LibreOffice HTML tidy-up app

But if you squint a bit it looks like a bloke with a pointy beard and long lower lashes looking to his left.

Looks more like a lady’s reproductive system to me.
Bad icon? Depends what you want to say…

Just showed the larger version (in the profile page) to the wife, and she independently came up with the same reproductive system comment.

(Minor edit, rubbish word use)

So really it’s a failure as an icon!

Hey – dragons are cool. Three million Welsh and 1.4 billion Chinese can’t be wrong…

a pair of red lips dribbling blood!

Oh, Elvira, dear, Halloween was last weekend.

Looks more like a ladies reproductive system to me.

Would that be “lady’s”? Or is it a composite?

Well, I can’t possibly comment on that without referring to the recent American election.

But I didn’t spot that one myself, despite being still active in that arena – same age as the “former” president!

But not so (insert own adverbial perjorative term here!)!

but lots of them have difficulty writing an X.

This is amusing for all the wrong reasons:
https://mobile.twitter.com/realDonaldTrump

despite being still active in that arena – same age as the “former” president!

TMI, John, T. M. I.

For Authentic Steve: http://clive.semmens.org.uk/RISCOS/XP1deBloat is ready for service.

It goes through a file copying everything until it finds ‘ ICON="’, whereupon it discards everything until it finds ‘"’, then starts copying again. Repeat ad nauseam, or strictly, until end of file. I’m not sure what you wanted to do with the resultant file, or whether it’ll work without the ICON=“…” sections – if not, I’ll find a nice short ICON= section, and stuff that in place of the long ones, and we can see if that works.

Or if you want a standard HTML file containing just the links, rather than a NetScape Bookmarks file, that’s a slightly bigger job but not hard.

• * * * *

This was a very quick and easy hack compared to most of my other apps. You could easily alter it to do completely different tricks. It’s pretty easy to edit the runfile yourself, as long as you don’t want to add options. All the actual work on the files is near the top of the program, and all the wimp stuff is out of the way down the bottom. If you don’t need to add options, there’s no need to mess with the wimp stuff at all.

Just to show what an easy little hack this is, here’s the guts of the program (the rest is bog standard WIMP stuff to drag and drop files on and off the iconbar icon):

DEF PROCXP1deBloat(in$,out$)
  PROCLoad(in$) : Ptr% = InBuff% : OutPtr% = Ptr%
  REPEAT
    IF ?Ptr% = &20 THEN
      IF FNIcon THEN PROCIconSkip
    ENDIF
    ?OutPtr% = ?Ptr% : OutPtr% += 1 : Ptr% += 1
  UNTIL Ptr% >= End% - 1
  PROCSave(out$)
  PROCReleaseBuff
ENDPROC
:
DEFFNIcon
  FOR n% = 0 TO 6 : Test%?n% = Ptr%?n% : NEXT
  IF $Test% = " ICON=""" THEN = TRUE
= FALSE
:
DEFPROCIconSkip
  Ptr% += 8
  REPEAT Ptr% += 1
  UNTIL ?Ptr% = &22
  Ptr% += 1
ENDPROC

<pre>
<code>

Code Here

< /code>
< /pre>

Cheers Jeff! Done it a different way already – and lost my indents… will try your way, see if it preserves them!

Great. It does.

Textile is hard, I still can’t work out how to show the < /code> tag without having an extra space to make it show!

&lt;code&gt; works outside a <pre> element, but I don’t know how to do it inside one.

I’m not sure what you wanted to do with the resultant file,

Part curiosity about how much the “dross” amounts to in a bookmarks file, part interest in seeing whether memory use reduces and whether things load faster etc.
Mostly because I’ve seen it and it has no function and it irritates me like that.

Textile is hard, I still can’t work out how to show the < /code> tag without having an extra space to make it show!

I think that, theoretically, the notextile tag does that – assuming textile isn’t having one of its quirky days.
Textile quirky days are notable for having a name with a “y” at the end.

" <code> works outside a < pre> element"

Except when textile decides you’re wrong, because “y”

Sorry, cannot resist it:

#! lua
local f = io.open (arg[1], "r")
local text = f:read "*all"
f:close ( )
io.write (text:gsub ([[ICON=%b""]], ""))

8~) Such is life. I suspect PERL would be good for this job, too. But better the language one knows than the language one half knows (PERL) or doesn’t know at all (Lua).

I’m not a StrongED user, either. Whereas Zap and BASIC are very familiar.

Years ago, I wrote an app in BASIC, with an assembler core for the heavy lifting, that could do several hundred overlapping search and replaces like this in parallel, to extract the content from a wide variety of different files from different word processors that were around in those days, and put them into Impression’s DDF (Document Description Format). This was for the Physiological Society’s journals – dozens of academic papers every month. My app processed papers to a reasonable state ready for the copy editors in a few seconds – versus overnight runs on a Mac for a similar process used by another publisher we knew publishing similar volumes of stuff.

I’ve never bothered to update the assembler core to 32-bit it…

I’m slowly getting more familiar with PERL, because it’s what I’ve got easy access to on the Mac.

Perl – bah!

Python if you want to retain a modicum of sanity.

BTW: I’m going to have to change my icon now aren’t I?

BTW: I’m going to have to change my icon now aren’t I?

Yes – at the size it is shown, it appears to be a crucifixion scene.

to retain a modicum of sanity

I’d have to acquire some before I could retain it.

Given that I know BASIC very well, I’m not sure I can be bothered with Python on the Pi. PERL is good on the Mac for doing search and replaces across numerous files all over a directory tree, which is all I’ve wanted it for so far. Anything I want to do on a single file that’s more than Atom’s search and replace can do easily, I just flip over to the Pi and do it in BASIC (or Zap).

Lua looks quite interesting, but again, I doubt I’ll bother.

If you are using the latest version of RiscLua the script can in fact be written in four lines:

#! lua
local f <close> = io.open (arg[1], "r")
local text = f:read "*all"
io.write (( text:gsub ([[ICON=%b""]], "") ))

It’s not as slick as Lua but:

*BasaltInit
ARGS "source,output" TO source$,output$
(strand$)=LOAD$(source$)
regex$=" ICON=""\a*"""
SEARCH (strand$),regex$
WHILE GROUPS
 (strand$)=REPLACE$("")
 SEARCH (strand$),regex$
ENDWHILE
SAVE (strand$),output$
DELETE (strand$) 
QUIT

;-)

Edit: added DELETE

Those suggesting alternative languages might perhaps note that this simple problem wasn’t what the thread was about. It was the more complex (but still not terribly difficult) problem of tidying up HTML files generated by LibreOffice. When LibreOffice saves HTML, it gives detailed formatting information that makes the output match what was in LibreOffice – whereas a web page probably doesn’t want most of that. It’s actually worse than that, in that LibreOffice has a tendency to leave matched pairs of tags lying around with nothing between them, and a few other oddities. Details in the help file in the app, and on the description page on my website.

The other thing is that I expect to be processing numerous such files in the future, so being able to simply drag files onto the icon bar and drag the product wherever I want it is important. I’ve no idea whether this would be easy with other languages.

You’re welcome to download the app and take a look inside. I’m quite interested to know what a complete drag and drop app like this would look like in Lua or other language – but I’ll probably continue using BASIC anyway, because I know it, and I can re-use large chunks of wimp handling code.

Also, for what it’s worth, BBC BASIC could be a lot more concise than what I wrote – probably not a lot different from Basalt – but I like the clarity of doing it the way I wrote it. No idea whether other people find it clearer, but that way of writing it makes it much easier for me to reread it in the future, for editing this app or re-using the code.

This script will do quite a lot of cleaning:

#! lua
local f  = io.open (arg[1], "r")
local s = f:read "*a"
f:close ( )
-- remove header
s = s:gsub ("<head.-</head>", "")
-- clean <body...>
s = s:gsub ("<body[^>]*", "<body")
-- remove ICON = "... "
s =  s:gsub ([[ICON=%s*%b""]], "")
-- clean <p...> tags
s = s:gsub ("<p[^>]*", "<p")
-- clean <hn...> tags
s = s:gsub ("<h(%d)[^>]*", "<h%1")
-- remove comments
s = s:gsub ("<!--.--->", "")
io.write ((s))

So much to play with.

Tired eyes though. Just done my first full day in the phased return.
Feed Freddie, Minnie and Merlyn – much less eye strain.

Gavin: a script isn’t a drag-and-drop app; and you’ve only dealt with a handful of the cases, without any attention to whether any of the detail of the cases is worth preserving…

Lua looks more concise than BASIC when it comes to a job like this, although how clear it would be for the more complex processing I do on LibreOffice files, I don’t know. I’ve no idea whether it’s quicker or slower, but since my app takes a few seconds to do multimegabyte files it’s not an issue I’m going to worry about. But the main thing is, while I could doubtless learn to write scripts like that, it’d be a whole different ball game writing a drag-and-drop app to do it. BASIC isn’t beautiful to write them in, but I already know how to do it – for RISCOS. This is my main attachment to RISCOS. If I was going to learn to write drag-and-drop in a new language, it’d be for MacOS, not for RISCOS.

If your handling of tags is logically the same as mine, it’s not clear (to a non-Lua programmer, anyway) at all. I don’t remove comments at all, because LibreOffice doesn’t insert any; I didn’t deal with NetScape Bookmark files’ ICON=" garbage because, again, LibreOffice doesn’t create them – they cause my program to go tits up because they’re more than 254 bytes long, and my program assumes a maximum of about 254 bytes for the length of a tag – I might fix that, because while no LibreOffice file I’ve handled had a tag anywhere near that long, there’s no rule to say it can’t. A sufficiently long URL could do it to an <a...> tag.

You’ve done nothing to clean out empty pairs of tags, common in LibreOffice HTML if the work has been edited much, or tags that give useless information, like (for the nineteenth time) “this in in GB English” or “this is black text” or “this (body text) is Times New Roman” – which LibreOffice thinks is part of the content, but which you don’t want for a web page (or you shouldn’t, if you understand what HTML is supposed to be for) – and you certainly don’t want it repeated ad nauseam when it’s not even been anything else in between.

Sure, I could probably write it all in a Lua script – but can you then drag and drop files on it? And anyway, I know BASIC…

Just for laughs, I went and looked what it was that went wrong when I dropped the Firefox bookmarks file on XP1LO2web – it wasn’t the ICON=" that stuffed it, in fact that didn’t bother it at all, when I edited the bookmark file to clear the issue that actually stuffed it. It was the absence of a

</head>

I suppose I could make XP1LO2web accept “HTML” files without a header, but I’m not sure why I would. What it ought to accept, perhaps, but currently doesn’t, is a

</HEAD>

I fully take your point Clive. By drag and drop you mean dropping the document onto the application, whereas what I am doing is the reverse: dragging the application (i.e. script) onto the document, or at least a StrongED window displaying it. I am sorry if my relentless pushing of Lua seems irrelevant. I am doing it mostly for my own pleasure. So here is a script for removing empty tags – it is not recursive, alas.

#! lua
local f = io.open (arg[1], "r")
local s = f:read "*a"
f:close ( )
local empty = "<([^%s/>]-)[^>]*></%1>"
io.write (( s:gsub (empty, "") ))