Notes |
(0000728)
hockm0bm
03-03-09 11:02
|
Are you able to reproduce with other documents? |
| |
(0000753)
hockm0bm
03-23-09 17:25
|
ping LEGOManiac? |
| |
(0000769)
LEGOManiac
03-30-09 22:58
|
Sorry for the delay. My motherboard died and, while trying to reconfigure Ubuntu for a replacement PC, I screwed it up so badly I had to re-install. I haven't come back to installing PDFEdit yet, but I will and I'll try other documents, as suggested. At the time, I was only trying the insurance forms I had been working with. I'll let you know in a couple of days. |
| |
(0000770)
LEGOManiac
03-30-09 22:58
|
Sorry for the delay. My motherboard died and, while trying to reconfigure Ubuntu for a replacement PC, I screwed it up so badly I had to re-install. I haven't come back to installing PDFEdit yet, but I will and I'll try other documents, as suggested. At the time, I was only trying the insurance forms I had been working with. I'll let you know in a couple of days. |
| |
(0000771)
LEGOManiac
03-30-09 22:58
|
Sorry for the delay. My motherboard died and, while trying to reconfigure Ubuntu for a replacement PC, I screwed it up so badly I had to re-install. I haven't come back to installing PDFEdit yet, but I will and I'll try other documents, as suggested. At the time, I was only trying the insurance forms I had been working with. I'll let you know in a couple of days. |
| |
(0000775)
LEGOManiac
03-31-09 04:26
|
Here's a file that consistantly loses the underlying PDF when you try to add text. Curiously, I have another scan done on the same scanner two weeks later which can be annotated without problems.
In both cases they were scans made with XSane image scanner capturing directly to PDF.
I have had this happen with insurance forms from Blue Cross, which were not scans, but I have not been able to locate them. |
| |
(0000782)
hockm0bm
04-01-09 01:02
|
> Here's a file that consistantly loses the underlying PDF when you try to add
> text
I cannot find any document neither attached to this bug nor provided through a link. |
| |
(0000785)
LEGOManiac
04-02-09 04:51
|
OK. I've tried uploading it again. I had assumed the first time that if I selected a file AND added a note and that the file would be uploaded when the note was added. I've now done this as two steps. |
| |
(0000796)
hockm0bm
05-14-09 10:49
|
Sorry for long time, but I was quite busy with other stuff.
I have tried the attached document and really all changes to the page (line, text) cause the rest of the page to disappear.
The document, however, doesn't seem to be correct. Here are the messages printed on the console:
Error (31873): Illegal character '{'
2:KERNEL:ccontentstream.cc:parse:418: Invalid content stream...
2:KERNEL:cpagefonts.cc:getFontIdsAndNames:189: No resource dictionary.
The first line is xpdf error message which encounters unexpected character in the page content stream (compressed by FlateDecode). The second line is our (PDFedit) reaction on that. The last one is not related to this at all.
What is not clear to me is that why sole xpdf doesn't print this message and why we are able to display the stream and the problem occurs only after we try to make change.
What do you think Jozo? |
| |
(0000797)
hockm0bm
05-14-09 11:15
|
I have looked to the file closer and here are my observations:
Page 1 -> [6 0]
content stream -> [7 0]
content stream is not compressed directly, but it contains one large inner blob which is compressed by FlateDecode:
"
q
1 0 0 1 0 0 cm
1.000000 0.000000 -0.000000 1.000000 0 0 cm
600.960022 0 0 832.080017 0 0 cm
BI
/W 2504
/H 3467
/CS /G
/BPC 1
/F /FlateDecode
ID
x<9C><EC><BD>.......
"
Honestly, I have never seen such a structure. Do we support it at all? |
| |
(0000798)
hockm0bm
05-14-09 11:17
|
Btw. I have used tools/pdf_object_printer and it didn't show any parsing error. |
| |
(0000919)
misuj1am
08-10-09 01:21
|
cannot reproduce |
| |
(0000922)
hockm0bm
08-10-09 09:22
|
How did you try to reproduce? I am reproducing it even with the curren CVS snapshot. Just use attached document and try to add a simple text. All previous content just disappears. |
| |
(0000924)
LEGOManiac
08-10-09 19:34
|
For what it may be worth, I just thought I'd let you know that this document was made using XSane Immage Scanner under Ubuntu 8.10. I'm not sure what handles the underlying PDF creation - whether it's XSane itself or XSane calling another package - but I would suspect that anyone scanning using XSane is going to have a similar problem. Indeed, if XSane is using an external PDF creation library then the problem will be more wide-spread.
If you suspect that the underlying PDF structure is incorrect, and if you can figure out who's responsible for the actual PDF creation, you may be doing a lot of people a service if you fire off an email to the creators of the PDF libraries (or whatever) to point out the error(s) so they may be corrected at the source.
I'd do it myself, but, being strictly and end-user, I don't know enought about it to talk about it intelligently. |
| |
(0000925)
misuj1am
08-11-09 02:03
|
i added another tool (add_text) which adds text to pdf files, that is how i try to reproduce it on win32. after 5 successful addition on different places i give up. reassinging to michal who can reproduce it. |
| |
(0000928)
hockm0bm
08-12-09 11:00
|
I have used test.cc to add a simple text to the attached document and really, the text is visible and the original content is there as well.
I am getting the following error though:
Error (31873): Illegal character '{'
2:KERNEL:ccontentstream.cc:parse:417: Invalid content stream... |
| |
(0000929)
hockm0bm
08-12-09 11:01
|
Jozo, could you check why this error message is displayed and what it means?
Is this related to what I was asking about in comment 797?
What should we do about that? |
| |
(0000930)
hockm0bm
08-12-09 11:06
|
Btw. you need to have pdfedit-core-dev installed on your system in order to compile attached test case. |
| |
(0000931)
misuj1am
08-12-09 11:10
|
as i told you before, i cannot reproduce. no error message like "'{' invalid .." for me.
please, stop assigning this bug to me. for now, i cannot reproduce so i can close it as works for me.
the time spend by trying to reproduce is getting much bigger than for you to solve it.
/jozo |
| |
(0000933)
misuj1am
08-12-09 11:41
edited on: 08-12-09 11:42
|
i am not working on linux anymore - if the last comment was for me :).
|
| |
(0001075)
hockm0bm
05-12-10 14:43
|
OK, I found out why you were not able to reproduce without GUI. The problem is that this has nothing to do with adding text at all. We happened to see this in GUI just because every page content update goes through scripting and all those modification functions (e.g. _AdddTextSlot) call go() at the end and this function calls saveChange on all content streams on the page.
I have update test.cc test program to simply call saveChange on all content streams and then save the document and I am able to reproduce also without GUI.
The strange thing is that the content stream (object [7 0]) has an empty stream data in the changed document:
7 0 obj
<<
/Length 0
>>
stream
endstream
endobj
I am getting the very same error message during test program:
./test test.pdf
Error (31873): Illegal character '{'
2:KERNEL:ccontentstream.cc:parse:417: Invalid content stream...
Jozo, could you have a look at it? |
| |
(0001076)
misuj1am
05-12-10 14:46
|
is length 0 stream invalid?
it can happen because when we have 10 streams in 10 objects and we change one of them, we concatenate them together and put them to the first one |
| |
(0001079)
hockm0bm
05-12-10 15:03
|
> is length 0 stream invalid?
I don't know. There is nothing specific about 0-length in the specification AFAICS.
> it can happen because when we have 10 streams in 10 objects and we change one
> of them, we concatenate them together and put them to the first one
This is not a case here as the document has only 1 content stream. It looks like we are doing something wrong in the buffer -> string translation. |
| |