COMP 2406 2024F Assignment 1 Solutions

1. [2] Q: What directory does simpleserver2 use for serving content? How did you verify your answer was correct?

A: simpleserver2 uses ./www for serving content, i.e., the www subdirectory in the current directory (the one from which formdemo.js was run). We can see this by looking at line 66, which appends ./www to the path. You can verify this by creating a www directory, placing files in it, and observing that they are served when requested (and that no other files can be successfully requested). (1 for ./www, 1 for the verification method)


2. [2] Q: Does simpleserver2 support serving PDFs? (What does "support" in this context mean?) How did you verify that your answer was correct? 

A: simpleserver2 does not directly support the serving of PDFs, as it has no support for mapping files with the .pdf extension to the application/pdf MIME type. However, PDF files can be hosted and downloaded, they will just get the application/octet-stream, and many browsers will recognize the file type and handle the file appropriately.

I tested this by placing a PDF file in the www directory (one of my old papers) and accessed it in Firefox. The console log reported the type as application/octet-stream and the Firefox inspector showed the same; however, Firefox displayed the PDF using the built-in PDF reader as it normally does for downloaded PDFs despite the wrong MIME type.

(1 for the answer, 1 for the verification)


3. [3] Q: How does the behavior of simpleserver2 change if you make MIMEtype() an async function? (Be brief and specific.) How did you check this? And what could you change in simpleserver2 to make it work as before (without removing the async function declaration)?

A: When you make MIMEtype() an async function, all MIME types of returned files (except for ones that don't exist) is now [object Promise], i.e., the textual representation of a promise object. I verified this by changing the code and then retrieving different files stored in www.

To change the behavior back, I added an await to the one call to MIMEtype() on line 68:

    contentType = await MIMEtype(path);

I ran the same tests as before (downloading a few files) and all had the correct MIME type recorded in the console log.


4a. [2] Q: How could you modify simpleserver2 so that it only returns content on GET requests? When it receives a non-GET method, it should return the string "Method not implemented." and the status code 501.

A: Replace lines 83-87 with the following:

    if (req.method === "GET") {
	if (path === "/") {
            path = "/index.html";
	}
    
	var r = await fileData(path);
    } else {
	r = {
	    status: 501,
	    contentType: "text/plain",
	    contents: "Method not implemented.\n",
	};
    }

Lines 2-6 above is from the original code; here we've just enclosed it in an if statement that checks the value of req.method, and the else clause sets r to be the appropriate error response as required.

4b. [2] Q: What issues arose when implementing the answer to the previous question? How did you verify your answer was correct? (Hint: you'll probably want to use a command-line tool.)

A: My first implementation I set the value of r without creating an object, because I assumed there was an object already.

I used wget to test. First, to make sure I didn't break anything, I tried

  wget http://localhost:8000

which produced

200 GET text/html /

in the console log. (I had set up an index.html file in www.)

I then changed the method:

  wget --method=POST http://localhost:8000


In the terminal where I ran wget I got

--2024-10-06 15:32:50--  http://localhost:8000/
Resolving localhost (localhost)... 127.0.0.1
Connecting to localhost (localhost)|127.0.0.1|:8000... connected.
HTTP request sent, awaiting response... 501 Not Implemented
2024-10-06 15:32:50 ERROR 501: Not Implemented.


and I got the following in the console log:

501 POST text/plain /

However, I couldn't get wget to verify the "Method not implemented"
part. However, by viewing index.html from formdemo and hitting submit,
I got in the console:

501 POST text/plain /add

And "Method not implemented" on the screen in Firefox.


5a. [2] Q: How could you change simpleserver2 so that it loaded the page not found template from a file (rather than it being included in the source)? The template should be loaded when the server starts. Note that your solution should only use functions or methods that are in the tutorial code. What code needs to be added, changed, or removed? Be precise.

A: First, create the error page by running the web server, visiting a non-existent page, and then saving that page to a file, call it error.html in the same directory as simpleserver2.js.

Next, remove the functions template_notFound() and template_header(), those are no longer needed.

Then add this line just before the call to Deno.serve at the end of simpleserver2:

  const errorPage = await Deno.readTextFile("error.html");

(Note we used this function in Tutorial 1.)

Finally, change the old line 70 in fileData() from

  contents = template_notFound(path);

to

  contents = errorPage;

And that's it!

5b. [1] Q: Why do you think your solution is correct?

A: First, I verified that visiting a non-existent page brought up the same error page as before and existing documents could still be loaded.

To make sure I was running the new code, I added a console.log just before the Deno.readTextFile() call so I would see something different when the server started.

I then renamed the error.html file to something else and observed the server failed to start.


5c. [1] Q: What mistakes, errors, or other problems arose while you were developing your solution?

A: This problem was pretty straightforward, but I would expect there to be problems with 1) remembering how to load a file and 2) figuring out where to load the file, as it can be confusing to think about where that should happen.


6a. [2] Q: How could you change line 80 of simpleserver2.js (the assignment of origpath) so that it directly extracted the path (pathname) from the request URL string, instead of first converting it into a URL object? You can use any built-in JavaScript string manipluation facilities, including regular expressions. Make sure your solution does not include the protocol, hostname, port number, query, or fragment. Your solution should work for any host the the server is running on. What is your solution? Please give the precise code.

A: Note the below solution specifically avoids regular expressions, which is the tool that would normally be used for this problem. Instead, this one just uses split, join, and slice.

You could replace line 80 with the following lines:

    var pathPieces = req.url.split('/').slice(3);
    var pathWithOptions = '/' + pathPieces.join('/');
    var pathWithFragment = pathWithOptions.split('?')[0];
    var origpath = pathWithFragment.split('#')[0];

Note these all could be put into one line; the split is to try and make the conversion clearer. The first splits the string on /'s, with the slice(3) call discarding the first three elements of the split array, getting rid of the "http" or "https", "" (because we have a double slash), and the hostname. The second re-creates the path by adding back in the slashes. The third gets rid of the options (the ? and everything afterwards), and the fourth gets rid of the fragment (everything after a #).


6b. [1] Q: What resources did you use while developing your solution?

A: The wikipedia page on URLs makes the structure of URLs very clear:

  https://en.wikipedia.org/wiki/URL

But there are many other online sources.

6c. [1] Q: How did you test your solution to make sure it was correct?

A: I tried the code in the deno REPL, trying it with multiple URLs with various structure, some pasted from my regular browsing. But this was to verify that my understanding of the structure of URLs was correct.

6d. [1] What mistakes, errors, or other problems arose while you were developing your solution?

A: My original solution did not handle fragments; after re-reading the question, though, I realized what I had forgotten, updated the solution, and tested it on a URL with a fragment (a URL for a later section on the class wiki).