contrib/nvi2/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 #include <sys/types.h>
  13 #include <sys/queue.h>
  14 #include <sys/time.h>
  15
  16 #include <bitstring.h>
  17 #include <ctype.h>
  18 #include <limits.h>
  19 #include <stdio.h>
  20
  21 #include "../common/common.h"
  22 #include "vi.h"
  23
  24 /*
  25  * !!!
  26  * In historic vi, a sentence was delimited by a '.', '?' or '!' character
  27  * followed by TWO spaces or a newline.  One or more empty lines was also
  28  * treated as a separate sentence.  The Berkeley documentation for historical
  29  * vi states that any number of ')', ']', '"' and '\'' characters can be
  30  * between the delimiter character and the spaces or end of line, however,
  31  * the historical implementation did not handle additional '"' characters.
  32  * We follow the documentation here, not the implementation.
  33  *
  34  * Once again, historical vi didn't do sentence movements associated with
  35  * counts consistently, mostly in the presence of lines containing only
  36  * white-space characters.
  37  *
  38  * This implementation also permits a single tab to delimit sentences, and
  39  * treats lines containing only white-space characters as empty lines.
  40  * Finally, tabs are eaten (along with spaces) when skipping to the start
  41  * of the text following a "sentence".
  42  */
  43
  44 /*
  45  * v_sentencef -- [count])
  46  *      Move forward count sentences.
  47  *
  48  * PUBLIC: int v_sentencef(SCR *, VICMD *);
  49  */
  50 int
  51 v_sentencef(SCR *sp, VICMD *vp)
  52 {
  53         enum { BLANK, NONE, PERIOD } state;
  54         VCS cs;
  55         size_t len;
  56         u_long cnt;
  57
  58         cs.cs_lno = vp->m_start.lno;
  59         cs.cs_cno = vp->m_start.cno;
  60         if (cs_init(sp, &cs))
  61                 return (1);
  62
  63         cnt = F_ISSET(vp, VC_C1SET) ? vp->count : 1;
  64
  65         /*
  66          * !!!
  67          * If in white-space, the next start of sentence counts as one.
  68          * This may not handle "  .  " correctly, but it's real unclear
  69          * what correctly means in that case.
  70          */
  71         if (cs.cs_flags == CS_EMP || (cs.cs_flags == 0 && isblank(cs.cs_ch))) {
  72                 if (cs_fblank(sp, &cs))
  73                         return (1);
  74                 if (--cnt == 0) {
  75                         if (vp->m_start.lno != cs.cs_lno ||
  76                             vp->m_start.cno != cs.cs_cno)
  77                                 goto okret;
  78                         return (1);
  79                 }
  80         }
  81
  82         for (state = NONE;;) {
  83                 if (cs_next(sp, &cs))
  84                         return (1);
  85                 if (cs.cs_flags == CS_EOF)
  86                         break;
  87                 if (cs.cs_flags == CS_EOL) {
  88                         if ((state == PERIOD || state == BLANK) && --cnt == 0) {
  89                                 if (cs_next(sp, &cs))
  90                                         return (1);
  91                                 if (cs.cs_flags == 0 &&
  92                                     isblank(cs.cs_ch) && cs_fblank(sp, &cs))
  93                                         return (1);
  94                                 goto okret;
  95                         }
  96                         state = NONE;
  97                         continue;
  98                 }
  99                 if (cs.cs_flags == CS_EMP) {    /* An EMP is two sentences. */
 100                         if (--cnt == 0)
 101                                 goto okret;
 102                         if (cs_fblank(sp, &cs))
 103                                 return (1);
 104                         if (--cnt == 0)
 105                                 goto okret;
 106                         state = NONE;
 107                         continue;
 108                 }
 109                 switch (cs.cs_ch) {
 110                 case '.':
 111                 case '?':
 112                 case '!':
 113                         state = PERIOD;
 114                         break;
 115                 case ')':
 116                 case ']':
 117                 case '"':
 118                 case '\'':
 119                         if (state != PERIOD)
 120                                 state = NONE;
 121                         break;
 122                 case '\t':
 123                         if (state == PERIOD)
 124                                 state = BLANK;
 125                         /* FALLTHROUGH */
 126                 case ' ':
 127                         if (state == PERIOD) {
 128                                 state = BLANK;
 129                                 break;
 130                         }
 131                         if (state == BLANK && --cnt == 0) {
 132                                 if (cs_fblank(sp, &cs))
 133                                         return (1);
 134                                 goto okret;
 135                         }
 136                         /* FALLTHROUGH */
 137                 default:
 138                         state = NONE;
 139                         break;
 140                 }
 141         }
 142
 143         /* EOF is a movement sink, but it's an error not to have moved. */
 144         if (vp->m_start.lno == cs.cs_lno && vp->m_start.cno == cs.cs_cno) {
 145                 v_eof(sp, NULL);
 146                 return (1);
 147         }
 148
 149 okret:  vp->m_stop.lno = cs.cs_lno;
 150         vp->m_stop.cno = cs.cs_cno;
 151
 152         /*
 153          * !!!
 154          * Historic, uh, features, yeah, that's right, call 'em features.
 155          * If the starting and ending cursor positions are at the first
 156          * column in their lines, i.e. the movement is cutting entire lines,
 157          * the buffer is in line mode, and the ending position is the last
 158          * character of the previous line.  Note check to make sure that
 159          * it's not within a single line.
 160          *
 161          * Non-motion commands move to the end of the range.  Delete and
 162          * yank stay at the start.  Ignore others.  Adjust the end of the
 163          * range for motion commands.
 164          */
 165         if (ISMOTION(vp)) {
 166                 if (vp->m_start.cno == 0 &&
 167                     (cs.cs_flags != 0 || vp->m_stop.cno == 0)) {
 168                         if (vp->m_start.lno < vp->m_stop.lno) {
 169                                 if (db_get(sp,
 170                                     --vp->m_stop.lno, DBG_FATAL, NULL, &len))
 171                                         return (1);
 172                                 vp->m_stop.cno = len ? len - 1 : 0;
 173                         }
 174                         F_SET(vp, VM_LMODE);
 175                 } else
 176                         --vp->m_stop.cno;
 177                 vp->m_final = vp->m_start;
 178         } else
 179                 vp->m_final = vp->m_stop;
 180         return (0);
 181 }
 182
 183 /*
 184  * v_sentenceb -- [count](
 185  *      Move backward count sentences.
 186  *
 187  * PUBLIC: int v_sentenceb(SCR *, VICMD *);
 188  */
 189 int
 190 v_sentenceb(SCR *sp, VICMD *vp)
 191 {
 192         VCS cs;
 193         recno_t slno;
 194         size_t len, scno;
 195         u_long cnt;
 196         int last;
 197
 198         /*
 199          * !!!
 200          * Historic vi permitted the user to hit SOF repeatedly.
 201          */
 202         if (vp->m_start.lno == 1 && vp->m_start.cno == 0)
 203                 return (0);
 204
 205         cs.cs_lno = vp->m_start.lno;
 206         cs.cs_cno = vp->m_start.cno;
 207         if (cs_init(sp, &cs))
 208                 return (1);
 209
 210         cnt = F_ISSET(vp, VC_C1SET) ? vp->count : 1;
 211
 212         /*
 213          * !!!
 214          * In empty lines, skip to the previous non-white-space character.
 215          * If in text, skip to the prevous white-space character.  Believe
 216          * it or not, in the paragraph:
 217          *      ab cd.
 218          *      AB CD.
 219          * if the cursor is on the 'A' or 'B', ( moves to the 'a'.  If it
 220          * is on the ' ', 'C' or 'D', it moves to the 'A'.  Yes, Virginia,
 221          * Berkeley was once a major center of drug activity.
 222          */
 223         if (cs.cs_flags == CS_EMP) {
 224                 if (cs_bblank(sp, &cs))
 225                         return (1);
 226                 for (;;) {
 227                         if (cs_prev(sp, &cs))
 228                                 return (1);
 229                         if (cs.cs_flags != CS_EOL)
 230                                 break;
 231                 }
 232         } else if (cs.cs_flags == 0 && !isblank(cs.cs_ch))
 233                 for (;;) {
 234                         if (cs_prev(sp, &cs))
 235                                 return (1);
 236                         if (cs.cs_flags != 0 || isblank(cs.cs_ch))
 237                                 break;
 238                 }
 239
 240         for (last = 0;;) {
 241                 if (cs_prev(sp, &cs))
 242                         return (1);
 243                 if (cs.cs_flags == CS_SOF)      /* SOF is a movement sink. */
 244                         break;
 245                 if (cs.cs_flags == CS_EOL) {
 246                         last = 1;
 247                         continue;
 248                 }
 249                 if (cs.cs_flags == CS_EMP) {
 250                         if (--cnt == 0)
 251                                 goto ret;
 252                         if (cs_bblank(sp, &cs))
 253                                 return (1);
 254                         last = 0;
 255                         continue;
 256                 }
 257                 switch (cs.cs_ch) {
 258                 case '.':
 259                 case '?':
 260                 case '!':
 261                         if (!last || --cnt != 0) {
 262                                 last = 0;
 263                                 continue;
 264                         }
 265
 266 ret:                    slno = cs.cs_lno;
 267                         scno = cs.cs_cno;
 268
 269                         /*
 270                          * Move to the start of the sentence, skipping blanks
 271                          * and special characters.
 272                          */
 273                         do {
 274                                 if (cs_next(sp, &cs))
 275                                         return (1);
 276                         } while (!cs.cs_flags &&
 277                             (cs.cs_ch == ')' || cs.cs_ch == ']' ||
 278                             cs.cs_ch == '"' || cs.cs_ch == '\''));
 279                         if ((cs.cs_flags || isblank(cs.cs_ch)) &&
 280                             cs_fblank(sp, &cs))
 281                                 return (1);
 282
 283                         /*
 284                          * If it was ".  xyz", with the cursor on the 'x', or
 285                          * "end.  ", with the cursor in the spaces, or the
 286                          * beginning of a sentence preceded by an empty line,
 287                          * we can end up where we started.  Fix it.
 288                          */
 289                         if (vp->m_start.lno != cs.cs_lno ||
 290                             vp->m_start.cno > cs.cs_cno)
 291                                 goto okret;
 292
 293                         /*
 294                          * Well, if an empty line preceded possible blanks
 295                          * and the sentence, it could be a real sentence.
 296                          */
 297                         for (;;) {
 298                                 if (cs_prev(sp, &cs))
 299                                         return (1);
 300                                 if (cs.cs_flags == CS_EOL)
 301                                         continue;
 302                                 if (cs.cs_flags == 0 && isblank(cs.cs_ch))
 303                                         continue;
 304                                 break;
 305                         }
 306                         if (cs.cs_flags == CS_EMP)
 307                                 goto okret;
 308
 309                         /* But it wasn't; try again. */
 310                         ++cnt;
 311                         cs.cs_lno = slno;
 312                         cs.cs_cno = scno;
 313                         last = 0;
 314                         break;
 315                 case '\t':
 316                         last = 1;
 317                         break;
 318                 default:
 319                         last =
 320                             cs.cs_flags == CS_EOL || isblank(cs.cs_ch) ||
 321                             cs.cs_ch == ')' || cs.cs_ch == ']' ||
 322                             cs.cs_ch == '"' || cs.cs_ch == '\'' ? 1 : 0;
 323                 }
 324         }
 325
 326 okret:  vp->m_stop.lno = cs.cs_lno;
 327         vp->m_stop.cno = cs.cs_cno;
 328
 329         /*
 330          * !!!
 331          * If the starting and stopping cursor positions are at the first
 332          * columns in the line, i.e. the movement is cutting an entire line,
 333          * the buffer is in line mode, and the starting position is the last
 334          * character of the previous line.
 335          *
 336          * All commands move to the end of the range.  Adjust the start of
 337          * the range for motion commands.
 338          */
 339         if (ISMOTION(vp))
 340                 if (vp->m_start.cno == 0 &&
 341                     (cs.cs_flags != 0 || vp->m_stop.cno == 0)) {
 342                         if (db_get(sp,
 343                             --vp->m_start.lno, DBG_FATAL, NULL, &len))
 344                                 return (1);
 345                         vp->m_start.cno = len ? len - 1 : 0;
 346                         F_SET(vp, VM_LMODE);
 347                 } else
 348                         --vp->m_start.cno;
 349         vp->m_final = vp->m_stop;
 350         return (0);
 351 }