diff options
| -rw-r--r-- | garcon.py | 50 | ||||
| -rw-r--r-- | helld.py | 93 |
2 files changed, 129 insertions, 14 deletions
@@ -1,22 +1,31 @@ -# assumes utf8 -# began writing this on 2024-02-20 -# found mozz-archiver on 2024-02-21 - +from sys import stdin, stdout, stderr import datetime +import hashlib import socket import ssl -import uuid -import hashlib import urllib.parse -from sys import argv, stdout +import uuid +import traceback # based on: # https://tildegit.org/solderpunk/gemini-demo-1/src/branch/master/gemini-demo.py # https://tildegit.org/solderpunk/AV-98/src/branch/master/src/av98/client.py # TODO ciphers etc +SIZE_LIMIT = 4 * 1024 * 1024 # 4MB seems reasonable + outf = stdout.buffer +# directly stolen from gemini-demo +def absolutise_url(base, relative): + # Absolutise relative links + if "://" not in relative: + # Python's URL tools somehow only work with known schemes? + base = base.replace("gemini://","http://") + relative = urllib.parse.urljoin(base, relative) + relative = relative.replace("http://", "gemini://") + return relative + def header(k, v): v = str(v) assert '\n' not in v @@ -52,7 +61,9 @@ def request_raw(host, port, url): cert = s.getpeercert(True) fp = s.makefile("rb") - payload = fp.read() + payload = fp.read(SIZE_LIMIT) + truncated = fp.read() != b'' + print(truncated) fp.close() s.close() @@ -70,12 +81,14 @@ def request_raw(host, port, url): header("WARC-IP-Address", peername[0]) header("WARC-Target-URI", url) header("Content-Type", "application/gemini; msgtype=response") # as in mozz-archiver + if trunacted: + header("WARC-Truncated", "length") # my extensions header("X-Server-Fingerprint", 'sha256:' + hashlib.sha256(cert).hexdigest()) outf.write(b'\r\n') - outf.write(payload) +#outf.write(payload) outf.write(b'\r\n\r\n') # TODO check for close_notify @@ -87,14 +100,23 @@ def request_url(url): return request_raw(p.hostname, p.port or 1965, url) def request_url_loop(url): - while True: + # i only allow 3 redirects, so detecting loops isn't really necessary + for _ in range(3): res = request_url(url) header = res.split(b'\r\n')[0] if 2 + 1 + 1024 < len(header): break - if header[0] == ord('3'): - url = header.split(b' ', 2)[1].decode('utf-8') + if len(header) > 0 and header[0] == ord('3'): + newurl = header.split(b' ', 2)[1].decode('utf-8') + url = absolutise_url(url, newurl) else: break -warcinfo() -request_url_loop(argv[1]) +if __name__ == '__main__': + warcinfo() + outf.flush() + for line in stdin: + try: + request_url_loop(line.rstrip('\r\n').rstrip('\n')) + outf.flush() + except: + print(traceback.format_exc(), file=stderr) diff --git a/helld.py b/helld.py new file mode 100644 index 0000000..68d24ac --- /dev/null +++ b/helld.py @@ -0,0 +1,93 @@ +# Hellish Gemini daemon, in both behaviour and code quality. +# +# openssl req -newkey rsa:4096 -nodes -keyout key.pem -nodes -x509 -out cert.pem +# python3 helld.py cert.pem key.pem + +import socket +import threading +import socketserver +import ssl +import sys +import time +import random +from urllib.parse import urlparse + +handlers = [] + +def register(addr): + def dec(fn): + handlers.append((addr, fn)) + return dec + +@register('/normal') +def normal(s, _): + s.sendall(b'20 text/gemini\r\n# hello\nnothing strange going on here\n') + +@register('/noclose') +def noclose(s, rh): + s.sendall(b'20 text/gemini\r\nquick before the connection is killed i need to tell you tha') + close(rh.request) + +@register('/sleep1') +def sleep1(s, rh): + s.sendall(b'20 text/gemini\r\nabout to take a nap') + time.sleep(120) + +@register('/sleep2') +def sleep(s, rh): + time.sleep(120) + s.sendall(b'20 text/gemini\r\ntook a nap') + +@register('/128mb') +def lotsofdata(s, rh): + s.sendall(b'20 text/gemini\r\n') + for n in range(128): + s.sendall(b'lol ' * (1024 * 1024 // 4)) + +@register('/4mb') +def sensibleamountofdata(s, rh): + buffer = b'20 text/gemini\r\n' + b'lol ' * (1024 * 1024) + buffer = buffer[:4*1024*1024] + s.sendall(buffer) + +@register('/over4mb') +def nonsensibleamountofdata(s, rh): + buffer = b'20 text/gemini\r\n' + b'lol ' * (1024 * 1024) + buffer = buffer[:4*1024*1024+1] + s.sendall(buffer) + +@register('/loop') +def loop(s, rh): + s.sendall(f'30 /loop/{random.randint(0, 2**32)}\r\n'.encode('utf-8')) + +@register('/index') +def index(s, rh): + s.sendall(b'20 text/gemini\r\n') + for addr, _ in handlers: + s.sendall(f'=> {addr}\n'.encode('utf-8')) + +@register('') +def toindex(s, rh): + s.sendall(b'30 /index\r\n') + +class ThreadedTCPRequestHandler(socketserver.BaseRequestHandler): + def handle(self): + s = context.wrap_socket(self.request, server_side=True) + req = s.read().decode('utf-8').rstrip('\r\n') + print(req) + req = urlparse(req) + for addr, fn in handlers: + if req.path.startswith(addr): + fn(s, self) + break + +class ThreadedTCPServer(socketserver.ThreadingMixIn, socketserver.TCPServer): + pass + +if __name__ == "__main__": + context = ssl.SSLContext(ssl.PROTOCOL_TLS_SERVER) + context.load_cert_chain(sys.argv[1], sys.argv[2]) + + ThreadedTCPServer.allow_reuse_address = True + server = ThreadedTCPServer(('0.0.0.0', 1965), ThreadedTCPRequestHandler) + server.serve_forever() |