Merge pull request #98 from pbthif/main

Add support for virtual hosted style S3 URLs in String.asS3ObjectIdentifier()
amzn · Aug 31, 2021 · e45bc7f · e45bc7f
2 parents a31e737 + da243c5
commit e45bc7f
Show file tree

Hide file tree

Showing 2 changed files with 64 additions and 40 deletions.
diff --git a/Sources/S3Client/String+asS3ObjectIdentifier.swift b/Sources/S3Client/String+asS3ObjectIdentifier.swift
@@ -13,6 +13,7 @@ public struct S3ObjectIdentifer: Equatable {
     internal static let s3Prefix = "s3://"
     internal static let httpsPrefix = "https://"
     internal static let httpPrefix = "http://"
+    internal static let s3EndpointRegex = #"^https?:\/\/(.+\.)?s3[.-][a-z0-9-]+\."#
 
     public let bucketName: String
     public let keyPath: String
@@ -41,43 +42,46 @@ public extension String {
             let nonPrefixedUrl = self.dropFirst(S3ObjectIdentifer.s3Prefix.count)
 
             return asS3ObjectIdentifierFromNonPrefixedUrl(nonPrefixedUrl: nonPrefixedUrl)
-        } else if self.starts(with: S3ObjectIdentifer.httpsPrefix) {
-            // get the url without the scheme - of the form {host}/{bucket}/{key+}
-            let droppedPrefix = self.dropFirst(S3ObjectIdentifer.httpsPrefix.count)
-
-            // get the index of the separator between the host and the bucket
-            guard let nextUrlSeparator = getIndexOfNextUrlSeparator(url: droppedPrefix) else {
-                return nil
-            }
-
-            let bucketStartIndex = droppedPrefix.index(nextUrlSeparator,
-                                                               offsetBy: 1)
-            // get the url without the scheme or the host -
-            // of the form {bucket}/{key+}
-            let nonPrefixedUrl = droppedPrefix[bucketStartIndex...]
-
-            return asS3ObjectIdentifierFromNonPrefixedUrl(nonPrefixedUrl: nonPrefixedUrl)
-        } else if self.starts(with: S3ObjectIdentifer.httpPrefix) {
-            // get the url without the scheme - of the form {host}/{bucket}/{key+}
-            let droppedPrefix = self.dropFirst(S3ObjectIdentifer.httpPrefix.count)
-
-            // get the index of the separator between the host and the bucket
-            guard let nextUrlSeparator = getIndexOfNextUrlSeparator(url: droppedPrefix) else {
-                return nil
-            }
-
-            let bucketStartIndex = droppedPrefix.index(nextUrlSeparator,
-                                                               offsetBy: 1)
-            // get the url without the scheme or the host -
-            // of the form {bucket}/{key+}
-            let nonPrefixedUrl = droppedPrefix[bucketStartIndex...]
-
-            return asS3ObjectIdentifierFromNonPrefixedUrl(nonPrefixedUrl: nonPrefixedUrl)
+        } else if self.starts(with: S3ObjectIdentifer.httpsPrefix) || self.starts(with: S3ObjectIdentifer.httpPrefix) {
+            return asS3ObjectIdentifierFromHttpOrHttps()
         }
 
         return nil
     }
 
+    /// Tries to parse the bucket and key names from an HTTP or HTTPS URL.
+    private func asS3ObjectIdentifierFromHttpOrHttps() -> S3ObjectIdentifer? {
+        guard let url = URL(string: self) else {
+            return nil
+        }
+
+        let urlPath = url.path.dropFirst()
+
+        guard let regex = try? NSRegularExpression(pattern: S3ObjectIdentifer.s3EndpointRegex, options: []) else {
+            return nil
+        }
+
+        let searchRange = NSRange(self.startIndex..<self.endIndex, in: self)
+        let match = regex.firstMatch(in: self, options: [], range: searchRange)
+        if let match = match,
+            match.numberOfRanges > 0,
+            let bucketRange = Range(match.range(at: 1), in: self),
+            !bucketRange.isEmpty,
+            self[bucketRange].count > 1 {
+            // If the capture group for the regex is not empty, the URL is the virtual hosted style, for example:
+            // https://bucket.s3.amazonaws.com/key
+            // The capture group is the bucket name (with trailing dot) and the URL path is the key name
+            let bucketName = String(self[bucketRange].dropLast())
+            let keyName = String(urlPath)
+            return S3ObjectIdentifer(bucketName: bucketName, keyPath: keyName)
+        }
+
+        // If the regex capture group is empty, the URL is in the path style, for example:
+        // https://s3.amazonaws.com/bucket/key
+        // Both the bucket and key names are in the URL path
+        return asS3ObjectIdentifierFromNonPrefixedUrl(nonPrefixedUrl: Substring(urlPath))
+    }
+
     private func getIndexOfNextUrlSeparator(url: Substring) -> String.Index? {
         #if swift(>=4.2)
         return url.firstIndex(of: "/")

diff --git a/Tests/S3ClientTests/S3ClientTests.swift b/Tests/S3ClientTests/S3ClientTests.swift
@@ -26,25 +26,45 @@ class S3ClientTests: XCTestCase {
     }
 
     func testValidHttpsUri() throws {
-        let s3Uri = "https://host/bucketName/the/key/path"
-
-        let identifier = s3Uri.asS3ObjectIdentifier()
+        let s3Uris = [
+            "https://s3.amazonaws.com/bucketName/the/key/path",
+            "https://s3-abc.amazonaws.com/bucketName/the/key/path",
+            "https://s3.us-east-1.amazonaws.com/bucketName/the/key/path",
+            "https://s3-abc.us-east-1.amazonaws.com/bucketName/the/key/path",
+            "https://bucketName.s3.amazonaws.com/the/key/path",
+            "https://bucketName.s3-abc.amazonaws.com/the/key/path",
+            "https://bucketName.s3.us-east-1.amazonaws.com/the/key/path",
+            "https://bucketName.s3-abc.us-east-1.amazonaws.com/the/key/path",
+        ]
 
         let expected = S3ObjectIdentifer(bucketName: "bucketName",
                                          keyPath: "the/key/path")
 
-        XCTAssertEqual(expected, identifier)
+        for s3Uri in s3Uris {
+            let identifier = s3Uri.asS3ObjectIdentifier()
+            XCTAssertEqual(expected, identifier)
+        }
     }
 
     func testValidHttpUri() throws {
-        let s3Uri = "http://host/bucketName/the/key/path"
-
-        let identifier = s3Uri.asS3ObjectIdentifier()
+        let s3Uris = [
+            "http://s3.amazonaws.com/bucketName/the/key/path",
+            "http://s3-abc.amazonaws.com/bucketName/the/key/path",
+            "http://s3.us-east-1.amazonaws.com/bucketName/the/key/path",
+            "http://s3-abc.us-east-1.amazonaws.com/bucketName/the/key/path",
+            "http://bucketName.s3.amazonaws.com/the/key/path",
+            "http://bucketName.s3-abc.amazonaws.com/the/key/path",
+            "http://bucketName.s3.us-east-1.amazonaws.com/the/key/path",
+            "http://bucketName.s3-abc.us-east-1.amazonaws.com/the/key/path",
+        ]
 
         let expected = S3ObjectIdentifer(bucketName: "bucketName",
                                          keyPath: "the/key/path")
 
-        XCTAssertEqual(expected, identifier)
+    for s3Uri in s3Uris {
+            let identifier = s3Uri.asS3ObjectIdentifier()
+            XCTAssertEqual(expected, identifier)
+        }
     }
 
     func testInvalidS3UriPrefix() throws {