main.c

harvest是一个下载html网页得机器人

/*
 * (c) Copyright 1993 by Panagiotis Tsirigotis
 * All rights reserved.  The file named COPYRIGHT specifies the terms 
 * and conditions for redistribution.
 */

static char RCSid[] = "main.c,v 1.8 1994/08/16 23:00:20 hardy Exp";
static char version[] = VERSION;

#include <fcntl.h>

void exit();

#include "sio.h"
#include "str.h"

#include "options.h"

#include "defs.h"

int line_count = 1;

typedef void (*ps_process) ();

/*
 * Available processes
 */
void dvi_process();
void psdit_process();
void troff_process();
void ditroff_process();
void wordperfect_process();
void groff_process();

PRIVATE ps_process find_process();


struct process_string {
	char *ps_string;
	char *ps_name;
	ps_process ps_func;
	strs_h ps_sh;
};

static struct process_string procstr[] =
{
	{".dvi", "dvi file", dvi_process},
	{"/wpdict", "wordperfect document", wordperfect_process},
	{"idraw", "idraw", troff_process},
	{"psdit", "ditroff through psdit", psdit_process},
	{"tpscript", "Kolstad's hack", ditroff_process},
	{"ditroff", "ditroff file", ditroff_process},
	{"troff", "troff file", troff_process},
	{"groff", "GNU groff file", groff_process},
	{NULL}
};


struct process_name {
	char *pn_name;
	ps_process pn_func;
};


static struct process_name procnames[] =
{
	{"dvi", dvi_process},
	{"psdit", psdit_process},
	{"dit", ditroff_process},
	{"troff", troff_process},
	{"wp", wordperfect_process},
	{"groff", groff_process},
	{NULL}
};


int main(argc, argv)
int argc;
char *argv[];
{
	int first_arg = opt_recognize(argc, argv);
	int in_fd;
	ps_process process;
	void printout_flush();

	if (d_option)
		Sbuftype(1, SIO_NOBUF);
	/* 
    *  do it anyway; fixes bug on OSF/1.  Panos' Sprint stuff doesn't
    *  seems to work properly with exit().  His flush() operation
    *  never gets called, so the output just sits in the buffering.
    *  So set the library to non-buffering i/o.  Must put this fix
    *  at the top; otherwise, things like the -l option won't work.
    */
	Sbuftype(1, SIO_NOBUF);	

	if (argc - first_arg > 1)	/* at most 1 arg */
		usage();

	if (first_arg == argc)
		in_fd = 0;
	else {
		extern int errno;
		char *file = argv[first_arg];

		in_fd = open(file, O_RDONLY);
		if (in_fd == -1)
			error("Failed to open file %s (errno = %d)\n", file, errno);
	}

	if (l_option) {
		struct process_name *pnp;

		Sprint(1, "Recognized process types:\n");
		for (pnp = procnames; pnp->pn_name; pnp++)
			Sprint(1, "\t%s\n", pnp->pn_name);
		exit(0);
	}

	if (p_option) {
		struct process_name *pnp;

		for (pnp = procnames;; pnp++) {
			if (pnp->pn_name == NULL)
				error("Unknown process: %s\n", p_option_arg);
			if (strcmp(pnp->pn_name, p_option_arg) == 0) {
				process = pnp->pn_func;
				break;
			}
		}
	} else
		process = find_process(in_fd);

	if (!n_option) {
		(*process) (in_fd);
		printout_flush();
	}
	exit(0);
	/* NOTREACHED */
}

/*
 * Determine what process to use to parse this file.
 */
PRIVATE ps_process find_process(fd)
int fd;
{
	char *line;
	struct process_string *psp;
	struct process_string *found = NULL;

	/*
	 * Create string matchers
	 */
	for (psp = procstr; psp->ps_string; psp++) {
		psp->ps_sh = strs_setup(STRS_BF + STRS_NOMALLOC, psp->ps_string);
		if (psp->ps_sh == NULL)
			error("out of memory\n");
	}

	/*
	 * Try to determine how the file was derived.
	 * Method:
	 *              Read lines looking for one of the identifier strings
	 *              If an identifier string is found, continue reading lines
	 *      until a non-comment line is found.
	 */
	while (line = Srdline(fd)) {
		line_count++;
		if (line[0] != '%' && found)
			break;

		/*
		 * Check what process (if any) is identified by this line
		 */
		for (psp = procstr; psp->ps_string; psp++)
			if (strs_match(psp->ps_sh, line, SIOLINELEN(fd))) {
				/*
				 * psdit process takes precedence over ditroff_process
				 */
				if (!found || found->ps_func == ditroff_process &&
				    psp->ps_func == psdit_process)
					found = psp;
			}
	}
	if (!found && !e_option) {
		error("Failed to identify how this file was derived\n");
	}
	if (v_option)
		Sprint(2, "Postscript file derived from %s\n", found->ps_name);
	if (e_option)
		exit(!found);
	return (found->ps_func);
}