# Problem with \`replace\` function making too many replacements

**URL:** <https://discourse.julialang.org/t/problem-with-replace-function-making-too-many-replacements/117022>\
**Category:** General Usage\
**Tags:** question\
**Created:** [July 13, 2024, 11:52pm UTC](https://discourse.julialang.org/t/problem-with-replace-function-making-too-many-replacements/117022 "2024-07-13T23:52:55Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![00krishna](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/00krishna/32/8843_2.png) [@00krishna](https://discourse.julialang.org/u/00krishna)\
**Post date:** [July 13, 2024, 11:52pm UTC](https://discourse.julialang.org/t/problem-with-replace-function-making-too-many-replacements/117022/1 "2024-07-13T23:52:55Z")

</div>

I have a dataset where one of the columns provides the names of university departments, such as “economics department” and “statistics department”, etc. However there are cases where some rows list the name as “statistics department” while other row list the name as “statistics dept”, etc. Hence I am trying to clean up and standardize the names a bit so that “statistics department” and “statistics dept” are not treated as 2 separate departments.

I am running into some trouble with my replacements where it seems like multiple replacements get applied to a single input value. This is leading to some weird outputs. I have an MWE below. I did use the `count=1` argument in the function, but even then I am hitting this issue. I suppose there is some circularity in the replacements as I try to hit all the variations I can think of. Perhaps someone can think of a better way.

```julia
using DataFrames

d = [ "atm, oceanic & space sci.",
 "atm, oceanic and space sciences",
 "biologic and materials science",
 "biologic and materials sciences"]

rpl = [ " sci " => " sciences ", 
        " sci" => " sciences",
        " sci." => " sciences", 
        " science " => " sciences ",
        " science" => " sciences",
        " sciencess" => " sciences"]

map(x -> replace(x, rpl ..., count=1), d)

```

The output that I get looks like:

```julia
4-element Vector{String}:
 "atm, oceanic & space sciences."
 "atm, oceanic and space sciencesences"
 "biologic and materials sciencesence"
 "biologic and materials sciencesences"

```

Some replacements work correctly, while others end up with these strange overlapping replacements. Can anyone indicate where I am going wrong and how to fix it?

Thanks for the assistance.

---

<div class="post-metadata">

**Author:** ![savq](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/savq/32/22063_2.png) [@savq](https://discourse.julialang.org/u/savq)\
**Post date:** [July 14, 2024, 3:32am UTC](https://discourse.julialang.org/t/problem-with-replace-function-making-too-many-replacements/117022/2 "2024-07-14T03:32:24Z")

</div>

I’m not entirely sure what the problem is, but you can use a word boundary assertion `\b` to avoid dealing with spaces, which fixes the problem.

```julia-repl
julia> depts = [
           "atm, oceanic & space sci.",
           "atm, oceanic and space sciences",
           "biologic and materials science",
           "biologic and materials sciences"
       ];

julia> replacements = [
           r"\bsci\b\.?" => "sciences",
           r"\bsciences*\b" => "sciences",
       ];

julia> map(str -> replace(str, replacements...), depts)
4-element Vector{String}:
 "atm, oceanic & space sciences"
 "atm, oceanic and space sciences"
 "biologic and materials sciences"
 "biologic and materials sciences"

```

---

<div class="post-metadata">

**Author:** ![00krishna](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/00krishna/32/8843_2.png) [@00krishna](https://discourse.julialang.org/u/00krishna)\
**Post date:** [July 14, 2024, 4:18am UTC](https://discourse.julialang.org/t/problem-with-replace-function-making-too-many-replacements/117022/3 "2024-07-14T04:18:23Z")

</div>

@savq that is very interesting, I had not seen that before. I don’t work with regular expressions that often, so I don’t know all the tricks. But I will certainly try this.

---

<div class="post-metadata">

**Author:** ![jules](https://avatars.discourse-cdn.com/v4/letter/j/41988e/32.png) [@jules](https://discourse.julialang.org/u/jules)\
**Post date:** [July 15, 2024, 7:00pm UTC](https://discourse.julialang.org/t/problem-with-replace-function-making-too-many-replacements/117022/4 "2024-07-15T19:00:46Z")

</div>

The problem is that `" sci"` matches all four entries, the first is only correct “by chance” because even ignoring the period, the output is right when the word is cut off anyway.

You have to order your replacement pairs such that the long ones come first, that way no match is the beginning of a later one.

But the word boundary is a better solution anyway.

---

<div class="post-metadata">

**Author:** ![DNF](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/dnf/32/10191_2.png) [@DNF](https://discourse.julialang.org/u/DNF)\
**Post date:** [July 15, 2024, 7:06pm UTC](https://discourse.julialang.org/t/problem-with-replace-function-making-too-many-replacements/117022/5 "2024-07-15T19:06:51Z")

</div>

> [@00krishna](#):
>
> ```julia
> rpl = [ " sci " => " sciences ", 
> " sci" => " sciences",
> " sci." => " sciences", 
> " science " => " sciences ",
> " science" => " sciences",
> " sciencess" => " sciences"]
> 
> ```

Instead of trying to find all versions of all strings combined with different leading and trailing spaces and punctuation, &, etc (leading to a combinatorial explosion), isn’t it better to divide the problem into several steps?

First remove leading, trailing, and multiple spaces. Remove unneeded punctuation, replace “&” with “and”, etc.

Only then turn “sci” into “sciences”, etc.