text parsing in a field

[ken]

RE: text parsing in a field

yes you are right but such cases are very rare and anyway I check them carefully before saving.
Unfortunately, not all correspondents use the semicolon.
anyway it would be much easier to make some minor corrections instead to copy and paste thousands of lines.

regards,
Corneliu

[ken]

RE: text parsing in a field

yes, it works now but it does not parse website in foll format

Website: www.crownint.biz

I think parsing of www. is enough to assume that it is a web site even without http

Thanks vm for your efforts.

regards
Corneliu

[ken]

RE: text parsing in a field

Hi Steve,

I have done a lot of testings last night and the routine parses correctly in most situations. However most people indicate their web site address in the format www.****.****
not http://www.******
I have modified the events routine like that
webs = filter_string(memtest:mem.value,"www.")
and it works in most situations

I have tried to use a regular expression here to accomodate both alternatives but is does not work
webs = filter_string(memtest:mem.value,"www.|http")

Is it possible to use a regular expression here and how can it be done?

Also I need a checkbox on the form D0NT_PARSE when checked
the text from memo field not to be parsed.
Maybe in some situations parsing is not needed.

Also in case I use the short form www.****.**** is it possible in the WEB field to add the correct prefix "http" so that in the database the web site address is stored in the correct format?

Regards,
Corneliu

[ken]

RE: text parsing in a field

Hi Steve,

I discovered a case where parsing is not done corectly like in foll example:

MSN: [email protected]

E-MAIL: [email protected]
[email protected]

in this case only the second email is parsed correctly but
the result is forwarded to EMAIL3 instead of EMAIL1 or EMAIL2

Regards,
Corneliu

[ken]

RE: text parsing in a field

Hi Steve,

regarding my previous msg reporting parsing errors, I think it has to do with empty lines because the program counts all lines even the empty ones.
I think it is a good idea at the beginning of the routine to delete all empty line and all empty spaces at the beginning of the lines.
This is would be useful anyway for formating the text.

regards,
Corneliu

[Steve Andrews]

RE: text parsing in a field

Corneliu,

Attached is an update to the test application. You are right that it is difficult or impossible to check for one substring or another with one call to filter_string(). However, we can concatenate multiple calls into one final list.

The only problem with that approach is that any line that included both substrings would appear twice. We can alleviate that situation with the word_unique() function. Thus, the new attachment has the following:

?st
= An Email:: som@none
[email protected]
email: not@here
Web http://www.unimart.cl

webs = word_unique(filter_string(st,"http:")+crlf()+filter_string(st,"www.")+crlf()+filter_string(st,".com")+crlf()+filter_string(st,".net"))

?webs
= Web http://www.unimart.cl
[email protected]

If not for the word_unique() function, the last line would be included twice (once for the "http:" substring and again for "www."). I threw in the ".net" and ".com" substrings for the cases you mentioned which did not get parsed correctly before.

Steve

[ken]

RE: text parsing in a field

Hi Steve,

First of all I want to thank you very much for your efforts in solving this case.
This new test application is better but however it does not catch all www.*** sites

I will provide below some examples not parsed by the program

www.brouwership.de
WEB-SITE: WWW.AZOVBALT.RU

A strange case is the foll

[email protected]

[email protected]

[email protected]

where 2 of the emails are parsed correctly and one is parsed as a web site

after deleting +crlf()+filter_string(st,".com")
from the routine, the web site dissapear but still only 2 emails are parsed correctly
After deleting the empty lines, all are parsed correctly.

I think the empty lines should be deleted before the start of parsing process.

Also there is no need to check for 2 web sites, because it is unusual that a company has 2 web sites. so you can delete checking for the second web site.

It would be useful to add instead a field called COMPANY
and the rule is that any text is written on the first line
is parsed as "company"
Of course, this line should not be empty, not contain numbers, or emails or web sites.

Thanks and regards,
Corneliu

[Steve Andrews]

RE: text parsing in a field

Corneliu,

You are right on two accounts. I should not have put the ".com" substring in for the web addresses. DUH? And also there are blank lines generated from the concatenation.

As for the blank lines, I removed them with the statement:

webs = word_subtract(webs,"",crlf())

In addition, I noticed that more than one trailing blanks could pose a problem. Hence, I added:

webs = *for_each(foo,alltrim(foo),webs)

I hope this helps out.

Steve

[ken]

RE: text parsing in a field

Congratulations!. It works much better than before now.
I had however to delete .net from the concatenation string
to avoid emails ending in .net to be parsed as web sites

one problem still remains, the routine can not detect the the space after the email or web site and is parsing everything after @ as email even if it is separated by a space.

regards,
Corneliu

[Steve Andrews]

RE: text parsing in a field

Cornneliu,

This brings up another consideration about the way I set it up. Namely, I assumed that the dta you wanted would be the last space-delimited word on a selected line. If you have spaces sometimes thrown in, it will be difficult to determine what pieces you want to parse out.

line = "my email: [email protected]"

?word(line,-1," ")
= "[email protected]"

line = "my email: solar@ solarshipping.com"

?word(line,-1," ")
= "solarshipping.com"

Thus, for my approach to work, the substring which you want to retrieve must be the last substring on the line separated by a space. How else would you determine whether you wanted the last substring or the last two?

Steve

[martinwcole]

RE: text parsing in a field

On a routine similar to this, I have a mirror table I call exceptions - all the ones I can't parse I copy to the exceptions table and send an alert so they can manually deal with the exceptions

[ken]

RE: text parsing in a field

No, I meant is that in a line like this

[email protected] xyz aaaa

the parsing result is aaaa

in a line like this

[email protected] xyz

the parsing result is xyz

so practically the email must be the last word on the line to be parsed correctly.

I thought that maybe there is a way to detect the space between the email and the next word on the line.

Regards,
Corneliu

[ken]

RE: text parsing in a field

I would be interested to see a sample of your routine may I can get some ideas

Regards,
Corneliu

[Steve Andrews]

RE: text parsing in a field

Corneliu,

I didn't realize you may have other text following the desired substring. One approach may be to replace the word(txt,-1," ") function call with another filter_string() call:

txt = "vsvs [email protected] xyz aaaa"

?word(txt,-1," ")
= "aaaa"

?filter_string(txt,"@"," ")
= "[email protected]"

I can take a closer look at this later.

Steve

[ken]

RE: text parsing in a field

yes it may happen that the email is not the last word in the line. Take you time. I do not want to abuse of your kindness. You have also other things to do.
Maybe you can take a look at it later.
Anyway you alredy did a very fine job!
Thanks a lot!

Sincerely yours,
Corneliu