/* Here's a complete sample program in C, written at the end of this week's C
and C++ training, as several delegates will be using some C++ but more C, and we
were looking to pull all the various elements together.

The program opens and reads a web server access log line by line, and checks if
each line in turn is a search arrival from Google. If it is, the data is stored
into a structure for later processing.

In this sample, I chose not to do a great deal with the data once we had read it
in, as the reading and settig up of the structure is a good example, and the further
processing would be more code along the same lines. But I did report on a sample
of 20 of the records stored, to show that they were set up correctly. */

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <regex.h>

#define CAPSTEP 10
#define NSAMPS 20

/* Sample data line ... */

/* 98.249.0.47 - - [28/Jun/2010:03:31:47 +0100] "GET /mouth/158_MySQL-
LEFT-JOIN-and-RIGHT-JOIN-INNER-JOIN-and-OUTER-JOIN.html HTTP/1.1" 200
65651 "http://www.google.com/search?hl=&q=left+join+vs+right+join&sour
ceid=navclient-ff&rlz=1B3GGLL_enUS358US358&ie=UTF-8" "Mozilla/5.0 (Win
dows; U; Windows NT 6.1; en-US; rv:1.9.2.4) Gecko/20100611 Firefox/3.6
.4 GTB7.0" */

typedef struct {
        char * ip;
        char * searchstring;
        int status;
        int size;
        } Arrival ;

/* Extract a character string from within another string and return its
address on the heap (so that it remains in memory until released with free
or cfree, or at the end of the program */

char * extract (char * start, int len)
         {
        char * here = malloc((len + 1) * sizeof(char)); /* Allocate Memory */
        strncpy(here,start,len); /* Copy string to the heap */
        *(here+len) = '\0'; /* Add null terminator */
        return here;
        }

main () {

        FILE *fh;

        Arrival *google; /* Pointer to an array of arrivals */
        int capacity = 0; /* Current capacity of arrivals area */

        char buffer[12049]; /* Read buffer and counters */
        int records = 0;
        int gooey_records = 0;
        int stepper;
        int k;

                                /* Regular Expression to match arrival */
        char * querystring = "\\.google\\..*[?&]q=([^\"& ]+)";
        regex_t query;
        regmatch_t matches[2];
        char *qs;

                                /* Regular expression for status and byte count */
        char * resultstring = "\" ([0-9]+) ([0-9]+) \"";
        regex_t result;
        regmatch_t info[3];

        int fspace; /* Temporary variables for setting up each Arrival record */
        int status;
        char * ipadd;
        char * ipheap;

/* Open File */
        fh = fopen("ac_20100630","r"); /* Should read name from command line? */

/* Set up for a whole lot (unknown number) of arrivals */
        google = calloc(0,sizeof(Arrival));

        regcomp(&query,querystring,REG_EXTENDED); /* Compile regexes just once! */
        regcomp(&result,resultstring,REG_EXTENDED);

/* Loop through file */
        while (fgets(buffer,12048,fh)) {
                records ++;

                /* If it's a google arrival, find the bits we want in the line */

                if (! regexec(&query,buffer,2,matches,0)) {
                        regexec(&result,buffer,3,info,0);
                        ipadd = strtok(buffer," ");

                        /* Extract those bits into clean variables */

                        ipheap = malloc((strlen(ipadd)+1) * sizeof(char));
                        strcpy(ipheap,ipadd);

                        status = atoi(extract(buffer + (int)info[1].rm_so,3));
                        fspace = atoi(extract(buffer + (int)info[2].rm_so,
                                (int)info[2].rm_eo - (int)info[2].rm_so));
                        qs = extract(buffer + (int)matches[1].rm_so,
                                (int)matches[1].rm_eo - (int)matches[1].rm_so);

/* Do we have room already allocated to store result? If not, increase capacity */

                        if (capacity <= gooey_records) {
                                google = realloc (google,
                                        (capacity + CAPSTEP) * sizeof(Arrival));
                                capacity += CAPSTEP;
                                }

/* If it's an arrival, add it to saved ones */

                        (google+gooey_records)->status = status;
                        (google+gooey_records)->size = fspace;
                        (google+gooey_records)->searchstring = qs;
                        (google+gooey_records)->ip = ipheap;

                        gooey_records++;
                        }

                }

/* Prove it's there! - Sample it*/

        stepper = gooey_records / NSAMPS;

        for (k=0; k<gooey_records; k += stepper) {
                printf ("%6d %3d %7d %16s %s\n", k,
                        (google+k)->status ,
                        (google+k)->size ,
                        (google+k)->ip ,
                        (google+k)->searchstring ) ;
        }

        printf("There were %d records\n",records);
        printf("There were %d Google arrivals\n",gooey_records);

        }

/* Here is some sample output:

wizzard:c graham$ ./loglook
     0 200 73390 88.77.57.79 perl+script+md5
   766 200 35818 116.197.178.83 perl+how+to+unset+a+env+variable
  1532 200 116072 115.113.232.135 telnet+program+using+tcl
  2298 200 65755 202.164.34.26 inner+join+in+mysql
  3064 200 13699 72.179.17.70 php+add+1+variable
  3830 200 57144 24.131.249.242 programming+an+image+uploader
  4596 200 17786 168.159.160.57 monitoring+process+using+perl
  5362 200 17493 196.213.166.66 python+elif
  6128 200 16317 213.55.110.198 c%2B%2B+function+example
  6894 200 22856 221.120.194.162 learn+linux+administration
  7660 200 19953 192.71.200.230 java+running+system+commands
  8426 200 66187 82.69.172.105 left+join+inner+join
  9192 200 25809 222.124.114.65 Python+thread
  9958 200 13875 188.129.69.221 python+gui+example
 10724 200 30441 67.127.45.246 php+adding+%22%5C%22
 11490 200 38712 208.50.255.30 load+word+ribbon+with+database+connection
 12256 200 20468 167.206.168.187 splitting+string+in+python+example
 13022 200 35531 198.206.219.38 what+is+_+in+perl
 13788 200 33595 129.74.86.78 perl+stdout+%3E
 14554 200 37758 41.238.231.219 debug+cgi+perl+file
 15320 200 17356 202.32.73.47 dopost+response
There were 157995 records
There were 15325 Google arrivals
wizzard:c graham$

*/