feat(users/Profpatsch): add read-http r/2150

reads a http request or response from stdin, and parses its headers
into a netencoded record.

Darn rust code took way too long to write.

Change-Id: Ie99faa6d4bbd4996fa4e43fb119a11d85b611c99
Reviewed-on: https://cl.tvl.fyi/c/depot/+/2447
Reviewed-by: Profpatsch <mail@profpatsch.de>
Tested-by: BuildkiteCI

3 files changed, 217 insertions, 0 deletions

diff --git a/users/Profpatsch/read-http/default.nix b/users/Profpatsch/read-http/default.nix
new file mode 100644
index 000000000000..20f8675b3fd4
--- /dev/null
+++ b/users/Profpatsch/read-http/default.nix
@@ -0,0 +1,19 @@
+{ depot, pkgs, ... }:
+
+let
+
+  # reads a http request (stdin), and writes all headers to stdout, as netencoded dict
+  read-http = depot.users.Profpatsch.writers.rustSimple {
+    name = "read-http";
+    dependencies = [
+      depot.users.Profpatsch.rust-crates.httparse
+      depot.users.Profpatsch.netencode.netencode-rs
+      depot.users.Profpatsch.arglib.netencode.rust
+    ];
+  } (builtins.readFile ./read-http.rs);
+
+in {
+  inherit
+    read-http
+    ;
+}
diff --git a/users/Profpatsch/read-http/read-http.rs b/users/Profpatsch/read-http/read-http.rs
new file mode 100644
index 000000000000..6d419542c707
--- /dev/null
+++ b/users/Profpatsch/read-http/read-http.rs
@@ -0,0 +1,191 @@
+extern crate httparse;
+extern crate netencode;
+extern crate arglib_netencode;
+
+use std::os::unix::io::FromRawFd;
+use std::io::Read;
+use std::io::Write;
+
+use netencode::{U, T};
+
+enum What {
+    Request,
+    Response
+}
+
+fn main() -> std::io::Result<()> {
+    fn die<T: std::fmt::Display>(msg: T) -> ! {
+        eprintln!("{}", msg);
+        std::process::exit(1);
+    }
+
+    let what : What = match arglib_netencode::arglib_netencode(None).unwrap() {
+        T::Record(rec) => match rec.get("what") {
+            Some(T::Text(t)) => match t.as_str() {
+                "request" => What::Request,
+                "response" => What::Response,
+                _ => die("read-http arglib: what should be either t:request or t:response"),
+            },
+            Some(o) => die(format!("read-http arglib: expected a record of text, got {:#?}", o)),
+            None => {
+                eprintln!("read-http arglib: no `what` given, defaulting to Response");
+                What::Response
+            }
+        }
+        o => die(format!("read-http arglib: expected a record, got {:#?}", o))
+    };
+
+    fn read_stdin_to_complete<F>(mut parse: F) -> ()
+        where F: FnMut(&[u8]) -> httparse::Result<usize>
+    {
+        let mut res = httparse::Status::Partial;
+        loop {
+            if let httparse::Status::Complete(_) = res {
+                return;
+            }
+            let mut buf = [0; 2048];
+            match std::io::stdin().read(&mut buf[..]) {
+                Ok(size) => if size == 0 {
+                    break;
+                },
+                Err(err) => panic!("could not read from stdin, {:?}", err)
+            }
+            match parse(&buf) {
+                Ok(status) => {
+                    res = status;
+                }
+                Err(err) => die(format!("httparse parsing failed: {:#?}", err))
+            }
+        }
+    }
+
+
+    fn lowercase_headers<'a>(headers: &'a [httparse::Header]) -> Vec<(String, &'a [u8])> {
+        let mut res = vec![];
+        for httparse::Header { name, value } in headers {
+            // lowercase the headers, since the standard doesn’t care
+            // and we want unique strings to match agains
+            res.push((name.to_lowercase(), *value))
+        }
+        res
+    }
+
+    // tries to read until the end of the http header (deliniated by two newlines "\r\n\r\n")
+    fn read_till_end_of_header<R: Read>(buf: &mut Vec<u8>, reader: R) -> Option<()> {
+        let mut chunker = Chunkyboi::new(reader, 4096);
+        loop {
+            match chunker.next() {
+                Some(Ok(chunk)) => {
+                    buf.extend_from_slice(&chunk);
+                    if chunk.windows(4).any(|c| c == b"\r\n\r\n" ) {
+                        return Some(());
+                    }
+                },
+                Some(Err(err)) => die(format!("error reading from stdin: {:?}", err)),
+                None => return None
+            }
+        }
+    }
+
+    // max header size chosen arbitrarily
+    let mut headers = [httparse::EMPTY_HEADER; 128];
+    let stdin = std::io::stdin();
+
+    match what {
+        Request => {
+            let mut req = httparse::Request::new(&mut headers);
+            let mut buf: Vec<u8> = vec![];
+            match read_till_end_of_header(&mut buf, stdin.lock()) {
+                Some(()) => match req.parse(&buf) {
+                    Ok(httparse::Status::Complete(_body_start)) => {},
+                    Ok(httparse::Status::Partial) => die("httparse should have gotten a full header"),
+                    Err(err) => die(format!("httparse response parsing failed: {:#?}", err))
+                },
+                None => die(format!("httparse end of stdin reached before able to parse request headers"))
+            }
+            let method = req.method.expect("method must be filled on complete parse");
+            let path = req.path.expect("path must be filled on complete parse");
+            write_dict_req(method, path, &lowercase_headers(req.headers))
+        },
+        Response => {
+            let mut resp = httparse::Response::new(&mut headers);
+            let mut buf: Vec<u8> = vec![];
+            match read_till_end_of_header(&mut buf, stdin.lock()) {
+                Some(()) => match resp.parse(&buf) {
+                    Ok(httparse::Status::Complete(_body_start)) => {},
+                    Ok(httparse::Status::Partial) => die("httparse should have gotten a full header"),
+                    Err(err) => die(format!("httparse response parsing failed: {:#?}", err))
+                },
+                None => die(format!("httparse end of stdin reached before able to parse response headers"))
+            }
+            let code = resp.code.expect("code must be filled on complete parse");
+            let reason = resp.reason.expect("reason must be filled on complete parse");
+            write_dict_resp(code, reason, &lowercase_headers(resp.headers))
+        }
+    }
+}
+
+fn write_dict_req<'buf>(method: &'buf str, path: &'buf str, headers: &[(String, &[u8])]) -> std::io::Result<()> {
+    let mut http = vec![
+        ("method", U::Text(method.as_bytes())),
+        ("path", U::Text(path.as_bytes())),
+    ];
+    write_dict(http, headers)
+}
+
+fn write_dict_resp<'buf>(code: u16, reason: &'buf str, headers: &[(String, &[u8])]) -> std::io::Result<()> {
+    let mut http = vec![
+        ("status", U::N6(code as u64)),
+        ("status-text", U::Text(reason.as_bytes())),
+    ];
+    write_dict(http, headers)
+}
+
+
+fn write_dict<'buf, 'a>(mut http: Vec<(&str, U<'a>)>, headers: &'a[(String, &[u8])]) -> std::io::Result<()> {
+    http.push(("headers", U::Record(
+        headers.iter().map(
+            |(name, value)|
+            (name.as_str(), U::Binary(value))
+        ).collect::<Vec<_>>()
+    )));
+
+    netencode::encode(
+        &mut std::io::stderr(),
+        U::Record(http)
+    )?;
+    Ok(())
+}
+
+
+// iter helper
+
+struct Chunkyboi<T> {
+    inner: T,
+    buf: Vec<u8>,
+}
+
+impl<R: Read> Chunkyboi<R> {
+    fn new(inner: R, chunksize: usize) -> Self {
+        let buf = vec![0; chunksize];
+        Chunkyboi {
+            inner,
+            buf
+        }
+    }
+}
+
+impl<R: Read> Iterator for Chunkyboi<R> {
+    type Item = std::io::Result<Vec<u8>>;
+
+    fn next(&mut self) -> Option<std::io::Result<Vec<u8>>> {
+        match self.inner.read(&mut self.buf) {
+            Ok(0) => None,
+            Ok(read) => {
+                // clone a new buffer so we can reuse the internal one
+                Some(Ok(self.buf[..read].to_owned()))
+            }
+            Err(err) => Some(Err(err))
+        }
+    }
+}
diff --git a/users/Profpatsch/rust-crates.nix b/users/Profpatsch/rust-crates.nix
index ecf3c960775b..bb65625fc758 100644
--- a/users/Profpatsch/rust-crates.nix
+++ b/users/Profpatsch/rust-crates.nix
@@ -82,4 +82,11 @@ rec {
     sha256 = "0fcknyvknglwwk1pdzdlb4m0ry2dym1yx8r5prf2v00pxnjk0hv2";
   };
 
+  httparse = pkgs.buildRustCrate {
+    pname = "httparse";
+    version = "1.3.4";
+    crateName = "httparse";
+    sha256 = "0dggj4s0cq69bn63q9nqzzay5acmwl33nrbhjjsh5xys8sk2x4jw";
+  };
+
 }