1 #include <string.h> 2 #include <sys/stat.h> 3 #include <cstdio> 4 #include <vector> 5 6 #include <utils/Log.h> 7 8 #include "minikin/Hyphenator.h" 9 10 using minikin::HyphenationType; 11 using minikin::Hyphenator; 12 13 Hyphenator* loadHybFile(const char* fn, int minPrefix, int minSuffix, const char* language) { 14 struct stat statbuf; 15 int status = stat(fn, &statbuf); 16 if (status < 0) { 17 fprintf(stderr, "error opening %s\n", fn); 18 return nullptr; 19 } 20 size_t size = statbuf.st_size; 21 FILE* f = fopen(fn, "rb"); 22 if (f == NULL) { 23 fprintf(stderr, "error opening %s\n", fn); 24 return nullptr; 25 } 26 uint8_t* buf = new uint8_t[size]; 27 size_t read_size = fread(buf, 1, size, f); 28 fclose(f); 29 if (read_size < size) { 30 fprintf(stderr, "error reading %s\n", fn); 31 delete[] buf; 32 return nullptr; 33 } 34 return Hyphenator::loadBinary(buf, minPrefix, minSuffix, language); 35 } 36 37 int main(int argc, char** argv) { 38 Hyphenator* hyph = loadHybFile("/tmp/en.hyb", 2, 3, "en"); // should also be configurable 39 std::vector<HyphenationType> result; 40 std::vector<uint16_t> word; 41 if (argc < 2) { 42 fprintf(stderr, "usage: hyphtool word\n"); 43 return 1; 44 } 45 char* asciiword = argv[1]; 46 size_t len = strlen(asciiword); 47 for (size_t i = 0; i < len; i++) { 48 uint32_t c = asciiword[i]; 49 if (c == '-') { 50 c = 0x00AD; 51 } 52 // ASCII (or possibly ISO Latin 1), but kinda painful to do utf conversion :( 53 word.push_back(c); 54 } 55 hyph->hyphenate(word, &result); 56 for (size_t i = 0; i < len; i++) { 57 if (result[i] != HyphenationType::DONT_BREAK) { 58 printf("-"); 59 } 60 printf("%c", word[i]); 61 } 62 printf("\n"); 63 return 0; 64 } 65
