close #210

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: stringi
 Version: 1.0-2
-Date: 2016-01-29
+Date: 2016-01-30
 Title: Character String Processing Facilities
 Description: Allows for fast, correct, consistent, portable,
     as well as convenient character string/text processing in every locale

NEWS

@@ -13,7 +13,9 @@ http://bugs.icu-project.org/trac/ticket/11554]
 
 * [BUGFIX] #214: allow a regex pattern like `.*`  to match an empty string.
 
-* ... #210: stri_replace_all_fixed(c("1", "NULL"), "NULL", NA) should result in c("1", NA)
+* [BUGFIX] #210: `stri_replace_all_fixed(c("1", "NULL"), "NULL", NA)`
+now results in `c("1", NA)`
+
 * ... #207
 * ... #199
 * ... #175

R/search_replace_4.R

@@ -103,6 +103,8 @@
 #' stri_replace_all_fixed(s, " ", "#")
 #' stri_replace_all_fixed(s, "o", "0")
 #'
+#' stri_replace_all_fixed(c("1", "NULL", "3"), "NULL", NA)
+#'
 #' stri_replace_all_regex(s, " .*? ", "#")
 #' stri_replace_all_regex(s, "(el|s)it", "1234")
 #' stri_replace_all_regex('abaca', 'a', c('!', '*'))

devel/testthat/test-replace-charclass.R

@@ -6,7 +6,7 @@ test_that("stri_replace_all_charclass-vectorize_all=FALSE", {
    expect_error(stri_replace_all_charclass("b",character(0),c("a", "b"), vectorize_all=FALSE))
    expect_error(stri_replace_all_charclass("b",c("", ""),"a", vectorize_all=FALSE))
    expect_identical(stri_replace_all_charclass(NA,c("\\p{WHITE_SPACE}", "[a]"),"?", vectorize_all=FALSE),NA_character_)
-   expect_identical(stri_replace_all_charclass("X",c("\\p{WHITE_SPACE}", "[a]"),NA, vectorize_all=FALSE),NA_character_)
+   expect_identical(stri_replace_all_charclass(c("X", " ", "a"),c("\\p{WHITE_SPACE}", "[a]"),NA, vectorize_all=FALSE),c("X",NA,NA))
    expect_identical(stri_replace_all_charclass("X",NA,"?", vectorize_all=FALSE),NA_character_)
    expect_identical(stri_replace_all_charclass(NA,NA,"?", vectorize_all=FALSE),NA_character_)
    expect_identical(stri_replace_all_charclass(NA,NA,NA, vectorize_all=FALSE),NA_character_)
@@ -24,14 +24,16 @@ test_that("stri_replace_all_charclass-vectorize_all=FALSE", {
    expect_identical(stri_replace_all_charclass("aaa bbb,   ccc", c("\\p{L}", "\\p{Z}"), merge=TRUE,
       c("xxxxx", ""), vectorize_all=FALSE), "xxxxxxxxxx,xxxxx")
 
+   expect_identical(stri_replace_all_charclass(c("1RR", "NURR", "3"), c("[R]", "[L]"), c("L", NA), vectorize_all=FALSE), c(NA, NA, "3"))
+
 })
 
 test_that("stri_replace_all_charclass", {
    expect_identical(stri_replace_all_charclass(character(0),"\\p{Z}",""),character(0))
    expect_identical(stri_replace_all_charclass("b",character(0),"a"),character(0))
    expect_error(stri_replace_all_charclass("b","","a"))
    expect_identical(stri_replace_all_charclass(NA,"\\p{WHITE_SPACE}","?"),NA_character_)
-   expect_identical(stri_replace_all_charclass("X","\\p{WHITE_SPACE}",NA),NA_character_)
+   expect_identical(stri_replace_all_charclass(c(" ","X"),"\\p{WHITE_SPACE}",NA),c(NA,"X"))
    expect_identical(stri_replace_all_charclass("X",NA,"?"),NA_character_)
    expect_identical(stri_replace_all_charclass(NA,NA,"?"),NA_character_)
    expect_identical(stri_replace_all_charclass(NA,NA,NA),NA_character_)
@@ -45,13 +47,15 @@ test_that("stri_replace_all_charclass", {
    expect_identical(stri_replace_all_charclass(c(stri_dup("Y ", 0:1500)), "\\p{Wspace}", "X"), stri_dup("YX", 0:1500))
    expect_identical(stri_replace_all_charclass("ala   ma \t    \n kota      ", "\\p{WHITESPACE}", " ", merge=TRUE),
          "ala ma kota ")
+
+   expect_identical(stri_replace_all_charclass(c("1", "N", "3"), "\\p{Lu}", NA), c("1", NA, "3"))
 })
 
 
 test_that("stri_replace_first_charclass", {
    expect_identical(stri_replace_first_charclass(character(0),"\\p{Z}",""),character(0))
    expect_identical(stri_replace_first_charclass(NA,"\\p{Wspace}","?"),NA_character_)
-   expect_identical(stri_replace_first_charclass("X","\\p{Wspace}",NA),NA_character_)
+   expect_identical(stri_replace_first_charclass(c(" ","X"),"\\p{Wspace}",NA),c(NA_character_, "X"))
    expect_identical(stri_replace_first_charclass("X",NA,"?"),NA_character_)
    expect_identical(stri_replace_first_charclass(NA,NA,"?"),NA_character_)
    expect_identical(stri_replace_first_charclass(NA,NA,NA),NA_character_)
@@ -60,13 +64,14 @@ test_that("stri_replace_first_charclass", {
    expect_identical(stri_replace_first_charclass(c("a a ", " aa ", "aa  ", "aa"), "\\p{Wspace}", "X"), c("aXa ", "Xaa ", "aaX ", "aa"))
    expect_identical(stri_replace_first_charclass("a1 ", c("\\p{Z}", "\\p{Nd}", "\\p{Ll}", "\\p{P}"), "X"), c("a1X", "aX ", "X1 ", "a1 "))
    expect_identical(stri_replace_first_charclass("a1 ", "\\P{Z}", c("X", "Y")), c("X1 ", "Y1 "))
+   expect_identical(stri_replace_first_charclass(c("1", "N", "3"), "\\p{Lu}", NA), c("1", NA, "3"))
 })
 
 
 test_that("stri_replace_last_charclass", {
    expect_identical(stri_replace_last_charclass(character(0),"\\p{Z}",""),character(0))
    expect_identical(stri_replace_last_charclass(NA,"\\p{Wspace}","?"),NA_character_)
-   expect_identical(stri_replace_last_charclass("X","\\p{Wspace}",NA),NA_character_)
+   expect_identical(stri_replace_last_charclass(c(" ","X"),"\\p{Wspace}",NA),c(NA_character_, "X"))
    expect_identical(stri_replace_last_charclass("X",NA,"?"),NA_character_)
    expect_identical(stri_replace_last_charclass(NA,NA,"?"),NA_character_)
    expect_identical(stri_replace_last_charclass(NA,NA,NA),NA_character_)
@@ -75,4 +80,6 @@ test_that("stri_replace_last_charclass", {
    expect_identical(stri_replace_last_charclass(c("a a ", " aa ", "aa  ", "aa", " aa"), "\\p{Wspace}", "X"), c("a aX", " aaX", "aa X", "aa", "Xaa"))
    expect_identical(stri_replace_last_charclass("a1 ", c("\\p{Z}", "\\p{Nd}", "\\p{Ll}", "\\p{P}"), "X"), c("a1X", "aX ", "X1 ", "a1 "))
    expect_identical(stri_replace_last_charclass("a1 ", "\\P{Z}", c("X", "Y")), c("aX ", "aY "))
+
+   expect_identical(stri_replace_all_charclass(c("1", "N", "3"), "\\p{Lu}", NA), c("1", NA, "3"))
 })

devel/testthat/test-replace-coll.R

@@ -8,7 +8,7 @@ test_that("stri_replace_all_coll", {
    expect_warning(stri_replace_all_coll("a", c("a", "b", "c"), c("b", "d"), vectorize_all=FALSE))
    expect_equivalent(stri_replace_all_coll("a", c("a", NA), c("b", "d"), vectorize_all=FALSE), c(NA_character_))
    expect_equivalent(stri_replace_all_coll(c("a", "b"), c("a", NA), c("b", "d"), vectorize_all=FALSE), c(NA_character_, NA_character_))
-   expect_equivalent(stri_replace_all_coll(c("aba", "bbbb"), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), c(NA, NA_character_))
+   expect_equivalent(stri_replace_all_coll(c("aba", "bbbb"), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), c(NA, "bbbb"))
    expect_equivalent(stri_replace_all_coll(character(0), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), character(0))
    expect_equivalent(stri_replace_all_coll(c("", "", ""), c("a", "c"), c("e", "d"), vectorize_all=FALSE), c("", "", ""))
    expect_equivalent(stri_replace_all_coll(c("abacada", "aaa", "fdsueo"), c("a", "b"), c("x", "y"), vectorize_all = FALSE),
@@ -19,7 +19,9 @@ test_that("stri_replace_all_coll", {
    expect_equivalent(stri_replace_all_coll("The quick brown fox jumped over the lazy dog.",
       c("quick", "brown", "fox", "dog"), c(""), vectorize_all = FALSE),
       "The    jumped over the lazy .")
-   expect_identical(stri_replace_all_coll("X",c("a", "b"),NA, vectorize_all=FALSE),NA_character_)
+   expect_identical(stri_replace_all_coll(c("X", "Y"),c("a", "X", "b"),NA, vectorize_all=FALSE),c(NA, "Y"))
+
+   expect_identical(stri_replace_all_coll(c("1RR", "NURR", "3"), c("RR", "NULL"), c("LL", NA), vectorize_all=FALSE), c("1LL", NA, "3"))
 })
 
 test_that("stri_replace_all_coll", {
@@ -38,6 +40,8 @@ test_that("stri_replace_all_coll", {
    expect_identical(stri_replace_all_coll("ALA MA  KOTA",c(" ", "A", NA) ,""), c("ALAMAKOTA", "L M  KOT", NA))
    expect_identical(stri_replace_all_coll("ALA","BF","HA"),"ALA")
    expect_identical(stri_replace_all_coll("ALA","ALA", c("","RYBA")), c("", "RYBA"))
+
+   expect_identical(stri_replace_all_coll(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
 })
 
 
@@ -55,6 +59,8 @@ test_that("stri_replace_first_coll", {
    expect_identical(stri_replace_first_coll("ALA MA  KOTA",c(" ", "A", NA) ,""), c("ALAMA  KOTA", "LA MA  KOTA", NA))
    expect_identical(stri_replace_first_coll("ALA","BF","HA"),"ALA")
    expect_identical(stri_replace_first_coll("ALA","ALA", c("","RYBA")), c("", "RYBA"))
+
+   expect_identical(stri_replace_first_coll(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
 })
 
 
@@ -73,4 +79,6 @@ test_that("stri_replace_last_coll", {
    expect_identical(stri_replace_last_coll("ALA MA  KOTA",c(" ", "A", NA) ,""), c("ALA MA KOTA", "ALA MA  KOT", NA))
    expect_identical(stri_replace_last_coll("ALA","BF","HA"),"ALA")
    expect_identical(stri_replace_last_coll("ALA","ALA", c("","RYBA")), c("", "RYBA"))
+
+   expect_identical(stri_replace_last_coll(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
 })

devel/testthat/test-replace-fixed.R

@@ -8,7 +8,7 @@ test_that("stri_replace_all_fixed", {
    expect_warning(stri_replace_all_fixed("a", c("a", "b", "c"), c("b", "d"), vectorize_all=FALSE))
    expect_equivalent(stri_replace_all_fixed("a", c("a", NA), c("b", "d"), vectorize_all=FALSE), c(NA_character_))
    expect_equivalent(stri_replace_all_fixed(c("a", "b"), c("a", NA), c("b", "d"), vectorize_all=FALSE), c(NA_character_, NA_character_))
-   expect_equivalent(stri_replace_all_fixed(c("aba", "bbbb"), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), c(NA, NA_character_))
+   expect_equivalent(stri_replace_all_fixed(c("aba", "bbbb"), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), c(NA, "bbbb"))
    expect_equivalent(stri_replace_all_fixed(character(0), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), character(0))
    expect_equivalent(stri_replace_all_fixed(c("", "", ""), c("a", "c"), c("e", "d"), vectorize_all=FALSE), c("", "", ""))
    expect_equivalent(stri_replace_all_fixed(c("abacada", "aaa", "fdsueo"), c("a", "b"), c("x", "y"), vectorize_all = FALSE),
@@ -19,7 +19,9 @@ test_that("stri_replace_all_fixed", {
    expect_equivalent(stri_replace_all_fixed("The quick brown fox jumped over the lazy dog.",
       c("quick", "brown", "fox", "dog"), c(""), vectorize_all = FALSE),
       "The    jumped over the lazy .")
-   expect_identical(stri_replace_all_fixed("X",c("a", "b"),NA, vectorize_all=FALSE),NA_character_)
+   expect_identical(stri_replace_all_fixed(c("Y", "X"),c("a", "b", "X"),NA, vectorize_all=FALSE), c("Y", NA))
+
+   expect_identical(stri_replace_all_fixed(c("1RR", "NURR", "3"), c("RR", "NULL"), c("LL", NA), vectorize_all=FALSE), c("1LL", NA, "3"))
 })
 
 
@@ -38,13 +40,17 @@ test_that("stri_replace_all_fixed", {
    expect_identical(stri_replace_all_fixed(NA,NA,NA),NA_character_)
    expect_warning(stri_replace_all_fixed('fasgasgas',c(" ","o"),1:3))
 
+   expect_identical(stri_replace_all_fixed(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
+
    expect_identical(stri_replace_all_fixed("ALA MA  KOTA",c(" ", "A", NA) ,""), c("ALAMAKOTA", "L M  KOT", NA))
    expect_identical(stri_replace_all_fixed("ALA","BF","HA"),"ALA")
    expect_identical(stri_replace_all_fixed("ALA","ALA", c("","RYBA")), c("", "RYBA"))
 })
 
 
 test_that("stri_replace_first_fixed", {
+   expect_identical(stri_replace_first_fixed(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
+
    expect_identical(stri_replace_first("abcde", fixed="bcd", replacement=""), "ae")
    expect_identical(stri_replace_first_fixed(character(0),1,2),character(0))
    expect_identical(stri_replace_first_fixed("abab123 a","a",1),"1bab123 a")
@@ -69,6 +75,8 @@ test_that("stri_replace_first_fixed", {
 
 
 test_that("stri_replace_last_fixed", {
+   expect_identical(stri_replace_last_fixed(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
+
    expect_identical(stri_replace_last("abcde", fixed="bcd", replacement=""), "ae")
    expect_identical(stri_replace_last_fixed(character(0),1,2),character(0))
    expect_identical(stri_replace_last_fixed("abab123 a","a",1),"abab123 1")

devel/testthat/test-replace-regex.R

@@ -8,7 +8,7 @@ test_that("stri_replace_all_regex [vectorize_all=FALSE]", {
    expect_warning(stri_replace_all_regex("a", c("a", "b", "c"), c("b", "d"), vectorize_all=FALSE))
    expect_equivalent(stri_replace_all_regex("a", c("a", NA), c("b", "d"), vectorize_all=FALSE), c(NA_character_))
    expect_equivalent(stri_replace_all_regex(c("a", "b"), c("a", NA), c("b", "d"), vectorize_all=FALSE), c(NA_character_, NA_character_))
-   expect_equivalent(stri_replace_all_regex(c("aba", "bbbb"), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), c(NA, NA_character_))
+   expect_equivalent(stri_replace_all_regex(c("aba", "bbbb"), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), c(NA, "bbbb"))
    expect_equivalent(stri_replace_all_regex(character(0), c("a", "c"), c(NA, "d"), vectorize_all=FALSE), character(0))
    expect_equivalent(stri_replace_all_regex(c("", "", ""), c("a", "c"), c("e", "d"), vectorize_all=FALSE), c("", "", ""))
    expect_equivalent(stri_replace_all_regex(c("abacada", "aaa", "fdsueo"), c("a+", "b"), c("x", "y"), vectorize_all = FALSE),
@@ -19,7 +19,7 @@ test_that("stri_replace_all_regex [vectorize_all=FALSE]", {
    expect_equivalent(stri_replace_all_regex("The quick brown fox jumped over the lazy dog.",
       c("quick", "brown", "fox", "dog"), c(""), vectorize_all = FALSE),
       "The    jumped over the lazy .")
-   expect_identical(stri_replace_all_regex("X",c("a", "b"),NA, vectorize_all=FALSE),NA_character_)
+   expect_identical(stri_replace_all_regex(c("X","Y"),c("a", "b", "X"),NA, vectorize_all=FALSE),c(NA, "Y"))
 
    expect_identical(stri_replace_all_regex("",c("^.*$","h"),c("hey!", "y"), vectorize_all=FALSE),"yey!")
    expect_identical(stri_replace_all_regex("  ",c("^.*$","h"),c("hey!", "y"), vectorize_all=FALSE),"yey!")
@@ -30,6 +30,8 @@ test_that("stri_replace_all_regex [vectorize_all=FALSE]", {
    # expect_identical(stri_replace_all_regex("ABC", "(.*)", "ONE($1)"), "ONE(A)")
    # expect_identical(stri_replace_all_regex("A", ".*", "ONE($0)"), "ONE(A)")
    # expect_identical(stri_replace_all_regex("A", "^.*", "ONE($0)"), "ONE(A)")
+
+   expect_identical(stri_replace_all_regex(c("1RR", "NURR", "3"), c("RR", "NULL"), c("LL", NA), vectorize_all=FALSE), c("1LL", NA, "3"))
 })
 
 test_that("stri_replace_all_regex", {
@@ -46,6 +48,8 @@ test_that("stri_replace_all_regex", {
    expect_identical(stri_replace_all_regex(NA,NA,NA),NA_character_)
    expect_warning(stri_replace_all_regex('fasgasgas',c(" ","o"),1:3))
 
+   expect_identical(stri_replace_all_regex(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
+
    expect_identical(stri_replace_all_regex("","^.*$","hey!"),"hey!")
    expect_identical(stri_replace_all_regex("  ","^.*$","hey!"),"hey!")
 
@@ -109,6 +113,8 @@ test_that("stri_replace_first_regex", {
       c("\u0105\u0106\u0108\u0107", "\u0105\u0107")) # match of zero length:
    expect_identical(stri_replace_first_regex("","^.*$","hey!"),"hey!")
    expect_identical(stri_replace_first_regex("  ","^.*$","hey!"),"hey!")
+
+   expect_identical(stri_replace_first_regex(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
 })
 
 
@@ -135,4 +141,6 @@ test_that("stri_replace_last_regex", {
 
    expect_identical(stri_replace_last_regex("","^.*$","hey!"),"hey!")
    expect_identical(stri_replace_last_regex("  ","^.*$","hey!"),"hey!")
+
+   expect_identical(stri_replace_last_regex(c("1", "NULL", "3"), "NULL", NA), c("1", NA, "3"))
 })

src/stri_search_class_replace.cpp

@@ -71,6 +71,9 @@ using namespace std;
  *
  * @version 0.3-1 (Marek Gagolewski, 2014-11-04)
  *    Issue #112: str_prepare_arg* retvals were not PROTECTed from gc
+ *
+ * @version 1.0-2 (Marek Gagolewski, 2016-01-30)
+ *    Issue #210: Allow NA replacement
  */
 SEXP stri__replace_all_charclass_yes_vectorize_all(SEXP str, SEXP pattern, SEXP replacement, SEXP merge)
 {
@@ -95,7 +98,7 @@ SEXP stri__replace_all_charclass_yes_vectorize_all(SEXP str, SEXP pattern, SEXP
          i != pattern_cont.vectorize_end();
          i = pattern_cont.vectorize_next(i))
    {
-      if (str_cont.isNA(i) || replacement_cont.isNA(i) || pattern_cont.isNA(i)) {
+      if (str_cont.isNA(i) || pattern_cont.isNA(i)) {
          SET_STRING_ELT(ret, i, NA_STRING);
          continue;
       }
@@ -114,6 +117,11 @@ SEXP stri__replace_all_charclass_yes_vectorize_all(SEXP str, SEXP pattern, SEXP
          continue;
       }
 
+      if (replacement_cont.isNA(i)) {
+         SET_STRING_ELT(ret, i, NA_STRING);
+         continue;
+      }
+
       R_len_t     replacement_cur_n = replacement_cont.get(i).length();
       R_len_t buf_need = str_cur_n+(R_len_t)occurrences.size()*replacement_cur_n-sumbytes;
       buf.resize(buf_need, false/*destroy contents*/);
@@ -150,6 +158,9 @@ SEXP stri__replace_all_charclass_yes_vectorize_all(SEXP str, SEXP pattern, SEXP
  *
  * @version 0.3-1 (Marek Gagolewski, 2014-11-04)
  *    Issue #112: str_prepare_arg* retvals were not PROTECTed from gc
+ *
+ * @version 1.0-2 (Marek Gagolewski, 2016-01-30)
+ *    Issue #210: Allow NA replacement
  */
 SEXP stri__replace_all_charclass_no_vectorize_all(SEXP str, SEXP pattern, SEXP replacement, SEXP merge)
 {
@@ -191,7 +202,7 @@ SEXP stri__replace_all_charclass_no_vectorize_all(SEXP str, SEXP pattern, SEXP r
 
    for (R_len_t i = 0; i<pattern_n; ++i)
    {
-      if (pattern_cont.isNA(i) || replacement_cont.isNA(i)) {
+      if (pattern_cont.isNA(i)) {
          STRI__UNPROTECT_ALL
          return stri__vector_NA_strings(str_n);
       }
@@ -208,6 +219,14 @@ SEXP stri__replace_all_charclass_no_vectorize_all(SEXP str, SEXP pattern, SEXP r
             false /* byte-based indices */
          );
 
+         if (occurrences.size() == 0)
+            continue;
+
+         if (replacement_cont.isNA(i)) {
+            str_cont.setNA(j);
+            continue;
+         }
+
          R_len_t     replacement_cur_n = replacement_cont.get(i).length();
          R_len_t buf_need = str_cur_n+(R_len_t)occurrences.size()*replacement_cur_n-sumbytes;
          buf.resize(buf_need, false/*destroy contents*/);
@@ -296,7 +315,7 @@ SEXP stri__replace_firstlast_charclass(SEXP str, SEXP pattern, SEXP replacement,
          i != pattern_cont.vectorize_end();
          i = pattern_cont.vectorize_next(i))
    {
-      if (str_cont.isNA(i) || replacement_cont.isNA(i) || pattern_cont.isNA(i)) {
+      if (str_cont.isNA(i) || pattern_cont.isNA(i)) {
          SET_STRING_ELT(ret, i, NA_STRING);
          continue;
       }
@@ -337,6 +356,11 @@ SEXP stri__replace_firstlast_charclass(SEXP str, SEXP pattern, SEXP replacement,
          continue;
       }
 
+      if (replacement_cont.isNA(i)) {
+         SET_STRING_ELT(ret, i, NA_STRING);
+         continue;
+      }
+
       R_len_t     replacement_cur_n = replacement_cont.get(i).length();
       const char* replacement_cur_s = replacement_cont.get(i).c_str();
       R_len_t buf_need = str_cur_n+replacement_cur_n-(j-jlast);