I ran into an odd issue when trying to modify a chef recipe. I have an attribute that contains a large hash of hashes. For each of those sub-hashes, I wanted to add a new key/value to a 'tags' hash within. In my recipe, I create a 'tags' local variable for each of those large hashes and assign the tags hash to that local variable.
I wanted to add a modification to the tags hash, but the modification had to be done at compile time since the value was dependent on a value stored in an input json. My first attempt was to do this:
tags = node['attribute']['tags']
tags['new_key'] = json_value
However, this resulted in a spec error that indicated I should use node.default, or the equivalent attribute assignment function. So I tried that:
tags = node['attribute']['tags']
node.normal['attribute']['tags']['new_key'] = json_value
While I did not have a spec error, the new key/value was not sticking.
At this point I reached my "throw stuff at a wall" phase and used the hash.merge function, which I used to think was functionally identical to hash['new_key'] for a single key/value pair addition:
tags = node['attribute']['tags']
tags.merge({ 'new_key' => 'json_value' })
This ultimately worked, but I do not understand why. What functional difference is there between the two methods that causes one to be seen as a modification of the original chef attribute, but not the other?
The issue is you can't use node['foo'] like that. That accesses the merged view of all attribute levels. If you then want to set things, it wouldn't know where to put them. So you need to lead off by tell it where to put the data:
tags = node.normal['attribute']['tags']
tags['new_key'] = json_value
Or just:
node.normal['attribute']['tags']['new_key'] = json_value
Beware of setting things at the normal level though, it is not reset at the start of each run which is probably what you want here, but it does mean that even if you remove the recipe code doing the set, the value will still be in place on any node that already ran it. If you want to actually remove things, you have to do it explicitly.
Related
I have several processes with almost same flow like "Get some parameters, extract data from database according to them and upload them to target". The parameters vary slightly across processes as well as targets but only a bit. Most of the process is the same. I would like to extract those differences to parameter-context and dynamically load them. My idea is to have parameters defined following way and then using them.
So core of question is:
How to dynamically choose which parameter group load and use?
Having several parameter contexts with same-named/different-valued parameters and dynamically switching them would be probably the best, but it is not possible as far as I know.
Also duplicating flows is out-of-the-table. Any error correction would be spread out over several places and maintenance would be a nightmare.
Moreover, I know I can do it like "In GenetrateFlowFile for process A set value1=#{A_value1} and in GenetrateFlowFile for process B set value1=#{B_value1}. But this is tedious, error-prone and scales kinda bad. Not speaking of situation when I can have dozens of parameters and several processes. Also it is a kind of hardcoding, not configuring...
I was hoping for something like defining group=A and then using it like value1=#{ ${ group:append('_value1') } } but this does not work - it is evaluated as parameter literally named ${ group:append('_value1') }.
TL;DR: Use evaluateELString().
The actual solution is to set in GenetrateFlowFile processor group=A and in next UpdateAttribute processor set the following:
value1=${ group:prepend('hash{ '):append('_value1 }'):replace('hash', '#'):evaluateELString() }
The magic being done here is "Take value of group slap around it #{ and _value1 } to make it valid NiFi Expression Language statement and then evaluate it." (Notice - the word hash and function replace is there since I didn´t manage to escape the # char right before {.)
If you would like to have your value1 at the beginning of the statement then you can use following code. The result is same, it is easier to use (often-changed value value1 is at the beginning of the statement) and is less readable "what is really going on?"-wise.
value1=${ literal('value1'):prepend('_'):prepend(${ group }):prepend('hash{ '):append(' }'):replace('hash', '#'):evaluateELString() }
When I run my code one error occurs with my code setindex! not defined for WeakRefStrings.StringArray{String,1}
CSV File here.
using CSV
EVDdata =CSV.read(raw"wikipediaEVDdatesconverted.csv")
EVDdata[end-9:end,:]
And the Error Code is here
rows, cols = size(EVDdata)
for j =1:cols
for i = 1:rows
if !isdigit(string(EVDdata[i, j])[1])
EVDdata[i,j] = 0
end
end
end
I am working with Julia 1.4.1 on Jupter Notebook
setindex!(collection, item, inds...) is the function that colection[inds...] = item gets lowered to. The error comes from the fact that CSV.read makes an immutable collection.
As Oscar says in his answer, setindex! tries to mutate its arguments, i.e. change the contents of your column in place. When you do CSV.read(), by default immutable columns of type CSV.Column are returned. This is done for performance reason, as it means columns don't have to be copied after parsing.
To get around this, you can do two things:
Pass the keyword argument CSV.read(raw"wikipediaEVDdatesconverted.csv", copycols = true) - this will copy the columns and therefore make them mutable; or
Achieve the same by using DataFrame((raw"wikipediaEVDdatesconverted.csv"))
The second way is the preferred way as CSV.read will be deprecated in the CSV.jl package.
You can see that it's current implementation is basically doing the same thing I listed in (2) above in the source here. Removing this method will allow CSV.jl not to depend on DataFrames.jl anymore.
It could also be done this way
col1dt = Vector{Dates.DateTime}(undef, length(col1))
for v = 1:length(col1)
col1dt[v] = Dates.DateTime(col1[v], "d-u-y")
end
I'm writing a ruby webdriver test that needs to store a dynamic value such as an order ID to use later on in the text. I think I need to extract the value from the string and then store it as a variable to call for future use.
The string looks like this and I just need to extract/store the numeric value.
<span class="receiptNum hidden-xs">Receipt #: 12303430</span>
Any tips or examples on how to extract that value and create a variable for future use would be great!
To extract the text (only numbers) out of the this element, try using the following code:
#numbers = #driver.find_element(:css=>'receiptNum').text.scan(\d+)
Currently I am saving this number in an instance variable which can be used again in the same test as it will flow around with the test object till the test finishes.
Other option include saving it in a temp txt file and reading from it when required.
Note: Fetching data now and using it later is not a good practice, try not to use this very frequently.
Hope it helps!!!
I used to have a series of independent arrays (e.g. name(), id(), description() ). I used to be able to check whether a value existed in a specific array by doing name.include?("Mark")
Now that I moved to a MUCH MORE elegant way to manage different these independent arrays (here for background: How do I convert an Array with a JSON string into a JSON object (ruby)) I am trying to figure out how I do the same.
In short I put all the independent arrays in a single structure so that I can reference the content as object().name, object().id, object().description.
However I am missing now how I can check whether the object array has a value "Mark" in its name structure.
I have tried object.name.include?("Mark") but it doesn't quite like it.
I have also tried to use has_value?but that doesn't seem to be working either (likely because it used to be an hash before I imported it into the structure but right now is no longer a hash - see here: How do I convert an Array with a JSON string into a JSON object (ruby))
Thoughts? How can I check whether object.name contains a certain string?
Thanks.
If you want to find all customers called Mark you can write the following:
customers_named_mark = array_of_customers.select{|c| c.name == 'Mark' }
This will return a potentially empty array.
If you want to find the first customer named Mark, write
customer_named_mark = array_of_customers.detect{|c| c.name == 'Mark' }
This will return the first matching item or nil.
I'm using Jammit to package assets up for a Rails application and I have a few asset files that I'd like to be included in each of a few groups. For example, I'd like Sammy and its plugins to be in both my mobile and screen JS packages.
I've tried this:
sammy: &SAMMY
- public/javascripts/vendor/sammy.js
- public/javascripts/vendor/sammy*.js
mobile:
<<: *SAMMY
- public/javascripts/something_else.js
and this:
mobile:
- *SAMMY
but both put the Sammy JS files in a nested Array, which Jammit can't understand. Is there a syntax for including the elements of an Array directly in another Array?
NB: I realize that in this case there are only two elements in the SAMMY Array, so it wouldn't be too bad to give each an alias and reference both in each package. That's fine for this case, but quickly gets unmaintainable when there are five or ten elements that have a specific load order.
Closest solution I know of is this one:
sammy:
- &SAMMY1
public/javascripts/vendor/sammy.js
- &SAMMY2
public/javascripts/vendor/sammy*.js
mobile:
- *SAMMY1
- *SAMMY2
- public/javascripts/something_else.js
Alternatively, as already suggested, flatten the nested lists in a code snippet.
Note: according to yaml-online-parser, your first suggestion is not a valid use of << (used to merge keys from two dictionaries. The anchor then has to point to another dictionary I believe.
If you want mobile to be equal to sammy, you can just do:
mobile: *SAMMY
However if you want mobile to contain other elements in addition to those in sammy, there's no way to do that in YAML to the best of my knowledge.
Your example is valid YAML (a convenient place to check is YPaste), but it's not defined what the merge does. Per the spec, a merge key can have a value:
A mapping, in which case it's merged into the parent mapping.
A sequence of mappings, in which case each is merged, one-by-one, into the parent mapping.
There's no way of merging sequences on YAML level.
You can, however, do this in code. Using the YAML from your second idea:
mobile:
- *SAMMY
you'll get nested sequences - so flatten them! Assuming you have a mapping of such nested sequences:
data = YAML::load(File.open('test.yaml'))
data.each_pair { |key, value| value.flatten! }
(Of course, if you have a more complicated YAML file, and you don't want every sequence flattened (or they're not all sequences), you'll have to do some filtering.)
This solution is for Symfony/PHP only (considerations for other languages, see below)
Note about array keys from the PHP array manual page:
Strings containing valid decimal ints, unless the number is preceded by a + sign, will be cast to the int type. E.g. the key "8" will actually be stored under 8. [...]
This means that if you actually index your anchor array with integer keys, you can simply add new keys by continuing the initial list. So your solution would look like this:
sammy: &SAMMY
1: public/javascripts/vendor/sammy.js
2: public/javascripts/vendor/sammy*.js
mobile:
<<: *SAMMY
3: public/javascripts/something_else.js
You can even overwrite keys and still add new ones:
laptop:
<<: *SAMMY
1: public/javascripts/sammy_laptop.js
3: public/javascripts/something_else.js
In both cases the end result is a perfectly valid indexed array, just like before.
Other programming languages
Depending on your YAML implementation and how you iterate over your array, this could conceivably also be used in other programming languages. Though with a caveat.
For instance, in JS you can access numerical string keys by their integer value as well:
const sammy = {"1": "public/javascripts/vendor/sammy.js"}
sammy["1"]; // "public/javascripts/vendor/sammy.js"
sammy[1]; // "public/javascripts/vendor/sammy.js"
But you'd need to keep in mind, that your initial array is now an object, and that you would need to iterate over it accordingly, e.g.:
Object.keys(sammy).forEach(key => console.log(sammy[key]))
As it has been suggested, when you need to flatten a list, at least in ruby, it is trivial to add a "!flatten" type specifier to mobile and implement a class that extends Array, adds the yaml_tag and flattens the coder seq on init_with.