Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-27 Thread michael . vancanneyt



On Thu, 27 Jan 2011, Max Vlasov wrote:


On Wed, Jan 26, 2011 at 3:06 PM, michael.vancann...@wisa.be wrote:








For example, can I rapidly get all function descriptions with space as
part of the description, which could subsequently be filtered (e.g. if I
were interested in text operations I could quickly ignore anything with
file in the description)?
...
That's somewhat abbreviated so that it's not screwed too badly by wrap.



Creating such an index requires additional keywords, which are simply not
present now.

That said, I am still looking for a good search engine *written in Object
pascal* for the FPC html docs. Any hints/tips are appreciated.



Michael, are the docs hand-prepared or automatically generated from some
db/xml?


They are generated from fpdoc XML files.


If the former, can it be safely assumed that the url
 
http://www.freepascal.org/docs-html/rtl/{unitname}/{functionname}.htmlhttp://www.freepascal.org/docs-html/rtl/%7Bunitname%7D/%7Bfunctionname%7D.html
is standardized?


Yes, it is safe to assume so. I have no plans to change this :-)

Michael.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-27 Thread michael . vancanneyt



On Thu, 27 Jan 2011, Sven Barth wrote:


Am 26.01.2011 13:06, schrieb michael.vancann...@wisa.be:

That said, I am still looking for a good search engine *written in Object
pascal* for the FPC html docs. Any hints/tips are appreciated.


I have never tested it, but I once stumpled upon the Ioda search engine ( 
http://ioda.sourceforge.net/ ) which fulfills your criteria and also is 
already compileable by FPC (according to the author).


I know this engine, but the sources are not available. Not for download, not
from CVS.

Michael.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-27 Thread Mark Morgan Lloyd

michael.vancann...@wisa.be wrote:

On Thu, 27 Jan 2011, Sven Barth wrote:


Am 26.01.2011 13:06, schrieb michael.vancann...@wisa.be:
That said, I am still looking for a good search engine *written in 
Object

pascal* for the FPC html docs. Any hints/tips are appreciated.


I have never tested it, but I once stumpled upon the Ioda search 
engine ( http://ioda.sourceforge.net/ ) which fulfills your criteria 
and also is already compileable by FPC (according to the author).


I know this engine, but the sources are not available. Not for download, 
not

from CVS.


I don't know whether this is complete but I note 
http://sourceforge.net/projects/ioda/files/ioda/1.3/ioda-1.3-src.tar.bz2/download


--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk

[Opinions above are the author's, not those of his employers or colleagues]
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-27 Thread Sven Barth

Am 27.01.2011 10:14, schrieb Mark Morgan Lloyd:

michael.vancann...@wisa.be wrote:

On Thu, 27 Jan 2011, Sven Barth wrote:


Am 26.01.2011 13:06, schrieb michael.vancann...@wisa.be:

That said, I am still looking for a good search engine *written in
Object
pascal* for the FPC html docs. Any hints/tips are appreciated.


I have never tested it, but I once stumpled upon the Ioda search
engine ( http://ioda.sourceforge.net/ ) which fulfills your criteria
and also is already compileable by FPC (according to the author).


I know this engine, but the sources are not available. Not for
download, not
from CVS.


I don't know whether this is complete but I note
http://sourceforge.net/projects/ioda/files/ioda/1.3/ioda-1.3-src.tar.bz2/download


This should be the complete and latest one (I have attached that in the 
other mail). Must have been my internet connection, cause I couldn't 
download it some minutes ago. ^^


Regards,
Sven
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-27 Thread michael . vancanneyt



On Thu, 27 Jan 2011, Sven Barth wrote:


Am 27.01.2011 10:14, schrieb Mark Morgan Lloyd:

michael.vancann...@wisa.be wrote:

On Thu, 27 Jan 2011, Sven Barth wrote:


Am 26.01.2011 13:06, schrieb michael.vancann...@wisa.be:

That said, I am still looking for a good search engine *written in
Object
pascal* for the FPC html docs. Any hints/tips are appreciated.


I have never tested it, but I once stumpled upon the Ioda search
engine ( http://ioda.sourceforge.net/ ) which fulfills your criteria
and also is already compileable by FPC (according to the author).


I know this engine, but the sources are not available. Not for
download, not
from CVS.


I don't know whether this is complete but I note
http://sourceforge.net/projects/ioda/files/ioda/1.3/ioda-1.3-src.tar.bz2/download


This should be the complete and latest one (I have attached that in the other 
mail). Must have been my internet connection, cause I couldn't download it 
some minutes ago. ^^


Duh... how could I have missed that ?

Thanks, I downloaded it. Let's see what we can do with this.

Michael.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


[fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Mark Morgan Lloyd
As a comparative beginner, I still find myself stumbling across RTL 
functions that I've previously hand-coded.


Is there an easy way of generating a complete permuted index from the 
one-line descriptions of the functions in the RTL (and optionally FCL 
and LCL)?


For example, can I rapidly get all function descriptions with space as 
part of the description, which could subsequently be filtered (e.g. if I 
were interested in text operations I could quickly ignore anything with 
file in the description)?


Copy2Space Returns  first space character
Copy2SpaceDel  Deletes  first space character
DelSpace   Delete   a space from
DelSpace1  Reduces of space characters
DelSpace1  Reduces  1 space character
IntToBin   Converts inserting spaces at
IsEmptyStr Check disregaring whitespace characters

That's somewhat abbreviated so that it's not screwed too badly by wrap.

--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk

[Opinions above are the author's, not those of his employers or colleagues]
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread michael . vancanneyt



On Wed, 26 Jan 2011, Mark Morgan Lloyd wrote:

As a comparative beginner, I still find myself stumbling across RTL functions 
that I've previously hand-coded.


Is there an easy way of generating a complete permuted index from the 
one-line descriptions of the functions in the RTL (and optionally FCL and 
LCL)?


Shortly said: no.



For example, can I rapidly get all function descriptions with space as part 
of the description, which could subsequently be filtered (e.g. if I were 
interested in text operations I could quickly ignore anything with file in 
the description)?


Copy2Space Returns  first space character
Copy2SpaceDel  Deletes  first space character
DelSpace   Delete   a space from
DelSpace1  Reduces of space characters
DelSpace1  Reduces  1 space character
IntToBin   Converts inserting spaces at
IsEmptyStr Check disregaring whitespace characters

That's somewhat abbreviated so that it's not screwed too badly by wrap.


Creating such an index requires additional keywords, which are simply not
present now.

That said, I am still looking for a good search engine *written in Object
pascal* for the FPC html docs. Any hints/tips are appreciated.

Michael.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Marco van de Voort
In our previous episode, michael.vancann...@wisa.be said:
  IntToBin   Converts inserting spaces at
  IsEmptyStr Check disregaring whitespace characters
 
  That's somewhat abbreviated so that it's not screwed too badly by wrap.
 
 Creating such an index requires additional keywords, which are simply not
 present now.

Categorization of routines (string,file etc) would be a good place to start.
I have thought about this before (for chm), but while the CHM format
supports this (building e.g. a list of string routines over multiple
independantly compiled CHMs), the Windows viewer's support for this is too
basic, and the *nix viewers are even worse.

It seems to be functionality intended for MSDN that never was finished
because of the move to later formats.
 
 That said, I am still looking for a good search engine *written in Object
 pascal* for the FPC html docs. Any hints/tips are appreciated.

CHM has a basic phrase indexer, and it operates on the html. The search part
is maybe part of the viewer, and one would have to see how complex it is. 
(Iow if the complexity is in index or search, if the search is relative
simple one could look at kchmviewer)


___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Mark Morgan Lloyd

Marco van de Voort wrote:

In our previous episode, michael.vancann...@wisa.be said:

IntToBin   Converts inserting spaces at
IsEmptyStr Check disregaring whitespace characters

That's somewhat abbreviated so that it's not screwed too badly by wrap.

Creating such an index requires additional keywords, which are simply not
present now.


Categorization of routines (string,file etc) would be a good place to start.
I have thought about this before (for chm), but while the CHM format
supports this (building e.g. a list of string routines over multiple
independantly compiled CHMs), the Windows viewer's support for this is too
basic, and the *nix viewers are even worse.

It seems to be functionality intended for MSDN that never was finished
because of the move to later formats.
 

That said, I am still looking for a good search engine *written in Object
pascal* for the FPC html docs. Any hints/tips are appreciated.


CHM has a basic phrase indexer, and it operates on the html. The search part
is maybe part of the viewer, and one would have to see how complex it is. 
(Iow if the complexity is in index or search, if the search is relative

simple one could look at kchmviewer)


I used something called Perlfect Search for a while- which might fulfil 
Michael's criterion if Lazarus could handle Perl plugins :-)


As shipped though I think it was oriented towards word (rather than 
phrase) indexing, and it had a limit of 64K files. 
http://www.perlmonks.org/index.pl?node_id=27509


--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk

[Opinions above are the author's, not those of his employers or colleagues]
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread michael . vancanneyt



On Wed, 26 Jan 2011, Marco van de Voort wrote:


In our previous episode, michael.vancann...@wisa.be said:

IntToBin   Converts inserting spaces at
IsEmptyStr Check disregaring whitespace characters

That's somewhat abbreviated so that it's not screwed too badly by wrap.


Creating such an index requires additional keywords, which are simply not
present now.


Categorization of routines (string,file etc) would be a good place to start.
I have thought about this before (for chm), but while the CHM format
supports this (building e.g. a list of string routines over multiple
independantly compiled CHMs), the Windows viewer's support for this is too
basic, and the *nix viewers are even worse.

It seems to be functionality intended for MSDN that never was finished
because of the move to later formats.


That said, I am still looking for a good search engine *written in Object
pascal* for the FPC html docs. Any hints/tips are appreciated.


CHM has a basic phrase indexer, and it operates on the html. The search part
is maybe part of the viewer, and one would have to see how complex it is.
(Iow if the complexity is in index or search, if the search is relative
simple one could look at kchmviewer)


Secretly, I am still waiting for Michael Hess to open source his IDKSM indexer. 
It was/is exactly what we need... Maybe the Lazarus devels can exert some

mild pressure :-)

Michael.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread michael . vancanneyt



On Wed, 26 Jan 2011, Mark Morgan Lloyd wrote:


Marco van de Voort wrote:

In our previous episode, michael.vancann...@wisa.be said:

IntToBin   Converts inserting spaces at
IsEmptyStr Check disregaring whitespace characters

That's somewhat abbreviated so that it's not screwed too badly by wrap.

Creating such an index requires additional keywords, which are simply not
present now.


Categorization of routines (string,file etc) would be a good place to 
start.

I have thought about this before (for chm), but while the CHM format
supports this (building e.g. a list of string routines over multiple
independantly compiled CHMs), the Windows viewer's support for this is too
basic, and the *nix viewers are even worse.

It seems to be functionality intended for MSDN that never was finished
because of the move to later formats.


That said, I am still looking for a good search engine *written in Object
pascal* for the FPC html docs. Any hints/tips are appreciated.


CHM has a basic phrase indexer, and it operates on the html. The search 
part
is maybe part of the viewer, and one would have to see how complex it is. 
(Iow if the complexity is in index or search, if the search is relative

simple one could look at kchmviewer)


I used something called Perlfect Search for a while- which might fulfil 
Michael's criterion if Lazarus could handle Perl plugins :-)


The criterion is: *written* in object pascal. 
Not 'callable from Object pascal' :-)


And I will rather spend an eternity in hell than use a perl tool... ;)

Michael.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Marco van de Voort
In our previous episode, Mark Morgan Lloyd said:
  simple one could look at kchmviewer)
 
 I used something called Perlfect Search for a while- which might fulfil 
 Michael's criterion if Lazarus could handle Perl plugins :-)

Pascal, and pascal only. We might make exceptions for extremely common C
libraries on Unix but that is it.

If we were lenient with that, you now had to install Perl, Python, Ruby,
PHP, Java, .NET and LUA as well as cygwin just to be able run FPC/Lazarus on
Windows.
 
Programming is about programming, not about smartshopping. (unless you can
find somebody crazy enough to foot the bill for integration and support)

 As shipped though I think it was oriented towards word (rather than 
 phrase) indexing, and it had a limit of 64K files. 
 http://www.perlmonks.org/index.pl?node_id=27509

There should me more than enough decent C libs to port. No need to go to
extremes like Perl.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Marco van de Voort
In our previous episode, michael.vancann...@wisa.be said:
  I used something called Perlfect Search for a while- which might fulfil 
  Michael's criterion if Lazarus could handle Perl plugins :-)
 
 The criterion is: *written* in object pascal. 
 Not 'callable from Object pascal' :-)
 
 And I will rather spend an eternity in hell than use a perl tool... ;)

I still wakeup screaming in the night sometimes because of the last one. The
little 1.5MB script that converted Latex to html. :-)
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread michael . vancanneyt



On Wed, 26 Jan 2011, Marco van de Voort wrote:


In our previous episode, michael.vancann...@wisa.be said:

I'm sure that somebody familiar with XML (which I'm afraid doesn't include
me) could knock out something half-way decent in a few minutes.

Or is the problem actually XML in this case, i.e. it's obvious that we want
to extract the short description but that leaves an uncertain context?


The short description is not a problem. But the Returns Deletes Reduces
etc ?


IMHO that is overkill. It only works comprehensively for a handful routines.


Well, a set of tags (not in XML sense, but in keyword sense) 
for each routine would be sufficient. This can be added easily, 
but the main problem is updating the many thousands of existing

identifiers.



Categorizing routines, with subcategs if need be will be trouble enough,
specially if you want to do it future proof (and also be able to flag the static
method approach that Delphi current favours)


OK, that was totally chinese. I'll have to ask my chinese pillow to translate 
this.

Michael.
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Marco van de Voort
In our previous episode, michael.vancann...@wisa.be said:

  Categorizing routines, with subcategs if need be will be trouble enough,
  specially if you want to do it future proof (and also be able to flag the 
  static
  method approach that Delphi current favours)
 
 OK, that was totally chinese. I'll have to ask my chinese pillow to
 translate this.

I mean procedures and functions is not enough. It must be possible to add
static methods to this categorization system too (to be visible in the same
overviews as normal functions), because newer delphi versions stuff all new
routines as static methods into classes, and sooner or later we will have to
deal with that too.

With subcateg(ory)s I meant that simply one dimensional tagging might not be
enough.  You might want to have a broad category with several smaller ones
nested in it.

I hope you and your chinese pillow will be allright :-)
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Mark Morgan Lloyd

Marco van de Voort wrote:

In our previous episode, michael.vancann...@wisa.be said:


Categorizing routines, with subcategs if need be will be trouble enough,
specially if you want to do it future proof (and also be able to flag the static
method approach that Delphi current favours)

OK, that was totally chinese. I'll have to ask my chinese pillow to
translate this.


I mean procedures and functions is not enough. It must be possible to add
static methods to this categorization system too (to be visible in the same
overviews as normal functions), because newer delphi versions stuff all new
routines as static methods into classes, and sooner or later we will have to
deal with that too.

With subcateg(ory)s I meant that simply one dimensional tagging might not be
enough.  You might want to have a broad category with several smaller ones
nested in it.


The problems with tagging etc. are that (a) somebody's got to do the 
work retroactively and (b) there's a temptation to fit things into 
existing categories (should Soundex support go into sorting, hashing, or 
its own category?).


As a user with limited experience, I'd settle for something that 
generated a simple KWIC list, even if I had to manually exclude some 
possibilities.


--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk

[Opinions above are the author's, not those of his employers or colleagues]
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Andrew Haines
On 01/26/11 07:41, Marco van de Voort wrote:
 CHM has a basic phrase indexer, and it operates on the html. The search part
 is maybe part of the viewer, and one would have to see how complex it is. 
 (Iow if the complexity is in index or search, if the search is relative
 simple one could look at kchmviewer)
 
 

As far as I remember the chm search only indexes whole words. So
searching for space would not return LotsOfSpace I'm not sure if it
would find SpaceFoo since it starts with space.

Does changing the indexer to extract words from capitalized words make
sense?  i.e. DoSomethingCool becomes do, something, cool ?

Regards,

Andrew Haines
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal


Re: [fpc-pascal] Permuted index (KWIC) of function descriptions

2011-01-26 Thread Max Vlasov
On Wed, Jan 26, 2011 at 3:06 PM, michael.vancann...@wisa.be wrote:






 For example, can I rapidly get all function descriptions with space as
 part of the description, which could subsequently be filtered (e.g. if I
 were interested in text operations I could quickly ignore anything with
 file in the description)?
 ...
 That's somewhat abbreviated so that it's not screwed too badly by wrap.


 Creating such an index requires additional keywords, which are simply not
 present now.

 That said, I am still looking for a good search engine *written in Object
 pascal* for the FPC html docs. Any hints/tips are appreciated.


Michael, are the docs hand-prepared or automatically generated from some
db/xml?

If the former, can it be safely assumed that the url
  
http://www.freepascal.org/docs-html/rtl/{unitname}/{functionname}.htmlhttp://www.freepascal.org/docs-html/rtl/%7Bunitname%7D/%7Bfunctionname%7D.html
is standardized?

if the latter, is it possible to download this data from somewhere for
experimenting?

Thanks

Max Vlasov
___
fpc-pascal maillist  -  fpc-pascal@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-pascal