path: root/www/publicfile/files/publicfile-0.52-commonlog-2.patch

diff -rNC3 publicfile-0.52.orig/Makefile publicfile-0.52/Makefile
*** publicfile-0.52.orig/Makefile	Tue Nov  9 08:23:46 1999
--- publicfile-0.52/Makefile	Fri Nov  9 10:09:45 2001
***************
*** 450,460 ****
  stralloc.a: \
  makelib stralloc_cat.o stralloc_catb.o stralloc_cats.o \
  stralloc_copy.o stralloc_eady.o stralloc_opyb.o stralloc_opys.o \
! stralloc_pend.o stralloc_num.o
  	./makelib stralloc.a stralloc_cat.o stralloc_catb.o \
  	stralloc_cats.o stralloc_copy.o stralloc_eady.o \
  	stralloc_opyb.o stralloc_opys.o stralloc_pend.o \
! 	stralloc_num.o
  
  stralloc_cat.o: \
  compile stralloc_cat.c byte.h stralloc.h gen_alloc.h
--- 450,460 ----
  stralloc.a: \
  makelib stralloc_cat.o stralloc_catb.o stralloc_cats.o \
  stralloc_copy.o stralloc_eady.o stralloc_opyb.o stralloc_opys.o \
! stralloc_pend.o stralloc_num.o stralloc_opytrim.o
  	./makelib stralloc.a stralloc_cat.o stralloc_catb.o \
  	stralloc_cats.o stralloc_copy.o stralloc_eady.o \
  	stralloc_opyb.o stralloc_opys.o stralloc_pend.o \
! 	stralloc_num.o stralloc_opytrim.o
  
  stralloc_cat.o: \
  compile stralloc_cat.c byte.h stralloc.h gen_alloc.h
***************
*** 484,489 ****
--- 484,493 ----
  stralloc_opyb.o: \
  compile stralloc_opyb.c stralloc.h gen_alloc.h byte.h
  	./compile stralloc_opyb.c
+ 
+ stralloc_opytrim.o: \
+ compile stralloc_opytrim.c stralloc.h
+ 	./compile stralloc_opytrim.c
  
  stralloc_opys.o: \
  compile stralloc_opys.c byte.h str.h stralloc.h gen_alloc.h
diff -rNC3 publicfile-0.52.orig/README.log_combined publicfile-0.52/README.log_combined
*** publicfile-0.52.orig/README.log_combined	Thu Jan  1 01:00:00 1970
--- publicfile-0.52/README.log_combined	Fri Nov  9 10:09:45 2001
***************
*** 0 ****
--- 1,36 ----
+ This version of publicfile has been patched to write *additional* log 
+ information to STDERR. This additional information conforms to Apache's
+ and Netscape's "common" or "combined" log formats, i. e. a line may look
+ like this (common log format):
+ 
+ 127.0.0.1 - - [23/Oct/2001:14:04:56 +0200] "HEAD /~conrad/hello_world_gen.swf HTTP/1.0" 200 0
+ 
+ or like this (combined log format):
+ 
+ 127.0.0.1 - - [23/Oct/2001:14:04:56 +0200] "HEAD /~conrad/hello_world_gen.swf HTTP/1.0" 200 0 "http://localhost/~conrad/hello_world.html" "Mozilla 4"
+ 
+ The meaning of the (space-separated) fields is as follows:
+ 
+ 1. Client IP-address
+ 2. Remote logname (from ident (RFC-...) lookup - always "-" in publicfile)
+ 3. Authenticated username (from Basic-Auth - always "-" in publicfile)
+ 4. Human-readable timestamp enclosed in []
+ 5. The request line enclosed in ""
+ 6. HTTP status code
+ 7. Number of bytes sent excluding HTTP headers
+ 8. Referrer information sent by the browser
+ 9. User-Agent information sent by the browser
+ 
+ These log lines can easily be distinguished from other log lines using the
+ second field: this will always be "-" for common and combined log lines,
+ and "read" or "dir" for publicfile standard log lines (see
+ http://cr.yp.to/publicfile/log.html). The script "splitPublicfileLog.pl"
+ may serve as an example how to split the log into interesting parts.
+ 
+ The logformat to be used can be selected via environment variables:
+ 
+  - HTTPD_LOG_COMMON selects common log format
+  - HTTPD_LOG_COMBINED selects combined log format
+ 
+ The latter takes precedence over the former.
+ 
diff -rNC3 publicfile-0.52.orig/httpd.c publicfile-0.52/httpd.c
*** publicfile-0.52.orig/httpd.c	Tue Nov  9 08:23:46 1999
--- publicfile-0.52/httpd.c	Fri Nov  9 10:10:51 2001
***************
*** 15,20 ****
--- 15,102 ----
  #include "substdio.h"
  #include "error.h"
  #include "getln.h"
+ #include "caltime.h"
+ #include "subfd.h"
+ #include "env.h"
+ 
+ char logformat = 0;
+ char *remoteip;
+ char *months[] = {"Jan", "Feb", "Mar", "Apr", "May", "Jun",
+ 		  "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"};
+ stralloc referer = {0};
+ stralloc agent = {0};
+ 
+ char strnum[FMT_ULONG];
+ stralloc method = {0};
+ stralloc url = {0};
+ stralloc protocol = {0};
+ stralloc host = {0};
+ 
+ void log(char *code, unsigned long length) {
+   struct caltime ct;
+   int dummy1, dummy2;
+   struct tai now;
+ 
+   if (!logformat) return;
+ 
+   substdio_puts(subfderr,remoteip);
+   substdio_puts(subfderr," - - [");
+   tai_now(&now);
+   caltime_utc(&ct, &now, &dummy1, &dummy2);
+   if (ct.date.day < 10)
+     substdio_puts(subfderr,"0");
+   dummy1 = fmt_ulong(strnum,ct.date.day);
+   substdio_put(subfderr,strnum,dummy1);
+   substdio_puts(subfderr,"/");
+   if (ct.date.month > 0 && ct.date.month < 13)
+     substdio_puts(subfderr,months[ct.date.month-1]);
+   substdio_puts(subfderr,"/");
+   dummy1 = fmt_ulong(strnum,ct.date.year);
+   substdio_put(subfderr,strnum,dummy1);
+   substdio_puts(subfderr,":");
+   if (ct.hour < 10)
+     substdio_puts(subfderr,"0");
+   dummy1 = fmt_ulong(strnum,ct.hour);
+   substdio_put(subfderr,strnum,dummy1);
+   substdio_puts(subfderr,":");
+   if (ct.minute < 10)
+     substdio_puts(subfderr,"0");
+   dummy1 = fmt_ulong(strnum,ct.minute);
+   substdio_put(subfderr,strnum,dummy1);
+   substdio_puts(subfderr,":");
+   if (ct.second < 10)
+     substdio_puts(subfderr,"0");
+   dummy1 = fmt_ulong(strnum,ct.second);
+   substdio_put(subfderr,strnum,dummy1);
+   substdio_puts(subfderr," +0000] \"");
+   substdio_put(subfderr,method.s,method.len - 1); /* len includes \0 */
+   substdio_puts(subfderr," ");
+   substdio_put(subfderr,url.s,url.len);
+   substdio_puts(subfderr," ");
+   substdio_put(subfderr,protocol.s,protocol.len > 0 ? protocol.len - 1 : 0); /* len includes \0 */
+   substdio_puts(subfderr,"\" ");
+   substdio_put(subfderr,code,str_len(code) < 3 ? str_len(code) : 3);
+   substdio_puts(subfderr," ");
+   if (length > 0) {
+     dummy1 = fmt_ulong(strnum,length);
+     substdio_put(subfderr,strnum,dummy1);
+   } else {
+     substdio_puts(subfderr,"-");
+   }
+ 
+   if (logformat > 1) {
+     substdio_puts(subfderr," \"");
+     substdio_put(subfderr,referer.s,referer.len);
+     substdio_puts(subfderr,"\" \"");
+     substdio_put(subfderr,agent.s,agent.len);
+     substdio_puts(subfderr,"\"");
+   }
+ 
+   substdio_puts(subfderr," \"");
+   substdio_put(subfderr,host.s,host.len);
+   substdio_puts(subfderr,"\"\n");
+   substdio_flush(subfderr);
+ }
  
  int safewrite(int fd,char *buf,int len)
  {
***************
*** 42,54 ****
    substdio_flush(&out);
  }
  
- char strnum[FMT_ULONG];
- 
- stralloc protocol = {0};
  int protocolnum = 0;
- stralloc method = {0};
- stralloc url = {0};
- stralloc host = {0};
  stralloc path = {0};
  stralloc ims = {0};
  int flagbody = 1;
--- 124,130 ----
***************
*** 93,98 ****
--- 169,175 ----
      out_puts("</body></html>\r\n");
    }
    out_flush();
+   log(code, str_len(message) + 28);
    if (protocolnum >= 2) {
      shutdown(1,1);
      sleep(1); /* XXX */
***************
*** 105,108 ****
--- 182,186 ----
  
  void get(void)
  {
+   char *code;
    unsigned long length;
***************
*** 133,143 ****
      tai_now(&now);
      if (!httpdate(&mtimestr,&mtime)) _exit(21);
      if ((ims.len < mtimestr.len) || byte_diff(mtimestr.s,mtimestr.len,ims.s))
!       header("200 ","OK");
      else {
!       header("304 ","OK");
        flagbody = 0;
      }
      if (tai_less(&mtime,&now)) {
        tai_sub(&mtimeage,&now,&mtime);
        if (tai_approx(&mtimeage) >= 60.0) {
--- 211,222 ----
      tai_now(&now);
      if (!httpdate(&mtimestr,&mtime)) _exit(21);
      if ((ims.len < mtimestr.len) || byte_diff(mtimestr.s,mtimestr.len,ims.s))
!       code = "200 ";
      else {
!       code = "304 ";
        flagbody = 0;
      }
+     header(code,"OK");
      if (tai_less(&mtime,&now)) {
        tai_sub(&mtimeage,&now,&mtime);
        if (tai_approx(&mtimeage) >= 60.0) {
***************
*** 166,171 ****
--- 245,253 ----
          if (r == 0) break;
          out_put(filebuf,r);
        }
+     else
+       length = 0;
+     log(code,length);
      out_flush();
      _exit(0);
    }
***************
*** 180,185 ****
--- 262,270 ----
        out_put(filebuf,r);
        out_puts("\r\n");
      }
+   else
+     length = 0;
+   log(code,length);
  
    close(fd);
  }
***************
*** 216,221 ****
--- 301,314 ----
  
    sig_pipeignore();
  
+   if (env_get("HTTPD_LOG_COMMON"))
+     logformat = 1;
+   if (env_get("HTTPD_LOG_COMBINED"))
+     logformat = 2;
+   remoteip = env_get("TCPREMOTEIP");
+   if (!remoteip)
+     remoteip = "0";
+ 
    for (;;) {
      readline();
  
***************
*** 227,232 ****
--- 320,327 ----
      if (!stralloc_copys(&path,"")) _exit(21);
      if (!stralloc_copys(&protocol,"")) _exit(21);
      if (!stralloc_copys(&ims,"")) _exit(21);
+     if (!stralloc_copys(&referer,"")) _exit(21);
+     if (!stralloc_copys(&agent,"")) _exit(21);
      protocolnum = 2;
  
      spaces = 0;
***************
*** 294,299 ****
--- 389,398 ----
              barf("412 ","I do not accept If-None-Match");
            if (case_startb(field.s,field.len,"if-unmodified-since:"))
              barf("412 ","I do not accept If-Unmodified-Since");
+           if (case_startb(field.s,field.len,"referer:"))
+ 	    if (!stralloc_copytrim(&referer,field.s + 8,field.len - 8)) _exit(21);
+           if (case_startb(field.s,field.len,"user-agent:"))
+ 	    if (!stralloc_copytrim(&agent,field.s + 11,field.len - 11)) _exit(21);
            if (case_startb(field.s,field.len,"host:"))
              if (!host.len)
                for (i = 5;i < field.len;++i)
diff -rNC3 publicfile-0.52.orig/splitPublicfileLog.pl publicfile-0.52/splitPublicfileLog.pl
*** publicfile-0.52.orig/splitPublicfileLog.pl	Thu Jan  1 01:00:00 1970
--- publicfile-0.52/splitPublicfileLog.pl	Fri Nov  9 10:09:45 2001
***************
*** 0 ****
--- 1,119 ----
+ #!/usr/bin/perl -w
+ 
+ # This script will read lines from stdin. All lines starting with
+ # "a.b.c.d - " (for an IP-number a.b.c.d) will be written to the
+ # "access logfile" given as the first argument. Anything else will be
+ # prepended with a timestamp and
+ # written to the "error logfile" given as the second argument.
+ # Note that the messages usually are not really *error* messages, but
+ # simply the standard publicfile output.
+ 
+ use strict;
+ use Symbol;
+ 
+ if ($#ARGV < 0 || $#ARGV > 3) {
+     &usage();
+ }
+ 
+ local $main::logdir;
+ local $main::accessfile;
+ local $main::errorfile;
+ local $main::multilog;
+ local %main::handles = ();
+ 
+ if ($ARGV[0] eq "-m") {
+     $main::multilog = 1;
+     if ($#ARGV != 3) { &usage(); }
+     $main::logdir = $ARGV[1]."/";
+     $main::accessfile = $main::logdir.$ARGV[2];
+     $main::errorfile = $main::logdir.$ARGV[3];
+ } else {
+     $main::multilog = 0;
+     if ($#ARGV != 1) { &usage(); }
+     $main::accessfile = $ARGV[0];
+     $main::errorfile = $ARGV[1];
+ }
+ 
+ #open(ACCESS, ">>".$ARGV[0]) || die("Can't open access logfile!");
+ #open(ERROR, ">>".$ARGV[1]) || die("Can't open error logfile!");
+ $SIG{PIPE} = \&closeAndExit;
+ $SIG{TERM} = \&closeAndExit;
+ $SIG{QUIT} = \&closeAndExit;
+ $SIG{INT} = \&closeAndExit;
+ $SIG{HUP} = \&closeAll;
+ 
+ while($_ = <STDIN>) {
+     if (/^\d+\.\d+\.\d+\.\d+ - /) {
+ 	my $host = "";
+ 	if ($main::multilog && / \"([^\"\/]*)\"[\r\n]*$/) {
+ 	    $_ = $`."\n";
+ 	    $host = $1;
+ 	}
+ 	my $fh = &getHandle($host);
+ 	print $fh $_;
+     } else {
+ 	my $fh = &getHandle($main::errorfile);
+ 	print $fh time()." $_";
+     }
+ }
+ 
+ closeAndExit("End of input");
+ 
+ sub closeAll {
+     foreach my $key (keys %main::handles) {
+ 	my $fh = $main::handles{$key};
+ 	close $fh;
+ 	delete $main::handles{$key};
+     }
+ }
+ 
+ sub closeAndExit {
+     my $reason = shift;
+     my $now = time();
+ 
+     my $fh = &getHandle($main::errorfile);
+     print $fh "$now $reason received\n";
+     print STDERR "$now $reason received\n";
+     &closeAll();
+     exit 0;
+ }
+ 
+ sub getHandle {
+     my $key = shift;
+     my $res = 0;
+ 
+     if (defined($main::handles{$key})) {
+ 	return $main::handles{$key};
+     }
+ 
+     my $fh = Symbol::gensym();
+     if ($key eq $main::errorfile) {
+ 	open($fh, ">>".$main::errorfile) || die("Can't open error logfile!");
+     } else {
+ 	if ($main::multilog) {
+             my $fn = $main::logdir.$key;
+             if (-f $fn && -w $fn) {
+ 	        $res = open($fh, ">>$fn");
+ 	    }
+         }
+ 	if (!$res) {
+ 	    open($fh, ">>".$main::accessfile) || die("Can't open error logfile!");
+ 	}
+     }
+ 
+     $main::handles{$key} = $fh;
+     return $fh;
+ }
+ 
+ sub usage {
+     print STDERR "Usage: $0 <access-path> <error-path>\n";
+     print STDERR "   or: $0 -m <logdir> <access-file> <error-file>\n";
+     print STDERR "The first form will separate publicfile standard messages from \"common\" log\n";
+     print STDERR "lines. The second form will write \"common\" lines to <logdir>/<hostname>\n";
+     print STDERR "where <hostname> is the host given in the request. The file <logdir>/<hostname>\n";
+     print STDERR "must exist before it will be used. Anything for which no suitable logfile can\n";
+     print STDERR "be found will be logged to <logdir>/<access-path>.\n";
+ 
+     exit 1;
+ }
+ 
diff -rNC3 publicfile-0.52.orig/stralloc.h publicfile-0.52/stralloc.h
*** publicfile-0.52.orig/stralloc.h	Tue Nov  9 08:23:46 1999
--- publicfile-0.52/stralloc.h	Fri Nov  9 10:09:45 2001
***************
*** 12,17 ****
--- 12,18 ----
  extern int stralloc_copys();
  extern int stralloc_cats();
  extern int stralloc_copyb();
+ extern int stralloc_copytrim();
  extern int stralloc_catb();
  extern int stralloc_append(); /* beware: this takes a pointer to 1 char */
  extern int stralloc_starts();
diff -rNC3 publicfile-0.52.orig/stralloc_opytrim.c publicfile-0.52/stralloc_opytrim.c
*** publicfile-0.52.orig/stralloc_opytrim.c	Thu Jan  1 01:00:00 1970
--- publicfile-0.52/stralloc_opytrim.c	Fri Nov  9 10:09:45 2001
***************
*** 0 ****
--- 1,17 ----
+ #include "stralloc.h"
+ 
+ int stralloc_copytrim(sa,s,n)
+ stralloc *sa;
+ char *s;
+ unsigned int n;
+ {
+   while (n > 0 &&
+ 	 (s[n] == ' ' || s[n] == '\t' || s[n] == '\n' || s[n] == '\r')) n--;
+   while (n > 0 &&
+ 	 (s[0] == ' ' || s[0] == '\t' || s[0] == '\n' || s[0] == '\r')) {
+     n--;
+     s++;
+   }
+ 
+   return stralloc_copyb(sa,s,n);
+ }