split.c

   1 /*
   2  * wiggle - apply rejected patches
   3  *
   4  * Copyright (C) 2003-2013 Neil Brown <neilb@suse.de>
   5  * Copyright (C) 2014-2020 Neil Brown <neil@brown.name>
   6  *
   7  *
   8  *    This program is free software; you can redistribute it and/or modify
   9  *    it under the terms of the GNU General Public License as published by
  10  *    the Free Software Foundation; either version 2 of the License, or
  11  *    (at your option) any later version.
  12  *
  13  *    This program is distributed in the hope that it will be useful,
  14  *    but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  *    GNU General Public License for more details.
  17  *
  18  *    You should have received a copy of the GNU General Public License
  19  *    along with this program.
  20  *
  21  *    Author: Neil Brown
  22  *    Email: <neil@brown.name>
  23  */
  24
  25 /*
  26  * Split a stream into words or lines
  27  *
  28  * A word is one of:
  29  *    string of [A-Za-z0-9_]
  30  *    or string of [ \t]
  31  *    or single char (i.e. punctuation and newlines).
  32  *
  33  * A line is any string that ends with \n
  34  *
  35  * As a special case to allow proper aligning of multiple chunks
  36  * in a patch, a word starting \0 will include 20+ chars with a newline
  37  * second from the end.
  38  *
  39  * We make two passes through the stream.
  40  * Firstly we count the number of item so an array can be allocated,
  41  * then we store start and length of each item in the array
  42  *
  43  */
  44
  45 #include        "wiggle.h"
  46 #include        <stdlib.h>
  47 #include        <ctype.h>
  48 #include        <stdlib.h>
  49
  50 #include "ccan/hash/hash.h"
  51
  52 static int wiggle_split_internal(char *start, char *end, int type,
  53                           struct elmnt *list)
  54 {
  55         int cnt = 0;
  56
  57         while (start < end) {
  58                 char *cp = start;
  59                 char *cp2;
  60                 int prefix = 0;
  61
  62                 if ((type & ByWord) && (type & IgnoreBlanks))
  63                         while (cp < end &&
  64                                (*cp == ' ' || *cp == '\t')) {
  65                                 prefix++;
  66                                 cp++;
  67                         }
  68                 start = cp;
  69
  70                 if (*cp == '\0' && cp+19 < end) {
  71                         /* special word */
  72                         cp += 19;
  73                         cp += strlen(cp) + 1;
  74                 } else
  75                         switch (type & ByMask) {
  76                         case ByLine:
  77                                 while (cp < end && *cp != '\n')
  78                                         cp++;
  79                                 if (cp < end)
  80                                         cp++;
  81                                 break;
  82                         case ByWord:
  83                                 if (*cp == ' ' || *cp == '\t') {
  84                                         do
  85                                                 cp++;
  86                                         while (cp < end
  87                                                && (*cp == ' '
  88                                                    || *cp == '\t'));
  89                                 } else if ((type & WholeWord) ||
  90                                            isalnum(*cp) || *cp == '_') {
  91                                         do
  92                                                 cp++;
  93                                         while (cp < end
  94                                                && (((type & WholeWord)
  95                                                     && *cp != ' ' && *cp != '\t'
  96                                                     && *cp != '\n')
  97                                                    || isalnum(*cp)
  98                                                    || *cp == '_'));
  99                                 } else
 100                                         cp++;
 101                                 break;
 102                         }
 103                 cp2 = cp;
 104                 if ((type & ByWord) && (type & IgnoreBlanks) &&
 105                     *start && *start != '\n')
 106                         while (cp2 < end &&
 107                                (*cp2 == ' ' || *cp2 == '\t' || *cp2 == '\n')) {
 108                                 cp2++;
 109                                 if (cp2[-1] == '\n')
 110                                         break;
 111                         }
 112                 if (list) {
 113                         list->start = start;
 114                         list->len = cp-start;
 115                         list->plen = cp2-start;
 116                         list->prefix = prefix;
 117                         if (*start)
 118                                 list->hash = hash(start, list->len, 0);
 119                         else
 120                                 list->hash = atoi(start+1);
 121                         list++;
 122                 }
 123                 cnt++;
 124                 start = cp2;
 125         }
 126         return cnt;
 127 }
 128
 129 struct file wiggle_split_stream(struct stream s, int type)
 130 {
 131         int cnt;
 132         struct file f;
 133         char *c, *end;
 134
 135         if (!s.body) {
 136                 f.list = NULL;
 137                 f.elcnt = 0;
 138                 return f;
 139         }
 140         end = s.body+s.len;
 141         c = s.body;
 142
 143         cnt = wiggle_split_internal(c, end, type, NULL);
 144         f.list = wiggle_xmalloc(cnt*sizeof(struct elmnt));
 145
 146         f.elcnt = wiggle_split_internal(c, end, type, f.list);
 147         return f;
 148 }