Merge branch 'protocol-refactor' into 'main'

Protocol refactor

See merge request reidrac/spacebeans!2

2 files changed, 628 insertions, 0 deletions

diff --git a/server/test/src/handlers/GeminiHandlerSpec.scala b/server/test/src/handlers/GeminiHandlerSpec.scala
new file mode 100644
index 0000000..f3545a0
--- /dev/null
+++ b/server/test/src/handlers/GeminiHandlerSpec.scala
@@ -0,0 +1,528 @@
+package net.usebox.gemini.server.handlers
+
+import java.net.URI
+
+import org.scalatest.flatspec.AnyFlatSpec
+import org.scalatest.matchers.should.Matchers
+
+import net.usebox.gemini.server._
+
+class GeminiHandlerSpec extends AnyFlatSpec with Matchers {
+
+  val handler = new GeminiHandler(TestData.conf)
+  val handlerUserDir = new GeminiHandler(TestData.confUserDir)
+  val handlerCgi = new GeminiHandler(TestData.cgiConf)
+
+  def handleWith(
+      geminiHandler: GeminiHandler,
+      req: String,
+      remoteAddr: String = "127.0.0.1"
+  ): Response =
+    geminiHandler.handle(req, URI.create(req), remoteAddr)
+
+  behavior of "handle"
+
+  it should "handle host case insensitive" in {
+    handleWith(handler, "gemini://LOCALHOST/") should be(a[Success])
+  }
+
+  it should "return proxy request refused when the vhost is not found" in {
+    handleWith(handler, "gemini://otherhost/") should be(
+      a[ProxyRequestRefused]
+    )
+  }
+
+  it should "return bad request when user info is present" in {
+    handleWith(handler, "gemini://user@localhost/") should be(
+      a[BadRequest]
+    )
+  }
+
+  it should "return bad request when the path is out of root dir" in {
+    handleWith(handler, "gemini://localhost/../../") should be(
+      a[BadRequest]
+    )
+  }
+
+  it should "redirect to normalize the URL" in {
+    handleWith(handler, "gemini://localhost/./") should be(
+      a[PermanentRedirect]
+    )
+  }
+
+  it should "return not found if the path doesn't exist" in {
+    handleWith(
+      handler,
+      "gemini://localhost/doesnotexist"
+    ) should be(a[NotFound])
+  }
+
+  it should "return not found if a dot file" in {
+    handleWith(handler, "gemini://localhost/.dotfile") should be(
+      a[NotFound]
+    )
+  }
+
+  it should "return success on reading file" in {
+    handleWith(
+      handler,
+      "gemini://localhost/index.gmi"
+    ) should matchPattern {
+      case Success(_, "text/gemini", Some(_), 25L) =>
+    }
+  }
+
+  it should "redirect and normalize request on a directory" in {
+    handleWith(handler, "gemini://localhost/dir") should be(
+      a[PermanentRedirect]
+    )
+  }
+
+  it should "return an existing index file when requesting a directory" in {
+    handleWith(handler, "gemini://localhost/") should matchPattern {
+      case Success(_, "text/gemini", Some(_), 25L) =>
+    }
+  }
+
+  it should "include gemini params for gemini MIME type" in {
+    handleWith(
+      new GeminiHandler(
+        TestData.conf.copy(virtualHosts =
+          List(TestData.conf.virtualHosts(0).copy(geminiParams = Some("test")))
+        )
+      ),
+      "gemini://localhost/index.gmi"
+    ) should matchPattern {
+      case Success(_, "text/gemini; test", Some(_), 25L) =>
+    }
+  }
+
+  behavior of "handler, directory listings"
+
+  it should "return a directory listing if is enabled and no index" in {
+    handleWith(handler, "gemini://localhost/dir/") should be(
+      a[DirListing]
+    )
+  }
+
+  it should "return a directory listing, directory listing flags: vhost flag false, directories flag true" in {
+    handleWith(
+      new GeminiHandler(
+        ServiceConf.initConf(
+          TestData.conf.copy(virtualHosts =
+            List(
+              TestData.conf
+                .virtualHosts(0)
+                .copy(
+                  directoryListing = false,
+                  directories = List(
+                    Directory(
+                      "dir/",
+                      directoryListing = Some(true),
+                      allowCgi = None
+                    )
+                  )
+                )
+            )
+          )
+        )
+      ),
+      "gemini://localhost/dir/"
+    ) should be(a[DirListing])
+  }
+
+  it should "return not found with no index, directory listing flags: vhost flag true, directories flag false" in {
+    handleWith(
+      new GeminiHandler(
+        ServiceConf.initConf(
+          TestData.conf.copy(virtualHosts =
+            List(
+              TestData.conf
+                .virtualHosts(0)
+                .copy(
+                  directoryListing = true,
+                  directories = List(
+                    Directory(
+                      "dir/",
+                      directoryListing = Some(false),
+                      allowCgi = None
+                    )
+                  )
+                )
+            )
+          )
+        )
+      ),
+      "gemini://localhost/dir/"
+    ) should be(a[NotFound])
+  }
+
+  it should "not apply directory listing override to subdirectories" in {
+    handleWith(
+      new GeminiHandler(
+        ServiceConf.initConf(
+          TestData.conf.copy(virtualHosts =
+            List(
+              TestData.conf
+                .virtualHosts(0)
+                .copy(
+                  directoryListing = false,
+                  directories = List(
+                    Directory(
+                      "dir/",
+                      directoryListing = Some(true),
+                      allowCgi = None
+                    )
+                  )
+                )
+            )
+          )
+        )
+      ),
+      "gemini://localhost/dir/sub/"
+    ) should be(
+      a[NotFound]
+    )
+  }
+
+  it should "return not found if directory listing is not enabled and no index" in {
+    handleWith(
+      new GeminiHandler(
+        TestData.conf.copy(virtualHosts =
+          List(TestData.conf.virtualHosts(0).copy(directoryListing = false))
+        )
+      ),
+      "gemini://localhost/dir/"
+    ) should be(a[NotFound])
+  }
+
+  behavior of "handler, user directories"
+
+  it should "return success on reading file" in {
+    handleWith(
+      handlerUserDir,
+      "gemini://localhost/~username/index.gmi"
+    ) should matchPattern {
+      case Success(_, "text/gemini", Some(_), 38L) =>
+    }
+  }
+
+  it should "return redirect accessing the user directory without ending slash" in {
+    handleWith(handlerUserDir, "gemini://localhost/~username") should be(
+      a[PermanentRedirect]
+    )
+  }
+
+  it should "return success accessing the user directory index" in {
+    handleWith(
+      handlerUserDir,
+      "gemini://localhost/~username/"
+    ) should matchPattern {
+      case Success(_, "text/gemini", Some(_), 38L) =>
+    }
+  }
+
+  it should "return bad request trying to exit the root directory" in {
+    handleWith(handlerUserDir, "gemini://localhost/~username/../../") should be(
+      a[BadRequest]
+    )
+  }
+
+  it should "return redirect to the virtual host root when leaving the user dir" in {
+    handleWith(handlerUserDir, "gemini://localhost/~username/../") should be(
+      a[PermanentRedirect]
+    )
+  }
+
+  it should "not translate root if used an invalid user pattern" in {
+    handleWith(handlerUserDir, "gemini://localhost/~username../") should be(
+      a[NotFound]
+    )
+
+    handleWith(handlerUserDir, "gemini://localhost/~0invalid/") should be(
+      a[NotFound]
+    )
+  }
+
+  it should "not translate root if no user directory path was provided" in {
+    handleWith(
+      new GeminiHandler(
+        TestData.conf.copy(virtualHosts =
+          List(
+            TestData.conf
+              .virtualHosts(0)
+              .copy(userDirectories = true)
+          )
+        )
+      ),
+      "gemini://localhost/~username/",
+      "127.0.0.1"
+    ) should be(a[NotFound])
+  }
+
+  it should "not execute a CGI if the target resource is not executable" in {
+    handleWith(
+      handlerCgi,
+      "gemini://localhost/dir/file.txt"
+    ) should matchPattern {
+      case Success(_, "text/plain", Some(_), 5L) =>
+    }
+  }
+
+  it should "not execute a CGI if the target resource is a directory" in {
+    handleWith(handlerCgi, "gemini://localhost/dir/sub/") should be(
+      a[DirListing]
+    )
+  }
+
+  it should "not apply allow CGI to subdirectories" in {
+    handleWith(
+      handlerCgi,
+      "gemini://localhost/dir/sub/cgi"
+    ) should matchPattern {
+      case Success(_, "text/plain", Some(_), 72) =>
+    }
+  }
+
+  it should "execute a CGI" in {
+    val cgi = handleWith(handlerCgi, "gemini://localhost/dir/cgi")
+      .asInstanceOf[Cgi]
+
+    cgi.status should be(20)
+    cgi.meta should be("text/gemini")
+    cgi.body should include("GATEWAY_INTERFACE=CGI/1.1")
+  }
+
+  it should "execute a CGI: empty parameters, host and port" in {
+    handleWith(
+      handlerCgi,
+      "gemini://localhost/dir/cgi"
+    ) should matchPattern {
+      case Cgi(
+            _,
+            _,
+            "",
+            "",
+            "cgi",
+            TestData.host,
+            TestData.portStr,
+            _,
+            _
+          ) =>
+    }
+  }
+
+  it should "execute a CGI: host is case-insensitive and the value in the conf is used" in {
+    handleWith(
+      handlerCgi,
+      "gemini://LOCALHOST/dir/cgi"
+    ) should matchPattern {
+      case Cgi(
+            _,
+            _,
+            "",
+            "",
+            "cgi",
+            TestData.host,
+            TestData.portStr,
+            _,
+            _
+          ) =>
+    }
+  }
+
+  it should "execute a CGI: query string" in {
+    handleWith(
+      handlerCgi,
+      "gemini://localhost/dir/cgi?query&string"
+    ) should matchPattern {
+      case Cgi(
+            _,
+            _,
+            "query&string",
+            "",
+            "cgi",
+            TestData.host,
+            TestData.portStr,
+            _,
+            _
+          ) =>
+    }
+  }
+
+  it should "execute a CGI: path info" in {
+    handleWith(
+      handlerCgi,
+      "gemini://localhost/dir/cgi/path/info"
+    ) should matchPattern {
+      case Cgi(
+            _,
+            _,
+            "",
+            "/path/info",
+            "cgi",
+            TestData.host,
+            TestData.portStr,
+            _,
+            _
+          ) =>
+    }
+  }
+
+  it should "execute a CGI: query string and path info" in {
+    handleWith(
+      handlerCgi,
+      "gemini://localhost/dir/cgi/path/info?query=string"
+    ) should matchPattern {
+      case Cgi(
+            _,
+            _,
+            "query=string",
+            "/path/info",
+            "cgi",
+            TestData.host,
+            TestData.portStr,
+            _,
+            _
+          ) =>
+    }
+  }
+
+  it should "not execute an executable if allow CGI is off" in {
+    handleWith(
+      handler,
+      "gemini://localhost/dir/cgi"
+    ) should matchPattern {
+      case Success(_, "text/plain", Some(_), _) =>
+    }
+  }
+
+  it should "response with an error if the CGI exits with non 0" in {
+    val bad =
+      handleWith(handlerCgi, "gemini://localhost/dir/bad-cgi")
+        .asInstanceOf[Cgi]
+
+    val meta = "Error executing CGI"
+    bad.status should be(50)
+    bad.meta should be(meta)
+    bad.body should include(meta)
+  }
+
+  it should "return a response with an error if the CGI exits with non 0" in {
+    val bad =
+      handleWith(handlerCgi, "gemini://localhost/dir/bad-cgi")
+        .asInstanceOf[Cgi]
+
+    val meta = "Error executing CGI"
+    bad.status should be(50)
+    bad.meta should be(meta)
+    bad.body should include(meta)
+  }
+
+  it should "return a response with an error if the CGI response is invalid" in {
+    val bad =
+      handleWith(handlerCgi, "gemini://localhost/dir/bad-response")
+        .asInstanceOf[Cgi]
+
+    val meta = "Invalid response from CGI"
+    bad.status should be(40)
+    bad.meta should be(meta)
+    bad.body should include(meta)
+  }
+
+  it should "environment variables are optional" in {
+    handleWith(
+      handlerCgi,
+      "gemini://localhost/dir/cgi/"
+    ) should matchPattern {
+      case Cgi(
+            _,
+            _,
+            _,
+            _,
+            "cgi",
+            TestData.host,
+            TestData.portStr,
+            _,
+            m
+          ) if m == Map() =>
+    }
+  }
+
+  it should "pass environment variables to the CGI" in {
+    handleWith(
+      new GeminiHandler(TestData.cgiEnvConf),
+      "gemini://localhost/dir/cgi/"
+    ) should matchPattern {
+      case Cgi(
+            _,
+            _,
+            _,
+            _,
+            "cgi",
+            TestData.host,
+            TestData.portStr,
+            _,
+            m
+          ) if m == Map("env1" -> "value") =>
+    }
+  }
+
+  it should "execute a CGI with the environment variables" in {
+    val cgi = handleWith(
+      new GeminiHandler(TestData.cgiEnvConf),
+      "gemini://localhost/dir/cgi"
+    ).asInstanceOf[Cgi]
+
+    cgi.status should be(20)
+    cgi.meta should be("text/gemini")
+    cgi.body should include("env1=value")
+  }
+
+  it should "execute a CGI when it is the index document" in {
+    val cgi = handleWith(
+      new GeminiHandler(TestData.cgiIndexConf),
+      "gemini://localhost/dir/"
+    ).asInstanceOf[Cgi]
+
+    cgi.status should be(20)
+    cgi.meta should be("text/gemini")
+    cgi.body should include("GATEWAY_INTERFACE=CGI/1.1")
+  }
+
+  it should "execute a CGI when it is the index document (full name)" in {
+    val cgi = handleWith(
+      new GeminiHandler(TestData.cgiIndexConf),
+      "gemini://localhost/dir/cgi"
+    ).asInstanceOf[Cgi]
+
+    cgi.status should be(20)
+    cgi.meta should be("text/gemini")
+    cgi.body should include("GATEWAY_INTERFACE=CGI/1.1")
+  }
+
+  it should "execute a CGI when it is the index document (full name, path info)" in {
+    val cgi = handleWith(
+      new GeminiHandler(TestData.cgiIndexConf),
+      "gemini://localhost/dir/cgi/path/info"
+    ).asInstanceOf[Cgi]
+
+    cgi.status should be(20)
+    cgi.meta should be("text/gemini")
+    cgi.body should include("GATEWAY_INTERFACE=CGI/1.1")
+    cgi.body should include("PATH_INFO=/path/info")
+  }
+
+  it should "resolve CGI directories from more to less specific" in {
+    // issue: https://gitlab.com/reidrac/spacebeans/-/issues/2
+    val cgi = handleWith(
+      new GeminiHandler(TestData.cgiPrefConf),
+      "gemini://localhost/dir/sub/cgiOk/path/info"
+    ).asInstanceOf[Cgi]
+
+    cgi.status should be(20)
+    cgi.meta should be("text/gemini")
+    cgi.body should include("GATEWAY_INTERFACE=CGI/1.1")
+    cgi.body should include("PATH_INFO=/path/info")
+  }
+}
diff --git a/server/test/src/handlers/ProtocolHandlerSpec.scala b/server/test/src/handlers/ProtocolHandlerSpec.scala
new file mode 100644
index 0000000..d35c9f2
--- /dev/null
+++ b/server/test/src/handlers/ProtocolHandlerSpec.scala
@@ -0,0 +1,100 @@
+package net.usebox.gemini.server.handlers
+
+import java.nio.file.FileSystems
+import java.net.URI
+
+import org.scalatest.flatspec.AnyFlatSpec
+import org.scalatest.matchers.should.Matchers
+
+import net.usebox.gemini.server.{ServiceConf, Response}
+import net.usebox.gemini.server.TestData
+
+class ProtocolHandlerSpec extends AnyFlatSpec with Matchers {
+
+  def getPath(value: String) = FileSystems.getDefault().getPath(value)
+
+  class TestHandler(conf: ServiceConf) extends ProtocolHandler(conf) {
+    def handle(req: String, uri: URI, remoteAddr: String): Response = ???
+  }
+
+  val handler = new TestHandler(TestData.conf)
+
+  behavior of "guessMimeType using the internal resolver"
+
+  it should "resolve a known MIME type" in {
+    handler
+      .guessMimeType(
+        getPath("file.html"),
+        None
+      ) shouldBe "text/html"
+  }
+
+  it should "resolve de default MIME type for unknown types" in {
+    handler
+      .guessMimeType(
+        getPath("unknow"),
+        None
+      ) shouldBe TestData.conf.defaultMimeType
+  }
+
+  it should "resolve gemini MIME type" in {
+    handler
+      .guessMimeType(
+        getPath("file.gmi"),
+        None
+      ) shouldBe "text/gemini"
+    handler
+      .guessMimeType(
+        getPath("file.gemini"),
+        None
+      ) shouldBe "text/gemini"
+  }
+
+  it should "resolve gemini MIME type, including parameters" in {
+    handler
+      .guessMimeType(
+        getPath("file.gmi"),
+        Some("param")
+      ) shouldBe "text/gemini; param"
+    handler
+      .guessMimeType(
+        getPath("file.gemini"),
+        Some("param")
+      ) shouldBe "text/gemini; param"
+  }
+
+  it should "gemini MIME type parameters are sanitized" in {
+    handler
+      .guessMimeType(
+        getPath("file.gmi"),
+        Some("     ; param")
+      ) shouldBe "text/gemini; param"
+  }
+
+  behavior of "guessMimeType using the configured types"
+
+  it should "resolve a known MIME type" in {
+    new TestHandler(TestData.conf.copy(mimeTypes = TestData.mimeTypes))
+      .guessMimeType(
+        getPath("file.gmi"),
+        None
+      ) shouldBe "config"
+  }
+
+  it should "include parameters for text/gemini MIME types" in {
+    new TestHandler(
+      TestData.conf.copy(mimeTypes = Some(Map("text/gemini" -> List(".gmi"))))
+    ).guessMimeType(
+      getPath("file.gmi"),
+      Some("param")
+    ) shouldBe "text/gemini; param"
+  }
+
+  it should "resolve de default MIME type for unknown types" in {
+    new TestHandler(TestData.conf.copy(mimeTypes = TestData.mimeTypes))
+      .guessMimeType(
+        getPath("unknow"),
+        None
+      ) shouldBe TestData.conf.defaultMimeType
+  }
+}