pquery.c

harvest是一个下载html网页得机器人

        p->exclusion = NULL;

    if (!lex (li))
    {
        li->error = YAZ_PQF_ERROR_MISSING;
        return NULL;
    }
    p->distance = (int *)odr_malloc (o, sizeof(*p->distance));
    *p->distance = atoi (li->lex_buf);

    if (!lex (li))
    {
        li->error = YAZ_PQF_ERROR_MISSING;
        return NULL;
    }
    p->ordered = (int *)odr_malloc (o, sizeof(*p->ordered));
    *p->ordered = atoi (li->lex_buf);
    
    if (!lex (li))
    {
        li->error = YAZ_PQF_ERROR_MISSING;
        return NULL;
    }
    p->relationType = (int *)odr_malloc (o, sizeof(*p->relationType));
    *p->relationType = atoi (li->lex_buf);

    if (!lex (li))
    {
        li->error = YAZ_PQF_ERROR_MISSING;
        return NULL;
    }
    if (*li->lex_buf == 'k')
        p->which = 0;
    else if (*li->lex_buf == 'p')
        p->which = 1;
    else
        p->which = atoi (li->lex_buf);

    if (!lex (li))
    {
        li->error = YAZ_PQF_ERROR_MISSING;
        return NULL;
    }
    p->which = Z_ProximityOperator_known;
    p->u.known = (int *)odr_malloc (o, sizeof(*p->u.known));
    *p->u.known = atoi (li->lex_buf);
    return p;
}

static Z_Complex *rpn_complex (struct yaz_pqf_parser *li, ODR o, oid_proto proto,
                               int num_attr, int max_attr, 
                               int *attr_list, char **attr_clist,
			       oid_value *attr_set)
{
    Z_Complex *zc;
    Z_Operator *zo;

    zc = (Z_Complex *)odr_malloc (o, sizeof(*zc));
    zo = (Z_Operator *)odr_malloc (o, sizeof(*zo));
    zc->roperator = zo;
    switch (li->query_look)
    {
    case 'a':
        zo->which = Z_Operator_and;
        zo->u.and_not = odr_nullval();
        break;
    case 'o':
        zo->which = Z_Operator_or;
        zo->u.and_not = odr_nullval();
        break;
    case 'n':
        zo->which = Z_Operator_and_not;
        zo->u.and_not = odr_nullval();
        break;
    case 'p':
        zo->which = Z_Operator_prox;
        zo->u.prox = rpn_proximity (li, o);
        if (!zo->u.prox)
            return NULL;
        break;
    default:
        /* we're only called if one of the above types are seens so
           this shouldn't happen */
        li->error = YAZ_PQF_ERROR_INTERNAL;
        return NULL;
    }
    lex (li);
    if (!(zc->s1 =
          rpn_structure (li, o, proto, num_attr, max_attr, attr_list,
			 attr_clist, attr_set)))
        return NULL;
    if (!(zc->s2 =
          rpn_structure (li, o, proto, num_attr, max_attr, attr_list,
                         attr_clist, attr_set)))
        return NULL;
    return zc;
}

static void rpn_term_type (struct yaz_pqf_parser *li, ODR o)
{
    if (!li->query_look)
        return ;
    if (compare_term (li, "general", 0))
        li->term_type = Z_Term_general;
    else if (compare_term (li, "numeric", 0))
        li->term_type = Z_Term_numeric;
    else if (compare_term (li, "string", 0))
        li->term_type = Z_Term_characterString;
    else if (compare_term (li, "oid", 0))
        li->term_type = Z_Term_oid;
    else if (compare_term (li, "datetime", 0))
        li->term_type = Z_Term_dateTime;
    else if (compare_term (li, "null", 0))
        li->term_type = Z_Term_null;
    else if (compare_term(li, "range", 0))
    {
	/* prepare for external: range search .. */
        li->term_type = Z_Term_external;
	li->external_type = VAL_MULTISRCH2;
    }
    lex (li);
}
                           
static Z_RPNStructure *rpn_structure (struct yaz_pqf_parser *li, ODR o,
                                      oid_proto proto, 
                                      int num_attr, int max_attr, 
                                      int *attr_list,
				      char **attr_clist,
				      oid_value *attr_set)
{
    Z_RPNStructure *sz;

    sz = (Z_RPNStructure *)odr_malloc (o, sizeof(*sz));
    switch (li->query_look)
    {
    case 'a':
    case 'o':
    case 'n':
    case 'p':
        sz->which = Z_RPNStructure_complex;
        if (!(sz->u.complex =
              rpn_complex (li, o, proto, num_attr, max_attr, attr_list,
                           attr_clist, attr_set)))
            return NULL;
        break;
    case 't':
    case 's':
        sz->which = Z_RPNStructure_simple;
        if (!(sz->u.simple =
              rpn_simple (li, o, proto, num_attr, attr_list,
                          attr_clist, attr_set)))
            return NULL;
        break;
    case 'l':
        lex (li);
        if (!li->query_look)
        {
            li->error = YAZ_PQF_ERROR_MISSING;
            return 0;
        }
        if (num_attr >= max_attr)
        {
            li->error = YAZ_PQF_ERROR_TOOMANY;
            return 0;
        }
	if (!p_query_parse_attr(li, o, num_attr, attr_list,
                                attr_clist, attr_set))
            return 0;
	num_attr++;
        lex (li);
        return
            rpn_structure (li, o, proto, num_attr, max_attr, attr_list,
			   attr_clist,  attr_set);
    case 'y':
	lex (li);
        rpn_term_type (li, o);
        return
            rpn_structure (li, o, proto, num_attr, max_attr, attr_list,
			   attr_clist, attr_set);
    case 0:                /* operator/operand expected! */
        li->error = YAZ_PQF_ERROR_MISSING;
        return 0;
    }
    return sz;
}

Z_RPNQuery *p_query_rpn_mk (ODR o, struct yaz_pqf_parser *li, oid_proto proto,
                            const char *qbuf)
{
    Z_RPNQuery *zq;
    int attr_array[1024];
    char *attr_clist[512];
    oid_value attr_set[512];
    oid_value topSet = VAL_NONE;

    zq = (Z_RPNQuery *)odr_malloc (o, sizeof(*zq));
    lex (li);
    if (li->query_look == 'r')
    {
        lex (li);
        topSet = query_oid_getvalbyname (li);
        if (topSet == VAL_NONE)
        {
            li->error = YAZ_PQF_ERROR_ATTSET;
            return NULL;
        }

        lex (li);
    }
    if (topSet == VAL_NONE)
        topSet = p_query_dfset;
    if (topSet == VAL_NONE)
        topSet = VAL_BIB1;

    zq->attributeSetId = yaz_oidval_to_z3950oid(o, CLASS_ATTSET, topSet);

    if (!zq->attributeSetId)
    {
        li->error = YAZ_PQF_ERROR_ATTSET;
	return 0;
    }

    if (!(zq->RPNStructure = rpn_structure (li, o, proto, 0, 512,
                                            attr_array, attr_clist, attr_set)))
        return 0;
    if (li->query_look)
    {
        li->error = YAZ_PQF_ERROR_EXTRA;
        return 0;
    }
    return zq;
}

Z_RPNQuery *p_query_rpn (ODR o, oid_proto proto,
                         const char *qbuf)
{
    struct yaz_pqf_parser li;

    li.error = 0;
    li.left_sep = "{\"";
    li.right_sep = "}\"";
    li.escape_char = '@';
    li.term_type = Z_Term_general;
    li.query_buf = li.query_ptr = qbuf;
    li.lex_buf = 0;
    return p_query_rpn_mk (o, &li, proto, qbuf);
}


Z_AttributesPlusTerm *p_query_scan_mk (struct yaz_pqf_parser *li,
                                       ODR o, oid_proto proto,
                                       Odr_oid **attributeSetP,
                                       const char *qbuf)
{
    int attr_list[1024];
    char *attr_clist[512];
    oid_value attr_set[512];
    int num_attr = 0;
    int max_attr = 512;
    oid_value topSet = VAL_NONE;
    Z_AttributesPlusTerm *apt;

    lex (li);
    if (li->query_look == 'r')
    {
        lex (li);
        topSet = query_oid_getvalbyname (li);

        lex (li);
    }
    if (topSet == VAL_NONE)
        topSet = p_query_dfset;
    if (topSet == VAL_NONE)
        topSet = VAL_BIB1;

    *attributeSetP = yaz_oidval_to_z3950oid (o, CLASS_ATTSET, topSet);

    while (1)
    {
        if (li->query_look == 'l')
        {
            lex (li);
            if (!li->query_look)
            {
                li->error = YAZ_PQF_ERROR_MISSING;
                return 0;
            }
            if (num_attr >= max_attr)
            {
                li->error = YAZ_PQF_ERROR_TOOMANY;
                return 0;
            }
            if (!p_query_parse_attr(li, o, num_attr, attr_list,
                                    attr_clist, attr_set))
                return 0;
            num_attr++;
            lex (li);
        }
        else if (li->query_look == 'y')
        {
            lex (li);
            rpn_term_type (li, o);
        }
        else
            break;
    }
    if (!li->query_look)
    {
        li->error = YAZ_PQF_ERROR_MISSING;
        return 0;
    }
    apt = rpn_term (li, o, proto, num_attr, attr_list, attr_clist, attr_set);

    lex (li);

    if (li->query_look != 0)
    {
        li->error = YAZ_PQF_ERROR_EXTRA;
        return 0;
    }
    return apt;
}

Z_AttributesPlusTerm *p_query_scan (ODR o, oid_proto proto,
                                    Odr_oid **attributeSetP,
                                    const char *qbuf)
{
    struct yaz_pqf_parser li;

    li.error = 0;
    li.left_sep = "{\"";
    li.right_sep = "}\"";
    li.escape_char = '@';
    li.term_type = Z_Term_general;
    li.query_buf = li.query_ptr = qbuf;
    li.lex_buf = 0;

    return p_query_scan_mk (&li, o, proto, attributeSetP, qbuf);
}

int p_query_attset (const char *arg)
{
    p_query_dfset = oid_getvalbyname (arg);
    return (p_query_dfset == VAL_NONE) ? -1 : 0;
}

YAZ_PQF_Parser yaz_pqf_create (void)
{
    YAZ_PQF_Parser p = (YAZ_PQF_Parser) xmalloc (sizeof(*p));

    p->error = 0;
    p->left_sep = "{\"";
    p->right_sep = "}\"";
    p->escape_char = '@';
    p->term_type = Z_Term_general;

    return p;
}

void yaz_pqf_destroy (YAZ_PQF_Parser p)
{
    xfree (p);
}

Z_RPNQuery *yaz_pqf_parse (YAZ_PQF_Parser p, ODR o, const char *qbuf)
{
    if (!p)
        return 0;
    p->query_buf = p->query_ptr = qbuf;
    p->lex_buf = 0;
    return p_query_rpn_mk (o, p, PROTO_Z3950, qbuf);
}

Z_AttributesPlusTerm *yaz_pqf_scan (YAZ_PQF_Parser p, ODR o,
                                    Odr_oid **attributeSetP,
                                    const char *qbuf)
{
    if (!p)
        return 0;
    p->query_buf = p->query_ptr = qbuf;
    p->lex_buf = 0;
    return p_query_scan_mk (p, o, PROTO_Z3950, attributeSetP, qbuf);
}

int yaz_pqf_error (YAZ_PQF_Parser p, const char **msg, size_t *off)
{
    switch (p->error)
    {
    case YAZ_PQF_ERROR_NONE:
        *msg = "no error"; break;
    case YAZ_PQF_ERROR_EXTRA:
        *msg = "extra token"; break;
    case YAZ_PQF_ERROR_MISSING:
        *msg = "missing token"; break;
    case YAZ_PQF_ERROR_ATTSET:
        *msg = "unknown attribute set"; break;
    case YAZ_PQF_ERROR_TOOMANY:
        *msg = "too many attributes"; break;
    case YAZ_PQF_ERROR_BADATTR:
        *msg = "bad attribute specification"; break;
    case YAZ_PQF_ERROR_INTERNAL:
        *msg = "internal error"; break;
    default:
        *msg = "unknown error"; break;
    }
    *off = p->query_ptr - p->query_buf;
    return p->error;
}