str.c 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345
  1. #include "pocketpy/common/str.h"
  2. #include "pocketpy/common/sstream.h"
  3. #include "pocketpy/common/utils.h"
  4. #include <assert.h>
  5. #include <string.h>
  6. #include <stdlib.h>
  7. #include <ctype.h>
  8. #include <stdio.h>
  9. c11_string* c11_string__new(const char* data) { return c11_string__new2(data, strlen(data)); }
  10. c11_string* c11_string__new2(const char* data, int size) {
  11. c11_string* retval = malloc(sizeof(c11_string) + size + 1);
  12. c11_string__ctor2(retval, data, size);
  13. return retval;
  14. }
  15. void c11_string__ctor(c11_string* self, const char* data) {
  16. c11_string__ctor2(self, data, strlen(data));
  17. }
  18. void c11_string__ctor2(c11_string* self, const char* data, int size) {
  19. self->size = size;
  20. char* p = (char*)self->data;
  21. memcpy(p, data, size);
  22. p[size] = '\0';
  23. }
  24. c11_string* c11_string__copy(c11_string* self) {
  25. int total_size = sizeof(c11_string) + self->size + 1;
  26. c11_string* retval = malloc(total_size);
  27. memcpy(retval, self, total_size);
  28. return retval;
  29. }
  30. void c11_string__delete(c11_string* self) { free(self); }
  31. c11_sv c11_string__sv(c11_string* self) { return (c11_sv){self->data, self->size}; }
  32. c11_string* c11_sv__replace(c11_sv self, char old, char new_) {
  33. c11_string* retval = c11_string__new2(self.data, self.size);
  34. char* p = (char*)retval->data;
  35. for(int i = 0; i < retval->size; i++) {
  36. if(p[i] == old) p[i] = new_;
  37. }
  38. return retval;
  39. }
  40. c11_string* c11_sv__replace2(c11_sv self, c11_sv old, c11_sv new_){
  41. c11_sbuf buf;
  42. c11_sbuf__ctor(&buf);
  43. int start = 0;
  44. while(true) {
  45. int i = c11_sv__index2(self, old, start);
  46. if(i == -1) break;
  47. c11_sv tmp = c11_sv__slice2(self, start, i);
  48. c11_sbuf__write_sv(&buf, tmp);
  49. c11_sbuf__write_sv(&buf, new_);
  50. start = i + old.size;
  51. }
  52. c11_sv tmp = c11_sv__slice2(self, start, self.size);
  53. c11_sbuf__write_sv(&buf, tmp);
  54. return c11_sbuf__submit(&buf);
  55. }
  56. int c11_string__u8_length(c11_string* self) {
  57. return c11__byte_index_to_unicode(self->data, self->size);
  58. }
  59. c11_sv c11_string__u8_getitem(c11_string* self, int i) {
  60. i = c11__unicode_index_to_byte(self->data, i);
  61. int size = c11__u8_header(self->data[i], false);
  62. return c11_sv__slice2(c11_string__sv(self), i, i + size);
  63. }
  64. c11_string* c11_string__u8_slice(c11_string* self, int start, int stop, int step) {
  65. c11_sbuf ss;
  66. c11_sbuf__ctor(&ss);
  67. assert(step != 0);
  68. for(int i = start; step > 0 ? i < stop : i > stop; i += step) {
  69. c11_sv unicode = c11_string__u8_getitem(self, i);
  70. c11_sbuf__write_sv(&ss, unicode);
  71. }
  72. return c11_sbuf__submit(&ss);
  73. }
  74. /////////////////////////////////////////
  75. c11_sv c11_sv__slice(c11_sv sv, int start) { return c11_sv__slice2(sv, start, sv.size); }
  76. c11_sv c11_sv__slice2(c11_sv sv, int start, int stop) {
  77. if(start < 0) start = 0;
  78. if(stop < start) stop = start;
  79. if(stop > sv.size) stop = sv.size;
  80. return (c11_sv){sv.data + start, stop - start};
  81. }
  82. c11_sv c11_sv__strip(c11_sv sv, bool left, bool right) {
  83. int L = 0;
  84. int R = sv.size;
  85. const char* data = sv.data;
  86. if(left) {
  87. while(L < R && (data[L] == ' ' || data[L] == '\t' || data[L] == '\n' || data[L] == '\r'))
  88. L++;
  89. }
  90. if(right) {
  91. while(L < R && (data[R - 1] == ' ' || data[R - 1] == '\t' || data[R - 1] == '\n' ||
  92. data[R - 1] == '\r'))
  93. R--;
  94. }
  95. return c11_sv__slice2(sv, L, R);
  96. }
  97. int c11_sv__index(c11_sv self, char c) {
  98. for(int i = 0; i < self.size; i++) {
  99. if(self.data[i] == c) return i;
  100. }
  101. return -1;
  102. }
  103. int c11_sv__index2(c11_sv self, c11_sv sub, int start) {
  104. if(sub.size == 0) return start;
  105. int max_end = self.size - sub.size;
  106. for(int i = start; i <= max_end; i++) {
  107. int res = memcmp(self.data + i, sub.data, sub.size);
  108. if(res == 0) return i;
  109. }
  110. return -1;
  111. }
  112. int c11_sv__count(c11_sv self, c11_sv sub) {
  113. if(sub.size == 0) return self.size + 1;
  114. int cnt = 0;
  115. int start = 0;
  116. while(true) {
  117. int i = c11_sv__index2(self, sub, start);
  118. if(i == -1) break;
  119. cnt++;
  120. start = i + sub.size;
  121. }
  122. return cnt;
  123. }
  124. c11_vector /* T=c11_sv */ c11_sv__split(c11_sv self, char sep) {
  125. c11_vector retval;
  126. c11_vector__ctor(&retval, sizeof(c11_sv));
  127. const char* data = self.data;
  128. int i = 0;
  129. for(int j = 0; j < self.size; j++) {
  130. if(data[j] == sep) {
  131. if(j > i) {
  132. c11_sv tmp = {data + i, j - i};
  133. c11_vector__push(c11_sv, &retval, tmp);
  134. }
  135. i = j + 1;
  136. continue;
  137. }
  138. }
  139. if(self.size > i) {
  140. c11_sv tmp = {data + i, self.size - i};
  141. c11_vector__push(c11_sv, &retval, tmp);
  142. }
  143. return retval;
  144. }
  145. c11_vector /* T=c11_sv */ c11_sv__split2(c11_sv self, c11_sv sep) {
  146. c11_vector retval;
  147. c11_vector__ctor(&retval, sizeof(c11_sv));
  148. int start = 0;
  149. const char* data = self.data;
  150. while(true) {
  151. int i = c11_sv__index2(self, sep, start);
  152. if(i == -1) break;
  153. c11_sv tmp = {data + start, i - start};
  154. if(tmp.size != 0) c11_vector__push(c11_sv, &retval, tmp);
  155. start = i + sep.size;
  156. }
  157. c11_sv tmp = {data + start, self.size - start};
  158. if(tmp.size != 0) c11_vector__push(c11_sv, &retval, tmp);
  159. return retval;
  160. }
  161. int c11__unicode_index_to_byte(const char* data, int i) {
  162. int j = 0;
  163. while(i > 0) {
  164. j += c11__u8_header(data[j], false);
  165. i--;
  166. }
  167. return j;
  168. }
  169. int c11__byte_index_to_unicode(const char* data, int n) {
  170. int cnt = 0;
  171. for(int i = 0; i < n; i++) {
  172. if((data[i] & 0xC0) != 0x80) cnt++;
  173. }
  174. return cnt;
  175. }
  176. //////////////
  177. int c11_sv__cmp(c11_sv self, c11_sv other) {
  178. int res = strncmp(self.data, other.data, c11__min(self.size, other.size));
  179. if(res != 0) return res;
  180. return self.size - other.size;
  181. }
  182. int c11_sv__cmp2(c11_sv self, const char* other) {
  183. int size = strlen(other);
  184. int res = strncmp(self.data, other, c11__min(self.size, size));
  185. if(res != 0) return res;
  186. return self.size - size;
  187. }
  188. bool c11__streq(const char* a, const char* b) { return strcmp(a, b) == 0; }
  189. bool c11__sveq(c11_sv a, c11_sv b) {
  190. if(a.size != b.size) return false;
  191. return memcmp(a.data, b.data, a.size) == 0;
  192. }
  193. bool c11__sveq2(c11_sv a, const char* b) {
  194. int size = strlen(b);
  195. if(a.size != size) return false;
  196. return memcmp(a.data, b, size) == 0;
  197. }
  198. // clang-format off
  199. static const int kLoRangeA[] = {170,186,443,448,660,1488,1519,1568,1601,1646,1649,1749,1774,1786,1791,1808,1810,1869,1969,1994,2048,2112,2144,2208,2230,2308,2365,2384,2392,2418,2437,2447,2451,2474,2482,2486,2493,2510,2524,2527,2544,2556,2565,2575,2579,2602,2610,2613,2616,2649,2654,2674,2693,2703,2707,2730,2738,2741,2749,2768,2784,2809,2821,2831,2835,2858,2866,2869,2877,2908,2911,2929,2947,2949,2958,2962,2969,2972,2974,2979,2984,2990,3024,3077,3086,3090,3114,3133,3160,3168,3200,3205,3214,3218,3242,3253,3261,3294,3296,3313,3333,3342,3346,3389,3406,3412,3423,3450,3461,3482,3507,3517,3520,3585,3634,3648,3713,3716,3718,3724,3749,3751,3762,3773,3776,3804,3840,3904,3913,3976,4096,4159,4176,4186,4193,4197,4206,4213,4238,4352,4682,4688,4696,4698,4704,4746,4752,4786,4792,4800,4802,4808,4824,4882,4888,4992,5121,5743,5761,5792,5873,5888,5902,5920,5952,5984,5998,6016,6108,6176,6212,6272,6279,6314,6320,6400,6480,6512,6528,6576,6656,6688,6917,6981,7043,7086,7098,7168,7245,7258,7401,7406,7413,7418,8501,11568,11648,11680,11688,11696,11704,11712,11720,11728,11736,12294,12348,12353,12447,12449,12543,12549,12593,12704,12784,13312,19968,40960,40982,42192,42240,42512,42538,42606,42656,42895,42999,43003,43011,43015,43020,43072,43138,43250,43259,43261,43274,43312,43360,43396,43488,43495,43514,43520,43584,43588,43616,43633,43642,43646,43697,43701,43705,43712,43714,43739,43744,43762,43777,43785,43793,43808,43816,43968,44032,55216,55243,63744,64112,64285,64287,64298,64312,64318,64320,64323,64326,64467,64848,64914,65008,65136,65142,65382,65393,65440,65474,65482,65490,65498,65536,65549,65576,65596,65599,65616,65664,66176,66208,66304,66349,66370,66384,66432,66464,66504,66640,66816,66864,67072,67392,67424,67584,67592,67594,67639,67644,67647,67680,67712,67808,67828,67840,67872,67968,68030,68096,68112,68117,68121,68192,68224,68288,68297,68352,68416,68448,68480,68608,68864,69376,69415,69424,69600,69635,69763,69840,69891,69956,69968,70006,70019,70081,70106,70108,70144,70163,70272,70280,70282,70287,70303,70320,70405,70415,70419,70442,70450,70453,70461,70480,70493,70656,70727,70751,70784,70852,70855,71040,71128,71168,71236,71296,71352,71424,71680,71935,72096,72106,72161,72163,72192,72203,72250,72272,72284,72349,72384,72704,72714,72768,72818,72960,72968,72971,73030,73056,73063,73066,73112,73440,73728,74880,77824,82944,92160,92736,92880,92928,93027,93053,93952,94032,94208,100352,110592,110928,110948,110960,113664,113776,113792,113808,123136,123214,123584,124928,126464,126469,126497,126500,126503,126505,126516,126521,126523,126530,126535,126537,126539,126541,126545,126548,126551,126553,126555,126557,126559,126561,126564,126567,126572,126580,126585,126590,126592,126603,126625,126629,126635,131072,173824,177984,178208,183984,194560};
  200. static const int kLoRangeB[] = {170,186,443,451,660,1514,1522,1599,1610,1647,1747,1749,1775,1788,1791,1808,1839,1957,1969,2026,2069,2136,2154,2228,2237,2361,2365,2384,2401,2432,2444,2448,2472,2480,2482,2489,2493,2510,2525,2529,2545,2556,2570,2576,2600,2608,2611,2614,2617,2652,2654,2676,2701,2705,2728,2736,2739,2745,2749,2768,2785,2809,2828,2832,2856,2864,2867,2873,2877,2909,2913,2929,2947,2954,2960,2965,2970,2972,2975,2980,2986,3001,3024,3084,3088,3112,3129,3133,3162,3169,3200,3212,3216,3240,3251,3257,3261,3294,3297,3314,3340,3344,3386,3389,3406,3414,3425,3455,3478,3505,3515,3517,3526,3632,3635,3653,3714,3716,3722,3747,3749,3760,3763,3773,3780,3807,3840,3911,3948,3980,4138,4159,4181,4189,4193,4198,4208,4225,4238,4680,4685,4694,4696,4701,4744,4749,4784,4789,4798,4800,4805,4822,4880,4885,4954,5007,5740,5759,5786,5866,5880,5900,5905,5937,5969,5996,6000,6067,6108,6210,6264,6276,6312,6314,6389,6430,6509,6516,6571,6601,6678,6740,6963,6987,7072,7087,7141,7203,7247,7287,7404,7411,7414,7418,8504,11623,11670,11686,11694,11702,11710,11718,11726,11734,11742,12294,12348,12438,12447,12538,12543,12591,12686,12730,12799,19893,40943,40980,42124,42231,42507,42527,42539,42606,42725,42895,42999,43009,43013,43018,43042,43123,43187,43255,43259,43262,43301,43334,43388,43442,43492,43503,43518,43560,43586,43595,43631,43638,43642,43695,43697,43702,43709,43712,43714,43740,43754,43762,43782,43790,43798,43814,43822,44002,55203,55238,55291,64109,64217,64285,64296,64310,64316,64318,64321,64324,64433,64829,64911,64967,65019,65140,65276,65391,65437,65470,65479,65487,65495,65500,65547,65574,65594,65597,65613,65629,65786,66204,66256,66335,66368,66377,66421,66461,66499,66511,66717,66855,66915,67382,67413,67431,67589,67592,67637,67640,67644,67669,67702,67742,67826,67829,67861,67897,68023,68031,68096,68115,68119,68149,68220,68252,68295,68324,68405,68437,68466,68497,68680,68899,69404,69415,69445,69622,69687,69807,69864,69926,69956,70002,70006,70066,70084,70106,70108,70161,70187,70278,70280,70285,70301,70312,70366,70412,70416,70440,70448,70451,70457,70461,70480,70497,70708,70730,70751,70831,70853,70855,71086,71131,71215,71236,71338,71352,71450,71723,71935,72103,72144,72161,72163,72192,72242,72250,72272,72329,72349,72440,72712,72750,72768,72847,72966,72969,73008,73030,73061,73064,73097,73112,73458,74649,75075,78894,83526,92728,92766,92909,92975,93047,93071,94026,94032,100343,101106,110878,110930,110951,111355,113770,113788,113800,113817,123180,123214,123627,125124,126467,126495,126498,126500,126503,126514,126519,126521,126523,126530,126535,126537,126539,126543,126546,126548,126551,126553,126555,126557,126559,126562,126564,126570,126578,126583,126588,126590,126601,126619,126627,126633,126651,173782,177972,178205,183969,191456,195101};
  201. // clang-format on
  202. bool c11__is_unicode_Lo_char(int c) {
  203. if(c == 0x1f955) return true;
  204. int index;
  205. c11__lower_bound(const int, kLoRangeA, 476, c, c11__less, &index);
  206. if(index == 476) return false;
  207. if(c == kLoRangeA[index]) return true;
  208. index -= 1;
  209. if(index < 0) return false;
  210. return c >= kLoRangeA[index] && c <= kLoRangeB[index];
  211. }
  212. int c11__u8_header(unsigned char c, bool suppress) {
  213. if((c & 0b10000000) == 0) return 1;
  214. if((c & 0b11100000) == 0b11000000) return 2;
  215. if((c & 0b11110000) == 0b11100000) return 3;
  216. if((c & 0b11111000) == 0b11110000) return 4;
  217. if((c & 0b11111100) == 0b11111000) return 5;
  218. if((c & 0b11111110) == 0b11111100) return 6;
  219. if(!suppress) PK_FATAL_ERROR("invalid utf8 char\n");
  220. return 0;
  221. }
  222. IntParsingResult c11__parse_uint(c11_sv text, int64_t* out, int base) {
  223. *out = 0;
  224. c11_sv prefix = {.data = text.data, .size = c11__min(2, text.size)};
  225. if(base == -1) {
  226. if(c11__sveq2(prefix, "0b"))
  227. base = 2;
  228. else if(c11__sveq2(prefix, "0o"))
  229. base = 8;
  230. else if(c11__sveq2(prefix, "0x"))
  231. base = 16;
  232. else
  233. base = 10;
  234. }
  235. if(base == 10) {
  236. // 10-base 12334
  237. if(text.size == 0) return IntParsing_FAILURE;
  238. for(int i = 0; i < text.size; i++) {
  239. char c = text.data[i];
  240. if(c >= '0' && c <= '9') {
  241. *out = (*out * 10) + (c - '0');
  242. } else {
  243. return IntParsing_FAILURE;
  244. }
  245. }
  246. // "9223372036854775807".__len__() == 19
  247. if(text.size > 19) return IntParsing_OVERFLOW;
  248. return IntParsing_SUCCESS;
  249. } else if(base == 2) {
  250. // 2-base 0b101010
  251. if(c11__sveq2(prefix, "0b")) {
  252. // text.remove_prefix(2);
  253. text = (c11_sv){text.data + 2, text.size - 2};
  254. }
  255. if(text.size == 0) return IntParsing_FAILURE;
  256. for(int i = 0; i < text.size; i++) {
  257. char c = text.data[i];
  258. if(c == '0' || c == '1') {
  259. *out = (*out << 1) | (c - '0');
  260. } else {
  261. return IntParsing_FAILURE;
  262. }
  263. }
  264. // "111111111111111111111111111111111111111111111111111111111111111".__len__() == 63
  265. if(text.size > 63) return IntParsing_OVERFLOW;
  266. return IntParsing_SUCCESS;
  267. } else if(base == 8) {
  268. // 8-base 0o123
  269. if(c11__sveq2(prefix, "0o")) {
  270. // text.remove_prefix(2);
  271. text = (c11_sv){text.data + 2, text.size - 2};
  272. }
  273. if(text.size == 0) return IntParsing_FAILURE;
  274. for(int i = 0; i < text.size; i++) {
  275. char c = text.data[i];
  276. if(c >= '0' && c <= '7') {
  277. *out = (*out << 3) | (c - '0');
  278. } else {
  279. return IntParsing_FAILURE;
  280. }
  281. }
  282. // "777777777777777777777".__len__() == 21
  283. if(text.size > 21) return IntParsing_OVERFLOW;
  284. return IntParsing_SUCCESS;
  285. } else if(base == 16) {
  286. // 16-base 0x123
  287. if(c11__sveq2(prefix, "0x")) {
  288. // text.remove_prefix(2);
  289. text = (c11_sv){text.data + 2, text.size - 2};
  290. }
  291. if(text.size == 0) return IntParsing_FAILURE;
  292. for(int i = 0; i < text.size; i++) {
  293. char c = text.data[i];
  294. if(c >= '0' && c <= '9') {
  295. *out = (*out << 4) | (c - '0');
  296. } else if(c >= 'a' && c <= 'f') {
  297. *out = (*out << 4) | (c - 'a' + 10);
  298. } else if(c >= 'A' && c <= 'F') {
  299. *out = (*out << 4) | (c - 'A' + 10);
  300. } else {
  301. return IntParsing_FAILURE;
  302. }
  303. }
  304. // "7fffffffffffffff".__len__() == 16
  305. if(text.size > 16) return IntParsing_OVERFLOW;
  306. return IntParsing_SUCCESS;
  307. }
  308. return IntParsing_FAILURE;
  309. }