English Language & Usage: Multi-Layered Discourse Room

Cerberus

00:05

Hey.

@Vitaly Which ones are bothering you? And what exactly are you doing?

Vitaly

@Cerberus — I was thinking of trying to convert it appropriately (with the macrons and whatever), but I don't understand half of what is there in the entries. It would take you longer to enlighten me than to learn regexps yourself.

And for example, “Α῎βροι”

WTH is the “῎”?

Cerberus

Ah. Yes, those are two accents that need to be to the left of the A, almost on top.

Vitaly

And why is it “Mir” instead of “M. Ir.” and “cymr” instead of “Welsh”?

That's the very first entry.

Cerberus

That is probably the German abbreviation for "Mittelirlandisch" or something?

And "cymr", I have no idea...

Vitaly

I have. It's Cymraeg.

But WHY is it Cymraeg in the original?

How come whoever converted that replaced it with “Welsh”?

Maybe it wasn't the original after all?

Cerberus

00:11

Hmm...

The "source" is weird already.

All those English translations of German words...

Seventh Element

I am captain Shakespeare. Who are you?

Cerberus

Welcome!

I am the cook, Epicurus.

Seventh Element

You took your time. How are you Epicurus?

Cerberus

Hungry!

Seventh Element

Is that a fysical reaction to my presence?

Cerberus

00:16

It depends. Do you carry food?

Seventh Element

I'm digesting some of it as we speak.

Cerberus

Okay. If you will lie down on this table, the "surgeon" will be right with you.

Seventh Element

This isn't a chat room. It's a medieval dungeon!

Cerberus

Oh, stop your word games. What's the difference?

Seventh Element

Medieval dungeons didn't have a "close" button in their top right corner.

Cerberus

00:23

Are you sure it's a close button, not a teleport-me-to-the-underworld button?

Vitaly here is trying to convert a dictionary of Proto-Indo-European into the right format for a dictionary program...

@Vitaly: I have checked the source version, and it seems to be correct in all respects, as far as I could see (the partial translation is weird but doesn't interfere with the contents).

That is, it contains many incidental errors, but no structural ones that I could see so far.

Vitaly

00:44

36 mins ago, by Vitaly

@Cerberus — I was thinking of trying to convert it appropriately (with the macrons and whatever), but I don't understand half of what is there in the entries. It would take you longer to enlighten me than to learn regexps yourself.

Cerberus

But I wouldn't even know where to begin! I have no idea how html works, and I have no idea what dsl is supposed to do.

Vitaly

Guess what, I have no idea about the meaning of those characters.

Cerberus

What if each article were just marked thus: one label for the title, one for the text. That would already work very well? Those colours in GD are nice, but not really necessary for the reader. Or is it impossible to format it that way?

@Vitaly If you just assume that everything in the source is correct, could it be converted then? Or do you need to check everything?

Vitaly

I don't know what they would end up like. And I cannot proofread the result for errors, because I have no idea what it should be like (and cross-checking it with the “source” is too much trouble).

Cerberus

Hmm... so it could happen that a certain character was displayed quite differently from the source?

Vitaly

00:52

Seriously, could you just open any DSL in a text editor?

Cerberus

I have it open right now!

If you regexed it all into a format that would display at all in GD (just with titles / the rest), I could try and check that as best I could?

Vitaly

…

Cerberus

But I don't mean to push you into doing this stuff. If it's too complicated, don't.

It's just that I thought you were already busy trying...

Vitaly

Cerberus

With your code2000 font, I can read the entries perfectly in GD (there's only the wrong spaces, so far, and the wrong characters in the titles/menu, but those aren't real problems).

Vitaly

00:58

With the Code2000 font, it's still u-dieresis-accent instead of a-macron-accent.

Cerberus

Not for me!

Vitaly

What?..

Cerberus

It's only those Greek capital accents that are in the wrong place.

Vitaly

erk, dude, erk

and I have no idea what to do with those Greek accents

Cerberus

Wait this is very weird.

Vitaly

01:02

It is not. The characters in the dictionary itself (DSL) are misconverted from the source.

Cerberus

In the GD version, those accents on Grk. Sītânes are nearly correct; in the source, they are wrong.

Vitaly

So it will invariably be u-dieresis-accent.

Just post a screenshot of whatever you have for erk

I mean kʷetu̯ er-

Cerberus

Oh. I forgot: I had converted those ü's to ā's in my dsl, and they look good now with the Code2000 font (not with the other fonts I tried).

Vitaly

That's what I am talking about.

Who knows how many similar misconvertings are there.

Cerberus

Huh... this is even weirder! The ā is not there in erk, but it is there in kwet-

Or shouldn't it be there in erk- in the first place? I don't remember.

Vitaly

01:08

No, I said “erk” because I remembered that it was the query for the kwetu entry.

Cerberus

Oh, ok.

So there shouldn't be a macron in erk-.

You just meant it as en example of something that was wrong (before I edited that).

Vitaly

Yes

Cerberus

If I change the place of the accent to the left, it is too far to the left, but at least it's at the right side:

I've only changed the first one.

Vitaly

I don't even know what you are talking about.

Cerberus

You will see that the accents can be before or after the A.

They should be before the A, but closer to it than under "after".

I think I saw the accents at the right place in some other word; let me see if I can find that...

Vitaly

01:19

Cool.

Cerberus

Hmm those accents are only completely right with lower-case letters; but those have a single unicode symbol for letter+accent.

I also saw some more wrong ü's, and wrong fusions of ā and ă that should simply be ă's.

MrHen

@Cerberus: Haha, so I accidentally started a raid without realizing what it was :)

Cerberus

01:37

Well... nothing happens is you are defeated!

MrHen

What happens if you win?

Vitaly

02:14

@Cerberus: ifile.it/eyjd8rv

Cerberus

Hey!

@Vitaly What is this...? You didn't...?

Vitaly

No.

Open it in a plain text editor and GD.

Cerberus

What a lovely entry do I get when I search for e!

I like that you've included your portrait.

Vitaly

I even included some sounds.

Cerberus

Hey! You bastard! You have bugged my house? How did you record my voice?

And what is that car?

Vitaly

02:22

You did open it in a plain text editor, didn't you?

Cerberus

Yes, I see [m1] introduces a paragraph... or is that just the indentation, and is the paragraph indicated by a return?

Vitaly

A tab introduces one. The [mN] is indentation.

Cerberus

Ah ok.

Vitaly

The simplest dictionary possible:

#NAME "Example Dictionary"
#INDEX_LANGUAGE "English"
#CONTENTS_LANGUAGE "English"
entry
→definition

(The “→” should be replaced with the tab character.)

Cerberus

Right.

Vitaly

02:25

So in GD, when you search for entry, it would just say “definition” in the body.

Cerberus

So would you need regex to convert the bare bones of the html file to a dictionary file?

Vitaly

Of course. In HTML, it's italics. In DSL, it's [i]italics[/i].

Cerberus

Okay, I mean, would simple find-and-replace suffice for everything, if you don't know regex?

Vitaly

It could suffice. Or more likely, it wouldn't.

Because you would still need the tabbed structure.

Cerberus

Eh...

Vitaly

02:31

The headline for an entry is on its own line without any tabs.

The body of an entry should be preceded by tabs, or else it would be recognized as a headline.

I have no idea how you are going to put tabs where they are required without regexps or manual labour.

Cerberus

Oh? I think that should be possible? Let me try...

Hmm... I'd need to know more about html.

How are paragraphs indicated in html?

It seems by returns?

Vitaly

w3schools.com/tags/tag_p.asp

For that particular source, it's useless.

Cerberus

Oh.

Vitaly

Anyway, about this Α῎βροι

In the HTML source, these denote all those special characters. GD won't recognise those, so I am going to convert them with regexes and upload the resulting HTML file.

Cerberus

I think I'd first replace each return by a tab + a return; then I'd replace the horizontal bar in the original + tab by a return. Would that work?

Vitaly

02:39

IDK.

Cerberus

@Vitaly Ah, so that's the problem.

Unless you found a way to match each string of numbers with a character automatically, that would be too much work...

It appears the horizontal bar is <hr />.

I should get the real Pokorny from the library to check things against.

Vitaly

While my computer is busy decoding the numbers/characters, try pasting this into Notepad with Code2000 selected as the font: c‍t s‍t

Cerberus

I am trying to set the display font. It seems Code2000 displays question marks for those characters.

When I tried Lucida, the program crashed.

Notepad++

Vitaly

Notepad, not Notepad++…

Cerberus

Why Notepad?

Vitaly

02:54

because that's what works for me with c‍t s‍t and Code2000

Cerberus

The characters are not visible.

I see it just as in this chat.

Vitaly

You know your computer is screwed up?

Cerberus

I am on XP, that could be it.

Vitaly

You pasted them or typed them?

Cerberus

Pasted, obviously.

Notepad++ gives question marks in some fonts, square in others, nothings in others again.

Vitaly

03:00

Wow.

Cerberus

Doesn't even work in Word.

Vitaly

In any case, it has successfully decoded all the numbers.

Cerberus

Have you tried copying and pasting those letters from this chat to your Notepad yourself?

Could be that the chat is doing something to them?

In chat I only see c‍t s‍t too, by the way.

@Vitaly Oh! Good!

Vitaly

@Cerberus — No. It works for me.

Cerberus

Mysterious...

Vitaly

03:02

ifile.it/wma2szg

Cerberus

Cool!

I can see these from Wikipedia, by the way: ﬆ

But that's one character.

Wow, nice.

The characters look good.

Ἄ

Vitaly

Now there's an easier source for you to convert. :P

Cerberus

Heh, thanks!

If I find the courage, I will...

I got the idea to simply copy the right letter from a unicode Greek inputter: Ἄ. But I think that is in the wrong code for GD?

I can't even paste it into your file.

Vitaly

You have found the courage to trash the belief that the laws of nature respond to wailings and incantations in the age of anti-intellectualism, and you cannot find the courage to convert a file right away?..

Cerberus

Heh.

Vitaly

03:08

I can, but I don't have the slightest idea what to do with it.

Cerberus

This courage I shall have to muster is about finding out what all those html things do.

Hah it works, I can use that Ἄ in a GD file.

Works.

Hey I wonder, is there still copyright on Pokorny?

Hmm you probably won't know that either.

Vitaly

Haha. Good guess.

I know it's from the 1959 version.

Cerberus

Yeah...

Vitaly

Tabbify everything that doesn't start with Root / lemma: (Hint: use regexps)

Cerberus

I will consider learning it...

By the way, I say that Root / Lemma also in the text of some articles, I think.

Vitaly

03:19

> doesn't start

Cerberus

I was rather thinking of replacing that separator with something.

Yeah is bold.

Vitaly

I mean, there is a regex operator for start-of-line.

What bothers me is that some headlines contain something other than a proper headline. E.g.:

> Root / lemma: akʷā- (more properly ǝkʷā): ēkʷ-

Though the original converter didn't bother removing it either.

Cerberus

I think the separators / horizontal lines should work.

Vitaly

The “more properly”.

Cerberus

Hmm, why should that "more properly" be removed? It is just part of the text?

Vitaly

03:27

Why should it be in the headline?

Cerberus

I am more surprised that "root" etc. should be under "Notes"?

I have started replacing the bold etc. tags.

Vitaly

How many entries there are in total?

Cerberus

I haven't got that far yet...

Takes more than a minute to replace all , for example...

Vitaly

OK, I just thought you would know. I will just see that in the annotation file.

Cerberus

Would it make sense to do anything with the ?

Better just delete them?

Vitaly

03:40

If I replace every line that starts with “Root / lemma: ” with exact same line, there are 2195 total lines replaced.

Cerberus

Ah ok.

Vitaly

And the annotation says there are 2221 headwords.

So I am losing some of the headwords, but WHERE???

Cerberus

I get 2221

If I replace all the separators.

Vitaly

The source file isn't uniform.

Really?

^Root / lemma: (.*?)$ → Root / lemma: \1

gives 2195 occurrences replaced…

Cerberus

Perhaps some do not start with Root, somehow?

Vitaly

03:44

And how am I supposed to find them?

Cerberus

Yeah... I am just replacing the separators, see how that goes.

Now how to paste a return...

NP++ refuses to let me do that.

Vitaly

\n with regexps

Cerberus

Ah ok!

Vitaly

^Root(.*?)$ → Root\1 still gives 2195

FFFFFFFFFUUUUUUUUUUUUUUUUUUUU…

Cerberus

And tab is \t?

Vitaly

03:47

yes…

Cerberus

@Vitaly We would only find out whether those missing entries were actual entries if we read it all!

Vitaly

you know why I like CD databases better?

the formatting in them is at least uniform...

Cerberus

I can imagine...

Vitaly

great, removed all tabs and “2370 strings were replaced”

FFFFFFFFFUUUUUUUUUUUUUU x2

Cerberus

Hey, what's this: Schwyzer Gr. Gr. I 4654.

Superscript?

Vitaly

03:51

superscript

[sup][/sup] in DSL

Cerberus

Hmm can I do that in GD?

Ahh ok.

Does that work with all similar things?

Vitaly

no

Cerberus

Is this of any use, I think it is a line break:

Vitaly

idk

Cerberus

OK.

Vitaly

03:55

does it always say “Page(s)” once per every entry?

2222 strings for all instances of Page(s)

FFFFFFFFFUUUUUUUUUU x3

Cerberus

Hmm probably...

What are you trying to do?

Vitaly

I am trying to exclude all instances of headwords from my tabbifying regex

Cerberus

Oh, and is this subscript: , and does [sub] work?

@Vitaly Hmm... it's just so poorly formatted...

Vitaly

it invariably gives 2195 for “Root / lemma: ” and 2370 for “Root / lemma:” (without the space after the colon), but there is 2222 instances of “Page(s)”

@Cerberus — Yes.

Cerberus

Yeah there are 2221 separators....

@Vitaly Ok thanks.

What is <S>?

Vitaly

04:06

strikethrough, deprecated

Cerberus

Oh, ok.

Does GD have that too?

Vitaly

No

Cerberus

Okay... so I will make it red or something.

Vitaly

use a combining strikethrough character from Unicode or something

Cerberus

Hmm I could do that... lemme check what those struck-through things are anyway.

Oh, but I have a problem:

There are some [i]s and [s]s already in the text.

What to do with those?

Vitaly

04:11

Eh?

Cerberus

There are few enough of them to do by hand I think.

Vitaly

Example Dictionary, [m1]This is how to put \[, \], \{, and \} in the body of an entry.[/m]

Cerberus

Square brackets are used to indicate supplied letters or something... oh, I should put slashes before them, like this: /[ and /]?

Ah, yes, jinx.

Vitaly

No, backslashes.

By the way, I give up. It's just too poorly formatted. :P

Cerberus

Oh, right.

Vitaly

04:15

And another mistake of the previous converter:

Original: FlN Avo[s] > span.
Converted: FlN Avo[s] > spO.N.

Cerberus

Huh that looks odd...

How does O.N resemble an?

Vitaly

Some replacement that captured all instances of “an.” perhaps.

Since he was replacing “cymr.” with “Welsh.” etc.

Cerberus

Ah, it must be his replacing German with English then, some abbreviation gone wrong...

Hey, is there a way to replace several different strings at once?

Vitaly

Which is why the source should be better formatted.

It's impossible to think of everything when constructing a regexp for such a source.

Cerberus

Say replace with [b] and with [i] in one go?

Vitaly

04:18

@Cerberus — Yeah, with regexps. The | operator.

Cerberus

Ahh... so how would that work? | in the search field, and [b]|[i] in the result field?

Vitaly

Scratch that though. XD

I think I got confused.

I don't think there is a way to do that in my text editor with built-in regexps.

Cerberus

Hmm...

Vitaly

And in any case, [ and ] are functional characters in regular expressions, so I usually leave the 's, 's and so on until the structure of the DSL is correct (tabbed and indented and whatever).

Because < and > aren't functional characters in regexps. Less hassle to construct expressions.

With ['s, you would need to escape them like so: \[.

Cerberus

I understand... and can't [ and ] be escaped in regex?

Ah.

Vitaly

04:24

They can be escaped, but it makes regular expressions even more unreadable than they already are!

Cerberus

Hmm it would really save time if I had a way of replacing several separate things at once.

@Vitaly Understandble!

Vitaly

04:45

Hmmm…

Cerberus

Hey, does "[bla bla]" do anything if "bla bla" isn't a single character?

I mean, if it isn't a recognized command thing?

Not i or b or sup etc.

Vitaly

Yeah, destroys the body of the entry in GD.

[ shouldn't be used if unescaped. And so shouldn't ].

Cerberus

Ouch.

OK.

Are the ones you mentioned in your example the only ones that need escaping?

Vitaly

In any case, I figured out how to tabbify the whole file based on the horizontal rules.

Cerberus

Oh?

Vitaly

04:52

@Cerberus — I dunno, my help files mention ~ and ^ as well, but it's vague.

Better check them all in GD.

Also, a single backslash is done like so: \\

Cerberus

On no....

So I should have done that first?

Vitaly

C'est la vie.

Cerberus

Ah: ~ is replaced by the title of the entry.

So better escape those too.

Vitaly

@Cer: tabbified.

Each headword is preceded by <headword> (as a temporary measure)

instead of [m1]'s, there are <m1>'s

escape all special characters you can find, and then proceed to formatting

or whatever might need fixing first

Cerberus

Right, those [m1]s, I almost forgot 'em.

Thanks a bunch!

Vitaly

05:11

What's the proper reply to “thanks a bunch”?

Cerberus

That that thanks a bunch? The reply would be an equally sarcastic "no prob!"...

But I didn't mean it like that, as you well know.

Vitaly

So neat, if only the formatting was uniform

Did I mention formatting doesn't work in headwords? (e.g. [sub][/sub] won't work in that particular headword in the picture)

Cerberus

Ohno...

But superscript w and h are probably their own character?

Vitaly

yup

oh, I also accidentally added a bunch of currency signs to the end of each headword line

but that should be easy to delete :PP

Cerberus

No prob.

I'm nearly done.

Just removing the incidental html stuff.

Such as hyperlinks sigh

Vitaly

05:42

4 hours ago, by Vitaly

I think italicised lemmae should be coloured like so.

Cerberus

Mmyes? Nearly finished.

Vitaly

@Cerberus — Don't forget to post a screenshot of your “finished” DSL loaded in GD. :P

Cerberus

I will!

Give me twenty seconds...

Vitaly

lemmata*

Cerberus

@Vitaly Thanks.

Vitaly

05:47

@Cerberus — Eh? I was correcting myself.

Cerberus

And I thanked you for doing so.

Vitaly

No prob!

Cerberus

Ack. You won't like this screenshot.

Vitaly

I can imagine. :P

Cerberus

How do I get the formatting out of the titles?

Vitaly

05:50

With regexes, obviously.

Cerberus

Yeah buttt....

Vitaly

:D

Cerberus

I will try to found out how if you don't happen to know it...

Vitaly

Gimme a representative sample or something. It's obviously not my version of the dictionary.

Never mind.

You obviously managed to tabbify it yourself.

Unless you removed the <headword>'s.

So, what do we do with that…

Cerberus

I did. That's not the problem; I can put those headwords there if it helps you?

And how would you like the scrap of text?

typewith.me/XrXWSDLmfh

I have added the headwords back in.

Same link.

Vitaly

05:58

It doesn't load.

Cerberus

Huh?

Vitaly

You can get rid of them.

Cerberus

You're suppose to see me typing live...

Vitaly

Try something like: ^((?!(\t)))(.*?)\[b\]Root / lemma: (.*?)$ → \3\4 (based on your screenshot)

Transcript for

English Language & Usage: Multi-Layer…