Bookmark backup size difference? What Data got lost?
I just got a new SSD and did a clean Ubuntu 19.10 install. Before deleting the old system I did a backup of my Firefox (71) bookmarks to a JSON file. That JSON file has a size of 6.4 MB.
Now on the new install (FF 72.0.1) I restored my bookmarks from that JSON file. But when I did my daily bookmark backup shortly after, the new JSON file only has 5.5 MB.
How come the 0.9 MB difference? How can I tell what data was lost in the process?
Zmodyfikowany przez blabot w dniu
Wszystkie odpowiedzi (10)
The name of an automatically created JSON bookmarks backup in the bookmarkbackups folder includes a total item count (folders and separators included) and an hash value to prevent saving the same backup more than once. You also see this count in the Restore menu drop-down list.
- bookmarks-YYYY-MM-DD_<item count>_<hash>.jsonlz4.
One reason for size difference could be the loss of site favicons.
@cor-el Your answer does not have much to do with my question. I am also not talking about automatic backups.
I did a manual bookmark backup to a json-file RIGHT AFTER restoring from the json-file created for migration purpose.
I did this to check whether the importing/restoring was successful, because the restoration of my ~18k bookmarks took Firefox a moment.
And I am now worried why the new file, that should have the same exact data, is sized 5.5 MB instead of the original 6.4 MB.
@crankygoat
So you are claiming that the bookmark-backup json-files contain the picture data for every website bookmarked...
Is that just a wild guess or where can we find information on what exactly Firefox includes in the bookmark files?
Because WHY WOULD FIREFOX include these graphical data in a bookmark backup and then don't use that data when using that allegedly-oversized json file for restoring?
The problem is reproducible btw: the 6.4 MB always turn into 5.5 MB after immediate countercheck.
What other cause could this disparity have? I just want to be sure that no vital information was lost!
Zmodyfikowany przez blabot w dniu
Why don't you compare these files in Notepad++?
TyDraniu said
Why don't you compare these files in Notepad++?
I now installed Notepad++ and the JSON Viewer plugin. I opened both json files with it. I have no experience on how to investigate this but what I see is:
6.4 MB file has: length: 8.9kk and lines: 210k 5.5 MB file has: length: 7.9kk and lines: 197k
I conclude that I have lost a ninth of my bookmark data just by inserting the json file into a (clean installation of) Firefox and extracting it again.
I mean has nobody posted a similar problem here before? What am I doing wrong? How can I prevent this disparity?
My problem stated in my first two posts remains: What part of my data is Firefox killing? Why is Firefox doing this mess?
Firefox should have creates new automatic backups by now and that would allow to see how many bookmarks there are.
You didn't backup any other profile data from your previous install ? It is hard to compare data in JSON format, so maybe use this tool to extract the bookmarks.
Yes I only backup bookmarks. The relevance of other profile data has not occurred to me yet.
Using that Jefferson website does not look safe to me from a privacy perspective.
crankygoat mentioned that the 0.9 MB difference could be because of the favicons.
Indeed the favicons are all absent, when one does a bookmark restore from file on a fresh OS+browser installation.
Now looking at the two json files in notepad++ I do notice that the 5.5 MB file is missing most of the lines that contain the word "iconuri".
In fact I identified a decrease from 9286 to 164 counts of "iconuri".
If that's really the cause of the 0.9 MB difference, then I would have no problem, since I don't really need or care about the icons.
Can you confirm that a 9k of these simple iconuri lines translate to 0.9 MB?
But looking at the exact difference in lines between the two files, it is 13472, while the difference in iconuri counts is (9286 - 164 = 9122).
So there is still a difference of 4350 missing lines unexplained.
And ultimately: WHY WOULD FIREFOX a) backup the favicon urls to json file, but b) not restore them when restoring from his json file?!
Doesn't that either make a) unnecessary big or b) an incomplete process?
Zmodyfikowany przez blabot w dniu
A JSON file is a file with one long line and no line breaks, so I assume you talk about displaying the file in an editor with a specific width or can your editor Pretty Print a JSON file ?
cor-el said
A JSON file is a file with one long line and no line breaks, so I assume you talk about displaying the file in an editor with a specific width or can your editor Pretty Print a JSON file ?
I installed Notepad++. Within Notepad++ I installed the "JSON Viewer" plugin. And then within that plugin I clicked on "Format JSON".
Late answer to your question about favicons: It is behavior i have noticed repeatedly in the past, where favicons disappear after a bookmark import. It seems, as you have noticed, that absent an import of the local favicons cache itself, the URIs in the bookmarks file will not be present upon a new backup. Moving favicons.sqlite (or the entire profile) to your new install obviates these and other potential issues.