hyph.c

   1 /* hyphenation */
   2 #include <ctype.h>
   3 #include <stdio.h>
   4 #include <string.h>
   5 #include "roff.h"
   6 #include "hyen.h"
   7
   8 #define HYPATLEN        (NHYPHS * 16)   /* hyphenation pattern length */
   9
  10 static void hcode_strcpy(char *d, char *s, int *map, int dots);
  11
  12 /* the hyphenation dictionary (.hw) */
  13
  14 static char hwword[HYPATLEN];   /* buffer for .hw words */
  15 static char hwhyph[HYPATLEN];   /* buffer for .hw hyphenations */
  16 static int hwword_len;          /* used hwword[] length */
  17 static struct dict hwdict;      /* map words to their index in hwoff[] */
  18 static int hwoff[NHYPHS];       /* the offset of words in hwword[] */
  19 static int hw_n;                /* the number of dictionary words */
  20
  21 /* insert word s into hwword[] and hwhyph[] */
  22 static void hw_add(char *s)
  23 {
  24         char *p = hwword + hwword_len;
  25         char *n = hwhyph + hwword_len;
  26         int len = strlen(s) + 1;
  27         int i = 0, c;
  28         if (hw_n == NHYPHS || hwword_len + len > sizeof(hwword))
  29                 return;
  30         memset(n, 0, len);
  31         while ((c = (unsigned char) *s++)) {
  32                 if (c == '-')
  33                         n[i] = 1;
  34                 else
  35                         p[i++] = c;
  36         }
  37         p[i] = '\0';
  38         hwoff[hw_n] = hwword_len;
  39         dict_put(&hwdict, hwword + hwoff[hw_n], hw_n);
  40         hwword_len += i + 1;
  41         hw_n++;
  42 }
  43
  44 static int hw_lookup(char *word, char *hyph)
  45 {
  46         char word2[WORDLEN] = {0};
  47         char *hyph2;
  48         int map[WORDLEN] = {0};
  49         int i, j, idx = -1;
  50         hcode_strcpy(word2, word, map, 0);
  51         i = dict_prefix(&hwdict, word2, &idx);
  52         if (i < 0)
  53                 return 1;
  54         hyph2 = hwhyph + hwoff[i];
  55         for (j = 0; word2[j]; j++)
  56                 if (hyph2[j])
  57                         hyph[map[j]] = hyph2[j];
  58         return 0;
  59 }
  60
  61 void tr_hw(char **args)
  62 {
  63         int i;
  64         for (i = 1; i < NARGS && args[i]; i++)
  65                 hw_add(args[i]);
  66 }
  67
  68 /* the tex hyphenation algorithm */
  69
  70 static int hyinit;              /* hyphenation data initialized */
  71 static char hypats[HYPATLEN];   /* hyphenation patterns */
  72 static char hynums[HYPATLEN];   /* hyphenation pattern numbers */
  73 static int hypats_len;          /* used hypats[] and hynums[] length */
  74 static struct dict hydict;      /* map patterns to their index in hyoff[] */
  75 static int hyoff[NHYPHS];       /* the offset of this pattern in hypats[] */
  76 static int hy_n;                /* the number of patterns */
  77
  78 /* find the patterns matching s and update hyphenation values in n */
  79 static void hy_find(char *s, char *n)
  80 {
  81         int plen;
  82         char *p, *np;
  83         int i, j;
  84         int idx = -1;
  85         while ((i = dict_prefix(&hydict, s, &idx)) >= 0) {
  86                 p = hypats + hyoff[i];
  87                 np = hynums + (p - hypats);
  88                 plen = strlen(p) + 1;
  89                 for (j = 0; j < plen; j++)
  90                         if (n[j] < np[j])
  91                                 n[j] = np[j];
  92         }
  93 }
  94
  95 /* mark the hyphenation points of word in hyph */
  96 static void hy_dohyph(char *hyph, char *word, int flg)
  97 {
  98         char n[WORDLEN] = {0};
  99         char w[WORDLEN] = {0};
 100         int c[WORDLEN];                 /* start of the i-th character in w */
 101         int wmap[WORDLEN] = {0};        /* word[wmap[i]] is w[i] */
 102         int nc = 0;
 103         int i, wlen;
 104         hcode_strcpy(w, word, wmap, 1);
 105         wlen = strlen(w);
 106         for (i = 0; i < wlen - 1; i += utf8len((unsigned char) w[i]))
 107                 c[nc++] = i;
 108         for (i = 0; i < nc - 1; i++)
 109                 hy_find(w + c[i], n + c[i]);
 110         memset(hyph, 0, wlen * sizeof(hyph[0]));
 111         for (i = 3; i < nc - 2; i++)
 112                 if (n[c[i]] % 2 && w[c[i - 1]] != '.' &&
 113                                 w[c[i - 2]] != '.' && w[c[i + 1]] != '.' &&
 114                                 (~flg & HY_FINAL2 || w[c[i + 2]] != '.') &&
 115                                 (~flg & HY_FIRST2 || w[c[i - 3]] != '.'))
 116                         hyph[wmap[c[i]]] = 1;
 117 }
 118
 119 /* insert pattern s into hypats[] and hynums[] */
 120 static void hy_add(char *s)
 121 {
 122         char *p = hypats + hypats_len;
 123         char *n = hynums + hypats_len;
 124         int len = strlen(s) + 1;
 125         int i = 0, c;
 126         if (hy_n >= NHYPHS || hypats_len + len >= sizeof(hypats))
 127                 return;
 128         memset(n, 0, len);
 129         while ((c = (unsigned char) *s++)) {
 130                 if (c >= '0' && c <= '9')
 131                         n[i] = c - '0';
 132                 else
 133                         p[i++] = c;
 134         }
 135         p[i] = '\0';
 136         hyoff[hy_n] = hypats_len;
 137         dict_put(&hydict, hypats + hyoff[hy_n], hy_n);
 138         hypats_len += i + 1;
 139         hy_n++;
 140 }
 141
 142 /* .hcode request */
 143 static struct dict hcodedict;
 144 static char hcodesrc[NHCODES][GNLEN];
 145 static char hcodedst[NHCODES][GNLEN];
 146 static int hcode_n;
 147
 148 /* replace the character in s after .hcode mapping; returns s's new length */
 149 static int hcode_mapchar(char *s)
 150 {
 151         int i = dict_get(&hcodedict, s);
 152         if (i >= 0)
 153                 strcpy(s, hcodedst[i]);
 154         else if (isalpha((unsigned char) *s))
 155                 *s = tolower(*s);
 156         return strlen(s);
 157 }
 158
 159 /* copy s to d after .hcode mappings; s[map[j]] corresponds to d[j] */
 160 static void hcode_strcpy(char *d, char *s, int *map, int dots)
 161 {
 162         int di = 0, si = 0, len;
 163         if (dots)
 164                 d[di++] = '.';
 165         while (di < WORDLEN - GNLEN && s[si]) {
 166                 len = utf8len((unsigned char) s[si]);
 167                 map[di] = si;
 168                 memcpy(d + di, s + si, len);
 169                 si += len;
 170                 di += hcode_mapchar(d + di);
 171         }
 172         if (dots)
 173                 d[di++] = '.';
 174         d[di] = '\0';
 175 }
 176
 177 static void hcode_add(char *c1, char *c2)
 178 {
 179         int i = dict_get(&hcodedict, c1);
 180         if (i >= 0) {
 181                 strcpy(hcodedst[i], c2);
 182         } else if (hcode_n < NHCODES) {
 183                 strcpy(hcodesrc[hcode_n], c1);
 184                 strcpy(hcodedst[hcode_n], c2);
 185                 dict_put(&hcodedict, hcodesrc[hcode_n], hcode_n);
 186                 hcode_n++;
 187         }
 188 }
 189
 190 void tr_hcode(char **args)
 191 {
 192         char c1[GNLEN], c2[GNLEN];
 193         char *s = args[1];
 194         while (s && utf8read(&s, c1) && utf8read(&s, c2))
 195                 hcode_add(c1, c2);
 196 }
 197
 198 static void hyph_readpatterns(char *s)
 199 {
 200         char word[WORDLEN];
 201         char *d;
 202         while (*s) {
 203                 d = word;
 204                 while (*s && !isspace((unsigned char) *s))
 205                         *d++ = *s++;
 206                 *d = '\0';
 207                 hy_add(word);
 208                 while (*s && isspace((unsigned char) *s))
 209                         s++;
 210         }
 211 }
 212
 213 static void hyph_readexceptions(char *s)
 214 {
 215         char word[WORDLEN];
 216         char *d;
 217         while (*s) {
 218                 d = word;
 219                 while (*s && !isspace((unsigned char) *s))
 220                         *d++ = *s++;
 221                 *d = '\0';
 222                 hw_add(word);
 223                 while (*s && isspace((unsigned char) *s))
 224                         s++;
 225         }
 226 }
 227
 228 void hyphenate(char *hyph, char *word, int flg)
 229 {
 230         if (!hyinit) {
 231                 hyinit = 1;
 232                 hyph_readpatterns(en_patterns);
 233                 hyph_readexceptions(en_exceptions);
 234         }
 235         if (hw_lookup(word, hyph))
 236                 hy_dohyph(hyph, word, flg);
 237 }
 238
 239 void tr_hpfa(char **args)
 240 {
 241         char tok[ILNLEN], c1[ILNLEN], c2[ILNLEN];
 242         FILE *filp;
 243         hyinit = 1;
 244         /* load english hyphenation patterns with no arguments */
 245         if (!args[1]) {
 246                 hyph_readpatterns(en_patterns);
 247                 hyph_readexceptions(en_exceptions);
 248         }
 249         /* reading patterns */
 250         if (args[1] && (filp = fopen(args[1], "r"))) {
 251                 while (fscanf(filp, "%s", tok) == 1)
 252                         if (strlen(tok) < WORDLEN)
 253                                 hy_add(tok);
 254                 fclose(filp);
 255         }
 256         /* reading exceptions */
 257         if (args[2] && (filp = fopen(args[2], "r"))) {
 258                 while (fscanf(filp, "%s", tok) == 1)
 259                         if (strlen(tok) < WORDLEN)
 260                                 hw_add(tok);
 261                 fclose(filp);
 262         }
 263         /* reading hcode mappings */
 264         if (args[3] && (filp = fopen(args[3], "r"))) {
 265                 while (fscanf(filp, "%s", tok) == 1) {
 266                         char *s = tok;
 267                         if (utf8read(&s, c1) && utf8read(&s, c2) && !*s)
 268                                 hcode_add(c2, c1);      /* inverting */
 269                 }
 270                 fclose(filp);
 271         }
 272 }
 273
 274 void hyph_init(void)
 275 {
 276         dict_init(&hwdict, NHYPHS, -1, 0, 1);
 277         dict_init(&hydict, NHYPHS, -1, 0, 1);
 278         dict_init(&hcodedict, NHYPHS, -1, 0, 1);
 279 }
 280
 281 void tr_hpf(char **args)
 282 {
 283         /* reseting the patterns */
 284         hypats_len = 0;
 285         hy_n = 0;
 286         dict_done(&hydict);
 287         /* reseting the dictionary */
 288         hwword_len = 0;
 289         hw_n = 0;
 290         dict_done(&hwdict);
 291         /* reseting hcode mappings */
 292         hcode_n = 0;
 293         dict_done(&hcodedict);
 294         /* reading */
 295         hyph_init();
 296         tr_hpfa(args);
 297 }