/parsley.c

http://github.com/fizx/parsley · C · 793 lines · 681 code · 90 blank · 22 comment · 275 complexity · 6b7359155add83edb50d0657e01ab025 MD5 · raw file

  1. #include <stdio.h>
  2. #include <stdlib.h>
  3. #include <argp.h>
  4. #include <stdarg.h>
  5. #include <json/json.h>
  6. #include "parsley.h"
  7. #include "y.tab.h"
  8. #include <json/printbuf.h>
  9. #include "functions.h"
  10. #include "util.h"
  11. #include <string.h>
  12. #include <errno.h>
  13. #include <ctype.h>
  14. #include <stdbool.h>
  15. #include <libxslt/xslt.h>
  16. #include <libxslt/xsltInternals.h>
  17. #include <libxslt/transform.h>
  18. #include <libxml/tree.h>
  19. #include <libxml/parser.h>
  20. #include <libxml/debugXML.h>
  21. #include <libxml/HTMLparser.h>
  22. #include <libxml/HTMLtree.h>
  23. #include <libxml/xmlwriter.h>
  24. #include <libxml/xmlerror.h>
  25. #include <libxml/debugXML.h>
  26. #include <libexslt/exslt.h>
  27. int yywrap(void){
  28. return 1;
  29. }
  30. struct ll {
  31. xmlChar *name;
  32. struct ll *next;
  33. };
  34. static char*
  35. arepl(char* orig, char* old, char* new) {
  36. // printf("y\n");
  37. char* ptr = strdup(orig);
  38. int nlen = strlen(new);
  39. int olen = strlen(old);
  40. char* i;
  41. struct printbuf * buf = printbuf_new();
  42. while((i = strstr(ptr, old)) != NULL) {
  43. printbuf_memappend(buf, ptr, i - ptr);
  44. printbuf_memappend(buf, new, nlen);
  45. ptr = i + olen;
  46. }
  47. printbuf_memappend(buf, ptr, strlen(ptr));
  48. ptr = strdup(buf->buf);
  49. printbuf_free(buf);
  50. return ptr;
  51. }
  52. static char *
  53. full_key_name(contextPtr c) {
  54. if(c == NULL || c->parent == NULL) return strdup("/");
  55. static struct ll * last = NULL;
  56. while(c->parent != NULL) {
  57. if(c->tag != NULL) {
  58. struct ll * ptr = calloc(sizeof(struct ll), 1);
  59. ptr->name = c->tag;
  60. ptr->next = last;
  61. last = ptr;
  62. }
  63. c = c->parent;
  64. }
  65. struct printbuf *buf = printbuf_new();
  66. while(last != NULL) {
  67. sprintbuf(buf, "/%s", last->name);
  68. last = last->next;
  69. }
  70. char *out = strdup(buf->buf);
  71. printbuf_free(buf);
  72. return out;
  73. }
  74. void parsed_parsley_free(parsedParsleyPtr ptr) {
  75. if(ptr->xml != NULL) xmlFree(ptr->xml);
  76. if(ptr->error != NULL) free(ptr->error);
  77. free(ptr);
  78. }
  79. static parsedParsleyPtr parse_error(char* format, ...) {
  80. parsedParsleyPtr ptr = (parsedParsleyPtr) calloc(sizeof(parsed_parsley), 1);
  81. ptr->xml = NULL;
  82. va_list args;
  83. va_start(args, format);
  84. vasprintf(&ptr->error, format, args);
  85. va_end(args);
  86. return ptr;
  87. }
  88. parsedParsleyPtr parsley_parse_file(parsleyPtr parsley, char* file, int flags) {
  89. xmlSetGenericErrorFunc(NULL , parsleyXsltError);
  90. bool html = flags & PARSLEY_OPTIONS_HTML;
  91. char * encoding = flags & PARSLEY_OPTIONS_FORCE_UTF8 ? "UTF-8" : NULL;
  92. if(html) {
  93. htmlParserCtxtPtr htmlCtxt = htmlNewParserCtxt();
  94. htmlDocPtr html = htmlCtxtReadFile(htmlCtxt, file, encoding, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
  95. htmlFreeParserCtxt(htmlCtxt);
  96. if(html == NULL) return parse_error("Couldn't parse file: %s\n", file);
  97. parsedParsleyPtr out = parsley_parse_doc(parsley, html, flags);
  98. xmlFreeDoc(html);
  99. return out;
  100. } else {
  101. xmlParserCtxtPtr ctxt = xmlNewParserCtxt();
  102. xmlDocPtr xml = xmlCtxtReadFile(ctxt, file, encoding, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
  103. xmlFreeParserCtxt(ctxt);
  104. if(xml == NULL) return parse_error("Couldn't parse file: %s\n", file);
  105. parsedParsleyPtr out = parsley_parse_doc(parsley, xml, flags);
  106. xmlFreeDoc(xml);
  107. return out;
  108. }
  109. }
  110. parsedParsleyPtr parsley_parse_string(parsleyPtr parsley, char* string, size_t size, char* base_uri, int flags) {
  111. xmlSetGenericErrorFunc(NULL , parsleyXsltError);
  112. bool html = flags & PARSLEY_OPTIONS_HTML;
  113. char * encoding = flags & PARSLEY_OPTIONS_FORCE_UTF8 ? "UTF-8" : NULL;
  114. if(base_uri == NULL) base_uri = "http://parselets.com/in-memory-string";
  115. if(html) {
  116. htmlParserCtxtPtr htmlCtxt = htmlNewParserCtxt();
  117. htmlDocPtr html = htmlCtxtReadMemory(htmlCtxt, string, size, base_uri, encoding, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
  118. if(html == NULL) return parse_error("Couldn't parse string");
  119. parsedParsleyPtr out = parsley_parse_doc(parsley, html, flags);
  120. xmlFreeDoc(html);
  121. return out;
  122. } else {
  123. xmlParserCtxtPtr ctxt = xmlNewParserCtxt();
  124. xmlDocPtr xml = xmlCtxtReadMemory(ctxt, string, size, base_uri, encoding, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
  125. if(xml == NULL) return parse_error("Couldn't parse string");
  126. parsedParsleyPtr out = parsley_parse_doc(parsley, xml, flags);
  127. xmlFreeDoc(xml);
  128. return out;
  129. }
  130. }
  131. static char *
  132. xpath_of(xmlNodePtr node) {
  133. if(node == NULL || node->name == NULL || node->parent == NULL) return strdup("/");
  134. struct ll * ptr = (struct ll *) calloc(sizeof(struct ll), 1);
  135. while(node->name != NULL && node->parent != NULL) {
  136. if(node->ns == NULL) {
  137. struct ll * tmp = (struct ll *) calloc(sizeof(struct ll), 1);
  138. tmp->name = node->name;
  139. tmp->next = ptr;
  140. ptr = tmp;
  141. }
  142. node = node->parent;
  143. }
  144. struct printbuf *buf = printbuf_new();
  145. sprintbuf(buf, "");
  146. while(ptr->name != NULL) {
  147. sprintbuf(buf, "/%s", ptr->name);
  148. struct ll * last = ptr;
  149. ptr = ptr->next;
  150. free(last);
  151. }
  152. free(ptr);
  153. char *str = strdup(strlen(buf->buf) ? buf->buf : "/");
  154. printbuf_free(buf);
  155. return str;
  156. }
  157. static bool
  158. is_root(xmlNodePtr node) {
  159. return node != NULL && node->ns != NULL && !strcmp(node->ns->prefix, "parsley") && !strcmp(node->name, "root");
  160. }
  161. int compare_pos (const void * a, const void * b)
  162. {
  163. char* as = xmlGetProp(*(xmlNodePtr*)a, "position");
  164. char* bs = xmlGetProp(*(xmlNodePtr*)b, "position");
  165. return atoi(as) - atoi(bs);
  166. }
  167. static void
  168. _xmlAddChild(xmlNodePtr parent, xmlNodePtr child) {
  169. xmlNodePtr node = parent->children;
  170. if(node == NULL) {
  171. parent->children = child;
  172. return;
  173. }
  174. while(node->next != NULL){
  175. node = node->next;
  176. }
  177. node->next = child;
  178. }
  179. static int
  180. _xmlChildElementCount(xmlNodePtr n) {
  181. xmlNodePtr child = n->children;
  182. int i = 0;
  183. while(child != NULL) {
  184. i++;
  185. child = child->next;
  186. }
  187. return i;
  188. }
  189. static bool
  190. xml_empty(xmlNodePtr xml) {
  191. // fprintf(stderr, "%s\n", xml->name);
  192. xmlNodePtr child = xml->children;
  193. while(child != NULL) {
  194. if(child->type != XML_TEXT_NODE) return false;
  195. if(strlen(child->content)) return false;
  196. child = child->next;
  197. }
  198. // printf("hello!\n");
  199. return true;
  200. }
  201. static void
  202. collate(xmlNodePtr xml) {
  203. if(xml == NULL) return ;
  204. if(xml->type != XML_ELEMENT_NODE) return;
  205. if(xml->ns != NULL && !strcmp(xml->ns->prefix, "parsley") && !strcmp(xml->name, "zipped")){
  206. xmlNodePtr parent = xml->parent;
  207. xmlNodePtr child = xml->children;
  208. if(child == NULL) return;
  209. int n = _xmlChildElementCount(xml);
  210. xmlNodePtr* name_nodes = calloc(n, sizeof(xmlNodePtr));
  211. xmlNodePtr* lists = calloc(n, sizeof(xmlNodePtr));
  212. bool* empty = calloc(n, sizeof(bool));
  213. bool* multi = calloc(n, sizeof(bool));
  214. bool* optional = calloc(n, sizeof(bool));
  215. int len = 0;
  216. for(int i = 0; i < n; i++) {
  217. name_nodes[i] = child;
  218. if(child->children == NULL) {
  219. lists[i] = NULL;
  220. } else {
  221. lists[i] = child->children;
  222. multi[i] = false;
  223. optional[i] = xmlGetProp(name_nodes[i], "optional") != NULL;
  224. if(lists[i] != NULL && !strcmp(lists[i]->name, "groups")) {
  225. lists[i] = lists[i]->children;
  226. multi[i] = true;
  227. }
  228. if(lists[i] != NULL) {
  229. lists[i]->parent->extra = i;
  230. len += _xmlChildElementCount(lists[i]->parent);
  231. }
  232. child->children = NULL;
  233. }
  234. child = child->next;
  235. }
  236. xml->children = NULL;
  237. xmlNodePtr* sortable = malloc(len * sizeof(xmlNodePtr));
  238. int j = 0;
  239. for(int i = 0; i < n; i++) {
  240. xmlNodePtr node = lists[i];
  241. while(node != NULL){
  242. sortable[j++] = node;
  243. // printf("%d/%d: %d/%d\n", i, n, j, len);
  244. node = node->next;
  245. }
  246. }
  247. for(int i = 0; i < len; i++) {
  248. sortable[i]->next = NULL;
  249. }
  250. qsort(sortable, len, sizeof(xmlNodePtr), compare_pos);
  251. xmlNodePtr groups = xml->parent;
  252. groups->children = NULL;
  253. xmlNodePtr group;
  254. xmlNodePtr* targets = calloc(sizeof(xmlNodePtr), n);
  255. for(j = 0; j < len; j++) {
  256. int i = sortable[j]->parent->extra;
  257. if (j == 0 || (!empty[i] && !multi[i] && !optional[i])) { // first or full
  258. xmlNodePtr group = xmlNewChild(groups, xml->ns, "group", NULL); //new group
  259. xmlSetProp(group, "optional", "true");
  260. for(int k = 0; k < n; k++) {
  261. empty[k] = true;
  262. targets[k] = xmlCopyNode(name_nodes[k], 2);
  263. _xmlAddChild(group, targets[k]);
  264. if(multi[k]) targets[k] = xmlNewChild(targets[k], xml->ns, "groups", NULL);
  265. }
  266. }
  267. if(!multi[i]) sortable[j] = sortable[j]->children;
  268. if(empty[i] || multi[i]) _xmlAddChild(targets[i], sortable[j]);
  269. empty[i] = false;
  270. }
  271. free(targets);
  272. free(name_nodes);
  273. free(lists);
  274. free(optional);
  275. free(empty);
  276. free(multi);
  277. free(sortable);
  278. collate(groups);
  279. } else {
  280. xmlNodePtr child = xml->children;
  281. while(child != NULL){
  282. collate(child);
  283. child = child->next;
  284. }
  285. }
  286. }
  287. void
  288. parsley_set_user_agent(char const * agent) {
  289. _parsley_set_user_agent(agent);
  290. }
  291. static void
  292. unlink(xmlNodePtr parent, xmlNodePtr child) {
  293. if(child == NULL || parent == NULL) return;
  294. xmlNodePtr ptr = parent->children;
  295. if(ptr == child) {
  296. parent->children = child->next;
  297. if(child->next != NULL) {
  298. child->next->prev = NULL;
  299. }
  300. } else {
  301. while(ptr != NULL) {
  302. if(ptr->next == child) {
  303. ptr->next = child->next;
  304. if(child->next) child->next->prev = ptr;
  305. }
  306. ptr = ptr->next;
  307. }
  308. }
  309. child->next = NULL;
  310. child->prev = NULL;
  311. child->parent = NULL;
  312. }
  313. static void
  314. visit(parsedParsleyPtr ptr, xmlNodePtr xml, char* err);
  315. static void
  316. prune(parsedParsleyPtr ptr, xmlNodePtr xml, char* err) {
  317. if(xml == NULL || is_root(xml)) return;
  318. bool optional = xmlGetProp(xml, "optional") != NULL;
  319. if(optional) {
  320. xmlNodePtr parent = xml->parent;
  321. unlink(parent, xml);
  322. free(err);
  323. err = NULL;
  324. visit(ptr, parent, err);
  325. return;
  326. } else {
  327. if(err == NULL) asprintf(&err, "%s was empty", xpath_of(xml));
  328. if(!is_root(xml->parent)) {
  329. // fprintf(stderr, "prune up: %s\n", xml->parent->name);
  330. prune(ptr, xml->parent, err);
  331. } else {
  332. // fprintf(stderr, "error out\n");
  333. ptr->error = err;
  334. }
  335. }
  336. }
  337. static void
  338. visit(parsedParsleyPtr ptr, xmlNodePtr xml, char* err) {
  339. if(xml == NULL) return;
  340. // printf("trying to visit: %s\n", xml->name);
  341. if(xml->type != XML_ELEMENT_NODE) return;
  342. xmlNodePtr child = xml->children;
  343. xmlNodePtr parent = xml->parent;
  344. if(parent == NULL) return;
  345. // printf("passed guard clause: %s\n", xml->name);
  346. if(xml_empty(xml)) {
  347. if(err == NULL) asprintf(&err, "%s was empty", xpath_of(xml));
  348. prune(ptr, xml, err);
  349. } else if(err != NULL) {
  350. free(err);
  351. }
  352. while(err == NULL && child != NULL){
  353. child->parent = xml;
  354. visit(ptr, child, err);
  355. child = child->next;
  356. }
  357. }
  358. static parsedParsleyPtr current_ptr = NULL;
  359. void
  360. parsleyXsltError(void * ctx, const char * msg, ...) {
  361. if(current_ptr == NULL) return;
  362. va_list ap;
  363. va_start(ap, msg);
  364. if(current_ptr->error == NULL) {
  365. char *tmp;
  366. char *tmp2;
  367. vasprintf(&tmp, msg, ap);
  368. tmp2 = arepl(tmp, "xmlXPathCompOpEval: ", "");
  369. current_ptr->error = arepl(tmp2, "\n", "");
  370. free(tmp);
  371. free(tmp2);
  372. }
  373. va_end(ap);
  374. }
  375. static bool
  376. hasDefaultNS(xmlDocPtr doc) {
  377. return xmlSearchNs(doc, doc->children, NULL) != NULL;
  378. }
  379. static void
  380. _killDefaultNS(xmlNodePtr node) {
  381. if(node == NULL) return;
  382. xmlNsPtr ns = node->nsDef;
  383. if(ns != NULL) {
  384. if(ns->prefix == NULL) node->nsDef = ns->next;
  385. xmlNsPtr prev = ns;
  386. while(ns = ns->next) {
  387. if(ns->prefix == NULL) prev->next = ns->next;
  388. }
  389. }
  390. ns = node->ns;
  391. if(ns != NULL) {
  392. if(ns->prefix == NULL) node->ns = ns->next;
  393. xmlNsPtr prev = ns;
  394. while(ns = ns->next) {
  395. if(ns->prefix == NULL) prev->next = ns->next;
  396. }
  397. }
  398. _killDefaultNS(node->children);
  399. _killDefaultNS(node->next);
  400. }
  401. void
  402. killDefaultNS(xmlDocPtr doc) {
  403. if(hasDefaultNS(doc)) {
  404. _killDefaultNS(doc->children);
  405. }
  406. }
  407. parsedParsleyPtr parsley_parse_doc(parsleyPtr parsley, xmlDocPtr doc, int flags) {
  408. killDefaultNS(doc);
  409. parsedParsleyPtr ptr = (parsedParsleyPtr) calloc(sizeof(parsed_parsley), 1);
  410. ptr->error = NULL;
  411. ptr->parsley = parsley;
  412. parsley_io_set_mode(flags);
  413. xsltTransformContextPtr ctxt = xsltNewTransformContext(parsley->stylesheet, doc);
  414. xmlSetGenericErrorFunc(ctxt, parsleyXsltError);
  415. current_ptr = ptr;
  416. if(flags & PARSLEY_OPTIONS_SGWRAP) {
  417. doc = parsley_apply_span_wrap(doc);
  418. }
  419. ptr->xml = xsltApplyStylesheetUser(parsley->stylesheet, doc, NULL, NULL, NULL, ctxt);
  420. xsltFreeTransformContext(ctxt);
  421. current_ptr = NULL;
  422. if(ptr->error == NULL) {
  423. if(ptr->xml != NULL && ptr->error == NULL) {
  424. if(flags & PARSLEY_OPTIONS_COLLATE) collate(ptr->xml->children);
  425. if(flags & PARSLEY_OPTIONS_PRUNE) visit(ptr, ptr->xml->children, NULL);
  426. }
  427. if(ptr->xml == NULL && ptr->error == NULL) { // == NULL
  428. ptr->error = strdup("Internal runtime error");
  429. }
  430. }
  431. return ptr;
  432. }
  433. static bool
  434. json_invalid_object(parsleyPtr ptr, struct json_object *json) {
  435. json_object_object_foreach(json, key, val) {
  436. if(val==NULL) ptr->error = strdup("Parselets can only be made up of strings, arrays, and objects.");
  437. switch(json_object_get_type(val)) {
  438. case json_type_string:
  439. break;
  440. case json_type_array:
  441. if(json_object_array_length(val) != 1) {
  442. ptr->error = strdup("Parselet arrays should have length 1.");
  443. return true;
  444. }
  445. struct json_object * inner = json_object_array_get_idx(val, 0);
  446. switch(json_object_get_type(inner)) {
  447. case json_type_string:
  448. break;
  449. case json_type_object:
  450. if(json_invalid_object(ptr, inner)) return true;
  451. break;
  452. default:
  453. ptr->error = strdup("Arrays may contain either a single string or an object.");
  454. return true;
  455. }
  456. break;
  457. case json_type_object:
  458. if(json_invalid_object(ptr, val)) {
  459. return true;
  460. }
  461. break;
  462. default:
  463. ptr->error = strdup("Parselets can only be made up of strings, arrays, and objects.");
  464. }
  465. if(val == NULL || !json_object_is_type(val, json_type_string)) return false;
  466. }
  467. return false;
  468. }
  469. static bool
  470. json_invalid(parsleyPtr ptr, struct json_object *json) {
  471. if(!json_object_is_type(json, json_type_object)) {
  472. ptr->error = strdup("The parselet root must be an object");
  473. return true;
  474. }
  475. return json_invalid_object(ptr, json);
  476. }
  477. static void free_context(contextPtr c) {
  478. if(c == NULL) return;
  479. if(c->tag != NULL) free(c->tag);
  480. if(c->filter != NULL) pxpath_free(c->filter);
  481. if(c->expr != NULL) pxpath_free(c->expr);
  482. if(c->parent != NULL && c->parent->child != NULL) {
  483. if(c->parent->child == c) {
  484. c->parent->child = NULL;
  485. } else {
  486. contextPtr ptr = c->parent->child;
  487. while(ptr->next != NULL) {
  488. if(ptr->next == c) {
  489. ptr->next = NULL;
  490. } else {
  491. ptr = ptr->next;
  492. }
  493. }
  494. }
  495. }
  496. if(c->next != NULL) free_context(c->next);
  497. if(c->child != NULL) free_context(c->child);
  498. free(c);
  499. }
  500. static contextPtr
  501. new_context(struct json_object * json, xmlNodePtr node) {
  502. contextPtr c = calloc(sizeof(parsley_context), 1);
  503. c->node = node;
  504. c->ns = node->ns;
  505. c->tag = strdup("root");
  506. c->expr = pxpath_new_path(1, "/");
  507. c->json = json;
  508. return c;
  509. }
  510. parsleyPtr parsley_compile(char* parsley_str, char* incl) {
  511. parsleyPtr parsley = (parsleyPtr) calloc(sizeof(compiled_parsley), 1);
  512. if(last_parsley_error != NULL) {
  513. free(last_parsley_error);
  514. last_parsley_error = NULL;
  515. }
  516. registerEXSLT();
  517. // struct json_tokener *tok = json_tokener_new();
  518. // struct json_object *json = json_tokener_parse_ex(tok, parsley_str);
  519. //
  520. struct json_tokener *tok = json_tokener_new();
  521. struct json_object *json = json_tokener_parse_ex(tok, parsley_str, -1);
  522. int error_offset = tok->char_offset;
  523. if(tok->err != json_tokener_success)
  524. json = error_ptr(-tok->err);
  525. json_tokener_free(tok);
  526. if(is_error(json)) {
  527. asprintf(&(parsley->error), "Your parselet is not valid json: %s at char:%d", json_tokener_errors[-(unsigned long) json], error_offset);
  528. return parsley;
  529. }
  530. if(json_invalid(parsley, json)) {
  531. // fprintf(stderr, "Invalid parselet structure: %s\n", parsley->error);
  532. return parsley;
  533. }
  534. xmlNodePtr node = new_stylesheet_skeleton(incl);
  535. contextPtr context = new_context(json, node);
  536. __parsley_recurse(context);
  537. json_object_put(json); // frees json
  538. parsley->error = last_parsley_error;
  539. if(parsley->error == NULL) {
  540. parsley->stylesheet = xsltParseStylesheetDoc(node->doc);
  541. }
  542. free_context(context);
  543. return parsley;
  544. }
  545. contextPtr deeper_context(contextPtr context, char* key, struct json_object * val) {
  546. contextPtr c = (contextPtr) calloc(sizeof(parsley_context), 1);
  547. c->node = context->node;
  548. c->ns = context->ns;
  549. c->parent = context;
  550. c->tag = parsley_key_tag(key);
  551. c->flags = parsley_key_flags(key);
  552. parsley_parsing_context = c;
  553. c->array = val != NULL && json_object_is_type(val, json_type_array);
  554. c->json = c->array ? json_object_array_get_idx(val, 0) : val;
  555. c->string = val != NULL && json_object_is_type(c->json, json_type_string);
  556. c->filter = parsley_key_filter(key);
  557. c->magic = context->array && context->filter == NULL;
  558. c->expr = c->string ? myparse(json_object_get_string(c->json)) : NULL;
  559. if(context->child == NULL) {
  560. context->child = c;
  561. } else {
  562. contextPtr tmp = context->child;
  563. while(tmp->next != NULL) tmp = tmp->next;
  564. tmp->next = c;
  565. }
  566. // printf(stderr, "json: %s\ntag: %s\nexpr: %s\nfilter: %s\n\n", json_object_get_string(c->json), c->tag, pxpath_to_string(c->expr), pxpath_to_string(c->filter));
  567. return c;
  568. }
  569. void parsley_free(parsleyPtr ptr) {
  570. if(ptr->error != NULL)
  571. free(ptr->error);
  572. if(ptr->stylesheet != NULL)
  573. xsltFreeStylesheet(ptr->stylesheet);
  574. free(ptr);
  575. }
  576. void yyerror(const char * s) {
  577. struct printbuf *buf = printbuf_new();
  578. if(last_parsley_error !=NULL) sprintbuf(buf, "%s\n", last_parsley_error);
  579. sprintbuf(buf, "%s in key: %s", s, full_key_name(parsley_parsing_context));
  580. last_parsley_error = strdup(buf->buf);
  581. printbuf_free(buf);
  582. }
  583. static bool
  584. all_strings(struct json_object * json) {
  585. json_object_object_foreach(json, key, val) {
  586. if(val == NULL || !json_object_is_type(val, json_type_string)) return false;
  587. }
  588. return true;
  589. }
  590. static char *
  591. inner_path(pxpathPtr p) {
  592. if(p == NULL) return NULL;
  593. if(p->type == PXPATH_PATH) return pxpath_to_string(p);
  594. char *tmp = NULL;
  595. pxpathPtr ptr = p->child;
  596. while(ptr != NULL) {
  597. tmp = inner_path(p->child);
  598. if(tmp != NULL) return tmp;
  599. ptr = ptr->next;
  600. }
  601. return NULL;
  602. }
  603. //TODO: tree-based replace, instead of naive string-based
  604. static char *
  605. inner_path_to_dot(pxpathPtr p) {
  606. char *outer = pxpath_to_string(p);
  607. char *inner = inner_path(p);
  608. char *repl = NULL;
  609. if(inner != NULL) {
  610. repl = arepl(outer, inner, ".");
  611. free(inner);
  612. }
  613. free(outer);
  614. return repl;
  615. }
  616. static bool
  617. inner_path_transform(contextPtr c) {
  618. return c->filter == NULL && c->expr != NULL && inner_path(c->expr) != NULL;
  619. }
  620. static char *
  621. resolve_filter(contextPtr c) {
  622. return inner_path_transform(c) ? inner_path(c->expr) : pxpath_to_string(c->filter);
  623. }
  624. static char *
  625. resolve_expr(contextPtr c) {
  626. return inner_path_transform(c) ? inner_path_to_dot(c->expr) : pxpath_to_string(c->expr);
  627. }
  628. static void
  629. render(contextPtr c) {
  630. char *filter = resolve_filter(c);
  631. char *expr = resolve_expr(c);
  632. char *scope = filter == NULL ? expr : filter;
  633. bool magic_children = c->array && filter == NULL;
  634. bool simple_array = c->array && filter != NULL;
  635. bool filtered = filter != NULL;
  636. bool multiple = (c->array || c->magic) && !magic_children;
  637. // printf("node %s\n", c->node->name);
  638. xmlNsPtr parsley = c->ns;
  639. xmlNsPtr xsl = xmlDocGetRootElement(c->node->doc)->ns;
  640. if(c->array) c->node = xmlNewChild(c->node, parsley, "groups", NULL);
  641. if(filtered) {
  642. c->node = xmlNewChild(c->node, xsl, "for-each", NULL);
  643. xmlSetProp(c->node, "select", filter);
  644. }
  645. if(filtered && !multiple) {
  646. c->node = xmlNewChild(c->node, xsl, "if", NULL);
  647. xmlSetProp(c->node, "test", "position() = 1");
  648. }
  649. if(multiple) {
  650. c->node = xmlNewChild(c->node, parsley, "group", NULL);
  651. if (!(c->flags & PARSLEY_BANG)) xmlSetProp(c->node, "optional", "true");
  652. }
  653. xmlNodePtr attr = xmlNewChild(c->node, xsl, "attribute", NULL);
  654. xmlSetProp(attr, "name", "position");
  655. xmlNodePtr counter = xmlNewChild(attr, xsl, "value-of", NULL);
  656. xmlSetProp(counter, "select", "count(preceding::*) + count(ancestor::*)");
  657. if(c->string) {
  658. c->node = xmlNewChild(c->node, xsl, "value-of", NULL);
  659. xmlSetProp(c->node, "select", expr);
  660. } else {
  661. if(magic_children) c->node = xmlNewChild(c->node, parsley, "zipped", NULL);
  662. __parsley_recurse(c);
  663. }
  664. if(filter !=NULL) free(filter);
  665. if(expr !=NULL) free(expr);
  666. }
  667. void __parsley_recurse(contextPtr context) {
  668. contextPtr c;
  669. if(context->json == NULL) return;
  670. // printf(stderr, "<%s> %s\n", context->tag, context->node->name);
  671. json_object_object_foreach(context->json, key, val) {
  672. c = deeper_context(context, key, val);
  673. // printf(stderr, "<%s>\n", c->tag);
  674. c->node = xmlAddChild(c->node, xmlNewNode(NULL, c->tag));
  675. if (c->flags & PARSLEY_OPTIONAL) xmlSetProp(c->node, "optional", "true");
  676. render(c);
  677. }
  678. }
  679. // static char* full_expr(contextPtr context, char* expr) {
  680. // if(expr == NULL) return context->full_expr;
  681. // char* merged = arepl(expr, ".", context->full_expr);
  682. // return arepl(merged, "///", "//");
  683. // }
  684. static char*
  685. inner_key_each(struct json_object * json);
  686. static char* inner_key_of(struct json_object * json) {
  687. switch(json_object_get_type(json)) {
  688. case json_type_string:
  689. return json_object_get_string(json);
  690. case json_type_array:
  691. return NULL;
  692. case json_type_object:
  693. return inner_key_each(json);
  694. }
  695. }
  696. static char*
  697. inner_key_each(struct json_object * json) {
  698. json_object_object_foreach(json, key, val) {
  699. char* inner = inner_key_of(val);
  700. if(inner != NULL) return inner;
  701. }
  702. return NULL;
  703. }