jimp.h 8.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313
  1. #ifndef JIMP_H_
  2. #define JIMP_H_
  3. // TODO: move all diagnostics reporting outside of the library
  4. // So the user has more options on how to report things
  5. typedef enum {
  6. JIMP_INVALID,
  7. JIMP_EOF,
  8. // Puncts
  9. JIMP_OCURLY,
  10. JIMP_CCURLY,
  11. JIMP_OBRACKET,
  12. JIMP_CBRACKET,
  13. JIMP_COMMA,
  14. JIMP_COLON,
  15. // Symbols
  16. JIMP_TRUE,
  17. JIMP_FALSE,
  18. JIMP_NULL,
  19. // Values
  20. JIMP_STRING,
  21. JIMP_NUMBER,
  22. } Jimp_Token;
  23. typedef struct {
  24. const char *file_path;
  25. const char *start;
  26. const char *end;
  27. const char *point;
  28. Jimp_Token token;
  29. const char *token_start;
  30. char *string;
  31. size_t string_count;
  32. size_t string_capacity;
  33. double number;
  34. const char *member;
  35. } Jimp;
  36. // TODO: how do null-s fit into this entire system?
  37. bool jimp_bool(Jimp *jimp, bool *boolean);
  38. bool jimp_number(Jimp *jimp, double *number);
  39. bool jimp_string(Jimp *jimp, const char **string);
  40. bool jimp_object_begin(Jimp *jimp);
  41. bool jimp_object_member(Jimp *jimp);
  42. bool jimp_object_end(Jimp *jimp);
  43. void jimp_unknown_member(Jimp *jimp);
  44. bool jimp_array_begin(Jimp *jimp);
  45. bool jimp_array_item(Jimp *jimp);
  46. bool jimp_array_end(Jimp *jimp);
  47. void jimp_diagf(Jimp *jimp, const char *fmt, ...);
  48. #endif // JIMP_H_
  49. #ifdef JIMP_IMPLEMENTATION
  50. static bool jimp__expect_token(Jimp *jimp, Jimp_Token token);
  51. static bool jimp__get_and_expect_token(Jimp *jimp, Jimp_Token token);
  52. static const char *jimp__token_kind(Jimp_Token token);
  53. static bool jimp__get_token(Jimp *jimp);
  54. static void jimp__skip_whitespaces(Jimp *jimp);
  55. static void jimp__append_to_string(Jimp *jimp, char x);
  56. static void jimp__append_to_string(Jimp *jimp, char x)
  57. {
  58. if (jimp->string_count >= jimp->string_capacity) {
  59. if (jimp->string_capacity == 0) jimp->string_capacity = 1024;
  60. else jimp->string_capacity *= 2;
  61. jimp->string = realloc(jimp->string, jimp->string_capacity);
  62. }
  63. jimp->string[jimp->string_count++] = x;
  64. }
  65. static void jimp__skip_whitespaces(Jimp *jimp)
  66. {
  67. while (jimp->point < jimp->end && isspace(*jimp->point)) {
  68. jimp->point += 1;
  69. }
  70. }
  71. static Jimp_Token jimp__puncts[256] = {
  72. ['{'] = JIMP_OCURLY,
  73. ['}'] = JIMP_CCURLY,
  74. ['['] = JIMP_OBRACKET,
  75. [']'] = JIMP_CBRACKET,
  76. [','] = JIMP_COMMA,
  77. [':'] = JIMP_COLON,
  78. };
  79. static struct {
  80. Jimp_Token token;
  81. const char *symbol;
  82. } jimp__symbols[] = {
  83. { .token = JIMP_TRUE, .symbol = "true" },
  84. { .token = JIMP_FALSE, .symbol = "false" },
  85. { .token = JIMP_NULL, .symbol = "null" },
  86. };
  87. #define jimp__symbols_count (sizeof(jimp__symbols)/sizeof(jimp__symbols[0]))
  88. static bool jimp__get_token(Jimp *jimp)
  89. {
  90. jimp__skip_whitespaces(jimp);
  91. jimp->token_start = jimp->point;
  92. if (jimp->point >= jimp->end) {
  93. jimp->token = JIMP_EOF;
  94. return false;
  95. }
  96. jimp->token = jimp__puncts[(unsigned char)*jimp->point];
  97. if (jimp->token) {
  98. jimp->point += 1;
  99. return true;
  100. }
  101. for (size_t i = 0; i < jimp__symbols_count; ++i) {
  102. const char *symbol = jimp__symbols[i].symbol;
  103. if (*symbol == *jimp->point) {
  104. while (*symbol && jimp->point < jimp->end && *symbol++ == *jimp->point++) {}
  105. if (*symbol) {
  106. jimp->token = JIMP_INVALID;
  107. jimp_diagf(jimp, "ERROR: invalid symbol\n");
  108. return false;
  109. } else {
  110. jimp->token = jimp__symbols[i].token;
  111. return true;
  112. }
  113. }
  114. }
  115. char *endptr = NULL;
  116. jimp->number = strtod(jimp->point, &endptr); // TODO: this implies that jimp->end is a valid address and *jimp->end == 0
  117. if (jimp->point != endptr) {
  118. jimp->point = endptr;
  119. jimp->token = JIMP_NUMBER;
  120. return true;
  121. }
  122. if (*jimp->point == '"') {
  123. jimp->point++;
  124. jimp->string_count = 0;
  125. while (jimp->point < jimp->end) {
  126. // TODO: support all the JSON escape sequences defined in the spec
  127. // Yes, including those dumb suroggate pairs. Spec is spec.
  128. if (*jimp->point == '"') {
  129. jimp->point++;
  130. jimp__append_to_string(jimp, '\0');
  131. jimp->token = JIMP_STRING;
  132. return true;
  133. } else {
  134. char x = *jimp->point++;
  135. jimp__append_to_string(jimp, x);
  136. }
  137. }
  138. jimp->token = JIMP_INVALID;
  139. jimp_diagf(jimp, "ERROR: unfinished string\n");
  140. return false;
  141. }
  142. jimp->token = JIMP_INVALID;
  143. jimp_diagf(jimp, "ERROR: invalid token\n");
  144. return false;
  145. }
  146. void jimp_diagf(Jimp *jimp, const char *fmt, ...)
  147. {
  148. long line_number = 0;
  149. const char *line_start = jimp->start;
  150. const char *point = jimp->start;
  151. while (point < jimp->token_start) {
  152. char x = *point++;
  153. if (x == '\n') {
  154. line_start = point;
  155. line_number += 1;
  156. }
  157. }
  158. fprintf(stderr, "%s:%ld:%ld: ", jimp->file_path, line_number + 1, point - line_start + 1);
  159. va_list args;
  160. va_start(args, fmt);
  161. vfprintf(stderr, fmt, args);
  162. va_end(args);
  163. }
  164. static const char *jimp__token_kind(Jimp_Token token)
  165. {
  166. switch (token) {
  167. case JIMP_EOF: return "end of input";
  168. case JIMP_INVALID: return "invalid";
  169. case JIMP_OCURLY: return "{";
  170. case JIMP_CCURLY: return "}";
  171. case JIMP_OBRACKET: return "[";
  172. case JIMP_CBRACKET: return "]";
  173. case JIMP_COMMA: return ",";
  174. case JIMP_COLON: return ":";
  175. case JIMP_TRUE: return "true";
  176. case JIMP_FALSE: return "false";
  177. case JIMP_NULL: return "null";
  178. case JIMP_STRING: return "string";
  179. case JIMP_NUMBER: return "number";
  180. }
  181. return temp_sprintf("<<UNKNOWN: %u>>", token);
  182. }
  183. bool jimp_array_begin(Jimp *jimp)
  184. {
  185. return jimp__get_and_expect_token(jimp, JIMP_OBRACKET);
  186. }
  187. bool jimp_array_end(Jimp *jimp)
  188. {
  189. return jimp__get_and_expect_token(jimp, JIMP_CBRACKET);
  190. }
  191. bool jimp_array_item(Jimp *jimp)
  192. {
  193. const char *point = jimp->point;
  194. if (!jimp__get_token(jimp)) return false;
  195. if (jimp->token == JIMP_COMMA) return true;
  196. if (jimp->token == JIMP_CBRACKET) {
  197. jimp->point = point;
  198. return false;
  199. }
  200. jimp->point = point;
  201. return true;
  202. }
  203. void jimp_unknown_member(Jimp *jimp)
  204. {
  205. jimp_diagf(jimp, "ERROR: unexpected object member `%s`\n", jimp->member);
  206. }
  207. bool jimp_object_begin(Jimp *jimp)
  208. {
  209. return jimp__get_and_expect_token(jimp, JIMP_OCURLY);
  210. }
  211. bool jimp_object_member(Jimp *jimp)
  212. {
  213. const char *point = jimp->point;
  214. if (!jimp__get_token(jimp)) return false;
  215. if (jimp->token == JIMP_COMMA) {
  216. if (!jimp__get_and_expect_token(jimp, JIMP_STRING)) return false;
  217. jimp->member = strdup(jimp->string); // TODO: memory leak
  218. if (!jimp__get_and_expect_token(jimp, JIMP_COLON)) return false;
  219. return true;
  220. }
  221. if (jimp->token == JIMP_CCURLY) {
  222. jimp->point = point;
  223. return false;
  224. }
  225. if (!jimp__expect_token(jimp, JIMP_STRING)) return false;
  226. jimp->member = strdup(jimp->string); // TODO: memory leak
  227. if (!jimp__get_and_expect_token(jimp, JIMP_COLON)) return false;
  228. return true;
  229. }
  230. bool jimp_object_end(Jimp *jimp)
  231. {
  232. return jimp__get_and_expect_token(jimp, JIMP_CCURLY);
  233. }
  234. bool jimp_string(Jimp *jimp, const char **string)
  235. {
  236. if (!jimp__get_and_expect_token(jimp, JIMP_STRING)) return false;
  237. *string = strdup(jimp->string);
  238. return true;
  239. }
  240. bool jimp_bool(Jimp *jimp, bool *boolean)
  241. {
  242. jimp__get_token(jimp);
  243. if (jimp->token == JIMP_TRUE) {
  244. *boolean = true;
  245. } else if (jimp->token == JIMP_FALSE) {
  246. *boolean = false;
  247. } else {
  248. jimp_diagf(jimp, "ERROR: expected boolean, but got `%s`\n", jimp__token_kind(jimp->token));
  249. return false;
  250. }
  251. return true;
  252. }
  253. bool jimp_number(Jimp *jimp, double *number)
  254. {
  255. if (!jimp__get_and_expect_token(jimp, JIMP_NUMBER)) return false;
  256. *number = jimp->number;
  257. return true;
  258. }
  259. static bool jimp__get_and_expect_token(Jimp *jimp, Jimp_Token token)
  260. {
  261. if (!jimp__get_token(jimp)) return false;
  262. return jimp__expect_token(jimp, token);
  263. }
  264. static bool jimp__expect_token(Jimp *jimp, Jimp_Token token)
  265. {
  266. if (jimp->token != token) {
  267. jimp_diagf(jimp, "ERROR: expected %s, but got %s\n", jimp__token_kind(token), jimp__token_kind(jimp->token));
  268. return false;
  269. }
  270. return true;
  271. }
  272. #endif // JIMP_IMPLEMENTATION