(search) Don't show addresses with URLencoding in the UI

Optimize the code by adding a simple heuristic for guessing whether we need to repair the URI before we pass it to Java's parser.

code/common/model/java/nu/marginalia/model/EdgeUrl.java

@@ -31,7 +31,7 @@ public class EdgeUrl implements Serializable {
 
     private static URI parseURI(String url) throws URISyntaxException {
         try {
-            return EdgeUriFactory.uriFromString(url);
+            return EdgeUriFactory.parseURILenient(url);
         } catch (URISyntaxException ex) {
             throw new URISyntaxException("Failed to parse URI '" + url + "'", ex.getMessage());
         }
@@ -112,11 +112,32 @@ public class EdgeUrl implements Serializable {
             sb.append(port);
         }
 
+        EdgeUriFactory.urlencodePath(sb, path);
+
+        if (param != null) {
+            EdgeUriFactory.urlencodeQuery(sb, param);
+        }
+
+        return sb.toString();
+    }
+
+
+    public String toDisplayString() {
+        StringBuilder sb = new StringBuilder(256);
+
+        sb.append(proto);
+        sb.append("://");
+        sb.append(domain);
+
+        if (port != null) {
+            sb.append(':');
+            sb.append(port);
+        }
+
         sb.append(path);
 
         if (param != null) {
-            sb.append('?');
-            sb.append(param);
+            sb.append('?').append(param);
         }
 
         return sb.toString();
@@ -194,16 +215,19 @@ public class EdgeUrl implements Serializable {
 
 }
 
-/* Java's URI parser is a bit too strict in throwing exceptions when there's an error.
-
-   Here on the Internet, standards are like the picture on the box of the frozen pizza,
-   and what you get is more like what's on the inside, we try to patch things instead,
-   just give it a best-effort attempt att cleaning out broken or unnecessary constructions
-   like bad or missing URLEncoding
- */
 class EdgeUriFactory {
-    public static URI uriFromString(String url) throws URISyntaxException {
-        var s = new StringBuilder();
+    public static URI parseURILenient(String url) throws URISyntaxException {
+
+        if (shouldOmitUrlencodeRepair(url)) {
+            try {
+                return new URI(url);
+            }
+            catch (URISyntaxException ex) {
+                // ignore and run the lenient parser
+            }
+        }
+
+        var s = new StringBuilder(url.length()+8);
 
         int pathIdx = findPathIdx(url);
         if (pathIdx < 0) { // url looks like http://marginalia.nu
@@ -218,14 +242,18 @@ class EdgeUriFactory {
         int queryIdx = url.indexOf('?');
         if (queryIdx < 0) queryIdx = end;
 
-        recombinePaths(s, url.substring(pathIdx, queryIdx));
+        urlencodePath(s, url.substring(pathIdx, queryIdx));
         if (queryIdx < end) {
-            recombineQueryString(s, url.substring(queryIdx + 1, end));
+            urlencodeQuery(s, url.substring(queryIdx + 1, end));
         }
         return new URI(s.toString());
     }
 
-    private static void recombinePaths(StringBuilder sb, String path) {
+    /** Break apart the path element of an URI into its components, and then
+     * urlencode any component that needs it, and recombine it into a single
+     * path element again.
+     */
+    public static void urlencodePath(StringBuilder sb, String path) {
         if (path == null || path.isEmpty()) {
             return;
         }
@@ -236,12 +264,22 @@ class EdgeUriFactory {
             return;
         }
 
+        boolean shouldUrlEncode = false;
         for (String pathPart : pathParts) {
             if (pathPart.isEmpty()) continue;
 
             if (needsUrlEncode(pathPart)) {
+                shouldUrlEncode = true;
+                break;
+            }
+        }
+
+        for (String pathPart : pathParts) {
+            if (pathPart.isEmpty()) continue;
+
+            if (shouldUrlEncode) {
                 sb.append('/');
-                sb.append(URLEncoder.encode(pathPart, StandardCharsets.UTF_8));
+                sb.append(URLEncoder.encode(pathPart, StandardCharsets.UTF_8).replace("+", "%20"));
             } else {
                 sb.append('/');
                 sb.append(pathPart);
@@ -254,44 +292,64 @@ class EdgeUriFactory {
 
     }
 
-    private static void recombineQueryString(StringBuilder sb, String param) {
+    /** Break apart the query element of a URI into its components, and then
+     * urlencode any component that needs it, and recombine it into a single
+     * query element again.
+     */
+    public static void urlencodeQuery(StringBuilder sb, String param) {
         if (param == null || param.isEmpty()) {
             return;
         }
 
-        sb.append('?');
-        String[] pathParts = StringUtils.split(param, '&');
+        String[] queryParts = StringUtils.split(param, '&');
+
+        boolean shouldUrlEncode = false;
+        for (String queryPart : queryParts) {
+            if (queryPart.isEmpty()) continue;
+
+            if (needsUrlEncode(queryPart)) {
+                shouldUrlEncode = true;
+                break;
+            }
+        }
+
         boolean first = true;
-        for (String pathPart : pathParts) {
-            if (pathPart.isEmpty()) continue;
+        for (String queryPart : queryParts) {
+            if (queryPart.isEmpty()) continue;
 
             if (first) {
+                sb.append('?');
                 first = false;
             } else {
                 sb.append('&');
             }
-            if (needsUrlEncode(pathPart)) {
-                sb.append(URLEncoder.encode(pathPart, StandardCharsets.UTF_8));
+
+            if (shouldUrlEncode) {
+                int idx = queryPart.indexOf('=');
+                if (idx < 0) {
+                    sb.append(URLEncoder.encode(queryPart, StandardCharsets.UTF_8));
+                } else {
+                    sb.append(URLEncoder.encode(queryPart.substring(0, idx), StandardCharsets.UTF_8));
+                    sb.append('=');
+                    sb.append(URLEncoder.encode(queryPart.substring(idx + 1), StandardCharsets.UTF_8));
+                }
             } else {
-                sb.append(pathPart);
+                sb.append(queryPart);
             }
         }
     }
 
-
     /** Test if the url element needs URL encoding.
      * <p></p>
      * Note we may have been given an already encoded path element,
      * so we include % and + in the list of good characters
      */
-    private static boolean needsUrlEncode(String urlElement) {
+    static boolean needsUrlEncode(String urlElement) {
         for (int i = 0; i < urlElement.length(); i++) {
             char c = urlElement.charAt(i);
 
-            if (c >= 'a' && c <= 'z') continue;
-            if (c >= 'A' && c <= 'Z') continue;
-            if (c >= '0' && c <= '9') continue;
-            if ("-_.~+?=&".indexOf(c) >= 0) continue;
+            if (isUrlSafe(c)) continue;
+            if ("+".indexOf(c) >= 0) continue;
             if (c == '%' && i + 2 < urlElement.length()) {
                 char c1 = urlElement.charAt(i + 1);
                 char c2 = urlElement.charAt(i + 2);
@@ -307,14 +365,90 @@ class EdgeUriFactory {
         return false;
     }
 
-    private static boolean isHexDigit(char c) {
+
+    static boolean isUrlSafe(int c) {
+        if (c >= 'a' && c <= 'z') return true;
+        if (c >= 'A' && c <= 'Z') return true;
+        if (c >= '0' && c <= '9') return true;
+        if (c == '-' || c == '_' || c == '.' || c == '~') return true;
+
+        return false;
+    }
+
+    /** Test if the URL is a valid URL that does not need to be
+     * urlencoded.
+     * <p></p>
+     * This is a very simple heuristic test that does not guarantee
+     * that the URL is valid, but it will identify cases where we
+     * are fairly certain that the URL does not need encoding,
+     * so we can skip a bunch of allocations and string operations
+     * that would otherwise be needed to fix the URL.
+     */
+    static boolean shouldOmitUrlencodeRepair(String url) {
+        int idx = 0;
+        final int len = url.length();
+
+        // Validate the scheme
+        while (idx < len - 2) {
+            char c = url.charAt(idx++);
+            if (c == ':') break;
+            if (!isAsciiAlphabetic(c)) return false;
+        }
+        if (url.charAt(idx++) != '/') return false;
+        if (url.charAt(idx++) != '/') return false;
+
+        // Validate the authority
+        while (idx < len) {
+            char c = url.charAt(idx++);
+            if (c == '/') break;
+            if (c == ':') continue;
+            if (c == '@') continue;
+            if (!isUrlSafe(c)) return false;
+        }
+
+        // Validate the path
+        if (idx >= len) return true;
+
+        while (idx < len) {
+            char c = url.charAt(idx++);
+            if (c == '?') break;
+            if (c == '/') continue;
+            if (c == '#') return true;
+            if (!isUrlSafe(c)) return false;
+        }
+
+        if (idx >= len) return true;
+
+        // Validate the query
+        while (idx < len) {
+            char c = url.charAt(idx++);
+            if (c == '&') continue;
+            if (c == '=') continue;
+            if (c == '#') return true;
+            if (!isUrlSafe(c)) return false;
+        }
+
+        return true;
+    }
+
+
+    private static boolean isAsciiAlphabetic(int c) {
+        return (c >= 'a' && c <= 'f') || (c >= 'A' && c <= 'F');
+    }
+
+    private static boolean isHexDigit(int c) {
         return (c >= '0' && c <= '9') || (c >= 'a' && c <= 'f') || (c >= 'A' && c <= 'F');
     }
 
+    /** Find the index of the path element in a URL.
+     * <p></p>
+     * The path element starts after the scheme and authority part of the URL,
+     * which is everything up to and including the first slash after the colon.
+     */
     private static int findPathIdx(String url) throws URISyntaxException {
         int colonIdx = url.indexOf(':');
         if (colonIdx < 0 || colonIdx + 3 >= url.length()) {
-            throw new URISyntaxException(url, "Lacking protocol");
+            throw new URISyntaxException(url, "Lacking scheme");
         }
         return url.indexOf('/', colonIdx + 3);
     }

code/common/model/test/nu/marginalia/model/EdgeUrlTest.java

@@ -24,23 +24,66 @@ class EdgeUrlTest {
 
     @Test
     void testUriFromString() throws URISyntaxException {
-        Assertions.assertEquals("https://www.example.com/", EdgeUriFactory.uriFromString("https://www.example.com/").toString());
-        Assertions.assertEquals("https://www.example.com/", EdgeUriFactory.uriFromString("https://www.example.com/#heredoc").toString());
-        Assertions.assertEquals("https://www.example.com/trailingslash/", EdgeUriFactory.uriFromString("https://www.example.com/trailingslash/").toString());
-        Assertions.assertEquals("https://www.example.com/%25-sign", EdgeUriFactory.uriFromString("https://www.example.com/%-sign").toString());
-        Assertions.assertEquals("https://www.example.com/%22-sign", EdgeUriFactory.uriFromString("https://www.example.com/%22-sign").toString());
-        Assertions.assertEquals("https://www.example.com/%0A+%22huh%22", EdgeUriFactory.uriFromString("https://www.example.com/\n \"huh\"").toString());
-        Assertions.assertEquals("https://en.wikipedia.org/wiki/S%C3%A1mi", EdgeUriFactory.uriFromString("https://en.wikipedia.org/wiki/Sámi").toString());
+        // We test these URLs several times as we perform URLEncode-fixing both when parsing the URL and when
+        // converting it back to a string, we want to ensure there is no changes along the way.
+
+        Assertions.assertEquals("/", EdgeUriFactory.parseURILenient("https://www.example.com/").getPath());
+        Assertions.assertEquals("https://www.example.com/", EdgeUriFactory.parseURILenient("https://www.example.com/").toString());
+        Assertions.assertEquals("https://www.example.com/", new EdgeUrl("https://www.example.com/").toString());
+
+        Assertions.assertEquals("/", EdgeUriFactory.parseURILenient("https://www.example.com/#heredoc").getPath());
+        Assertions.assertEquals("https://www.example.com/", EdgeUriFactory.parseURILenient("https://www.example.com/#heredoc").toString());
+        Assertions.assertEquals("https://www.example.com/", new EdgeUrl("https://www.example.com/#heredoc").toString());
+
+        Assertions.assertEquals("/trailingslash/", EdgeUriFactory.parseURILenient("https://www.example.com/trailingslash/").getPath());
+        Assertions.assertEquals("https://www.example.com/trailingslash/", EdgeUriFactory.parseURILenient("https://www.example.com/trailingslash/").toString());
+        Assertions.assertEquals("https://www.example.com/trailingslash/", new EdgeUrl("https://www.example.com/trailingslash/").toString());
+
+        Assertions.assertEquals("/%-sign", EdgeUriFactory.parseURILenient("https://www.example.com/%-sign").getPath());
+        Assertions.assertEquals("https://www.example.com/%25-sign", EdgeUriFactory.parseURILenient("https://www.example.com/%-sign").toString());
+        Assertions.assertEquals("https://www.example.com/%25-sign", new EdgeUrl("https://www.example.com/%-sign").toString());
+
+        Assertions.assertEquals("/%-sign/\"-sign", EdgeUriFactory.parseURILenient("https://www.example.com//%-sign/\"-sign").getPath());
+        Assertions.assertEquals("https://www.example.com/%25-sign/%22-sign", EdgeUriFactory.parseURILenient("https://www.example.com//%-sign/\"-sign").toString());
+        Assertions.assertEquals("https://www.example.com/%25-sign/%22-sign", new EdgeUrl("https://www.example.com//%-sign/\"-sign").toString());
+
+        Assertions.assertEquals("/\"-sign", EdgeUriFactory.parseURILenient("https://www.example.com/%22-sign").getPath());
+        Assertions.assertEquals("https://www.example.com/%22-sign", EdgeUriFactory.parseURILenient("https://www.example.com/%22-sign").toString());
+        Assertions.assertEquals("https://www.example.com/%22-sign", new EdgeUrl("https://www.example.com/%22-sign").toString());
+
+        Assertions.assertEquals("/\n \"huh\"", EdgeUriFactory.parseURILenient("https://www.example.com/\n \"huh\"").getPath());
+        Assertions.assertEquals("https://www.example.com/%0A%20%22huh%22", EdgeUriFactory.parseURILenient("https://www.example.com/\n \"huh\"").toString());
+        Assertions.assertEquals("https://www.example.com/%0A%20%22huh%22", new EdgeUrl("https://www.example.com/\n \"huh\"").toString());
+
+        Assertions.assertEquals("/wiki/Sámi", EdgeUriFactory.parseURILenient("https://en.wikipedia.org/wiki/Sámi").getPath());
+        Assertions.assertEquals("https://en.wikipedia.org/wiki/S%C3%A1mi", EdgeUriFactory.parseURILenient("https://en.wikipedia.org/wiki/Sámi").toString());
+        Assertions.assertEquals("https://en.wikipedia.org/wiki/S%C3%A1mi", new EdgeUrl("https://en.wikipedia.org/wiki/Sámi").toString());
+
+        Assertions.assertEquals("https://www.prijatelji-zivotinja.hr/index.en.php?id=2301k", new EdgeUrl("https://en.wikipedia.org/wiki/Sámi").toString());
     }
 
     @Test
     void testParms() throws URISyntaxException {
         Assertions.assertEquals("id=123", new EdgeUrl("https://search.marginalia.nu/?id=123").param);
+        Assertions.assertEquals("https://search.marginalia.nu/?id=123", new EdgeUrl("https://search.marginalia.nu/?id=123").toString());
+
         Assertions.assertEquals("t=123", new EdgeUrl("https://search.marginalia.nu/?t=123").param);
+        Assertions.assertEquals("https://search.marginalia.nu/?t=123", new EdgeUrl("https://search.marginalia.nu/?t=123").toString());
+
         Assertions.assertEquals("v=123", new EdgeUrl("https://search.marginalia.nu/?v=123").param);
+        Assertions.assertEquals("https://search.marginalia.nu/?v=123", new EdgeUrl("https://search.marginalia.nu/?v=123").toString());
+
         Assertions.assertEquals("id=1", new EdgeUrl("https://memex.marginalia.nu/showthread.php?id=1&count=5&tracking=123").param);
+        Assertions.assertEquals("https://memex.marginalia.nu/showthread.php?id=1",
+                new EdgeUrl("https://memex.marginalia.nu/showthread.php?id=1&count=5&tracking=123").toString());
+
+
         Assertions.assertEquals("id=1&t=5", new EdgeUrl("https://memex.marginalia.nu/shöwthrëad.php?id=1&t=5&tracking=123").param);
+        Assertions.assertEquals("https://memex.marginalia.nu/sh%C3%B6wthr%C3%ABad.php?id=1&t=5", new EdgeUrl("https://memex.marginalia.nu/shöwthrëad.php?id=1&t=5&tracking=123").toString());
+
         Assertions.assertEquals("id=1&t=5", new EdgeUrl("https://memex.marginalia.nu/shöwthrëad.php?trëaking=123&id=1&t=5&").param);
+        Assertions.assertEquals("https://memex.marginalia.nu/sh%C3%B6wthr%C3%ABad.php?id=1&t=5", new EdgeUrl("https://memex.marginalia.nu/shöwthrëad.php?trëaking=123&id=1&t=5&").toString());
+
         Assertions.assertNull(new EdgeUrl("https://search.marginalia.nu/?m=123").param);
         Assertions.assertNull(new EdgeUrl("https://search.marginalia.nu/?follow=123").param);
     }

code/services-application/search-service/java/nu/marginalia/search/model/UrlDetails.java

@@ -180,7 +180,7 @@ public class UrlDetails implements Comparable<UrlDetails> {
      * semantically meaningful codepoints into entity codes */
     public String displayUrl() {
         StringBuilder sb = new StringBuilder();
-        String urlStr = url.toString();
+        String urlStr = url.toDisplayString();
         for (int i = 0; i < urlStr.length(); i++) {
             char c = urlStr.charAt(i);
 

code/services-application/status-service/java/nu/marginalia/status/StatusModule.java

@@ -20,6 +20,6 @@ public class StatusModule extends AbstractModule {
         bind(String.class)
                 .annotatedWith(Names.named("searchEngineTestQuery"))
                 .toInstance(System.getProperty("status-service.public-query",
-                        "https://search.marginalia.nu/search?query=plato&ref=marginalia-automatic-metrics"));
+                        "https://marginalia-search.com/search?query=plato&ref=marginalia-automatic-metrics"));
     }
 }

deploy.txt

@@ -3,4 +3,5 @@
 2025-01-08:  Deploy executor.
 2025-01-07:  Deploy executor.
 2025-04-24:  Deploy executor.
-2025-04-24:  Deploy assistant.
+2025-04-24:  Deploy assistant.
+2025-04-24:  Deploy qs, search and api-services.

tools/deployment/deployment.py

@@ -314,6 +314,13 @@ if __name__ == '__main__':
             deploy_tier=0,
             groups={"all", "core"}
         ),
+        'status': ServiceConfig(
+            gradle_target=':code:services-application:status-service:docker',
+            docker_name='status-service',
+            instances=None,
+            deploy_tier=4,
+            groups={"all"}
+        ),
         'query': ServiceConfig(
             gradle_target=':code:services-core:query-service:docker',
             docker_name='query-service',