zebrash.c

harvest是一个下载html网页得机器人

    return rc;
}

static int cmd_find( char *args[], WRBUF outbuff)
{
    char *setname=DEFAULTRESULTSET;
    int rc;
    int hits=0;
    WRBUF qry=wrbuf_alloc();
    if (0==strstr(args[0],"@attr"))
        wrbuf_puts(qry, "@attr 1=/ ");
    wrbuf_puts(qry,restargs(args,1));
    if (!zh)
	onecommand("quickstart", outbuff, "");
    wrbuf_printf(outbuff, "find %s\n",wrbuf_buf(qry));
    rc=zebra_search_PQF(zh, wrbuf_buf(qry), setname, &hits);
    if (0==rc)
    {
        wrbuf_printf(outbuff,"%d hits found\n",hits);
        nextrecno=1;
    }
    wrbuf_free(qry,1);
    return rc;
}

static int cmd_show( char *args[], WRBUF outbuff)
{
    int start=defargint(args[1], nextrecno);
    int nrecs=defargint(args[2],1);
    char *setname=defarg(args[3],DEFAULTRESULTSET);
    int rc=0;
    ZebraRetrievalRecord *recs;
    ODR odr;
    Z_RecordComposition *pcomp=0;
    int i;
    oid_value format;

    odr=odr_createmem(ODR_ENCODE);
    recs= odr_malloc(odr,sizeof(ZebraRetrievalRecord)*nrecs);
    rc =z_RecordComposition(odr, &pcomp, 0,"recordComposition");
    format=oid_getvalbyname ("xml"); /*FIXME - let the user specify*/
    for (i=0;i<nrecs;i++)
        recs[i].position=start+i;

    rc = zebra_records_retrieve (zh, odr, setname,
				 pcomp, format, nrecs,recs);
    if (0==rc)
    {
        for (i=0;i<nrecs;i++)
        {
            printf("Err %d: %d\n",i,recs[i].errCode);
            if (recs[i].buf)
            {
                wrbuf_printf(outbuff,"Record %d\n", recs[i].position);
                wrbuf_write(outbuff, recs[i].buf, recs[i].len);
                wrbuf_puts(outbuff, "\n");
            } else
                wrbuf_printf(outbuff,"NO Record %d\n", recs[i].position);
        }
        nextrecno=start+nrecs+1;
    }
    odr_destroy(odr);
    return rc;
}
/**************************************)
 * Command table, parser, and help 
 */

struct cmdstruct
{
    char *cmd;
    char *args;
    char *explanation;
    int (*testfunc)(char *args[], WRBUF outbuff);
} ;

 
struct cmdstruct cmds[] = {
    /* special cases:
     *   if text is 0, does not list the command
     *   if cmd is "", adds the args (and newline) in command listing
     */
    { "", "Starting and stopping:", "", 0 },
    { "zebra_start", 
      "[configfile]", 
      "starts the zebra service. You need to call this first\n"
      "if no configfile is given, assumes " DEFAULTCONFIG, 
      cmd_zebra_start },
    { "zebra_stop",   "", 
      "stops the zebra service", 
      cmd_zebra_stop },
    { "zebra_open", "",  
      "starts a zebra session. Once you have called zebra_start\n"
      "you can call zebra_open to start working", 
      cmd_zebra_open },
    { "zebra_close", "", 
      "closes a zebra session", 
      cmd_zebra_close }, 
    { "quickstart", "[configfile]", 
      "Does a zebra_start, zebra_open, and sets up the log", 
      cmd_quickstart }, 
  
    { "", "Log file:","", 0},  
    { "yaz_log_file", 
      "[filename]",
      "Directs the log to filename (or stderr)",
      cmd_yaz_log_file },
    { "yaz_log_level", 
      "[level]",
      "Sets the logging level (or returns to default)",
      cmd_yaz_log_level },
    { "yaz_log_prefix", 
      "[prefix]",
      "Sets the log prefix",
      cmd_yaz_log_prefix},    
    { "logf", 
      "[level] text...",
      "writes an entry in the log",
      cmd_logf},    

    { "", "Error handling:","", 0},
    { "err",  "",
      "Displays zebra's error status (code, str, add)",
      cmd_err},    
    { "errcode",  "",
      "Displays zebra's error code",
      cmd_errcode},    
    { "errstr",  "",
      "Displays zebra's error string",
      cmd_errstr},    
    { "erradd",  "",
      "Displays zebra's additional error message",
      cmd_erradd},    
  
    { "", "Admin:","", 0}, 
    { "init",  "",
      "Initializes the zebra database, destroying all data in it",
      cmd_init},    
    { "select_database",  "basename",
      "Selects a database",
      cmd_select_database},    
    { "create_database", "basename",
      "Create database",
      cmd_create_database},
    { "drop_database", "basename",
      "Drop database",
      cmd_drop_database},
    { "begin_trans", "[rw]",
      "Begins a transaction. rw=1 means write, otherwise read-only",
      cmd_begin_trans},
    { "end_trans","",
      "Ends a transaction",
      cmd_end_trans},

    { "","Updating:","",0},
    { "record_insert","record",
      "inserts an sgml record into Default",
      cmd_record_insert},
    { "exchange_record","database record-id action record",
      "inserts (1), updates (2), or deletes (3) a record \n"
      "record-id must be a unique identifier for the record",
      cmd_exchange_record},
    { "","Searching and retrieving:","",0},
    { "search_pqf","setname query",
      "search ",
      cmd_search_pqf},
    { "find","query",
      "simplified search",
      cmd_find},
    { "f","query",
      "simplified search",
      cmd_find},
    { "show","[start] [numrecs] [resultset]",
      "shows a result",
      cmd_show},
    { "s","[start] [numrecs] [resultset]",
      "shows a result",
      cmd_show},
    { "", "Misc:","", 0}, 
    { "echo", "string", 
      "ouputs the string", 
      cmd_echo },
    { "q", "", 
      "exits the program", 
      cmd_quit },
    { "quit", "", 
      "exits the program", 
      cmd_quit },
    { "help", "[command]", 
      "Gives help on command, or lists them all", 
      cmd_help },
    { "", "help [command] gives more info on command", "",0 },   
  
    {0,0,0,0} /* end marker */
};
 
int onecommand( 
		char *line,     /* input line */
		WRBUF outbuff,  /* output goes here */
		const char *prevout) /* prev output, for 'expect' */
{
    int i;
    char *args[MAX_NO_ARGS];
    int nargs;
    char argbuf[MAX_ARG_LEN];
    logf(LOG_APP,"%s",line);
    strncpy(argbuf,line, MAX_ARG_LEN-1);
    argbuf[MAX_ARG_LEN-1]='\0'; /* just to be sure */
    /*memset(args,'\0',MAX_NO_ARGS*sizeof(char *));*/
    nargs=split_args(argbuf, args);
    
#if 0
    for (i = 0; i <= n; i++)
    {
	const char *cp = args[i];
	printf ("args %d :%s:\n", i, cp ? cp : "<null>");
    }
#endif
    if (0==nargs)
	    return -90; /* no command on line, too bad */

    if (0==strcmp(args[0],"expect")) 
    {
	char *rest;
        if (nargs>1) /* args[0] is not yet set, can't use restargs */
            rest= line + (args[1]-argbuf); /* rest of the line */
        else
            return -1; /* need something to expect */
	if (0==strstr(prevout,rest))
	{
	    printf( "Failed expectation, '%s' not found\n", rest);
            exit(9); 
	}
	return 0;
    }
    for (i=0;cmds[i].cmd;i++)
	if (0==strcmp(cmds[i].cmd, args[0])) 
	{
	    if (nargs>1)
		args[0]= line + (args[1]-argbuf); /* rest of the line */
	    else
		args[0]=""; 
	    return ((cmds[i].testfunc)(args,outbuff));
	}
    wrbuf_printf(outbuff, "Unknown command '%s'. Try help\n",args[0]);
    logf(LOG_APP,"Unknown command");
    return -90; 
}
 
static int cmd_help( char *args[], WRBUF outbuff)
{ 
    int i;
    int linelen;
    if (args[1]) 
    { /* help for a single command */ 
	for (i=0;cmds[i].cmd;i++)
	    if (0==strcmp(cmds[i].cmd, args[1])) 
	    {
                wrbuf_printf(outbuff,"%s  %s\n%s\n",
		             cmds[i].cmd, cmds[i].args, 
			     cmds[i].explanation);
		return 0;
	    }
	wrbuf_printf(outbuff, "Unknown command '%s'", args[1]);
    }
    else 
    { /* list all commands */
        linelen=9999;
	for (i=0;cmds[i].cmd;i++)
        {
            if (*cmds[i].cmd)
            { /* ordinary command */
                if (linelen>50)
                {
                    wrbuf_puts(outbuff,"\n   ");
                    linelen=0;
                }
                linelen += strlen(cmds[i].cmd) + 2;
                wrbuf_printf(outbuff,"%s ", cmds[i].cmd);
            } else
            { /* section head */
                wrbuf_printf(outbuff,"\n%s\n   ",cmds[i].args);
                linelen=0;
            }
	    } /* for */
        wrbuf_puts(outbuff,"\n");
    }
    return 0;
}
 
/* If Zebra reports an error after an operation,
 * append it to the outbuff and log it */
static void Zerrors ( WRBUF outbuff)
{
    int ec;
    if (!zh)
	return ;
    ec=zebra_errCode (zh);
    if (ec)
    {
	logf(LOG_APP, "   Zebra error %d: %s, (%s)",
	     ec, zebra_errString (zh),
	     zebra_errAdd (zh) );
	wrbuf_printf(outbuff, "   Zebra error %d: %s, (%s)\n",
		     ec, zebra_errString (zh),
		     zebra_errAdd (zh) );
	zebra_clearError(zh);
    }
}

/************************************** 
 * The shell
 */
 
void shell()
{
    int rc=0;
    WRBUF outbuff=wrbuf_alloc();
    char prevout[MAX_OUT_BUFF]=""; /* previous output for 'expect' */
    wrbuf_puts(outbuff,"Zebrash at your service");
    while (rc!=-99)
    {
	char *nl_cp;
	char buf[MAX_ARG_LEN];
	char* line_in = 0;
#if HAVE_READLINE_READLINE_H
	if (isatty(0)) {
	    line_in=readline(PROMPT);
	    if (!line_in)
		break;
#if HAVE_READLINE_HISTORY_H
	    if (*line_in)
		add_history(line_in);
#endif
	}
#endif
	/* line_in != NULL if readine is present and input is a tty */
	
	printf (PROMPT); 
	fflush (stdout);
	if (line_in)
	{
	    if(strlen(line_in) > MAX_ARG_LEN-1) {
		fprintf(stderr,"Input line too long\n");
		break;
	    }
	    strcpy(buf,line_in);
	    free (line_in);
	}
	else 
	{
	    if (!fgets (buf, MAX_ARG_LEN-1, stdin))
		break; 
	}
	
	/* get rid of \n in line */
	if ((nl_cp = strchr(buf, '\n')))
	    *nl_cp = '\0';
	strncpy(prevout, wrbuf_buf(outbuff), MAX_OUT_BUFF);
        wrbuf_rewind(outbuff);
	rc=onecommand(buf, outbuff, prevout);
	if (rc==0)
	{
	    wrbuf_puts(outbuff, "   OK\n");
	    logf(LOG_APP, "OK");
	}
	else if (rc>-90)
	{
	    wrbuf_printf(outbuff, "   command returned %d\n",rc);
	} 
	Zerrors(outbuff);
	printf("%s\n", wrbuf_buf(outbuff));
    } /* while */
    wrbuf_free(outbuff,1);
} /* shell() */


/**************************************
 * Main 
 */
 
int main (int argc, char ** args)
{
    shell();
    return 0;
} /* main */