eval: port v:collate

Cherry-picked from patch v8.2.0988. Required for patch v8.2.1933.

7 files changed, 55 insertions, 7 deletions

diff --git a/runtime/doc/eval.txt b/runtime/doc/eval.txt
index cff87b2fed..01803041a8 100644
--- a/runtime/doc/eval.txt
+++ b/runtime/doc/eval.txt
@@ -1495,6 +1495,15 @@ v:cmdarg	This variable is used for two purposes:
 		   the argument for the ":hardcopy" command.  This can be used
 		   in 'printexpr'.
 
+						*v:collate* *collate-variable*
+v:collate	The current locale setting for collation order of the runtime
+		environment.  This allows Vim scripts to be aware of the
+		current locale encoding.  Technical: it's the value of
+		LC_COLLATE.  When not using a locale the value is "C".
+		This variable can not be set directly, use the |:language|
+		command.
+		See |multi-lang|.
+
 					*v:cmdbang* *cmdbang-variable*
 v:cmdbang	Set like v:cmdarg for a file read/write command.  When a "!"
 		was used the value is 1, otherwise it is 0.  Note that this
diff --git a/runtime/doc/mlang.txt b/runtime/doc/mlang.txt
index 5217b2c160..7ed574d9b8 100644
--- a/runtime/doc/mlang.txt
+++ b/runtime/doc/mlang.txt
@@ -31,6 +31,7 @@ use of "-" and "_".
 :lan[guage] mes[sages]
 :lan[guage] cty[pe]
 :lan[guage] tim[e]
+:lan[guage] col[late]
 			Print the current language (aka locale).
 			With the "messages" argument the language used for
 			messages is printed.  Technical: LC_MESSAGES.
@@ -38,15 +39,19 @@ use of "-" and "_".
 			character encoding is printed.  Technical: LC_CTYPE.
 			With the "time" argument the language used for
 			strftime() is printed.  Technical: LC_TIME.
+			With the "collate" argument the language used for
+			collation order is printed.  Technical: LC_COLLATE.
 			Without argument all parts of the locale are printed
 			(this is system dependent).
 			The current language can also be obtained with the
-			|v:lang|, |v:ctype| and |v:lc_time| variables.
+			|v:lang|, |v:ctype|, |v:collate| and |v:lc_time|
+			variables.
 
 :lan[guage] {name}
 :lan[guage] mes[sages] {name}
 :lan[guage] cty[pe] {name}
 :lan[guage] tim[e] {name}
+:lan[guage] col[late] {name}
 			Set the current language (aka locale) to {name}.
 			The locale {name} must be a valid locale on your
 			system.  Some systems accept aliases like "en" or
@@ -66,7 +71,10 @@ use of "-" and "_".
 			With the "time" argument the language used for time
 			and date messages is set.  This affects strftime().
 			This sets $LC_TIME.
-			Without an argument both are set, and additionally
+			With the "collate" argument the language used for the
+			collation order is set.  This affects sorting of
+			characters. This sets $LC_COLLATE.
+			Without an argument all are set, and additionally
 			$LANG is set.
 			The LC_NUMERIC value will always be set to "C" so
 			that floating point numbers use '.' as the decimal
diff --git a/src/nvim/eval.c b/src/nvim/eval.c
index 05d429c7d5..370ffd152e 100644
--- a/src/nvim/eval.c
+++ b/src/nvim/eval.c
@@ -225,6 +225,7 @@ static struct vimvar {
   VV(VV_EVENT,          "event",            VAR_DICT, VV_RO),
   VV(VV_ECHOSPACE,      "echospace",        VAR_NUMBER, VV_RO),
   VV(VV_ARGV,           "argv",             VAR_LIST, VV_RO),
+  VV(VV_COLLATE,        "collate",          VAR_STRING, VV_RO),
   VV(VV_EXITING,        "exiting",          VAR_NUMBER, VV_RO),
   // Neovim
   VV(VV_STDERR,         "stderr",           VAR_NUMBER, VV_RO),
diff --git a/src/nvim/eval.h b/src/nvim/eval.h
index 3da4bb8655..ef5f90ed8f 100644
--- a/src/nvim/eval.h
+++ b/src/nvim/eval.h
@@ -154,6 +154,7 @@ typedef enum {
     VV_EVENT,
     VV_ECHOSPACE,
     VV_ARGV,
+    VV_COLLATE,
     VV_EXITING,
     // Neovim
     VV_STDERR,
diff --git a/src/nvim/ex_cmds2.c b/src/nvim/ex_cmds2.c
index 950a1a436f..2821074b0d 100644
--- a/src/nvim/ex_cmds2.c
+++ b/src/nvim/ex_cmds2.c
@@ -3627,6 +3627,14 @@ void set_lang_var(void)
   loc = get_locale_val(LC_TIME);
 # endif
   set_vim_var_string(VV_LC_TIME, loc, -1);
+
+# ifdef HAVE_GET_LOCALE_VAL
+  loc = get_locale_val(LC_COLLATE);
+# else
+  // setlocale() not supported: use the default value
+  loc = "C";
+# endif
+  set_vim_var_string(VV_COLLATE, loc, -1);
 }
 
 #ifdef HAVE_WORKING_LIBINTL
@@ -3667,6 +3675,10 @@ void ex_language(exarg_T *eap)
       what = LC_TIME;
       name = skipwhite(p);
       whatstr = "time ";
+    } else if (STRNICMP(eap->arg, "collate", p - eap->arg) == 0) {
+      what = LC_COLLATE;
+      name = skipwhite(p);
+      whatstr = "collate ";
     }
   }
 
@@ -3711,7 +3723,7 @@ void ex_language(exarg_T *eap)
       // Reset $LC_ALL, otherwise it would overrule everything.
       os_setenv("LC_ALL", "", 1);
 
-      if (what != LC_TIME) {
+      if (what != LC_TIME && what != LC_COLLATE) {
         // Tell gettext() what to translate to.  It apparently doesn't
         // use the currently effective locale.
         if (what == LC_ALL) {
@@ -3726,7 +3738,7 @@ void ex_language(exarg_T *eap)
         }
       }
 
-      // Set v:lang, v:lc_time and v:ctype to the final result.
+      // Set v:lang, v:lc_time, v:collate and v:ctype to the final result.
       set_lang_var();
       maketitle();
     }
@@ -3811,12 +3823,15 @@ char_u *get_lang_arg(expand_T *xp, int idx)
   if (idx == 2) {
     return (char_u *)"time";
   }
+  if (idx == 3) {
+    return (char_u *)"collate";
+  }
 
   init_locales();
   if (locales == NULL) {
     return NULL;
   }
-  return locales[idx - 3];
+  return locales[idx - 4];
 }
 
 /// Function given to ExpandGeneric() to obtain the available locales.
diff --git a/src/nvim/ex_docmd.c b/src/nvim/ex_docmd.c
index ae5c334592..555029c1fb 100644
--- a/src/nvim/ex_docmd.c
+++ b/src/nvim/ex_docmd.c
@@ -3642,7 +3642,8 @@ const char * set_one_cmd_context(
     } else {
       if (strncmp(arg, "messages", p - arg) == 0
           || strncmp(arg, "ctype", p - arg) == 0
-          || strncmp(arg, "time", p - arg) == 0) {
+          || strncmp(arg, "time", p - arg) == 0
+          || strncmp(arg, "collate", p - arg) == 0) {
         xp->xp_context = EXPAND_LOCALES;
         xp->xp_pattern = skipwhite((const char_u *)p);
       } else {
diff --git a/src/nvim/testdir/test_cmdline.vim b/src/nvim/testdir/test_cmdline.vim
index 489b2477e6..a1968807ac 100644
--- a/src/nvim/testdir/test_cmdline.vim
+++ b/src/nvim/testdir/test_cmdline.vim
@@ -615,10 +615,20 @@ func Test_cmdline_complete_bang()
 endfunc
 
 funct Test_cmdline_complete_languages()
+  let lang = substitute(execute('language time'), '.*"\(.*\)"$', '\1', '')
+  call assert_equal(lang, v:lc_time)
+
+  let lang = substitute(execute('language ctype'), '.*"\(.*\)"$', '\1', '')
+  call assert_equal(lang, v:ctype)
+
+  let lang = substitute(execute('language collate'), '.*"\(.*\)"$', '\1', '')
+  call assert_equal(lang, v:collate)
+
   let lang = substitute(execute('language messages'), '.*"\(.*\)"$', '\1', '')
+  call assert_equal(lang, v:lang)
 
   call feedkeys(":language \<c-a>\<c-b>\"\<cr>", 'tx')
-  call assert_match('^"language .*\<ctype\>.*\<messages\>.*\<time\>', @:)
+  call assert_match('^"language .*\<collate\>.*\<ctype\>.*\<messages\>.*\<time\>', @:)
 
   if has('unix')
     " TODO: these tests don't work on Windows. lang appears to be 'C'
@@ -633,6 +643,9 @@ funct Test_cmdline_complete_languages()
 
     call feedkeys(":language time \<c-a>\<c-b>\"\<cr>", 'tx')
     call assert_match('^"language .*\<' . lang . '\>', @:)
+
+    call feedkeys(":language collate \<c-a>\<c-b>\"\<cr>", 'tx')
+    call assert_match('^"language .*\<' . lang . '\>', @:)
   endif
 endfunc