Add gemini protocol support

The gemini protocol's "speculative specification" is documented  at the
following URLs:

  https://gemini.circumlunar.space/docs/spec-spec.txt
  gopher://gemini.circumlunar.space/1/docs/spec-spec.txt
  gemini://gemini.circumlunar.space/docs/spec-spec.txt

This patch introduces preliminary support for version 0.11.0 of the
specification, as of March 1st 2020.

7 files changed, 227 insertions, 1 deletions

diff --git a/lib/Makefile.inc b/lib/Makefile.inc
index 08b0f73a3..e4789b5b5 100644
--- a/lib/Makefile.inc
+++ b/lib/Makefile.inc
@@ -51,6 +51,7 @@ LIB_CFILES = altsvc.c amigaos.c asyn-ares.c asyn-thread.c base64.c            \
   curl_ntlm_core.c curl_ntlm_wb.c curl_path.c curl_range.c curl_rtmp.c        \
   curl_sasl.c curl_sspi.c curl_threads.c dict.c dotdot.c easy.c escape.c      \
   file.c fileinfo.c formdata.c ftp.c url.c ftplistparser.c getenv.c getinfo.c \
+  gemini.c \
   gopher.c hash.c hmac.c hostasyn.c hostcheck.c hostip.c hostip4.c hostip6.c  \
   hostsyn.c http.c http2.c http_chunks.c http_digest.c http_negotiate.c       \
   http_ntlm.c http_proxy.c idn_win32.c if2ip.c imap.c inet_ntop.c inet_pton.c \
@@ -70,6 +71,7 @@ LIB_HFILES = altsvc.h amigaos.h arpa_telnet.h asyn.h conncache.h connect.h    \
   curl_printf.h curl_range.h curl_rtmp.h curl_sasl.h curl_sec.h curl_setup.h  \
   curl_setup_once.h curl_sha256.h curl_sspi.h curl_threads.h curlx.h dict.h   \
   dotdot.h easyif.h escape.h file.h fileinfo.h formdata.h ftp.h url.h         \
+  gemini.h \
   ftplistparser.h getinfo.h gopher.h hash.h hostcheck.h hostip.h http.h       \
   http2.h http_chunks.h http_digest.h http_negotiate.h http_ntlm.h            \
   http_proxy.h if2ip.h imap.h inet_ntop.h inet_pton.h llist.h memdebug.h      \
diff --git a/lib/curl_setup.h b/lib/curl_setup.h
index a86a235b3..a102e0519 100644
--- a/lib/curl_setup.h
+++ b/lib/curl_setup.h
@@ -196,6 +196,9 @@
 #  ifndef CURL_DISABLE_GOPHER
 #    define CURL_DISABLE_GOPHER
 #  endif
+#  ifndef CURL_DISABLE_GEMINI
+#    define CURL_DISABLE_GEMINI
+#  endif
 #  ifndef CURL_DISABLE_SMB
 #    define CURL_DISABLE_SMB
 #  endif
diff --git a/lib/gemini.c b/lib/gemini.c
new file mode 100644
index 000000000..bd1953d82
--- /dev/null
+++ b/lib/gemini.c
@@ -0,0 +1,179 @@
+/***************************************************************************
+ *                                  _   _ ____  _
+ *  Project                     ___| | | |  _ \| |
+ *                             / __| | | | |_) | |
+ *                            | (__| |_| |  _ <| |___
+ *                             \___|\___/|_| \_\_____|
+ *
+ * Copyright (C) 1998 - 2020, Daniel Stenberg, <daniel@haxx.se>, et al.
+ *
+ * This software is licensed as described in the file COPYING, which
+ * you should have received as part of this distribution. The terms
+ * are also available at https://curl.haxx.se/docs/copyright.html.
+ *
+ * You may opt to use, copy, modify, merge, publish, distribute and/or sell
+ * copies of the Software, and permit persons to whom the Software is
+ * furnished to do so, under the terms of the COPYING file.
+ *
+ * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
+ * KIND, either express or implied.
+ *
+ ***************************************************************************/
+
+#include "curl_setup.h"
+
+#ifndef CURL_DISABLE_GEMINI
+
+#include "urldata.h"
+#include <curl/curl.h>
+#include "transfer.h"
+#include "sendf.h"
+#include "connect.h"
+#include "progress.h"
+#include "gemini.h"
+#include "select.h"
+#include "strdup.h"
+#include "url.h"
+#include "escape.h"
+#include "warnless.h"
+#include "curl_printf.h"
+#include "curl_memory.h"
+#include "vtls/vtls.h"
+/* The last #include file should be: */
+#include "memdebug.h"
+
+/*
+ * Forward declarations.
+ */
+
+static CURLcode gemini_do(struct connectdata *conn, bool *done);
+static CURLcode gemini_connecting(struct connectdata *conn, bool *done);
+CURLcode Curl_gemini_connect(struct connectdata *conn, bool *done);
+
+/*
+ * Gemini protocol handler.
+ */
+
+const struct Curl_handler Curl_handler_gemini = {
+  "GEMINI",                             /* scheme */
+  ZERO_NULL,                            /* setup_connection */
+  gemini_do,                            /* do_it */
+  ZERO_NULL,                            /* done */
+  ZERO_NULL,                            /* do_more */
+  Curl_gemini_connect,                  /* connect_it */
+  gemini_connecting,                    /* connecting */
+  ZERO_NULL,                            /* doing */
+  ZERO_NULL,                            /* proto_getsock */
+  ZERO_NULL,                            /* doing_getsock */
+  ZERO_NULL,                            /* domore_getsock */
+  ZERO_NULL,                            /* perform_getsock */
+  ZERO_NULL,                            /* disconnect */
+  ZERO_NULL,                            /* readwrite */
+  ZERO_NULL,                            /* connection_check */
+  PORT_GEMINI,                          /* defport */
+  CURLPROTO_GEMINI,                     /* protocol */
+  PROTOPT_NONE                          /* flags */
+};
+
+static CURLcode gemini_do(struct connectdata *conn, bool *done)
+{
+  CURLcode result = CURLE_OK;
+  struct Curl_easy *data = conn->data;
+  curl_socket_t sockfd = conn->sock[FIRSTSOCKET];
+  timediff_t timeout_ms;
+  ssize_t amount, k;
+  size_t len;
+  int what;
+  char *req = NULL;
+  char *u = NULL;
+
+  *done = TRUE; /* unconditionally */
+
+  /* format the request URL */
+  if (curl_url_get(data->state.uh, CURLUPART_URL, &u, 0) != CURLUE_OK)
+    return CURLE_URL_MALFORMAT;
+
+  /* build the request line */
+  req = aprintf("%s\r\n", u);
+  if (!req)
+    return CURLE_OUT_OF_MEMORY;
+
+  len = strlen(req);
+
+  /* We use Curl_write instead of Curl_sendf to make sure the entire buffer is
+     sent, which could be sizeable with long selectors. */
+  k = curlx_uztosz(len);
+
+  for(;;) {
+    result = Curl_write(conn, sockfd, req, k, &amount);
+    if(!result) { /* Which may not have written it all! */
+      result = Curl_client_write(conn, CLIENTWRITE_HEADER, req, amount);
+      if(result)
+        break;
+
+      k -= amount;
+      req += amount;
+      if(k < 1)
+        break; /* but it did write it all */
+    }
+    else
+      break;
+
+    timeout_ms = Curl_timeleft(conn->data, NULL, FALSE);
+    if(timeout_ms < 0) {
+      result = CURLE_OPERATION_TIMEDOUT;
+      break;
+    }
+    if(!timeout_ms)
+      timeout_ms = TIMEDIFF_T_MAX;
+
+    /* Don't busyloop. The entire loop thing is a work-around as it causes a
+       BLOCKING behavior which is a NO-NO. This function should rather be
+       split up in a do and a doing piece where the pieces that aren't
+       possible to send now will be sent in the doing function repeatedly
+       until the entire request is sent.
+    */
+    what = SOCKET_WRITABLE(sockfd, timeout_ms);
+    if(what < 0) {
+      result = CURLE_SEND_ERROR;
+      break;
+    }
+    else if(!what) {
+      result = CURLE_OPERATION_TIMEDOUT;
+      break;
+    }
+  }
+
+  Curl_setup_transfer(data, FIRSTSOCKET, -1, FALSE, -1);
+  return CURLE_OK;
+}
+
+static CURLcode gemini_connecting(struct connectdata *conn, bool *done)
+{
+  CURLcode result;
+  DEBUGASSERT((conn) && (conn->handler->flags & PROTOPT_SSL));
+
+  /* perform SSL initialization for this socket */
+  result = Curl_ssl_connect_nonblocking(conn, FIRSTSOCKET, done);
+  if(result)
+    connclose(conn, "Failed TLS connection");
+
+  return result;
+}
+
+/*
+ * Curl_http_connect() performs HTTP stuff to do at connect-time, called from
+ * the generic Curl_connect().
+ */
+CURLcode Curl_gemini_connect(struct connectdata *conn, bool *done)
+{
+  CURLcode result;
+
+  /* perform SSL initialization */
+  result = gemini_connecting(conn, done);
+  if(result)
+    return result;
+
+  return CURLE_OK;
+}
+#endif /*CURL_DISABLE_GEMINI*/
diff --git a/lib/gemini.h b/lib/gemini.h
new file mode 100644
index 000000000..54cc6afd0
--- /dev/null
+++ b/lib/gemini.h
@@ -0,0 +1,29 @@
+#ifndef HEADER_CURL_GEMINI_H
+#define HEADER_CURL_GEMINI_H
+/***************************************************************************
+ *                                  _   _ ____  _
+ *  Project                     ___| | | |  _ \| |
+ *                             / __| | | | |_) | |
+ *                            | (__| |_| |  _ <| |___
+ *                             \___|\___/|_| \_\_____|
+ *
+ * Copyright (C) 1998 - 2019, Daniel Stenberg, <daniel@haxx.se>, et al.
+ *
+ * This software is licensed as described in the file COPYING, which
+ * you should have received as part of this distribution. The terms
+ * are also available at https://curl.haxx.se/docs/copyright.html.
+ *
+ * You may opt to use, copy, modify, merge, publish, distribute and/or sell
+ * copies of the Software, and permit persons to whom the Software is
+ * furnished to do so, under the terms of the COPYING file.
+ *
+ * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
+ * KIND, either express or implied.
+ *
+ ***************************************************************************/
+
+#ifndef CURL_DISABLE_GEMINI
+extern const struct Curl_handler Curl_handler_gemini;
+#endif
+
+#endif /* HEADER_CURL_GEMINI_H */
diff --git a/lib/url.c b/lib/url.c
index 0173dc88a..e2b63cf72 100644
--- a/lib/url.c
+++ b/lib/url.c
@@ -113,6 +113,7 @@ bool curl_win32_idn_to_ascii(const char *in, char **out);
 #include "inet_ntop.h"
 #include "http_ntlm.h"
 #include "curl_rtmp.h"
+#include "gemini.h"
 #include "gopher.h"
 #include "mqtt.h"
 #include "http_proxy.h"
@@ -145,7 +146,7 @@ static unsigned int get_protocol_family(unsigned int protocol);
  * Protocol table. Schemes (roughly) in 2019 popularity order:
  *
  * HTTPS, HTTP, FTP, FTPS, SFTP, FILE, SCP, SMTP, LDAP, IMAPS, TELNET, IMAP,
- * LDAPS, SMTPS, TFTP, SMB, POP3, GOPHER POP3S, RTSP, RTMP, SMBS, DICT
+ * LDAPS, SMTPS, TFTP, SMB, POP3, GOPHER POP3S, RTSP, RTMP, SMBS, DICT, GEMINI
  */
 static const struct Curl_handler * const protocols[] = {
 
@@ -249,6 +250,10 @@ static const struct Curl_handler * const protocols[] = {
   &Curl_handler_dict,
 #endif
 
+#ifndef CURL_DISABLE_GEMINI
+  &Curl_handler_gemini,
+#endif
+
   (struct Curl_handler *) NULL
 };
 
@@ -4093,6 +4098,10 @@ static unsigned int get_protocol_family(unsigned int protocol)
     family = CURLPROTO_GOPHER;
     break;
 
+  case CURLPROTO_GEMINI:
+    family = CURLPROTO_GEMINI;
+    break;
+
   case CURLPROTO_SMB:
   case CURLPROTO_SMBS:
     family = CURLPROTO_SMB;
diff --git a/lib/urldata.h b/lib/urldata.h
index 9b4ce5f5b..46321c5fd 100644
--- a/lib/urldata.h
+++ b/lib/urldata.h
@@ -49,6 +49,7 @@
 #define PORT_RTMPT PORT_HTTP
 #define PORT_RTMPS PORT_HTTPS
 #define PORT_GOPHER 70
+#define PORT_GEMINI 1965
 #define PORT_MQTT 1883
 
 #define DICT_MATCH "/MATCH:"
diff --git a/lib/version.c b/lib/version.c
index 14e509606..87c69a187 100644
--- a/lib/version.c
+++ b/lib/version.c
@@ -250,6 +250,9 @@ static const char * const protocols[] = {
 #if defined(USE_SSL) && !defined(CURL_DISABLE_FTP)
   "ftps",
 #endif
+#ifndef CURL_DISABLE_GEMINI
+  "gemini",
+#endif
 #ifndef CURL_DISABLE_GOPHER
   "gopher",
 #endif