123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330 |
- #include "codec/parser_nt.h"
- #include "codec/tokens_nt.h"
- #define YYCTYPE uint8_t
- #define YYCURSOR it->cur
- #define YYMARKER it->mar
- #define YYLIMIT it->lim
- #define YYFILL fill(it) == 0
- typedef struct {
- FILE * fh; // Input file handle.
- YYCTYPE buf[CHUNK_SIZE], // Start of buffer.
- * lim, // Position after the last available
- // input character (YYLIMIT).
- * cur, // Next input character to be read
- // (YYCURSOR)
- * mar, // Most recent match (YYMARKER)
- * tok, // Start of current token.
- * bol; // Address of the beginning of the
- // current line (for debugging).
- unsigned line; // Current line no. (for debugging).
- unsigned ct; // Number of parsed triples.
- bool eof; // if we have reached EOF.
- /*!stags:re2c format = "YYCTYPE *@@;"; */
- } ParseIterator;
- static int fill(ParseIterator *it)
- {
- if (it->eof) {
- return 1;
- }
- const size_t shift = it->tok - it->buf;
- if (shift < 1) {
- return 2;
- }
- log_debug ("Shifting bytes: %lu", shift);
- memmove(it->buf, it->tok, it->lim - it->tok);
- it->lim -= shift;
- it->cur -= shift;
- it->mar -= shift;
- it->tok -= shift;
- it->lim += fread(it->lim, 1, shift, it->fh);
- /*!stags:re2c format = "if (it->@@) it->@@ -= shift; "; */
- it->lim[0] = 0;
- it->eof |= it->lim < it->buf + CHUNK_SIZE - 1;
- return 0;
- }
- static void parse_init(ParseIterator *it, FILE *fh)
- {
- it->fh = fh;
- it->cur = it->mar = it->tok = it->lim = it->buf + CHUNK_SIZE - 1;
- it->line = 1;
- it->bol = it->buf;
- it->ct = 0;
- it->eof = 0;
- /*!stags:re2c format = "it->@@ = NULL; "; */
- fill (it);
- }
- // Parser interface. Required here to silence linters.
- void *NTParseAlloc();
- void NTParse();
- void NTParseFree();
- #ifdef DEBUG
- void NTParseTrace();
- #endif
- // Lexer.
- static int lex (ParseIterator *it, LSUP_Term **term)
- {
- const YYCTYPE *lit_data_e, *dtype_s, *lang_s;
- loop:
- it->tok = it->cur;
- *term = NULL;
- /*!re2c
- re2c:eof = 0;
- re2c:flags:8 = 1;
- re2c:flags:tags = 1;
- re2c:tags:expression = "it->@@";
- re2c:api:style = functions;
- re2c:define:YYFILL:naked = 1;
- // For unresolved and partially resolved inconsistencies of the spec, see
- // https://lists.w3.org/Archives/Public/public-rdf-comments/2017Jun/0000.html
- _WS = [\x09\x20];
- WS = _WS+;
- EOL = [\x0D\x0A] (_WS | [\x0D\x0A])*;
- DOT = ".";
- HEX = [0-9A-Fa-f];
- ECHAR = [\\] [tbnrf"'\\];
- UCHAR = "\\u" HEX{4} | "\\U" HEX{8};
- PN_CHARS_BASE = [A-Za-z\u00C0-\u00D6\u00D8-\u00F6\u00F8-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\U00010000-\U000EFFFF];
- PN_CHARS_U = PN_CHARS_BASE | '_' | ':';
- PN_CHARS = PN_CHARS_U | '-' | [0-9\u00B7\u0300-\u036F\u203F-\u2040];
- IRI_CHARS = ([^\x00-\x20<>"{}|^`\\] | UCHAR)*;
- LITERAL_QUOTE = ["] ([^\x22\x5C\x0A\x0D] | ECHAR|UCHAR)* ["];
- LANGTAG = [@] [a-zA-Z]+ ("-" [a-zA-Z0-9]+)*;
- IRIREF = [<] IRI_CHARS [>];
- LITERAL = LITERAL_QUOTE @lit_data_e _WS* ("^^" _WS* @dtype_s IRIREF | @lang_s LANGTAG)?;
- BNODE = "_:" ((PN_CHARS_U | [0-9]) ((PN_CHARS | ".")* PN_CHARS)?);
- COMMENT = "#" .*;
- EOL {
- it->line ++;
- it->bol = YYCURSOR;
- log_debug ("New line: #%u.", it->line);
- return T_EOL;
- }
- $ {
- log_debug ("End of buffer.");
- return T_EOF;
- }
- IRIREF {
- YYCTYPE *data = unescape_unicode (it->tok + 1, YYCURSOR - it->tok - 2);
- log_debug ("URI data: %s", data);
- *term = LSUP_iriref_new ((char*)data, NULL);
- free (data);
- if (!UNLIKELY (term)) return -1;
- return T_IRIREF;
- }
- LITERAL {
- // Only unescape Unicode from data.
- size_t size = lit_data_e - it->tok - 2;
- YYCTYPE *data = unescape_unicode (it->tok + 1, size);
- log_trace ("Literal data: %s", data);
- char *metadata = NULL;
- const YYCTYPE *md_marker;
- LSUP_TermType type = LSUP_TERM_LITERAL;
- if (dtype_s) {
- md_marker = dtype_s;
- size = YYCURSOR - md_marker - 1;
- } else if (lang_s) {
- type = LSUP_TERM_LT_LITERAL;
- md_marker = lang_s;
- size = YYCURSOR - md_marker;
- } else md_marker = NULL;
- if (md_marker) {
- metadata = malloc (size);
- memcpy (metadata, md_marker + 1, size);
- metadata [size - 1] = '\0';
- log_trace ("metadata: %s", metadata);
- }
- if (type == LSUP_TERM_LITERAL) {
- LSUP_Term *dtype;
- dtype = (
- metadata ? LSUP_iriref_new ((char *) metadata, NULL) : NULL);
- *term = LSUP_literal_new ((char *) data, dtype);
- } else *term = LSUP_lt_literal_new ((char *) data, (char *) metadata);
- free (data);
- free (metadata);
- if (!UNLIKELY (term)) return -1;
- return T_LITERAL;
- }
- BNODE {
- YYCTYPE *data = unescape_unicode (it->tok + 2, YYCURSOR - it->tok - 2);
- log_debug ("BNode data: %s", data);
- *term = LSUP_term_new (LSUP_TERM_BNODE, (char*)data, NULL);
- free (data);
- if (!UNLIKELY (term)) return -1;
- return T_BNODE;
- }
- DOT {
- log_debug ("End of triple.");
- it->ct ++;
- return T_DOT;
- }
- WS {
- log_debug ("Separator.");
- return T_WS;
- }
- COMMENT {
- size_t size = YYCURSOR - it->tok + 1;
- YYCTYPE *data = malloc (size);
- memcpy (data, it->tok, size);
- data [size - 1] = '\0';
- log_debug ("Comment: `%s`", data);
- free (data);
- goto loop;
- }
- * {
- log_debug (
- "Invalid token @ %lu: %s (\\x%x)",
- YYCURSOR - it->buf - 1, it->tok, *it->tok);
- return -1;
- }
- */
- }
- LSUP_rc
- LSUP_nt_parse_term (const char *rep, const LSUP_NSMap *map, LSUP_Term **term)
- {
- FILE *fh = fmemopen ((void *)rep, strlen (rep), "r");
- ParseIterator it;
- parse_init (&it, fh);
- int ttype = lex (&it, term);
- fclose (fh);
- switch (ttype) {
- case T_IRIREF:
- case T_LITERAL:
- case T_BNODE:
- return LSUP_OK;
- default:
- return LSUP_PARSE_ERR;
- }
- }
- LSUP_rc
- LSUP_nt_parse_doc (FILE *fh, LSUP_Graph **gr_p, size_t *ct, char **err_p)
- {
- *err_p = NULL;
- *gr_p = NULL;
- ParseIterator parse_it;
- parse_init (&parse_it, fh);
- #ifdef DEBUG
- NTParseTrace (stdout, "NT Parser > ");
- #endif
- void *parser = NTParseAlloc (malloc);
- LSUP_rc rc;
- LSUP_Graph *gr = LSUP_graph_new (NULL, NULL, NULL);
- if (UNLIKELY (!gr)) return LSUP_MEM_ERR;
- LSUP_GraphIterator *it = LSUP_graph_add_init (gr);
- if (UNLIKELY (!it)) {
- LSUP_graph_free (gr);
- return LSUP_MEM_ERR;
- }
- LSUP_Term *term = NULL;
- for (;;) {
- int ttype = lex (&parse_it, &term);
- if (ttype == -1) {
- char token[16] = {};
- strncpy (token, (const char *)parse_it.tok, 15);
- char *err_start = "Parse error near token `";
- char err_info [64];
- sprintf(
- err_info, "[...]' at line %u, character %ld.\n",
- parse_it.line, parse_it.cur - parse_it.bol);
- size_t err_size = strlen (err_start) + 16 + strlen(err_info);
- char *err_str = malloc (err_size);
- sprintf (err_str, "%s%s%s", err_start, token, err_info);
- rc = LSUP_PARSE_ERR;
- *err_p = err_str;
- log_error (err_str);
- goto finally;
- }
- NTParse (parser, ttype, term, it);
- if (ttype == T_EOF) break;
- };
- if (ct) *ct = parse_it.ct;
- log_info ("Parsed %u triples.", parse_it.ct);
- log_debug ("Graph size: %lu", LSUP_graph_size (gr));
- rc = parse_it.ct > 0 ? LSUP_OK : LSUP_NORESULT;
- *gr_p = gr;
- finally: ;
- NTParse (parser, 0, NULL, it);
- NTParseFree (parser, free);
- LSUP_graph_add_done (it);
- LSUP_term_free (term);
- if (rc < 0) LSUP_graph_free (gr);
- return rc;
- }
|