dvi2tty.c

harvest是一个下载html网页得机器人

/******************************************************************************
 * Kang-Jin Lee: 23sep2000 ([e-mail])
 *               Don't append ".dvi" to dvi file name without ".dvi" extension.
 * Kang-Jin Lee: 26jan2000 ([e-mail])
 *               Minor changes to integrate with Harvest.
 * Marcel Mol: 27may1996  (Internet: [e-mail])
 *               From: [e-mail] (Manuel Lemos): Port to AMIGA.
 * Marcel Mol: 24mar1996  (Internet: [e-mail])
 *               From: Alex Viskovatoff <[e-mail]>: Port to Macintosh
 *               From: Anton vd Repe  1995-6-9     (UUCP: [e-mail]):
 *               Added option to skip translation of Scandinavian/Finnish
 *               and skip the constraint to ASCII characters only.
 *               Added option to include font name in output when new font
 *               is chosen.
 *               Limit line width of 132 expanded to 332.
 * Marcel Mol: 23jun1995  (Internet: [e-mail])
 *               Included patch form Hideki ISOZAKI ([e-mail])
 *               to allow japanese fonts generated by NTT JTeX (March 1992).
 *               Added cleanups (protoize + includes) by D.U. Zoller
 *               (1995-07-17 [e-mail])
 *               Version 5.1
 * Marcel Mol: 09sep1993  (Internet: [e-mail])
 *               In dvistuff.c: added auto detection for 'ttfont'.
 *               This make -t flag unneeded, but it is still left in for
 *               people who like to force 'verbatim' mode.
 *               Thanks to: Diomidis Spinellis <[e-mail]>
 * Marcel Mol: 05sep1993  (Internet: [e-mail])
 *               In dvistuff.c: put some extra chars arround the 'ttfont'
 *               flag in the 'normchar' function.
 * Marcel Mol: 01oct1991  (Internet: [e-mail])
 *               (From Igor Romanenko, [e-mail].)
 *               Fixed unneccesary * in *envp++.
 * Marcel Mol: 24jul1991  (Internet: [e-mail])
 *               Suggestion From: [e-mail]   [uwo]
 *               vertical bars "|" and the tilde "~" print properly in
 *               typewriter font.
 * Marcel Mol: 1990-11-29  (UUCP: [e-mail])
 *               Fixed bug in option handling. Cleanup of code.
 *               Other minor changes/optimizations.
 * Marcel Mol: 1990-11-13  (UUCP: [e-mail])
 *               Included patches of Bernard Gaulle in my version.
 *               Version 4.4
 * Bernard Gaulle: 1990-10-16 ([e-mail])
 *               Add -t option (assuming tt fonts were used)
 *               (Version 3.3)
 * Marcel Mol: 1990-05-03  (UUCP: [e-mail])
 *               Now displays accents etc. on characters again. Added option
 *               (-a) to switch this off. If switched off it now uses proper
 *               spacing.
 *               Cleaned up and show version number in usage();
 *               Version 4.3
 * Marcel Mol: 1990-05-02  (UUCP: [e-mail])
 *               Included patches for VAX/VMS VAXC, done by Robert Schneider.
 *               Version 4.2
 * Marcel Mol: 1990-03-27  (UUCP: [e-mail])
 *               Fixed bug that causes the program to hang when it finds a
 *               fontname with an 's' in it not followed by an 'y'.
 *               Thanks to Paul Orgren ([e-mail]).
 *               Version 4.1
 * Marcel Mol: 1990-02-04  (UUCP: [e-mail])
 *               First attempt to recognize symbol fonts, so bullets (in
 *               itemized lists) are translated to a proper character instead
 *               an awfull ligature.
 *               Version 4.0.
 * Marcel Mol: 1990-02-01  (UUCP: [e-mail])
 *               Included port to VMS (off Joseph Vasallo and Seppo Rantala)
 *               into latest version. Hope things still work, cannot test it ...
 * Joseph Vasallo & Seppo Rantala: 1989-09-05 (Internet: [e-mail])
 *		 Ported to work under VAX/VMS V4.4 & VAXC V2.4 or higher.
 *		 Fixed bugs in using Swedish/Finnish characters.
 * Marcel Mol: 1989-02-14  (UUCP: duteca!marcel)
 *               Fixed check for .dvi extension.
 *               Allowed more ligatures.
 *               Fixed side effect bugs (2 gets as function arguments).
 *               Version 3.2.
 * Marcel Mol: 1989-01-19  (UUCP: duteca!marcel)
 *               Changed in option handling, no change
 *               in user interface (only the undocumented
 *               feature -e).
 *               Version 3.1.
 * Marcel Mol: 1989-01-11  (UUCP: duteca!marcel)
 *               Changed some longs to ints.
 *               It now also runs on MSDOS Microsoft C 5.1
 *               New version: 3.0
 * Marcel Mol: 1989-01-03  (UUCP: duteca!marcel)
 *               Fixed a bugs concerning pager programs
 *               and scanning environment variable DVI2TTY.
 * Marcel Mol: 1988-10-25  (UUCP: duteca!marcel)
 *               Renamed to dvi2tty and converted to C.
 *               Filenames: dvi2tty.c dvi2tty.h dvistuff.c commands.h
 *               Improved spacing between words/characters.
 * bogart:/usr/alla/zap/dvitty/dvitty.p  1986-08-15 20:24:31,
 *               Version to be sent to mod.sources ready.
 * New option since last version:
 *   -Fprog      Pipe output to prog. Can be used to get a different
 *               pager than the default.
 * bogart:/usr/alla/zap/dvitty/dvitty.p  1986-01-13 21:49:31,
 *   Environment variable DVITTY is read and options can be set from it.
 * bogart:/usr/alla/zap/dvitty/dvitty.p  1986-01-10 18:51:03,
 *   Argument parsing, and random access functions (external, in C)
 *   and other OS-dependent stuff (in C). Removed private 'pager' &
 *   tries to pipe through PAGER (environment var) or, if PAGER not
 *   defined, /usr/ucb/more. Some changes for efficency.
 * bogart:/usr/alla/svante/dvitty/dvitty.p  1985-07-15 20:51:00,
 *   The code for processing dvi-files running on UNIX (UCB-Pascal)
 *   but no argument parsing.
 * VERA::SS:<SVANTE-LINDAHL.WORK>DVITTY.PAS.140, 30-Mar-85 05:43:56,
 *   Edit: Svante Lindahl
 * VERA::SS:<SVANTE-LINDAHL.WORK>DVITTY.PAS.136, 15-Jan-85 13:52:59,
 *   Edit: Svante Lindahl, final Twenex version !!!??
 * VERA::SS:<SVANTE-LINDAHL.WORK>DVITTY.PAS.121, 14-Jan-85 03:10:22,
 *   Edit: Svante Lindahl, cleaned up and fixed a lot of little things
 * VERA::SS:<SVANTE-LINDAHL.WORK>DVITTY.PAS.25, 15-Dec-84 05:29:56,
 *   Edit: Svante Lindahl, COMND-interface, including command line scanning
 * VERA::SS:<SVANTE-LINDAHL.WORK>DVITTY.PAS.23, 10-Dec-84 21:24:41,
 *   Edit: Svante Lindahl, added command line scanning with Rscan-JSYS
 * VERA::<SVANTE-LINDAHL.DVITTY>DVITTY.PAS.48,  8-Oct-84 13:26:30,
 *  Edit: Svante Lindahl, fixed switch-parsing, destroyed by earlier patches
 * VERA::<SVANTE-LINDAHL.DVITTY>DVITTY.PAS.45, 29-Sep-84 18:29:53,
 *  Edit: Svante Lindahl
 *
 * dvitty - get an ascii representation of a dvi-file, suitable for ttys
 *
 * This program, and any documentation for it, is copyrighted by Svante
 * Lindahl. It may be copied for non-commercial use only, provided that
 * any and all copyright notices are preserved.
 *
 *
 * UUCP: {seismo,mcvax,cernvax,diku,ukc,unido}!enea!ttds!zap
 * ARPA: enea!ttds![e-mail]
 *  or   [e-mail]
 * EAN:  [e-mail]
 */

/*
 * Include files
 */

#include "dvi2tty.h"
#if defined(VMS)
# include unixio
#endif
#if defined(THINK_C)
# include <unix.h>
# include "macintosh.h"
#endif



/*
 * Constant definitons
 */
    /*-----------------------------------------------------------------------*/
    /* The following constants may be toggled before compilation to          */
    /* customize the default behaviour of the program for your site.         */
    /* Whichever their settings are, the defaults can be overridden at       */
    /* runtime.                                                              */
    /*-----------------------------------------------------------------------*/

#define DEFSCAND  FALSE       /* default is Scandinavian, toggle this if you */
                              /* don't have terminals with Scand. nat. chars */
#define WANTPAGER FALSE       /* default: try to pipe through a pager (like  */
                              /* more) if stdout is tty and no -o switch     */
#define DEFPAGER  "/usr/bin/less"   /* CHANGE TO YOUR LOCAL PAGER            */

    /*------------------ end of customization constants ---------------------*/

#define MAXLEN          100    /* size of char-arrays for strings            */
#if defined(MSDOS) || defined(VMS) || defined(AMIGA)
#define OPTSET      "haJwepPousltvbc" /* legal options                       */
#define OPTWARG     "wepPovb"  /* options with argument                      */
#else
#define OPTSET      "haJwepPousqlfFtvbc"/* legal options                     */
#define OPTWARG     "wepPoFvb"     /* options with argument                  */
#endif

/*
 * USAGE CODES
 */

#define wrnge  1                /* width switch arg out of range     */
#define ign    2                /* ignore cause, print 'Usage:..'    */
#define nan    3                /* not a number where one expected   */
#define gae    4                /* garbage at end                    */
#define bdlst  5                /* bad page-numberlist               */
#define onef   6                /* only one dvifile allowed          */
#define bdopt  7                /* bad option                        */
#define onepp  8                /* only one page list allowed        */
#define noarg  9                /* argument expected                 */
#if defined(THINK_C)
#define nored  10               /* if no input file, redirect stdin  */
#endif



/*
 * Variable definitions
 */

char *dvi2tty = "@(#) dvi2tty.c  5.12 27may1996 M.J.E. Mol (c) 1989-1996";
#define VERSION 5.3


printlisttype * currentpage;    /* current page to print                     */
printlisttype * firstpage;      /* first page selected                       */
printlisttype * lastpage;       /* last page selected                        */

FILE *          DVIfile;
FILE *          output;
bool            outputtofile;   /* tells if output goes to file or stdout    */
#if defined(THINK_C)
bool            inputfromfile;  /* tells if input comes from file or stdin   */
#endif
int             ttywidth;       /* max nr of chars per printed line          */
int             espace;         /* to fake calcs with ttywidth               */

long            foo;            /* utility variable, "register"              */
#if !defined(MSDOS) && !defined(VMS) && !defined(THINK_C) && !defined(AMIGA)
bool            pager;          /* tells if output is piped to a pager       */
char  *         path;           /* name of the pager to run                  */
#endif
char  *         progname;       /* our name                                  */
int             Argc;
char **         Argv;
char            DVIfilename[MAXLEN];
char *          OUTfilename;
char            optch;          /* for option handling                       */



/*
 * Function declarations
 */

#if defined(MSDOS)
void    setoption (char *);
void    getargs   (void);
void    getpages  (int, char *);
void    plcnxt    (int);
void    getfname  (char *);
int     getinteger(int *, int *, char *);
int     getlong   (long *, int *, char *);
void    usage     (int);
#else
void    setoption (char * optarg);
void    getargs   (void);
void    getpages  (int j, char * str);
void    plcnxt    (int pagnr);
void    getfname  (char * str);
int     getinteger(int * dest, int * j, char * str);
int     getlong   (long * dest, int * j, char * str);
void    usage     (int uerr);
#endif


/****************************************************************************/
/*                                                                          */
/*                                 M A I N                                  */
/*                                                                          */
/****************************************************************************/

int main(int argc, char **argv)
{

#if defined(THINK_C)
    argc = process_dvi_command_line(&argv);
#endif

    progname = *argv;
    Argc = argc;
    Argv = argv;

    getargs();                              /* read command line arguments   */
#if defined(THINK_C)
    if (inputfromfile) {
#endif
#if defined(MSDOS)
        if ((DVIfile = fopen(DVIfilename, "rb")) == NULL)
#else
# if defined(VMS)
        if ((DVIfile = fopen(DVIfilename, "r","ctx=rec")) == NULL)
# else
        if ((DVIfile = fopen(DVIfilename, "r")) == NULL)
# endif
#endif
            errorexit(filop);               /* can't open dvifile            */
#if defined(THINK_C)
    }
    else
        DVIfile = stdin;
#endif

    if (outputtofile) {                     /* open the outfile, if needed   */
        if ((output = fopen(OUTfilename, "w")) == NULL)
            errorexit(filcr);
#if defined(THINK_C)
        else
            set_creator((unsigned char*) OUTfilename);
#endif