static void yaz_rpnstructure_to_wrbuf(WRBUF b, const Z_RPNStructure *zs)
{
if (zs->which == Z_RPNStructure_complex)
{
Z_Operator *op = zs->u.complex->roperator;
wrbuf_printf(b, "@%s ", complex_op_name(op) );
if (op->which== Z_Operator_prox)
{
if (!op->u.prox->exclusion)
wrbuf_putc(b, 'n');
else if (*op->u.prox->exclusion)
wrbuf_putc(b, '1');
else
wrbuf_putc(b, '0');
wrbuf_printf(b, " " ODR_INT_PRINTF " %d "
ODR_INT_PRINTF " ", *op->u.prox->distance,
*op->u.prox->ordered,
*op->u.prox->relationType);
switch(op->u.prox->which)
{
case Z_ProximityOperator_known:
wrbuf_putc(b, 'k');
break;
case Z_ProximityOperator_private:
wrbuf_putc(b, 'p');
break;
default:
wrbuf_printf(b, "%d", op->u.prox->which);
}
if (op->u.prox->u.known)
wrbuf_printf(b, " " ODR_INT_PRINTF " ", *op->u.prox->u.known);
else
wrbuf_printf(b, " 0 ");
}
yaz_rpnstructure_to_wrbuf(b,zs->u.complex->s1);
yaz_rpnstructure_to_wrbuf(b,zs->u.complex->s2);
}
else if (zs->which == Z_RPNStructure_simple)
{
if (zs->u.simple->which == Z_Operand_APT)
yaz_apt_to_wrbuf(b, zs->u.simple->u.attributesPlusTerm);
else if (zs->u.simple->which == Z_Operand_resultSetId)
{
wrbuf_printf(b, "@set ");
yaz_encode_pqf_term(b, zs->u.simple->u.resultSetId,
strlen(zs->u.simple->u.resultSetId));
}
else
wrbuf_puts(b, "(unknown simple structure)");
}
else
wrbuf_puts(b, "(unknown structure)");
}
void yaz_encode_pqf_term(WRBUF b, const char *term, int len)
{
int i;
for (i = 0; i < len; i++)
if (strchr(" \"{", term[i]))
break;
if (len > 0 && i == len)
{
if (term[0] == '@')
wrbuf_putc(b, '\\');
for (i = 0; i < len; i++)
{
if (term[i] == '\\')
wrbuf_putc(b, '\\');
wrbuf_putc(b, term[i]);
}
}
else
{
wrbuf_putc(b, '"');
for (i = 0; i < len; i++)
{
if (term[i] == '"' || term[i] == '\\')
wrbuf_putc(b, '\\');
wrbuf_putc(b, term[i]);
}
wrbuf_putc(b, '"');
}
wrbuf_putc(b, ' ');
}
static int yaz_gets(int (*getbyte)(void *client_data),
void (*ungetbyte)(int b, void *client_data),
void *client_data,
WRBUF w)
{
size_t sz = 0;
int ch = getbyte(client_data);
while (ch != '\0' && ch != '\r' && ch != '\n')
{
wrbuf_putc(w, ch);
sz++;
ch = getbyte(client_data);
}
if (ch == '\r')
{
ch = getbyte(client_data);
if (ch != '\n' && ch != '\0')
ungetbyte(ch, client_data);
}
else if (ch == '\n')
{
ch = getbyte(client_data);
if (ch != '\r' && ch != '\0')
ungetbyte(ch, client_data);
}
if (sz)
{
return 1;
}
return 0;
}
/* original tokenizer with our tokenize interface, but we
add +1 to ensure no '\0' are in our string (except for EOF)
*/
static const char *pp2_charset_token_a_to_z(pp2_charset_token_t prt)
{
const char *cp = prt->cp;
int c;
/* skip white space */
while (*cp && (c = raw_char(tolower(*(const unsigned char *)cp))) < 0)
cp++;
if (*cp == '\0')
{
prt->cp = cp;
prt->last_cp = 0;
return 0;
}
/* now read the term itself */
prt->last_cp = cp;
wrbuf_rewind(prt->norm_str);
while (*cp && (c = raw_char(tolower(*cp))) >= 0)
{
wrbuf_putc(prt->norm_str, c);
cp++;
}
prt->cp = cp;
return wrbuf_cstr(prt->norm_str);
}
int yaz_tok_move(yaz_tok_parse_t tp)
{
yaz_tok_cfg_t t = tp->cfg;
const char *cp;
int ch = get_byte(tp);
/* skip white space */
while (ch && strchr(t->white_space, ch))
ch = get_byte(tp);
if (!ch)
ch = YAZ_TOK_EOF;
else if (strchr(t->comment, ch))
ch = YAZ_TOK_EOF;
else if ((cp = strchr(t->single_tokens, ch)))
ch = *cp; /* single token match */
else if ((cp = strchr(t->quote_tokens_begin, ch)))
{ /* quoted string */
int end_ch = t->quote_tokens_end[cp - t->quote_tokens_begin];
ch = get_byte(tp);
wrbuf_rewind(tp->wr_string);
while (ch && ch != end_ch)
wrbuf_putc(tp->wr_string, ch);
if (!ch)
ch = YAZ_TOK_ERROR;
else
ch = YAZ_TOK_QSTRING;
}
else
{ /* unquoted string */
wrbuf_rewind(tp->wr_string);
while (ch && !strchr(t->white_space, ch)
&& !strchr(t->single_tokens, ch)
&& !strchr(t->comment, ch))
{
wrbuf_putc(tp->wr_string, ch);
ch = get_byte(tp);
}
unget_byte(tp, ch);
ch = YAZ_TOK_STRING;
}
tp->look = ch;
return ch;
}
static void json_indent(WRBUF result, int indent)
{
size_t l = wrbuf_len(result);
if (l == 0 || wrbuf_buf(result)[l-1] == '\n')
{
int i;
for (i = 0; i < indent; i++)
wrbuf_putc(result, ' ');
}
}
void yaz_encode_pqf_term(WRBUF b, const char *term, int len)
{
int i;
for (i = 0; i < len; i++)
if (strchr(" \"{", term[i]))
break;
if (i == len && i)
wrbuf_printf(b, "%.*s ", len, term);
else
{
wrbuf_putc(b, '"');
for (i = 0; i<len; i++)
{
if (term[i] == '"')
wrbuf_putc(b, '\\');
wrbuf_putc(b, term[i]);
}
wrbuf_printf(b, "\" ");
}
}
static void write_metadata(WRBUF w, struct conf_service *service,
struct record_metadata **ml, int full, int indent)
{
int imeta;
for (imeta = 0; imeta < service->num_metadata; imeta++)
{
struct conf_metadata *cmd = &service->metadata[imeta];
struct record_metadata *md;
if (!cmd->brief && !full)
continue;
for (md = ml[imeta]; md; md = md->next)
{
struct record_metadata_attr *attr = md->attributes;
int i;
for (i = 0; i < indent; i++)
wrbuf_putc(w, ' ');
wrbuf_printf(w, "<md-%s", cmd->name);
for (; attr; attr = attr->next)
{
wrbuf_printf(w, " %s=\"", attr->name);
wrbuf_xmlputs(w, attr->value);
wrbuf_puts(w, "\"");
}
wrbuf_puts(w, ">");
switch (cmd->type)
{
case Metadata_type_generic:
wrbuf_xmlputs(w, md->data.text.disp);
break;
case Metadata_type_year:
wrbuf_printf(w, "%d", md->data.number.min);
if (md->data.number.min != md->data.number.max)
wrbuf_printf(w, "-%d", md->data.number.max);
break;
default:
wrbuf_puts(w, "[can't represent]");
break;
}
wrbuf_printf(w, "</md-%s>\n", cmd->name);
}
}
}
static int convert_solrmarc(void *info, WRBUF record, WRBUF wr_error)
{
WRBUF w = wrbuf_alloc();
const char *buf = wrbuf_buf(record);
size_t i, sz = wrbuf_len(record);
for (i = 0; i < sz; i++)
{
int ch;
if (buf[i] == '#' && i < sz - 3 && buf[i+3] == ';'
&& atoi_n_check(buf+i+1, 2, &ch))
i += 3;
else
ch = buf[i];
wrbuf_putc(w, ch);
}
wrbuf_rewind(record);
wrbuf_write(record, wrbuf_buf(w), wrbuf_len(w));
wrbuf_destroy(w);
return 0;
}
static struct json_node *do_parse_from_stdin(void)
{
FILE *f = stdin;
WRBUF w = wrbuf_alloc();
struct json_node *n;
size_t pos;
const char *json_str;
const char *err_msg;
int c;
while ((c = getc(f)) != EOF)
wrbuf_putc(w, c);
json_str = wrbuf_cstr(w);
n = json_parse2(json_str, &err_msg, &pos);
if (!n)
{
fprintf(stderr, "JSON parse error: %s\nLeading text was:\n", err_msg);
fwrite(json_str, 1, pos, stderr);
fprintf(stderr, "^\n");
}
wrbuf_destroy(w);
return n;
}
int yaz_marc_write_iso2709(yaz_marc_t mt, WRBUF wr)
{
struct yaz_marc_node *n;
int indicator_length;
int identifier_length;
int length_data_entry;
int length_starting;
int length_implementation;
int data_offset = 0;
const char *leader = 0;
WRBUF wr_dir, wr_head, wr_data_tmp;
int base_address;
for (n = mt->nodes; n; n = n->next)
if (n->which == YAZ_MARC_LEADER)
leader = n->u.leader;
if (!leader)
return -1;
if (!atoi_n_check(leader+10, 1, &indicator_length))
return -1;
if (!atoi_n_check(leader+11, 1, &identifier_length))
return -1;
if (!atoi_n_check(leader+20, 1, &length_data_entry))
return -1;
if (!atoi_n_check(leader+21, 1, &length_starting))
return -1;
if (!atoi_n_check(leader+22, 1, &length_implementation))
return -1;
wr_data_tmp = wrbuf_alloc();
wr_dir = wrbuf_alloc();
for (n = mt->nodes; n; n = n->next)
{
int data_length = 0;
struct yaz_marc_subfield *s;
switch(n->which)
{
case YAZ_MARC_DATAFIELD:
wrbuf_printf(wr_dir, "%.3s", n->u.datafield.tag);
data_length += indicator_length;
wrbuf_rewind(wr_data_tmp);
for (s = n->u.datafield.subfields; s; s = s->next)
{
/* write dummy IDFS + content */
wrbuf_iconv_putchar(wr_data_tmp, mt->iconv_cd, ' ');
wrbuf_iconv_puts(wr_data_tmp, mt->iconv_cd, s->code_data);
marc_iconv_reset(mt, wr_data_tmp);
}
/* write dummy FS (makes MARC-8 to become ASCII) */
wrbuf_iconv_putchar(wr_data_tmp, mt->iconv_cd, ' ');
marc_iconv_reset(mt, wr_data_tmp);
data_length += wrbuf_len(wr_data_tmp);
break;
case YAZ_MARC_CONTROLFIELD:
wrbuf_printf(wr_dir, "%.3s", n->u.controlfield.tag);
wrbuf_rewind(wr_data_tmp);
wrbuf_iconv_puts(wr_data_tmp, mt->iconv_cd,
n->u.controlfield.data);
marc_iconv_reset(mt, wr_data_tmp);
wrbuf_iconv_putchar(wr_data_tmp, mt->iconv_cd, ' ');/* field sep */
marc_iconv_reset(mt, wr_data_tmp);
data_length += wrbuf_len(wr_data_tmp);
break;
case YAZ_MARC_COMMENT:
break;
case YAZ_MARC_LEADER:
break;
}
if (data_length)
{
wrbuf_printf(wr_dir, "%0*d", length_data_entry, data_length);
wrbuf_printf(wr_dir, "%0*d", length_starting, data_offset);
data_offset += data_length;
}
}
/* mark end of directory */
wrbuf_putc(wr_dir, ISO2709_FS);
/* base address of data (comes after leader+directory) */
base_address = 24 + wrbuf_len(wr_dir);
wr_head = wrbuf_alloc();
/* write record length */
wrbuf_printf(wr_head, "%05d", base_address + data_offset + 1);
/* from "original" leader */
wrbuf_write(wr_head, leader+5, 7);
/* base address of data */
wrbuf_printf(wr_head, "%05d", base_address);
/* from "original" leader */
wrbuf_write(wr_head, leader+17, 7);
wrbuf_write(wr, wrbuf_buf(wr_head), 24);
wrbuf_write(wr, wrbuf_buf(wr_dir), wrbuf_len(wr_dir));
wrbuf_destroy(wr_head);
wrbuf_destroy(wr_dir);
wrbuf_destroy(wr_data_tmp);
for (n = mt->nodes; n; n = n->next)
{
struct yaz_marc_subfield *s;
switch(n->which)
{
case YAZ_MARC_DATAFIELD:
wrbuf_write(wr, n->u.datafield.indicator, indicator_length);
for (s = n->u.datafield.subfields; s; s = s->next)
{
wrbuf_putc(wr, ISO2709_IDFS);
wrbuf_iconv_puts(wr, mt->iconv_cd, s->code_data);
marc_iconv_reset(mt, wr);
}
wrbuf_putc(wr, ISO2709_FS);
break;
case YAZ_MARC_CONTROLFIELD:
wrbuf_iconv_puts(wr, mt->iconv_cd, n->u.controlfield.data);
marc_iconv_reset(mt, wr);
wrbuf_putc(wr, ISO2709_FS);
break;
case YAZ_MARC_COMMENT:
break;
case YAZ_MARC_LEADER:
break;
}
}
wrbuf_printf(wr, "%c", ISO2709_RS);
return 0;
}
static int emit_term(solr_transform_t ct, WRBUF w, Z_Term *term, Odr_int trunc)
{
size_t lterm = 0;
const char *sterm = 0;
switch (term->which)
{
case Z_Term_general:
lterm = term->u.general->len;
sterm = (const char *) term->u.general->buf;
break;
case Z_Term_numeric:
wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
break;
case Z_Term_characterString:
sterm = term->u.characterString;
lterm = strlen(sterm);
break;
default:
return YAZ_BIB1_TERM_TYPE_UNSUPP;
}
if (sterm)
{
size_t i;
int must_quote = 0;
for (i = 0 ; i < lterm; i++)
if (sterm[i] == ' ')
must_quote = 1;
if (must_quote)
wrbuf_puts(w, "\"");
if (trunc == 2 || trunc == 3)
wrbuf_puts(w, "*");
for (i = 0 ; i < lterm; i++)
{
if (sterm[i] == '\\' && i < lterm - 1)
{
i++;
if (strchr(SOLR_SPECIAL, sterm[i]))
wrbuf_putc(w, '\\');
wrbuf_putc(w, sterm[i]);
}
else if (sterm[i] == '?' && trunc == 104)
{
wrbuf_putc(w, '*');
}
else if (sterm[i] == '#' && trunc == 104)
{
wrbuf_putc(w, '?');
}
else if (strchr(SOLR_SPECIAL, sterm[i]))
{
wrbuf_putc(w, '\\');
wrbuf_putc(w, sterm[i]);
}
else
wrbuf_putc(w, sterm[i]);
}
if (trunc == 1 || trunc == 3)
wrbuf_puts(w, "*");
if (must_quote)
wrbuf_puts(w, "\"");
}
return 0;
}
static int rpn2solr_simple(solr_transform_t ct,
void (*pr)(const char *buf, void *client_data),
void *client_data,
Z_Operand *q, WRBUF w)
{
int ret = 0;
if (q->which != Z_Operand_APT)
{
ret = -1;
solr_transform_set_error(ct, YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM, 0);
}
else
{
Z_AttributesPlusTerm *apt = q->u.attributesPlusTerm;
Z_Term *term = apt->term;
const char *sterm = 0;
size_t lterm = 0;
Odr_int trunc = get_truncation(apt);
wrbuf_rewind(w);
ret = rpn2solr_attr(ct, apt->attributes, w);
if (trunc == 0 || trunc == 1 || trunc == 100 || trunc == 104)
;
else
{
solr_transform_set_error(ct, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, 0);
return -1;
}
switch (term->which)
{
case Z_Term_general:
lterm = term->u.general->len;
sterm = (const char *) term->u.general->buf;
break;
case Z_Term_numeric:
wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
break;
case Z_Term_characterString:
sterm = term->u.characterString;
lterm = strlen(sterm);
break;
default:
ret = -1;
solr_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
}
if (sterm)
{
size_t i;
int must_quote = 0;
for (i = 0 ; i < lterm; i++)
if (sterm[i] == ' ')
must_quote = 1;
if (must_quote)
wrbuf_puts(w, "\"");
for (i = 0 ; i < lterm; i++)
{
if (sterm[i] == '\\' && i < lterm - 1)
{
i++;
if (strchr(SOLR_SPECIAL, sterm[i]))
wrbuf_putc(w, '\\');
wrbuf_putc(w, sterm[i]);
}
else if (sterm[i] == '?' && trunc == 104)
{
wrbuf_putc(w, '*');
}
else if (sterm[i] == '#' && trunc == 104)
{
wrbuf_putc(w, '?');
}
else if (strchr(SOLR_SPECIAL, sterm[i]))
{
wrbuf_putc(w, '\\');
wrbuf_putc(w, sterm[i]);
}
else
wrbuf_putc(w, sterm[i]);
}
if (trunc == 1)
wrbuf_puts(w, "*");
if (must_quote)
wrbuf_puts(w, "\"");
}
if (ret == 0)
pr(wrbuf_cstr(w), client_data);
}
return ret;
}
static int rpn2cql_simple(cql_transform_t ct,
void (*pr)(const char *buf, void *client_data),
void *client_data,
Z_Operand *q, WRBUF w)
{
int ret = 0;
if (q->which != Z_Operand_APT)
{
ret = -1;
cql_transform_set_error(ct, YAZ_BIB1_RESULT_SET_UNSUPP_AS_A_SEARCH_TERM, 0);
}
else
{
Z_AttributesPlusTerm *apt = q->u.attributesPlusTerm;
Z_Term *term = apt->term;
const char *sterm = 0;
size_t lterm = 0;
Odr_int trunc = lookup_truncation(apt->attributes);
size_t i;
wrbuf_rewind(w);
ret = rpn2cql_attr(ct, apt->attributes, w);
switch (term->which)
{
case Z_Term_general:
lterm = term->u.general->len;
sterm = (const char *) term->u.general->buf;
break;
case Z_Term_numeric:
wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
break;
case Z_Term_characterString:
sterm = term->u.characterString;
lterm = strlen(sterm);
break;
default:
cql_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
return -1;
}
if (trunc <= 3 || trunc == 100 || trunc == 102 || trunc == 104)
{
int quote_it = 0;
for (i = 0 ; i < lterm; i++)
if (strchr(" ()=></", sterm[i]))
{
quote_it = 1;
break;
}
if (lterm == 0)
quote_it = 1;
if (quote_it)
wrbuf_puts(w, "\"");
if (trunc == 2 || trunc == 3)
wrbuf_puts(w, "*");
for (i = 0; i < lterm; i++)
{
if (sterm[i] == '\\' && i < lterm - 1)
{
i++;
if (strchr("*?\"\\", sterm[i]))
wrbuf_putc(w, '\\');
wrbuf_putc(w, sterm[i]);
}
else if (trunc == 102 && sterm[i] == '.' && sterm[i+1] == '*')
{
wrbuf_putc(w, '*');
i++;
}
else if (trunc == 102 && sterm[i] == '.')
wrbuf_putc(w, '?');
else if (trunc == 104 && sterm[i] == '?')
wrbuf_putc(w, '*');
else if (trunc == 104 && sterm[i] == '#')
wrbuf_putc(w, '?');
else if (strchr("*?\"", sterm[i]))
{
wrbuf_putc(w, '\\');
wrbuf_putc(w, sterm[i]);
}
else
wrbuf_putc(w, sterm[i]);
}
if (trunc == 1 || trunc == 3)
wrbuf_puts(w, "*");
if (quote_it)
wrbuf_puts(w, "\"");
}
else
{
cql_transform_set_error(
ct, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, 0);
ret = -1;
}
if (ret == 0)
pr(wrbuf_cstr(w), client_data);
}
return ret;
}
