Merge branch 'master' into issue6556

Author: MichaelChirico

.ci/atime/tests.R

@@ -231,7 +231,20 @@ test.list <- atime::atime_test_list(
     },
     expr = data.table:::melt(DT, measure.vars = measure.vars),
     Slow = "fd24a3105953f7785ea7414678ed8e04524e6955", # Parent of the merge commit (https://github.com/Rdatatable/data.table/commit/ed72e398df76a0fcfd134a4ad92356690e4210ea) of the PR (https://github.com/Rdatatable/data.table/pull/5054) that fixes the issue
-    Fast = "ed72e398df76a0fcfd134a4ad92356690e4210ea"), # Merge commit of the PR (https://github.com/Rdatatable/data.table/pull/5054) that fixes the issue
+    Fast = "ed72e398df76a0fcfd134a4ad92356690e4210ea"), # Merge commit of the PR (https://github.com/Rdatatable/data.table/pull/5054) that fixes the issue  # Test case created directly using the atime code below (not adapted from any other benchmark), based on the issue/fix PR https://github.com/Rdatatable/data.table/pull/5054#issue-930603663 "melt should be more efficient when there are missing input columns."
+
+  # Test case created from @tdhock's comment https://github.com/Rdatatable/data.table/pull/6393#issuecomment-2327396833, in turn adapted from @philippechataignon's comment https://github.com/Rdatatable/data.table/pull/6393#issuecomment-2326714012
+  "fwrite refactored in #6393" = atime::atime_test(
+    setup = {
+      set.seed(1)
+      NC = 10L
+      L <- data.table(i=1:N)
+      L[, paste0("V", 1:NC) := replicate(NC, rnorm(N), simplify=FALSE)]
+      out.csv <- tempfile()
+    },
+    expr = data.table::fwrite(L, out.csv, compress="gzip"),
+    Before = "f339aa64c426a9cd7cf2fcb13d91fc4ed353cd31", # Parent of the first commit https://github.com/Rdatatable/data.table/commit/fcc10d73a20837d0f1ad3278ee9168473afa5ff1 in the PR https://github.com/Rdatatable/data.table/pull/6393/commits with major change to fwrite with gzip.
+    PR = "3630413ae493a5a61b06c50e80d166924d2ef89a"), # Close-to-last merge commit in the PR.
 
   tests=extra.test.list)
 # nolint end: undesirable_operator_linter.

.github/workflows/code-quality.yaml

@@ -70,6 +70,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
+      - uses: r-lib/actions/setup-r@v2
       - name: Lint
         run: for (f in list.files('.ci/linters/md', full.names=TRUE)) source(f)
         shell: Rscript {0}

NEWS.md

@@ -69,6 +69,10 @@ rowwiseDT(
 
 6. `fread()` gains `logicalYN` argument to read columns consisting only of strings `Y`, `N` as `logical` (as opposed to character), [#4563](https://github.com/Rdatatable/data.table/issues/4563). The default is controlled by option `datatable.logicalYN`, itself defaulting to `FALSE`, for back-compatibility -- some smaller tables (especially sharded tables) might inadvertently read a "true" string column as `logical` and cause bugs. This is particularly important for tables with a column named `y` or `n` -- automatic header detection under `logicalYN=TRUE` will see these values in the first row as being "data" as opposed to column names. A parallel option was not included for `fwrite()` at this time -- users looking for a compact representation of logical columns can still use `fwrite(logical01=TRUE)`. We also opted for now to check only `Y`, `N` and not `Yes`/`No`/`YES`/`NO`.
 
+7. `fwrite()` with `compress="gzip"` produces compatible gz files when composed of multiple independent chunks owing to parallelization, [#6356](https://github.com/Rdatatable/data.table/issues/6356).  Earlier `fwrite()` versions could have issues with HTTP upload using `Content-Encoding: gzip` and `Transfer-Encoding: chunked`. Thanks to @oliverfoster for report and @philippechataignon for the fix.
+
+8. `fwrite()` gains a new parameter `compressLevel` to control compression level for gzip, [#5506](https://github.com/Rdatatable/data.table/issues/5506). This parameter balances compression speed and total compression, and corresponds directly to the analogous command-line parameter, e.g. `compressLevel=4` corresponds to passing `-4`; the default, `6`, matches the command-line default, i.e. equivalent to passing `-6`. Thanks @mgarbuzov for the request and @philippechataignon for implementing.
+
 ## BUG FIXES
 
 1. `fwrite()` respects `dec=','` for timestamp columns (`POSIXct` or `nanotime`) with sub-second accuracy, [#6446](https://github.com/Rdatatable/data.table/issues/6446). Thanks @kav2k for pointing out the inconsistency and @MichaelChirico for the PR.
@@ -304,7 +308,7 @@ rowwiseDT(
 
 5. Input files are now kept open during `mmap()` when running under Emscripten, [emscripten-core/emscripten#20459](https://github.com/emscripten-core/emscripten/issues/20459). This avoids an error in `fread()` when running in WebAssembly, [#5969](https://github.com/Rdatatable/data.table/issues/5969). Thanks to @maek-ies for the report and @georgestagg for the PR.
 
-6. `dcast()` improves behavior for the situation that the `fun.aggregate` value of `length()` is used but not provided by the user. 
+6. `dcast()` improves behavior for the situation that the `fun.aggregate` value of `length()` is used but not provided by the user.
 
     a. This now triggers a warning, not a message, since relying on this default often signals unexpected duplicates in the data, [#5386](https://github.com/Rdatatable/data.table/issues/5386). The warning is classed as `dt_missing_fun_aggregate_warning`, allowing for more targeted handling in user code. Thanks @MichaelChirico for the suggestion and @Nj221102 for the fix.
 
@@ -1019,7 +1023,7 @@ rowwiseDT(
 
 14. The options `datatable.print.class` and `datatable.print.keys` are now `TRUE` by default. They have been available since v1.9.8 (Nov 2016) and v1.11.0 (May 2018) respectively.
 
-15. Thanks to @ssh352, Václav Tlapák, Cole Miller, András Svraka and Toby Dylan Hocking for reporting and bisecting a significant performance regression in dev. This was fixed before release thanks to a PR by Jan Gorecki, [#5463](https://github.com/Rdatatable/data.table/pull/5463). 
+15. Thanks to @ssh352, Václav Tlapák, Cole Miller, András Svraka and Toby Dylan Hocking for reporting and bisecting a significant performance regression in dev. This was fixed before release thanks to a PR by Jan Gorecki, [#5463](https://github.com/Rdatatable/data.table/pull/5463).
 
 16. `key(x) <- value` is now fully deprecated (from warning to error). Use `setkey()` to set a table's key. We started warning not to use this approach in 2012, with a stronger warning starting in 2019 (1.12.2). This function will be removed in the next release.
 

R/fwrite.R

@@ -10,6 +10,7 @@ fwrite = function(x, file="", append=FALSE, quote="auto",
            buffMB=8, nThread=getDTthreads(verbose),
            showProgress=getOption("datatable.showProgress", interactive()),
            compress = c("auto", "none", "gzip"),
+           compressLevel = 6L,
            yaml = FALSE,
            bom = FALSE,
            verbose=getOption("datatable.verbose", FALSE),
@@ -18,18 +19,17 @@ fwrite = function(x, file="", append=FALSE, quote="auto",
   if (length(encoding) != 1L || !encoding %chin% c("", "UTF-8", "native")) {
     stopf("Argument 'encoding' must be '', 'UTF-8' or 'native'.")
   }
-  if (missing(qmethod)) qmethod = qmethod[1L]
-  if (missing(compress)) compress = compress[1L]
-  if (missing(dateTimeAs)) { dateTimeAs = dateTimeAs[1L] }
-  else if (length(dateTimeAs)>1L) stopf("dateTimeAs must be a single string")
-  dateTimeAs = chmatch(dateTimeAs, c("ISO","squash","epoch","write.csv"))-1L
-  if (is.na(dateTimeAs)) stopf("dateTimeAs must be 'ISO','squash','epoch' or 'write.csv'")
+  qmethod = match.arg(qmethod)
+  compress = match.arg(compress)
+  dateTimeAs = match.arg(dateTimeAs)
+  dateTimeAs = chmatch(dateTimeAs, c("ISO", "squash", "epoch", "write.csv")) - 1L
   if (!is.null(logicalAsInt)) {
     stopf("logicalAsInt has been renamed logical01 for consistency with fread.")
   }
   scipen = if (is.numeric(scipen)) as.integer(scipen) else 0L
   buffMB = as.integer(buffMB)
   nThread = as.integer(nThread)
+  compressLevel = as.integer(compressLevel)
   # write.csv default is 'double' so fwrite follows suit. write.table's default is 'escape'
   # validate arguments
   if (is.matrix(x)) { # coerce to data.table if input object is matrix
@@ -42,7 +42,8 @@ fwrite = function(x, file="", append=FALSE, quote="auto",
       x = as.data.table(x)
     }
   }
-  stopifnot(is.list(x),
+  stopifnot(
+    is.list(x),
     identical(quote,"auto") || isTRUEorFALSE(quote),
     is.character(sep) && length(sep)==1L && (nchar(sep) == 1L || identical(sep, "")),
     is.character(sep2) && length(sep2)==3L && nchar(sep2[2L])==1L,
@@ -51,14 +52,15 @@ fwrite = function(x, file="", append=FALSE, quote="auto",
     is.character(eol) && length(eol)==1L,
     length(qmethod) == 1L && qmethod %chin% c("double", "escape"),
     length(compress) == 1L && compress %chin% c("auto", "none", "gzip"),
+    length(compressLevel) == 1L && 0 <= compressLevel && compressLevel <= 9,
     isTRUEorFALSE(col.names), isTRUEorFALSE(append), isTRUEorFALSE(row.names),
     isTRUEorFALSE(verbose), isTRUEorFALSE(showProgress), isTRUEorFALSE(logical01),
     isTRUEorFALSE(bom),
     length(na) == 1L, #1725, handles NULL or character(0) input
     is.character(file) && length(file)==1L && !is.na(file),
     length(buffMB)==1L && !is.na(buffMB) && 1L<=buffMB && buffMB<=1024L,
     length(nThread)==1L && !is.na(nThread) && nThread>=1L
-    )
+  )
 
   is_gzip = compress == "gzip" || (compress == "auto" && endsWithAny(file, ".gz"))
 
@@ -115,7 +117,7 @@ fwrite = function(x, file="", append=FALSE, quote="auto",
   file = enc2native(file) # CfwriteR cannot handle UTF-8 if that is not the native encoding, see #3078.
   .Call(CfwriteR, x, file, sep, sep2, eol, na, dec, quote, qmethod=="escape", append,
         row.names, col.names, logical01, scipen, dateTimeAs, buffMB, nThread,
-        showProgress, is_gzip, bom, yaml, verbose, encoding)
+        showProgress, is_gzip, compressLevel, bom, yaml, verbose, encoding)
   invisible()
 }
 

inst/tests/tests.Rraw

@@ -159,6 +159,9 @@ TZnotUTC = !identical(tt,"") && !is_utc(tt)
 #   (3) function factory for matching messages exactly by substituting anything between delimiters [delim, fmt=TRUE]
 #   (4) function factory for matching messages exactly by substituting a generic string [fmt=string]
 get_msg = function(e, delim, fmt=FALSE) {
+  ufq = options(useFancyQuotes = FALSE) # otherwise we get angled quotes, hard to match robustly
+  on.exit(options(ufq))
+
   condition = tryCatch({e; NULL}, error=identity, warning=identity)
   if (is.null(condition)) return(condition)
   msg = condition$message
@@ -170,17 +173,13 @@ get_msg = function(e, delim, fmt=FALSE) {
     sprintf("%s%s%s", delim[1L], if (fmt) "%s" else ".+", delim[2L]),
     msg
   )
-  if (fmt) return(function(x) sprintf(msg, x))
+  if (fmt) return(function(...) sprintf(msg, ...))
   return(msg)
 }
 base_messages = list(
   missing_object = get_msg(`__dt_test_missing_` + 1, "'", fmt=TRUE),
   missing_function = get_msg(`__dt_test_missing_`(), '"', fmt=TRUE),
-  missing_coerce_method = get_msg(delim = '"', {
-    old = options(useFancyQuotes = FALSE) # otherwise we get angled quotes, hard to match robustly
-    on.exit(options(old))
-    methods::as(TRUE, 'foo')
-  }),
+  missing_coerce_method = get_msg(methods::as(TRUE, 'foo'), delim = '"'),
   missing_dispatch_method = get_msg(conditionMessage(structure(1, class="foo")), '[\'"]'),
   invalid_arg_unary_operator = get_msg(-'a'),
   invalid_arg_binary_operator = get_msg(1 + 'a'),
@@ -199,6 +198,8 @@ base_messages = list(
   stopifnot = get_msg(stopifnot(FALSE), fmt="FALSE"),
   not_yet_used = get_msg(.NotYetUsed("abc"), "'", fmt=TRUE), # NB: need fmt= because the English message has '(yet)' --> parens in regex
   ambiguous_date_fmt = get_msg(as.Date('xxx')),
+  match_arg_length = get_msg(match.arg(c('a', 'b'), letters)),
+  match_arg_4_choices = get_msg(match.arg('e', letters[1:4]), delim='"', fmt=TRUE),
   NULL
 )
 
@@ -10014,7 +10015,7 @@ test(1658.27, fwrite(DT, na="NA", verbose=TRUE), output='Writing bom .false., ya
 test(1658.28, fwrite(ok_dt, 1), error=base_messages$stopifnot("is.character(file) && length(file) == 1L && !is.na(file)"))
 test(1658.29, fwrite(ok_dt, quote=123), error="identical\\(quote.*auto.*FALSE.*TRUE")
 test(1658.30, fwrite(ok_dt, sep="..."), error="nchar(sep)")
-test(1658.31, fwrite(ok_dt, qmethod=c("double", "double")), error="length(qmethod)")
+test(1658.31, fwrite(ok_dt, qmethod=c("double", "double")), error=base_messages$match_arg_length)
 test(1658.32, fwrite(ok_dt, col.names="foobar"), error="isTRUEorFALSE(col.names)")
 
 # null data table (no columns)
@@ -10056,8 +10057,12 @@ if (!haszlib()) {
   test(1658.423, file.info(f1)$size < file.info(f2)$size)  # 74 < 804  (file.size() isn't available in R 3.1.0)
   if (test_R.utils) test(1658.43, fread(f1), DT)  # use fread to decompress gz (works cross-platform)
   fwrite(DT, file=f3<-tempfile(), compress="gzip")   # compress to filename not ending .gz
+  fwrite(DT, file=f4<-tempfile(), compress="gzip", compressLevel=1)   # test compressLevel
+  fwrite(DT, file=f5<-tempfile(), compress="gzip", compressLevel=9)
   test(1658.441, file.info(f3)$size, file.info(f1)$size)
-  unlink(c(f1,f2,f3))
+  test(1658.442, file.info(f4)$size >= file.info(f1)$size)
+  test(1658.443, file.info(f1)$size >= file.info(f5)$size)
+  unlink(c(f1,f2,f3,f4,f5))
 }
 DT = data.table(a=1:3, b=list(1:4, c(3.14, 100e10), c("foo", "bar", "baz")))
 test(1658.45, fwrite(DT), output=c("a,b","1,1|2|3|4","2,3.14|1e+12","3,foo|bar|baz"))
@@ -10106,6 +10111,23 @@ test(1658.58, fwrite(DT), output='a,b\n1,0\\+1i\n2,-1-1i\n3,$')
 test(1658.59, fwrite(data.table(a=list('a')), verbose=TRUE),
      output='fields will be quoted if the field contains either sep.*sep2.*list column')
 test(1658.60, fwrite(data.table(r=as.raw(0))), error = "'raw' - not yet implemented")
+## naLen is bigger than col width
+test(1658.61, fwrite(data.table(a="a"), na="VERY LONG MISSING VALUE STRING !", quote=FALSE, verbose=TRUE),
+     output="maxLineLen=66")
+## eol="" error
+test(1658.62, fwrite(data.table(a=1), tempfile(), eol=''), error='eol must be 1 or more bytes')
+
+## buffMB < single line width and < header width
+f = tempfile()
+test(1658.63, fwrite(data.table(a=strrep('x', 2**21)), f, buffMB=1.0), NULL)
+test(1658.64, file.size(f) > 2**20) # almost exactly 2**21, but exact number will vary by platform. we just care we didn't truncate at 1MiB.
+DT=data.table(1L)
+setnames(DT, strrep('y', 2**21))
+test(1658.65, fwrite(DT, f, buffMB=1.0, nThread=1L), NULL)
+test(1658.66, file.size(f) > 2**20)
+unlink(f)
+
+test(1658.67, fwrite(data.table(a=numeric()), verbose=TRUE), output='No data rows present')
 
 options(oldverbose)
 ## End fwrite tests
@@ -10977,9 +10999,9 @@ DT = data.table(
   D = as.POSIXct(dt<-paste(d,t), tz="UTC"),
   E = as.POSIXct(paste0(dt,c(".999",".0",".5",".111112",".123456",".023",".0",".999999",".99",".0009")), tz="UTC"))
 
-test(1740.0, fwrite(DT,dateTimeAs="iso"), error="dateTimeAs must be 'ISO','squash','epoch' or 'write.csv'")
-test(1740.1, fwrite(DT,dateTimeAs=c("ISO","squash")), error="dateTimeAs must be a single string")
-test(1740.2, capture.output(fwrite(DT,dateTimeAs="ISO")), c(
+test(1740.1, fwrite(DT,dateTimeAs="iso"), error=base_messages$match_arg_4_choices("ISO", "squash", "epoch", "write.csv"))
+test(1740.2, fwrite(DT,dateTimeAs=c("ISO","squash")), error=base_messages$match_arg_length)
+test(1740.3, capture.output(fwrite(DT,dateTimeAs="ISO")), c(
 "A,B,C,D,E",
 "1907-10-21,1907-10-21,23:59:59,1907-10-21T23:59:59Z,1907-10-21T23:59:59.999Z",
 "1907-10-22,1907-10-22,00:00:00,1907-10-22T00:00:00Z,1907-10-22T00:00:00Z",
@@ -10991,7 +11013,7 @@ test(1740.2, capture.output(fwrite(DT,dateTimeAs="ISO")), c(
 "1999-12-31,1999-12-31,01:23:45,1999-12-31T01:23:45Z,1999-12-31T01:23:45.999999Z",
 "2000-02-29,2000-02-29,23:59:59,2000-02-29T23:59:59Z,2000-02-29T23:59:59.990Z",
 "2016-09-12,2016-09-12,01:30:30,2016-09-12T01:30:30Z,2016-09-12T01:30:30.000900Z"))
-test(1740.3, capture.output(fwrite(DT,dateTimeAs="squash")), c(
+test(1740.4, capture.output(fwrite(DT,dateTimeAs="squash")), c(
 "A,B,C,D,E",
 "19071021,19071021,235959,19071021235959000,19071021235959999",
 "19071022,19071022,000000,19071022000000000,19071022000000000",
@@ -11003,7 +11025,7 @@ test(1740.3, capture.output(fwrite(DT,dateTimeAs="squash")), c(
 "19991231,19991231,012345,19991231012345000,19991231012345999",
 "20000229,20000229,235959,20000229235959000,20000229235959990",
 "20160912,20160912,013030,20160912013030000,20160912013030000"))
-test(1740.4, capture.output(fwrite(DT,dateTimeAs="epoch")), c(
+test(1740.5, capture.output(fwrite(DT,dateTimeAs="epoch")), c(
 "A,B,C,D,E",
 "-22718,-22718,86399,-1962748801,-1962748800.001",
 "-22717,-22717,0,-1962748800,-1962748800",

man/fwrite.Rd

@@ -18,6 +18,7 @@ fwrite(x, file = "", append = FALSE, quote = "auto",
   buffMB = 8L, nThread = getDTthreads(verbose),
   showProgress = getOption("datatable.showProgress", interactive()),
   compress = c("auto", "none", "gzip"),
+  compressLevel = 6L,
   yaml = FALSE,
   bom = FALSE,
   verbose = getOption("datatable.verbose", FALSE),
@@ -58,6 +59,7 @@ fwrite(x, file = "", append = FALSE, quote = "auto",
   \item{nThread}{The number of threads to use. Experiment to see what works best for your data on your hardware.}
   \item{showProgress}{ Display a progress meter on the console? Ignored when \code{file==""}. }
   \item{compress}{If \code{compress = "auto"} and if \code{file} ends in \code{.gz} then output format is gzipped csv else csv. If \code{compress = "none"}, output format is always csv. If \code{compress = "gzip"} then format is gzipped csv. Output to the console is never gzipped even if \code{compress = "gzip"}. By default, \code{compress = "auto"}.}
+  \item{compressLevel}{Level of compression between 1 and 9, 6 by default. See \url{https://linux.die.net/man/1/gzip} for details.}
   \item{yaml}{If \code{TRUE}, \code{fwrite} will output a CSVY file, that is, a CSV file with metadata stored as a YAML header, using \code{\link[yaml]{as.yaml}}. See \code{Details}. }
   \item{bom}{If \code{TRUE} a BOM (Byte Order Mark) sequence (EF BB BF) is added at the beginning of the file; format 'UTF-8 with BOM'.}
   \item{verbose}{Be chatty and report timings?}

src/data.table.h

@@ -296,7 +296,7 @@ SEXP chmatch_R(SEXP, SEXP, SEXP);
 SEXP chmatchdup_R(SEXP, SEXP, SEXP);
 SEXP chin_R(SEXP, SEXP);
 SEXP freadR(SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP);
-SEXP fwriteR(SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP);
+SEXP fwriteR(SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP, SEXP);
 SEXP rbindlist(SEXP, SEXP, SEXP, SEXP, SEXP);
 SEXP setlistelt(SEXP, SEXP, SEXP);
 SEXP setS4elt(SEXP, SEXP, SEXP);