kson.c 6.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253
  1. #include <string.h>
  2. #include <stdlib.h>
  3. #include <stdarg.h>
  4. #include <assert.h>
  5. #include <ctype.h>
  6. #include <stdio.h>
  7. #include "kson.h"
  8. /*************
  9. *** Parse ***
  10. *************/
  11. kson_node_t *kson_parse_core(const char *json, long *_n, int *error, long *parsed_len)
  12. {
  13. long *stack = 0, top = 0, max = 0, n_a = 0, m_a = 0, i, j;
  14. kson_node_t *a = 0, *u;
  15. const char *p, *q;
  16. size_t *tmp;
  17. #define __push_back(y) do { \
  18. if (top == max) { \
  19. max = max? max<<1 : 4; \
  20. stack = (long*)realloc(stack, sizeof(long) * max); \
  21. } \
  22. stack[top++] = (y); \
  23. } while (0)
  24. #define __new_node(z) do { \
  25. if (n_a == m_a) { \
  26. long old_m = m_a; \
  27. m_a = m_a? m_a<<1 : 4; \
  28. a = (kson_node_t*)realloc(a, sizeof(kson_node_t) * m_a); \
  29. memset(a + old_m, 0, sizeof(kson_node_t) * (m_a - old_m)); \
  30. } \
  31. *(z) = &a[n_a++]; \
  32. } while (0)
  33. assert(sizeof(size_t) == sizeof(kson_node_t*));
  34. *error = KSON_OK;
  35. for (p = json; *p; ++p) {
  36. while (*p && isspace(*p)) ++p;
  37. if (*p == 0) break;
  38. if (*p == ',') { // comma is somewhat redundant
  39. } else if (*p == '[' || *p == '{') {
  40. int t = *p == '['? -1 : -2;
  41. if (top < 2 || stack[top-1] != -3) { // unnamed internal node
  42. __push_back(n_a);
  43. __new_node(&u);
  44. __push_back(t);
  45. } else stack[top-1] = t; // named internal node
  46. } else if (*p == ']' || *p == '}') {
  47. long i, start, t = *p == ']'? -1 : -2;
  48. for (i = top - 1; i >= 0 && stack[i] != t; --i);
  49. if (i < 0) { // error: an extra right bracket
  50. *error = KSON_ERR_EXTRA_RIGHT;
  51. break;
  52. }
  53. start = i;
  54. u = &a[stack[start-1]];
  55. u->key = u->v.str;
  56. u->n = top - 1 - start;
  57. u->v.child = (kson_node_t**)malloc(u->n * sizeof(kson_node_t*));
  58. tmp = (size_t*)u->v.child;
  59. for (i = start + 1; i < top; ++i)
  60. tmp[i - start - 1] = stack[i];
  61. u->type = *p == ']'? KSON_TYPE_BRACKET : KSON_TYPE_BRACE;
  62. if ((top = start) == 1) break; // completed one object; remaining characters discarded
  63. } else if (*p == ':') {
  64. if (top == 0 || stack[top-1] == -3) {
  65. *error = KSON_ERR_NO_KEY;
  66. break;
  67. }
  68. __push_back(-3);
  69. } else {
  70. int c = *p;
  71. // get the node to modify
  72. if (top >= 2 && stack[top-1] == -3) { // we have a key:value pair here
  73. --top;
  74. u = &a[stack[top-1]];
  75. u->key = u->v.str; // move old value to key
  76. } else { // don't know if this is a bare value or a key:value pair; keep it as a value for now
  77. __push_back(n_a);
  78. __new_node(&u);
  79. }
  80. // parse string
  81. if (c == '\'' || c == '"') {
  82. for (q = ++p; *q && *q != c; ++q)
  83. if (*q == '\\') ++q;
  84. } else {
  85. for (q = p; *q && *q != ']' && *q != '}' && *q != ',' && *q != ':' && *q != '\n'; ++q)
  86. if (*q == '\\') ++q;
  87. }
  88. u->v.str = (char*)malloc(q - p + 1); strncpy(u->v.str, p, q - p); u->v.str[q-p] = 0; // equivalent to u->v.str=strndup(p, q-p)
  89. u->type = c == '\''? KSON_TYPE_SGL_QUOTE : c == '"'? KSON_TYPE_DBL_QUOTE : KSON_TYPE_NO_QUOTE;
  90. p = c == '\'' || c == '"'? q : q - 1;
  91. }
  92. }
  93. while (*p && isspace(*p)) ++p; // skip trailing blanks
  94. if (parsed_len) *parsed_len = p - json;
  95. if (top != 1) *error = KSON_ERR_EXTRA_LEFT;
  96. for (i = 0; i < n_a; ++i)
  97. for (j = 0, u = &a[i], tmp = (size_t*)u->v.child; j < (long)u->n; ++j)
  98. u->v.child[j] = &a[tmp[j]];
  99. free(stack);
  100. *_n = n_a;
  101. return a;
  102. }
  103. void kson_destroy(kson_t *kson)
  104. {
  105. long i;
  106. if (kson == 0) return;
  107. for (i = 0; i < kson->n_nodes; ++i) {
  108. free(kson->root[i].key); free(kson->root[i].v.str);
  109. }
  110. free(kson->root); free(kson);
  111. }
  112. kson_t *kson_parse(const char *json)
  113. {
  114. kson_t *kson;
  115. int error;
  116. kson = (kson_t*)calloc(1, sizeof(kson_t));
  117. kson->root = kson_parse_core(json, &kson->n_nodes, &error, 0);
  118. if (error) {
  119. kson_destroy(kson);
  120. return 0;
  121. }
  122. return kson;
  123. }
  124. /*************
  125. *** Query ***
  126. *************/
  127. const kson_node_t *kson_by_path(const kson_node_t *p, int depth, ...)
  128. {
  129. va_list ap;
  130. va_start(ap, depth);
  131. while (p && depth > 0) {
  132. if (p->type == KSON_TYPE_BRACE) {
  133. p = kson_by_key(p, va_arg(ap, const char*));
  134. } else if (p->type == KSON_TYPE_BRACKET) {
  135. p = kson_by_index(p, va_arg(ap, long));
  136. } else break;
  137. --depth;
  138. }
  139. va_end(ap);
  140. return p;
  141. }
  142. /**************
  143. *** Fromat ***
  144. **************/
  145. void kson_format_recur(const kson_node_t *p, int depth)
  146. {
  147. long i;
  148. if (p->key) printf("\"%s\":", p->key);
  149. if (p->type == KSON_TYPE_BRACKET || p->type == KSON_TYPE_BRACE) {
  150. putchar(p->type == KSON_TYPE_BRACKET? '[' : '{');
  151. if (p->n) {
  152. putchar('\n'); for (i = 0; i <= depth; ++i) fputs(" ", stdout);
  153. for (i = 0; i < (long)p->n; ++i) {
  154. if (i) {
  155. int i;
  156. putchar(',');
  157. putchar('\n'); for (i = 0; i <= depth; ++i) fputs(" ", stdout);
  158. }
  159. kson_format_recur(p->v.child[i], depth + 1);
  160. }
  161. putchar('\n'); for (i = 0; i < depth; ++i) fputs(" ", stdout);
  162. }
  163. putchar(p->type == KSON_TYPE_BRACKET? ']' : '}');
  164. } else {
  165. if (p->type != KSON_TYPE_NO_QUOTE)
  166. putchar(p->type == KSON_TYPE_SGL_QUOTE? '\'' : '"');
  167. fputs(p->v.str, stdout);
  168. if (p->type != KSON_TYPE_NO_QUOTE)
  169. putchar(p->type == KSON_TYPE_SGL_QUOTE? '\'' : '"');
  170. }
  171. }
  172. void kson_format(const kson_node_t *root)
  173. {
  174. kson_format_recur(root, 0);
  175. putchar('\n');
  176. }
  177. /*********************
  178. *** Main function ***
  179. *********************/
  180. #ifdef KSON_MAIN
  181. #define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
  182. int main(int argc, char *argv[])
  183. {
  184. kson_t *kson = 0;
  185. if (argc > 1) {
  186. FILE *fp;
  187. int len = 0, max = 0, tmp, i;
  188. char *json = 0, buf[0x10000];
  189. if ((fp = fopen(argv[1], "rb")) != 0) {
  190. // read the entire file into a string
  191. while ((tmp = fread(buf, 1, 0x10000, fp)) != 0) {
  192. if (len + tmp + 1 > max) {
  193. max = len + tmp + 1;
  194. kroundup32(max);
  195. json = (char*)realloc(json, max);
  196. }
  197. memcpy(json + len, buf, tmp);
  198. len += tmp;
  199. }
  200. fclose(fp);
  201. // parse
  202. kson = kson_parse(json);
  203. free(json);
  204. if (kson) {
  205. kson_format(kson->root);
  206. if (argc > 2) {
  207. // path finding
  208. const kson_node_t *p = kson->root;
  209. for (i = 2; i < argc && p; ++i) {
  210. if (p->type == KSON_TYPE_BRACKET)
  211. p = kson_by_index(p, atoi(argv[i]));
  212. else if (p->type == KSON_TYPE_BRACE)
  213. p = kson_by_key(p, argv[i]);
  214. else p = 0;
  215. }
  216. if (p) {
  217. if (kson_is_internal(p)) printf("Reached an internal node\n");
  218. else printf("Value: %s\n", p->v.str);
  219. } else printf("Failed to find the slot\n");
  220. }
  221. } else printf("Failed to parse\n");
  222. }
  223. } else {
  224. kson = kson_parse("{'a' : 1,'b':[0,'isn\\'t',true],'d':[{\n\n\n}]}");
  225. if (kson) {
  226. const kson_node_t *p = kson_by_path(kson->root, 2, "b", 1);
  227. if (p) printf("*** %s\n", p->v.str);
  228. else printf("!!! not found\n");
  229. kson_format(kson->root);
  230. } else {
  231. printf("Failed to parse\n");
  232. }
  233. }
  234. kson_destroy(kson);
  235. return 0;
  236. }
  237. #endif