tor-browser

The Tor Browser
git clone https://git.dasho.dev/tor-browser.git
Log | Files | Refs | README | LICENSE

test_unicode-url.js (7572B)

      1 /* Any copyright is dedicated to the Public Domain.
      2   http://creativecommons.org/publicdomain/zero/1.0/ */
      3 
      4 "use strict";
      5 
      6 /**
      7 * Tests utility functions contained in `unicode-url.js`
      8 */
      9 
     10 const {
     11  getUnicodeUrl,
     12  getUnicodeUrlPath,
     13  getUnicodeHostname,
     14 } = require("resource://devtools/client/shared/unicode-url.js");
     15 
     16 // List of URLs used to test Unicode URL conversion
     17 const TEST_URLS = [
     18  // Type:     Readable ASCII URLs
     19  // Expected: All of Unicode versions should equal to the raw.
     20  {
     21    raw: "https://example.org",
     22    expectedUnicode: "https://example.org",
     23  },
     24  {
     25    raw: "http://example.org",
     26    expectedUnicode: "http://example.org",
     27  },
     28  {
     29    raw: "ftp://example.org",
     30    expectedUnicode: "ftp://example.org",
     31  },
     32  {
     33    raw: "https://example.org.",
     34    expectedUnicode: "https://example.org.",
     35  },
     36  {
     37    raw: "https://example.org/",
     38    expectedUnicode: "https://example.org/",
     39  },
     40  {
     41    raw: "https://example.org/test",
     42    expectedUnicode: "https://example.org/test",
     43  },
     44  {
     45    raw: "https://example.org/test.html",
     46    expectedUnicode: "https://example.org/test.html",
     47  },
     48  {
     49    raw: "https://example.org/test.html?one=1&two=2",
     50    expectedUnicode: "https://example.org/test.html?one=1&two=2",
     51  },
     52  {
     53    raw: "https://example.org/test.html#here",
     54    expectedUnicode: "https://example.org/test.html#here",
     55  },
     56  {
     57    raw: "https://example.org/test.html?one=1&two=2#here",
     58    expectedUnicode: "https://example.org/test.html?one=1&two=2#here",
     59  },
     60  // Type:     Unreadable URLs with either Punycode domain names or URI-encoded
     61  //           paths
     62  // Expected: Unreadable domain names and URI-encoded paths should be converted
     63  //           to readable Unicode.
     64  {
     65    raw: "https://xn--g6w.xn--8pv/test.html",
     66    // Do not type Unicode characters directly, because this test file isn't
     67    // specified with a known encoding.
     68    expectedUnicode: "https://\u6e2c.\u672c/test.html",
     69  },
     70  {
     71    raw: "https://example.org/%E6%B8%AC%E8%A9%A6.html",
     72    // Do not type Unicode characters directly, because this test file isn't
     73    // specified with a known encoding.
     74    expectedUnicode: "https://example.org/\u6e2c\u8a66.html",
     75  },
     76  {
     77    raw: "https://example.org/test.html?One=%E4%B8%80",
     78    // Do not type Unicode characters directly, because this test file isn't
     79    // specified with a known encoding.
     80    expectedUnicode: "https://example.org/test.html?One=\u4e00",
     81  },
     82  {
     83    raw: "https://example.org/test.html?%E4%B8%80=1",
     84    // Do not type Unicode characters directly, because this test file isn't
     85    // specified with a known encoding.
     86    expectedUnicode: "https://example.org/test.html?\u4e00=1",
     87  },
     88  {
     89    raw:
     90      "https://xn--g6w.xn--8pv/%E6%B8%AC%E8%A9%A6.html" +
     91      "?%E4%B8%80=%E4%B8%80" +
     92      "#%E6%AD%A4",
     93    // Do not type Unicode characters directly, because this test file isn't
     94    // specified with a known encoding.
     95    expectedUnicode:
     96      "https://\u6e2c.\u672c/\u6e2c\u8a66.html" + "?\u4e00=\u4e00" + "#\u6b64",
     97  },
     98  // Type:     data: URIs
     99  // Expected: All should not be converted.
    100  {
    101    raw: "data:text/plain;charset=UTF-8;Hello%20world",
    102    expectedUnicode: "data:text/plain;charset=UTF-8;Hello%20world",
    103  },
    104  {
    105    raw: "data:text/plain;charset=UTF-8;%E6%B8%AC%20%E8%A9%A6",
    106    expectedUnicode: "data:text/plain;charset=UTF-8;%E6%B8%AC%20%E8%A9%A6",
    107  },
    108  {
    109    raw:
    110      "data:image/png;base64,iVBORw0KGgoAAA" +
    111      "ANSUhEUgAAAAUAAAAFCAYAAACNbyblAAAAHElEQVQI12P4" +
    112      "//8/w38GIAXDIBKE0DHxgljNBAAO9TXL0Y4OHwAAAABJRU" +
    113      "5ErkJggg==",
    114    expectedUnicode:
    115      "data:image/png;base64,iVBORw0KGgoAAA" +
    116      "ANSUhEUgAAAAUAAAAFCAYAAACNbyblAAAAHElEQVQI12P4" +
    117      "//8/w38GIAXDIBKE0DHxgljNBAAO9TXL0Y4OHwAAAABJRU" +
    118      "5ErkJggg==",
    119  },
    120  // Type:     Malformed URLs
    121  // Expected: All should not be converted.
    122  {
    123    raw: "://example.org/test",
    124    expectedUnicode: "://example.org/test",
    125  },
    126  {
    127    raw: "://xn--g6w.xn--8pv/%E6%B8%AC%E8%A9%A6.html" + "?%E4%B8%80=%E4%B8%80",
    128    expectedUnicode:
    129      "://xn--g6w.xn--8pv/%E6%B8%AC%E8%A9%A6.html" + "?%E4%B8%80=%E4%B8%80",
    130  },
    131  {
    132    // %E8%A9 isn't a valid UTF-8 code, so this URL is malformed.
    133    raw: "https://xn--g6w.xn--8pv/%E6%B8%AC%E8%A9",
    134    expectedUnicode: "https://xn--g6w.xn--8pv/%E6%B8%AC%E8%A9",
    135  },
    136 ];
    137 
    138 // List of hostanmes used to test Unicode hostname conversion
    139 const TEST_HOSTNAMES = [
    140  // Type:     Readable ASCII hostnames
    141  // Expected: All of Unicode versions should equal to the raw.
    142  {
    143    raw: "example",
    144    expectedUnicode: "example",
    145  },
    146  {
    147    raw: "example.org",
    148    expectedUnicode: "example.org",
    149  },
    150  // Type:     Unreadable Punycode hostnames
    151  // Expected: Punycode should be converted to readable Unicode.
    152  {
    153    raw: "xn--g6w",
    154    // Do not type Unicode characters directly, because this test file isn't
    155    // specified with a known encoding.
    156    expectedUnicode: "\u6e2c",
    157  },
    158  {
    159    raw: "xn--g6w.xn--8pv",
    160    // Do not type Unicode characters directly, because this test file isn't
    161    // specified with a known encoding.
    162    expectedUnicode: "\u6e2c.\u672c",
    163  },
    164 ];
    165 
    166 // List of URL paths used to test Unicode URL path conversion
    167 const TEST_URL_PATHS = [
    168  // Type:     Readable ASCII URL paths
    169  // Expected: All of Unicode versions should equal to the raw.
    170  {
    171    raw: "test",
    172    expectedUnicode: "test",
    173  },
    174  {
    175    raw: "/",
    176    expectedUnicode: "/",
    177  },
    178  {
    179    raw: "/test",
    180    expectedUnicode: "/test",
    181  },
    182  {
    183    raw: "/test.html?one=1&two=2#here",
    184    expectedUnicode: "/test.html?one=1&two=2#here",
    185  },
    186  // Type:     Unreadable URI-encoded URL paths
    187  // Expected: URL paths should be converted to readable Unicode.
    188  {
    189    raw: "/%E6%B8%AC%E8%A9%A6",
    190    // Do not type Unicode characters directly, because this test file isn't
    191    // specified with a known encoding.
    192    expectedUnicode: "/\u6e2c\u8a66",
    193  },
    194  {
    195    raw: "/%E6%B8%AC%E8%A9%A6.html",
    196    // Do not type Unicode characters directly, because this test file isn't
    197    // specified with a known encoding.
    198    expectedUnicode: "/\u6e2c\u8a66.html",
    199  },
    200  {
    201    raw:
    202      "/%E6%B8%AC%E8%A9%A6.html" +
    203      "?%E4%B8%80=%E4%B8%80&%E4%BA%8C=%E4%BA%8C" +
    204      "#%E6%AD%A4",
    205    // Do not type Unicode characters directly, because this test file isn't
    206    // specified with a known encoding.
    207    expectedUnicode:
    208      "/\u6e2c\u8a66.html" + "?\u4e00=\u4e00&\u4e8c=\u4e8c" + "#\u6b64",
    209  },
    210  // Type:     Malformed URL paths
    211  // Expected: All should not be converted.
    212  {
    213    // %E8%A9 isn't a valid UTF-8 code, so this URL is malformed.
    214    raw: "/%E6%B8%AC%E8%A9",
    215    expectedUnicode: "/%E6%B8%AC%E8%A9",
    216  },
    217 ];
    218 
    219 function run_test() {
    220  // Test URLs
    221  for (const url of TEST_URLS) {
    222    const result = getUnicodeUrl(url.raw);
    223    equal(
    224      result,
    225      url.expectedUnicode,
    226      "Test getUnicodeUrl: " +
    227        url.raw +
    228        " should be unicodized to " +
    229        url.expectedUnicode
    230    );
    231  }
    232 
    233  // Test hostnames
    234  for (const hostname of TEST_HOSTNAMES) {
    235    const result = getUnicodeHostname(hostname.raw);
    236    equal(
    237      result,
    238      hostname.expectedUnicode,
    239      "Test getUnicodeHostname: " +
    240        hostname.raw +
    241        " should be unicodized to " +
    242        hostname.expectedUnicode
    243    );
    244  }
    245 
    246  // Test URL paths
    247  for (const urlPath of TEST_URL_PATHS) {
    248    const result = getUnicodeUrlPath(urlPath.raw);
    249    equal(
    250      result,
    251      urlPath.expectedUnicode,
    252      "Test getUnicodeUrlPath: " +
    253        urlPath.raw +
    254        " should be unicodized to " +
    255        urlPath.expectedUnicode
    256    );
    257  }
    258 }