Link Checker

1
<?php
2
/*
3
  Takes a URL on the command line and parses it for links and URLs
4
  Outputs the status of every link and URL as either OK or BROKEN
5
*/
6

7
function error( $str ) {
8
  fwrite(STDERR, $str);
9
  exit(1);
10
}
11

12
function test_url( $url ) {
13
  fwrite(STDOUT, "        Checking \"$url\": ");
14
  $handle = @fopen($url, 'r');
15
  if ($handle) {
16
    fwrite(STDOUT, "OK\n");
17
    fclose($handle);
18
  }
19
  else {
20
    fwrite(STDOUT, "*BROKEN*\n");
21
  }
22
}
23

24
if ($argc != 2) {
25
  error("syntax: url_tester.php url\n");
26
}
27

28
$url = $argv[1];
29
/* prefix with http:// */
30
$url = preg_replace("/^www\./","http://www.",$url);
31
fwrite(STDOUT, "Testing $url for broken links:\n");
32
$file_contents = @file_get_contents($url);
33

34
if (!$file_contents) {
35
  error("Error reading from $url.  Try again later");
36
}
37

38
/* finds all anchor (<a href=) links and ends when they hit a quote */
39
$url_pattern = "!<a href=(?:\")?([^\" >]+)!i";
40
preg_match_all($url_pattern, $file_contents, $url_list, PREG_PATTERN_ORDER);
41
fwrite(STDOUT, "    ANCHOR (<a href=) URLS\n");
42
foreach($url_list[1] as $link) {
43
  if (preg_match("!^(http://|www.)!i", $link)) {
44
    /* prefix with http:// */
45
    $link = preg_replace("/^www\./","http://www.",$link);
46
    test_url($link);
47
  }
48
  else {
49
    if (preg_match("!/$!",$url) and $link[0] == '/' ) {
50
      test_url($url . substr($link,1));
51
    }
52
    else if (preg_match("!/$!",$url) xor $link[0] == '/' ) {
53
      test_url($url . $link);
54
    }
55
    else {
56
      echo $url, '---', $link;
57
      test_url($url . $link);
58
    }
59
  }
60
}
61

62
/* finds a URL in the source, and ends it when it hits a quote or space */
63
$url_pattern = "!http://(?:[^/\" >:]*)(?::(?:[0-9]*))?(?:/[^ >\"]*)?!i";
64
preg_match_all($url_pattern, $file_contents, $url_list, PREG_PATTERN_ORDER);
65
fwrite(STDOUT, "    ALL URLS\n");
66
foreach($url_list[0] as $link) {
67
  test_url($link);
68
}
69
?>