data1.h

harvest是一个下载html网页得机器人

	struct 
	{
	    char *tag;
	    data1_element *element;
	    int no_data_requested;
	    int get_bytes;
	    unsigned node_selected : 1;
	    unsigned make_variantlist : 1;
            data1_xattr *attributes;
	} tag;

	struct
	{
	    char *data;      /* filename or data */
	    int len;
            /* text inclusion */
#define DATA1I_inctxt 1
            /* binary data inclusion */
#define DATA1I_incbin 2
        /* text data */
#define DATA1I_text 3 
            /* numerical data */
#define DATA1I_num 4
            /* object identifier */
#define DATA1I_oid 5
            /* XML text */
#define DATA1I_xmltext 6
            unsigned what:7;
	    unsigned formatted_text : 1;   /* newlines are significant */
	} data;

	struct
	{
	    data1_vartype *type;
	    char *value;
	} variant;

        struct
        {
            char *target;
            data1_xattr *attributes;
        } preprocess;
    } u;

    void (*destroy)(struct data1_node *n);
#define DATA1_LOCALDATA 12
    char lbuf[DATA1_LOCALDATA]; /* small buffer for local data */
    struct data1_node *next;
    struct data1_node *child;
    struct data1_node *last_child;
    struct data1_node *parent;
    struct data1_node *root;
} data1_node;

YAZ_EXPORT data1_handle data1_create (void);

#define DATA1_FLAG_XML  1
YAZ_EXPORT data1_handle data1_createx (int flags);

YAZ_EXPORT void data1_destroy(data1_handle dh);
YAZ_EXPORT data1_node *get_parent_tag(data1_handle dh, data1_node *n);
YAZ_EXPORT data1_node *data1_read_node(data1_handle dh, const char **buf,
				       NMEM m);
YAZ_EXPORT data1_node *data1_read_nodex (data1_handle dh, NMEM m,
					 int (*get_byte)(void *fh), void *fh,
					 WRBUF wrbuf);
YAZ_EXPORT data1_node *data1_read_record(data1_handle dh, 
					 int (*rf)(void *, char *, size_t),
					 void *fh, NMEM m);
YAZ_EXPORT data1_absyn *data1_read_absyn(data1_handle dh, const char *file,
                                         int file_must_exist);
YAZ_EXPORT data1_tag *data1_gettagbynum(data1_handle dh,
					data1_tagset *s,
					int type, int value);
YAZ_EXPORT data1_tagset *data1_empty_tagset (data1_handle dh);
YAZ_EXPORT data1_tagset *data1_read_tagset(data1_handle dh, 
					   const char *file,
					   int type);
YAZ_EXPORT data1_element *data1_getelementbytagname(data1_handle dh, 
						    data1_absyn *abs,
						    data1_element *parent,
						    const char *tagname);
YAZ_EXPORT Z_GenericRecord *data1_nodetogr(data1_handle dh, data1_node *n,
					   int select, ODR o,
					   int *len);
YAZ_EXPORT data1_tag *data1_gettagbyname(data1_handle dh, data1_tagset *s,
					 const char *name);
YAZ_EXPORT void data1_free_tree(data1_handle dh, data1_node *t);
YAZ_EXPORT char *data1_nodetobuf(data1_handle dh, data1_node *n,
				 int select, int *len);
YAZ_EXPORT data1_node *data1_mk_tag_data_wd(data1_handle dh,
                                            data1_node *at,
                                            const char *tagname, NMEM m);
YAZ_EXPORT data1_node *data1_mk_tag_data(data1_handle dh, data1_node *at,
                                         const char *tagname, NMEM m);
YAZ_EXPORT data1_datatype data1_maptype(data1_handle dh, char *t);
YAZ_EXPORT data1_varset *data1_read_varset(data1_handle dh, const char *file);
YAZ_EXPORT data1_vartype *data1_getvartypebyct(data1_handle dh,
					       data1_varset *set,
					       char *zclass, char *type);
YAZ_EXPORT Z_Espec1 *data1_read_espec1(data1_handle dh, const char *file);
YAZ_EXPORT int data1_doespec1(data1_handle dh, data1_node *n, Z_Espec1 *e);
YAZ_EXPORT data1_esetname *data1_getesetbyname(data1_handle dh, 
					       data1_absyn *a,
					       const char *name);
YAZ_EXPORT data1_element *data1_getelementbyname(data1_handle dh,
						 data1_absyn *absyn,
						 const char *name);
YAZ_EXPORT data1_node *data1_mk_node2(data1_handle dh, NMEM m,
                                      int type, data1_node *parent);

YAZ_EXPORT data1_node *data1_mk_tag (data1_handle dh, NMEM nmem, 
                                     const char *tag, const char **attr,
                                     data1_node *at);
YAZ_EXPORT data1_node *data1_mk_tag_n (data1_handle dh, NMEM nmem,
                                       const char *tag, size_t len,
                                       const char **attr,
                                       data1_node *at);
YAZ_EXPORT void data1_tag_add_attr (data1_handle dh, NMEM nmem,
                                    data1_node *res, const char **attr);

YAZ_EXPORT data1_node *data1_mk_text_n (data1_handle dh, NMEM mem,
                                        const char *buf, size_t len,
                                        data1_node *parent);
YAZ_EXPORT data1_node *data1_mk_text_nf (data1_handle dh, NMEM mem,
                                         const char *buf, size_t len,
                                         data1_node *parent);
YAZ_EXPORT data1_node *data1_mk_text (data1_handle dh, NMEM mem,
                                      const char *buf, data1_node *parent);

YAZ_EXPORT data1_node *data1_mk_comment_n (data1_handle dh, NMEM mem,
                                           const char *buf, size_t len,
                                           data1_node *parent);

YAZ_EXPORT data1_node *data1_mk_comment (data1_handle dh, NMEM mem,
                                         const char *buf, data1_node *parent);

YAZ_EXPORT data1_node *data1_mk_preprocess_n (data1_handle dh, NMEM nmem,
                                              const char *target, size_t len,
                                              const char **attr,
                                              data1_node *at);

YAZ_EXPORT data1_node *data1_mk_preprocess (data1_handle dh, NMEM nmem,
                                            const char *target,
                                            const char **attr,
                                            data1_node *at);

YAZ_EXPORT data1_node *data1_mk_root (data1_handle dh, NMEM nmem,
                                      const char *name);
YAZ_EXPORT void data1_set_root(data1_handle dh, data1_node *res,
                               NMEM nmem, const char *name);

YAZ_EXPORT data1_node *data1_mk_tag_data_int (data1_handle dh, data1_node *at,
                                              const char *tag, int num,
                                              NMEM nmem);
YAZ_EXPORT data1_node *data1_mk_tag_data_oid (data1_handle dh, data1_node *at,
                                              const char *tag, Odr_oid *oid,
                                              NMEM nmem);
YAZ_EXPORT data1_node *data1_mk_tag_data_text (data1_handle dh, data1_node *at,
                                               const char *tag,
                                               const char *str,
                                               NMEM nmem);
YAZ_EXPORT data1_node *data1_mk_tag_data_text_uni (data1_handle dh,
                                                   data1_node *at,
                                                   const char *tag,
                                                   const char *str,
                                                   NMEM nmem);

YAZ_EXPORT data1_absyn *data1_get_absyn (data1_handle dh, const char *name);

YAZ_EXPORT data1_node *data1_search_tag (data1_handle dh, data1_node *n,
                                         const char *tag);
YAZ_EXPORT data1_node *data1_mk_tag_uni (data1_handle dh, NMEM nmem, 
                                         const char *tag, data1_node *at);
YAZ_EXPORT data1_attset *data1_get_attset (data1_handle dh, const char *name);
YAZ_EXPORT data1_maptab *data1_read_maptab(data1_handle dh, const char *file);
YAZ_EXPORT data1_node *data1_map_record(data1_handle dh, data1_node *n,
					data1_maptab *map, NMEM m);
YAZ_EXPORT data1_marctab *data1_read_marctab (data1_handle dh,
					      const char *file);
YAZ_EXPORT char *data1_nodetomarc(data1_handle dh, data1_marctab *p,
				  data1_node *n, int selected, int *len);
YAZ_EXPORT char *data1_nodetoidsgml(data1_handle dh, data1_node *n,
				    int select, int *len);
YAZ_EXPORT Z_ExplainRecord *data1_nodetoexplain(data1_handle dh,
						data1_node *n, int select,
						ODR o);
YAZ_EXPORT Z_BriefBib *data1_nodetosummary(data1_handle dh, 
					   data1_node *n, int select,
					   ODR o);
YAZ_EXPORT char *data1_nodetosoif(data1_handle dh, data1_node *n, int select,
				  int *len);
YAZ_EXPORT void data1_set_tabpath(data1_handle dh, const char *path);
YAZ_EXPORT void data1_set_tabroot (data1_handle dp, const char *p);
YAZ_EXPORT const char *data1_get_tabpath(data1_handle dh);
YAZ_EXPORT const char *data1_get_tabroot(data1_handle dh);

YAZ_EXPORT WRBUF data1_get_wrbuf (data1_handle dp);
YAZ_EXPORT char **data1_get_read_buf (data1_handle dp, int **lenp);
YAZ_EXPORT char **data1_get_map_buf (data1_handle dp, int **lenp);
YAZ_EXPORT data1_absyn_cache *data1_absyn_cache_get (data1_handle dh);
YAZ_EXPORT data1_attset_cache *data1_attset_cache_get (data1_handle dh);
YAZ_EXPORT NMEM data1_nmem_get (data1_handle dh);
YAZ_EXPORT void data1_pr_tree (data1_handle dh, data1_node *n, FILE *out);
YAZ_EXPORT char *data1_insert_string (data1_handle dh, data1_node *res,
				      NMEM m, const char *str);
YAZ_EXPORT char *data1_insert_string_n (data1_handle dh, data1_node *res,
                                        NMEM m, const char *str, size_t len);
YAZ_EXPORT data1_node *data1_read_sgml (data1_handle dh, NMEM m,
					const char *buf);
YAZ_EXPORT data1_node *data1_read_xml (data1_handle dh,
                                       int (*rf)(void *, char *, size_t),
                                       void *fh, NMEM m);
YAZ_EXPORT void data1_absyn_trav (data1_handle dh, void *handle,
				  void (*fh)(data1_handle dh,
					     void *h, data1_absyn *a));

YAZ_EXPORT data1_attset *data1_attset_search_id (data1_handle dh, int id);

YAZ_EXPORT char *data1_getNodeValue(data1_node* node, char* pTagPath);
YAZ_EXPORT data1_node *data1_LookupNode(data1_node* node, char* pTagPath);
YAZ_EXPORT int data1_CountOccurences(data1_node* node, char* pTagPath);

YAZ_EXPORT FILE *data1_path_fopen (data1_handle dh, const char *file,
                                   const char *mode);

/* obsolete functions ... */

YAZ_EXPORT data1_node *data1_mk_node (data1_handle dh, NMEM m);
YAZ_EXPORT data1_node *data1_insert_taggeddata (data1_handle dh,
                                                data1_node *root,
                                                data1_node *at,
                                                const char *tagname, NMEM m);
YAZ_EXPORT data1_node *data1_mk_node_type (data1_handle dh, NMEM m, int type);
YAZ_EXPORT data1_node *data1_add_taggeddata (data1_handle dh, data1_node *root,
                                             data1_node *at,
                                             const char *tagname,
                                             NMEM m);

YAZ_EXPORT data1_node *data1_get_root_tag (data1_handle dh, data1_node *n);

YAZ_EXPORT int data1_iconv (data1_handle dh, NMEM m, data1_node *n,
                            const char *tocode, 
                            const char *fromcode);

YAZ_EXPORT const char *data1_get_encoding (data1_handle dh, data1_node *n);

YAZ_EXPORT int data1_is_xmlmode(data1_handle dh);

YAZ_EXPORT const char *data1_systag_lookup(data1_absyn *absyn, const char *tag,
                                           const char *default_value);

YAZ_EXPORT void data1_concat_text(data1_handle dh, NMEM m, data1_node *n);

YAZ_EXPORT void data1_absyn_destroy(data1_handle dh);

YAZ_END_CDECL

#endif