Fix fread clipboard handling on Windows (fixes #1292) by AmanKashyap0807 · Pull Request #7640 · Rdatatable/data.table

AmanKashyap0807 · 2026-02-09T22:05:54Z

Fix fread clipboard handling on Windows using readClipboard(), add test (#2366), and update NEWS.

Changed files :

R/fread.R : Added logic to detect clipboard inputs on Windows, reading it by using readClipboard(), handle errors and process clipboard content as a temporary file for parsing.
inst/tests/tests.Rraw : Added test 2366 for clipboard functionality.
Updated NEWS.md with a brief entry under BUG FIXES.

To keep the scope of issue #1292, I only solved it for Windows. I'd be happy to submit another PR for macOS and Linux later.

Reproduction:

# Write tab delimited data to clipboard
writeLines("a\tb\n1\t4\n2\t5\n3\t6\n", "clipboard")

# read.delim works
read.delim("clipboard-128")
# [1] read.delim.clipboard.128.
# <0 rows> (or 0-length row.names)
# Warning message:
# In read.table(file = file, header = header, sep = sep, quote = quote,  :
#   incomplete final line found by readTableHeader on 'clipboard'

# fread fails
library(data.table)
fread("clipboard-128")
# Error in fread("clipboard-128") : 
#   File 'clipboard-128' does not exist or is non-readable. getwd()=='D:/OpenSource/data.table'

fread() does not recognize "clipboard" / "clipboard-128" as clipboard input on Windows. Due to the long period since the issue was reported, there might have been some changes in fread() over time, which is why the error message differs now, but the core problem is the same, clipboard input is not detected.

Environment:

OS: Windows
R: 4.5.2

Please let me know if any changes are needed.

codecov · 2026-02-09T22:13:19Z

Codecov Report

❌ Patch coverage is 3.84615% with 25 lines in your changes missing coverage. Please review.
✅ Project coverage is 98.88%. Comparing base (67e7840) to head (b95ca4c).

Files with missing lines	Patch %	Lines
R/fread.R	3.84%	25 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #7640      +/-   ##
==========================================
- Coverage   99.02%   98.88%   -0.15%     
==========================================
  Files          87       87              
  Lines       16897    16923      +26     
==========================================
+ Hits        16733    16734       +1     
- Misses        164      189      +25

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

AmanKashyap0807 · 2026-02-09T23:18:44Z

Hi @MichaelChirico,

I’ve synced the branch with master.
The atime check is still hitting an 'invalid reference' error, which I suspect might be a CI issue with external forks. Also, the Codecov failure is likely due to my changes being Windows-specific while the CI runs on Linux.

Let me know if there’s anything else I should do on my end to help resolve these!

joshhwuu · 2026-02-12T15:28:28Z

R/fread.R

+        })
+      } else {
+        # Note: macOS (pbpaste) and Linux (xclip/xsel) support discussed in #1292
+        stopf("Clipboard reading is supported on Windows only.")


i wonder if we must stopf here

I added the explicit error to avoid this potentially confusing behavior on non-Windows systems.
I tested this by commenting out else-stopf() block on Ubuntu 20.04 (WSL). In that case, fread("clipboard") falls through and returns:

> temp <- fread("clipboard") > print(temp) Empty data.table (0 rows and 1 cols): clipboard

That said, if preserving the current fallback behavior is preferred for compatibility, I’m happy to remove the stopf() and let the normal control flow continue.

this probably isn't an issue but I'd prefer a warning at most, just so we're not diverging from current behaviour

Thanks for the clarification. I’ll drop the stopf(), and as I am adding macOS/Linux clipboard support per @ben-schwen suggestion, I’ll switch to a warning when no clipboard utility is found.

ben-schwen · 2026-02-16T08:41:34Z

R/fread.R

      # input is data itself containing at least one \n or \r
    } else if (startsWith(input, " ")) {
      stopf("input= contains no \\n or \\r, but starts with a space. Please remove the leading space, or use text=, file= or cmd=")
+    } else if (grepl("^clipboard(-[0-9]+)?$", tolower(input))) {


Why do we use grepl here and not simply only read from clipboard when the input is clipboard?

I used regex to consider both 'clipboard' and 'clipboard-128' because original issue #1292 shows the user trying fread('clipboard-128').
However your point is valid, since readClipboard() takes no argument and only read clipboard regardless of any suffix so we can use simple string matching.
but the user who are used to base R's read.delim('clipboard-128') syntax would have to use 'clipboard'. let me know if u want the change I will do it.

On Windows, R's own clipboard connections can be opened with the description clipboard-<buffer size> in order to set the size limit for writing. It's not needed to read the clipboard. On Unix, X11_primary, X11_secondary, X11_clipboard are also supported. We don't have to support anything except clipboard.

Yes as mentioned by Ivan, using a buffer size after clipboard- is a winwodws idiom, which we do not have to support.

I would go with identical(input, "clipboard") or input == "clipboard"

ben-schwen · 2026-02-16T08:43:00Z

R/fread.R

+    } else if (grepl("^clipboard(-[0-9]+)?$", tolower(input))) {
+      is_windows = identical(.Platform$OS.type, "windows")
+      if (is_windows) {
+        clip = tryCatch(utils::readClipboard(), error = identity)


we don't normally use namespacing for functions from utils, see e.g. head/tail

we don't have blanket utils import so it'll need to be included:

data.table/NAMESPACE

Line 148 in 1bd88cb

importFrom(utils, capture.output, contrib.url, download.file, flush.console, getS3method, head, packageVersion, tail, untar, unzip)

readClipboard() is documented as Windows‑only in the utils::clipboard help:
https://search.r-project.org/R/refmans/utils/help/clipboard.html
There are also reports on non‑Windows where readClipboard() simply isn’t available even with utils loaded,
(https://stackoverflow.com/questions/24365249/readclipboard-removed-from-utils-package)
so importing it unconditionally in NAMESPACE risks errors when the package is loaded on those platforms. That’s why I kept utils::readClipboard() inside the Windows guard.

Regarding head/tail, data.table::last() has a small fallback that explicitly calls utils::head and utils::tail instead of using them unqualified everywhere, and that pattern was in my mind when I treated utils::readClipboard() as a similar Windows‑specific special case. I’m happy to change this to plain readClipboard() with a NAMESPACE import if you prefer and are confident it behaves well on non‑Windows platforms.

R/fread.R

ben-schwen · 2026-02-16T11:06:54Z

I would also prefer to add support for all major systems once and not part after part.

The commands should be the following:

OS	Command
Windows	`readClipboard()`
Mac	`system("pbpaste", intern = TRUE)`
Linux (xsel)	`system("xsel --clipboard --output", intern = TRUE)`
Linux (xclip)	`system("xclip -o -selection clipboard", intern = TRUE)`

AmanKashyap0807 · 2026-02-17T03:35:22Z

@ben-schwen thanks for outlining the commands. I’ll work on adding support for macOS and Linux using these and push an update soon.

AmanKashyap0807 · 2026-02-18T19:15:36Z

Hi @ben-schwen, while looking into the Linux clipboard side I noticed that newer distros (Ubuntu 22.04+, Fedora) default to Wayland, so I’m planning to add wl-paste alongside xclip/xsel to keep clipboard reading working smoothly there.

For testing:

Linux: I'll test xclip/xsel and wl-paste on a WSL
macOS: I'll test pbpaste on a friend's Mac.

Let me know if you'd like me to do it differently or if you'd like me to add more commands.

…pported device

AmanKashyap0807 · 2026-02-19T11:08:10Z

I've pushed an update adding clipboard support for macOS and Linux as discussed.

macOS: Uses pbpaste.
Linux: Prioritizes wl-paste (for Wayland), then checks for xclip, and finally xsel (for X11)

Let me know if there any adjustment are required.

ben-schwen · 2026-02-19T11:49:09Z

NEWS.md


 5. Non-equi joins combining an equality condition with two inequality conditions on the same column (e.g., `on = .(id == id, val >= lo, val <= hi)`) no longer error, [#7641](https://github.com/Rdatatable/data.table/issues/7641). The internal `chmatchdup` remapping of duplicate `rightcols` was overwriting the original column indices, causing downstream code to reference non-existent columns. Thanks @tarun-t for the report and fix, and @aitap for the diagnosis.

+6. `fread("clipboard")` now works on Linux (via `xclip`, `xsel`, or `wl-paste`), macOS (via `pbpaste`), and Windows, [#1292](https://github.com/Rdatatable/data.table/issues/1292). Thanks @mbacou for the report, @ben-schwen for the suggestion, and @AmanKashyap0807 for the fix.


pls elaborate more what this means, e.g. that fread supports now reading from the clipboard via fread("clipboard")

ben-schwen · 2026-02-19T11:53:14Z

R/fread.R

+        )
+      } else if (identical(os_type, "unix")) {
+        sysname = Sys.info()[["sysname"]]
+        clip_cmd = if (identical(sysname, "Darwin")) {


wont this be uninitialized on Solaris and friends?

aitap · 2026-02-19T11:57:03Z

Let's not limit X11/Wayland clipboard support to just Linux. Yes, macOS is special, but anything else where .Platform$OS.type == 'unix' (FreeBSD, OpenBSD, NetBSD, Cygwin, OpenIndiana...) should work as well.

ben-schwen · 2026-02-19T11:59:32Z

R/fread.R

+      if (!length(clip) || !any(nzchar(trimws(clip)))) {
+        stopf("Clipboard is empty.")
+      }
+      input = paste(clip, collapse = "\n")


Suggested change

input = paste(clip, collapse = "\n")

writeLines(clip, tmpFile <- tempfile(tmpdir=tmpdir))

file = tmpFile

on.exit(unlink(tmpFile), add=TRUE)

Using paste might allocate a large string before dumping it to disk

ben-schwen · 2026-02-19T12:00:05Z

inst/tests/tests.Rraw

+    }, add=TRUE)
+    test(2366.2, fread("clipboard"), data.table(a=1L, b=2L))
+  }
+})


missing newline at end of test file

ben-schwen · 2026-02-19T12:07:09Z

R/fread.R

+      } else {
+        warning("Clipboard reading is not supported on this platform.", call. = FALSE)
+      }
+      if (!length(clip) || !any(nzchar(trimws(clip)))) {


on non windows and non-unix this will error because of trimws(clip) but clip was never assigned

Fix fread clipboard handling on Windows (fixes Rdatatable#1292)

546de99

AmanKashyap0807 requested a review from MichaelChirico as a code owner February 9, 2026 22:05

addressing linting

c68136d

joshhwuu reviewed Feb 12, 2026

View reviewed changes

ben-schwen reviewed Feb 16, 2026

View reviewed changes

R/fread.R Outdated Show resolved Hide resolved

ben-schwen reviewed Feb 16, 2026

View reviewed changes

R/fread.R Outdated Show resolved Hide resolved

ben-schwen reviewed Feb 16, 2026

View reviewed changes

R/fread.R Outdated Show resolved Hide resolved

Tweak fread clipboard checks and error handling on Windows

43b8a22

AmanKashyap0807 added 4 commits February 19, 2026 15:29

Add Linux (X11/Wayland) and macOS clipboard support to fread

b5a2a7c

Merge branch 'master' into fix-fread-clipboard-issue1292

a8abf6c

correct wl-paste redundancy and replace stopf with warning for non su…

6c34cba

…pported device

fix linter warning of CI workflow

b95ca4c

ben-schwen reviewed Feb 19, 2026

View reviewed changes


		5. Non-equi joins combining an equality condition with two inequality conditions on the same column (e.g., `on = .(id == id, val >= lo, val <= hi)`) no longer error, [#7641](https://github.com/Rdatatable/data.table/issues/7641). The internal `chmatchdup` remapping of duplicate `rightcols` was overwriting the original column indices, causing downstream code to reference non-existent columns. Thanks @tarun-t for the report and fix, and @aitap for the diagnosis.

		6. `fread("clipboard")` now works on Linux (via `xclip`, `xsel`, or `wl-paste`), macOS (via `pbpaste`), and Windows, [#1292](https://github.com/Rdatatable/data.table/issues/1292). Thanks @mbacou for the report, @ben-schwen for the suggestion, and @AmanKashyap0807 for the fix.

-      input = paste(clip, collapse = "\n")
+      writeLines(clip, tmpFile <- tempfile(tmpdir=tmpdir))
+      file = tmpFile
+      on.exit(unlink(tmpFile), add=TRUE)

Comments

Conversation

AmanKashyap0807 commented Feb 9, 2026

Uh oh!

codecov bot commented Feb 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

AmanKashyap0807 commented Feb 9, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ben-schwen commented Feb 16, 2026

Uh oh!

AmanKashyap0807 commented Feb 17, 2026

Uh oh!

AmanKashyap0807 commented Feb 18, 2026

Uh oh!

AmanKashyap0807 commented Feb 19, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

aitap commented Feb 19, 2026 via email

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

codecov bot commented Feb 9, 2026 •

edited

Loading