grammar_ttl.y 7.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249
  1. %include {
  2. /** @brief Lemon parser grammar for Turtle.
  3. *
  4. * The `lemon' parser generator executable must be in your PATH:
  5. * https://sqlite.org/src/doc/trunk/doc/lemon.html
  6. *
  7. * TTL EBNF: https://www.w3.org/TeamSubmission/turtle/#sec-grammar-grammar
  8. */
  9. #include "volksdata/codec.h"
  10. }
  11. %name TTLParse
  12. %stack_overflow {
  13. log_error ("Stack oveflow in TTL parsing.");
  14. state->rc = VOLK_MEM_ERR;
  15. }
  16. %parse_failure {
  17. log_error ("TTL parse error. Cannot continue.");
  18. state->rc = VOLK_PARSE_ERR;
  19. }
  20. %syntax_error {
  21. // Fail immediately on first error.
  22. yy_parse_failed (yypParser);
  23. }
  24. //%stack_size 1024
  25. %token_prefix "T_"
  26. %token_type { char * }
  27. %token_destructor { (void) state; free ($$); }
  28. %default_type { char * }
  29. %extra_argument { VOLK_TTLParserState *state }
  30. // Low- to high-priority.
  31. %left PERIOD .
  32. %left SEMICOLON .
  33. %left COMMA .
  34. %left STRING INTEGER DOUBLE DECIMAL BOOLEAN BNODE_ID QNAME IRIREF .
  35. %nonassoc LANGTAG PREFIX .
  36. %nonassoc COLON .
  37. /*
  38. * Rules.
  39. */
  40. turtleDoc ::= statements EOF .
  41. statements ::= statements statement .
  42. statements ::= .
  43. statement ::= prefixID .
  44. statement ::= base .
  45. statement ::= triples .
  46. statement ::= WS .
  47. prefixID ::= PREFIX(P) WS IRIREF(N) PERIOD . {
  48. VOLK_nsmap_add (P, N);
  49. free (P);
  50. free (N);
  51. }
  52. base ::= BASE WS IRIREF(D) PERIOD . {
  53. VOLK_term_free (state->base);
  54. state->base = VOLK_iriref_new (D);
  55. free (D);
  56. }
  57. triples ::= subject(S) ows predObjList(L) PERIOD . {
  58. size_t ct = VOLK_graph_add_link_map (state->it, L);
  59. state->ct += ct;
  60. state->rc = VOLK_OK;
  61. LOG_TRACE("Added %lu triples.", ct);
  62. VOLK_term_free (S);
  63. VOLK_link_map_free (L);
  64. }
  65. triples ::= subject(S) ows predObjList(L) SEMICOLON PERIOD . [PERIOD] {
  66. size_t ct = VOLK_graph_add_link_map (state->it, L);
  67. state->ct += ct;
  68. state->rc = VOLK_OK;
  69. LOG_TRACE("Added %lu triples.", ct);
  70. VOLK_term_free (S);
  71. VOLK_link_map_free (L);
  72. }
  73. %type predObjList { VOLK_LinkMap * }
  74. %destructor predObjList { VOLK_link_map_free ($$); }
  75. predObjList(A) ::= predicate(P) ows objectList(O) . [SEMICOLON] {
  76. VOLK_Term *s;
  77. if (state->lms) s = state->lms;
  78. else {
  79. // TODO This may be brittle. It is not verifying the
  80. // full BNode syntax.
  81. log_info ("Link map subject not present. Assuming BNode.");
  82. s = VOLK_bnode_new (NULL);
  83. }
  84. A = VOLK_link_map_new (s, VOLK_LINK_OUTBOUND);
  85. VOLK_link_map_add (A, P, O);
  86. if (s != state->lms) VOLK_term_free (s);
  87. }
  88. predObjList(A) ::= predObjList(L) SEMICOLON predicate(P) ows objectList(O) . {
  89. VOLK_link_map_add (L, P, O);
  90. A = L;
  91. }
  92. %type objectList { VOLK_TermSet * }
  93. %destructor objectList { VOLK_term_set_free ($$); }
  94. objectList(A) ::= objectList(L) COMMA object(O) . {
  95. if (VOLK_term_set_add (L, O, NULL) == VOLK_NOACTION)
  96. VOLK_term_free (O);
  97. A = L;
  98. }
  99. objectList(A) ::= object(O) . [IRIREF] {
  100. A = VOLK_term_set_new();
  101. VOLK_term_set_add (A, O, NULL);
  102. }
  103. %type subject { VOLK_Term * }
  104. %destructor subject { VOLK_term_free ($$); }
  105. subject ::= resource(D) . { state->lms = D; }
  106. subject ::= blank(D) . { state->lms = D; }
  107. %type predicate { VOLK_Term * }
  108. %destructor predicate { VOLK_term_free ($$); }
  109. predicate ::= resource .
  110. predicate(A)::= RDF_TYPE . { A = VOLK_iriref_new_ns ("rdf:type"); }
  111. %type object { VOLK_Term * }
  112. %destructor object { VOLK_term_free ($$); }
  113. object ::= resource .
  114. object ::= blank .
  115. object ::= literal .
  116. %type literal { VOLK_Term * }
  117. %destructor literal { VOLK_term_free ($$); }
  118. literal(A) ::= STRING(D) . {
  119. A = VOLK_literal_new (D, NULL);
  120. LOG_TRACE("Created plain literal: \"%s\"", A->data);
  121. free (D);
  122. }
  123. literal(A) ::= STRING(D) LANGTAG(L) . {
  124. A = VOLK_lt_literal_new (D, L);
  125. LOG_TRACE("Created LT-literal: \"%s\"@%s", A->data, A->lang);
  126. free (D);
  127. free (L);
  128. }
  129. literal(A) ::= STRING(D) DTYPE_MARKER resource(M) . {
  130. A = VOLK_literal_new (D, M);
  131. LOG_TRACE(
  132. "Created DT-literal: \"%s\"^^%s",
  133. A->data, A->datatype);
  134. free (D);
  135. }
  136. literal(A) ::= INTEGER(D) . {
  137. A = VOLK_literal_new (D, VOLK_iriref_new_ns ("xsd:integer"));
  138. free (D);
  139. }
  140. literal(A) ::= DOUBLE(D) . {
  141. A = VOLK_literal_new (D, VOLK_iriref_new_ns ("xsd:double"));
  142. free (D);
  143. }
  144. literal(A) ::= DECIMAL(D) . {
  145. A = VOLK_literal_new (D, VOLK_iriref_new_ns ("xsd:decimal"));
  146. free (D);
  147. }
  148. literal(A) ::= BOOLEAN(D) . {
  149. A = VOLK_literal_new (D, VOLK_iriref_new_ns ("xsd:boolean"));
  150. free (D);
  151. }
  152. %type blank { VOLK_Term * }
  153. %destructor blank { VOLK_term_free ($$); }
  154. blank(A) ::= BNODE_ID(D) . {
  155. A = VOLK_bnode_new (D);
  156. LOG_TRACE("Created blank node: _:%s", A->data);
  157. free (D);
  158. }
  159. blank(A) ::= ANON . [BNODE_ID] {
  160. LOG_TRACE ("Found empty BNode.");
  161. A = VOLK_bnode_new (NULL);
  162. LOG_TRACE("Created empty list BN: _:%s", A->data);
  163. }
  164. blank(A) ::= LBRACKET predObjList(L) RBRACKET . [BNODE_ID] {
  165. LOG_TRACE ("Found BNode with data.");
  166. A = VOLK_bnode_new (NULL);
  167. state->lms = A;
  168. state->ct += VOLK_graph_add_link_map (state->it, L);
  169. LOG_TRACE("Created list BN: _:%s", A->data);
  170. VOLK_link_map_free (L);
  171. }
  172. blank ::= collection . [BNODE_ID]
  173. blank(A) ::= LPAREN RPAREN . [BNODE_ID] {
  174. A = VOLK_iriref_new_ns ("rdf:nil");
  175. LOG_TRACE("Created list terminator: %s", A->data);
  176. }
  177. // "collection" is the subject of the first collection item.
  178. %type collection { VOLK_Term * }
  179. %destructor collection { VOLK_term_free ($$); }
  180. // Collection triples are added here to the graph.
  181. collection(A) ::= LPAREN itemList(L) RPAREN . {
  182. A = VOLK_bnode_add_collection (state->it, L);
  183. VOLK_term_set_free (L);
  184. }
  185. %type itemList { VOLK_TermSet * }
  186. %destructor itemList { VOLK_term_set_free ($$); }
  187. itemList(A) ::= itemList(L) ows object(O) . {
  188. if (VOLK_term_set_add (L, O, NULL) == VOLK_NOACTION)
  189. VOLK_term_free (O);
  190. A = L;
  191. }
  192. itemList(A) ::= object(O) . {
  193. A = VOLK_term_set_new ();
  194. VOLK_term_set_add (A, O, NULL);
  195. }
  196. %type resource { VOLK_Term * }
  197. %destructor resource { VOLK_term_free ($$); }
  198. resource(A) ::= IRIREF(D) . {
  199. VOLK_Term *rel_iri = VOLK_iriref_new (D);
  200. free (D);
  201. if (state->base) {
  202. A = VOLK_iriref_new_abs (rel_iri, state->base);
  203. VOLK_term_free (rel_iri);
  204. } else {
  205. A = rel_iri;
  206. }
  207. LOG_TRACE("Created IRI: <%s>", A->data);
  208. }
  209. resource(A) ::= QNAME(D) . {
  210. A = VOLK_iriref_new_ns (D);
  211. LOG_TRACE("Created IRI: %s", A->data);
  212. free (D);
  213. }
  214. ows ::= WS .
  215. ows ::= .