add man2html program

git-svn-id: svn://kolibrios.org@9278 a494cfbc-eb01-0410-851d-a64ba20cac60

programs/other/man2html/Makefile

@@ -0,0 +1,13 @@
+CC = gcc
+NAME = man2html
+
+CFLAGS = -DNOCGI
+
+SRC = man2html.c cgibase.c abbrev.c strdefs.c
+
+$(NAME): $(SRC)
+	$(CC) $(SRC) -o $(NAME)
+
+clean:
+	rm -f $(NAME)
+

programs/other/man2html/Makefile.kos

@@ -0,0 +1,15 @@
+TCC = /kolibrios/develop/tcc/tcc
+
+NAME = man2html
+
+SRC = man2html.c cgibase.c abbrev.c strdefs.c
+
+CFLAGS = -D_KOLIBRI -DNOCGI
+LIBS = -ltcc -lc.obj
+
+all:
+	$(KTCC) $(CFLAGS) $(SRC) $(LIBS) -o $(NAME)
+
+
+clean:
+	rm $(NAME)

programs/other/man2html/README

@@ -0,0 +1,91 @@
+This directory contains the following.
+
+1. man2html
+
+This is a pure manroff -> html converter.
+No manpath search etc.
+
+Call: man2html [-l | -H host.domain:port] [filename]
+
+The contents of FILENAME (or STDIN, in case FILENAME is "-" or absent)
+are converted from man-style nroff to html, and printed on STDOUT.
+
+With "-l" URLs of the form "lynxcgi:/home/httpd/cgi-bin/..." are generated.
+With "-H host" we make URLs of the form "http://host/cgi-bin/...".
+The default is "http://localhost/cgi-bin/...".
+
+2. A collection of scripts
+
+This part is not installed by "make install" of the global Makefile.
+There are security considerations: it is very unlikely that these
+scripts (still in alpha) are secure, so for the time being they
+should only be used where security is not a major concern.
+
+If you are not afraid, or are not running a httpd, do
+"make install-scripts" in this directory.
+This does three things: install man stuff, install glimpse stuff,
+and install user interface stuff.
+
+2A. man stuff
+
+This first part (that can be done separately with "make install-man-scripts")
+puts various scripts under /home/httpd/cgi-bin and /home/httpd/cgi-aux
+in a subdirectory man.
+It will create a directory /var/man2html to hold the indices.
+(This directory should be writable by the cgi scripts;
+probably that means that the owner should be nobody.
+Choose a group and add all non-httpd users that should be
+able to write this directory to that group.)
+
+Structure of the collection of scripts:
+	man2html is the main script.
+	It uses man.aux when called without arguments.
+	It uses manwhatis when asked for an index of manpages+descriptions.
+	It uses mansec when asked for a compact index of manpages.
+	It uses mansearch when asked for a glimpse search.
+	  In its turn mansearch uses mansearch.aux when called
+	  without arguments. It uses mansearchhelp (which uses
+	  mansearchhelp.aux) when asked for help.
+
+2B. glimpse stuff
+The second part (that can be done separately with
+"make install-glimpse-stuff") installs .glimpse_filters
+in /var/man2html, in order to tell glimpse what decompressors to use.
+
+2C. user interface stuff
+The third part (that can be done separately with "make install-hman")
+installs a user interface to these scripts in /usr/bin/hman.
+Now people can say
+    alias man=/usr/bin/hman
+and have a man that uses a html browser.
+The browser is chosen via environment variables - look at the script.
+
+3. Glimpse.
+
+For the glimpse part, I quote Michael Hamilton:
+----------------------------------------------------------------------
+To use the Glimpse full text searching, you will need to install
+glimpse in /usr/bin.  Redhat rpm users can get glimpse from 
+
+ ftp://ftp.redhat.com/pub/contrib/i386/glimpse-4.0-6.i386.rpm
+
+The glimpse home ftp site is cs.arizona.edu.  N.B. glimpse is not
+freely redistributable for commercial use, I'd be very interested in a
+more liberal alternative.  Having installed glimpse, you will need to
+build a glimpse index in /var/man2html.  This doesn't take too long -
+about 3 minutes on my 486DX2/66 16MB machine.  As root do:
+
+ /usr/bin/glimpseindex -z -H /var/man2html /usr/man/man* /usr/X11R6/man/man* \
+     /usr/local/man/man* /opt/man/man*
+ chmod ugo+r /var/man2html/.glimpse*
+
+The -z option causes glimpse to apply any filters (for decompression etc)
+specified in /var/man2html/.glimpse_filters.
+
+This could be set up as a cron job in /etc/crontab, e.g. (the following
+must be all on one line):
+
+  21 04 * * 1 root /usr/bin/glimpseindex -z -H /var/man2html /usr/man/man* 
+      /usr/X11R6/man/man* /usr/local/man/man* /opt/man/man* ;
+      chmod +r /var/man2html/.glimpse*
+--------------------------------------------------------------------------

programs/other/man2html/abbrev.c

@@ -0,0 +1,62 @@
+#include <string.h>
+#include "defs.h"
+/*
+ * lookup_abbrev() is used for TX macros - is that
+ * something SUN-specific?
+ */
+
+char *abbrev_list[] = {
+    "GSBG", "Getting Started ",
+    "SUBG", "Customizing SunOS",
+    "SHBG", "Basic Troubleshooting",
+    "SVBG", "SunView User's Guide",
+    "MMBG", "Mail and Messages",
+    "DMBG", "Doing More with SunOS",
+    "UNBG", "Using the Network",
+    "GDBG", "Games, Demos &amp; Other Pursuits",
+    "CHANGE", "SunOS 4.1 Release Manual",
+    "INSTALL", "Installing SunOS 4.1",
+    "ADMIN", "System and Network Administration",
+    "SECUR", "Security Features Guide",
+    "PROM", "PROM User's Manual",
+    "DIAG", "Sun System Diagnostics",
+    "SUNDIAG", "Sundiag User's Guide",
+    "MANPAGES", "SunOS Reference Manual",
+    "REFMAN", "SunOS Reference Manual",
+    "SSI", "Sun System Introduction",
+    "SSO", "System Services Overview",
+    "TEXT", "Editing Text Files",
+    "DOCS", "Formatting Documents",
+    "TROFF", "Using <B>nroff</B> and <B>troff</B>",
+    "INDEX", "Global Index",
+    "CPG", "C Programmer's Guide",
+    "CREF", "C Reference Manual",
+    "ASSY", "Assembly Language Reference",
+    "PUL", "Programming Utilities and Libraries",
+    "DEBUG", "Debugging Tools",
+    "NETP", "Network Programming",
+    "DRIVER", "Writing Device Drivers",
+    "STREAMS", "STREAMS Programming",
+    "SBDK", "SBus Developer's Kit",
+    "WDDS", "Writing Device Drivers for the SBus",
+    "FPOINT", "Floating-Point Programmer's Guide",
+    "SVPG", "SunView 1 Programmer's Guide",
+    "SVSPG", "SunView 1 System Programmer's Guide",
+    "PIXRCT", "Pixrect Reference Manual",
+    "CGI", "SunCGI Reference Manual",
+    "CORE", "SunCore Reference Manual",
+    "4ASSY", "Sun-4 Assembly Language Reference",
+    "SARCH", "<FONT SIZE=\"-1\">SPARC</FONT> Architecture Manual",
+    "KR", "The C Programming Language",
+    0, 0 };
+
+char *lookup_abbrev (char *s)
+{
+    int i=0;
+
+    if (!s)
+	 return "";
+    while (abbrev_list[i] && strcmp(s, abbrev_list[i]))
+	 i = i+2;
+    return abbrev_list[i] ? abbrev_list[i+1] : s;
+}

programs/other/man2html/cgibase.c

@@ -0,0 +1,123 @@
+/*
+ * Here are the routines of man2html that output a HREF string.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <time.h>
+#include <ctype.h>		/* tolower() */
+#include <string.h>		/* strlen() */
+#include "defs.h"
+
+/*
+ * The default is to use cgibase. With relative html style
+ * we generate URLs of the form "../manX/page.html".
+ */
+static int relat_html_style = 0;
+
+/*
+ * The default is to use cgibase. With current html style
+ * we generate URLs of the form "./page.html".
+ */
+static int current_html_style = 0;
+
+/*
+ * Either the user is non-local (or local, but using httpd),
+ * in which case we use http:/cgi-bin, or the user is local
+ * and uses lynx, and we use lynxcgi:/home/httpd/cgi-bin.
+ */
+
+static char *man2htmlpath = "/cgi-bin/man/man2html"; 	/* default */
+static char *cgibase_format = "http://%s"; 		/* host.domain:port */
+static char *cgibase_ll_format = "lynxcgi:%s"; 		/* directory */
+static char *cgibase = "http://localhost";		/* default */
+
+/*
+ * Separator between URL and argument string.
+ *
+ * With http:<path to script>/a/b?c+d+e the script is called
+ * with PATH_INFO=/a/b and QUERY_STRING=c+d+e and args $1=c, $2=d, $3=e.
+ * With lynxcgi:<full path to script>?c+d+e no PATH_INFO is possible.
+ */
+static char sep = '?';					/* or '/' */
+
+/* What shall we say in case of relat_html_style? */
+static char *signature = "<HR>\n"
+"This document was created by\n"
+"<A HREF=\"http://github.com/hamano/man2html/\">man2html</A>,\n"
+"using the manual pages.<BR>\n"
+"%s\n";
+
+#define TIMEFORMAT "%T GMT, %B %d, %Y"
+#define TIMEBUFSZ	500
+
+void print_sig()
+{
+    char timebuf[TIMEBUFSZ];
+    struct tm *timetm;
+    time_t clock;
+
+    timebuf[0] = 0;
+#ifdef TIMEFORMAT
+    sprintf(timebuf, "Time: ");
+    clock=time(NULL);
+    timetm=localtime(&clock);
+    snprintf(timebuf, TIMEBUFSZ, "%s%s", timebuf, asctime(timetm));
+    timebuf[TIMEBUFSZ-1] = 0;
+#endif
+    fprintf(out, signature, timebuf);
+}
+
+void
+include_file_html(char *g) {
+     fprintf(out, "<A HREF=\"file:/usr/include/%s\">%s</A>&gt;", g,g);
+}
+
+void
+man_page_html(char *sec, char *h) {
+	if (current_html_style) {
+		if (!h)
+			fprintf(out, "<A HREF=\"./\">"
+			       "Return to Main Contents</A>");
+		else
+			fprintf(out, "<A HREF=\"./%s.html\">%s</A>",
+			       h, h);
+	} else if (relat_html_style) {
+		if (!h)
+			fprintf(out, "<A HREF=\"../index.html\">"
+			       "Return to Main Contents</A>");
+		else
+			fprintf(out, "<A HREF=\"../man%s/%s.%s.html\">%s</A>",
+			       sec, h, sec, h);
+	} else {
+		if (!h)
+			fprintf(out, "<A HREF=\"%s%s\">Return to Main Contents</A>",
+			       cgibase, man2htmlpath);
+		else if (!sec)
+			fprintf(out, "<A HREF=\"%s%s%c%s\">%s</A>",
+			       cgibase, man2htmlpath, sep, h, h);
+		else
+			fprintf(out, "<A HREF=\"%s%s%c%s+%s\">%s</A>",
+			       cgibase, man2htmlpath, sep, sec, h, h);
+	}
+}
+
+void
+ftp_html(char *f) {
+     fprintf(out, "<A HREF=\"ftp://%s\">%s</A>", f, f);
+}
+
+void
+www_html(char *f) {
+     fprintf(out, "<A HREF=\"http://%s\">%s</A>", f, f);
+}
+
+void
+mailto_html(char *g) {
+     fprintf(out, "<A HREF=\"mailto:%s\">%s</A>", g, g);
+}
+
+void
+url_html(char *g) {
+     fprintf(out, "<A HREF=\"%s\">%s</A>", g, g);
+}

programs/other/man2html/defs.h

@@ -0,0 +1,43 @@
+#ifndef DEFS_H
+#define DEFS_H
+#include <stddef.h>
+#include <stdio.h>
+
+extern FILE *out;
+
+extern int nroff;
+extern int local_lynx;
+
+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;
+};
+
+extern STRDEF *chardef, *strdef, *defdef;
+extern INTDEF *intdef;
+
+#define V(A,B) ((A)*256+(B))
+
+extern void stdinit(void);
+extern void print_sig(void);
+extern char *lookup_abbrev(char *);
+extern void include_file_html(char *);
+extern void man_page_html(char*, char *);
+extern void ftp_html(char *);
+extern void www_html(char *);
+extern void mailto_html(char *);
+extern void url_html(char *);
+extern void *xmalloc(size_t size);
+extern void *xrealloc(void *ptr, size_t size);
+extern char *xstrdup(const char *s);
+#endif

programs/other/man2html/strdefs.c

@@ -0,0 +1,176 @@
+#include "defs.h"
+
+#ifndef NULL
+#define NULL	((void *) 0)
+#endif
+
+int nroff = 1;
+
+#define NROFF (-666)
+#define TROFF (-667)
+
+STRDEF *chardef, *strdef, *defdef;
+INTDEF *intdef;
+
+static INTDEF standardint[] = {
+    { V('n',' '), NROFF, 0, NULL },
+    { V('t',' '), TROFF, 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'), TROFF, 0, NULL },
+    { V('.','V'), 1,     0, NULL }, /* the me package tests for this */
+    { 0, 0, 0, NULL } };
+
+static STRDEF standardstring[] = {
+    { V('R',' '), 1, "®", NULL },
+    { V('l','q'), 2, "``", NULL },
+    { V('r','q'), 2, "''", NULL },
+    { 0, 0, NULL, NULL}
+};
+
+
+static STRDEF standardchar[] = {
+    { V('*','*'), 1, "*", NULL  },	/* math star */
+    { 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, "µ", 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('+','-'), 1, "±", NULL  },
+    { V('1','2'), 1, "½", NULL  },
+    { V('1','4'), 1, "¼", NULL  },
+    { V('3','4'), 1, "¾", NULL  },
+    { V('F','i'), 3, "ffi", NULL  },
+    { V('F','l'), 3, "ffl", NULL  },
+    { V('a','a'), 1, "´", NULL  },
+    { V('a','p'), 1, "~", NULL  },
+    { V('b','r'), 1, "|", NULL  },
+    { V('b','u'), 1, "*", NULL  }, 	/* bullet */
+    { V('b','v'), 1, "|", NULL  },
+    { V('c','i'), 1, "o", NULL  }, 	/* circle */
+    { V('c','o'), 1, "©", NULL  },
+    { V('c','t'), 1, "¢", NULL  },
+    { V('d','e'), 1, "°", NULL  },
+    { V('d','g'), 1, "+", NULL  }, 	/* dagger */
+    { V('d','i'), 1, "÷", NULL  },
+    { V('e','m'), 3, "---", NULL  }, 	/* em dash */
+    { V('e','n'), 1, "-", NULL }, 	/* en dash */
+    { V('e','q'), 1, "=", NULL  },
+    { V('e','s'), 1, "Ø", NULL  },
+    { V('f','f'), 2, "ff", NULL  },
+    { V('f','i'), 2, "fi", NULL  },
+    { V('f','l'), 2, "fl", NULL  },
+    { V('f','m'), 1, "´", NULL  },
+    { V('g','a'), 1, "`", NULL  },
+    { V('h','y'), 1, "-", NULL  },
+    { V('l','c'), 2, "|¯", NULL  },
+    { V('i','f'), 8, "Infinity", NULL }, /* infinity sign */
+    { V('i','s'), 8, "Integral", NULL }, /* integral sign */
+    { 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','d'), 1, "d", NULL }, 	/* partial derivative */
+    { 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('t','s'), 1, "s", NULL }, 	/* should be terminal sigma */
+    { V('u','l'), 1, "_", NULL  },
+    { V('>','='), 1, "&gt;", NULL },
+    { V('<','='), 1, "&lt;", NULL },
+    { 0, 0, NULL, NULL  }
+};
+
+void stdinit(void) {
+    STRDEF *stdf;
+    int i;
+
+    stdf = &standardchar[0];
+    i = 0;
+    while (stdf->nr) {
+	if (stdf->st) stdf->st = xstrdup(stdf->st);
+	stdf->next = &standardchar[i];
+	stdf = stdf->next;
+	i++;
+    }
+    chardef=&standardchar[0];
+
+    stdf=&standardstring[0];
+    i=0;
+    while (stdf->nr) {
+	 /* waste a little memory, and make a copy, to avoid
+	    the segfault when we free non-malloced memory */
+	if (stdf->st) stdf->st = xstrdup(stdf->st);
+	stdf->next = &standardstring[i];
+	stdf = stdf->next;
+	i++;
+    }
+    strdef=&standardstring[0];
+
+    intdef=&standardint[0];
+    i=0;
+    while (intdef->nr) {
+	if (intdef->nr == NROFF) intdef->nr = nroff; else
+	if (intdef->nr == TROFF) intdef->nr = !nroff;
+	intdef->next = &standardint[i];
+	intdef = intdef->next;
+	i++;
+    }
+    intdef = &standardint[0];
+    defdef = NULL;
+}