Fixing the namespace issue

[sadavarterohit]

No description provided.

[jtr13]

Can you explain why you took out removing the line breaks? (It may not be necessary to remove them but I haven't fully tested.)

[sadavarterohit]

I apolgize, I responded to another question instead of this about why I did this.

While I was testing on this example out of the readme, I accidentally selected #Before as well, and that put the mtcars variable in the comment, the line breaking did not persist some of the line breaks that already exist.
Since we're unlisting, breaking and putting the strings back together based on the operators in between, this didn't change anything. (atleast in the cases I tested.)

Removing the breaklines solved the issue like a charm, since the line break simply never got removed from the end of the comment.

# BEFORE
mtcars |> group_by(cyl) |> summarize(mean_mpg = mean(mpg)) |> ggplot(aes(factor(cyl),
mean_mpg)) + theme_bw(14) + geom_col() + labs(title = "mtcars", x = "number of cylinders", y = "average miles per gallon")

[jtr13]

That sounds right... I will do a little more testing as soon as I get a chance.

[jtr13]

Adding the regex here partially works -- the problem is that the splitwords are also used for the token column later on for the purpose of sorting. We'll need to think about this. A minor issue: \\s shouldn't be there because it's not necessary for there to be a space before a geom.

[sadavarterohit]

That's fair, I did make that assumption in most of the code I tested.
Let me try to come up with a different regex to solve this.

[jtr13]

I think it will work to keep "geom_" as a split word and add "\w*::geom_" as an additional one. Then in the section where df$token is created, remove \w*:: from the namespaced tokens so that all the tokens will be recognized as factor levels and the sort can be performed correctly.

[jtr13]

Thanks for tackling this issue... it's a good start! Please see the comments above. Here's an example that I tested it on:

mtcars |> group_by(cyl) |> summarize(mean_mpg = mean(mpg)) |> ggplot(aes(factor(cyl),
mean_mpg)) + theme_bw(14) + geom_col() + labs(title = "mtcars",
                                              
  x = "number of cylinders", y = "average miles per gallon")
 + ggalluvial::geom_alluvium()

For the reasons I detailed above the sort order is not correct:

mtcars |>
  group_by(cyl) |>
  summarize(mean_mpg = mean(mpg)) |>
  ggplot(aes(factor(cyl), mean_mpg)) +
  labs(title = "mtcars", x = "number of cylinders", y = "average miles per gallon") +
  theme_bw(14) +
  geom_col() +
  ggalluvial::geom_alluvium()

More specifically, the df$token column after sorting should be:

      token
1 firstline
2        |>
3        |>
4   ggplot(
6     geom_
8     geom_
7     labs(
5    theme_

but with the new code it's:

     token
1 firstline
2        |>
3        |>
4   ggplot(
7     labs(
5    theme_
6      <NA>
8      <NA>

since it doesn't recognize either of the geoms as tokens.

I'd be glad to meet so we can brainstorm how to address this if you're interested. (To be clear, you don't need to complete this for the community contribution!)

[jtr13]

Once you have the link, please add it here and then I'll merge it (whether or not the namespace issue is completed.)