writeonce/docs/examples/writeonce-framework/internal/parse.wo

-- internal/parse.wo — HTTP/1.1 request parsing over a net connection.
--
-- INTERNAL (iteration 17): a consumer cannot `use` this module — the
-- `internal/` segment makes that WO-E108. The connection-level parser, the
-- carry-state record, and the %XX/query decoders are the framework's own
-- business; `media_type`/`form_values` are the public half and live in
-- `http/form.wo`.
--
-- Bounded reads only (`net.read`), so requests are buffered to the header
-- terminator, then the body to exactly Content-Length. Keep-alive means
-- bytes past this request belong to the NEXT one: the caller passes the
-- carry-over in and gets the new remainder back in Parsed.rest.
--
-- Parsed is a three-state answer (no tuples in the language):
--   closed=true            peer ended the connection cleanly between requests
--   ok=false               malformed request — answer 400 and close
--   ok=true, req non-nil   one complete request
use net
use http -- Req lives in the public module now

const BODY_MAX = 1048576

pub typedef Parsed = {
  closed: Bool,
  ok:     Bool,
  ?req:   Req,
  rest:   Text
}

-- %XX decoding, '+' as space when plus_space (query strings only).
-- Malformed escapes pass through verbatim — parsing stays total.
fn hex_val(b: Int) -> Int {
  if b >= 48 and b <= 57 { return b - 48; }        -- 0-9
  if b >= 97 and b <= 102 { return b - 87; }       -- a-f
  if b >= 65 and b <= 70 { return b - 55; }        -- A-F
  return -1;
}

pub fn url_decode(t: Text, plus_space: Bool) -> Text {
  let out = "";
  let i = 0;
  let n = len(t);
  while i < n {
    let b = byte_at(t, i);
    if b == 37 and i + 2 < n {                     -- '%'
      let hi = hex_val(byte_at(t, i + 1));
      let lo = hex_val(byte_at(t, i + 2));
      if hi >= 0 and lo >= 0 {
        out = out .. char_of(hi * 16 + lo);
        i = i + 3;
        continue;
      }
    }
    if plus_space and b == 43 {                    -- '+'
      out = out .. " ";
      i = i + 1;
      continue;
    }
    out = out .. substr(t, i, 1);
    i = i + 1;
  }
  return out;
}

-- "a=1&b=hello+world" -> decoded pairs; a bare key maps to ""
pub fn parse_query(qs: Text) -> map<Text, Text> {
  let q: map<Text, Text> = {};
  if qs == "" { return q; }
  for pair in split(qs, "&") {
    if pair == "" { continue; }
    let eq = index_of(pair, "=");
    if eq < 0 {
      q[url_decode(pair, true)] = "";
    } else {
      q[url_decode(substr(pair, 0, eq), true)] = url_decode(substr(pair, eq + 1, len(pair) - eq - 1), true);
    }
  }
  return q;
}

fn malformed(rest: Text) -> Parsed {
  return Parsed { closed: false, ok: false, req: nil, rest: rest };
}

-- One request off the connection. `carry` = leftover bytes from the same
-- connection's previous request (keep-alive).
pub fn parse_request(c: net.Conn, carry: Text) -> Parsed {
  let buf = carry;
  let header_end = index_of(buf, "\r\n\r\n");
  while header_end == -1 {
    let got = net.read(c, 8192);
    if len(got) == 0 {
      -- peer closed: clean between requests (empty buffer), torn otherwise
      if trim(buf) == "" { return Parsed { closed: true, ok: true, req: nil, rest: "" }; }
      return malformed("");
    }
    buf = buf .. got;
    header_end = index_of(buf, "\r\n\r\n");
    if header_end == -1 and len(buf) > BODY_MAX { return malformed(""); }
  }

  let lines = split(substr(buf, 0, header_end), "\r\n");
  let req_line = split_ws(trim(lines[0]));
  if len(req_line) < 3 { return malformed(""); }
  let method = req_line[0];
  let target = req_line[1];

  -- path / query split, both %-decoded ('+' is a space only in the query)
  let path = target;
  let query: map<Text, Text> = {};
  let qm = index_of(target, "?");
  if qm >= 0 {
    path = substr(target, 0, qm);
    query = parse_query(substr(target, qm + 1, len(target) - qm - 1));
  }
  path = url_decode(path, false);

  let headers: map<Text, Text> = {};
  let cl_seen = 0;
  let i = 1;
  while i < len(lines) {
    let line = trim(lines[i]);
    i = i + 1;
    if line == "" { continue; }
    let colon = index_of(line, ":");
    if colon > 0 {
      let hname = to_lower(substr(line, 0, colon));
      -- v1 slice 2, the strict-ambiguity audit: a SECOND Content-Length
      -- header is request smuggling's favorite tool — reject the request
      -- outright instead of letting last-one-wins pick a body length
      -- (RFC 9112 §6.3: such a message MUST be treated as an error).
      if hname == "content-length" {
        cl_seen = cl_seen + 1;
        if cl_seen > 1 { return malformed(""); }
      }
      headers[hname] = trim(substr(line, colon + 1, len(line) - colon - 1));
    }
  }

  let want = 0;
  let cl = headers["content-length"];
  if cl != nil {
    let n = parse_int(cl);
    if n == nil { return malformed(""); }
    if n < 0 or n > BODY_MAX { return malformed(""); }
    want = n;
  }

  let body = substr(buf, header_end + 4, len(buf) - header_end - 4);
  while len(body) < want {
    let got = net.read(c, 8192);
    if len(got) == 0 { return malformed(""); }     -- peer died mid-body
    body = body .. got;
  }
  -- bytes past the declared body belong to the next request on this conn
  let rest = "";
  if len(body) > want {
    rest = substr(body, want, len(body) - want);
    body = substr(body, 0, want);
  }

  let req = Req { method: method, path: path, params: {}, query: query,
                  headers: headers, body: body, principal: "", ctx: {},
                  conn: c };
  return Parsed { closed: false, ok: true, req: req, rest: rest };
}