jdev - 2021-04-06

Office hours starting in 15 minutes! https://socialcoop.meet.coop/sam-pku-dud-niv

Sam, maybe im too stupid, but parsing blocks seems to be much more of a challenge than spans ✏

but im not giving up

only need a few more regexen

No no, the term is 'regexopodes' ✏

lovetox: I think I had a harder time with quotes in particular as well because they don't have any convenient "end" marker and I wanted to keep the line lengths and spaces and what not.

its the mixing that gets me

blocks that have ends, blocks that dont have ends

block where the end is defined as the start of another block

I mostly just ignore "plain" blocks I think, so it was really just quotes for me.

pre's were easy because you just scan until the end of the document or the end of whatever block you were already in

who needs nested quotes

if the xep aims at "good enough" maybe my parser should too

seems weird for such a very simple styling language to support unlimited nested quotes

I think I always count the ">"s at the beginning of each line to get the quote level, then whenever that number gets smaller than whatever it was before I call that the end of a block.

Originally I'd done that recursively where I had a byte reader that understood quotes and would only read up to the end of its own quote, sipping the ">" bytes and then I could recursively create a new decoder inside of each block, but I changed that because I wanted it to be lossless and not throw away spaces and newlines.

*stripping the ">" bytes, I mean

I think… I rewrote this thing like 5 times before I found an API that I liked, so I barely remember what bits I ended up with.

when clients quote, du they somehow identify the originator of the quote ?

and how do they do that

Not that I've seen

Plz XEPify `<in-reply-to id=$stanza-id/>` and use that 😉

sounds like something very basic i would want to have with a quote..

Or references I guess

I could go either way, but I'm with Zash. I think these quotes are mostly just to provide context in a MUC when multiple people are talking at once and it's not immediately clear which recent message you were relying to

If you need more, this can be done with some sort of threading XEP

`nick[:,]\n> lorem ipsum\n what you say‽` 🤷️

But a client would certainly be free to prepend "So and so said:" before the quote like email clients do

Yes, that :)

Yaxim has something too IIRC

so > ´´´sometext

this is a quote with a preformated block

this is valid or not

yes that's valid

and why do we need that? because its common to not quote the end ```

?

It just made it easier because you didn't have to scan for an end token to know if the pre was valid

i think i just use regex

for blocks

its easy to match a pre or quote block with regex

the only problem is the pre without end inside the quote block ..

hm or can i just write a regex for that too ..

lovetox: out of curiosity, what are you writing this for?

for Gajim

oh right, I think I knew you did gajim stuff. cool.

Sam in the xep is nothing said about empty blocks

> ``` for example

dont know if it should though

I think that's valid. It's not mentioned specifically, but I think the rules say that a pre-formatted block has to be started with "```" and runs until the end of the parent block or until the next line containing only "```", so it should be valid.

so in the end now parse the blocks with regex

works fine for some tests i made, tomorrow i will test all your test cases

Cool; glad it worked for you!