Challenge #259: Disenvowel

A solution to last week's challenge can be found here.

This week's challenge was submitted by @MarqueeCrew based on this post - Thanks for your submission! Be sure to check out the accompanying blog post!

In this challenge we need to create a workflow to remove vowels from text, excluding vowels that are the first letter in a word. Y is included as a vowel for this challenge...

challenge_259_start_file.yxmd

challenge_259_solution.yxmd

Parse

Join

Preparation

Intermediate

Data Preparation

Core

Transform

Accepted answers

All comments

Aaron_Harter

Arn cmpltd!

Spoiler

challenge_259_AB Data_Aaron.yxmd

Maskell_Rascal

Fun little RegEx challenge to start the week! Here is my solution.

Spoiler

REGEX_Replace([Input Data], "\B[aeiouyAEIOUY]", "")

challenge_259_Phil_Nelson.yxmd

MarqueeCrew

Here's my blog post about the original post:

https://community.alteryx.com/t5/Engine-Works/Accepted-Solution-Deep-Dive-to-RegEx-or-Not-to-RegEx/ba-p/733109

I do encourage everyone to explore that blog and the original post. Please do comment along the way.

cheers,

mark

Luke_C

Tried with the usual formula approach I would use, and also learned something new with the regex method mentioned in original thread

Spoiler

challenge_259_solved_lc.yxmd

Blake

All st on ths chllng

Spoiler

Chck ot m vd sltn blw

(Check out my video solution below)

Spoiler

challenge_259_BH.yxmd

jferrone

Spoiler

Chllng Cmpltd

This is a different approach as I haven't really used Tokenize. Results matched. Reattached solution workbook with both options.

challenge_259_jf_solution.yxmd

DeanWest

Fun one-tool challenge! :formula:

Spoiler

challenge_259_Solution_DeanWest.yxmd

patrick_digan

Spoiler

challenge_259_start_file.yxmd

RolandSchubert

Spoiler

challenge_259_solution_rsc.yxmd

HemalathaSundaram

Learnt a new syntax!

challenge_259_start_file.yxmd

BeginnerMindset

Spoiler

challenge_259_start_fileJS.yxmd

Toons

My solution :

Spoiler

challenge_259_Solution_Toons.yxmd

Kenda

Spoiler

I didn't know that \B was a thing until now! Here is my solution before looking at @MarqueeCrew 's response

challenge_259_start_file.yxmd

rcarog

Solution attached.

challenge_259_solution.yxmd

ulaimins

Tried 3 different quicker methods

Spoiler

challenge_259_start_file-ulaimins.yxmd

hellyars

@AYXAcademy

Okay...

My original question was motived by the need to simplify and shorten titles to facilitate charting etc.

Let me modify the scenario a bit to reflect real world data...

Using the example provided by @MarqueeCrew as a start point... I would not want You reduced to Y. How can us adjust for the type of scenarios? For example, what if the start point is "Universal UAS Ground Control". I would want the output to be "Unvrsl UAS Grnd Cntrl" and NOT "Unvrsl US Grnd Cntrl". How might you incorporate a list of "don't modify words" into your workflow?

MarqueeCrew

@hellyars ,

Did I mention that I am a UAS certified remote pilot?

Cheers,

mark

OllieClarke

Here's my solution:

Spoiler

challenge_259_OC_Solution.yxmd

philipmannering

My solution,

Spoiler

I was using,

(\b[aeiouy])|[aeiouy]&#8203;

and replacing with `\1`.

But better solution from @Maskell_Rascal

\B[aeiouy]

Challenge 259 - Disenvowel.yxmd

LHolmes

Ths ws an excllnt chllng!

Spoiler

challenge_259_solution_lmh.yxmd

apathetichell

I was trying to do this as a one tool REGEX and could get it to skip the first vowel on the line - but not in each word...

challenge 259 solution.yxmd

JP_SDAK

These are the only reason I look forward to Monday mornings 😊 Love the Regex Tokenizer - made this nice and easy.

Spoiler

challenge_259_start_file.yxmd

abrouwer

Solution attached

challenge_259_solution_file - AJB.yxmd

Milanz

m sltn 😉

Spoiler

challenge_259_ML.yxmd

Tony_Castillo

Spoiler

TonyA

Fun little exercise. But the best part was the name 😀.

Spoiler

Just looked at some of the other results. Never ran into \B before. A lot cleaner than [^\s].

That's why I do all of these -- even the simple ones. You can always learn something new.

challenge_259_TonyA.yxmd

MarqueeCrew

@hellyars ,

if you can trust capitalization ....

eliminate the AEIOUY and UAS stays UAS.

I'd use a find replace tool to identify text with words that I don't want to change. Theoretically if use a replacement (__001__), run the data through the tool and then another find replace to restore the word. I'd probably do this with the non-RegEx solution.

care to give that a try?

mike_w

This can be done as a one-liner! I suggest at least one more test case: "You and I" should become "Y an".

Spoiler

Substring([Input Data],0,1) + REGEX_Replace(Substring([Input Data],1), "[aeiouy]", "", 1)

challenge_259_mrm.yxmd

Qiu

Not so difficult even for me😁

Spoiler

challenge_259_start_file-Qiu.yxmd

Tyro_abc

Best Regards

Arundhuti

phottovy

Not my best solution but it gets the job done.

Spoiler

challenge_259_ph_solution.yxmd

AkimasaKajitani

My solution.

Spoiler

challenge_259_start_file_AK.yxmd

balajilolla

Solution Attached

challenge_259_start_file.yxmd

dhavaldoshi

How to do it as a Beginner?

Trying to start this approach at weekly challenges where I post solutions through the really simple route. This is my first attempt!

Spoiler

challenge_259_solution_DD_file.yxmd

Jon_Taylor

Spoiler

259 challenge.yxzp

kelly_gilbert

@hellyars, regarding your question HERE about not shortening specific words made me think of a database naming convention we had at a previous job... If each "word" in the fieldname was fewer than five letters, then you use it as-is, otherwise, you remove vowels. So, STORE_SKU would remain STORE_SKU, but STORE_ELEVATION would become STORE_ELVTN.

One way to do it would be to separate the input strings into tokens and then evaluate each token individually (i.e. if the token is longer than 5 chars, take out the vowels). This is sort of a mash-up of the regex solution and the solution in @MarqueeCrew 's video.

Using the Regex Parse tool with the Tokenize method splits the strings into tokens.

(.+?) = match any character one or more times (lazy/not greedy)
(?:\b|$) = this is a non-capturing group, meaning match it, but don't return it. Either a word boundary or the end of the string ($) indicates the end of a token.

Then, you can handle each token separately to decide whether or not to disemvowel it. Instead of using length to decide whether or not to disemvowel the token, you could also join to a text input (or file) that contains words you never want to shorten (such as the company name).

Then, you can use concatenate (with nothing in the separator) to put the strings back together again, grouping by the RecordID.

Now, the field name example I mentioned above gets tricky with regex, because we have to break the tokens at underscores. We want to split it into three tokens: STORE, _, and ELEVATION, so we can evaluate the lengths separately.

Underscores are actually included in "word" character class, so \b (word boundary) wouldn't create a break between STORE and _. The only way I know how to handle that (with regex) is to use lookarounds to identify the boundary.

Regex to tokenize with underscores: (.+?)(?:(?=\W|_)(?<!\W|_)|(?<=\W|_)(?!\W|_)|$)

(.+?) = match any character one or more times, ? means lazy/not greedy
(?:...) a non-capturing group (match it, but don't return it). A match in this group marks the end of a token.

Inside the non-capturing group:
(?=\W|_)(?<!\W|_) = the NEXT char is either a nonword or underscore, but the PREVIOUS char is not
| = OR
(?<=\W|_)(?!\W|_) = the PREVIOUS char is a nonword or underscore, but the NEXT char is not
| = OR
$ = end of string

That would correctly break STORE_ELEVATION into STORE, _, ELEVATION, so we can apply the desired logic to STORE and ELEVATION separately.

challenge_259_solution_klg_tokenized.yxmd

karlhansen

ed_hayter

Spoiler

challenge_259_flow.yxzp

Quick Links

This months top contributors

LanisC 335

hhaigh1 280

erich_shiino 280

FláviaB 262

JORGE4900 235