master 4eebf528fc: Add textsec predicates for different types of confusability

[Lars Ingebrigtsen]

branch: master
commit 4eebf528fca6f6f16168c4f76a653353f3598a35
Author: Lars Ingebrigtsen <larsi@gnus.org>
Commit: Lars Ingebrigtsen <larsi@gnus.org>

    Add textsec predicates for different types of confusability
    
    * lisp/international/textsec.el (textsec-resolved-script-set)
    (textsec-single-script-confusable-p)
    (textsec-mixed-script-confusable-p)
    (textsec-whole-script-confusable-p): New functions.
---
 lisp/international/textsec.el            | 30 ++++++++++++++++++++++++++++++
 test/lisp/international/textsec-tests.el | 19 ++++++++++++++++++-
 2 files changed, 48 insertions(+), 1 deletion(-)

diff --git a/lisp/international/textsec.el b/lisp/international/textsec.el
index 304d69cb89..8095733e09 100644
--- a/lisp/international/textsec.el
+++ b/lisp/international/textsec.el
@@ -192,6 +192,36 @@ This algorithm is described in:
                          (string char)))
                    (ucs-normalize-NFD-string string)))))
 
+(defun textsec-resolved-script-set (string)
+  "Return the resolved script set for STRING.
+This is the minimal covering script set for STRING, but is nil is
+STRING isn't a single script string."
+  (and (textsec-single-script-p string)
+       (textsec-covering-scripts string)))
+
+(defun textsec-single-script-confusable-p (string1 string2)
+  "Say whether STRING1 and STRING2 are single script confusables."
+  (and (equal (textsec-unconfuse-string string1)
+              (textsec-unconfuse-string string2))
+       ;; And they have to have at least one resolved script in
+       ;; common.
+       (seq-intersection (textsec-resolved-script-set string1)
+                         (textsec-resolved-script-set string2))))
+
+(defun textsec-mixed-script-confusable-p (string1 string2)
+  "Say whether STRING1 and STRING2 are mixed script confusables."
+  (and (equal (textsec-unconfuse-string string1)
+              (textsec-unconfuse-string string2))
+       ;; And they have no resolved scripts in common.
+       (null (seq-intersection (textsec-resolved-script-set string1)
+                               (textsec-resolved-script-set string2)))))
+
+(defun textsec-whole-script-confusable-p (string1 string2)
+  "Say whether STRING1 and STRING2 are whole script confusables."
+  (and (textsec-mixed-script-confusable-p string1 string2)
+       (textsec-single-script-p string1)
+       (textsec-single-script-p string2)))
+
 (provide 'textsec)
 
 ;;; textsec.el ends here
diff --git a/test/lisp/international/textsec-tests.el 
b/test/lisp/international/textsec-tests.el
index 50106bb955..15b6b21b34 100644
--- a/test/lisp/international/textsec-tests.el
+++ b/test/lisp/international/textsec-tests.el
@@ -86,11 +86,28 @@
   (should-not (textsec-mixed-numbers-p "8foo8"))
   (should (textsec-mixed-numbers-p "8foo৪")))
 
+(ert-deftest test-resolved ()
+  (should (equal (textsec-resolved-script-set "ljeto")
+                 '(latin)))
+  (should-not (textsec-resolved-script-set "Сirсlе")))
+
 (ert-deftest test-confusable ()
   (should (equal (textsec-unconfuse-string "ljeto") "ljeto"))
   (should (textsec-ascii-confusable-p "ljeto"))
   (should-not (textsec-ascii-confusable-p "ljeto"))
   (should (equal (textsec-unconfuse-string "～") "〜"))
-  (should-not (textsec-ascii-confusable-p "～")))
+  (should-not (textsec-ascii-confusable-p "～"))
+
+  (should (textsec-single-script-confusable-p "ljeto" "ljeto"))
+  (should-not (textsec-single-script-confusable-p "paypal" "pаypаl"))
+  (should-not (textsec-single-script-confusable-p "scope""ѕсоре"))
+
+  (should-not (textsec-mixed-script-confusable-p "ljeto" "ljeto"))
+  (should (textsec-mixed-script-confusable-p "paypal" "pаypаl"))
+  (should (textsec-mixed-script-confusable-p "scope""ѕсоре"))
+
+  (should-not (textsec-whole-script-confusable-p "ljeto" "ljeto"))
+  (should-not (textsec-whole-script-confusable-p "paypal" "pаypаl"))
+  (should (textsec-whole-script-confusable-p "scope""ѕсоре")))
 
 ;;; textsec-tests.el ends here