vi/v_sentence.c

   1 /*-
   2  * Copyright (c) 1992, 1993, 1994
   3  *      The Regents of the University of California.  All rights reserved.
   4  * Copyright (c) 1992, 1993, 1994, 1995, 1996
   5  *      Keith Bostic.  All rights reserved.
   6  *
   7  * See the LICENSE file for redistribution information.
   8  */
   9
  10 #include "config.h"
  11
  12 #ifndef lint
  13 static const char sccsid[] = "$Id: v_sentence.c,v 10.8 2000/04/21 19:00:41 skimo Exp $ (Berkeley) $Date: 2000/04/21 19:00:41 $";
  14 #endif /* not lint */
  15
  16 #include <sys/types.h>
  17 #include <sys/queue.h>
  18 #include <sys/time.h>
  19
  20 #include <bitstring.h>
  21 #include <ctype.h>
  22 #include <limits.h>
  23 #include <stdio.h>
  24
  25 #include "../common/common.h"
  26 #include "vi.h"
  27
  28 /*
  29  * !!!
  30  * In historic vi, a sentence was delimited by a '.', '?' or '!' character
  31  * followed by TWO spaces or a newline.  One or more empty lines was also
  32  * treated as a separate sentence.  The Berkeley documentation for historical
  33  * vi states that any number of ')', ']', '"' and '\'' characters can be
  34  * between the delimiter character and the spaces or end of line, however,
  35  * the historical implementation did not handle additional '"' characters.
  36  * We follow the documentation here, not the implementation.
  37  *
  38  * Once again, historical vi didn't do sentence movements associated with
  39  * counts consistently, mostly in the presence of lines containing only
  40  * white-space characters.
  41  *
  42  * This implementation also permits a single tab to delimit sentences, and
  43  * treats lines containing only white-space characters as empty lines.
  44  * Finally, tabs are eaten (along with spaces) when skipping to the start
  45  * of the text following a "sentence".
  46  */
  47
  48 /*
  49  * v_sentencef -- [count])
  50  *      Move forward count sentences.
  51  *
  52  * PUBLIC: int v_sentencef __P((SCR *, VICMD *));
  53  */
  54 int
  55 v_sentencef(sp, vp)
  56         SCR *sp;
  57         VICMD *vp;
  58 {
  59         enum { BLANK, NONE, PERIOD } state;
  60         VCS cs;
  61         size_t len;
  62         u_long cnt;
  63
  64         cs.cs_lno = vp->m_start.lno;
  65         cs.cs_cno = vp->m_start.cno;
  66         if (cs_init(sp, &cs))
  67                 return (1);
  68
  69         cnt = F_ISSET(vp, VC_C1SET) ? vp->count : 1;
  70
  71         /*
  72          * !!!
  73          * If in white-space, the next start of sentence counts as one.
  74          * This may not handle "  .  " correctly, but it's real unclear
  75          * what correctly means in that case.
  76          */
  77         if (cs.cs_flags == CS_EMP || cs.cs_flags == 0 && isblank(cs.cs_ch)) {
  78                 if (cs_fblank(sp, &cs))
  79                         return (1);
  80                 if (--cnt == 0) {
  81                         if (vp->m_start.lno != cs.cs_lno ||
  82                             vp->m_start.cno != cs.cs_cno)
  83                                 goto okret;
  84                         return (1);
  85                 }
  86         }
  87
  88         for (state = NONE;;) {
  89                 if (cs_next(sp, &cs))
  90                         return (1);
  91                 if (cs.cs_flags == CS_EOF)
  92                         break;
  93                 if (cs.cs_flags == CS_EOL) {
  94                         if ((state == PERIOD || state == BLANK) && --cnt == 0) {
  95                                 if (cs_next(sp, &cs))
  96                                         return (1);
  97                                 if (cs.cs_flags == 0 &&
  98                                     isblank(cs.cs_ch) && cs_fblank(sp, &cs))
  99                                         return (1);
 100                                 goto okret;
 101                         }
 102                         state = NONE;
 103                         continue;
 104                 }
 105                 if (cs.cs_flags == CS_EMP) {    /* An EMP is two sentences. */
 106                         if (--cnt == 0)
 107                                 goto okret;
 108                         if (cs_fblank(sp, &cs))
 109                                 return (1);
 110                         if (--cnt == 0)
 111                                 goto okret;
 112                         state = NONE;
 113                         continue;
 114                 }
 115                 switch (cs.cs_ch) {
 116                 case '.':
 117                 case '?':
 118                 case '!':
 119                         state = PERIOD;
 120                         break;
 121                 case ')':
 122                 case ']':
 123                 case '"':
 124                 case '\'':
 125                         if (state != PERIOD)
 126                                 state = NONE;
 127                         break;
 128                 case '\t':
 129                         if (state == PERIOD)
 130                                 state = BLANK;
 131                         /* FALLTHROUGH */
 132                 case ' ':
 133                         if (state == PERIOD) {
 134                                 state = BLANK;
 135                                 break;
 136                         }
 137                         if (state == BLANK && --cnt == 0) {
 138                                 if (cs_fblank(sp, &cs))
 139                                         return (1);
 140                                 goto okret;
 141                         }
 142                         /* FALLTHROUGH */
 143                 default:
 144                         state = NONE;
 145                         break;
 146                 }
 147         }
 148
 149         /* EOF is a movement sink, but it's an error not to have moved. */
 150         if (vp->m_start.lno == cs.cs_lno && vp->m_start.cno == cs.cs_cno) {
 151                 v_eof(sp, NULL);
 152                 return (1);
 153         }
 154
 155 okret:  vp->m_stop.lno = cs.cs_lno;
 156         vp->m_stop.cno = cs.cs_cno;
 157
 158         /*
 159          * !!!
 160          * Historic, uh, features, yeah, that's right, call 'em features.
 161          * If the starting and ending cursor positions are at the first
 162          * column in their lines, i.e. the movement is cutting entire lines,
 163          * the buffer is in line mode, and the ending position is the last
 164          * character of the previous line.  Note check to make sure that
 165          * it's not within a single line.
 166          *
 167          * Non-motion commands move to the end of the range.  Delete and
 168          * yank stay at the start.  Ignore others.  Adjust the end of the
 169          * range for motion commands.
 170          */
 171         if (ISMOTION(vp)) {
 172                 if (vp->m_start.cno == 0 &&
 173                     (cs.cs_flags != 0 || vp->m_stop.cno == 0)) {
 174                         if (vp->m_start.lno < vp->m_stop.lno) {
 175                                 if (db_get(sp,
 176                                     --vp->m_stop.lno, DBG_FATAL, NULL, &len))
 177                                         return (1);
 178                                 vp->m_stop.cno = len ? len - 1 : 0;
 179                         }
 180                         F_SET(vp, VM_LMODE);
 181                 } else
 182                         --vp->m_stop.cno;
 183                 vp->m_final = vp->m_start;
 184         } else
 185                 vp->m_final = vp->m_stop;
 186         return (0);
 187 }
 188
 189 /*
 190  * v_sentenceb -- [count](
 191  *      Move backward count sentences.
 192  *
 193  * PUBLIC: int v_sentenceb __P((SCR *, VICMD *));
 194  */
 195 int
 196 v_sentenceb(sp, vp)
 197         SCR *sp;
 198         VICMD *vp;
 199 {
 200         VCS cs;
 201         db_recno_t slno;
 202         size_t len, scno;
 203         u_long cnt;
 204         int last;
 205
 206         /*
 207          * !!!
 208          * Historic vi permitted the user to hit SOF repeatedly.
 209          */
 210         if (vp->m_start.lno == 1 && vp->m_start.cno == 0)
 211                 return (0);
 212
 213         cs.cs_lno = vp->m_start.lno;
 214         cs.cs_cno = vp->m_start.cno;
 215         if (cs_init(sp, &cs))
 216                 return (1);
 217
 218         cnt = F_ISSET(vp, VC_C1SET) ? vp->count : 1;
 219
 220         /*
 221          * !!!
 222          * In empty lines, skip to the previous non-white-space character.
 223          * If in text, skip to the prevous white-space character.  Believe
 224          * it or not, in the paragraph:
 225          *      ab cd.
 226          *      AB CD.
 227          * if the cursor is on the 'A' or 'B', ( moves to the 'a'.  If it
 228          * is on the ' ', 'C' or 'D', it moves to the 'A'.  Yes, Virginia,
 229          * Berkeley was once a major center of drug activity.
 230          */
 231         if (cs.cs_flags == CS_EMP) {
 232                 if (cs_bblank(sp, &cs))
 233                         return (1);
 234                 for (;;) {
 235                         if (cs_prev(sp, &cs))
 236                                 return (1);
 237                         if (cs.cs_flags != CS_EOL)
 238                                 break;
 239                 }
 240         } else if (cs.cs_flags == 0 && !isblank(cs.cs_ch))
 241                 for (;;) {
 242                         if (cs_prev(sp, &cs))
 243                                 return (1);
 244                         if (cs.cs_flags != 0 || isblank(cs.cs_ch))
 245                                 break;
 246                 }
 247
 248         for (last = 0;;) {
 249                 if (cs_prev(sp, &cs))
 250                         return (1);
 251                 if (cs.cs_flags == CS_SOF)      /* SOF is a movement sink. */
 252                         break;
 253                 if (cs.cs_flags == CS_EOL) {
 254                         last = 1;
 255                         continue;
 256                 }
 257                 if (cs.cs_flags == CS_EMP) {
 258                         if (--cnt == 0)
 259                                 goto ret;
 260                         if (cs_bblank(sp, &cs))
 261                                 return (1);
 262                         last = 0;
 263                         continue;
 264                 }
 265                 switch (cs.cs_ch) {
 266                 case '.':
 267                 case '?':
 268                 case '!':
 269                         if (!last || --cnt != 0) {
 270                                 last = 0;
 271                                 continue;
 272                         }
 273
 274 ret:                    slno = cs.cs_lno;
 275                         scno = cs.cs_cno;
 276
 277                         /*
 278                          * Move to the start of the sentence, skipping blanks
 279                          * and special characters.
 280                          */
 281                         do {
 282                                 if (cs_next(sp, &cs))
 283                                         return (1);
 284                         } while (!cs.cs_flags &&
 285                             (cs.cs_ch == ')' || cs.cs_ch == ']' ||
 286                             cs.cs_ch == '"' || cs.cs_ch == '\''));
 287                         if ((cs.cs_flags || isblank(cs.cs_ch)) &&
 288                             cs_fblank(sp, &cs))
 289                                 return (1);
 290
 291                         /*
 292                          * If it was ".  xyz", with the cursor on the 'x', or
 293                          * "end.  ", with the cursor in the spaces, or the
 294                          * beginning of a sentence preceded by an empty line,
 295                          * we can end up where we started.  Fix it.
 296                          */
 297                         if (vp->m_start.lno != cs.cs_lno ||
 298                             vp->m_start.cno != cs.cs_cno)
 299                                 goto okret;
 300
 301                         /*
 302                          * Well, if an empty line preceded possible blanks
 303                          * and the sentence, it could be a real sentence.
 304                          */
 305                         for (;;) {
 306                                 if (cs_prev(sp, &cs))
 307                                         return (1);
 308                                 if (cs.cs_flags == CS_EOL)
 309                                         continue;
 310                                 if (cs.cs_flags == 0 && isblank(cs.cs_ch))
 311                                         continue;
 312                                 break;
 313                         }
 314                         if (cs.cs_flags == CS_EMP)
 315                                 goto okret;
 316
 317                         /* But it wasn't; try again. */
 318                         ++cnt;
 319                         cs.cs_lno = slno;
 320                         cs.cs_cno = scno;
 321                         last = 0;
 322                         break;
 323                 case '\t':
 324                         last = 1;
 325                         break;
 326                 default:
 327                         last =
 328                             cs.cs_flags == CS_EOL || isblank(cs.cs_ch) ||
 329                             cs.cs_ch == ')' || cs.cs_ch == ']' ||
 330                             cs.cs_ch == '"' || cs.cs_ch == '\'' ? 1 : 0;
 331                 }
 332         }
 333
 334 okret:  vp->m_stop.lno = cs.cs_lno;
 335         vp->m_stop.cno = cs.cs_cno;
 336
 337         /*
 338          * !!!
 339          * If the starting and stopping cursor positions are at the first
 340          * columns in the line, i.e. the movement is cutting an entire line,
 341          * the buffer is in line mode, and the starting position is the last
 342          * character of the previous line.
 343          *
 344          * All commands move to the end of the range.  Adjust the start of
 345          * the range for motion commands.
 346          */
 347         if (ISMOTION(vp))
 348                 if (vp->m_start.cno == 0 &&
 349                     (cs.cs_flags != 0 || vp->m_stop.cno == 0)) {
 350                         if (db_get(sp,
 351                             --vp->m_start.lno, DBG_FATAL, NULL, &len))
 352                                 return (1);
 353                         vp->m_start.cno = len ? len - 1 : 0;
 354                         F_SET(vp, VM_LMODE);
 355                 } else
 356                         --vp->m_start.cno;
 357         vp->m_final = vp->m_stop;
 358         return (0);
 359 }