| Index: trunk/udplog/filters/lp-filter.c |
| — | — | @@ -2,18 +2,24 @@ |
| 3 | 3 | #include <stdio.h> |
| 4 | 4 | #include <string.h> |
| 5 | 5 | |
| 6 | | -char burl[]="wikimediafoundation.org/"; |
| | 6 | +char url1[] ="wikimediafoundation.org/"; |
| | 7 | +char url2[] ="donate.wikimedia.org/"; |
| 7 | 8 | |
| 8 | 9 | |
| 9 | 10 | main() { |
| 10 | | - char line[10240]; |
| 11 | | - char title[10240]; |
| 12 | | - char *urlstart, *urlend; |
| | 11 | + char line[65534]; |
| | 12 | + char title[65534]; |
| 13 | 13 | char *t = 0; |
| | 14 | + char *u = 0; |
| | 15 | + |
| | 16 | + //to cut down on the processing time: Guess first. |
| | 17 | + //longest filter is 25 characters. |
| | 18 | + //I'm allowing for one heck of a subdomain, here. |
| | 19 | + int search_length = 75; |
| 14 | 20 | |
| 15 | 21 | while (!feof(stdin)) { |
| 16 | 22 | char *r; |
| 17 | | - r=fgets(line, 10000, stdin); |
| | 23 | + r=fgets(line, 65534, stdin); |
| 18 | 24 | |
| 19 | 25 | int pos=0; |
| 20 | 26 | t = line; |
| — | — | @@ -28,13 +34,26 @@ |
| 29 | 35 | } |
| 30 | 36 | if (!t) |
| 31 | 37 | continue; |
| 32 | | - urlstart = t; |
| 33 | | - urlend = strstr(urlstart, " "); |
| 34 | | - strncpy(title, urlstart, urlend-urlstart); |
| 35 | | - title[urlend-urlstart]=0; |
| 36 | | - if (strstr(title, burl) ) |
| 37 | | - printf("%s", line); |
| 38 | 38 | |
| | 39 | + strncpy(title, t, search_length); |
| | 40 | + title[search_length]=0; |
| | 41 | + |
| | 42 | + if (strstr(title, url1) || strstr(title, url2) ){ |
| | 43 | + u = strstr(title, " "); |
| | 44 | + |
| | 45 | + if (!u){ //no spaces, just do it. |
| | 46 | + printf("%s", line); |
| | 47 | + } else { |
| | 48 | + //make sure it was before the first space. |
| | 49 | + t = strstr(title, url1); |
| | 50 | + if (!t) |
| | 51 | + t = strstr(title, url2); |
| | 52 | + if ( (t) && t < u ) { |
| | 53 | + printf("%s", line); |
| | 54 | + } |
| | 55 | + } |
| | 56 | + } |
| | 57 | + |
| 39 | 58 | } |
| 40 | 59 | |
| 41 | 60 | } |