I am playing with indexing an articlenumber in Elasticsearch.
Here i provide a working example:
https://found.no/play/gist/557202b3542be157d813
i dont understand why i get a different score if i change the value for "product_number.search"
if i change the value from "ak454" to "ak 454" the score changes.
i thought that if i am using a search_analyzer the value "ak 454" will be transformed to "ak454" (its mapped using the searchable_id).
you can also look at the analyses tab to see my tokenizer:
https://found.no/play/gist/557202b3542be157d813#analysis
thanks.
The term-query (and filter) does not do any text analysis.
The match-query does, and can achieve what you want.
I adapted the example: https://found.no/play/gist/2de967d844c5fbc14d2f
Setting explain to true is very useful when working with problems like this, as you see exactly what Lucene is doing when it's scoring.
Related
I saw in cytoscape application we have features like find and filter by keywords and degree. I tried a workaround following the original docs. Here you can see the demo webdemo.intolap.com/cytoscape (view-source for the source code or snippet). The filter works well partially. Example, "apple" will display apple and it's connected nodes (1st level) just what I am looking for.
But the problem I am facing is about resetting the graph and filter again with a
different keyword. It seems the filter function does not work after the text box is cleared and then keyed in a different keyword.
I mean when I clear the text box, it resets the graph to original which is correct. I did that using an init() function which reinstates the graph. But then if I search for "Ball" filter does not work. Any help please. Thanks!
actually there is a reasonably good explanation in the official docs here, but to be honest, I too struggled with this feature at first:
Basically, you can filter the specific collection you want to search by just inserting a filter query. So if you want to filter all nodes, you can use this:
cy.nodes(filterQuery);
If you want to filter all elements, just call this:
cy.elements(filterQuery);
If you want to make it easy, you can use this short version (short for cy.filter(...)):
cy.$(filterQuery);
The filter query itself is not that hard, you can do this (assuming that you have a node with the id "first" or an attribute like nodeColor "#2763c4"):
cy.$('[id != "first"]');
cy.$('[id = "first"]');
cy.$('[nodeColor = "#2763c4"]');
cy.$('[weight > 50]');
Additionally, you can specify the target collection within your filter query like this:
cy.$('node[id != "first"]');
Lastly, if you need complex filtering, you can use a function to apply that logic to the filter, for that just do this:
cy.$(function(element, i){
return element.isNode() && element.data('weight') > 50;
});
Sounds like you are trying to cy.filter on a cytoscape instance that no longer exists at that point. That's why it works the first time, but not the second time (after you reinstate the graph, which probably means destroy & create).
You need to make sure you point your filter handlers to the active cytoscape instance.
I have some records in a CMS that include HTML fragments with custom tags for a widget tool. The maker of the CMS has apparently updated their CMS without providing proper data conversion. Their widgets use keys for layout based on screen width such as block_lg, block_md, block_sm. The problem kicks in with the fact they used to have a block_xs and they have now shifted them all -- dropping the block_xs and instead placing a block_xl on the other end.
We don't really use these things, but their widget configurations do. What this means for us is the values for each key are identical. The problem occurs when the updated CMS code is looking for the 'block_xl' in any widget definition tags, it can't find it and errors out.
What I'm thinking then is that the new code will appear to 'ignore' the block_xs due to how it reads the tags. (and similarly, the old code will ignore block_xl) Since the values for each are identical, I need to basically read any widget definition and add a block_xl value to it matching the value of [any one of] the other width parameters.
Since the best place order-wise would be 'before' the block_lg value, it's probably easiest to do it as follows:
Replace any thing matching posix style regex matching /block_lg(="\d+,\d+")/ with: block_xl="$1" block_lg="$1"
Or whatever the equivalent of that would be.
Example of an existing CMS block with multiple widget definitions:
<div>{{widget type="CleverSoft\CleverBlock\Block\Widget"
widget_title="The Album" classes="highlight-bottom modish greenfont font52 fontlight"
enable_fullwidth="0" block_ids="127" lazyload="0"
block_lg="127,12," block_md="127,12," block_sm="127,12," block_xs="127,12,"
template="widget/block.phtml" scroll="0" background_overlay_o="0"}}</div>
<!-- Image Block -->
<div>{{widget type="CleverSoft\CleverBlock\Block\Widget"
widget_title="What’s Your Favorite Cover Style?"
classes="zoo-widget-style2 modish grey font26 fontlight"
enable_fullwidth="0" block_ids="126" lazyload="0"
block_lg="126,12," block_md="126,12," block_sm="126,12," block_xs="126,12,"
template="widget/block.phtml" scroll="0" background_overlay_o="0"}}</div>
What I would prefer to end up with from the above (adding block_xl):
<div>{{widget type="CleverSoft\CleverBlock\Block\Widget"
widget_title="The Album" classes="highlight-bottom modish greenfont font52 fontlight"
enable_fullwidth="0" block_ids="127" lazyload="0"
block_xl="127,12," block_lg="127,12," block_md="127,12," block_sm="127,12," block_xs="127,12,"
template="widget/block.phtml" scroll="0" background_overlay_o="0"}}</div>
<!-- Image Block -->
<div>{{widget type="CleverSoft\CleverBlock\Block\Widget"
widget_title="What’s Your Favorite Cover Style?"
classes="zoo-widget-style2 modish grey font26 fontlight"
enable_fullwidth="0" block_ids="126" lazyload="0"
block_xl="126,12," block_lg="126,12," block_md="126,12," block_sm="126,12," block_xs="126,12,"
template="widget/block.phtml" scroll="0" background_overlay_o="0"}}</div>
I know how to do it in php and if necessary, I will just replace it on my local DB and write an sql script to update the modified records, but the html blocks can be kind of big in some cases. It would be preferable, if it is possible, to make the substitutions right in the SQL but I'm not sure how to do it or if it's even possible to do.
And yes, there can be more than one instance of a widget in any given cms page or block. (i.e. there may be a need for more than one such substitutions with different local 'values' assigned to the block_lg)
If anyone can help me do it in SQL, it would be greatly appreciated.
for reference, the tables effected are called cms_page and cms_block, the name of the row in both cases is content
SW
I'm having trouble using the query/search capabilities of Cloud.photos, more specifically, finding a way of search for a string in title field.
After reading this: http://docs.appcelerator.com/cloud/latest/#!/guide/search_query, I was able to perform querys with $regex:
where: {title: {'$regex' : '^' + searchterm}},
But works only if the search term is in the beginning of the title, and it's case sensitive (I'm obviously a complete regex newbie).
Also tried search API with no luck. I didn't find which were the searchable fields.
So, I'm in some sort of dead end. I'm looking for a SQL like similar funcionality, where I can perform a search of a term and if matches regardless of the position and whether it is upper or lower case.
Thanks in advance.
see the release notes here for 11 Apr 2013
http://docs.appcelerator.com/cloud/latest/#!/guide/acs_releasenotes
it cannot be done any other way because the query would be too inefficient
I'm using the google custom search api and want to create a search using the siteSearch:
https://www.googleapis.com/customsearch/v1?key=k&cx=cx&q=cocos2d&siteSearch=www.cocos2d-iphone.org&siteSearchFilter=i
and it works fine (returns all the result only from the given site).
Then I want to specify TWO sites to search so I tried to change the :
siteSearch=www.cocos2d-iphone.org
to
siteSearch=www.cocos2d-iphone.org www.XXXXXXXX.org
siteSearch=www.cocos2d-iphone.org|www.XXXXXXXX.org
siteSearch=www.cocos2d-iphone.org||www.XXXXXXXX.org
but none of these works.
hope someone can help here, thanks:)
Currently I don't believe you can specify more site through the query param siteSearch.
nevertheless you can configure your Custom Search Engine here: https://www.google.com/cse/manage/all
in the "Site to search" area.
This also works for excluding, as you can read here: https://support.google.com/customsearch/bin/answer.py?hl=en&answer=2631038&topic=2601037&ctx=topic
You cannot do this with the as_sitesearch parameter as that only accepts a single value. But you can achieve what you want with the as_q parameter, setting it to some value like: "site:google.com OR site:microsoft.com" - that will work in a similar way to this search.
The as_q parameter is documented here as:
The as_q parameter provides search terms to check for in a document.
This parameter is also commonly used to allow users to specify
additional terms to search for within a set of search results.
Examples q=president&as_q=John+Adams
Use "space" as seperator
Below is sample PHP code which works for me
$url="https://www.googleapis.com/customsearch/v1?key=k&cx=cx&q=cocos2d&siteSearch=".urlencode("www.cocos2d-iphone.org www.XXXXXXXX.org")."&siteSearchFilter=i"
Thanks,
Ojal Suthar
I have a mongoid document which embeds other documents with a relation like
this:
embeds_many :blocks
Creating new blocks works fine, but I cannot manage to change the
order of existing embedded documents. For example I have three
embedded blocks and I want to move the last one to the first
position.What's the correct way to do that?
I had to deal with this with mongoid's recursively_embeds_many feature, but it's essentially the same. There's nothing wrong as far as I can tell with literally rewriting the document. Write a model method to do something like:
def reverse_blocks
reversed_blocks = blocks.to_a.reverse
blocks.clear
reversed_blocks.each do |b|
blocks.create b.attributes
end
save
end
That's not great code above, but it gives you an idea of how to do what you want to do. I'm not thrilled with having to go through that just to reorder stuff in an array, but there it is.
I think, that really correct way is make in your embedded docs field "weight" and query them with asc(:weight) or desc(:weight). You don't rely on the order of persisted non-embedded docs, so you shouldn't in embedded.
But if you urgently need to make this, your embedded docs in mongoid are just array, so you can do such way:
doc.embedded_docs = [doc.embedded_docs.last] + doc.embedded_docs[0..-2]