Add max string length option

[jankatins]

If set to a number, strings longer than this number will be shortend with ....

E.g. `options(tibble.print_string_max = 10) will only print up to 10 characters
of each string.

[codecov-io]

Current coverage is 99.83%

Merging #104 into master will increase coverage by <.01%

@@             master       #104   diff @@
==========================================
  Files            14         14          
  Lines           590        595     +5   
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
+ Hits            589        594     +5   
  Misses            1          1          
  Partials          0          0          

Powered by Codecov. Last updated by 7c00abf...bf131a6

[jankatins]

Unfortunately, <data.frame [4 x 2]> is already 20 chars long...So the string formatting should probably done before the obj_sum function is applied...?

[krlmlr]

I'm not generally opposed to limiting string length, but I think #89 was about adaptive shortening that is applied as needed if the output is too wide.

[jankatins]

@krlmlr so the algo would be if printed_width > available_width: set string columns to max(nchar(names),min_string_length) and apply the "shortener" (=don't redistribute the leftover space), but this could potentially lead to a bit of leftover space. But that's already happening now...

I will do the changes, if that's what you want :-)

[krlmlr]

It would help to see an example, see the expect_output() calls in the tests.

[jankatins]

This is an example with the current implementation and options(tibble.print_string_max = 20)...

Source: local data frame [216 x 23]
Groups: tags [7]

# A tibble: 216 x 23
   longitude.x latitude.x                    date  company eyeColor   age                  guid.x index product
         <chr>      <chr>                   <chr>    <chr>    <chr> <int>                   <chr> <int>   <int>
1    27.339158 -85.314797 2000-03-17T23:45:01....  FLYBOYZ    brown    20 d3982b08-2a5e-4abc-a...     0       2
2    27.339158 -85.314797 2000-03-17T23:45:01....  FLYBOYZ    brown    20 d3982b08-2a5e-4abc-a...     0       2
3    27.339158 -85.314797 2000-03-17T23:45:01....  FLYBOYZ    brown    20 d3982b08-2a5e-4abc-a...     0       2
4   -72.514214  46.225443 2002-03-11T04:09:28....  BIOSPAN    brown    38 9afd44ba-5b0c-4302-8...     1       2
5   -72.514214  46.225443 2002-03-11T04:09:28....  BIOSPAN    brown    38 9afd44ba-5b0c-4302-8...     1       2
6   -72.514214  46.225443 2002-03-11T04:09:28....  BIOSPAN    brown    38 9afd44ba-5b0c-4302-8...     1       2
7   -31.964069  38.553859 1980-02-19T05:23:31....     ZOID    green    23 f1f4aa36-15e8-4208-a...     2       0
8   -31.964069  38.553859 1980-02-19T05:23:31....     ZOID    green    23 f1f4aa36-15e8-4208-a...     2       0
9   -31.964069  38.553859 1980-02-19T05:23:31....     ZOID    green    23 f1f4aa36-15e8-4208-a...     2       0
10  150.363342 -85.232234 1996-05-18T20:27:20.... PHOTOBIN     blue    37 b23bbf56-f6ae-4bb7-b...     3       1

[krlmlr]

Nice. If you could enhance the output tests, so that the "known output" becomes part of this PR? Just add a test to test-trunc-mat.r, the test file will be generated if it's missing.

See #100 for problems with "wide" characters in certain scripts. For instance, I'm getting:

> sprintf("%.*s...", 4, "合同录入日期")
[1] "合\xe5..."

We need a better method to shorten a string to a given "visible" width.

@hadley: Should string columns be limited to width 20 by default?

[hadley]

Shortening by default feels a bit too aggressive to me - I'd prefer to do only if there's not enough space. Ideally a one column df with long string would be truncated to screen width.

[krlmlr]

@JanSchulz: Would you like to contribute more to this PR? See #104 (comment).

[jankatins]

@krlmlr I should have time to work on this from wednesday next week... Sorry if that's too late :-/

Regarding the unicode stuff: is nchar(x, type = "width") the solutions? #100 mentions that this fails on windows?

So, to recap the requirements:

• don't shorten if everything fits the space
• one column should be shortend to exactly the screen width
• min width of a column is max(nchar(names),min_string_length)
• min_string_length is a setting

Algo:

• if columns do not fit, set str columns to min width
• calculate space and the maximum number of "fit in" (=visible) columns
• calculate "leftover" space and redistribute to the string visible columns

[krlmlr]

Thanks. No hurry.

nchar(type = "width") works in RStudio and RGui, at least for the examples shown in #100. It works the same in the R terminal on Windows, but the output is printed unhelpfully as "<U+....>", which renders useless the width calculation. At some point we may need to invent our own nchar() that takes this into account, so it's probably a good idea to encapsulate the width calculation.

The "shortening to screen width" part looks interesting to me in the context of #100. If everything else fails, you could split the strings to code points and calculate the width for each; I'd generally exclude code points with zero width if the next codepoint doesn't fit:

> c("成交日期", "合同录入日期") %>% strsplit("") %>% lapply(nchar, type = "width")
[[1]]
[1] 2 2 2 2

[[2]]
[1] 2 2 2 2 2 2

Otherwise, the requirements look good to me.

[krlmlr]

Two more points:

• You could use a Unicode ellipsis "\u2026", it renders as a dot in R terminal on Windows and as a single-char wide ellipsis elsewhere.
• Are you going to look into shortening of column names, too?

[jankatins]

regarding windows <U+....> printing: r-lib/evaluate#66 This problem seems to happen quite deep in R. We had quite a lot of fun in IRkernel (or better in repr) with that because we use sink (or better evaluate does) to get the output of a computation: https://github.com/IRkernel/repr/blob/master/R/repr_matrix_df.r#L16-L27

[krlmlr]

Shelving this for now, I think strings with limited width will be easier with #144.

[randomgambit]

hello everyone, thanks for your great work!

Just wondering if there are any plans about implementing that? I think its pretty useful. For instance, in Pandas one could simply do:

In [43]: df = pd.DataFrame(np.array([['foo', 'bar', 'bim', 'uncomfortably long string'],
   ....:                             ['horse', 'cow', 'banana', 'apple']]))
   ....: 

In [44]: pd.set_option('max_colwidth',40)

In [45]: df
Out[45]: 
       0    1       2                          3
0    foo  bar     bim  uncomfortably long string
1  horse  cow  banana                      apple

In [46]: pd.set_option('max_colwidth', 6)

In [47]: df
Out[47]: 
       0    1      2      3
0    foo  bar    bim  un...
1  horse  cow  ba...  apple

In [48]: pd.reset_option('max_colwidth')

which is really helpful when one prints a tibble that contains both text and numeric values.

[randomgambit]

hello there! any updates on this? I tried options(tibble.print_string_max = 20) but does not seem to work. Thanks!

[randomgambit]

hello there! any updates on this? I tried options(tibble.print_string_max = 20) but does not seem to work. Thanks!

[krlmlr]

Development of better column formatting has moved to https://github.com/hadley/colformat. Would you like to contribute there?

Comments to a closed issue are not very effective, because it is easy to ignore them.

[randomgambit]

thanks @krlmrl ! i ll have a look at it