static foreign_t
turtle_read_name(term_t C0, term_t Stream, term_t C, term_t Name)
{ int c;
charbuf b;
IOSTREAM *in;
if ( !PL_get_integer(C0, &c) )
return type_error(C0, "code");
if ( !wcis_name_start_char(c) )
return FALSE;
if ( !PL_get_stream_handle(Stream, &in) )
return FALSE;
init_charbuf(&b);
add_charbuf(&b, c);
for(;;)
{ int c = Sgetcode(in);
if ( wcis_name_char(c) )
{ add_charbuf(&b, c);
} else
{ int rc = ( PL_unify_integer(C, c) &&
PL_unify_wchars(Name, PL_ATOM, b.here-b.base, b.base) );
free_charbuf(&b);
PL_release_stream(in);
return rc;
}
}
}
static foreign_t
uri_query_components(term_t string, term_t list)
{ pl_wchar_t *s;
size_t len;
if ( PL_get_wchars(string, &len, &s, CVT_ATOM|CVT_STRING|CVT_LIST) )
{ return unify_query_string_components(list, len, s);
} else if ( PL_is_list(list) )
{ term_t tail = PL_copy_term_ref(list);
term_t head = PL_new_term_ref();
term_t nv = PL_new_term_refs(2);
charbuf out;
int rc;
fill_flags();
init_charbuf(&out);
while( PL_get_list(tail, head, tail) )
{ atom_t fname;
int arity;
if ( PL_is_functor(head, FUNCTOR_equal2) ||
PL_is_functor(head, FUNCTOR_pair2) )
{ _PL_get_arg(1, head, nv+0);
_PL_get_arg(2, head, nv+1);
} else if ( PL_get_name_arity(head, &fname, &arity) && arity == 1 )
{ PL_put_atom(nv+0, fname);
_PL_get_arg(1, head, nv+1);
} else
{ free_charbuf(&out);
return type_error("name_value", head);
}
if ( out.here != out.base )
add_charbuf(&out, '&');
if ( !add_encoded_term_charbuf(&out, nv+0, ESC_QNAME) )
{ free_charbuf(&out);
return FALSE;
}
add_charbuf(&out, '=');
if ( !add_encoded_term_charbuf(&out, nv+1, ESC_QVALUE) )
{ free_charbuf(&out);
return FALSE;
}
}
rc = PL_unify_wchars(string, PL_ATOM, out.here-out.base, out.base);
free_charbuf(&out);
return rc;
} else
{ return PL_get_wchars(string, &len, &s,
CVT_ATOM|CVT_STRING|CVT_LIST|CVT_EXCEPTION);
}
return FALSE;
}
static int
iri_add_encoded_charbuf(charbuf *cb, int c, int flags)
{ if ( iri_no_escape(c, flags) )
{ add_charbuf(cb, c);
} else
{ assert(c < 128);
add_charbuf(cb, '%');
add_charbuf(cb, hexdigit(c>>4));
add_charbuf(cb, hexdigit(c&0xf));
}
return TRUE;
}
static foreign_t
uri_authority_components(term_t Authority, term_t components)
{ pl_wchar_t *s;
size_t len;
if ( PL_get_wchars(Authority, &len, &s, CVT_ATOM|CVT_STRING|CVT_LIST) )
{ return unify_uri_authority_components(components, len, s);
} else if ( PL_is_functor(components, FUNCTOR_uri_authority4) )
{ charbuf b;
int rc;
init_charbuf(&b);
if ( (rc=get_text_arg(components, 1, &len, &s, TXT_EX_TEXT)) == TRUE )
{ add_nchars_charbuf(&b, len, s);
if ( (rc=get_text_arg(components, 2, &len, &s, TXT_EX_TEXT)) == TRUE )
{ add_charbuf(&b, ':');
add_nchars_charbuf(&b, len, s);
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
add_charbuf(&b, '@');
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
if ( (rc=get_text_arg(components, 3, &len, &s, TXT_EX_TEXT)) == TRUE )
{ add_nchars_charbuf(&b, len, s);
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
if ( (rc=get_text_arg(components, 4, &len, &s,
TXT_EX_TEXT|CVT_INTEGER)) == TRUE )
{ add_charbuf(&b, ':');
add_nchars_charbuf(&b, len, s);
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
rc = PL_unify_wchars(Authority, PL_ATOM, b.here-b.base, b.base);
free_charbuf(&b);
return rc;
} else
{ return PL_get_wchars(Authority, &len, &s,
CVT_ATOM|CVT_STRING|CVT_LIST|CVT_EXCEPTION);
}
}
static foreign_t
turtle_read_relative_uri(term_t C0, term_t Stream, term_t C, term_t Value)
{ int c;
charbuf b;
IOSTREAM *in;
if ( !PL_get_integer(C0, &c) )
return type_error(C0, "code");
if ( c != '<' )
return FALSE;
if ( !PL_get_stream_handle(Stream, &in) )
return FALSE;
init_charbuf(&b);
c = Sgetcode(in);
for(; ; c = Sgetcode(in))
{ if ( c == '>' )
{ int rc;
c = Sgetcode(in);
rc = (PL_unify_integer(C, c) &&
PL_unify_wchars(Value, PL_ATOM, b.here-b.base, b.base));
PL_release_stream(in);
free_charbuf(&b);
return rc;
} else if ( c == '\\' )
{ int esc;
c = Sgetcode(in);
if ( c == '>' )
{ add_charbuf(&b, c);
} else if ( string_escape(in, c, &esc) )
{ add_charbuf(&b, esc);
} else
{ free_charbuf(&b);
PL_release_stream(in);
return FALSE;
}
} else if ( c == -1 )
{ free_charbuf(&b);
PL_release_stream(in);
return syntax_error("eof_in_uri", in);
} else
{ add_charbuf(&b, c);
}
}
}
static int
add_decoded_range_charbuf(charbuf *cb, const range *r, int flags)
{ const pl_wchar_t *s = r->start;
while(s<r->end)
{ int c;
if ( *s == '%' )
{ const pl_wchar_t *e;
if ( (e=get_encoded_utf8(s, &c)) )
{ s = e;
} else if (hex(s+1, 2, &c) )
{ s += 3;
} else
{ c = *s++;
}
} else if ( *s == '+' && flags == ESC_QVALUE )
{ s++;
c = ' ';
} else
{ c = *s++;
}
add_charbuf(cb, c);
}
return TRUE;
}
static int
add_encoded_charbuf(charbuf *cb, int c, int flags)
{ if ( no_escape(c, flags) )
{ add_charbuf(cb, c);
} else
{ char tmp[6];
const char *end = utf8_put_char(tmp, c);
const char *s;
for(s=tmp; s<end; s++)
{ int b = s[0]&0xff;
add_charbuf(cb, '%');
add_charbuf(cb, hexdigit(b>>4));
add_charbuf(cb, hexdigit(b&0xf));
}
}
return TRUE;
}
static int
add_nchars_charbuf(charbuf *cb, size_t len, const pl_wchar_t *s)
{ if ( cb->here+len <= cb->end )
{ wcsncpy(cb->here, s, len);
cb->here += len;
} else
{ size_t n;
for(n=0; n<len; n++)
add_charbuf(cb, s[n]);
}
return TRUE;
}
static int
ranges_in_charbuf(charbuf *cb, uri_component_ranges *ranges)
{ if ( ranges->scheme.start )
{ add_verb_range_charbuf(cb, &ranges->scheme);
add_charbuf(cb, ':');
}
if ( ranges->authority.start )
{ add_charbuf(cb, '/');
add_charbuf(cb, '/');
add_verb_range_charbuf(cb, &ranges->authority);
}
add_verb_range_charbuf(cb, &ranges->path);
if ( ranges->query.start )
{ add_charbuf(cb, '?');
add_verb_range_charbuf(cb, &ranges->query);
}
if ( ranges->fragment.start )
{ add_charbuf(cb, '#');
add_verb_range_charbuf(cb, &ranges->fragment);
}
return TRUE;
}
static int
normalize_in_charbuf(charbuf *cb, uri_component_ranges *ranges, int iri)
{ fill_flags();
if ( ranges->scheme.start )
{ add_lwr_range_charbuf(cb, &ranges->scheme, iri, ESC_SCHEME);
add_charbuf(cb, ':');
}
if ( ranges->authority.start )
{ add_charbuf(cb, '/');
add_charbuf(cb, '/');
add_lwr_range_charbuf(cb, &ranges->authority, iri, ESC_AUTH);
}
if ( ranges->path.end > ranges->path.start )
{ charbuf pb;
charbuf path;
size_t len;
init_charbuf(&pb);
add_range_charbuf(&pb, &ranges->path, iri, ESC_PATH);
init_charbuf_at_size(&path, pb.here-pb.base);
len = removed_dot_segments(pb.here-pb.base, pb.base, path.base);
add_nchars_charbuf(cb, len, path.base);
free_charbuf(&path);
free_charbuf(&pb);
}
if ( ranges->query.start )
{ add_charbuf(cb, '?');
add_range_charbuf(cb, &ranges->query, iri, ESC_QUERY);
}
if ( ranges->fragment.start )
{ add_charbuf(cb, '#');
add_range_charbuf(cb, &ranges->fragment, iri, ESC_QVALUE);
}
return TRUE;
}
static foreign_t
turtle_read_string(term_t C0, term_t Stream, term_t C, term_t Value)
{ int c;
charbuf b;
IOSTREAM *in;
int endlen = 1;
if ( !PL_get_integer(C0, &c) )
return type_error(C0, "code");
if ( c != '"' )
return FALSE;
if ( !PL_get_stream_handle(Stream, &in) )
return FALSE;
init_charbuf(&b);
c = Sgetcode(in);
if ( c == '"' )
{ c = Sgetcode(in);
if ( c == '"' ) /* """...""" */
{ endlen = 3;
c = Sgetcode(in);
} else
{ PL_release_stream(in);
return (PL_unify_integer(C, c) &&
PL_unify_atom(Value, ATOM_));
}
}
for(;;c = Sgetcode(in))
{ if ( c == -1 )
{ free_charbuf(&b);
PL_release_stream(in);
return syntax_error("eof_in_string", in);
} else if ( c == '"' )
{ int count = 1;
for(count=1; count<endlen; )
{ if ( (c=Sgetcode(in)) == '"' )
count++;
else
break;
}
if ( count == endlen )
{ int rc;
c = Sgetcode(in);
rc = (PL_unify_integer(C, c) &&
PL_unify_wchars(Value, PL_ATOM, b.here-b.base, b.base));
free_charbuf(&b);
PL_release_stream(in);
return rc;
}
while(count-- > 0)
add_charbuf(&b, '"');
add_charbuf(&b, c);
} else if ( c == '\\' )
{ int esc;
c = Sgetcode(in);
if ( !string_escape(in, c, &esc) )
{ free_charbuf(&b);
PL_release_stream(in);
return FALSE;
}
add_charbuf(&b, esc);
} else
{ add_charbuf(&b, c);
}
}
}
static foreign_t
uri_components(term_t URI, term_t components)
{ pl_wchar_t *s;
size_t len;
if ( PL_get_wchars(URI, &len, &s, CVT_ATOM|CVT_STRING|CVT_LIST) )
{ uri_component_ranges ranges;
term_t rt = PL_new_term_refs(6);
term_t av = rt+1;
parse_uri(&ranges, len, s);
unify_range(av+0, &ranges.scheme);
unify_range(av+1, &ranges.authority);
unify_range(av+2, &ranges.path);
unify_range(av+3, &ranges.query);
unify_range(av+4, &ranges.fragment);
return (PL_cons_functor_v(rt, FUNCTOR_uri_components5, av) &&
PL_unify(components, rt));
} else if ( PL_is_functor(components, FUNCTOR_uri_components5) )
{ charbuf b;
int rc;
init_charbuf(&b);
/* schema */
if ( (rc=get_text_arg(components, 1, &len, &s, TXT_EX_TEXT)) == TRUE )
{ add_nchars_charbuf(&b, len, s);
add_charbuf(&b, ':');
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
/* authority */
if ( (rc=get_text_arg(components, 2, &len, &s, TXT_EX_TEXT)) == TRUE )
{ add_charbuf(&b, '/');
add_charbuf(&b, '/');
add_nchars_charbuf(&b, len, s);
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
/* path */
if ( (rc=get_text_arg(components, 3, &len, &s, TXT_EX_TEXT)) == TRUE )
{ add_nchars_charbuf(&b, len, s);
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
/* query */
if ( (rc=get_text_arg(components, 4, &len, &s, TXT_EX_TEXT)) == TRUE )
{ if ( len > 0 )
{ add_charbuf(&b, '?');
add_nchars_charbuf(&b, len, s);
}
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
/* fragment */
if ( (rc=get_text_arg(components, 5, &len, &s, TXT_EX_TEXT)) == TRUE )
{ add_charbuf(&b, '#');
add_nchars_charbuf(&b, len, s);
} else if ( rc == -1 )
{ free_charbuf(&b);
return FALSE;
}
rc = PL_unify_wchars(URI, PL_ATOM, b.here-b.base, b.base);
free_charbuf(&b);
return rc;
} else /* generate an error */
{ return PL_get_wchars(URI, &len, &s,
CVT_ATOM|CVT_STRING|CVT_LIST|CVT_EXCEPTION);
}
}
static foreign_t
resolve(term_t Rel, term_t Base, term_t URI, int normalize, int iri)
{ pl_wchar_t *s;
size_t slen;
uri_component_ranges s_ranges, t_ranges;
int rc;
size_t len;
charbuf out, pb, path;
init_charbuf(&pb); /* path-buffer */
if ( PL_get_wchars(Rel, &slen, &s,
CVT_ATOM|CVT_STRING|CVT_LIST|CVT_EXCEPTION) )
{ parse_uri(&s_ranges, slen, s);
if ( s_ranges.scheme.start )
{ t_ranges = s_ranges;
} else
{ const uri_component_ranges *b_ranges;
if ( !(b_ranges = base_ranges(Base)) )
return FALSE;
memset(&t_ranges, 0, sizeof(t_ranges));
if ( s_ranges.authority.start )
{ t_ranges.authority = s_ranges.authority;
t_ranges.path = s_ranges.path;
t_ranges.query = s_ranges.query;
} else
{ if ( s_ranges.path.start == s_ranges.path.end )
{ t_ranges.path = b_ranges->path;
if ( s_ranges.query.start )
t_ranges.query = s_ranges.query;
else
t_ranges.query = b_ranges->query;
} else
{ if ( s_ranges.path.start[0] == '/' )
{ t_ranges.path = s_ranges.path;
} else
{ if ( b_ranges->authority.start &&
b_ranges->path.start == b_ranges->path.end )
{ add_charbuf(&pb, '/');
add_verb_range_charbuf(&pb, &s_ranges.path);
} else
{ range path = b_ranges->path;
path.end = remove_last_segment(path.start, path.end);
add_verb_range_charbuf(&pb, &path);
add_verb_range_charbuf(&pb, &s_ranges.path);
t_ranges.path.start = pb.base;
t_ranges.path.end = pb.here;
}
}
t_ranges.query = s_ranges.query;
}
t_ranges.authority = b_ranges->authority;
}
t_ranges.scheme = b_ranges->scheme;
t_ranges.fragment = s_ranges.fragment;
}
} else
return FALSE;
init_charbuf(&out); /* output buffer */
if ( normalize )
{ normalize_in_charbuf(&out, &t_ranges, iri);
} else
{ init_charbuf_at_size(&path, t_ranges.path.end - t_ranges.path.start);
len = removed_dot_segments(t_ranges.path.end - t_ranges.path.start,
t_ranges.path.start,
path.base);
t_ranges.path.start = path.base;
t_ranges.path.end = path.base+len;
free_charbuf(&pb);
ranges_in_charbuf(&out, &t_ranges);
}
rc = PL_unify_wchars(URI, PL_ATOM, out.here-out.base, out.base);
free_charbuf(&out);
return rc;
}