dicomweb json on metadata malforfated json Error

[hakancicektr]

Hello i were testing 1.5.0c
i found issue with dicomweb json on metadata there is problem with creating json response.
json response is invalid. there some values lost and json format damaged.
when i tried with 2 images study . its working (OHIF)
but if image more than 2 images json is malformated. OHIF throw Sopclass error.

here is sample.
"0020000D":{"vr":"UI","Value":["1.2.},"000.7.572041167.76106034234"]},"" malformated part
"0020000E":{"vr":"UI","Value":["1.3.46.670589.11.72012.5.0.3812.2017031407141108960"]}
,"00200011":{"vr":"IS","Value":[101]},
"00200012":{"vr":"IS","Value":[1]}
,"00200013":{"vr":"IS","Value":[9]

looks like sprintf mixing while putting values. i can provide more samples.

i am using centos 7 i couldnt complie servertask due compile errors. i tried to dgate rather it still work fine.

Problem looks like on dgate.cpp

static int luaserialize(lua_State *L) function.

[marcelvanherk]

Hi can you share the images where it goes wrong? I assume you are using the released 150c

…

On Wed, 7 Sep 2022, 03:25 hakancicektr, ***@***.***> wrote: Hello i were testing 1.5.0c i found issue with dicomweb json on metadata there is problem with creating json response. json response is invalid. there some values lost and json format damaged. when i tried with 2 images study . its working (OHIF) but if image more than 2 images json is malformated. OHIF throw Sopclass error. here is sample. "0020000D":{"vr":"UI","Value":["1.2.},"000.7.572041167.76106034234"]},"" malformated part "0020000E":{"vr":"UI","Value":["1.3.46.670589.11.72012.5.0.3812.2017031407141108960"]} ,"00200011":{"vr":"IS","Value":[101]}, "00200012":{"vr":"IS","Value":[1]} ,"00200013":{"vr":"IS","Value":[9] looks like sprintf mixing while putting values. i can provide more samples. i am using centos 7 i couldnt complie servertask due compile errors. i tried to dgate rather it still work fine. — Reply to this email directly, view it on GitHub <#26>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAVDWJPY4KQXOEVAS6EHKQ3V474KBANCNFSM6AAAAAAQGK4VRA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[marcelvanherk]

Also interested in servertask compile errors. Marcel

…

On Wed, 7 Sep 2022, 03:25 hakancicektr, ***@***.***> wrote: Hello i were testing 1.5.0c i found issue with dicomweb json on metadata there is problem with creating json response. json response is invalid. there some values lost and json format damaged. when i tried with 2 images study . its working (OHIF) but if image more than 2 images json is malformated. OHIF throw Sopclass error. here is sample. "0020000D":{"vr":"UI","Value":["1.2.},"000.7.572041167.76106034234"]},"" malformated part "0020000E":{"vr":"UI","Value":["1.3.46.670589.11.72012.5.0.3812.2017031407141108960"]} ,"00200011":{"vr":"IS","Value":[101]}, "00200012":{"vr":"IS","Value":[1]} ,"00200013":{"vr":"IS","Value":[9] looks like sprintf mixing while putting values. i can provide more samples. i am using centos 7 i couldnt complie servertask due compile errors. i tried to dgate rather it still work fine. — Reply to this email directly, view it on GitHub <#26>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAVDWJPY4KQXOEVAS6EHKQ3V474KBANCNFSM6AAAAAAQGK4VRA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[marcelvanherk]

A quick code read shows nothing strange there (and code was tested on hundreds of studies in debian and windows). Can you give the steps to install on centos or create a Dockerfile ? I can then test it at home. Marcel

…

On Wed, Sep 7, 2022 at 3:25 AM hakancicektr ***@***.***> wrote: Hello i were testing 1.5.0c i found issue with dicomweb json on metadata there is problem with creating json response. json response is invalid. there some values lost and json format damaged. when i tried with 2 images study . its working (OHIF) but if image more than 2 images json is malformated. OHIF throw Sopclass error. here is sample. "0020000D":{"vr":"UI","Value":["1.2.},"000.7.572041167.76106034234"]},"" malformated part "0020000E":{"vr":"UI","Value":["1.3.46.670589.11.72012.5.0.3812.2017031407141108960"]} ,"00200011":{"vr":"IS","Value":[101]}, "00200012":{"vr":"IS","Value":[1]} ,"00200013":{"vr":"IS","Value":[9] looks like sprintf mixing while putting values. i can provide more samples. i am using centos 7 i couldnt complie servertask due compile errors. i tried to dgate rather it still work fine. — Reply to this email directly, view it on GitHub <#26>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAVDWJPY4KQXOEVAS6EHKQ3V474KBANCNFSM6AAAAAAQGK4VRA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[hakancicektr]

Hello,

in this part there " }, " between 1.2. and 000.7
is must not there .

"0020000D":{"vr":"UI","Value":["1.2.},"000.7.572041167.76106034234"]},""

should be correct one
"0020000D":{"vr":"UI","Value":["1.2.000.7.572041167.76106034234"]},""

i sent you files . Patient Images and Metadata
one correct metadata ( single image )
one incorrect (multi images)

i am using conquest linux 1417e with appox 1 million patients per month.

i would like to conturbate and share expriences.

If you interested you can contact me by +90 554 8848998 whatsapp or telegram (might better for fastest communication)

correct_json_metadata.txt
malformat_json_metada.txt
63791833_12.zip

[marcelvanherk]

Hi, the problem is definitively Centos specific, at least on windows the images load fine in OHIF. That does not mean it is not a bug on my side, such things may appear only with certain compilers. Can you share a Centos Dockerfile / install instructions so that I can make a test environment? I any case the effect is that 4 characters text randomly overwrite the output, may be related to socket issues. Can you also try the node.js api? Marcel Marcel

…

On Wed, Sep 7, 2022 at 10:19 AM hakancicektr ***@***.***> wrote: Hello, in this part there " }, " between 1.2. and 000.7 is must not there . "0020000D":{"vr":"UI","Value":["1.2.},"000.7.572041167.76106034234"]},"" should be correct one "0020000D":{"vr":"UI","Value":["1.2.000.7.572041167.76106034234"]},"" i sent you files . Patient Images and Metadata one correct metadata ( single image ) one incorrect (multi images) i am using conquest linux 1417e with appox 1 million patients per month. i would like to conturbate and share expriences. If you interested you can contact me by +90 554 8848998 whatsapp or telegram (might better for fastest communication) correct_json_metadata.txt <https://github.com/marcelvanherk/Conquest-DICOM-Server/files/9504642/correct_json_metadata.txt> malformat_json_metada.txt <https://github.com/marcelvanherk/Conquest-DICOM-Server/files/9504643/malformat_json_metada.txt> 63791833_12.zip <https://github.com/marcelvanherk/Conquest-DICOM-Server/files/9504655/63791833_12.zip> — Reply to this email directly, view it on GitHub <#26 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAVDWJP7OLLKPUQSJ2EDSFTV5BM2BANCNFSM6AAAAAAQGK4VRA> . You are receiving this because you commented.Message ID: ***@***.***>

[marcelvanherk]

Hi,

I think this is data corruption while transmitting from conquest to servertask. What compiler flags do you use?

Marcel

[hakancicektr]

its looks like problem with pointer usage
there is result pointer but some codes writing over it indirectly
while Vr->data formating its adding to result but there is index to where data will put. some how index moving backward and writing data to wrong position then is overwrite existing data.
that is what am susspected.
i am still analizing your codes.

[marcelvanherk]

Hi, I don't think so the same images and the same code work well on windows and debian. Can you try this: ./dgate "--lua:x=DicomObject:new();x:Read(dicomfilename);f=io.open('/tmp/x.json', 'w'); f:write(x:Serialize(true, true, false)); f:close()" This does the same serialisation but writes directly to file. Or you can modify the rquery.lua function getmetadata to write a copy of the json data to file there (returnfile is deleted and is hard to get). My guess is that the output of Serialize(true, true, false) is correct but that the issue is in socket communication. I had previous problems there. Marcel

…

On Wed, 7 Sep 2022, 21:39 hakancicektr, ***@***.***> wrote: its looks like problem with pointer usage there is result pointer but some codes writing over it indirectly while Vr->data formating its adding to result but there is index to where data will put. some how index moving backward and writing data to wrong position then is overwrite existing data. that is what am susspected. i am still analizing your codes. — Reply to this email directly, view it on GitHub <#26 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAVDWJI4EWAZQUA3KSJLFWDV5D4RJANCNFSM6AAAAAAQGK4VRA> . You are receiving this because you commented.Message ID: ***@***.***>

[marcelvanherk]

This issue appears in socket.cxx and buffer.cxx. You could roll them back to 1417d version to see what happens.

[hakancicektr]

i have changed default break size to oldvalue / 2
then all worked fine.

include "dicom.hpp"

#ifdef UNIX

define DEFAULT_BREAK_SIZE 4096

#else

define DEFAULT_BREAK_SIZE 16384

#endif

[hakancicektr]

but still there is corruption on image even i treid smaller value

[marcelvanherk]

Ah,

This is the same issue. I will send an update to socket.cxx tonight to try and fix it.

Edit: api/dicom uses a lot of internal tcp/ip communication and apparantly that is more sensitive for bugs in the socket layer. This is good, because I though they have been fixed, but apparently not. I think I will use luasockets code and link that to socket.cxx.

Marcel

[marcelvanherk]

Hi, these new source files for in 1.5.0c integrate the socket layer from luasocket into the dicom server. They are still a bit rough, but they compile and run under windows and Debian. Can you try on your Centos system? Servertask does not compile with them but you can use dgate there. With them BREAK_SIZE should not matter anymore. regards, Marcel

…

On Thu, Sep 8, 2022 at 8:42 AM hakancicektr ***@***.***> wrote: but still there is corruption on image even i treid smaller value [image: image] <https://user-images.githubusercontent.com/112996705/189064410-b1184120-a45b-4406-9ffd-5654991e7be8.png> — Reply to this email directly, view it on GitHub <#26 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAVDWJLNEYYL7624BJF6QOLV5GKHJANCNFSM6AAAAAAQGK4VRA> . You are receiving this because you commented.Message ID: ***@***.***>

[marcelvanherk]

ctest.zip

[hakancicektr]

Hello thank you for your efort

after complie dgate listener got error

***Error in Accept() function call
***Attemping to re-bind socket