codec_ttl.c 8.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306
  1. #include "lsup/codec/codec_ttl.h"
  2. /** @brief NT codec iterator.
  3. *
  4. * This iterator yields one or more triples at a time, one group per subject,
  5. * with the most compact form allowed by Turtle, e.g.
  6. *
  7. * :s :p1 :o1, :o2, o3; p2 o4, o5, <http://example.com/ext1> .
  8. */
  9. typedef struct {
  10. const LSUP_Codec * codec; ///< Codec that generated this iterator.
  11. const LSUP_Graph * gr; ///< Graph being encoded.
  12. LSUP_TermSet * subjects; ///< All subjects in the graph.
  13. size_t s_cur; ///< Term set cursor.
  14. LSUP_rc rc; ///< Internal return code.
  15. char * s_str; ///< Serialized subject block (output).
  16. char * p_str; ///< Serialized predicate block.
  17. char * o_str; ///< Serialized object block.
  18. } LSUP_TTLCodecIterator;
  19. /* * * Codec functions. * * */
  20. static LSUP_rc
  21. term_to_ttl (const LSUP_Term *term, char **out_p)
  22. {
  23. LSUP_rc rc;
  24. char *tmp = NULL, *out = NULL;
  25. char *metadata = NULL;
  26. size_t buf_len;
  27. LSUP_rc md_rc = LSUP_NORESULT;
  28. switch (term->type) {
  29. case LSUP_TERM_IRIREF:
  30. if (strcmp (term->data, LSUP_RDF_TYPE) == 0) {
  31. // Shorten RDF type
  32. buf_len = 2;
  33. tmp = "a";
  34. } else {
  35. md_rc = LSUP_nsmap_denormalize_uri (term->data, &tmp);
  36. PRCCK (md_rc);
  37. if (md_rc == LSUP_NORESULT) {
  38. // If URI counld not be shortened, add `<>`
  39. out = realloc (*out_p, strlen (tmp) + 3);
  40. if (UNLIKELY (!out)) return LSUP_MEM_ERR;
  41. sprintf (out, "<%s>", tmp);
  42. free (tmp);
  43. } else {
  44. free (*out_p);
  45. out = tmp;
  46. }
  47. }
  48. rc = LSUP_OK;
  49. break;
  50. case LSUP_TERM_LITERAL:
  51. // Calculate string length.
  52. if (escape_lit (term->data, &tmp) != LSUP_OK)
  53. return LSUP_ERROR;
  54. buf_len = strlen (tmp) + 3; // Room for "" and terminator
  55. // Data type.
  56. bool shorten = false;
  57. if (
  58. term->datatype != 0
  59. && term->datatype != LSUP_default_datatype
  60. ) {
  61. md_rc = LSUP_nsmap_denormalize_uri (
  62. term->datatype->data, &metadata);
  63. PRCCK (md_rc);
  64. unsigned padding = 0;
  65. // Shorten numeric and boolean types.
  66. if (strcmp (metadata, "xsd:integer") == 0) {
  67. // TODO check for valid format.
  68. shorten = true;
  69. } else if (strcmp (metadata, "xsd:double") == 0) {
  70. // TODO check for valid format.
  71. shorten = true;
  72. } else if (strcmp (metadata, "xsd:decimal") == 0) {
  73. // TODO check for valid format.
  74. shorten = true;
  75. } else if (strcmp (metadata, "xsd:boolean") == 0) {
  76. // TODO check for valid format.
  77. shorten = true;
  78. } else {
  79. // Room for `^^<>` for FQURI, `^^` for NS URI
  80. padding = md_rc == LSUP_NORESULT ? 4 : 2;
  81. }
  82. buf_len += strlen (metadata) + padding;
  83. }
  84. out = realloc (*out_p, buf_len);
  85. if (UNLIKELY (!out)) return LSUP_MEM_ERR;
  86. if (shorten) {
  87. strcpy (out, tmp);
  88. } else if (metadata) {
  89. char *fmt = (
  90. md_rc == LSUP_NORESULT ? "\"%s\"^^<%s>"
  91. : "\"%s\"^^%s");
  92. sprintf (out, fmt, tmp, metadata);
  93. }
  94. else {
  95. sprintf (out, "\"%s\"", tmp);
  96. }
  97. free (tmp);
  98. rc = LSUP_OK;
  99. break;
  100. case LSUP_TERM_LT_LITERAL:
  101. // Calculate string length.
  102. if (escape_lit (term->data, &tmp) != LSUP_OK)
  103. return LSUP_ERROR;
  104. buf_len = strlen (tmp) + 3; // Room for "" and terminator
  105. if (term->lang[0] != '\0') {
  106. metadata = strndup (term->lang, sizeof (LSUP_LangTag));
  107. buf_len += strlen (metadata) + 1; // Room for @
  108. }
  109. out = realloc (*out_p, buf_len);
  110. if (UNLIKELY (!out)) return LSUP_MEM_ERR;
  111. sprintf (out, "\"%s\"", tmp);
  112. free (tmp);
  113. // Add lang.
  114. if (metadata) out = strcat (strcat (out, "@"), metadata);
  115. rc = LSUP_OK;
  116. break;
  117. case LSUP_TERM_BNODE:
  118. out = realloc (*out_p, strlen (term->data) + 3);
  119. if (UNLIKELY (!out)) return LSUP_MEM_ERR;
  120. sprintf (out, "_:%s", term->data);
  121. rc = LSUP_OK;
  122. break;
  123. default:
  124. out = *out_p; // This is considered garbage.
  125. rc = LSUP_PARSE_ERR;
  126. }
  127. free (metadata);
  128. *out_p = out;
  129. return rc;
  130. }
  131. static void *
  132. gr_to_ttl_init (const LSUP_Graph *gr)
  133. {
  134. LSUP_TTLCodecIterator *it;
  135. CALLOC_GUARD (it, NULL);
  136. it->codec = &ttl_codec;
  137. it->gr = gr;
  138. it->subjects = LSUP_graph_unique_terms (gr, TRP_POS_S);
  139. // Sets the condition to build the prolog on 1st iteration.
  140. it->rc = LSUP_NORESULT;
  141. return it;
  142. }
  143. /// Build header and prolog.
  144. static LSUP_rc
  145. build_prolog (LSUP_TTLCodecIterator *it, char **res_p)
  146. {
  147. char *res = fmt_header ("# ");
  148. const char ***nsm = LSUP_nsmap_dump ();
  149. char *ns_tpl = "@prefix %s: <%s> .\n";
  150. // Prefix map.
  151. for (size_t i = 0; nsm[i]; i++) {
  152. const char **ns = nsm[i];
  153. size_t old_len = strlen (res);
  154. size_t ns_len = strlen (ns[0]) + strlen (ns[1]) + strlen (ns_tpl);
  155. char *tmp = realloc (res, old_len + ns_len + 1);
  156. if (UNLIKELY (!tmp)) return LSUP_MEM_ERR;
  157. res = tmp;
  158. sprintf (res + old_len, ns_tpl, ns[0], ns[1]);
  159. free (ns);
  160. }
  161. free (nsm);
  162. // Base.
  163. char *base_uri_str = LSUP_graph_uri (it->gr)->data;
  164. char *base_stmt_tpl = "\n@base <%s> .\n\n";
  165. char *base_stmt = malloc (
  166. strlen (base_stmt_tpl) + strlen (base_uri_str) + 1);
  167. if (!UNLIKELY (base_stmt)) return LSUP_MEM_ERR;
  168. sprintf (base_stmt, base_stmt_tpl, base_uri_str);
  169. res = realloc (res, strlen (res) + strlen (base_stmt) + 1);
  170. if (!UNLIKELY (res)) return LSUP_MEM_ERR;
  171. res = strcat (res, base_stmt);
  172. free (base_stmt);
  173. *res_p = res;
  174. it->rc = LSUP_OK;
  175. return LSUP_OK;
  176. }
  177. /// Encode all the triples for a single subject.
  178. static LSUP_rc
  179. gr_to_ttl_iter (void *h, char **res_p) {
  180. LSUP_TTLCodecIterator *it = h;
  181. if (it->rc == LSUP_NORESULT) return build_prolog (it, res_p);
  182. LSUP_Term *s = NULL;
  183. char *res = *res_p; // Result string will be reallocated.
  184. RCCK (LSUP_term_set_next (it->subjects, &it->s_cur, &s));
  185. term_to_ttl (s, &res);
  186. LSUP_LinkMap *lmap = LSUP_graph_connections (
  187. it->gr, s, LSUP_LINK_OUTBOUND);
  188. LSUP_LinkMapIterator *lmit = LSUP_link_map_iter_new (lmap);
  189. LSUP_Term *p = NULL;
  190. LSUP_TermSet *o_ts = NULL;
  191. char *p_join = " ";
  192. // Begin predicate loop.
  193. while (LSUP_link_map_next (lmit, &p, &o_ts) != LSUP_END) {
  194. // Add predicate representation.
  195. RCCK (term_to_ttl (p, &it->p_str));
  196. char *tmp = realloc (
  197. res, strlen (res) + strlen (it->p_str) + strlen (p_join) + 1);
  198. if (UNLIKELY (!tmp)) goto memfail;
  199. res = strcat (strcat (tmp, p_join), it->p_str);
  200. p_join = " ; ";
  201. // Add objects for predicate.
  202. size_t i = 0;
  203. LSUP_Term *o = NULL;
  204. char *o_join = " ";
  205. while (LSUP_term_set_next (o_ts, &i, &o) != LSUP_END) {
  206. it->rc = term_to_ttl (o, &it->o_str);
  207. RCCK (it->rc);
  208. char *tmp = realloc (
  209. res, strlen (res) + strlen (it->o_str) + strlen (o_join) + 1);
  210. if (UNLIKELY (!tmp)) goto memfail;
  211. res = strcat (strcat (tmp, o_join), it->o_str);
  212. o_join = " , ";
  213. }
  214. }
  215. char *s_sep = " .\n";
  216. char *tmp = realloc (res, strlen (res) + strlen (s_sep) + 1);
  217. if (UNLIKELY (!tmp)) goto memfail;
  218. *res_p = strcat (tmp, s_sep);
  219. LSUP_link_map_iter_free (lmit);
  220. LSUP_link_map_free (lmap);
  221. return it->rc;
  222. memfail:
  223. free (res);
  224. *res_p = NULL;
  225. return LSUP_MEM_ERR;
  226. }
  227. static void
  228. gr_to_ttl_done (void *h)
  229. {
  230. LSUP_TTLCodecIterator *it = h;
  231. LSUP_term_set_free (it->subjects);
  232. free (it->s_str);
  233. free (it->p_str);
  234. free (it->o_str);
  235. free (it);
  236. }
  237. const LSUP_Codec ttl_codec = {
  238. .name = "Turtle",
  239. .mimetype = "text/turtle",
  240. .extension = "ttl",
  241. .encode_term = term_to_ttl,
  242. .encode_graph_init = gr_to_ttl_init,
  243. .encode_graph_iter = gr_to_ttl_iter,
  244. .encode_graph_done = gr_to_ttl_done,
  245. //.decode_term = LSUP_ttl_parse_term,
  246. .decode_graph = LSUP_ttl_parse_doc,
  247. };