Perf

clj/src/vim_clojure_static/test.clj

@@ -90,6 +90,7 @@
         test-file (str "tmp/" name ".clj")
         syntable (gensym "syntable")]
     `(test/deftest ~name
+       (~io/make-parents ~test-file)
        (spit ~test-file "")
        (let [~syntable (syn-id-names ~test-file ~@strings)]
          ~@(map (fn [{:keys [fmt ss λs]}]

clj/test/vim_clojure_static/syntax_test.clj

@@ -9,6 +9,7 @@
 
 (defpredicates number :clojureNumber)
 (defpredicates kw :clojureKeyword)
+(defpredicates character :clojureCharacter)
 (defpredicates regexp :clojureRegexp)
 (defpredicates regexp-escape :clojureRegexpEscape)
 (defpredicates regexp-char-class :clojureRegexpCharClass)
@@ -37,6 +38,7 @@
     "0x0"        number "+0x0"  number "-0x0"  number ; Hexadecimal zero
     "3/2"        number "+3/2"  number "-3/2"  number ; Rational
     "0/0"        number "+0/0"  number "-0/0"  number ; Rational (not a syntax error)
+    "36r0XYZ"    number "16rFF" number "8r077" number ; Radix
     "2r1"        number "+2r1"  number "-2r1"  number ; Radix
     "36R1"       number "+36R1" number "-36R1" number ; Radix
 
@@ -49,7 +51,7 @@
     "1.0/1" !number
     "01/2" !number
     "1/02" !number
-    "2r2" !number
+    ; "2r2" !number ;; Removed for performance
     "1r0" !number
     "37r36" !number
 
@@ -70,7 +72,7 @@
     "08.9M" !number
     "0x1fM" !number
     "3/4M" !number
-    "2r1M" !number
+    ; "2r1M" !number ;; Removed for performance
 
     ;; Exponential notation
 
@@ -84,27 +86,46 @@
 
 (comment (test #'test-number-literals))
 
-;; TODO: Finish me! (this was in an old git stash)
-;; (defsyntaxtest keywords-test
-;;   (with-format "%s"
-;;     ":1" kw
-;;     ":A" kw
-;;     ":a" kw
-;;     ":αβγ" kw
-;;     "::a" kw
-;;     ":a/b" kw
-;;     ":a:b" kw
-;;     ":a:b/:c:b" kw
-;;     ":a/b/c/d" kw
-;;     "::a/b" !kw
-;;     "::" !kw
-;;     ":a:" !kw
-;;     ":a/" !kw
-;;     ":/" !kw
-;;     ":" !kw
-;;     ))
-;;
-;; (comment (test #'keywords-test))
+(defsyntaxtest test-character-literals
+  ["[%s]"
+   ["\\0"  character  "\\a"    character  "\\Z"     character
+    "\\."  character  "\\\\"   character  "\\❤"     character
+    "\\o7" character  "\\o07"  character  "\\o307"  character
+    "\\o8" !character "\\o477" !character "\\o0007" !character
+
+    "\\u09af" character
+
+    "\\u0"     !character "\\u01"    !character "\\u012" !character
+    "\\u01234" !character "\\u0123g" !character
+
+    "\\newline"   character "\\new" !character
+    "\\tab"       character "\\ta"  !character
+    "\\space"     character "\\sp"  !character
+    "\\return"    character "\\ret" !character
+    "\\backspace" character "\\bs"  !character
+    "\\formfeed"  character "\\ff"  !character]])
+
+(comment (test #'test-character-literals))
+
+(defsyntaxtest keywords-test
+  ["%s"
+   [":1" kw
+    ":A" kw
+    ":a" kw
+    ":αβγ" kw
+    "::a" kw
+    ":a/b" kw
+    ":a:b" kw
+    ":a:b/:c:b" kw
+    ":a/b/c/d" kw
+    "::a/b" kw
+    "::" !kw
+    ":a:" !kw
+    ":a/" !kw
+    ; ":/" !kw ; This is legal, but for simplicity we do not match it
+    ":" !kw]])
+
+(comment (test #'keywords-test))
 
 (defsyntaxtest test-java-regexp-literals
   ["#\"%s\""

ftplugin/clojure.vim

@@ -17,7 +17,7 @@ set cpo&vim
 
 let b:undo_ftplugin = 'setlocal iskeyword< define< formatoptions< comments< commentstring< lispwords<'
 
-setlocal iskeyword+=?,-,*,!,+,/,=,<,>,.,:,$
+setlocal iskeyword+=?,-,*,!,+,/,=,<,>,.,:,$,%,&,\|
 
 " There will be false positives, but this is better than missing the whole set
 " of user-defined def* definitions.

syntax/clojure.vim

@@ -68,40 +68,23 @@ delfunction s:syntax_keyword
 "   * Must not end in a : or /
 "   * Must not have two adjacent colons except at the beginning
 "   * Must not contain any reader metacharacters except for ' and #
-syntax match clojureKeyword "\v<:{1,2}%([^ \n\r\t()\[\]{}";@^`~\\%/]+/)*[^ \n\r\t()\[\]{}";@^`~\\%/]+:@<!>"
+syntax match clojureKeyword "\v<:{1,2}%([^ \n\r\t()\[\]{}";@^`~\\%/]+/)*[^ \n\r\t()\[\]{}";@^`~\\%/]+:@1<!>"
 
 syntax match clojureStringEscape "\v\\%([\\btnfr"]|u\x{4}|[0-3]\o{2}|\o{1,2})" contained
 
 syntax region clojureString matchgroup=clojureStringDelimiter start=/"/ skip=/\\\\\|\\"/ end=/"/ contains=clojureStringEscape,@Spell
 
-syntax match clojureCharacter "\\."
-syntax match clojureCharacter "\\o\%([0-3]\o\{2\}\|\o\{1,2\}\)"
-syntax match clojureCharacter "\\u\x\{4\}"
-syntax match clojureCharacter "\\space"
-syntax match clojureCharacter "\\tab"
-syntax match clojureCharacter "\\newline"
-syntax match clojureCharacter "\\return"
-syntax match clojureCharacter "\\backspace"
-syntax match clojureCharacter "\\formfeed"
+syntax match clojureCharacter "\v\\%(o%([0-3]\o{2}|\o{1,2})|u\x{4}|newline|tab|space|return|backspace|formfeed|.)"
 
-syntax match clojureSymbol "\v%([a-zA-Z!$&*_+=|<.>?-]|[^\x00-\x7F])+%(:?%([a-zA-Z0-9!#$%&*_+=|'<.>/?-]|[^\x00-\x7F]))*[#:]@<!"
+syntax match clojureSymbol "\v%([a-zA-Z!$&*_+=|<.>?-]|[^\x00-\x7F])+%(:?%([a-zA-Z0-9!#$%&*_+=|'<.>/?-]|[^\x00-\x7F]))*[#:]@1<!"
 
-let s:radix_chars = "0123456789abcdefghijklmnopqrstuvwxyz"
-for s:radix in range(2, 36)
-	execute 'syntax match clojureNumber "\v\c<[-+]?' . s:radix . 'r[' . strpart(s:radix_chars, 0, s:radix) . ']+>"'
-endfor
-unlet! s:radix_chars s:radix
-
-syntax match clojureNumber "\v<[-+]?%(0\o*|0x\x+|[1-9]\d*)N?>"
-syntax match clojureNumber "\v<[-+]?%(0|[1-9]\d*|%(0|[1-9]\d*)\.\d*)%(M|[eE][-+]?\d+)?>"
-syntax match clojureNumber "\v<[-+]?%(0|[1-9]\d*)/%(0|[1-9]\d*)>"
+" NB. Correct matching of radix literals was removed for better performance.
+syntax match clojureNumber "\v<[-+]?%(%([2-9]|[12]\d|3[0-6])[rR][[:alnum:]]+|%(0\o*|0x\x+|[1-9]\d*)N?|%(0|[1-9]\d*|%(0|[1-9]\d*)\.\d*)%(M|[eE][-+]?\d+)?|%(0|[1-9]\d*)/%(0|[1-9]\d*))>"
 
 syntax match clojureVarArg "&"
 
-syntax match clojureQuote "'"
-syntax match clojureQuote "`"
-syntax match clojureUnquote "\~"
-syntax match clojureUnquote "\~@"
+syntax match clojureQuote "\v['`]"
+syntax match clojureUnquote "\v\~\@?"
 syntax match clojureMeta "\^"
 syntax match clojureDeref "@"
 syntax match clojureDispatch "\v#[\^'=<_]?"
@@ -134,14 +117,14 @@ syntax match   clojureRegexpBoundary            "[$^]" contained display
 syntax match   clojureRegexpQuantifier          "[?*+][?+]\=" contained display
 syntax match   clojureRegexpQuantifier          "\v\{\d+%(,|,\d+)?}\??" contained display
 syntax match   clojureRegexpOr                  "|" contained display
-syntax match   clojureRegexpBackRef             "\v\\%([1-9]\d*|k\<[a-zA-z]+\>)" contained display
+syntax match   clojureRegexpBackRef             "\v\\%([1-9]\d*|k\<[[:alpha:]]+\>)" contained display
 
 " Mode modifiers, mode-modified spans, lookaround, regular and atomic
 " grouping, and named-capturing.
 syntax match clojureRegexpMod "\v\(@<=\?:" contained display
 syntax match clojureRegexpMod "\v\(@<=\?[xdsmiuU]*-?[xdsmiuU]+:?" contained display
 syntax match clojureRegexpMod "\v\(@<=\?%(\<?[=!]|\>)" contained display
-syntax match clojureRegexpMod "\v\(@<=\?\<[a-zA-Z]+\>" contained display
+syntax match clojureRegexpMod "\v\(@<=\?\<[[:alpha:]]+\>" contained display
 
 syntax region clojureRegexpGroup start="(" skip=/\\\\\|\\)/ end=")" matchgroup=clojureRegexpGroup contained contains=clojureRegexpMod,clojureRegexpQuantifier,clojureRegexpBoundary,clojureRegexpEscape,@clojureRegexpCharClasses
 syntax region clojureRegexp start=/\#"/ skip=/\\\\\|\\"/ end=/"/ contains=@clojureRegexpCharClasses,clojureRegexpEscape,clojureRegexpQuote,clojureRegexpBoundary,clojureRegexpQuantifier,clojureRegexpOr,clojureRegexpBackRef,clojureRegexpGroup keepend