man2html.c

android-w.song.android.widget

/*
 * 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.
 *
 * THIS IS A MODIFIED VERSION.  IT WAS MODIFIED BY [e-mail] FOR
 * USE BY BASH.
 */

/*
 * 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 Also out of date pages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.)
 */
#ifdef HAVE_CONFIG_H
#include <config.h>
#endif

#define NROFF 0

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

#define NULL_TERMINATED(n) ((n) + 1)

#define HUGE_STR_MAX  10000
#define LARGE_STR_MAX 2000
#define MED_STR_MAX   500
#define SMALL_STR_MAX 100
#define TINY_STR_MAX  10

#define MAX_MAN_PATHS 100	/* Max number of directories */
#define MAX_ZCATS     10	/* Max number of zcat style programs */
#define MAX_WORDLIST  100

#ifndef EXIT_SUCCESS
#define EXIT_SUCCESS 0
#endif
#ifndef EXIT_FAILURE
#define EXIT_FAILURE 1
#endif
#ifndef EXIT_USAGE
#define EXIT_USAGE 2
#endif

static char location_base[NULL_TERMINATED(MED_STR_MAX)] = "";

static char th_page_and_sec[128] = { '\0' };
static char th_datestr[128] = { '\0' };
static char th_version[128] = { '\0' };

char   *signature = "<HR>\nThis document was created by man2html from %s.<BR>\nTime: %s\n";

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

char *manpage;

/* BSD mandoc Bl/El lists to HTML list types */
#define BL_DESC_LIST   1
#define BL_BULLET_LIST 2
#define BL_ENUM_LIST   4

/* BSD mandoc Bd/Ed example(?) blocks */
#define BD_LITERAL  1
#define BD_INDENT   2

#ifndef HAVE_STRERROR
static char *
strerror(int e)
{
	static char emsg[40];

#if defined (HAVE_SYS_ERRLIST)
	extern int sys_nerr;
	extern char *sys_errlist[];

	if (e > 0 && e < sys_nerr)
		return (sys_errlist[e]);
	else
#endif				/* HAVE_SYS_ERRLIST */
	{
		sprintf(emsg, "Unknown system error %d", e);
		return (&emsg[0]);
	}
}
#endif				/* !HAVE_STRERROR */

static char *
strgrow(char *old, int len)
{
	char   *new = realloc(old, (strlen(old) + len + 1) * sizeof(char));

	if (!new) {
		fprintf(stderr, "man2html: out of memory");
		exit(EXIT_FAILURE);
	}
	return new;
}

static char *
stralloc(int len)
{
	/* allocate enough for len + NULL */
	char   *new = malloc((len + 1) * sizeof(char));

	if (!new) {
		fprintf(stderr, "man2html: out of memory");
		exit(EXIT_FAILURE);
	}
	return new;
}

/*
 * Some systems don't have strdup so lets use our own - which can also
 * check for out of memory.
 */
static char *
strduplicate(char *from)
{
	char   *new = stralloc(strlen(from));

	strcpy(new, from);
	return new;
}

/* Assumes space for n plus a null */
static char *
strmaxcpy(char *to, char *from, int n)
{
	int     len = strlen(from);

	strncpy(to, from, n);
	to[(len <= n) ? len : n] = '\0';
	return to;
}

static char *
strmaxcat(char *to, char *from, int n)
{
	int     to_len = strlen(to);

	if (to_len < n) {
		int     from_len = strlen(from);
		int     cp = (to_len + from_len <= n) ? from_len : n - to_len;

		strncpy(to + to_len, from, cp);
		to[to_len + cp] = '\0';
	}
	return to;
}

/* Assumes space for limit plus a null */
static char *
strlimitcpy(char *to, char *from, int n, int limit)
{
	int     len = n > limit ? limit : n;

	strmaxcpy(to, from, len);
	to[len] = '\0';
	return to;
}

/*
 * takes string and escapes all metacharacters.  should be used before
 * including string in system() or similar call.
 */
static char *
escape_input(char *str)
{
	int     i, j = 0;
	static char new[NULL_TERMINATED(MED_STR_MAX)];

	if (strlen(str) * 2 + 1 > MED_STR_MAX) {
		fprintf(stderr,
			"man2html: escape_input - str too long:\n%-80s...\n",
			str);
		exit(EXIT_FAILURE);
	}
	for (i = 0; i < strlen(str); i++) {
		if (!(((str[i] >= 'A') && (str[i] <= 'Z')) ||
		      ((str[i] >= 'a') && (str[i] <= 'z')) ||
		      ((str[i] >= '0') && (str[i] <= '9')))) {
			new[j] = '\\';
			j++;
		}
		new[j] = str[i];
		j++;
	}
	new[j] = '\0';
	return new;
}

static void
usage(void)
{
	fprintf(stderr, "man2html: usage: man2html filename\n");
}



/*
 * 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"

static char *fname;
static FILE *idxfile;

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

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

static 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}};

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


static 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 */


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

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

static char *scan_troff(char *c, int san, char **result);
static char *scan_troff_mandoc(char *c, int san, char **result);

static char **argument = NULL;

static char charb[TINY_STR_MAX];

static void
print_sig(void)
{
	char    datbuf[NULL_TERMINATED(MED_STR_MAX)];
	struct tm *timetm;
	time_t  clock;

	datbuf[0] = '\0';
	clock = time(NULL);
	timetm = localtime(&clock);
	strftime(datbuf, MED_STR_MAX, TIMEFORMAT, timetm);
	printf(signature, manpage, datbuf);
}

static 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';
	if (charb[0] == '<') {	/* Fix up <= */
		charb[4] = charb[1];
		strncpy(charb, "&lt;", 4);
		charb[5] = '\0';
	}
	curpos += 2;
	return charb;
}

static 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;
}

static char *
read_man_page(char *filename)
{
	char   *man_buf = NULL;
	int     i;
	FILE   *man_stream = NULL;
	struct stat stbuf;
	int     buf_size;

	if (stat(filename, &stbuf) == -1)
		return NULL;

	buf_size = stbuf.st_size;
	man_buf = stralloc(buf_size + 5);
	man_stream = fopen(filename, "r");
	if (man_stream) {
		man_buf[0] = '\n';
		if (fread(man_buf + 1, 1, buf_size, man_stream) == buf_size) {
			man_buf[buf_size] = '\n';
			man_buf[buf_size + 1] = man_buf[buf_size + 2] = '\0';
		} else {
			man_buf = NULL;
		}
		fclose(man_stream);
	}
	return man_buf;
}


static char outbuffer[NULL_TERMINATED(HUGE_STR_MAX)];
static int obp = 0;
static int no_newline_output = 0;
static int newline_for_fun = 0;
static int output_possible = 0;
static int out_length = 0;

/*
 * Add the links to the output. At the moment the following are
 * recognized:
 * 
#if 0
 *	name(*)                 -> ../man?/name.*
#endif
 *	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
 */
static void
add_links(char *c)
{
	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.");
#if 0
	idtest[4] = strchr(c + 1, '(');
#else
	idtest[4] = 0;
#endif
	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';
				fputs(c, stdout);
				*g = t;
				*h = '\0';
				printf("<A HREF=\"file:/usr/include/%s\">%s</A>&gt;", g, g);
				c = f + 6;
			} else {
				f[5] = '\0';
				fputs(c, stdout);
				f[5] = ';';
				c = f + 5;
			}
			break;
		case 4:	/* manpage */
#if 0
			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';
					fputs(c, stdout);
					*h = t;
					t = *e;
					*e = '\0';
					if (subsec)
						printf("<A HREF=\""
						       CGIBASE
						  "?man%c/%s.%c%c\">%s</A>",
						       sec, h, sec, tolower(subsec), h);
					else
						printf("<A HREF=\""
						       CGIBASE
						    "?man%c/%s.%c\">%s</A>",
						       sec, h, sec, h);
					*e = t;
					c = e;