Noja/examples/http.noja

scan = import("scan.noja");

newScanner = scan.newScanner;
Scanner = scan.Scanner;
hint = scan.hint;
consume = scan.consume;
current = scan.current;
isDigit = scan.isDigit;
consumeSpaces = scan.consumeSpaces;

fun isAlpha(char: String)
    return (string.ord(char) >= string.ord('a') and string.ord(char) <= string.ord('z'))
        or (string.ord(char) >= string.ord('A') and string.ord(char) <= string.ord('Z'));

fun isUpper(char: String)
    return string.ord(char) >= string.ord('A')
       and string.ord(char) <= string.ord('Z');

fun isUnreserved(char: String)
    return isAlpha(char)
        or isDigit(char)
        or char == "-"
        or char == "."
        or char == "_"
        or char == "~";

fun isSubdelim(char: String)
    return char == "!" or char == "$"
        or char == "&" or char == "'"
        or char == "(" or char == ")"
        or char == "*" or char == "+"
        or char == "," or char == ";"
        or char == "=";

fun isPChar(char: String)
    return isUnreserved(char)
        or isSubdelim(char)
        or char == ":" 
        or char == "@";

fun parseMethod(scan: Scanner) {
    
    char = current(scan);
    if char == none or not isUpper(char):
        return none, "Missing method";

    method = "";
    do {
        method = string.cat(method, char);
        consume(scan);
        char = current(scan);
    } while char != none and isUpper(char);

    return method, none;
}

fun parsePort(scan: Scanner) {

    port = none;
    char = current(scan);
    if char == ":" and isDigit(hint(scan, 1)): {

        consume(scan); # Skip the ":"
        char = current(scan);

        port = 0;
        do {
            digit = string.ord(char)
                  - string.ord("0");
            port = port * 10 + digit;
            consume(scan);
            char = current(scan);
        } while char != none and isDigit(char);
    }
    return port;
}

fun isHostname(char: String)
    return isUnreserved(char)
        or isSubdelim(char);

fun parseHost(scan: Scanner) {

    char = current(scan);
    if char == none:
        return none, "Missing host";

    if not isHostname(char):
        return none, "Invalid character in place of host";

    name = "";
    do {
        name = string.cat(name, char);
        consume(scan);
        char = current(scan);
    } while char != none and isHostname(char);

    port = parsePort(scan);
    return {name: name, port: port};
}

fun parsePath(scan: Scanner) {

    path = "";
    char = current(scan);
    if char != none and char == "/": {
        path = string.cat(path, char);
        consume(scan);
        char = current(scan);
    } else {
        if char == none or not isPChar(char):
            return none, "Invalid character in place of path";
    }

    while char != none and isPChar(char): {
        do {
            path = string.cat(path, char);
            consume(scan);
            char = current(scan);
        } while char != none and isPChar(char);
        if char == none or char != "/":
            break;
        path = string.cat(path, char);
        consume(scan);
        char = current(scan);
    }

    return path;
}

fun isQuery(char: ?String) {
    if char == none:
        return false;
    return isPChar(char) 
        or char == "/" 
        or char == "?";
}

fun parseQuery(scan: Scanner) {
    
    char = current(scan);
    if char != none and char == "?": {
        consume(scan); # Consume the "?"
        query = "";
        while isQuery(char = current(scan)): {
            query = string.cat(query, char);
            consume(scan);
        }
    } else
        query = none;

    return query;
}

fun isFragment(char: ?String) {
    if char == none:
        return false;
    return isPChar(char) or char == "/";
}

fun parseFragment(scan: Scanner) {
    
    char = current(scan);
    if char != none and char == "#": {
        consume(scan); # Consume the "#"
        fragment = "";
        while isFragment(char = current(scan)): {
            fragment = string.cat(fragment, char);
            consume(scan);
        }
    } else
        fragment = none;

    return fragment;
}

Host = { name: String, port: ?int };

URL = {
    schema: ?String,
    host: ?Host,
    path: ?String,
    query: ?String,
    fragment: ?String
};

fun isSchemaFirst(char: ?String) {
    return char != none and isAlpha(char);
}

fun isSchema(char: ?String)
    return char != none 
       and (isAlpha(char) 
        or isDigit(char)
        or char == "+" 
        or char == "-"
        or char == ".");

fun parseSchema(scan: Scanner) {

    char = current(scan);
    schema = none;
    if isSchemaFirst(char): {
        start = scan.i;
        schema = "";
        do {
            schema = string.cat(schema, char);
            consume(scan);
            char = current(scan);
        } while isSchema(char);

        if char == ":": 
            consume(scan);
        else {
            schema = none;
            scan.i = start;
        }
    }

    return schema;
}

fun followsAuthority(scan: Scanner)
    return current(scan) == "/" and hint(scan, 1) == "/";

fun parseURL(scan: Scanner) {

    schema = parseSchema(scan);
    
    if followsAuthority(scan): {
        consume(scan, 2);
        
        host, error = parseHost(scan);
        if error != none:
            return none, error;
        
        path = none;
        if current(scan) == "/":
            path = parsePath(scan);

    } else {
        
        host = none;

        char = current(scan);
        if char == none or char == "?" or char == "#":
            return none, "Missing path";

        path, error = parsePath(scan);
        if error != none:
            return none, error;
    }

    query = parseQuery(scan);
    fragment = parseFragment(scan);
    print(scan);

    return {
        host    : host,
        path    : path, 
        query   : query,
        schema  : schema, 
        fragment: fragment
    };
}

fun parseVersion(scan: Scanner) {

    if hint(scan, 0) != "H" or hint(scan, 1) != "T"
    or hint(scan, 2) != "T" or hint(scan, 3) != "P"
    or hint(scan, 4) != "/" or not isDigit(hint(scan, 5)):
        return none, "Invalid version token";

    consume(scan, 5); 
    major = string.ord(current(scan)) 
          - string.ord("0");

    if hint(scan, 1) != "." and isDigit(hint(scan, 2)): {
        consume(scan, 2);
        minor = string.ord(current(scan)) 
              - string.ord("0");
    } else
        minor = 0;
    consume(scan);

    return {major: major, minor: minor};
}

Version = {
    major: int,
    minor: int
};

Request = {
    method : String, 
    url    : URL, 
    version: Version
};

fun newRequest(method: String, url: URL, 
               version: Version, headers: Map = {})
    return {
        method: method, 
        url: url, 
        version: version,
        headers: headers
    };

fun parse(src: String) {
    
    scan = newScanner(src);
    
    method, error = parseMethod(scan);
    if error != none:
        return none, error;

    char = current(scan);
    if char == none or char != " ":
        return none, "Missing space after method";
    consume(scan);

    url, error = parseURL(scan);
    if error != none:
        return none, error;

    char = current(scan);
    if char != " ":
        return none, "Missing space after URL";
    consume(scan);

    version, error = parseVersion(scan);
    if error != none:
        return none, error;

    if hint(scan, 0) != "\r" or hint(scan, 1) != "\n":
        # Request with no headers or body
        return newRequest(method, url, version);
    consume(scan, 2); # Consume the "\r\n"

    headers = {};
    while (char = hint(scan, 0)) != none and (char != "\r" or hint(scan, 1) != "\n"): {

        # Header name until the next ":"
        name = "";
        do {
            name = string.cat(name, char);
            consume(scan);
            char = current(scan);
        } while char != ":" and char != none;
        
        if char == none:
            return none, "Invalid header (missing \":\" after name)";
        consume(scan);
        consumeSpaces(scan);
        char = current(scan);

        # Header body until \r
        body = "";
        do {
            body = string.cat(body, char);
            consume(scan);
            char = current(scan);
        } while char != "\r" and char != none;

        headers[name] = body;
    }

    if char != none:
        consume(scan, 2); # Consume the final "\r\n"

    return newRequest(method, url, version, headers);
}

sample = string.cat(
    "GET /search?client=firefox-b-d&q=http+request+dataset HTTP/2\r\n",
    "Host: www.google.com\r\n",
    "user-agent: curl/7.84.0\r\n",
    "accept: */*\r\n",
    "\r\n");

request, error = parse(sample);
print("request=", request, "\n");
print("error=", error, "\n");