curl, and one HTTP request from Python

02One client, one file

The supplied note contains owned note followed by LF, the one-byte newline 0a. The space counts too: its complete size is 11 bytes. The server sends those bytes as the body and adds HTTP status and headers around them. curl transfers that response; it does not render the body as a browser page.

The local server command is python3 -m http.server 8000 --bind 127.0.0.1 --directory lab-site. The -m option asks Python to run the named module. The next number selects the listening port. --bind selects the address, limiting this example to local connections; --directory selects the folder whose contents are served.

Create a fresh parent directory and place the downloaded fetch-note.py there. Create its lab-site child and save the supplied note as lab-site/note.txt. Keep the server directory limited to that note, with no symbolic links, which are filesystem references that could lead outside the folder. Python’s simple server can follow such links. It is a local teaching server, not a production service.

Python 3.12’s default server response version here is HTTP/1.0 even though a client can send an HTTP/1.1 request. A response version describes that message. In the recorded run, Content-type was text/plain, Content-Length was 11, and the body was the supplied note. Time-related headers depend on the run and file.

The Python client imports urlopen from urllib.request, a standard-library module for opening URLs. with closes the response after use; response.status is the numeric HTTP status. response.read() obtains body bytes, decode("utf-8") interprets them as text, and print(body, end="") avoids adding a second newline. The final if line checks whether this file was launched directly; if it was, it calls the main function.

1curl -i http://127.0.0.1:8000/note.txt

1HTTP/1.0 200 OK
2Content-type: text/plain
3Content-Length: 11
4 
5owned note
6 

03Headers in the output, body in a file

For the browser route, follow the commands as actions in the supplied record and predict each result before reading its recorded lines. For a later lab run, use the fresh folder described above. Keep the server in one terminal and the clients in a second terminal, both starting in that parent folder.

owned note

"""Read one loopback note. The printed result is 200, then owned note. An earlier desk-note run is a different file."""

from urllib.request import urlopen


def main():
    with urlopen("http://127.0.0.1:8000/note.txt") as response:
        print(response.status)
        body = response.read().decode("utf-8")
        print(body, end="")


if __name__ == "__main__":
    main()

Recorded local exchange for lab-site/note.txt, 2026-10-12.
Python 3.12.3 and curl 8.5.0. The header/body listing combines
separately captured response headers and body; it is not one verbatim
curl -i stdout record.
An earlier fetch of desk-note on port 8765
is a different file.

Command whose option includes headers in output
curl -i http://127.0.0.1:8000/note.txt

The handler's default protocol was HTTP/1.0. The response also sent
Server: SimpleHTTP/0.6 Python/3.12.3, a Date, and Last-Modified.
Last-Modified followed the file's modification time. Those three headers
are not needed for the interpretation below.

HTTP/1.0 200 OK
Content-type: text/plain
Content-Length: 11

owned note

Content-Length 11 is the byte length of lab-site/note.txt, including its final LF.
Content-type for this file was text/plain.
A request for /missing.txt returned HTTP/1.0 404 File not found.

Activity command
curl -o note-copy.txt http://127.0.0.1:8000/note.txt

Observed contents of note-copy.txt
owned note

Check: note-copy.txt
That file contains the body. The separate -i command requests headers as well.

python3 fetch-note.py against the same server
200
owned note

A file URL such as file:///home/desk/lab-site/note.txt reads the local file.
It is a different operation from the HTTP URL above.

1. In the first terminal, start python3 -m http.server 8000 --bind 127.0.0.1 --directory lab-site. It remains running to handle requests, so leave that terminal open. The browser route identifies these options in the transcript instead.
2. In the second terminal, enter curl -i http://127.0.0.1:8000/note.txt. Expect a status line, response headers, a blank line, then owned note. Compare the stable recorded lines: status 200, text/plain, length 11.
3. Use a fresh destination: enter curl -o note-copy.txt http://127.0.0.1:8000/note.txt only where note-copy.txt does not already exist. Open that copy in your editor. It should contain the body owned note and its final LF, without HTTP headers.
4. From the same parent folder, enter python3 fetch-note.py. Its recorded result is 200 followed by owned note. On the browser route, trace read and decode to explain those lines instead of claiming a new run.
5. In the first terminal, press Ctrl+C to stop the server. The shell prompt should return. Stop when you can explain the two curl outputs and the Python result; retain only your own notes or copies and leave no server running.

04Optional challenge

For about fifteen extra minutes in a later local lab attempt, request /missing.txt from the same server. The earlier record received HTTP/1.0 404 File not found, meaning the request reached a server that reported an absent file. Compare that with a refused connection, which supplies no HTTP response. Stop your server afterward.

05Fresh cases for the checks

Use these fictional cases for the checks below. They have different inputs from the worked examples. Read them here or make notes; no installation or external service is needed.

Server command: python3 -m http.server 8164 --bind 127.0.0.1 --directory leaf-site
Only supplied invented files belong in leaf-site; no symbolic links or private files.
Request goal: include response headers while retrieving http://127.0.0.1:8164/entry.txt with curl.
Stop the foreground local server afterward using the taught interrupt.

HTTP/1.0 200 OK
Content-Type: text/plain
Content-Length: 6
 
ready
The body is ready followed by LF; it has six bytes. This is an authored message, not a captured curl run.

06Explain your prediction

Optional written self-check: Record which lines are status, headers and body, plus the server directory and loopback boundary. A refused connection supplies no HTTP response, while an HTTP 404 comes from a responding server.