From bb9c5735dbb554a92f24b1fb8b941d87e8e788b5 Mon Sep 17 00:00:00 2001

From: =?UTF-8?q?Jaakko=20Kera=CC=88nen?= jaakko.keranen@iki.fi

Date: Sun, 11 Dec 2022 11:01:42 +0200

Subject: [PATCH 1/1] Report Unicode decode error to client

README.md | 1 +

gmcapsule/gemini.py | 36 ++++++++++++++++++++----------------

2 files changed, 21 insertions(+), 16 deletions(-)

diff --git a/README.md b/README.md

index a1af4b9..80b3f0a 100644

--- a/README.md

+++ b/README.md

@@ -48,6 +48,7 @@ The log can be viewed via journalctl (or syslog):

v0.2.3

• Requests exceeding 1024 bytes should result in an error code and not just be ignored.

+* Respond with an error code to malformed UTF-8 in the request.

v0.2.2

• Reduced required Python version to 3.6 (f-strings).

diff --git a/gmcapsule/gemini.py b/gmcapsule/gemini.py

index b086ec6..34ac7f6 100644

--- a/gmcapsule/gemini.py

+++ b/gmcapsule/gemini.py

@@ -376,22 +376,26 @@ class Worker(threading.Thread):

     req_mime = None

     incoming = safe_recv(stream, MAX_RECV)

•    while len(data) < MAX_RECV:
  

•        data += incoming
  

•        crlf_pos = data.find(b'\r\n')
  

•        if crlf_pos >= 0:
  

•            request = data[:crlf_pos].decode('utf-8')
  

•            data = data[crlf_pos + 2:]
  

•            break
  

•        elif len(data) > MAX_LEN:
  

•            # At this point we should have received the line terminator.
  

•            self.log(from_addr, 'sent a malformed request')
  

•            report_error(stream, 59, "Request exceeds maximum length")
  

•            return
  

•        incoming = safe_recv(stream, MAX_RECV - len(data))
  

•        if len(incoming) <= 0:
  

•            break
  

•    try:
  

•        while len(data) < MAX_RECV:
  

•            data += incoming
  

•            crlf_pos = data.find(b'\r\n')
  

•            if crlf_pos >= 0:
  

•                request = data[:crlf_pos].decode('utf-8')
  

•                data = data[crlf_pos + 2:]
  

•                break
  

•            elif len(data) > MAX_LEN:
  

•                # At this point we should have received the line terminator.
  

•                self.log(from_addr, 'sent a malformed request')
  

•                report_error(stream, 59, "Request exceeds maximum length")
  

•                return
  

•            incoming = safe_recv(stream, MAX_RECV - len(data))
  

•            if len(incoming) <= 0:
  

•                break
  

•    except UnicodeDecodeError:
  

•        report_error(stream, 59, "Request contains malformed UTF-8")
  

•        return
  

     if not request or len(data) > MAX_RECV:

         report_error(stream, 59, "Invalid request")

--

2.25.1