On 07/28/2014 08:28 PM, Aline Manera wrote:
>
> On 07/28/2014 06:15 AM, Sheldon wrote:
>> On 07/26/2014 05:01 AM, alinefm(a)linux.vnet.ibm.com wrote:
>>> From: Aline Manera <alinefm(a)linux.vnet.ibm.com>
>>>
>>> >From python documentation, base64.urlsafe_b64encode(s) substitutes
>>> - instead
>>> of + and _ instead of / in the standard Base64 alphabet, BUT the
>>> result can
>>> still contain = which is not safe in a URL query component.
>>> As token value is not decoded nowhere, replace = by A
>>
>> what about other character instead of A? such as "." or "~"
>>
>> This is the base64 alphabet:
>> 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/='
>>
>> "A" is in the alphabet.
>>
>
> Ok. I will use . to replace =
>
>>
>> or do not pad the encode string.
>> some base64 variant has no pad character.
>>
http://en.wikipedia.org/wiki/Base64#Implementations_and_history
>>
>
> I think we can have problem in not using the pad if you have the same
> code string to different sentence
> Example:
>
> "abcd" = "YWJjZA="
> "jdcf" = "YWJjZA"
in python, I have try:
In [51]: base64.urlsafe_b64encode("jdcf")
Out[51]: 'amRjZg=='
In [52]: base64.urlsafe_b64encode("abcd")
Out[52]: 'YWJjZA=='
and in UI I also try:
atob("YWJjZA")
"abcd"
atob("YWJjZA=")
[Exception... "String contains an invalid character" code: "5"
nsresult: "0x80530005 (InvalidCharacterError)" location:
"<unknown>"]
atob("YWJjZA==")
"abcd"
Ok. I will remove the padding encode
>
> If we ignore the = in the first string we will have 2 matches to the
> different values.
> Not sure it can happen in real world, but I think it is safe to use .
> instead of =
>
>>
>> Then in python, it can easy strip the "="
>> In [21]: s1 = base64.urlsafe_b64encode("abcd")
>> In [22]: s1.rstrip("=")
>> Out[22]: 'YWJjZA'
>>
>> and In js add a funtion to
>> $ git diff
>> diff --git a/ui/js/src/kimchi.utils.js b/ui/js/src/kimchi.utils.js
>> index 480b9b5..2d252a7 100644
>> --- a/ui/js/src/kimchi.utils.js
>> +++ b/ui/js/src/kimchi.utils.js
>> @@ -191,3 +191,8 @@ kimchi.urlSafeB64Decode = function(str) {
>> kimchi.urlSafeB64Encode = function(str) {
>> return btoa(str).replace(/\+/g, '-').replace(/\//g, '_');
>> }
>> +
>> +kimchi.padBase64 = function(str) {
>> + padLen = str.length % 4;
>> + return str + new Array(padLen? 4 - padLen + 1 : 0).join("=");
>> +}
>>
>>
>> Now test this method:
>> in python:
>> In [40]: base64.urlsafe_b64encode("abcd")
>> Out[40]: 'YWJjZA=='
>>
>> In [41]: base64.urlsafe_b64encode("abcde")
>> Out[41]: 'YWJjZGU='
>>
>> In UI:
>> kimchi.padBase64("YWJjZA")
>> "YWJjZA=="
>>
>> kimchi.padBase64("YWJjZGU")
>> "YWJjZGU="
>>
>>
>>>
>>> The problem with equals sign was only identified on Spice connections.
>>> noVNC can deal well with that.
>>>
>>> For reference:
https://docs.python.org/2/library/base64.html
>>>
>>> Signed-off-by: Aline Manera <alinefm(a)linux.vnet.ibm.com>
>>> ---
>>> src/kimchi/vnc.py | 9 ++++++++-
>>> ui/js/src/kimchi.api.js | 18 ++++++++++++++++--
>>> 2 files changed, 24 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/src/kimchi/vnc.py b/src/kimchi/vnc.py
>>> index 9380e21..4159049 100644
>>> --- a/src/kimchi/vnc.py
>>> +++ b/src/kimchi/vnc.py
>>> @@ -54,7 +54,14 @@ def new_ws_proxy():
>>>
>>> def add_proxy_token(name, port):
>>> with open(os.path.join(WS_TOKENS_DIR, name), 'w') as f:
>>> - name = base64.urlsafe_b64encode(name)
>>> + """
>>> + From python documentation base64.urlsafe_b64encode(s)
>>> + substitutes - instead of + and _ instead of / in the
>>> + standard Base64 alphabet, BUT the result can still
>>> + contain = which is not safe in a URL query component.
>>> + As token value is not decoded nowhere, replace = by A
>>> + """
>>> + name = base64.urlsafe_b64encode(name).replace('=',
'A')
>>> f.write('%s: localhost:%s' % (name.encode('utf-8'),
port))
>>>
>>>
>>> diff --git a/ui/js/src/kimchi.api.js b/ui/js/src/kimchi.api.js
>>> index 8f5b68f..30360c5 100644
>>> --- a/ui/js/src/kimchi.api.js
>>> +++ b/ui/js/src/kimchi.api.js
>>> @@ -352,7 +352,14 @@ var kimchi = {
>>> }).done(function() {
>>> url = 'https://' + location.hostname + ':'
+
>>> proxy_port;
>>> url +=
"/console.html?url=vnc_auto.html&port=" +
>>> proxy_port;
>>> - url += "&path=?token=" +
kimchi.urlSafeB64Encode(vm);
>>> + /*
>>> + * From python documentation
>>> base64.urlsafe_b64encode(s)
>>> + * substitutes - instead of + and _ instead of /
>>> in the
>>> + * standard Base64 alphabet, BUT the result can still
>>> + * contain = which is not safe in a URL query
>>> component.
>>> + * As token value is not decoded nowhere, replace
>>> = by A
>>> + * */
>>> + url += "&path=?token=" +
>>> kimchi.urlSafeB64Encode(vm).replace(/=/g, 'A');
>>> url += "&kimchi=" + location.port;
>>> url += '&encrypt=1';
>>> window.open(url);
>>> @@ -377,7 +384,14 @@ var kimchi = {
>>> url = 'https://' + location.hostname + ':'
+
>>> proxy_port;
>>> url += "/console.html?url=spice.html&port=" +
>>> proxy_port;
>>> url += "&listen=" + location.hostname;
>>> - url += "&token=" +
kimchi.urlSafeB64Encode(vm);
>>> + /*
>>> + * From python documentation
>>> base64.urlsafe_b64encode(s)
>>> + * substitutes - instead of + and _ instead of /
>>> in the
>>> + * standard Base64 alphabet, BUT the result can still
>>> + * contain = which is not safe in a URL query
>>> component.
>>> + * As token value is not decoded nowhere, replace
>>> = by A
>>> + * */
>>> + url += "&token=" +
>>> kimchi.urlSafeB64Encode(vm).replace(/=/g, 'A');
>>> url += "&kimchi=" + location.port;
>>> url += '&encrypt=1';
>>> window.open(url);
>>
>>
>
>
>