path: root/test/pyhttpd/nghttp.py

import re
import os
import subprocess
from datetime import datetime
from typing import Dict

from urllib.parse import urlparse

from .result import ExecResult


def _get_path(x):
    return x["path"]
    

class Nghttp:

    def __init__(self, path, connect_addr=None, tmp_dir="/tmp",
                 test_name: str = None):
        self.NGHTTP = path
        self.CONNECT_ADDR = connect_addr
        self.TMP_DIR = tmp_dir
        self._test_name = test_name

    @staticmethod
    def get_stream(streams, sid):
        sid = int(sid)
        if sid not in streams:
            streams[sid] = {
                    "id": sid,
                    "header": {},
                    "request": {
                        "id": sid,
                        "body": b''
                    },
                    "response": {
                        "id": sid,
                        "body": b''
                    },
                    "data_lengths": [],
                    "paddings": [],
                    "promises": []
            }
        return streams[sid] if sid in streams else None

    def run(self, urls, timeout, options):
        return self._baserun(urls, timeout, options)

    def complete_args(self, url, _timeout, options: [str]) -> [str]:
        if not isinstance(url, list):
            url = [url]
        u = urlparse(url[0])
        args = [self.NGHTTP]
        if self.CONNECT_ADDR:
            connect_host = self.CONNECT_ADDR
            args.append("--header=host: %s:%s" % (u.hostname, u.port))
        else:
            connect_host = u.hostname
        if options:
            args.extend(options)
        for xurl in url:
            xu = urlparse(xurl)
            nurl = "%s://%s:%s/%s" % (u.scheme, connect_host, xu.port, xu.path)
            if xu.query:
                nurl = "%s?%s" % (nurl, xu.query)
            args.append(nurl)
        return args

    def _baserun(self, url, timeout, options):
        return self._run(self.complete_args(url, timeout, options))
    
    def parse_output(self, btext) -> Dict:
        # getting meta data and response body out of nghttp's output
        # is a bit tricky. Without '-v' we just get the body. With '-v' meta
        # data and timings in both directions are listed. 
        # We rely on response :status: to be unique and on 
        # response body not starting with space.
        # Something not good enough for general purpose, but for these tests.
        output = {}
        body = ''
        streams = {}
        skip_indents = True
        # chunk output into lines. nghttp mixes text
        # meta output with bytes from the response body.
        lines = [l.decode() for l in btext.split(b'\n')]

        for lidx, l in enumerate(lines):
            if len(l) == 0:
                body += '\n'
                continue
            m = re.match(r'(.*)\[.*] recv \(stream_id=(\d+)\) (\S+): (\S*)', l)
            if m:
                body += m.group(1)
                s = self.get_stream(streams, m.group(2))
                hname = m.group(3)
                hval = m.group(4)
                print("stream %d header %s: %s" % (s["id"], hname, hval))
                header = s["header"]
                if hname in header: 
                    header[hname] += ", %s" % hval
                else:
                    header[hname] = hval
                continue

            m = re.match(r'(.*)\[.*] recv HEADERS frame <.* stream_id=(\d+)>', l)
            if m:
                body += m.group(1)
                s = self.get_stream(streams, m.group(2))
                if s:
                    print("stream %d: recv %d header" % (s["id"], len(s["header"])))
                    response = s["response"]
                    hkey = "header"
                    if "header" in response:
                        h = response["header"]
                        if ":status" in h and int(h[":status"]) >= 200:
                            hkey = "trailer"
                        else:
                            prev = {
                                "header": h
                            }
                            if "previous" in response:
                                prev["previous"] = response["previous"]
                            response["previous"] = prev
                    response[hkey] = s["header"]
                    s["header"] = {}
                    body = ''
                continue
            
            m = re.match(r'(.*)\[.*] recv DATA frame <length=(\d+), .*stream_id=(\d+)>', l)
            if m:
                body += m.group(1)
                s = self.get_stream(streams, m.group(3))
                blen = int(m.group(2))
                if s:
                    print(f'stream {s["id"]}: {blen} DATA bytes added via "{l}"')
                    padlen = 0
                    if len(lines) > lidx + 2:
                        mpad = re.match(r' +\(padlen=(\d+)\)', lines[lidx+2])
                        if mpad: 
                            padlen = int(mpad.group(1))
                    s["data_lengths"].append(blen)
                    s["paddings"].append(padlen)
                    blen -= padlen
                    s["response"]["body"] += body[-blen:].encode()
                body = ''
                skip_indents = True
                continue
                
            m = re.match(r'(.*)\[.*] recv PUSH_PROMISE frame <.* stream_id=(\d+)>', l)
            if m:
                body += m.group(1)
                s = self.get_stream(streams, m.group(2))
                if s:
                    # headers we have are request headers for the PUSHed stream
                    # these have been received on the originating stream, the promised
                    # stream id it mentioned in the following lines
                    print("stream %d: %d PUSH_PROMISE header" % (s["id"], len(s["header"])))
                    if len(lines) > lidx+2:
                        m2 = re.match(r'\s+\(.*promised_stream_id=(\d+)\)', lines[lidx+2])
                        if m2:
                            s2 = self.get_stream(streams, m2.group(1))
                            s2["request"]["header"] = s["header"]
                            s["promises"].append(s2)
                    s["header"] = {} 
                continue
                    
            m = re.match(r'(.*)\[.*] recv (\S+) frame <length=(\d+), .*stream_id=(\d+)>', l)
            if m:
                print("recv frame %s on stream %s" % (m.group(2), m.group(4)))
                body += m.group(1)
                skip_indents = True
                continue
                
            m = re.match(r'(.*)\[.*] send (\S+) frame <length=(\d+), .*stream_id=(\d+)>', l)
            if m:
                print("send frame %s on stream %s" % (m.group(2), m.group(4)))
                body += m.group(1)
                skip_indents = True
                continue
                
            if skip_indents and l.startswith('      '):
                continue
            
            if '[' != l[0]:
                skip_indents = None
                body += l + '\n'

        # the main request is done on the lowest odd numbered id
        main_stream = 99999999999
        for sid in streams:
            s = streams[sid]
            if "header" in s["response"] and ":status" in s["response"]["header"]:
                s["response"]["status"] = int(s["response"]["header"][":status"])
            if (sid % 2) == 1 and sid < main_stream:
                main_stream = sid
        
        output["streams"] = streams
        if main_stream in streams:
            output["response"] = streams[main_stream]["response"]
            output["paddings"] = streams[main_stream]["paddings"]
            output["data_lengths"] = streams[main_stream]["data_lengths"]
        return output

    def _raw(self, url, timeout, options):
        args = ["-v"]
        if self._test_name is not None:
            args.append(f'--header=AP-Test-Name: {self._test_name}')
        if options:
            args.extend(options)
        r = self._baserun(url, timeout, args)
        if 0 == r.exit_code:
            r.add_results(self.parse_output(r.outraw))
        return r

    def get(self, url, timeout=5, options=None):
        return self._raw(url, timeout, options)

    def assets(self, url, timeout=5, options=None):
        if not options:
            options = []
        options.extend(["-ans"])
        r = self._baserun(url, timeout, options)
        assets = []
        if 0 == r.exit_code:
            lines = re.findall(r'[^\n]*\n', r.stdout, re.MULTILINE)
            for lidx, l in enumerate(lines):
                m = re.match(r'\s*(\S+)\s+(\S+)\s+(\S+)\s+(\S+)\s+(\S+)\s+(\S+)\s+/(.*)', l)
                if m:
                    assets.append({
                        "path": m.group(7),
                        "status": int(m.group(5)),
                        "size": m.group(6)
                    })
        assets.sort(key=_get_path)
        r.add_assets(assets)
        return r

    def post_data(self, url, data, timeout=5, options=None):
        reqbody = ("%s/nghttp.req.body" % self.TMP_DIR)
        with open(reqbody, 'wb') as f:
            f.write(data.encode('utf-8'))
        if not options:
            options = []
        options.extend(["--data=%s" % reqbody])
        return self._raw(url, timeout, options)

    def post_name(self, url, name, timeout=5, options=None):
        reqbody = ("%s/nghttp.req.body" % self.TMP_DIR)
        with open(reqbody, 'w') as f:
            f.write("--DSAJKcd9876\r\n")
            f.write("Content-Disposition: form-data; name=\"value\"; filename=\"xxxxx\"\r\n")
            f.write("Content-Type: text/plain\r\n")
            f.write(f"\r\n{name}")
            f.write("\r\n--DSAJKcd9876\r\n")
        if not options:
            options = []
        options.extend([ 
            "--data=%s" % reqbody, 
            "-HContent-Type: multipart/form-data; boundary=DSAJKcd9876"])
        return self._raw(url, timeout, options)

    def upload(self, url, fpath, timeout=5, options=None):
        if not options:
            options = []
        options.extend(["--data=%s" % fpath])
        return self._raw(url, timeout, options)

    def upload_file(self, url, fpath, timeout=5, options=None):
        fname = os.path.basename(fpath)
        reqbody = ("%s/nghttp.req.body" % self.TMP_DIR)
        with open(fpath, 'rb') as fin:
            with open(reqbody, 'wb') as f:
                preamble = [
                    '--DSAJKcd9876',
                    'Content-Disposition: form-data; name="xxx"; filename="xxxxx"',
                    'Content-Type: text/plain',
                    '',
                    'testing mod_h2',
                    '\r\n--DSAJKcd9876',
                    f'Content-Disposition: form-data; name="file"; filename="{fname}"',
                    'Content-Type: application/octet-stream',
                    'Content-Transfer-Encoding: binary',
                    '', ''
                ]
                f.write('\r\n'.join(preamble).encode('utf-8'))
                f.write(fin.read())
                f.write('\r\n'.join([
                    '\r\n--DSAJKcd9876', ''
                ]).encode('utf-8'))
        if not options:
            options = []
        options.extend([ 
            "--data=%s" % reqbody, 
            "--expect-continue", 
            "-HContent-Type: multipart/form-data; boundary=DSAJKcd9876"])
        return self._raw(url, timeout, options)

    def _run(self, args) -> ExecResult:
        print(("execute: %s" % " ".join(args)))
        start = datetime.now()
        p = subprocess.run(args, capture_output=True, text=False)
        return ExecResult(args=args, exit_code=p.returncode,
                          stdout=p.stdout, stderr=p.stderr,
                          duration=datetime.now() - start)