man2html.c

很少见的linux下的红外口的工具

/*
** This program was written by Richard Verhoeven (NL:5482ZX35)
** at the Eindhoven University of Technology. Email: [e-mail]
**
** Permission is granted to distribute, modify and use this program as long
** as this comment is not removed or changed.
*/
/*
** If you want to use this program for your WWW server, adjust the line
** which defines the CGIBASE or compile it with the -DCGIBASE='"..."' option.
**
** You have to adjust the built-in manpath to your local system. Note that
** every directory should start and end with the '/' and that the first
** directory should be "/" to allow a full path as an argument.
**
** The program first check if PATH_INFO contains some information.
** If it does (t.i. man2html/some/thing is used), the program will look
** for a manpage called PATH_INFO in the manpath.
**
** Otherwise the manpath is searched for the specified command line argument,
** where the following options can be used:
**
** name      name of manpage (csh, printf, xv, troff)
** section   the section (1 2 3 4 5 6 7 8 9 n l 1v ...)
** -M path   an extra directory to look for manpages (replaces "/")
**
** If man2html finds multiple manpages that satisfy the options, an index
** is displayed and the user can make a choice. If only one page is
** found, that page will be displayed.
**
** man2html will add links to the converted manpages. The function add_links
** is used for that. At the moment it will add links as follows, where   
**     indicates what should match to start with:
** ^^^
** Recognition           Item            Link
** ----------------------------------------------------------
** name(*)               Manpage         ../man?/name.*
**     ^
** name@hostname         Email address   mailto:name@hostname
**     ^
** method://string       URL             method://string
**       ^^^
** www.host.name         WWW server      http://www.host.name
** ^^^^
** ftp.host.name         FTP server      ftp://ftp.host.name
** ^^^^
** <file.h>              Include file    file:/usr/include/file.h
**      ^^^
**
** Since man2html does not check if manpages, hosts or email addresses exist,
** some links might not work. For manpages, some extra checks are performed
** to make sure not every () pair creates a link. Also out of date pages
** might point to incorrect places.
**
** The program will not allow users to get system specific files, such as
** /etc/passwd. It will check that "man" is part of the specified file and
** that  "/../" isn't. Even if someone manages to get such file, man2html will
** handle it like a manpage and will usually not produce any output (or crash).
**
** If you find any bugs when normal manpages are converted, please report
** them to me ([e-mail]) after you have checked that man(1) can handle
** the manpage correct.
**
** Known bugs and missing features:
**
**  * Equations are not converted at all.
**  * Tables are converted but some features are not possible in html.
**  * The tabbing environment is converted by counting characters and adding
**    spaces. This might go wrong (outside <PRE>)
**  * Some pages look beter if man2html works in troff mode, especially pages
**    with tables. You can deside at compile time which made you want to use.
**
**    -DNROFF=0     troff mode
**    -DNROFF=1     nroff mode   (default)
**
**    if you install both modes, you should compile with the correct CGIBASE.
**  * Some manpages rely on the fact that troff/nroff is used to convert
**    them and use features which are not descripted in the man manpages.
**    (definitions, calculations, conditionals, requests). I can't guarantee
**    that all these features work on all manpages. (I didn't have the
**    time to look through all the available manpages.)
*/


#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <sys/stat.h>
#include <dirent.h>
#include <ctype.h>
#include <sys/types.h>
#include <time.h>
#include <sys/time.h>

#ifndef CGIBASE
#define CGIBASE "http://wsinwp01.win.tue.nl:1234/cgi-bin/man2html"
#endif

#ifndef NROFF
#define NROFF 1
#endif

char *signature = "<HR>\n"
"This document was created by\n"
"<A HREF=\""CGIBASE"\">man2html</A>,\n"
"using the manual pages.<BR>\n"
"Time: %s\n";

/* timeformat for signature */
#define TIMEFORMAT "%T GMT, %B %d, %Y"

char *manpath[] = { "/",
		    "/usr/X11/man/",
		    "/usr/man/",
		    "/usr/local/man/",
		    "/usr/exp/man/",
		    "/usr/openwin/man/",
		    "/usr/tex/man/",
		    "/usr/www/man/",
		    "/usr/lang/man/",
		    "/usr/gnu/man/",
		    "/usr/motif/man/",
		    "/usr/titools/man/",
		    "/usr/sunpc/man/",
		    "/usr/ncd/man/",
		    "/usr/newsprint/man/",
		    NULL };

char *sections = "123456789nl";

void usage(void)
{
    printf("Content-type: text/html\n\n"
	   "<HTML><HEAD>\n"
	   "<TITLE>Manual Pages</TITLE>\n"
	   "</HEAD><BODY>\n"
	   "<H1>Manual Pages</H1>\n"
	   "This is a HyperText interface to the UNIX manpages.\n"
	   "You can enter a program name, the section, an extra\n"
	   "directory (using -M) or a full name. For example\n"
	   "<UL><LI><TT>elm</TT>\n"
	   "<LI><TT>elm 1</TT>\n"
	   "<LI><TT>-M /usr/local/man elm</TT>\n"
	   "<LI><TT>/local/gcc/man/man1/gperf.1</TT>\n"
	   "</UL>\n"
	   "<ISINDEX>\n"
	   "<P>\n"
	   "This man2html converter was written by \n"
	   "<A HREF=\"http://wsinwp01.win.tue.nl:1234/index.html\">"
	   "Richard Verhoeven</A>\n"
	   "</BODY></HTML>\n");
    exit(0);
}

/* below this you should not change anything unless you know a lot
** about this program or about troff.
*/


typedef struct STRDEF STRDEF;
struct STRDEF {
    int nr,slen;
    char *st;
    STRDEF *next;
};

typedef struct INTDEF INTDEF;
struct INTDEF {
    int nr;
    int val;
    int incr;
    INTDEF *next;
};

static char NEWLINE[2]="\n";
static char idxlabel[6] = "ixAAA";

#define INDEXFILE "/tmp/manindex.list"

char *fname;
FILE *idxfile;

STRDEF *chardef, *strdef, *defdef;
INTDEF *intdef;

#define V(A,B) ((A)*256+(B))

INTDEF standardint[] = {
    { V('n',' '), NROFF,0, NULL },
    { V('t',' '), 1-NROFF,0, NULL },
    { V('o',' '), 1,0, NULL },
    { V('e',' '), 0,0, NULL },
    { V('.','l'), 70,0,NULL },
    { V('.','$'), 0,0, NULL },
    { V('.','A'), NROFF,0, NULL },
    { V('.','T'), 1-NROFF,0, NULL },
    { V('.','V'), 1,0, NULL }, /* the me package tests for this */
    { 0, 0, 0, NULL } };

STRDEF standardstring[] = {
    { V('R',' '), 1, "&#174;", NULL },
    { V('l','q'), 2, "``", NULL },
    { V('r','q'), 2, "''", NULL },
    { 0, 0, NULL, NULL}
};


STRDEF standardchar[] = {
    { V('*','*'), 1, "*", NULL  },
    { V('*','A'), 1, "A", NULL  },
    { V('*','B'), 1, "B", NULL  },
    { V('*','C'), 2, "Xi", NULL  },
    { V('*','D'), 5, "Delta", NULL  },
    { V('*','E'), 1, "E", NULL  },
    { V('*','F'), 3, "Phi", NULL  },
    { V('*','G'), 5, "Gamma", NULL  },
    { V('*','H'), 5, "Theta", NULL  },
    { V('*','I'), 1, "I", NULL  },
    { V('*','K'), 1, "K", NULL  },
    { V('*','L'), 6, "Lambda", NULL  },
    { V('*','M'), 1, "M", NULL  },
    { V('*','N'), 1, "N", NULL  },
    { V('*','O'), 1, "O", NULL  },
    { V('*','P'), 2, "Pi", NULL  },
    { V('*','Q'), 3, "Psi", NULL  },
    { V('*','R'), 1, "P", NULL  },
    { V('*','S'), 5, "Sigma", NULL  },
    { V('*','T'), 1, "T", NULL  },
    { V('*','U'), 1, "Y", NULL  },
    { V('*','W'), 5, "Omega", NULL  },
    { V('*','X'), 1, "X", NULL  },
    { V('*','Y'), 1, "H", NULL  },
    { V('*','Z'), 1, "Z", NULL  },
    { V('*','a'), 5, "alpha", NULL },
    { V('*','b'), 4, "beta", NULL },
    { V('*','c'), 2, "xi", NULL },
    { V('*','d'), 5, "delta", NULL },
    { V('*','e'), 7, "epsilon", NULL },
    { V('*','f'), 3, "phi", NULL },
    { V('*','g'), 5, "gamma", NULL },
    { V('*','h'), 5, "theta", NULL },
    { V('*','i'), 4, "iota", NULL },
    { V('*','k'), 5, "kappa", NULL },
    { V('*','l'), 6, "lambda", NULL },
    { V('*','m'), 1, "&#181;", NULL  },
    { V('*','n'), 2, "nu", NULL },
    { V('*','o'), 1, "o", NULL },
    { V('*','p'), 2, "pi", NULL },
    { V('*','q'), 3, "psi", NULL },
    { V('*','r'), 3, "rho", NULL },
    { V('*','s'), 5, "sigma", NULL },
    { V('*','t'), 3, "tau", NULL },
    { V('*','u'), 7, "upsilon", NULL },
    { V('*','w'), 5, "omega", NULL },
    { V('*','x'), 3, "chi", NULL },
    { V('*','y'), 3, "eta", NULL },
    { V('*','z'), 4, "zeta", NULL },
    { V('t','s'), 5, "sigma", NULL },
    { V('+','-'), 1, "&#177;", NULL  },
    { V('1','2'), 1, "&#189;", NULL  },
    { V('1','4'), 1, "&#188;", NULL  },
    { V('3','4'), 1, "&#190;", NULL  },
    { V('F','i'), 3, "ffi", NULL  },
    { V('F','l'), 3, "ffl", NULL  },
    { V('a','a'), 1, "&#180;", NULL  },
    { V('a','p'), 1, "~", NULL  },
    { V('b','r'), 1, "|", NULL  },
    { V('b','u'), 1, "*", NULL  },
    { V('b','v'), 1, "|", NULL  },
    { V('c','i'), 1, "o", NULL  },
    { V('c','o'), 1, "&#169;", NULL  },
    { V('c','t'), 1, "&#162;", NULL  },
    { V('d','e'), 1, "&#176;", NULL  },
    { V('d','g'), 1, "+", NULL  },
    { V('d','i'), 1, "&#247;", NULL  },
    { V('e','m'), 1, "-", NULL  },
    { V('e','m'), 3, "---", NULL },
    { V('e','q'), 1, "=", NULL  },
    { V('e','s'), 1, "&#216;", NULL  },
    { V('f','f'), 2, "ff", NULL  },
    { V('f','i'), 2, "fi", NULL  },
    { V('f','l'), 2, "fl", NULL  },
    { V('f','m'), 1, "&#180;", NULL  },
    { V('g','a'), 1, "`", NULL  },
    { V('h','y'), 1, "-", NULL  },
    { V('l','c'), 2, "|&#175;", NULL  },
    { V('l','f'), 2, "|_", NULL  },
    { V('l','k'), 1, "<FONT SIZE=+2>{</FONT>", NULL  },
    { V('m','i'), 1, "-", NULL  },
    { V('m','u'), 1, "&#215;", NULL  },
    { V('n','o'), 1, "&#172;", NULL  },
    { V('o','r'), 1, "|", NULL  },
    { V('p','l'), 1, "+", NULL  },
    { V('r','c'), 2, "&#175;|", NULL  },
    { V('r','f'), 2, "_|", NULL  },
    { V('r','g'), 1, "&#174;", NULL  },
    { V('r','k'), 1, "<FONT SIZE=+2>}</FONT>", NULL  },
    { V('r','n'), 1, "&#175;", NULL  },
    { V('r','u'), 1, "_", NULL  },
    { V('s','c'), 1, "&#167;", NULL  },
    { V('s','l'), 1, "/", NULL  },
    { V('s','q'), 2, "[]", NULL  },
    { V('u','l'), 1, "_", NULL  },
    { 0, 0, NULL, NULL  }
};

/* default: print code */


char eqndelimopen=0, eqndelimclose=0;
char escapesym='\\', nobreaksym='\'', controlsym='.', fieldsym=0, padsym=0;

char *buffer=NULL;
int buffpos=0, buffmax=0;
int scaninbuff=0;
int itemdepth=0;
int dl_set[20]= { 0 };
int still_dd=0;
int tabstops[20] = { 8, 16, 24, 32, 40, 48, 56, 64, 72, 80, 88, 96 };
int maxtstop=12;
int curpos=0;

extern char *scan_troff(char *c, int san, char **result);

static char **argument=NULL;

static char charb[3];

void print_sig(void)
{
    char datbuf[500];
    struct tm *timetm;
    time_t clock;
    datbuf[0]='\0';
#ifdef TIMEFORMAT
    clock=time(NULL);
    timetm=gmtime(&clock);
    strftime(datbuf,500,TIMEFORMAT, timetm);
#endif
    printf(signature, datbuf);
}

char *expand_char(int nr)
{
  STRDEF *h;
  h=chardef;
  if (!nr) return NULL;
  while (h)
      if (h->nr==nr) {
	  curpos+=h->slen;
	  return h->st;
      } else
	  h=h->next;
  charb[0]=nr/256;
  charb[1]=nr%256;
  charb[2]='\0';
  curpos+=2;
  return charb;
}

char *expand_string(int nr)
{
  STRDEF *h=strdef;
  if (!nr) return NULL;
  while (h)
      if (h->nr==nr) {
	  curpos+=h->slen;
	  return h->st;
    } else
	h=h->next;
  return NULL;
}


char outbuffer[1024];
int obp=0;
int no_newline_output=0;
int newline_for_fun=0;
int output_possible=0;
int out_length=0;

void add_links(char *c)
{
    /*
    ** Add the links to the output.
    ** At the moment the following are recognized:
    **
    ** name(*)                 -> ../man?/name.*
    ** method://string         -> method://string
    ** www.host.name           -> http://www.host.name
    ** ftp.host.name           -> ftp://ftp.host.name
    ** name@host               -> mailto:name@host
    ** <name.h>                -> file:/usr/include/name.h   (guess)
    **
    ** Other possible links to add in the future:
    **
    ** /dir/dir/file  -> file:/dir/dir/file
    */
    int i,j,nr;
    char *f, *g,*h;
    char *idtest[6]; /* url, mailto, www, ftp, manpage */
    out_length+=strlen(c);
    /* search for (section) */
    nr=0;
    idtest[0]=strstr(c+1,"://");
    idtest[1]=strchr(c+1,'@');
    idtest[2]=strstr(c,"www.");
    idtest[3]=strstr(c,"ftp.");
    idtest[4]=strchr(c+1,'(');
    idtest[5]=strstr(c+1,".h&gt;");
    for (i=0; i<6; i++) nr += (idtest[i]!=NULL);
    while (nr) {
	j=-1;
	for (i=0; i<6; i++)
	    if (idtest[i] && (j<0 || idtest[i]<idtest[j])) j=i;
	switch (j) {
	case 5: /* <name.h> */
	    f=idtest[5];
	    h=f+2;
	    g=f;
	    while (g>c && g[-1]!=';') g--;
	    if (g!=c) {
		char t;
		t=*g;
		*g='\0';
		printf("%s",c);
		*g=t;*h='\0';
		printf("<A HREF=\"file:/usr/include/%s\">%s</A>&gt;", g,g);
		c=f+6;
	    } else {
		f[5]='\0';
		printf("%s",c);
		f[5]=';';
		c=f+5;
	    }
	    break;
	case 4: /* manpage */
	    f=idtest[j];
	    /* check section */
	    g=strchr(f,')');
	    if (g && f-g<6 && (isalnum(f[-1]) || f[-1]=='>') &&
		((isdigit(f[1]) && f[1]!='0' &&
		  (f[2]==')' || (isalpha(f[2]) && f[3]==')') || f[2]=='X')) ||
		 (f[2]==')' && (f[1]=='n' || f[1]=='l')))) {
		/* this might be a link */
		h=f-1;
		/* skip html makeup */
		while (h>c && *h=='>') {
		    while (h!=c && *h!='<') h--;
		    if (h!=c) h--;
		}
		if (isalnum(*h)) {
		    char t,sec,subsec, *e;
		    e=h+1;
		    sec=f[1];
		    subsec=f[2];
		    if ((subsec=='X' && f[3]!=')')|| subsec==')') subsec='\0';
		    while (h>c && (isalnum(h[-1]) || h[-1]=='_' ||
				    h[-1]=='-' || h[-1]=='.'))
			h--;
		    t=*h;
		    *h='\0';
		    printf("%s", c);
		    *h=t;
		    t=*e;
		    *e='\0';
#if LIRC_RELATIVE_PATH
		    if (subsec)
			printf("<A HREF=\"../man%c/%s.%c%c\">%s</A>",
			       sec, h, sec, tolower(subsec), h);
		    else
			printf("<A HREF=\"../man%c/%s.%c\">%s</A>",
			       sec, h, sec, h);
#elif LIRC_ABSOLUTE_PATH
		    if (subsec)
			printf("<A HREF=\"file:///usr/man/man%c/%s.%c%c\">%s</A>",
			       sec, h, sec, tolower(subsec), h);
		    else
			printf("<A HREF=\"file:///usr/man/man%c/%s.%c\">%s</A>",
			       sec, h, sec, h);
#else
		    /* links to man pages make no sense for us */
		    printf("%s",h);
#endif
		    *e=t;
		    c=e;
		}
	    }
	    *f='\0';
	    printf("%s", c);
	    *f='(';
	    idtest[4]=f-1;
	    c=f;
	    break; /* manpage */
	case 3: /* ftp */
	case 2: /* www */
	    g=f=idtest[j];
	    while (*g && (isalnum(*g) || *g=='_' || *g=='-' || *g=='+' ||
			  *g=='.')) g++;
	    if (g[-1]=='.') g--;
	    if (g-f>4) {
		char t;
		t=*f; *f='\0';
		printf("%s",c);
		*f=t; t=*g;*g='\0';
		printf("<A HREF=\"%s://%s\">%s</A>", (j==3?"ftp":"http"),
		       f,f);
		*g=t;
		c=g;
	    } else {
		f[3]='\0';
		printf("%s",c);
		c=f+3;
		f[3]='.';
	    }
	    break;
	case 1: /* mailto */
	    g=f=idtest[1];
	    while (g>c && (isalnum(g[-1]) || g[-1]=='_' || g[-1]=='-' ||
			   g[-1]=='+' || g[-1]=='.' || g[-1]=='%')) g--;
	    h=f+1;
	    while (*h && (isalnum(*h) || *h=='_' || *h=='-' || *h=='+' ||
			  *h=='.')) h++;
	    if (*h=='.') h--;
	    if (h-f>4 && f-g>1) {
		char t;
		t=*g;
		*g='\0';
		printf("%s",c);
		*g=t;t=*h;*h='\0';
		printf("<A HREF=\"mailto:%s\">%s</A>",g,g);
		*h=t;
		c=h;
	    } else {
		*f='\0';
		printf("%s",c);
		*f='@';
		idtest[1]=c;
		c=f;
	    }
	    break;
	case 0: /* url */