* Creates an array that points into each word in the string and puts \0 terminators * after each word. */
| 24 | * after each word. |
| 25 | */ |
| 26 | static char **ArrayFromWordList(char *wordlist, int *len, bool onlyLineEnds = false) { |
| 27 | int prev = '\n'; |
| 28 | int words = 0; |
| 29 | // For rapid determination of whether a character is a separator, build |
| 30 | // a look up table. |
| 31 | bool wordSeparator[256]; |
| 32 | for (int i=0; i<256; i++) { |
| 33 | wordSeparator[i] = false; |
| 34 | } |
| 35 | wordSeparator['\r'] = true; |
| 36 | wordSeparator['\n'] = true; |
| 37 | if (!onlyLineEnds) { |
| 38 | wordSeparator[' '] = true; |
| 39 | wordSeparator['\t'] = true; |
| 40 | } |
| 41 | for (int j = 0; wordlist[j]; j++) { |
| 42 | int curr = static_cast<unsigned char>(wordlist[j]); |
| 43 | if (!wordSeparator[curr] && wordSeparator[prev]) |
| 44 | words++; |
| 45 | prev = curr; |
| 46 | } |
| 47 | char **keywords = new char *[words + 1]; |
| 48 | words = 0; |
| 49 | prev = '\0'; |
| 50 | size_t slen = strlen(wordlist); |
| 51 | for (size_t k = 0; k < slen; k++) { |
| 52 | if (!wordSeparator[static_cast<unsigned char>(wordlist[k])]) { |
| 53 | if (!prev) { |
| 54 | keywords[words] = &wordlist[k]; |
| 55 | words++; |
| 56 | } |
| 57 | } else { |
| 58 | wordlist[k] = '\0'; |
| 59 | } |
| 60 | prev = wordlist[k]; |
| 61 | } |
| 62 | keywords[words] = &wordlist[slen]; |
| 63 | *len = words; |
| 64 | return keywords; |
| 65 | } |
| 66 | |
| 67 | WordList::WordList(bool onlyLineEnds_) : |
| 68 | words(0), list(0), len(0), onlyLineEnds(onlyLineEnds_) { |