Re: text parsing

McKirahan wrote:
"Carolyn Marenger" <cajunk@xxxxxxxxxxxx> wrote in message
news:81d7b$47973b05$cf70133e$360@xxxxxxxxxxxx
McKirahan wrote:
"Carolyn Marenger" <cajunk@xxxxxxxxxxxx> wrote in message
news:7c0f$4795ea54$cf70133e$1079@xxxxxxxxxxxx
McKirahan wrote:
"Carolyn Marenger" <cajunk@xxxxxxxxxxxx> wrote in message
news:74fb1$479501d1$cf70133e$7458@xxxxxxxxxxxx
Can someone point me in the direction of some good documentation on
text
parsing?

I want to take a bunch of text files (rtf), read them in and dump the
contents in a database. The files are effectively a flat file
database,
with I suspect some fairly intricate programming needed to process
the
files. Unfortunately, they are laid out for human readability, not
data
conversion.
A few questions.

How many is a "bunch"?
What would the target database be -- MySQL?
What table and column structures do you envision?
Perhaps simply a single table with two columns:
filename (key) and a memo field containing the data?
What is the purpose behind doing this?

A few answers

A bunch is about a dozen. Basically one large file that was broken
into
sixteen subsets, following the initial letter for each record.

The target database would be MySQL

I haven't looked too closely at the data, but I think one main table
with a few linked tables for those cases where there may be more than
one piece of data for a category. There are about 25 categories to
each
record. Eventually there would be additional structure added around
the
imported data, but that isn't relevant to importing the data itself.
(I
will confirm this before beginning to code.

The purpose: I am a D&D fan and I run games. I would like to be able
to
reference the material and automate much of the process so I don't have
to lug and reference 20lbs of books.
Any chance the RTF files are online so I could look at them?

Perhaps http://www.wizards.com/default.asp?x=d20/article/srd35?
http://www.wizards.com/d20/files/v35/SRD.zip contains 88 RTF files.


Also, I gather, this might be a one-time effort; correct?

Not what you requested but ...

I've developed a VBScript solution that takes the following approach:
for a given folder, each RTF file is opened in MS-Word and saved
as a text file which is opened and read then saved in an MS-Access
database table containing 3 columns: id (AutoNumber), file, data.

Using those 86 RTF files it created a 10MB MS-Access database.

Yes, they are online. Yes, you can look at them. Yes, those are the
files except I only care about the 16 monster files. Yes, this is a one
time effort.

My goal is to create a encounter generation program - where I key in
climate, geography, season, encounter level, time of day, proximity to
civilization, and the application gives me a suggested random encounter
suited to the scenario. For example, if the party was wandering around
the city sewers on a hot summer night, they might encounter a pack of
giant rats being led by a were rat. I would then want the program to
determine how many rats, how many hit points each, and any other
pertinent variable data, including what weapons and treasure the wererat
was carrying and using.

Having the rtfs loaded into a database like your script does, would
enable faster searches, it would not go the next step and perform the
various calculations based on the results of the searches. It is a good
start, but if it has stripped any of the rtf encoding, it may make it
harder to have a script find the various 'fields'.

Thanks, Carolyn

I counted 17 "Monster" prefixed files.

My version creates ".txt" files which do strip "the rtf encoding".

An alternative version creates ".htm" files which retains the
formatting you want; I don't think you really want all of the
"rtf encoding" unless you fully understand the specification:
(search on "rtf specification".)

Perhaps, as an intermediate step, you would like all of the
"Monster" rtfs converted to HTML and made available via
an interface to open one or more for viewing.

As HTML files they consume 7.5MB.


There are a couple of the monster prefixed files that are not listings of monsters but other information, such as monsters as characters. Anyway, exact number of files is not overly important.

I just did a little test, and looking at the files, I think the easiest to work with may indeed be the text file.

Here is an example to illustrate: I am pulling the monster name, type and hit dice from each file format.

in rtf...
{
\par }{\fs36
\par DELVER
\par }\trowd \trgaph108\trleft-108\trbrdrh\brdrs\brdrw10 \trftsWidth1\trautofit1\trpaddl108\trpaddr108\trpaddfl3\trpaddfr3 \clvertalt\clbrdrb\brdrs\brdrw10 \cltxlrtb\clftsWidth1 \cellx1969\clvertalt\clbrdrb\brdrs\brdrw10 \cltxlrtb\clftsWidth3\clwWidth4871
\cellx6840\pard \ql \li0\ri0\nowidctlpar\intbl\faauto\rin0\lin0 {\b\fs20 }{\b\fs19 \cell }{\fs20 Huge Aberration}{\fs19 \cell }\pard \ql \li0\ri0\widctlpar\intbl\aspalpha\aspnum\faauto\adjustright\rin0\lin0 {\fs19 \trowd \trgaph108\trleft-108\trbrdrh
\brdrs\brdrw10 \trftsWidth1\trautofit1\trpaddl108\trpaddr108\trpaddfl3\trpaddfr3 \clvertalt\clbrdrb\brdrs\brdrw10 \cltxlrtb\clftsWidth1 \cellx1969\clvertalt\clbrdrb\brdrs\brdrw10 \cltxlrtb\clftsWidth3\clwWidth4871 \cellx6840\row }\trowd
\trgaph108\trleft-108\trbrdrh\brdrs\brdrw10 \trftsWidth1\trautofit1\trpaddl108\trpaddr108\trpaddfl3\trpaddfr3 \clvertalt\clbrdrt\brdrs\brdrw10 \clbrdrb\brdrs\brdrw10 \cltxlrtb\clftsWidth1 \cellx1969\clvertalt\clbrdrt\brdrs\brdrw10 \clbrdrb\brdrs\brdrw10
\cltxlrtb\clftsWidth3\clwWidth4871 \cellx6840\pard \ql \li0\ri0\nowidctlpar\intbl\faauto\rin0\lin0 {\b\fs20 Hit Dice:}{\b\fs19 \cell }{\fs20 15d8+78 (145 hp)}{\fs19 \cell }\pard \ql

----------
in .html...
<P STYLE="page-break-after: avoid"><FONT SIZE=5>DARKMANTLE</FONT></P>
<TABLE WIDTH=410 BORDER=1 BORDERCOLOR="#000000" CELLPADDING=7 CELLSPACING=0 FRAME=VOID RULES=ROWS>
<COL WIDTH=124>
<COL WIDTH=258>
<TR VALIGN=TOP>

<TD WIDTH=124>
<P CLASS="western">
</P>
</TD>
<TD WIDTH=258>
<P CLASS="western"><FONT SIZE=2>Small Magical Beast</FONT></P>
</TD>
</TR>
<TR VALIGN=TOP>

<TD WIDTH=124>
<P CLASS="western"><FONT SIZE=2><B>Hit Dice:</B></FONT></P>
</TD>
<TD WIDTH=258>
<P CLASS="western"><FONT SIZE=2>1d10+1 (6 hp)</FONT></P>
</TD>
</TR>

---------
in .txt...

DARKMANTLE

Small Magical Beast
Hit Dice:
1d10+1 (6 hp)


--------

So, looking at that and assuming the rest will be similar, the text version looks the easiest to deal with. If document styling such as 'title', 'heading' and 'subheading' had been used, maybe not, but in this case, a new line seems to denote either a field heading or field data. There are exceptions of course - particularly when denoting a category of monster.

That doies bring me a little closer to achievign my goal. Thanks for the assistance. :)

Carolyn
.

Relevant Pages

  • Re: text parsing
    ... What would the target database be -- MySQL? ... Any chance the RTF files are online so I could look at them? ... climate, geography, season, encounter level, time of day, proximity to ... I counted 17 "Monster" prefixed files. ...
    (comp.lang.php)
  • Re: text parsing
    ... What would the target database be -- MySQL? ... Any chance the RTF files are online so I could look at them? ... climate, geography, season, encounter level, time of day, proximity to ... "Monster" rtfs converted to HTML and made available via ...
    (comp.lang.php)
  • Re: text parsing
    ... I want to take a bunch of text files (rtf), read them in and dump the ... What would the target database be -- MySQL? ... A bunch is about a dozen. ... My goal is to create a encounter generation program - where I key in climate, geography, season, encounter level, time of day, proximity to civilization, and the application gives me a suggested random encounter suited to the scenario. ...
    (comp.lang.php)
  • Re: text parsing
    ... I want to take a bunch of text files (rtf), read them in and dump the ... What would the target database be -- MySQL? ... A bunch is about a dozen. ... My goal is to create a encounter generation program - where I key in climate, geography, season, encounter level, time of day, proximity to civilization, and the application gives me a suggested random encounter suited to the scenario. ...
    (comp.lang.php)
  • Re: text parsing
    ... I want to take a bunch of text files (rtf), read them in and dump the ... What would the target database be -- MySQL? ... A bunch is about a dozen. ... My goal is to create a encounter generation program - where I key in climate, geography, season, encounter level, time of day, proximity to civilization, and the application gives me a suggested random encounter suited to the scenario. ...
    (comp.lang.php)