GitLab

at 2024-06-03T14:38:08-07:00

Fix typo in the lower limit for char-upcase.

We use a lower limit of `#\a`, but we really should have used `#\``.

Fix equal-char-code to use CL lowercase

Previously, `equal-char-code` used the Unicode function
`unicode-lower` to convert the character to lower case.  That's not
what we want; we want to use the CL lower case character.  Thus, use
`case-mapping-lower-case` to get the desired character code.

Add a test for `char-equal` that shows we've fixed this.

Simplify equal-char-code as suggested.

Replace the nested `if`'s with a `cond`.  Change the bounds to be
inclusive to make it easier to reason about the range of characters.

Fix a typo and add a comment.

 (defmacro equal-char-code (character)

   `(let ((ch (char-code ,character)))

-     ;; Handle ASCII separately for bootstrapping and for unidata missing.

-     (if (< (char-code #\@) ch (char-code #\[))

-	 (+ ch 32)

-	 #-(and unicode (not unicode-bootstrap))

-	 ch

-	 #+(and unicode (not unicode-bootstrap))

-	 (if (> ch +ascii-limit+) (unicode-lower ch) ch))))

-

+     ;; Handle ASCII separately for bootstrapping.

+     (cond ((<= (char-code #\A) ch (char-code #\Z))

+            (logxor ch #x20))

+	   #+(and unicode (not unicode-bootstrap))

+           ((> ch +ascii-limit+)

+            (case-mapping-lower-case ch))

+           (t

+            ch))))

 (defun char-equal (character &rest more-characters)

   "Returns T if all of its arguments are the same character.

        x)

   #+(and unicode (not unicode-bootstrap))

   '(let ((m (char-code x)))

-    (cond ((< (char-code #\a) m (char-code #\{))

+    (cond ((< (char-code #\`) m (char-code #\{))

            (code-char (logxor m #x20)))

           ((> m lisp::+ascii-limit+)

            (code-char (lisp::case-mapping-upper-case m)))

+;; Tests of char functions

+

+(defpackage :char-tests

+  (:use :cl :lisp-unit))

+

+(in-package "CHAR-TESTS")

+

+(define-test char-equal

+  (:tag :issues)

+  (let ((test-codes

+          ;; Find all the codes where the CL lower case character

+          ;; doesn't match the Unicode lower case character.

+          (loop for code from 128 below char-code-limit

+               for ch = (code-char code)

+               when (/= (char-code (char-downcase ch)) (or (lisp::unicode-lower code) code))

+                 collect code)))

+    (dolist (code test-codes)

+      ;; Verify that we convert to the CL lower case character instead

+      ;; of the Unicode lower case character for the cases where these

+      ;; are different.

+      (assert-false (char-equal (code-char (lisp::unicode-lower code))

+                                (code-char code))

+                    code

+                    (lisp::unicode-lower code)))))

Raymond Toy pushed to branch issue-326-char-casing-cleanup at cmucl / cmucl

Commits:

3 changed files:

Changes: