tags:Problems and inconsistencies in how tags are created from folder names when importing bookmarks (from at least Firefox)

[znoteer]

Some multi-word folder names were split into multiple tags, one per word. Other folder names were left intact as multi-word tags.

Bookmarks -> Toolbar -> suivre -> Énergie hors réseau, chauffage alternatif -> woodgas

produced the tags

bookmarks ­­ · toolbar · suivre · Énergie · hors · réseau · chauffage · alternatif · woodgas

but,

Bookmarks -> Toolbar -> suivre -> Énergie hors réseau, chauffage alternatif -> Éolien

was broken apart differently giving

bookmarks · toolbar · suivre · Énergie hors réseau · chauffage alternatif · éolien

Notice also that the captial É of Énergie was preserved whereas the initial caps of Bookmarks and Éolien were converted to lower case.

Bookmarks -> Toolbar -> suivre -> Ordi, PCB, SoC -> PCBs (RaspPi, Arduino, Beagle, etc) -> Raspberry Pi computers

gave

bookmarks · toolbar · suivre · ordi · pcb · soc · pcbs · rasppi · arduino · beagle · etc · raspberry · pi · computers

whereas the same path for a different bookmark came out as

bookmarks · toolbar · suivre · ordi · pcb · soc · pcbs rasppi · arduino · beagle · etc · raspberry · pi · computers

where "PCBs (RaspPi" come out as a single tag.

Likewise,

Bookmarks -> Toolbar -> suivre -> DEL -> Lampes de poche, lanternes

for "Lampes de poche" generated tags "lampes", "de", "poche" and "lampes de poche" depending on the bookmark being imported.

Same thing for "Maison intelligente" and "file systems" in the paths

Bookmarks -> Toolbar -> suivre -> Maison intelligente, senseurs, sécurité

and

Bookmarks -> Computers -> file systems, synchronization

In the paths

Bookmarks -> Toolbar -> projetPBX -> VOIP -> IP-PBX, passerelles PSTN, ATA, cartes expansion and

Bookmarks -> Toolbar -> Search -> Phone, Area Codes

"passerelles PSTN", "cartes expansion" and "Area Codes" were all preserved as 2-word tags.

My preference would be for multi-word folder labels to be preserved as multi-word tags and for capitalisation to be preserved.

Some punctuation is also not handled gracefully. Some folder had datestamps as names. The ':' (colons) in the time part disappeared.

Other folders had a '/' (forward slash, divide-by sign) in the name. In that case the words on either side were concatenated into a single tag. For example, the folder name "one/two" would be converted to "onetwo". It would be preferable to have two tags.

Ampersands ('&') were handled not bad, being replaced by the word "amp". It would be more helpful if the '&' were converted to the word "and", and, as above, if the multi-word label were converted to a multi-word tag.

Finally, apostrophes ("'") were converted, for whatever reason to "39", so that "Can'n" and "l'espace" became "can39n" and "l39espace"

At the end of the process, I also ended up with the tags "0", "1" and "2". I have no idea where they came from. I looked through all my bookmark folder hierarchy and see no isolated digits that could have generated them. "0", if clicked, returns the complete list of all my bookmarks and url-less notes, despite having only 9 members according to the tag cloud. "1" and "2" return emply lists when clicked despite reportedly having 4 and 3 members respectively.

[znoteer]

New detail:

I've had a case where an existing multiword tag added to a new shaare is converted to multiple tags when the new shaare is saved. In other words, the multiword tag did not survive as a single tag.

Unfortunately, I can't say with certainty that I have seen an opposite case, where an existing multiword tag added to a new shaare does survive as a single tag after the save operation.

[ArthurHoaro]

Would you be able to provide a sample file to easily reproduce the inconsistencies that you are mentioning?
Ideally with only a few entries, but containing significant data.

[znoteer]

@ArthurHoaro what kind of a file? This was done using the import feature, and the "New detail" was done using the shaare+ button. I didn't start off with a file, so I'm not sure what kind of a file you'd like me to send. Please elaborate.

[nodiscc]

what kind of a file?

A Firefox bookmarks export that contains problematic bookmarks.

[znoteer]

@ArthurHoaro @nodiscc

I think it was more the folder names that were problematic than the bookmarks themselves. With that in mind, in order to try to make the file as small and relevant as possible, I've deleted all folders that didn't have any funky subfolders, and in subfolders with problematic names, I've deleted all but one or two bookmarks. I believe this will cover all the cases I brought up in the OP. If not, please let me know and I will send another less drastically editted file.

Hmm, github won't let me attach an html file to this comment, so I guess I'll have to post my file somewhere else.

https://thisis.fbx.one/bepasty/kufz3wmX

Edit: I just realized that you will need to download the file rather than view it. You need to download it in order to import it in your tests and because the above server won't render it because it's too big. Given that, I'm adding the extention ".txt" to the filename so that GitHub will accept it. That way there will be a copy on GitHub for posterity should anything happen to the above server.

bookmarks_for_shaarli_edited_02.html.txt

[znoteer]

@ArthurHoaro @nodiscc

Has any progress been made on this?

I had a data disaster recently and lost all shaares. I'm currently cleaning up my Firefox bookmarks in preparation for another run at importing them to shaarli (mainly removing dead links, and reclassifying others).

Is there anything I can do in this next eventual importation that would be useful to debugging this?