Safari acting weird. Not loading pages and showing some Chinese text

safari

Occasionally my Safari hangs on loading a page, and after some time shows this text:

㰡ⴭ⁌佃䅌䥚䕒区⁅慣栠汯捡汩穡扬攠灩散攠潦⁴桥⁰慧攠楳慲步搠睩瑨⁡⁣潭浥湴‭ⴾਊ㱈呍䰾਼ℭⴠ䱏䍁䱉婅剓㨠䥦⁩渠愠物杨琭瑯⵬敦琠汯捡汥Ⱐ慤搠摩爽≲瑬∠瑯⁴桥⁈呍䰠敬敭敮琮‭ⴾ਼䡅䅄㸊††㱌䥎䬠牥氽獴祬敳桥整⁴祰攽≴數琯捳猢⁨牥昽≰慧攭汯慤ⵥ牲潲献捳猢㸊††㰡ⴭ⁌佃䅌䥚䕒区⁙潵楧桴⁷慮琠瑯⁣桡湧攠瑨攠景湴⁦慭楬礮⁙潵⁣慮⁡汳漠慤搠獴祬敳⁴漠潶敲物摥⁳楺敳Ⱐ整挮‭ⴾਠ†‼協奌䔾ਠ†††⁂佄夠筦潮琭晡浩汹㨧䡥汶整楣愠乥略✻紊††㰯協奌䔾ਊ††㰡ⴭ⁌佃䅌䥚䕒区⁔桥數琠汩湥⁣潮瑡楮猠瑨攠灡来⁴楴汥⁴桡琠慰灥慲猠楮⁴桥⁷楮摯眧猠瑩瑬攠扡爠ⴭ㸊††㱔䥔䱅㻐鷐딠킲킴킰킻킾톁톏⃐닑雐듐뫑胐룑苐렠톁톂킾톀톖킽킺톃㰯呉呌䔾਼⽈䕁䐾ਊ㱂佄夾਼摩瘠捬慳猽≥牲潲ⵣ潮瑡楮敲∾਼摩瘠捬慳猽≩捯渢⁡汴㴢卡晡物⁉捯渢㸼⽤楶㸊㱤楶⁣污獳㴢瑥硴ⵣ潮瑡楮敲∾਼ℭⴠ浡楮⁴楴汥⁨敲攬⁲数敡瑥搠㌠瑩浥猠ⴭ㸊㱐⁣污獳㴢敲牯爭瑩瑬攠敲牯爭瑥硴ⵥ湧牡癩湧∾╀‼䄠楤㴢桥汰ⵢ畴瑯渢㸼⽡㸼⽐㸊㱐⁣污獳㴢敲牯爭瑩瑬攠敲牯爭瑥硴∾╀‼䄠楤㴢桥汰ⵢ畴瑯渢㸼⽡㸼⽐㸊㱐⁣污獳㴢敲牯爭瑩瑬攠敲牯爭瑥硴⵩湮敲⵳桡摯眢㸥䀠㱁⁩搽≨敬瀭扵瑴潮∠捬慳猽∥䀢⁈剅䘽❯灥渭桥汰ⵡ湣桯爺╀✾㰯愾㰯倾਼⽤楶㸊㱤楶⁣污獳㴢瑥硴ⵣ潮瑡楮敲∾਼ℭⴠ敲牯爠浥獳慧攠桥牥Ⱐ牥灥慴敤″⁴業敳‭ⴾ਼倠捬慳猽≥牲潲⵭敳獡来⁥牲潲⵴數琭敮杲慶楮朢㸥䀼⽐㸊㱐⁣污獳㴢敲牯爭浥獳慧攠敲牯爭瑥硴∾╀㰯倾਼倠捬慳猽≥牲潲⵭敳獡来⁥牲潲⵴數琭楮湥爭獨慤潷∾╀㰯倾਼⽤楶㸊㰯摩瘾਼⽂佄夾ਊ㰯䡔䵌㸊

Maybe someone knows what that text is? Or how to solve a problem at all?

I'll be very grateful for reply.

P.S. Also when some page hangs while loading and I hit refresh, it usually loads instantly. So it's now a problem with an internet connection.

Best Answer

The characters you posted are what you get if you take the text at the bottom of this post, encode it as UTF-8, and then (mistakenly) decode it as UTF-16BE.

Ultimately, Safari is decoding the data incorrectly (it should be decoded as UTF-8, but it is being decoded as UTF-16BE). It might do this for any one of several reasons:

  • The server incorrectly told Safari to decode as UTF-16BE (e.g. via a charset parameter in the Content-Type header of the server’s HTTP response).
  • The server incorrectly prepended a UTF-16BE BOM before the UTF-8 data.
  • Safari is configured to default to UTF-16BE.
    I am not sure how this would be actually be accomplished though. The only UTF entry in the manual encoding list† is “Unicode (UTF-8)”, which would seem to preclude forcing UTF-16BE (unlike TextEdit where you can actually select UTF-16BE, if you customize its list of encodings).
  • Safari auto-detected the encoding incorrectly.

Your particular problem seems likely to be a server (or proxy‡) problem since you say that it is only happening intermittently.

The Network tab of Safari’s Web Inspector can show you the server response headers (to confirm or deny an erroneous charset parameter), but it would be probably be annoying to always have to remember to open/enable it for each tab you open (to capture the information it must already be active before when the page of interest is loaded—which may be difficult to predict if your problem is intermittent).

A packet capture could identify an erroneous BOM (and/or an erroneous charset), but such tools are much less convenient (also usually useless for encrypted (HTTPS) requests). If you know that the problem occurs fairly often, you might be able to send many requests through curl and try to identify erroneous BOMs/charsets by looking at the data it can report (server response headers and the raw content bytes).

†You can manually choose an encoding (in Safari 5.1) via the View menu’s Text Encoding submenu.
‡Such a proxy might be a “transparent” proxy that you might not otherwise suspect.

<!-- LOCALIZERS: Each localizable piece of the page is marked with a comment -->

<HTML>
<!-- LOCALIZERS: If in a right-to-left locale, add dir="rtl" to the HTML element. -->
<HEAD>
    <LINK rel=stylesheet type="text/css" href="page-load-errors.css">
    <!-- LOCALIZERS: You might want to change the font family. You can also add styles to override sizes, etc. -->
    <STYLE>
        BODY {font-family:'Helvetica Neue';}
    </STYLE>

    <!-- LOCALIZERS: The next line contains the page title that appears in the window's title bar -->
    <TITLE>Не вдалося відкрити сторінку</TITLE>
</HEAD>

<BODY>
<div class="error-container">
<div class="icon" alt="Safari Icon"></div>
<div class="text-container">
<!-- main title here, repeated 3 times -->
<P class="error-title error-text-engraving">%@ <A id="help-button"></a></P>
<P class="error-title error-text">%@ <A id="help-button"></a></P>
<P class="error-title error-text-inner-shadow">%@ <A id="help-button" class="%@" HREF='open-help-anchor:%@'></a></P>
</div>
<div class="text-container">
<!-- error message here, repeated 3 times -->
<P class="error-message error-text-engraving">%@</P>
<P class="error-message error-text">%@</P>
<P class="error-message error-text-inner-shadow">%@</P>
</div>
</div>
</BODY>

</HTML>

Google translation suggests that the Cyrillic <title> text seems to be Ukrainian for “Could not open page”.

The comments indicated that this text only started showing up after a software update from Apple. The exact update was probably the 10.7.3 update which added localization for several languages, including Ukrainian.

I do not have a 10.7 system, but the above content (apart from the Ukrainian <title> text) is largely identical to the content of files from 10.6 (Safari 5.1.3) with names like /Applications/Safari.app/Contents/Resources/*.lproj/StandardErrorPage.html; all these files (on 10.6 Safari 5.1.3) are UTF-16LE encoded with a leading (UTF-16LE) BOM. This points strongly to the second possible reason I described: the “server” (actually just a local file) is supplying content with an incorrect BOM.

If you can identify the file used for your particular localization, you can probably “fix” it. As a guess, the Ukrainian file is probably something like …/ua.lproj/StandardErrorPage.html.

Note: Modifying this file might break Safari’s “code signature”. On 10.6, the English version of this file is listed in the …/Safari.app/Contents/_CodeSignature/CodeResources file as “optional”, so it might be okay (I have not tried it). If you do try editing/replacing the file, make a backup first! To be safe, you could make a copy of the whole Safari.app folder.

You can use a hex editor (e.g. HexFiend) to adjust the leading BOM bytes. If you are actually dealing with a file that starts with (hex) FE FF (UTF-16BE BOM) and is followed by UTF-8 encoded data (i.e. the following bytes are for <!-- LOCALIZERS: without NUL (hex 00) bytes between the characters), then you can probably just delete the first two bytes (Safari should then be able to auto-detect the content as UTF-8). Alternatively you could replace the leading, two-byte sequence FE FF with the three-byte sequence EF BB BF (i.e. the UTF-8 encoding of U+FEFF).

You can practice on a copy of the file if you want to try out the result before replacing the file that Safari actually uses. Just copy the file to your desktop (or wherever), hex edit that copy, then open it with Safari. After making the correct modifications (and reloading the page) you should see the Ukrainian title in the title bar and several lines of %@ in the page itself (instead of the old page of Chinese/Korean/etc. characters).

Hopefully, Apple will fix all this in some new Safari/Lion update.

Related Question