itsb/jq/helpers.jq

# Parse URLs into an object with {scheme, netloc, path, params, query, fragment}.
# Similar to Python's urllib.parse.urlparse.
def urlparse: capture("^(?:(?<scheme>[^:/?#]+):)?(?://(?<netloc>[^/?#]*))?(?:(?<path>(?:[^?#]+/)?[^?#;]*)(?:;(?<params>[^?#/]*))?)?(?:\\?(?<query>[^#]*))?(?:#(?<fragment>.*))?$");

# Parse URLs into an object with {scheme, netloc, path, query, fragment}. Path parameters are not parsed.
# Similar to Python's urllib.parse.urlsplit.
def urlsplit: capture("^(?:(?<scheme>[^:/?#]+):)?(?://(?<netloc>[^/?#]*))?(?<path>(?:[^?#]+/)?[^?#]*)?(?:\\?(?<query>[^#]*))?(?:#(?<fragment>.*))?$");

# Reverse operation of either urlparse or urlsplit.
def urlunparse:
    (if .scheme then .scheme + "://" else "" end)
    + (.netloc // "")
    + (.path // "")
    + (if .params then ";" + .params else "" end)
    + (if .query then "?" + .query else "" end)
    + (if .fragment then "#" + .fragment else "" end);

# Resolve a possibly relative URI into an absolute URI.
def urlresolve(base):
    (if type == "string" then urlsplit else . end) as $parsed
    # There is a scheme: this is an absolute URL
    | if $parsed.scheme then . else (
        base|(if type == "string" then urlsplit else . end) as $parsedbase
        # No scheme but a domain: use the base's scheme
        | $parsed
        | if .netloc then (
            .scheme = $parsedbase.scheme
        # No scheme and no domain: resolve the relative URI
        ) elif .path then (
            .scheme = $parsedbase.scheme
            | .netloc = $parsedbase.netloc
            # When the path does not start with a slash, make it relative to the base's path
            # Note that this assumes the base URL always points to a folder, even if it does not end with a /
            | if .path|startswith("/")|not then (
              .path = (($parsedbase.path|rtrimstr("/")) + "/" + ($parsed.path|ltrimstr("/")))
            ) else . end
        ) elif (.query // .fragment) then (
            .scheme = $parsedbase.scheme
            | .netloc = $parsedbase.netloc
            | .path = $parsedbase.path
        ) else . end
        | urlunparse
    ) end;