Changeset 18869 in josm

Timestamp:

2023-10-13T13:50:04+02:00 (8 months ago)

Author:

taylor.smock

Message:

Fix #23235: Allow unicode characters in URL paths

Location:

trunk

Files:

• src/org/openstreetmap/josm/data/validation/routines/UrlValidator.java (modified) (4 diffs)
• test/unit/org/openstreetmap/josm/data/validation/routines/UrlValidatorTest.java (modified) (1 diff)

trunk/src/org/openstreetmap/josm/data/validation/routines/UrlValidator.java

@@ -132 +132 @@
      * Protocol scheme (e.g. http, ftp, https).
      */
-    private static final String SCHEME_REGEX = "^\\p{Alpha}[\\p{Alnum}\\+\\-\\.]*";
+    private static final String SCHEME_REGEX = "^\\p{Alpha}[\\p{Alnum}+\\-.]*";
     private static final Pattern SCHEME_PATTERN = Pattern.compile(SCHEME_REGEX);
 
@@ -138 +138 @@
     // TODO does not allow for optional userinfo.
     // Validation of character set is done by isValidAuthority
-    private static final String AUTHORITY_CHARS_REGEX = "\\p{Alnum}\\-\\."; // allows for IPV4 but not IPV6
+    private static final String AUTHORITY_CHARS_REGEX = "\\p{Alnum}-\\."; // allows for IPV4 but not IPV6
     private static final String IPV6_REGEX = "[0-9a-fA-F:]+"; // do this as separate match because : could cause ambiguity with port prefix
 
@@ -151 +151 @@
             USERINFO_CHARS_REGEX + "*@"; // password may be absent
     private static final String AUTHORITY_REGEX =
-            "(?:\\[("+IPV6_REGEX+")\\]|(?:(?:"+USERINFO_FIELD_REGEX+")?([" + AUTHORITY_CHARS_REGEX + "]*)))(:\\d*)?(.*)?";
+            "(?:\\[("+IPV6_REGEX+")]|(?:(?:"+USERINFO_FIELD_REGEX+")?([" + AUTHORITY_CHARS_REGEX + "]*)))(:\\d*)?(.*)?";
     //             1                          e.g. user:pass@          2                                   3       4
     private static final Pattern AUTHORITY_PATTERN = Pattern.compile(AUTHORITY_REGEX);
@@ -164 +164 @@
     private static final int PARSE_AUTHORITY_EXTRA = 4;
 
-    private static final String PATH_REGEX = "^(/[-\\w:@&?=+,.!/~*'%$_;\\(\\)]*)?$";
-    private static final Pattern PATH_PATTERN = Pattern.compile(PATH_REGEX);
+    private static final String PATH_REGEX = "^(/[-\\w:@&?=+,.!/~*'%$_;()]*)?$";
+    private static final Pattern PATH_PATTERN = Pattern.compile(PATH_REGEX, Pattern.UNICODE_CHARACTER_CLASS);
 
     private static final String QUERY_REGEX = "^(.*)$";

trunk/test/unit/org/openstreetmap/josm/data/validation/routines/UrlValidatorTest.java

@@ -565 +565 @@
             new ResultPair("/test1/file", true),
             new ResultPair("/..//file", false),
-            new ResultPair("/test1//file", false)
+            new ResultPair("/test1//file", false),
+            new ResultPair("/projekte-und-themen/eulen-falken-und-deren-nistkästen/trafotürme", true)
     };