make-doc-pro: Version 1.0.3 beta update

Hi, well it's easy and simple ;-)). Anyway, just help me to remeber that stuff with load-thru. IIRC it has something to do with security things... right? Robert

Hi, ups sorry I'm still loading my internal library. For the released version, I normally comment this. Anyway, I uploaded a rather new version with some enhancements: - Now supports inline markup characters in bullets etc. (should work for most situations) - =file command support added There is a little quirk with tables and talbe header handling. Just to let you know ;-)) Robert

rebol [ Title: "Push Cache" File: %push-cache.r Date: 20-Mar-2002 Version: 1.0.6 Needs: [] Author: "Anton Rolls" Language: 'English Purpose: {To push files from local/ directory into the public cache} Usage: { site: http://your-site/rebol/ anton-lib: http://anton.idatam.com.au/rebol/library/ do load-thru anton-lib/include.r include [ anton-lib/dir-utils.r [version?] anton-lib/push-cache.r [push-cache] ] ; now ready to use push-cache in script or in console ; eg. push-cache %afile.r ;(don't need to specify path) } ToDo: { - somehow report success with /quiet for util/push-into-cache.r - if errors writing the file in copy-file (check for owner-write) show-file-modes - /force for copying older onto newer - incorporate into dir-utils? or maybe a new cache-utils? reb-utils? - make doc/push-cache-flow-diagram.r } History: [ 1.0.0 [6-Jan-2002 {First version} "Anton"] 1.0.1 [18-Feb-2002 {wrapped in context to comply with new include framework, filling it full of goodness, seems to function correctly } "Anton"] 1.0.2 [21-Feb-2002 {added copy-file helper function} "Anton"] 1.0.3 [27-Feb-2002 {fixed bug by moving copy-file into push-cache} Anton ] 1.0.4 [28-Feb-2002 {commented site, added usage, added cache-site, change-cache-site} "Anton"] 1.0.5 [1-Mar-2002 {fixed bug checking for empty? common-path, checking for existance of source file} "Anton"] 1.0.6 [20-Mar-2002 {reporting cache-file} "Anton"] ] Notes: { push-cache expects 'site to be set at the time this file is included (or you do it). Consequently, this value is stored in cache-site and won't change unless you use change-cache-site. } Public-Functions: [push-cache change-cache-site] ; for include ] ;site: http://anton.idatam.com.au/rebol/ ; <--- change to your site, push-cache expects this context [ ; protective wrap necessary for include framework cache-site: site change-cache-site: func [new-cache-site [url!]][cache-site: new-cache-site] push-cache: func [ "copy (push) a file from local directory into the public cache" file "file you would like to copy" /quiet "don't print anything" /local local-dir clean-file common-path cache-file info1 info2 hash version1 version2 copy-file report ][ report: func [arg] either quiet [[]][[ print arg ]] ;report ["cache-site:" cache-site] local-dir: join system/options/home %local/ copy-file: has [err][ make-dir/deep first split-path cache-file ; ensure directory exists either error? set/any 'err try [ write cache-file read/binary clean-file ][ report "problem writing the file" if not get-modes cache-file 'owner-write [ report join form cache-file " has false owner-write file-mode. (File is read-only.)" ] ][ report "wrote file successfully" ] ] if not exists? file [report "file not found" return] clean-file: clean-path file ; get absolute path to file either found? common-path: find/tail clean-file local-dir [ ; file has to be under local-dir either dir? clean-file [ report "it's a directory - no action taken" ][ cache-file: path-thru cache-site/:common-path report ["cache-file:" cache-file] ; copy file here either exists? cache-file [ report [common-path "already exists"] info1: info? clean-file info2: info? cache-file either all [ info1/size = info2/size (checksum read/binary clean-file) = (checksum read/binary cache-file) ][ ; contents same report "files the same - no action" ][ ; contents different hash: 2 * (to-integer none? script? clean-file) + (to-integer none? script? cache-file) switch hash [ 0 [ ; both are scripts if (version1: version? clean-file) > (version2: version? cache-file) [ report "newer version, copying" copy-file ] if version1 = version2 [ ; file contents different, but versions the same either info1/date > info2/date [ report "newer date, copying" copy-file ][ ; ignore report "cache file has same date or newer - no action" ] ] ; (if older version - no action) ] 3 [ ; neither are scripts either info1/date > info2/date [ report "newer date, copying" copy-file ][ ; ignore report "cache file has same date or newer - no action" ] ] 1 [ report "warning - one file is a script, the other not - no action." ] 2 [ report "warning - one file is a script, the other not - no action." ] ] ] ][ report "not in the cache, copying" copy-file ] ] ][ report ["file not under" local-dir] ] ] ] ; end of context -------- end -----------

Hi, it's annoying me too. I still have a linker tool on my to-do list. That will scan through my library files and make a textual include of the used functions into the final source-code file (static linking ;-)).

bad.) Thaks a lot. Can you send it to my private address as attachment, than I don't have to deal with the linebreak stuff? Thanks! Robert

Hi, there is the makedocpro.html file which discribes most of the new features. Further I have a document describing the new features as well. That's the doc I use to test theses features. I need to publish this. Anyhow, I welcome people who want to help me writing a good documentation for make-doc-pro. Every little piece helps, so if someone just want to suggest a good section structure, examples, etc. let me know. Robert

Hi, yep correct. I fixed it. See my Rebsite. Robert

Have a look at my Rebsite there you will find some docs. Robert

instead of =url iam thinking about here you get all the nice stuff: http://rebol.com what do you think? -volker

Hi, this is OK if you want to include the link directly but how do you specify a text that should appear instead of the link than? Robert

here you get all the nice stuff: http://rebol.com => <a href="http://rebol.com">here you get all the nice stuff</a> ;) -volker

specify a text that should appear instead of the link than? Something like this: My "web site" http://valley.150m.com/. which converts to: My [a href="http://valley.150m.com/"]web site[/a]. I've changed the tags to square brackets. It works for my eText dialect, which converts plain text to HTML. Have a look at my site where there's lots and lots of examples of it at work. Andrew Martin ICQ: 26227169 http://valley.150m.com/

Ok I see ;-). Anyway, this implies that the URL stuff is a single line. Howabout this for inlining URLs as well: This is an exmaple of an inline URL. Have a look at the =url http://rebol.com Rebol Homepage ASAP. What do you think? Robert

Yes that's an other possibility. Mentioning the =url tag explicit makes parsing a lot easiert and you know exactly what the meaning is. How you handle URLs with spaces in them? What do you do if you want exactly the same text to appear in the document? You have somehow to escape it... Robert

http://rebol.com "Rebol Homepage" ASAP. Yuck! :( This is much nicer: This is an example of an inline URL. Have a look at the "Rebol Homepage" http://rebol.com ASAP. Why the need to have special markup codes like "=url" when careful use of white space gets the same result? Andrew Martin ICQ: 26227169 http://valley.150m.com/

Well, since our two approaches forked you followed the whitespace and implicit meaning of things and I followed in the direction of make-doc and explicit markup mentioning. I don't think it's to hard to write =url. I'm more a friend of clear syntax and don't like to interpret to much from the context. Robert

Links are like doors. They should be obvious visible. In browser i have different colors, so they can be inlined. In text i like them emphasized by separate lines. also i like a paragraph which i can read without interruption, then one with links. doors are in the walls, not in the working area. link-doors between chair and coffeecup, and the =url, breaks my reading-flow. so here is a link: http://www.rebol.com and another not so important (not checked): http://www.somewhere.com Of course all is IMHO ;) Volker

exactly what the meaning is. That's right.

The same way browsers and Rebol handle them. A space stops the parsing of a URL. If there's meant to be a space in the URL, then substitute " " for the space.

** Script Error: with has no value ** Near: with Spaces.txt

== http://www.rebol.com/File with Spaces.txt If I type a URL with spaces into my MS Internet Explorer, it does changes spaces to . I hope that helps! Andrew Martin ICQ: 26227169 http://valley.150m.com/

This requires the users to know the encoding scheme for URLs. Further the emitter must decide how to convert this URL encoding for the output device. For example if I emit PDF I don't want to have in the output maybe better to underline the complete link, or wrap it in " chars. IMO there is no silver bullet to solve this. Adding minimal user-friendly syntax can help a lot... and that's what I'm trying to do in make-doc-pro (with being compatible to make-doc). Robert

Part of the problem with all these "pick the url from the data stream" suggestions is that it is not trivial to do. All of the following _could_ be URLs: http://www.rebol.com www.rebol.com ftp://rebol.com/a-script.r /a-folder/a-subfolder/a-document.txt /a-folder/a-doc.htm#a-fragment /x And, for any of them in a document, they may be meant to be quoted or meant to be a clickable link. Basically the only 100% accurate rule is: "it's a clickable URL if I mark it up as such". Robert's =url may not be perfect as it requires the URL to be on a line by itself, and it needs embedded spaces to be coded as , but it does the job. Sunanda.

===a little parser-part url-line: func [line /local text-end text url-start url rest] [ either parse line [ some [to ": " text-end: skip] url-start: ( text: copy/part line text-end either all [ not error? try [set [val rest] load/next url-start] find reduce [url! file!] type? val ] [ url: val ] [ rest: head line ] ) :rest ] [ rejoin [""build-tag compose [a href (url)] text </a>] ] [ line ] ] ;--samples foreach line reduce [ "some text pointing to: http://somewhere.dom garbage" "no url at all" "no real url: nowhere" "nothing behind: " "multiple colons: this and: that" "multiple colons: this and: that and url: http://somewhere.dom" "some text pointing to: http://somewhere.dom" {some text pointing to: %"http://somewhere/ with space.dom"} {a file-link demo: %index.html} ] [ print[">>" line newline "==" url-line line ] ] ;--no real urls inside

== some text pointing to: http://somewhere.dom garbage

== no url at all

== no real url: nowhere

== nothing behind:

== multiple colons: this and: that ;--with urls

== <a href="http://somewhere.dom">multiple colons: this and: that and url</a>

== <a href="http://somewhere.dom">some text pointing to</a>

== <a href="http://somewhere/ with space.dom">some text pointing to</a>

== <a href="index.html">a file-link demo</a>

grretings Volker

Hi, it will be possible to use it inline in one of the next versions. The URL part can be put into " and " if it contains spaces, so no need to use . =url "http://my url with spaces/" Test URL Most likely I will follow the same pattern for inlined URLs like: This is a text =url http://www.rebol.com "Rebol Homepage" with an embedded URL. Or something like this: This is a text =url "http://my url with spaces/" "My Website" with an embedded URL. Robert

As a user, I would be happy if it caught the easy ones without any intervention on my part. Other tools can do this, and url! is a native type that REBOL can identify, so there's virtually no cost there (though I haven't looked at the MDP parser so maybe there is). If I have some non-standard url, or new url formats are rolled out, then having an explicit way to mark them is a good thing to have around. For the normal case, though, why should I have to mark them up? I'm not writing the parser so I won't complain either way. :) --Gregg

I completely agree with your last comment there -- it's a very civilised way of developing code. We (the potential users) can request features or debate enhancements. But the developer (Robert in this case) has the vision for the code, and can add features as fits that vision. And, of course, the beauty of a language like Rebol is that us users can tweak the code to fit our needs -- my version of MDP has a couple of such tweaks that I've passed to Robert; he's free to incorporate them in the base code or not. On the specific point about interpreting URLs I'm still with Robert.... MDP is a "lightweight" mark-up language (light compared to HTML or XML), and all mark-ups require some indicator to distinguish them from plain content. To make an exception for URLs is, I think, non-orthogonal, and potentially confusing. Especially if the definition of what counts as a "normal case" URL changes between editions. Maybe an industrial-strength version of MDP will have a user-call function -- add your own function at this point to post-process the MDP output. URL highlighting would be an obvious add-on here for anyone who wants it. Sunanda.

Hi, first you can complain about anything in MDP you want :-)) I take it sportily.

Second, I'm really happy about anyone thinking for new features etc. it might take some time to get them implemented. The hardest part is to keep focused on the big picture (vision). Some features are very specific and I try to tranform them to a more general/generic concept so that most of you can use it and gain value out of them.

That's right. And sending me the code of your enhancements makes it easier for me to get the change integrated :-)). And I hope I don't miss anything from you. I'm using the bug-reporter on my IOS server to keep track of everything. So, I hope I got included all your change requests in the current version :-|

Thanks.

Yep. And if possible without a lot of exception handling. This makes the code (parser) and sometimes the emitter non-maintainable. I rewrote MDP because I reached such a state. The new code is much cleaner and follows the same concept whereevery possible. It's not perfect yet but good enough.

This should be already possible. You can hook into MDP at several levels: 1.) Replace a parser rule and emit an existing intermediate format or a new one. 2.) Replace the emitting function with something you like. 3.) Implement an emitting function for the new intermediat command. The intermediate format should be quite obvious. At the end there is a commented line, just uncomment to see the stack. Robert