[0032] 优化 char-alphabetic?, char-whitespace?, char-upcase, char-foldcase, char-downcase 的性能

bench/char-alphabetic-p.scm

@@ -0,0 +1,67 @@
+;; char-alphabetic? 性能测试
+
+(import (scheme base)
+  (scheme char)
+  (liii timeit)
+)
+
+(define iterations 100000)
+
+(define (bench-ascii-letter)
+  (timeit (lambda () (char-alphabetic? #\a)) '() iterations)
+)
+
+(define (bench-ascii-digit)
+  (timeit (lambda () (char-alphabetic? #\5)) '() iterations)
+)
+
+(define (bench-ascii-space)
+  (timeit (lambda () (char-alphabetic? #\space)) '() iterations)
+)
+
+(define (bench-unicode-cjk)
+  (timeit (lambda () (char-alphabetic? #\x4E2D)) '() iterations)
+)
+
+(define (bench-unicode-symbol)
+  (timeit (lambda () (char-alphabetic? #\x2600)) '() iterations)
+)
+
+(define (bench-beyond-max)
+  (timeit (lambda () (char-alphabetic? #\x40000)) '() iterations)
+)
+
+(display "=== char-alphabetic? 性能测试 ===\n")
+(display "迭代次数: ")
+(display iterations)
+(display "\n\n")
+
+(display "ASCII 字母 (#\\a) -> 命中:\n")
+(display "  ")
+(display (bench-ascii-letter))
+(display " 秒\n\n")
+
+(display "ASCII 数字 (#\\5) -> 未命中:\n")
+(display "  ")
+(display (bench-ascii-digit))
+(display " 秒\n\n")
+
+(display "ASCII 空格 -> 未命中:\n")
+(display "  ")
+(display (bench-ascii-space))
+(display " 秒\n\n")
+
+(display "Unicode CJK (#\\x4E2D) -> 命中:\n")
+(display "  ")
+(display (bench-unicode-cjk))
+(display " 秒\n\n")
+
+(display "Unicode 符号 (#\\x2600) -> 未命中:\n")
+(display "  ")
+(display (bench-unicode-symbol))
+(display " 秒\n\n")
+
+(display "超出上届 (#\\x40000) -> 快速拒绝:\n")
+(display "  ")
+(display (bench-beyond-max))
+(display " 秒\n")

bench/char-downcase.scm

@@ -0,0 +1,67 @@
+;; char-downcase 性能测试
+
+(import (scheme base)
+  (scheme char)
+  (liii timeit)
+)
+
+(define iterations 100000)
+
+(define (bench-ascii-upper)
+  (timeit (lambda () (char-downcase #\A)) '() iterations)
+)
+
+(define (bench-ascii-lower)
+  (timeit (lambda () (char-downcase #\a)) '() iterations)
+)
+
+(define (bench-ascii-digit)
+  (timeit (lambda () (char-downcase #\5)) '() iterations)
+)
+
+(define (bench-ascii-space)
+  (timeit (lambda () (char-downcase #\space)) '() iterations)
+)
+
+(define (bench-unicode-upper)
+  (timeit (lambda () (char-downcase #\x391)) '() iterations)
+)
+
+(define (bench-unicode-lower)
+  (timeit (lambda () (char-downcase #\x3B1)) '() iterations)
+)
+
+(display "=== char-downcase 性能测试 ===\n")
+(display "迭代次数: ")
+(display iterations)
+(display "\n\n")
+
+(display "ASCII 大写 (#\\A) -> 命中前部分支:\n")
+(display "  ")
+(display (bench-ascii-upper))
+(display " 秒\n\n")
+
+(display "ASCII 小写 (#\\a) -> 未命中:\n")
+(display "  ")
+(display (bench-ascii-lower))
+(display " 秒\n\n")
+
+(display "ASCII 数字 (#\\5) -> 未命中:\n")
+(display "  ")
+(display (bench-ascii-digit))
+(display " 秒\n\n")
+
+(display "ASCII 空格 -> 未命中:\n")
+(display "  ")
+(display (bench-ascii-space))
+(display " 秒\n\n")
+
+(display "Unicode 大写 (#\\x391) -> 命中中后部分支:\n")
+(display "  ")
+(display (bench-unicode-upper))
+(display " 秒\n\n")
+
+(display "Unicode 小写 (#\\x3B1) -> 未命中:\n")
+(display "  ")
+(display (bench-unicode-lower))
+(display " 秒\n")

bench/char-foldcase.scm

@@ -0,0 +1,67 @@
+;; char-foldcase 性能测试
+
+(import (scheme base)
+  (scheme char)
+  (liii timeit)
+)
+
+(define iterations 100000)
+
+(define (bench-ascii-upper)
+  (timeit (lambda () (char-foldcase #\A)) '() iterations)
+)
+
+(define (bench-ascii-lower)
+  (timeit (lambda () (char-foldcase #\a)) '() iterations)
+)
+
+(define (bench-ascii-digit)
+  (timeit (lambda () (char-foldcase #\5)) '() iterations)
+)
+
+(define (bench-ascii-space)
+  (timeit (lambda () (char-foldcase #\space)) '() iterations)
+)
+
+(define (bench-special-383)
+  (timeit (lambda () (char-foldcase #\x17F)) '() iterations)
+)
+
+(define (bench-special-304)
+  (timeit (lambda () (char-foldcase #\x130)) '() iterations)
+)
+
+(display "=== char-foldcase 性能测试 ===\n")
+(display "迭代次数: ")
+(display iterations)
+(display "\n\n")
+
+(display "ASCII 大写 (#\\A) -> 回退 char-downcase, 命中前部分支:\n")
+(display "  ")
+(display (bench-ascii-upper))
+(display " 秒\n\n")
+
+(display "ASCII 小写 (#\\a) -> 回退 char-downcase, 未命中:\n")
+(display "  ")
+(display (bench-ascii-lower))
+(display " 秒\n\n")
+
+(display "ASCII 数字 (#\\5) -> 回退 char-downcase, 未命中:\n")
+(display "  ")
+(display (bench-ascii-digit))
+(display " 秒\n\n")
+
+(display "ASCII 空格 -> 回退 char-downcase, 未命中:\n")
+(display "  ")
+(display (bench-ascii-space))
+(display " 秒\n\n")
+
+(display "特殊码点 #\\x17F -> 命中 foldcase 自身分支:\n")
+(display "  ")
+(display (bench-special-383))
+(display " 秒\n\n")
+
+(display "特殊码点 #\\x130 -> 命中 foldcase 自身分支:\n")
+(display "  ")
+(display (bench-special-304))
+(display " 秒\n")

bench/char-upcase.scm

@@ -0,0 +1,67 @@
+;; char-upcase 性能测试
+
+(import (scheme base)
+  (scheme char)
+  (liii timeit)
+)
+
+(define iterations 100000)
+
+(define (bench-ascii-lower)
+  (timeit (lambda () (char-upcase #\a)) '() iterations)
+)
+
+(define (bench-ascii-upper)
+  (timeit (lambda () (char-upcase #\A)) '() iterations)
+)
+
+(define (bench-ascii-digit)
+  (timeit (lambda () (char-upcase #\5)) '() iterations)
+)
+
+(define (bench-ascii-space)
+  (timeit (lambda () (char-upcase #\space)) '() iterations)
+)
+
+(define (bench-unicode-lower)
+  (timeit (lambda () (char-upcase #\x3B1)) '() iterations)
+)
+
+(define (bench-unicode-upper)
+  (timeit (lambda () (char-upcase #\x391)) '() iterations)
+)
+
+(display "=== char-upcase 性能测试 ===\n")
+(display "迭代次数: ")
+(display iterations)
+(display "\n\n")
+
+(display "ASCII 小写 (#\\a) -> 命中前部分支:\n")
+(display "  ")
+(display (bench-ascii-lower))
+(display " 秒\n\n")
+
+(display "ASCII 大写 (#\\A) -> 未命中:\n")
+(display "  ")
+(display (bench-ascii-upper))
+(display " 秒\n\n")
+
+(display "ASCII 数字 (#\\5) -> 未命中:\n")
+(display "  ")
+(display (bench-ascii-digit))
+(display " 秒\n\n")
+
+(display "ASCII 空格 -> 未命中:\n")
+(display "  ")
+(display (bench-ascii-space))
+(display " 秒\n\n")
+
+(display "Unicode 小写 (#\\x3B1) -> 命中中后部分支:\n")
+(display "  ")
+(display (bench-unicode-lower))
+(display " 秒\n\n")
+
+(display "Unicode 大写 (#\\x391) -> 未命中:\n")
+(display "  ")
+(display (bench-unicode-upper))
+(display " 秒\n")

bench/char-whitespace-p.scm

@@ -0,0 +1,67 @@
+;; char-whitespace? 性能测试
+
+(import (scheme base)
+  (scheme char)
+  (liii timeit)
+)
+
+(define iterations 100000)
+
+(define (bench-space)
+  (timeit (lambda () (char-whitespace? #\space)) '() iterations)
+)
+
+(define (bench-tab)
+  (timeit (lambda () (char-whitespace? #\tab)) '() iterations)
+)
+
+(define (bench-letter)
+  (timeit (lambda () (char-whitespace? #\a)) '() iterations)
+)
+
+(define (bench-digit)
+  (timeit (lambda () (char-whitespace? #\5)) '() iterations)
+)
+
+(define (bench-nbsp)
+  (timeit (lambda () (char-whitespace? #\xA0)) '() iterations)
+)
+
+(define (bench-ideographic-space)
+  (timeit (lambda () (char-whitespace? #\x3000)) '() iterations)
+)
+
+(display "=== char-whitespace? 性能测试 ===\n")
+(display "迭代次数: ")
+(display iterations)
+(display "\n\n")
+
+(display "ASCII 空格 -> 命中:\n")
+(display "  ")
+(display (bench-space))
+(display " 秒\n\n")
+
+(display "ASCII tab -> 命中:\n")
+(display "  ")
+(display (bench-tab))
+(display " 秒\n\n")
+
+(display "ASCII 字母 (#\\a) -> 未命中:\n")
+(display "  ")
+(display (bench-letter))
+(display " 秒\n\n")
+
+(display "ASCII 数字 (#\\5) -> 未命中:\n")
+(display "  ")
+(display (bench-digit))
+(display " 秒\n\n")
+
+(display "Unicode nbsp (#\\xA0) -> 命中:\n")
+(display "  ")
+(display (bench-nbsp))
+(display " 秒\n\n")
+
+(display "Unicode 全角空格 (#\\x3000) -> 命中:\n")
+(display "  ")
+(display (bench-ideographic-space))
+(display " 秒\n")

compat/char-alphabetic-p.scm

@@ -1,3 +1,5 @@
+(import (scheme char))
+
 (letrec
   ((loop
     (lambda (i end acc count)
@@ -22,10 +24,10 @@
             (if (< cp #x1000)
               (if (< cp #x100)
                 (if (< cp #x10)
-                  (display (string-append "000" (number->string cp 16)))
-                  (display (string-append "00" (number->string cp 16))))
-                (display (string-append "0" (number->string cp 16))))
-              (display (number->string cp 16)))
+                  (display (string-append "000" (string-upcase (number->string cp 16))))
+                  (display (string-append "00" (string-upcase (number->string cp 16)))))
+                (display (string-append "0" (string-upcase (number->string cp 16)))))
+              (display (string-upcase (number->string cp 16))))
             (display ",")
             (display (integer->char cp))
             (newline))