#4, #5: fix tests

NEWS

@@ -1,8 +1,11 @@
 # What Is New in *stringx*
 
+
 ## 0.1.2.9xxx (under development)
 
-* ...
+* [BUGFIX] #4: Fixed failing check with ICU 55.
+
+* [BUGFIX] #5: Fixed failing check under POSIX/C locale.
 
 
 ## 0.1.2 (2021-07-27)

R/grepl.R

@@ -128,15 +128,15 @@
 #'
 #' @examples
 #' x <- c("abc", "1237", "\U0001f602", "\U0001f603", "stringx\U0001f970", NA)
-#' grepl2(x, "\\p{EMOJI_PRESENTATION}")
-#' which(grepl2(x, "\\p{EMOJI_PRESENTATION}"))  # like grep
+#' grepl2(x, "\\p{L}")
+#' which(grepl2(x, "\\p{L}"))  # like grep
 #'
-#' # at least 1 letter, digit, or emoji:
-#' p <- c("\\p{L}", "\\p{N}", "\\p{EMOJI_PRESENTATION}")
+#' # at least 1 letter or digit:
+#' p <- c("\\p{L}", "\\p{N}")
 #' `dimnames<-`(outer(x, p, grepl2), list(x, p))
 #'
-#' x |> grepv2("\\p{EMOJI_PRESENTATION}")
-#' grepv2(x, "\\p{EMOJI_PRESENTATION}", invert=TRUE) <- "\U0001F496"
+#' x |> grepv2("\\p{L}")
+#' grepv2(x, "\\p{L}", invert=TRUE) <- "\U0001F496"
 #' print(x)
 #'
 #' @seealso

R/gsub.R

@@ -110,7 +110,7 @@
 #'
 #'
 #' @examples
-#' "change \U0001f602 me \U0001f603" |> gsub2("\\p{EMOJI_PRESENTATION}", "O_O")
+#' "change \U0001f602 me \U0001f603" |> gsub2("\\p{L}+", "O_O")
 #'
 #' x <- c("mario", "Mario", "M\u00E1rio", "M\u00C1RIO", "Mar\u00EDa", "Rosario", NA)
 #' sub2(x, "mario", "M\u00E1rio", fixed=NA, strength=1L)

inst/realtest/gregexpr-stringx.R

@@ -17,12 +17,14 @@ E(
 
 E(
     regexpr2(x, "mario", fixed=NA, ignore_case=TRUE)!=-1L,
-    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA)
+    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
 )
 
 E(
     regexpr2(x, "mario", fixed=NA, strength=1L)!=-1L,
-    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA)
+    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
 )
 
 
@@ -43,30 +45,36 @@ E(
 
 E(
     unlist(gregexpr2(x, "mario", fixed=NA, ignore_case=TRUE))!=-1L,
-    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA)
+    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
 )
 
 E(
     unlist(gregexpr2(x, "mario", fixed=NA, strength=1L))!=-1L,
-    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA)
+    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
 )
 
 E(
     unlist(regexec2(x, "mario", fixed=NA, ignore_case=TRUE))!=-1L,
-    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA)
+    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
 )
 
 E(
     unlist(regexec2(x, "mario", fixed=NA, strength=1L))!=-1L,
-    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA)
+    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
 )
 
 E(
     unlist(gregexec2(x, "mario", fixed=NA, ignore_case=TRUE))!=-1L,
-    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA)
+    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
 )
 
 E(
     unlist(gregexec2(x, "mario", fixed=NA, strength=1L))!=-1L,
-    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA)
+    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
 )

inst/realtest/gregextr-stringx.R

@@ -18,8 +18,16 @@ x <- c("mario", "Mario M\u00E1rio M\u00C1RIO Mar\u00EDa Marios", "Rosario", NA)
 E(regextr2(x, "mario", ignore_case=TRUE, fixed=TRUE), c("mario", "Mario", NA, NA))
 E(gregextr2(x, "mario", ignore_case=TRUE, fixed=TRUE), list("mario", c("Mario", "Mario"), character(0), NA_character_))
 
-E(regextr2(x, "mario", fixed=NA, strength=1L), c("mario", "Mario", NA, NA))
-E(gregextr2(x, "mario", fixed=NA, strength=1L), list("mario", c("Mario", "M\u00E1rio", "M\u00C1RIO", "Mario"), character(0), NA_character_))
+E(
+    regextr2(x, "mario", fixed=NA, strength=1L),
+    c("mario", "Mario", NA, NA),
+    bad=c("mario", NA, NA, NA)  # C locale
+)
+E(
+    gregextr2(x, "mario", fixed=NA, strength=1L),
+    list("mario", c("Mario", "M\u00E1rio", "M\u00C1RIO", "Mario"), character(0), NA_character_),
+    bad=list("mario", character(0), character(0), NA_character_)  # C locale
+)
 
 E(regextr2(x, "m(?<a>[a\u00E1])rio(?<plural>s)?", ignore_case=TRUE), c("mario", "Mario", NA, NA))
 E(gregextr2(x, "m(?<a>[a\u00E1])rio(?<plural>s)?", ignore_case=TRUE), list("mario", c("Mario", "M\u00E1rio", "M\u00C1RIO", "Marios"), character(0), NA_character_))
@@ -44,7 +52,11 @@ E(
     )
 )
 
-E(regextr2(x, "mario", fixed=NA, strength=1L, capture_groups=TRUE), list("mario", "Mario", NA_character_, NA_character_))
+E(
+    regextr2(x, "mario", fixed=NA, strength=1L, capture_groups=TRUE),
+    list("mario", "Mario", NA_character_, NA_character_),
+    bad=list("mario", NA_character_, NA_character_, NA_character_)  # C locale
+)
 
 E(
     gregextr2(x, "mario", fixed=NA, strength=1L, capture_groups=TRUE),
@@ -53,7 +65,13 @@ E(
         cbind("Mario", "M\u00E1rio", "M\u00C1RIO", "Mario"),
         cbind(NA_character_)[, -1, drop=FALSE],
         cbind(NA_character_)
-    )
+    ),
+    bad=list(
+        cbind("mario"),
+        cbind(NA_character_)[, -1, drop=FALSE],
+        cbind(NA_character_)[, -1, drop=FALSE],
+        cbind(NA_character_)
+    )  # C locale
 )
 
 E(`regextr2<-`(x, "[mM]\\w+", value="x"), c("x", "x M\u00E1rio M\u00C1RIO Mar\u00EDa Marios", "Rosario", NA))

inst/realtest/grepl-stringx.R

@@ -5,8 +5,16 @@ x <- c("mario", "Mario", "M\u00E1rio", "M\u00C1RIO", "Mar\u00EDa", "Rosario", NA
 E(grepl2(x, "mario"), c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA))
 E(grepl2(x, "mario", fixed=NA), c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA))
 E(grepl2(x, "mario", ignore_case=TRUE), c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA))
-E(grepl2(x, "mario", fixed=NA, ignore_case=TRUE), c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA))
-E(grepl2(x, "mario", fixed=NA, strength=1L), c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA))
+E(
+    grepl2(x, "mario", fixed=NA, ignore_case=TRUE),
+    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
+)
+E(
+    grepl2(x, "mario", fixed=NA, strength=1L),
+    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
+)
 
 E(grepl2("abc", ".{3,3}"), TRUE)  # default fixed=FALSE
 

inst/realtest/nchar-all.R

@@ -39,13 +39,15 @@ E(
 E(
     nchar(x, "width", keepNA=FALSE),
     c(1L, 0L, 2L, 1L, 8L, 6L, 2L, 2L),
-    bad=c(1L, 0L, 2L, 1L, 8L, 6L, 2L, 4L)
+    bad=c(1L, 0L, 2L, 1L, 8L, 6L, 2L, 4L),
+    bad=c(1L, 0L, 2L, 1L, 8L, 6L, 2L, 3L)
 )
 
 E(
     nchar(x, "width", keepNA=TRUE),
     c(1L, 0L, NA, 1L, 8L, 6L, 2L, 2L),
-    bad=c(1L, 0L, NA, 1L, 8L, 6L, 2L, 4L)
+    bad=c(1L, 0L, NA, 1L, 8L, 6L, 2L, 4L),
+    bad=c(1L, 0L, NA, 1L, 8L, 6L, 2L, 3L)
 )
 
 E(

inst/realtest/startswith-stringx.R

@@ -8,8 +8,26 @@ E(endsWith(x, "mario", fixed=NA), c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)
 E(startsWith(x, "mario", ignore_case=TRUE), c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA))
 E(endsWith(x, "mario", ignore_case=TRUE), c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA))
 
-E(startsWith(x, "mario", fixed=NA, ignore_case=TRUE), c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA))
-E(endsWith(x, "mario", fixed=NA, ignore_case=TRUE), c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA))
+E(
+    startsWith(x, "mario", fixed=NA, ignore_case=TRUE),
+    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
+)
 
-E(startsWith(x, "mario", fixed=NA, strength=1L), c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA))
-E(endsWith(x, "mario", fixed=NA, strength=1L), c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA))
+E(
+    endsWith(x, "mario", fixed=NA, ignore_case=TRUE),
+    c(TRUE, TRUE, FALSE, FALSE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
+)
+
+E(
+    startsWith(x, "mario", fixed=NA, strength=1L),
+    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
+)
+
+E(
+    endsWith(x, "mario", fixed=NA, strength=1L),
+    c(TRUE, TRUE, TRUE, TRUE, FALSE, FALSE, NA),
+    bad=c(TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, NA)  # C locale
+)

inst/realtest/strcoll-stringx.R

@@ -19,4 +19,3 @@ E(x %x==% y, x == y)
 E(x %x!=% y, x != y)
 E(x %x>%  y, x >  y)
 E(x %x<=% y, x <= y)
-

inst/realtest/strptime-all.R

@@ -83,7 +83,7 @@ E(
 )
 
 
-t <- ISOdate(2021, 05, 27)
+t <- ISOdatetime(2021, 05, 27, 12, 0, 0)  # default time zone
 
 E(
     strftime(t),

inst/realtest/strptime-stringx.R

@@ -1,4 +1,4 @@
-t <- structure(ISOdate(2021, 05, 27), names="t")
+t <- structure(ISOdatetime(2021, 05, 27, 12, 0, 0), names="t")  # default time zone
 
 # different calendar/locale - easy with stringx:
 
@@ -17,6 +17,9 @@ E(
     c(t="Donnerstag")
 )
 
+
+t <- structure(ISOdatetime(2021, 05, 27, 12, 0, 0, tz="GMT"), names="t")
+
 E(
     strftime(t, "%H:%M:%S", tz="UTC", usetz=TRUE),
     P(c(t="12:00:00"), warning=TRUE)

inst/realtest/strtrim-all.R

@@ -61,6 +61,13 @@ E(
     x,
     bad=c("\U0001F4A9", "\U0001F64D", "\U0001F64D", "\u26F9\U0001F3FF\u200D",
         "\U0001F3F4\U000E0067\U000E0062\U000E0073\U000E0063\U000E0074\U000E007F"),
+    bad=c(
+        "\U0001F4A9",
+        "\U0001F64D\U0001F3FC\U0000200D",
+        "\U0001F64D\U0001F3FB\U0000200D",
+        "\U000026F9\U0001F3FF\U0000200D",
+        "\U0001F3F4\U000E0067\U000E0062\U000E0073\U000E0063\U000E0074\U000E007F"
+    ),
     worst=c("<U+", "<U+", "<U+", "<U+", "<U+")
 )
 E(
@@ -69,6 +76,13 @@ E(
     bad=c("\U0001F4A9", "\U0001F64D\U0001F3FC\u200D",
         "\U0001F64D\U0001F3FB\u200D", "\u26F9\U0001F3FF\u200D\u2640\uFE0F",
         "\U0001F3F4\U000E0067\U000E0062\U000E0073\U000E0063\U000E0074\U000E007F"),
+    bad=c(
+        "\U0001F4A9",
+        "\U0001F64D\U0001F3FC\U0000200D",
+        "\U0001F64D\U0001F3FB\U0000200D",
+        "\U000026F9\U0001F3FF\U0000200D",
+        "\U0001F3F4\U000E0067\U000E0062\U000E0073\U000E0063\U000E0074\U000E007F"
+    ),
     worst=c("<U+0", "<U+0", "<U+0", "<U+2", "<U+0")
 )
 E(

inst/realtest/strwrap-all.R

@@ -30,8 +30,8 @@ E(strwrap(character(0), simplify=FALSE), list())
 
 E(stringi::stri_width(strwrap(strrep("az ", 20), 60)), 59L)
 E(stringi::stri_width(strwrap(strrep("\u0105\u20AC ", 20), 60)), 59L)
-E(stringi::stri_width(strwrap(strrep("\U0001F643 ", 20), 60)), 59L)
-E(stringi::stri_width(strwrap(strrep("\U0001F3F3\U0000FE0F\U0000200D\U0001F308 ", 20), 60)), 59L, bad=c(44L, 14L))
+E(stringi::stri_width(strwrap(strrep("\U0001F643 ", 20), 60)), 59L, bad=39L)
+E(stringi::stri_width(strwrap(strrep("\U0001F3F3\U0000FE0F\U0000200D\U0001F308 ", 20), 60)), 59L, bad=c(44L, 14L), bad=c(74L, 24L), bad=c(59L, 39L))
 
 # attribute preservation (lack thereof)
 E(attributes(strwrap(structure(c(a="aaa", attrib1="value1")), simplify=TRUE)), NULL)

tests/realtest.R

@@ -10,7 +10,23 @@ library(this_package, character.only=TRUE)
 if (require("realtest", quietly=TRUE)) {
     f <- file.path(path.package(this_package), "realtest")
     r <- test_dir(f, ".*\\.R$")
-    s <- summary(r)
-    print(s)
-    stopifnot(all(s[["match"]] != "fail"))
+    x <- summary(r)
+
+    names(r) <- row.names(x)
+    fails <- x[x[["match"]] == "fail", , drop=FALSE]
+    if (nrow(fails) > 0) {
+        fails2 <- as.data.frame(fails[,
+            !(names(fails) %in% ".expr") & !sapply(fails, function(x) all(is.na(x))),
+            drop=FALSE])
+        for (i in row.names(fails2)) {
+            cat(sprintf("%s:%d:\n", fails2[i, ".file"], fails2[i, ".line"]))
+            cat("    value: "); str(r[[i]][["object"]][["value"]], indent.str="    ")
+            cat("    sides: "); str(r[[i]][["object"]][["sides"]], indent.str="    ")
+        }
+
+        if (require("stringi", quietly=TRUE))
+            cat(sprintf("%s; %s\n", stri_info(TRUE), Sys.getlocale()))
+
+        stop("some tests failed")
+    }
 }