[ic] Re: table drive component template for complex searches
Barry Treahy, Jr.
interchange-users@icdevgroup.org
Wed Jun 12 20:13:01 2002
Barry Treahy, Jr. wrote:
> Ed LaFrance wrote:
>
>> At 08:42 PM 6/10/2002 -0700, you wrote:
>>
>>> Barry Treahy, Jr. wrote:
>>>
>>>> I need to create a component for page construction, preferably
>>>> table driven, what will allow for choices to be made from radio,
>>>> check box, and entered values. There are enough similarities
>>>> between our product lines that being table driven could make this
>>>> task much simpler in the long run.
>>>>
>>>> An subset example from one of our product lines would need these
>>>> types of choices:
>>>>
>>>> Frequency:
>>>> [ ] 26.5 Ghz
>>>> [X] 18.0 Ghz
>>>> [X] 12.4 Ghz
>>>> [ ] 11.0 Ghz
>>>>
>>>> Power Rating:
>>>> [X] 2 watts
>>>> [X] 5 watts
>>>> [ ]O 10 watts
>>>>
>>>> Interface Genders:
>>>> * Male/Female
>>>> O Female/Female
>>>> O Male/Male
>>>>
>>>> Low VSWR: [ ]
>>>>
>>>> Attenuation value: ______
>>>>
>>>> And then with the input from this dynamically created form, a
>>>> search would need to be constructed to scan through the products
>>>> table. A side note, the intent isn't to search for a precise hit
>>>> through a description or title column as individual metric columns
>>>> for each parameter exist, but rather reduce the 30,000 SKU's and
>>>> infinite choice combinations down to a manageable few SKU's and
>>>> then present those SKU's to the prospective buyer. Before I attempt
>>>> to do something that has already been created by someone else who
>>>> wouldn't mind sharing their efforts, has any one done any thing
>>>> like this?
>>>>
>>>> Can anyone provide a push start on this or at least a recommended
>>>> direction to head?
>>>
>>>
>>>
>>> I've been forging ahead on this, perhaps a case of the
>>> blind-leading-ones-self, but so far I have a form establish which is
>>> not data driven as I had hoped, at least not yet :-)
>>>
>>> I'm still trying to figure out how to handle the content selection.
>>> I have the following search:
>>>
>>> <input type=hidden name="mv_searchtype" value="db">
>>> <input type=hidden name="mv_sort_field" value="category">
>>> <input type=hidden name="mv_sort_field" value="sku">
>>> <input type=hidden name="mv_sort_field" value="">
>>> <input type=hidden name="mv_coordinated" value="yes">
>>> <input type=hidden name="mv_search_map" value="
>>> mv_searchspec=FREQMAX
>>> mv_searchspec=POWER
>>> mv_searchspec=VSWR
>>> mv_searchspec=BODY
>>> mv_searchspec=GENDERS
>>> mv_searchspec=INTERFACES
>>> mv_searchspec=DBRANGES">
>>> <input type=hidden name="mv_search_field" value="freqmax">
>>> <input type=hidden name="mv_search_field" value="power">
>>> <input type=hidden name="mv_search_field" value="vswr">
>>> <input type=hidden name="mv_search_field" value="body">
>>> <input type=hidden name="mv_search_field" value="genders">
>>> <input type=hidden name="mv_search_field" value="interfaces">
>>> <input type=hidden name="mv_search_field" value="dbranges">
>>>
>>> by gleaning hints from the archives and docs, it does select product
>>> but the challenge I'm running into is that first, this should be an
>>> AND reduction between the fields, yet the returned results appear to
>>> be an OR.
>>>
>>> Additionally, the water is murkied because within the field such as
>>> POWER, if the person checks that he wants to see both 2 and 5 watt
>>> choices, the search seems to be applying an AND to the field search
>>> where I really need the logic to operate as an OR.
>>>
>>> So, so summarize where I need to reach, each field test if non-black
>>> needs to reduce the results selected in an AND test between
>>> themselves. Within each field, if multiple choices are available,
>>> each choice needs to be considered as an OR test:
>>>
>>> (FREQMAX eq 18 OR FREQMAX eq 12.4) AND (POWER eq 2 or POWER eq 5)
>>> and BODY eq STD
>>>
>>> for example...
>>>
>>> Anyone???
>>>
>>> Barry
>>
>>
>>
>> Given the proliferation of ANDing and ORing, you really should
>> consider going SQL with this - it is a better vehicle for complex
>> searches IMHO. In either case, I would also suggest that you collect
>> all search data as a separate step - evaluate it (which allows you to
>> perform manipulations if needed), then pass it to a search. There are
>> some examples of this in the UI and elsewhere in foundation. A direct
>> search form post is too restrictive I think.
>>
>> - Ed L.
>
>
> Thanks Ed and Paul, you both recommended a collection page and then
> passing on the results. Ed also recommened SQL, which isn't an option
> at the moment, so I'm forging ahead from there.
>
> Before I dive into this development further, I need to ask what may be
> a profoundly dumb question, but here it goes.
>
> Looking at the contents of the CGI values from dump (I've pasted some
> of the values below), I see the results from the multiple check box
> selections delimited by a null. Could I not replace the null with the
> pipe, isn't that the equivalent of OR in a regexp, and then pass that
> value into a loop search? Wouldn't the fact that the nulls are gone
> stop IC from parsing each element into a unique set of 'se/op/sf' for
> scan?
>
>
> Thanks...
>
> Barry
>
> ###### END ENVIRONMENT #####
> ###### CGI VALUES #####
> {
> 'mv_nextpage' => 'ATTsearch',
> 'FREQMAX' => '26\018\012.4\011\08\04\02',
> 'mv_session_id' => 'rIX5IGoR',
> 'INTERFACES' => 'SMA,SMA',
> 'POWER' => '2\05\010\020',
> 'VSWR' => 'LOW',
> 'GENDERS' => 'M,F',
> 'DBRANGES' => '3',
> 'BODY' => 'STD'
> }
>
> ###### END CGI VALUES #####
> ###### SESSION #####
> {
> 'last_error' => 'search error: Limit subroutine creation: bad
> limit creation code in coordinated search, probably search group
> without search specification. at /ibin/interchange/lib/Vend/Search.pm
> line 768.
> ',
> 'last_search' =>
> 'scan/fi=products/se=Attenuator/se=26/se=18/se=12%2e4/se=11/se=8/se=4/se=2/se=2/se=5/se=10/se=20/se=LOW/se=STD/se=M%2cF/se=SMA%2cSMA/op=rm/op=rm/op=rm/op=rm/op=rm/op=rm/op=rm/lb=loop2ATTsearch/sf=prod_group/sf=freqmax/sf=power/sf=vswr/sf=body/sf=genders/sf=interfaces/tf=sort/co=yes/ac=0/ac=0/st=db',
>
Any ideas why this segment is not actually substituting the null for the
pipe?
[set FREQMAX][cgi FREQMAX][/set]
[set POWER][cgi POWER][/set]
[set VSWR][cgi VSWR][/set]
[set BODY][cgi BODY][/set]
[set GENDERS][cgi GENDERS][/set]
[set INTERFACES][cgi INTERFACES][/set]
[perl]
$Scratch->{FREQMAX} =~ s/\0/|/g;
$Scratch->{POWER} =~ s/\0/|/g;
$Scratch->{VSWR} =~ s/\0/|/g;
$Scratch->{BODY} =~ s/\0/|/g;
$Scratch->{GENDERS} =~ s/\0/|/g;
$Scratch->{INTERFACES} =~ s/\0/|/g;
[/perl]
Barry
--
Barry Treahy, Jr * Midwest Microwave * Vice President & CIO
E-mail: Treahy@mmaz.com * Phone: 480/314-1320 * FAX: 480/661-7028