|
| 1 | +package org.w3c.css.util; |
| 2 | + |
| 3 | +import java.io.FilterReader; |
| 4 | +import java.io.IOException; |
| 5 | +import java.io.Reader; |
| 6 | + |
| 7 | +public class UnescapeFilterReader extends FilterReader { |
| 8 | + |
| 9 | + public UnescapeFilterReader(Reader r) { |
| 10 | + super(r); |
| 11 | + } |
| 12 | + |
| 13 | + @Override |
| 14 | + public int read() |
| 15 | + throws IOException { |
| 16 | + int esc; |
| 17 | + int c = in.read(); |
| 18 | + // https://www.w3.org/TR/css-syntax-3/#input-preprocessing |
| 19 | + if (c == 13) { // U+000D CARRIAGE RETURN (CR) |
| 20 | + mark(1); |
| 21 | + c = in.read(); |
| 22 | + // eat any LF |
| 23 | + if (c != 10) { // U+000A LINE FEED (LF) |
| 24 | + reset(); |
| 25 | + } |
| 26 | + return 10; // U+000A LINE FEED (LF) |
| 27 | + } |
| 28 | + if (c == 12) { //U+000C FORM FEED (FF) |
| 29 | + return 10;// U+000A LINE FEED (LF) |
| 30 | + } |
| 31 | + if (c == 0) { // U+0000 NULL |
| 32 | + return 65533; // U+FFFD REPLACEMENT CHARACTER |
| 33 | + } |
| 34 | + |
| 35 | + // now specific case of CSS unicode escape for ascii values [A-Za-z0-9]. |
| 36 | + if (c != '\\') { |
| 37 | + return c; |
| 38 | + } |
| 39 | + mark(6); |
| 40 | + int val = 0; |
| 41 | + for (int i = 0; i < 6; i++) { |
| 42 | + esc = in.read(); |
| 43 | + // 0-9 |
| 44 | + if (esc > 47 && esc < 58) { |
| 45 | + val = (val << 4) + (esc - 48); |
| 46 | + } else if (esc > 64 && esc < 71) { |
| 47 | + // A_F |
| 48 | + val = (val << 4) + (esc - 55); |
| 49 | + } else if (esc > 96 && esc < 103) { |
| 50 | + val = (val << 4) + (esc - 87); |
| 51 | + } else if (esc == 10 || esc == 9 || esc == 32) { // CSS whitespace. |
| 52 | + // U+000A LINE FEED, U+0009 CHARACTER TABULATION, or U+0020 SPACE. |
| 53 | + if ((val > 96 && val < 124) || (val > 64 && val < 91) || (val > 47 && val < 58)) { |
| 54 | + return val; |
| 55 | + } |
| 56 | + } else { |
| 57 | + if ((val > 96 && val < 124) || (val > 64 && val < 91) || (val > 47 && val < 58)) { |
| 58 | + //we must unread 1 |
| 59 | + reset(); |
| 60 | + i++; |
| 61 | + for (int j = 0; j < i; j++) { |
| 62 | + in.read(); |
| 63 | + } |
| 64 | + return val; |
| 65 | + } |
| 66 | + reset(); |
| 67 | + return c; |
| 68 | + } |
| 69 | + } |
| 70 | + // we read up to 6 char test value first |
| 71 | + if ((val <= 96 || val >= 124) && (val <= 64 || val >= 91) && (val <= 47 || val >= 58)) { |
| 72 | + reset(); |
| 73 | + return c; |
| 74 | + } |
| 75 | + mark(1); |
| 76 | + c = in.read(); |
| 77 | + // not a CSS WHITESPACE |
| 78 | + if (c != 10 && c != 9 && c != 32) { |
| 79 | + reset(); |
| 80 | + } |
| 81 | + return val; |
| 82 | + } |
| 83 | + |
| 84 | + @Override |
| 85 | + public int read(char[] cbuf, int off, int len) throws IOException { |
| 86 | + int i, j, k, l, cki; |
| 87 | + char[] chars = new char[len]; |
| 88 | + in.mark(len); |
| 89 | + l = super.read(chars, 0, len); |
| 90 | + if (l <= 0) { |
| 91 | + return l; |
| 92 | + } |
| 93 | + for (i = 0, j = 0; i < l; i++) { |
| 94 | + // pre-processing |
| 95 | + if (chars[i] == 13) { |
| 96 | + chars[j++] = 10; |
| 97 | + // test for CRLF |
| 98 | + if (i + 1 < l && chars[i + 1] == 10) { |
| 99 | + i++; |
| 100 | + } |
| 101 | + } else if (chars[i] == 12) { |
| 102 | + chars[j++] = 10; |
| 103 | + } else if (chars[i] == 0) { |
| 104 | + chars[j++] = 65533; |
| 105 | + } |
| 106 | + // escaping |
| 107 | + |
| 108 | + if (chars[i] == '\\') { |
| 109 | + int val = 0; |
| 110 | + boolean escaped = false; |
| 111 | + for (k = 1; k < 7 && k + i < l; k++) { |
| 112 | + cki = chars[k + i]; |
| 113 | + // 0-9 |
| 114 | + if (cki > 47 && cki < 58) { |
| 115 | + val = (val << 4) + (cki - 48); |
| 116 | + } else if (cki > 64 && cki < 71) { |
| 117 | + // A_F |
| 118 | + val = (val << 4) + (cki - 55); |
| 119 | + } else if (cki > 96 && cki < 103) { |
| 120 | + val = (val << 4) + (cki - 87); |
| 121 | + } else if (cki == 10 || cki == 9 || cki == 32) { // CSS whitespace. |
| 122 | + // U+000A LINE FEED, U+0009 CHARACTER TABULATION, or U+0020 SPACE. |
| 123 | + if ((val > 96 && val < 124) || (val > 64 && val < 91) || (val > 47 && val < 58)) { |
| 124 | + chars[j++] = (char) val; |
| 125 | + escaped = true; |
| 126 | + i += k; |
| 127 | + break; |
| 128 | + } |
| 129 | + } else { |
| 130 | + if (val == 0) { |
| 131 | + if ((cki > 96 && cki < 124) || (cki > 64 && cki < 91)) { |
| 132 | + // so we found a regular char, just remove the escaping |
| 133 | + break; |
| 134 | + } |
| 135 | + } |
| 136 | + if ((val > 96 && val < 124) || (val > 64 && val < 91) || (val > 47 && val < 58)) { |
| 137 | + chars[j++] = (char) val; |
| 138 | + escaped = true; |
| 139 | + i += k - 1; |
| 140 | + break; |
| 141 | + } |
| 142 | + } |
| 143 | + } |
| 144 | + if (k == 7 && !escaped) { |
| 145 | + if ((val > 96 && val < 124) || (val > 64 && val < 91) || (val > 47 && val < 58)) { |
| 146 | + chars[j++] = (char) val; |
| 147 | + escaped = true; |
| 148 | + i += k - 1; |
| 149 | + if (i + 1 < l) { |
| 150 | + cki = chars[i + 1]; |
| 151 | + // skip extra space |
| 152 | + if (cki == 10 || cki == 9 || cki == 32) { |
| 153 | + i++; |
| 154 | + } |
| 155 | + } |
| 156 | + } else { |
| 157 | + // do nothing |
| 158 | + chars[j++] = chars[i]; |
| 159 | + } |
| 160 | + } else { |
| 161 | + // we reached the end, unescaping didn't happen let's stop here |
| 162 | + // unless we are the last |
| 163 | + if (!escaped) { |
| 164 | + if (j != 0) { |
| 165 | + in.reset(); |
| 166 | + in.skip(i); |
| 167 | + break; |
| 168 | + } else { |
| 169 | + chars[j++] = chars[i]; |
| 170 | + } |
| 171 | + } |
| 172 | + } |
| 173 | + } else { |
| 174 | + chars[j++] = chars[i]; |
| 175 | + } |
| 176 | + } |
| 177 | + |
| 178 | + System.arraycopy(chars, 0, cbuf, off, j); |
| 179 | + return j; |
| 180 | + } |
| 181 | +} |
0 commit comments