vim-patch:7.4.1783

Problem:    The old regexp engine doesn't handle character classes correctly.
            (Manuel Ortega)
Solution:   Use regmbc() instead of regc().  Add a test.

https://github.com/vim/vim/commit/af98a49dd0ef1661b4998f118151fddbf6e4df75

3 files changed, 91 insertions, 23 deletions

diff --git a/src/nvim/regexp.c b/src/nvim/regexp.c
index 122b589d2f..6613d284d7 100644
--- a/src/nvim/regexp.c
+++ b/src/nvim/regexp.c
@@ -2311,14 +2311,14 @@ collection:
             case CLASS_ALNUM:
               for (cu = 1; cu < 128; cu++) {
                 if (isalnum(cu)) {
-                  regc(cu);
+                  regmbc(cu);
                 }
               }
               break;
             case CLASS_ALPHA:
               for (cu = 1; cu < 128; cu++) {
                 if (isalpha(cu)) {
-                  regc(cu);
+                  regmbc(cu);
                 }
               }
               break;
@@ -2327,36 +2327,44 @@ collection:
               regc('\t');
               break;
             case CLASS_CNTRL:
-              for (cu = 1; cu <= 255; cu++)
-                if (iscntrl(cu))
-                  regc(cu);
+              for (cu = 1; cu <= 255; cu++) {
+                if (iscntrl(cu)) {
+                  regmbc(cu);
+                }
+              }
               break;
             case CLASS_DIGIT:
-              for (cu = 1; cu <= 255; cu++)
-                if (ascii_isdigit(cu))
-                  regc(cu);
+              for (cu = 1; cu <= 255; cu++) {
+                if (ascii_isdigit(cu)) {
+                  regmbc(cu);
+                }
+              }
               break;
             case CLASS_GRAPH:
-              for (cu = 1; cu <= 255; cu++)
-                if (isgraph(cu))
-                  regc(cu);
+              for (cu = 1; cu <= 255; cu++) {
+                if (isgraph(cu)) {
+                  regmbc(cu);
+                }
+              }
               break;
             case CLASS_LOWER:
               for (cu = 1; cu <= 255; cu++) {
                 if (vim_islower(cu) && cu != 170 && cu != 186) {
-                  regc(cu);
+                  regmbc(cu);
                 }
               }
               break;
             case CLASS_PRINT:
-              for (cu = 1; cu <= 255; cu++)
-                if (vim_isprintc(cu))
-                  regc(cu);
+              for (cu = 1; cu <= 255; cu++) {
+                if (vim_isprintc(cu)) {
+                  regmbc(cu);
+                }
+              }
               break;
             case CLASS_PUNCT:
               for (cu = 1; cu < 128; cu++) {
                 if (ispunct(cu)) {
-                  regc(cu);
+                  regmbc(cu);
                 }
               }
               break;
@@ -2366,14 +2374,18 @@ collection:
               regc(' ');
               break;
             case CLASS_UPPER:
-              for (cu = 1; cu <= 255; cu++)
-                if (vim_isupper(cu))
-                  regc(cu);
+              for (cu = 1; cu <= 255; cu++) {
+                if (vim_isupper(cu)) {
+                  regmbc(cu);
+                }
+              }
               break;
             case CLASS_XDIGIT:
-              for (cu = 1; cu <= 255; cu++)
-                if (ascii_isxdigit(cu))
-                  regc(cu);
+              for (cu = 1; cu <= 255; cu++) {
+                if (ascii_isxdigit(cu)) {
+                  regmbc(cu);
+                }
+              }
               break;
             case CLASS_TAB:
               regc('\t');
diff --git a/src/nvim/testdir/test_expr_utf8.vim b/src/nvim/testdir/test_expr_utf8.vim
index 9ea6d8872b..6ec3a00154 100644
--- a/src/nvim/testdir/test_expr_utf8.vim
+++ b/src/nvim/testdir/test_expr_utf8.vim
@@ -35,3 +35,59 @@ func Test_strcharpart_utf8()
   call assert_equal('̀', strcharpart('àxb', 1, 1))
   call assert_equal('x', strcharpart('àxb', 2, 1))
 endfunc
+
+func s:classes_test()
+  call assert_equal('Motörhead', matchstr('Motörhead', '[[:print:]]\+'))
+
+  let alphachars = ''
+  let lowerchars = ''
+  let upperchars = ''
+  let alnumchars = ''
+  let printchars = ''
+  let punctchars = ''
+  let xdigitchars = ''
+  let i = 1
+  while i <= 255
+    let c = nr2char(i)
+    if c =~ '[[:alpha:]]'
+      let alphachars .= c
+    endif
+    if c =~ '[[:lower:]]'
+      let lowerchars .= c
+    endif
+    if c =~ '[[:upper:]]'
+      let upperchars .= c
+    endif
+    if c =~ '[[:alnum:]]'
+      let alnumchars .= c
+    endif
+    if c =~ '[[:print:]]'
+      let printchars .= c
+    endif
+    if c =~ '[[:punct:]]'
+      let punctchars .= c
+    endif
+    if c =~ '[[:xdigit:]]'
+      let xdigitchars .= c
+    endif
+    let i += 1
+  endwhile
+
+  call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alphachars)
+  call assert_equal('abcdefghijklmnopqrstuvwxyzµßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', lowerchars)
+  call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ', upperchars)
+  call assert_equal('0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alnumchars)
+  call assert_equal(' !"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ ¡¢£¤¥¦§¨©ª«¬­®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ', printchars)
+  call assert_equal('!"#$%&''()*+,-./:;<=>?@[\]^_`{|}~', punctchars)
+  call assert_equal('0123456789ABCDEFabcdef', xdigitchars)
+endfunc
+
+func Test_classes_re1()
+  set re=1
+  call s:classes_test()
+endfunc
+
+func Test_classes_re2()
+  set re=2
+  call s:classes_test()
+endfunc
diff --git a/src/nvim/version.c b/src/nvim/version.c
index 6736b5149d..27736d9e98 100644
--- a/src/nvim/version.c
+++ b/src/nvim/version.c
@@ -657,7 +657,7 @@ static int included_patches[] = {
   // 1786 NA
   // 1785,
   // 1784 NA
-  // 1783,
+  1783,
   1782,
   // 1781,
   1780,