| 1 |
using System; |
|
| 2 |
using System.Collections.Generic; |
|
| 3 |
using System.Linq; |
|
| 4 |
using System.Text; |
|
| 5 |
|
|
| 6 |
namespace TurnerSoftware.RobotsExclusionTools.Helpers |
|
| 7 |
{
|
|
| 8 |
public class PathComparisonUtility |
|
| 9 |
{
|
|
| 10 |
public bool IsAllowed(SiteAccessEntry accessEntry, Uri requestUri) |
|
| 11 | 1 |
{
|
| 12 | 1 |
var requestPath = requestUri.PathAndQuery; |
| 13 |
|
|
| 14 |
//Robots file is always unrestricted
|
|
| 15 |
if (requestUri.LocalPath == "/robots.txt") |
|
| 16 | 1 |
{
|
| 17 | 1 |
return true; |
| 18 |
}
|
|
| 19 |
|
|
| 20 |
//If no entry is defined, the robot is allowed access by default
|
|
| 21 |
if (accessEntry == null) |
|
| 22 | 1 |
{
|
| 23 | 1 |
return true; |
| 24 |
}
|
|
| 25 |
|
|
| 26 | 1 |
foreach (var rule in accessEntry.PathRules) |
| 27 | 1 |
{
|
| 28 |
if (rule.RuleType == PathRuleType.Disallow && rule.Path == string.Empty) |
|
| 29 | 1 |
{
|
| 30 | 1 |
return true; |
| 31 |
}
|
|
| 32 |
else if (PathMatch(rule.Path, requestPath, StringComparison.InvariantCulture)) |
|
| 33 | 1 |
{
|
| 34 | 1 |
return rule.RuleType == PathRuleType.Allow; |
| 35 |
}
|
|
| 36 | 1 |
}
|
| 37 |
|
|
| 38 | 1 |
return true; |
| 39 | 1 |
}
|
| 40 |
|
|
| 41 |
public bool PathMatch(string sourceRecord, string uriPath, StringComparison comparison) |
|
| 42 | 1 |
{
|
| 43 | 1 |
var sourcePieces = sourceRecord.Split(new[] { '*' }, StringSplitOptions.RemoveEmptyEntries).ToArray(); |
| 44 | 1 |
var lastPiece = sourcePieces.LastOrDefault(); |
| 45 | 1 |
var mustMatchToEnd = false; |
| 46 | 1 |
var mustMatchToStart = true; |
| 47 |
|
|
| 48 |
if (lastPiece.EndsWith("$")) |
|
| 49 | 1 |
{
|
| 50 |
//Remove the last dollar sign from the last piece
|
|
| 51 | 1 |
lastPiece = lastPiece.Substring(0, lastPiece.Length - 1); |
| 52 | 1 |
sourcePieces[sourcePieces.Length - 1] = lastPiece; |
| 53 | 1 |
mustMatchToEnd = true; |
| 54 | 1 |
}
|
| 55 |
|
|
| 56 |
if (sourceRecord.StartsWith("*")) |
|
| 57 | 1 |
{
|
| 58 | 1 |
mustMatchToStart = false; |
| 59 | 1 |
}
|
| 60 |
|
|
| 61 | 1 |
var offsetPosition = 0; |
| 62 |
|
|
| 63 |
for (int i = 0, l = sourcePieces.Length; i < l; i++) |
|
| 64 | 1 |
{
|
| 65 | 1 |
var piece = sourcePieces[i]; |
| 66 | 1 |
var indexPosition = uriPath.IndexOf(piece, offsetPosition, comparison); |
| 67 |
|
|
| 68 |
if (mustMatchToStart && offsetPosition == 0 && indexPosition > 0) |
|
| 69 | 1 |
{
|
| 70 | 1 |
return false; |
| 71 |
}
|
|
| 72 |
|
|
| 73 |
if (indexPosition >= offsetPosition) |
|
| 74 | 1 |
{
|
| 75 | 1 |
offsetPosition = piece.Length; |
| 76 | 1 |
}
|
| 77 |
else
|
|
| 78 | 1 |
{
|
| 79 | 1 |
return false; |
| 80 |
}
|
|
| 81 |
|
|
| 82 | 1 |
}
|
| 83 |
|
|
| 84 |
if (mustMatchToEnd) |
|
| 85 | 1 |
{
|
| 86 | 1 |
var endOffset = uriPath.Length - lastPiece.Length; |
| 87 |
if (uriPath.IndexOf(lastPiece, endOffset, comparison) == -1) |
|
| 88 | 1 |
{
|
| 89 | 1 |
return false; |
| 90 |
}
|
|
| 91 | 1 |
}
|
| 92 |
|
|
| 93 | 1 |
return true; |
| 94 | 1 |
}
|
| 95 |
}
|
|
| 96 |
}
|
Read our documentation on viewing source code .