate 102*7c478bd9Sstevel@tonic-gate punct(i) 103*7c478bd9Sstevel@tonic-gate { 104*7c478bd9Sstevel@tonic-gate if (!i || alph(i)) 105*7c478bd9Sstevel@tonic-gate return(0); 106*7c478bd9Sstevel@tonic-gate else 107*7c478bd9Sstevel@tonic-gate return(1); 108*7c478bd9Sstevel@tonic-gate } 109*7c478bd9Sstevel@tonic-gate 110*7c478bd9Sstevel@tonic-gate 111*7c478bd9Sstevel@tonic-gate alph(i) 112*7c478bd9Sstevel@tonic-gate { 113*7c478bd9Sstevel@tonic-gate if (i >= 'a' && i <= 'z' || i >= 'A' && i <= 'Z') 114*7c478bd9Sstevel@tonic-gate return(1); 115*7c478bd9Sstevel@tonic-gate else 116*7c478bd9Sstevel@tonic-gate return(0); 117*7c478bd9Sstevel@tonic-gate } 118*7c478bd9Sstevel@tonic-gate 119*7c478bd9Sstevel@tonic-gate 120*7c478bd9Sstevel@tonic-gate caseht() 121*7c478bd9Sstevel@tonic-gate { 122*7c478bd9Sstevel@tonic-gate thresh = THRESH; 123*7c478bd9Sstevel@tonic-gate if (skip()) 124*7c478bd9Sstevel@tonic-gate return; 125*7c478bd9Sstevel@tonic-gate noscale++; 126*7c478bd9Sstevel@tonic-gate thresh = atoi(); 127*7c478bd9Sstevel@tonic-gate noscale = 0; 128*7c478bd9Sstevel@tonic-gate } 129*7c478bd9Sstevel@tonic-gate 130*7c478bd9Sstevel@tonic-gate 131*7c478bd9Sstevel@tonic-gate casehw() 132*7c478bd9Sstevel@tonic-gate { 133*7c478bd9Sstevel@tonic-gate register i, k; 134*7c478bd9Sstevel@tonic-gate register char *j; 135*7c478bd9Sstevel@tonic-gate tchar t; 136*7c478bd9Sstevel@tonic-gate 137*7c478bd9Sstevel@tonic-gate k = 0; 138*7c478bd9Sstevel@tonic-gate while (!skip()) { 139*7c478bd9Sstevel@tonic-gate if ((j = nexth) >= (hbuf + NHEX - 2)) 140*7c478bd9Sstevel@tonic-gate goto full; 141*7c478bd9Sstevel@tonic-gate for (; ; ) { 142*7c478bd9Sstevel@tonic-gate if (ismot(t = getch())) 143*7c478bd9Sstevel@tonic-gate continue; 144*7c478bd9Sstevel@tonic-gate i = cbits(t); 145*7c478bd9Sstevel@tonic-gate if (i == ' ' || i == '\n') { 146*7c478bd9Sstevel@tonic-gate *j++ = 0; 147*7c478bd9Sstevel@tonic-gate nexth = j; 148*7c478bd9Sstevel@tonic-gate *j = 0; 149*7c478bd9Sstevel@tonic-gate if (i == ' ') 150*7c478bd9Sstevel@tonic-gate break; 151*7c478bd9Sstevel@tonic-gate else 152*7c478bd9Sstevel@tonic-gate return; 153*7c478bd9Sstevel@tonic-gate } 154*7c478bd9Sstevel@tonic-gate if (i == '-') { 155*7c478bd9Sstevel@tonic-gate k = HY_BIT; 156*7c478bd9Sstevel@tonic-gate continue; 157*7c478bd9Sstevel@tonic-gate } 158*7c478bd9Sstevel@tonic-gate *j++ = maplow(i) | k; 159*7c478bd9Sstevel@tonic-gate k = 0; 160*7c478bd9Sstevel@tonic-gate if (j >= (hbuf + NHEX - 2)) 161*7c478bd9Sstevel@tonic-gate goto full; 162*7c478bd9Sstevel@tonic-gate } 163*7c478bd9Sstevel@tonic-gate } 164*7c478bd9Sstevel@tonic-gate return; 165*7c478bd9Sstevel@tonic-gate full: 166*7c478bd9Sstevel@tonic-gate errprint(gettext("exception word list full.")); 167*7c478bd9Sstevel@tonic-gate *nexth = 0; 168*7c478bd9Sstevel@tonic-gate } 169*7c478bd9Sstevel@tonic-gate 170*7c478bd9Sstevel@tonic-gate 171*7c478bd9Sstevel@tonic-gate exword() 172*7c478bd9Sstevel@tonic-gate { 173*7c478bd9Sstevel@tonic-gate register tchar *w; 174*7c478bd9Sstevel@tonic-gate register char *e; 175*7c478bd9Sstevel@tonic-gate char *save; 176*7c478bd9Sstevel@tonic-gate 177*7c478bd9Sstevel@tonic-gate e = hbuf; 178*7c478bd9Sstevel@tonic-gate while (1) { 179*7c478bd9Sstevel@tonic-gate save = e; 180*7c478bd9Sstevel@tonic-gate if (*e == 0) 181*7c478bd9Sstevel@tonic-gate return(0); 182*7c478bd9Sstevel@tonic-gate w = wdstart; 183*7c478bd9Sstevel@tonic-gate while (*e && w <= hyend && (*e & 0177) == maplow(cbits(*w))) { 184*7c478bd9Sstevel@tonic-gate e++; 185*7c478bd9Sstevel@tonic-gate w++; 186*7c478bd9Sstevel@tonic-gate }; 187*7c478bd9Sstevel@tonic-gate if (!*e) { 188*7c478bd9Sstevel@tonic-gate if (w-1 == hyend || (w == wdend && maplow(cbits(*w)) == 's')) { 189*7c478bd9Sstevel@tonic-gate w = wdstart; 190*7c478bd9Sstevel@tonic-gate for (e = save; *e; e++) { 191*7c478bd9Sstevel@tonic-gate if (*e & HY_BIT) 192*7c478bd9Sstevel@tonic-gate *hyp++ = w; 193*7c478bd9Sstevel@tonic-gate if (hyp > (hyptr + NHYP - 1)) 194*7c478bd9Sstevel@tonic-gate hyp = hyptr + NHYP - 1; 195*7c478bd9Sstevel@tonic-gate w++; 196*7c478bd9Sstevel@tonic-gate } 197*7c478bd9Sstevel@tonic-gate return(1); 198*7c478bd9Sstevel@tonic-gate } else { 199*7c478bd9Sstevel@tonic-gate e++; 200*7c478bd9Sstevel@tonic-gate continue; 201*7c478bd9Sstevel@tonic-gate } 202*7c478bd9Sstevel@tonic-gate } else 203*7c478bd9Sstevel@tonic-gate while (*e++) 204*7c478bd9Sstevel@tonic-gate ; 205*7c478bd9Sstevel@tonic-gate } 206*7c478bd9Sstevel@tonic-gate } 207*7c478bd9Sstevel@tonic-gate 208*7c478bd9Sstevel@tonic-gate 209*7c478bd9Sstevel@tonic-gate suffix() 210*7c478bd9Sstevel@tonic-gate { 211*7c478bd9Sstevel@tonic-gate register tchar *w; 212*7c478bd9Sstevel@tonic-gate register char *s, *s0; 213*7c478bd9Sstevel@tonic-gate tchar i; 214*7c478bd9Sstevel@tonic-gate extern char *suftab[]; 215*7c478bd9Sstevel@tonic-gate extern tchar *chkvow(); 216*7c478bd9Sstevel@tonic-gate 217*7c478bd9Sstevel@tonic-gate again: 218*7c478bd9Sstevel@tonic-gate if (!alph(cbits(i = cbits(*hyend)))) 219*7c478bd9Sstevel@tonic-gate return(0); 220*7c478bd9Sstevel@tonic-gate if (i < 'a') 221*7c478bd9Sstevel@tonic-gate i -= 'A' - 'a'; 222*7c478bd9Sstevel@tonic-gate if ((s0 = suftab[i-'a']) == 0) 223*7c478bd9Sstevel@tonic-gate return(0); 224*7c478bd9Sstevel@tonic-gate for (; ; ) { 225*7c478bd9Sstevel@tonic-gate if ((i = *s0 & 017) == 0) 226*7c478bd9Sstevel@tonic-gate return(0); 227*7c478bd9Sstevel@tonic-gate s = s0 + i - 1; 228*7c478bd9Sstevel@tonic-gate w = hyend - 1; 229*7c478bd9Sstevel@tonic-gate while (s > s0 && w >= wdstart && (*s & 0177) == maplow(cbits(*w))) { 230*7c478bd9Sstevel@tonic-gate s--; 231*7c478bd9Sstevel@tonic-gate w--; 232*7c478bd9Sstevel@tonic-gate } 233*7c478bd9Sstevel@tonic-gate if (s == s0) 234*7c478bd9Sstevel@tonic-gate break; 235*7c478bd9Sstevel@tonic-gate s0 += i; 236*7c478bd9Sstevel@tonic-gate } 237*7c478bd9Sstevel@tonic-gate s = s0 + i - 1; 238*7c478bd9Sstevel@tonic-gate w = hyend; 239*7c478bd9Sstevel@tonic-gate if (*s0 & HY_BIT) 240*7c478bd9Sstevel@tonic-gate goto mark; 241*7c478bd9Sstevel@tonic-gate while (s > s0) { 242*7c478bd9Sstevel@tonic-gate w--; 243*7c478bd9Sstevel@tonic-gate if (*s-- & HY_BIT) { 244*7c478bd9Sstevel@tonic-gate mark: 245*7c478bd9Sstevel@tonic-gate hyend = w - 1; 246*7c478bd9Sstevel@tonic-gate if (*s0 & 0100) 247*7c478bd9Sstevel@tonic-gate continue; 248*7c478bd9Sstevel@tonic-gate if (!chkvow(w)) 249*7c478bd9Sstevel@tonic-gate return(0); 250*7c478bd9Sstevel@tonic-gate *hyp++ = w; 251*7c478bd9Sstevel@tonic-gate } 252*7c478bd9Sstevel@tonic-gate } 253*7c478bd9Sstevel@tonic-gate if (*s0 & 040) 254*7c478bd9Sstevel@tonic-gate return(0); 255*7c478bd9Sstevel@tonic-gate if (exword()) 256*7c478bd9Sstevel@tonic-gate return(1); 257*7c478bd9Sstevel@tonic-gate goto again; 258*7c478bd9Sstevel@tonic-gate } 259*7c478bd9Sstevel@tonic-gate 260*7c478bd9Sstevel@tonic-gate 261*7c478bd9Sstevel@tonic-gate maplow(i) 262*7c478bd9Sstevel@tonic-gate register int i; 263*7c478bd9Sstevel@tonic-gate { 264*7c478bd9Sstevel@tonic-gate if (ischar(i) && isupper(i)) 265*7c478bd9Sstevel@tonic-gate i = tolower(i); 266*7c478bd9Sstevel@tonic-gate return(i); 267*7c478bd9Sstevel@tonic-gate } 268*7c478bd9Sstevel@tonic-gate 269*7c478bd9Sstevel@tonic-gate 270*7c478bd9Sstevel@tonic-gate vowel(i) 271*7c478bd9Sstevel@tonic-gate int i; 272*7c478bd9Sstevel@tonic-gate { 273*7c478bd9Sstevel@tonic-gate switch (maplow(i)) { 274*7c478bd9Sstevel@tonic-gate case 'a': 275*7c478bd9Sstevel@tonic-gate case 'e': 276*7c478bd9Sstevel@tonic-gate case 'i': 277*7c478bd9Sstevel@tonic-gate case 'o': 278*7c478bd9Sstevel@tonic-gate case 'u': 279*7c478bd9Sstevel@tonic-gate case 'y': 280*7c478bd9Sstevel@tonic-gate return(1); 281*7c478bd9Sstevel@tonic-gate default: 282*7c478bd9Sstevel@tonic-gate return(0); 283*7c478bd9Sstevel@tonic-gate } 284*7c478bd9Sstevel@tonic-gate } 285*7c478bd9Sstevel@tonic-gate 286*7c478bd9Sstevel@tonic-gate 287*7c478bd9Sstevel@tonic-gate tchar *chkvow(w) 288*7c478bd9Sstevel@tonic-gate tchar *w; 289*7c478bd9Sstevel@tonic-gate { 290*7c478bd9Sstevel@tonic-gate while (--w >= wdstart) 291*7c478bd9Sstevel@tonic-gate if (vowel(cbits(*w))) 292*7c478bd9Sstevel@tonic-gate return(w); 293*7c478bd9Sstevel@tonic-gate return(0); 294*7c478bd9Sstevel@tonic-gate } 295*7c478bd9Sstevel@tonic-gate 296*7c478bd9Sstevel@tonic-gate 297*7c478bd9Sstevel@tonic-gate digram() 298*7c478bd9Sstevel@tonic-gate { 299*7c478bd9Sstevel@tonic-gate register tchar *w; 300*7c478bd9Sstevel@tonic-gate register val; 301*7c478bd9Sstevel@tonic-gate tchar * nhyend, *maxw; 302*7c478bd9Sstevel@tonic-gate int maxval; 303*7c478bd9Sstevel@tonic-gate extern char bxh[26][13], bxxh[26][13], xxh[26][13], xhx[26][13], hxx[26][13]; 304*7c478bd9Sstevel@tonic-gate 305*7c478bd9Sstevel@tonic-gate again: 306*7c478bd9Sstevel@tonic-gate if (!(w = chkvow(hyend + 1))) 307*7c478bd9Sstevel@tonic-gate return; 308*7c478bd9Sstevel@tonic-gate hyend = w; 309*7c478bd9Sstevel@tonic-gate if (!(w = chkvow(hyend))) 310*7c478bd9Sstevel@tonic-gate return; 311*7c478bd9Sstevel@tonic-gate nhyend = w; 312*7c478bd9Sstevel@tonic-gate maxval = 0; 313*7c478bd9Sstevel@tonic-gate w--; 314*7c478bd9Sstevel@tonic-gate while ((++w < hyend) && (w < (wdend - 1))) { 315*7c478bd9Sstevel@tonic-gate val = 1; 316*7c478bd9Sstevel@tonic-gate if (w == wdstart) 317*7c478bd9Sstevel@tonic-gate val *= dilook('a', cbits(*w), bxh); 318*7c478bd9Sstevel@tonic-gate else if (w == wdstart + 1) 319*7c478bd9Sstevel@tonic-gate val *= dilook(cbits(*(w-1)), cbits(*w), bxxh); 320*7c478bd9Sstevel@tonic-gate else 321*7c478bd9Sstevel@tonic-gate val *= dilook(cbits(*(w-1)), cbits(*w), xxh); 322*7c478bd9Sstevel@tonic-gate val *= dilook(cbits(*w), cbits(*(w+1)), xhx); 323*7c478bd9Sstevel@tonic-gate val *= dilook(cbits(*(w+1)), cbits(*(w+2)), hxx); 324*7c478bd9Sstevel@tonic-gate if (val > maxval) { 325*7c478bd9Sstevel@tonic-gate maxval = val; 326*7c478bd9Sstevel@tonic-gate maxw = w + 1; 327*7c478bd9Sstevel@tonic-gate } 328*7c478bd9Sstevel@tonic-gate } 329*7c478bd9Sstevel@tonic-gate hyend = nhyend; 330*7c478bd9Sstevel@tonic-gate if (maxval > thresh) 331*7c478bd9Sstevel@tonic-gate *hyp++ = maxw; 332*7c478bd9Sstevel@tonic-gate goto again; 333*7c478bd9Sstevel@tonic-gate } 334*7c478bd9Sstevel@tonic-gate 335*7c478bd9Sstevel@tonic-gate 336*7c478bd9Sstevel@tonic-gate dilook(a, b, t) 337*7c478bd9Sstevel@tonic-gate int a, b; 338*7c478bd9Sstevel@tonic-gate char t[26][13]; 339*7c478bd9Sstevel@tonic-gate { 340*7c478bd9Sstevel@tonic-gate register i, j; 341*7c478bd9Sstevel@tonic-gate 342*7c478bd9Sstevel@tonic-gate i = t[maplow(a)-'a'][(j = maplow(b)-'a')/2]; 343*7c478bd9Sstevel@tonic-gate if (!(j & 01)) 344*7c478bd9Sstevel@tonic-gate i >>= 4; 345*7c478bd9Sstevel@tonic-gate return(i & 017); 346*7c478bd9Sstevel@tonic-gate } 347*7c478bd9Sstevel@tonic-gate 348*7c478bd9Sstevel@tonic-gate 349