Return to html_entity_decode3.phpt CVS log

Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / php / ext / standard / tests / strings

Annotation of embedaddon/php/ext/standard/tests/strings/html_entity_decode3.phpt, revision 1.1.1.1

1.1       misho       1: --TEST--
                      2: html_entity_decode: Do not decode numerical entities that refer to non-SGML or otherwise disallowed chars
                      3: --FILE--
                      4: <?php
                      5: 
                      6: $tests = array(
                      7:     "&#0;", //C0
                      8:     "&#1;",
                      9:     "&#x09;",
                     10:     "&#x0A;",
                     11:     "&#x0B;",
                     12:     "&#x0C;",
                     13:     "&#x0D;", //note that HTML5 is unique in that it forbids this entity, but allows a literal U+0D
                     14:     "&#x0E;",
                     15:     "&#x1F;",
                     16:     "&#x20;", //allowed always
                     17:     "&#x7F;", //DEL
                     18:     "&#x80;", //C1
                     19:     "&#x9F;",
                     20:     "&#xA0;", //allowed always
                     21:     "&#xD7FF;", //surrogates
                     22:     "&#xD800;",
                     23:     "&#xDFFF;",
                     24:     "&#xE000;", //allowed always
                     25:     "&#xFFFE;", //nonchar
                     26:     "&#xFFFF;",
                     27:     "&#xFDCF;", //allowed always
                     28:     "&#xFDD0;", //nonchar
                     29:     "&#xFDEF;",
                     30:     "&#xFDF0;", //allowed always
                     31:     "&#x2FFFE;", //nonchar
                     32:     "&#x2FFFF;",
                     33: );
                     34: 
                     35: echo "*** HTML 4.01  ***\n";
                     36: 
                     37: foreach ($tests as $t) {
                     38:     $dec = html_entity_decode($t, ENT_QUOTES | ENT_HTML401, "UTF-8");
                     39:     if ($t == $dec) {
                     40:         echo "$t\tNOT DECODED\n";
                     41:     } else {
                     42:         echo "$t\tDECODED\n";
                     43:     }
                     44: }
                     45: 
                     46: echo "\n*** XHTML 1.0  ***\n";
                     47: 
                     48: foreach ($tests as $t) {
                     49:     $dec = html_entity_decode($t, ENT_QUOTES | ENT_XHTML, "UTF-8");
                     50:     if ($t == $dec) {
                     51:         echo "$t\tNOT DECODED\n";
                     52:     } else {
                     53:         echo "$t\tDECODED\n";
                     54:     }
                     55: }
                     56: 
                     57: echo "\n*** HTML5  ***\n";
                     58: 
                     59: foreach ($tests as $t) {
                     60:     $dec = html_entity_decode($t, ENT_QUOTES | ENT_HTML5, "UTF-8");
                     61:     if ($t == $dec) {
                     62:         echo "$t\tNOT DECODED\n";
                     63:     } else {
                     64:         echo "$t\tDECODED\n";
                     65:     }
                     66: }
                     67: 
                     68: echo "\n*** XML 1.0  ***\n";
                     69: 
                     70: foreach ($tests as $t) {
                     71:     $dec = html_entity_decode($t, ENT_QUOTES | ENT_XML1, "UTF-8");
                     72:     if ($t == $dec) {
                     73:         echo "$t\tNOT DECODED\n";
                     74:     } else {
                     75:         echo "$t\tDECODED\n";
                     76:     }
                     77: }
                     78: 
                     79: echo "\nDone.\n";
                     80: --EXPECT--
                     81: *** HTML 4.01  ***
                     82: &#0;   NOT DECODED
                     83: &#1;   NOT DECODED
                     84: &#x09; DECODED
                     85: &#x0A; DECODED
                     86: &#x0B; NOT DECODED
                     87: &#x0C; NOT DECODED
                     88: &#x0D; DECODED
                     89: &#x0E; NOT DECODED
                     90: &#x1F; NOT DECODED
                     91: &#x20; DECODED
                     92: &#x7F; NOT DECODED
                     93: &#x80; NOT DECODED
                     94: &#x9F; NOT DECODED
                     95: &#xA0; DECODED
                     96: &#xD7FF;       DECODED
                     97: &#xD800;       NOT DECODED
                     98: &#xDFFF;       NOT DECODED
                     99: &#xE000;       DECODED
                    100: &#xFFFE;       DECODED
                    101: &#xFFFF;       DECODED
                    102: &#xFDCF;       DECODED
                    103: &#xFDD0;       DECODED
                    104: &#xFDEF;       DECODED
                    105: &#xFDF0;       DECODED
                    106: &#x2FFFE;      DECODED
                    107: &#x2FFFF;      DECODED
                    108: 
                    109: *** XHTML 1.0  ***
                    110: &#0;   NOT DECODED
                    111: &#1;   NOT DECODED
                    112: &#x09; DECODED
                    113: &#x0A; DECODED
                    114: &#x0B; NOT DECODED
                    115: &#x0C; NOT DECODED
                    116: &#x0D; DECODED
                    117: &#x0E; NOT DECODED
                    118: &#x1F; NOT DECODED
                    119: &#x20; DECODED
                    120: &#x7F; DECODED
                    121: &#x80; DECODED
                    122: &#x9F; DECODED
                    123: &#xA0; DECODED
                    124: &#xD7FF;       DECODED
                    125: &#xD800;       NOT DECODED
                    126: &#xDFFF;       NOT DECODED
                    127: &#xE000;       DECODED
                    128: &#xFFFE;       NOT DECODED
                    129: &#xFFFF;       NOT DECODED
                    130: &#xFDCF;       DECODED
                    131: &#xFDD0;       DECODED
                    132: &#xFDEF;       DECODED
                    133: &#xFDF0;       DECODED
                    134: &#x2FFFE;      DECODED
                    135: &#x2FFFF;      DECODED
                    136: 
                    137: *** HTML5  ***
                    138: &#0;   NOT DECODED
                    139: &#1;   NOT DECODED
                    140: &#x09; DECODED
                    141: &#x0A; DECODED
                    142: &#x0B; NOT DECODED
                    143: &#x0C; DECODED
                    144: &#x0D; NOT DECODED
                    145: &#x0E; NOT DECODED
                    146: &#x1F; NOT DECODED
                    147: &#x20; DECODED
                    148: &#x7F; NOT DECODED
                    149: &#x80; NOT DECODED
                    150: &#x9F; NOT DECODED
                    151: &#xA0; DECODED
                    152: &#xD7FF;       DECODED
                    153: &#xD800;       NOT DECODED
                    154: &#xDFFF;       NOT DECODED
                    155: &#xE000;       DECODED
                    156: &#xFFFE;       NOT DECODED
                    157: &#xFFFF;       NOT DECODED
                    158: &#xFDCF;       DECODED
                    159: &#xFDD0;       NOT DECODED
                    160: &#xFDEF;       NOT DECODED
                    161: &#xFDF0;       DECODED
                    162: &#x2FFFE;      NOT DECODED
                    163: &#x2FFFF;      NOT DECODED
                    164: 
                    165: *** XML 1.0  ***
                    166: &#0;   NOT DECODED
                    167: &#1;   NOT DECODED
                    168: &#x09; DECODED
                    169: &#x0A; DECODED
                    170: &#x0B; NOT DECODED
                    171: &#x0C; NOT DECODED
                    172: &#x0D; DECODED
                    173: &#x0E; NOT DECODED
                    174: &#x1F; NOT DECODED
                    175: &#x20; DECODED
                    176: &#x7F; DECODED
                    177: &#x80; DECODED
                    178: &#x9F; DECODED
                    179: &#xA0; DECODED
                    180: &#xD7FF;       DECODED
                    181: &#xD800;       NOT DECODED
                    182: &#xDFFF;       NOT DECODED
                    183: &#xE000;       DECODED
                    184: &#xFFFE;       NOT DECODED
                    185: &#xFFFF;       NOT DECODED
                    186: &#xFDCF;       DECODED
                    187: &#xFDD0;       DECODED
                    188: &#xFDEF;       DECODED
                    189: &#xFDF0;       DECODED
                    190: &#x2FFFE;      DECODED
                    191: &#x2FFFF;      DECODED
                    192: 
                    193: Done.