URIUtils.java

/*
 * This file is part of vospace-rest
 * Copyright (C) 2021 Istituto Nazionale di Astrofisica
 * SPDX-License-Identifier: GPL-3.0-or-later
 */
package it.inaf.oats.vospace;

import it.inaf.oats.vospace.exception.InternalFaultException;
import it.inaf.oats.vospace.exception.InvalidURIException;
import java.net.URI;
import java.net.URISyntaxException;
import java.util.regex.Pattern;
import net.ivoa.xml.vospace.v2.Node;

public class URIUtils {

    // It's different from the one in NodeUtils!
    // Slashes are treated separately
    private static final Pattern FORBIDDEN_CHARS = Pattern.compile("[\\x00\\x08\\x0B\\x0C\\x0E-\\x1F" + Pattern.quote("<>?\":\\|'`*") + "]");
    private static final String SCHEME = "vos";
    
    public static boolean isURIInternal(String URI) {
        if(URI == null)
            throw new IllegalArgumentException("URI can't be null");
        
        return URI.toLowerCase().startsWith(SCHEME);
    }

    public static String returnURIFromVosPath(String vosPath, String authority) {
        String result = null;

        try {
            URI uri = new URI(
                    SCHEME,
                    authority,
                    vosPath,
                    null,
                    null
            );

            result = uri.toASCIIString();

        } catch (URISyntaxException e) {
            throw new InternalFaultException("unable to percent encode URI from authority and path: "
                    + authority + " , " + vosPath);
        }

        return result;
    }

    public static String returnVosPathFromNodeURI(Node myNode, String authority) {
        return returnVosPathFromNodeURI(myNode.getUri(), authority);
    }

    // This method validates the URI too
    public static String returnVosPathFromNodeURI(String nodeURI, String authority) {

        String resultPath = null;

        try {
            
            if(nodeURI == null)
                throw new IllegalArgumentException("URI string is null");
            
            URI uri = new URI(nodeURI);

            // Check scheme
            if (!uri.isAbsolute()
                    || uri.isOpaque()
                    || !uri.getRawSchemeSpecificPart().startsWith("//")
                    || !uri.getScheme().equalsIgnoreCase(SCHEME)) {
                throw new InvalidURIException(nodeURI);
            }

            // Check authority
            if (!uri.getAuthority().replace("~", "!").equals(authority)) {
                throw new InvalidURIException(nodeURI);
            }

            // Check path
            String rawPath = uri.getRawPath();

            // Check if raw Path is null or contains percent encoded slashes or multiple
            // separators
            if (rawPath == null
                    || rawPath.contains("//")
                    || rawPath.contains("%2F")
                    || rawPath.contains("%2f")) {
                throw new InvalidURIException(nodeURI);
            }

            resultPath = uri.getPath();

            if (resultPath.isBlank()
                    || FORBIDDEN_CHARS.matcher(resultPath).find()
                    || (!resultPath.equals("/") && resultPath.endsWith("/"))) {
                throw new InvalidURIException(nodeURI);
            }

        } catch (URISyntaxException e) {
            throw new InvalidURIException(nodeURI);
        } 

        return resultPath;

    }

}