[Cuis-dev] [DEFECT] Can't zip Cuis' sources file

[Gerald Klix]

Hi Juan,

thanks a lot for your efforts.
I will test it as soon as I have checked in my latest work.

Did I get this right, this sample text ('ლორემ იფსუმ დოლორ')
is Georgian, isn't it? I wonder what it means.


Best Regards,

Gerald



On 11/23/23 3:17 PM, Juan Vuletich wrote:
> Hi Gerald,
>
> The simple fact is that ZIP files don't store "strings", they only 
> store bytes. There were a few places in the Compression.pck.st that 
> still made the old assumption that Strings and ByteArrays were "the 
> same". I fixed those I found. I also added #addBytes:as:, that is what 
> you usually want.
>
> Now, the following examples work:
>
> ZipArchive new
>     zipFileComment: 'ZIP file test';
>     addString: 'ლორემ იფსუმ დოლორ' as: 'someText.bin';
>     writeToFileNamed: 'someText.zip'.
>
> ZipArchive new
>     zipFileComment: 'ZIP file test';
>     addBytes: 'ლორემ იფსუმ დოლორ' asUtf8Bytes as: 'someTextBytes2.bin';
>     writeToFileNamed: 'someTextBytes2.zip'.
>
> Additionally, this example also works:
>
> ZipArchive new
>     zipFileComment: 'ZIP file test';
>     addFile: 
> 'Cuis-Smalltalk-Dev/CoreUpdates/5922-FixedClassMethodPushDown-TomásSpognardi-2023Jul09-20h38m-TS.001.cs.st' 
> as: 'testOutx.bin';
>     writeToFileNamed: 'testbinx.zip'.
>
> So, I guess the problem you had with this file got also fixed along 
> the way.
>
> All 3 zip files seem to unzip properly on my Mac.
>
> Please test again, and keep reporting further problems.
>
> Cheers,
>
> On 11/22/2023 8:13 AM, Gerald Klix via Cuis-dev wrote:
>> Hi Juan,
>>
>> I got tried to work around this problem by reading the
>> license files in binary mode and converting them to a string.
>> However this leads to as CRC error:
>>
>>  extracting: LICENSES.txt             bad CRC 46294e2b  (should be 
>> c42f8133)
>>
>>
>> As promised I skimmed the ZIP file spec at 
>> https://pkware.cachefly.net/webdocs/casestudies/APPNOTE.TXT
>> I only saw provisions for UTF-8 file names.
>> Indeed we have problems storing such files; adding
>> this one 
>> https://github.com/Cuis-Smalltalk/Cuis-Smalltalk-Dev/blob/master/CoreUpdates/5922-FixedClassMethodPushDown-Tom%C3%A1sSpognardi-2023Jul09-20h38m-TS.001.cs.st
>> yields the following error when listing the contents withunzip -l
>>
>>      1195  2023-07-19 20:39 
>> Image/CoreUpdates/5922-FixedClassMethodPushDown-TomásSpognardi-2023Jul09-20h38m-TS.001.cs.s
>> error:  expected central file header signature not found (file #459).
>>   (please check that you have transferred or created the zipfile in the
>>
>>
>> *The biggest problem however is this one:
>>
>> *Execute the following code snippet:
>>
>> 'testIn.bin' asFileEntry forceWriteStreamDo: [ :ws |
>>     ws binary.
>>     ws nextPutAll: #[16rF1 16rF7] ].
>>
>>
>> ZipArchive new
>>     zipFileComment: 'ZIP file test';
>>     addFile: 'testIn.bin' as: 'testOut.bin';
>>     writeToFileNamed: 'testbin.zip'
>>
>> Then unzip the file with:
>>
>> bear at speedy ~/s/c/Environments> unzip testbin.zip
>> Archive:  testbin.zip
>> ZIP file test
>>   inflating: testOut.bin
>>
>> This yields the following:
>>
>> bear at speedy ~/s/c/Environments> ls -l test*.bin
>> -rw-r--r-- 1 bear bear 2 Nov 22 12:01 testIn.bin
>> -rw-rw-rw- 1 bear bear 4 Nov 22 12:01 testOut.bin
>>
>> Invoking hexdump on both files shows a "slight" difference:
>>
>> bear at speedy ~/s/c/Environments> hexdump testIn.bin
>> 0000000 f7f1
>> 0000002
>> bear at speedy ~/s/c/Environments> hexdump testOut.bin
>> 0000000 b1c3 b7c3
>> 0000004
>>
>> Obviously some UTF8 conversion happened.
>> Needless to say that ZipArchive can't be used to compress binary
>> files and thus I can't create a distribution of Haver
>> with my current tool-set.
>>
>>
>> Best Regards,
>>
>> Gerald
>>
>>
>>
>> On 11/22/23 10:21 AM, Gerald Klix via Cuis-dev wrote:
>>> Thank's Juan,
>>>
>>> this works now.
>>>
>>> Alas when I tried to create a new Haver release I ran into another 
>>> problem.
>>> I also add strings to the ZIP-archive, for example a generated file 
>>> that contains
>>> all the contents of all the license files found. The resulting 
>>> string contains
>>> Unicode code-points, which makes the CRC computation fail.
>>>
>>> In essence when sending #updateCrc:from:to:in: the primitive fails – 
>>> for
>>> obvious reasons – and the failure code kicks in. This in turn 
>>> (indirectly) sends #bitXor:
>>> to a UnicodePoint instance, which results in a DNU exception.
>>>
>>> I will try to do some research on how UTF8 data is supposed
>>> to be stored in a ZIP file.
>>>
>>>
>>> Thank's again and Best Regards,
>>>
>>> Gerald
>>>
>>>
>>>
>>> On 11/21/23 9:12 PM, Juan Vuletich wrote:
>>>> On 11/14/2023 10:20 AM, Gerald Klix via Cuis-dev wrote:
>>>>> Hi all, Hi Juan,
>>>>>
>>>>> The following snippet does not terminate (within reasonable time, 
>>>>> 20h):
>>>>>
>>>>> (z1 := ZipArchive new) zipFileComment: 'ZIP test'.
>>>>> z1
>>>>>     addFile: SourceFiles first fileEntry as: 'f1';
>>>>>     writeToFileNamed: 'test.zip';
>>>>>     close.
>>>>>
>>>>> Adding small test files, created like this: echo test > test.txt ,
>>>>> however does work:
>>>>>
>>>>> (z3 := ZipArchive new) zipFileComment: 'ZIP test'.
>>>>> z3
>>>>>     addFile: Smalltalk imageName as: 'f3';
>>>>>     writeToFileNamed: 'test.zip';
>>>>>     close.
>>>>>
>>>>>
>>>>>
>>>>> Best Regards,
>>>>>
>>>>> Gerald
>>>>
>>>> Hi Gerald,
>>>>
>>>> I just pushed a fix to GitHub. Thanks for reporting.
>>>>
>>>> Cheers,
>>>>
>>>
>>
>
>