Commit fb1db93
John J. Aylward
Changes encoding to better match the XML spec section 2.2
1 parent adb0478 commit fb1db93
1 file changed
Lines changed: 27 additions & 1 deletion
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
@@ -137,7 +137,7 @@ public static String escape(String string) { | |||
| 137 | 137 | sb.append("'"); | |
| 138 | 138 | break; | |
| 139 | 139 | default: | |
| 140 | - if (Character.isISOControl(cp)) { | ||
| 140 | + if (mustEscape(cp)) { | ||
| 141 | 141 | sb.append("&#x"); | |
| 142 | 142 | sb.append(Integer.toHexString(cp)); | |
| 143 | 143 | sb.append(";"); | |
@@ -149,6 +149,32 @@ public static String escape(String string) { | |||
| 149 | 149 | return sb.toString(); | |
| 150 | 150 | } | |
| 151 | 151 | ||
| 152 | + /** | ||
| 153 | + * @param cp code point to test | ||
| 154 | + * @return true if the code point is not valid for an XML | ||
| 155 | + */ | ||
| 156 | + private static boolean mustEscape(int cp) { | ||
| 157 | + /* Valid range from https://www.w3.org/TR/REC-xml/#charsets | ||
| 158 | + * | ||
| 159 | + * #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF] | ||
| 160 | + * | ||
| 161 | + * any Unicode character, excluding the surrogate blocks, FFFE, and FFFF. | ||
| 162 | + */ | ||
| 163 | + // isISOControl is true when (cp >= 0 && cp <= 0x1F) || (cp >= 0x7F && cp <= 0x9F) | ||
| 164 | + // all ISO control characters are out of range except tabs and new lines | ||
| 165 | + return (Character.isISOControl(cp) | ||
| 166 | + && cp != 0x9 | ||
| 167 | + && cp != 0xA | ||
| 168 | + && cp != 0xD | ||
| 169 | + ) || !( | ||
| 170 | + // valid the range of acceptable characters that aren't control | ||
| 171 | + (cp >= 0x20 && cp <= 0xD7FF) | ||
| 172 | + || (cp >= 0xE000 && cp <= 0xFFFD) | ||
| 173 | + || (cp >= 0x10000 && cp <= 0x10FFFF) | ||
| 174 | + ) | ||
| 175 | + ; | ||
| 176 | + } | ||
| 177 | + | ||
| 152 | 178 | /** | |
| 153 | 179 | * Removes XML escapes from the string. | |
| 154 | 180 | * | |
0 commit comments