Implement special sigma casing rule (bug#24603)

In Greek, a sigma character has two lower case forms which depend on
their position in the word.  Implement logic determining it.

* src/casefiddle.c (struct casing_context, case_character_impl): Don’t
assume inword is true when flag is CASE_UP and false when flag is
CASE_DOWN.  For final sigma detection we need this information tracked
reliably;.
(CAPITAL_SIGMA, SMALL_SIGMA, SMALL_FINAL_SIGMA): New macros defining
Unicode code point of different forms of sigma letter.
(case_character): Implement support for final sigma casing.
(do_casify_multibyte_string, do_casify_multibyte_region): Update after
changes to case_character.

* test/src/casefiddle-tests.el (casefiddle-tests-casing): Add test
cases for final sigma.

diff --git a/etc/NEWS b/etc/NEWS
index cc02e07f5621ed3d45bd6a8ba47168204cae3a22..3574d0fb694595ce7b5fe446befa0a5a254f7c49 100644 (file)
--- a/etc/NEWS
+++ b/etc/NEWS
@@ -367,6 +367,11 @@ of incorrect DŽungla).
 *** Characters which turn into multiple ones when cased are correctly handled.
 For example, fi ligature is converted to FI when upper cased.
 
+*** Greek small sigma is correctly handled when at the end of the word.
+Strings such as ΌΣΟΣ are now correctly converted to Όσος when
+capitalized instead of incorrect Όσοσ (compare lowercase sigma at the
+end of the word).
+
 \f
 * Changes in Specialized Modes and Packages in Emacs 26.1
 

diff --git a/src/casefiddle.c b/src/casefiddle.c
index 10674d963ecf5454fd3682ec87634d7ef14362f9..6fe584b8302c8c21590c7ca557cdfe4be24fe8d3 100644 (file)
--- a/src/casefiddle.c
+++ b/src/casefiddle.c
@@ -46,9 +46,7 @@ struct casing_context {
      When run on a buffer, syntax_prefix_flag_p is taken into account when
      determined inword flag. */
   bool inbuffer;
-  /* Conceptually, this denotes whether we are inside of a word except
-     that if flag is CASE_UP it’s always false and if flag is CASE_DOWN
-     this is always true. */
+  /* Whether we are inside of a word. */
   bool inword;
 };
 
@@ -59,7 +57,7 @@ prepare_casing_context (struct casing_context *ctx,
 {
   ctx->flag = flag;
   ctx->inbuffer = inbuffer;
-  ctx->inword = flag == CASE_DOWN;
+  ctx->inword = false;
   ctx->titlecase_char_table = (int)flag < (int)CASE_CAPITALIZE ? Qnil :
     uniprop_table (intern_c_string ("titlecase"));
   ctx->specialcase_char_tables[CASE_UP] = flag == CASE_DOWN ? Qnil :
@@ -101,15 +99,16 @@ case_character_impl (struct casing_str_buf *buf,
 
   /* Update inword state */
   was_inword = ctx->inword;
-  if ((int) ctx->flag >= (int) CASE_CAPITALIZE)
-    ctx->inword = SYNTAX (ch) == Sword &&
-      (!ctx->inbuffer || was_inword || !syntax_prefix_flag_p (ch));
+  ctx->inword = SYNTAX (ch) == Sword &&
+    (!ctx->inbuffer || was_inword || !syntax_prefix_flag_p (ch));
 
   /* Normalise flag so its one of CASE_UP, CASE_DOWN or CASE_CAPITALIZE. */
-  if (!was_inword)
-    flag = ctx->flag == CASE_UP ? CASE_UP : CASE_CAPITALIZE;
+  if (ctx->flag == CASE_CAPITALIZE)
+    flag = (enum case_action)((int)ctx->flag - was_inword);
   else if (ctx->flag != CASE_CAPITALIZE_UP)
-    flag = CASE_DOWN;
+    flag = ctx->flag;
+  else if (!was_inword)
+    flag = CASE_CAPITALIZE;
   else
     {
       cased = ch;
@@ -150,7 +149,18 @@ case_character_impl (struct casing_str_buf *buf,
   buf->len_bytes = CHAR_STRING (cased, buf->data);
   return cased != ch;
 }
+\f
+/* In Greek, lower case sigma has two forms: one when used in the middle and one
+   when used at the end of a word.  Below is to help handle those cases when
+   casing.
+
+   The rule does not conflict with any other casing rules so while it is
+   a conditional one, it is independent on language. */
 
+#define CAPITAL_SIGMA     0x03A3
+#define SMALL_SIGMA       0x03C3
+#define SMALL_FINAL_SIGMA 0x03C2
+\f
 /* Based on CTX, case character CH accordingly.  Update CTX as necessary.
    Return cased character.
 
@@ -164,12 +174,34 @@ case_single_character (struct casing_context *ctx, int ch)
 }
 
 /* Save in BUF result of casing character CH.  Return whether casing changed the
-   character.  This is like case_single_character but also handles one-to-many
-   casing rules. */
-static inline bool
-case_character (struct casing_str_buf *buf, struct casing_context *ctx, int ch)
+   character.
+
+   If not-NULL, NEXT points to the next character in the cased string.  If NULL,
+   it is assumed current character is the last one being cased.  This is used to
+   apply some rules which depend on proceeding state.
+
+   This is like case_single_character but also handles one-to-many casing
+   rules. */
+static bool
+case_character (struct casing_str_buf *buf, struct casing_context *ctx,
+               int ch, const unsigned char *next)
 {
-  return case_character_impl (buf, ctx, ch);
+  bool changed, was_inword;
+
+  was_inword = ctx->inword;
+  changed = case_character_impl (buf, ctx, ch);
+
+  /* If we have just down-cased a capital sigma and the next character no longer
+     has a word syntax (i.e. current character is end of word), use final
+     sigma. */
+  if (was_inword && ch == CAPITAL_SIGMA && changed &&
+      (!next || SYNTAX (STRING_CHAR (next)) != Sword))
+    {
+      buf->len_bytes = CHAR_STRING (SMALL_FINAL_SIGMA, buf->data);
+      buf->len_chars = 1;
+    }
+
+  return changed;
 }
 \f
 static Lisp_Object
@@ -231,7 +263,7 @@ do_casify_multibyte_string (struct casing_context *ctx, Lisp_Object obj)
       if (dst_end - o < sizeof(struct casing_str_buf))
        string_overflow ();
       ch = STRING_CHAR_ADVANCE (src);
-      case_character ((void *)o, ctx, ch);
+      case_character ((void *)o, ctx, ch, size > 1 ? src : NULL);
       n += ((struct casing_str_buf *)o)->len_chars;
       o += ((struct casing_str_buf *)o)->len_bytes;
     }
@@ -382,12 +414,17 @@ do_casify_multibyte_region (struct casing_context *ctx,
   ptrdiff_t pos = *startp, pos_byte = CHAR_TO_BYTE (pos), size = *endp - pos;
   ptrdiff_t opoint = PT, added = 0;
   struct casing_str_buf buf;
-  int ch, cased, len;
+  bool changed;
+  int ch, len;
 
   for (; size; --size)
     {
       ch = STRING_CHAR_AND_LENGTH (BYTE_POS_ADDR (pos_byte), len);
-      if (!case_character (&buf, ctx, ch))
+      changed = case_character (
+         &buf, ctx, ch,
+         size > 1 ? BYTE_POS_ADDR (pos_byte + len) : NULL);
+
+      if (!changed)
        {
          pos_byte += len;
          ++pos;

diff --git a/test/src/casefiddle-tests.el b/test/src/casefiddle-tests.el
index dd260633f4c63417ff076e8b89dc7fdcdc2237a6..234d233c71a501d4699252eeaddf2a9048469d76 100644 (file)
--- a/test/src/casefiddle-tests.el
+++ b/test/src/casefiddle-tests.el
@@ -195,13 +195,16 @@
         ("define" "DEFINE" "define" "Define" "Define")
         ("fish" "FISH" "fish" "Fish" "Fish")
         ("Straße" "STRASSE" "straße" "Straße" "Straße")
-        ;; FIXME(bug#24603): Everything below is broken at the moment.
-        ;; Here’s what should happen:
-        ;;("ΌΣΟΣ" "ΌΣΟΣ" "όσος" "Όσος" "Όσος")
-        ;; And here’s what is actually happening:
-        ("ΌΣΟΣ" "ΌΣΟΣ" "όσοσ" "Όσοσ" "ΌΣΟΣ")
 
-        ("όσος" "ΌΣΟΣ" "όσος" "Όσος" "Όσος"))))))
+        ;; The word repeated twice to test behaviour at the end of a word
+        ;; inside of an input string as well as at the end of the string.
+        ("ΌΣΟΣ ΌΣΟΣ" "ΌΣΟΣ ΌΣΟΣ" "όσος όσος" "Όσος Όσος" "ΌΣΟΣ ΌΣΟΣ")
+        ;; What should be done with sole sigma?  It is ‘final’ but on the
+        ;; other hand it does not form a word.  We’re using regular sigma.
+        ("Σ Σ" "Σ Σ" "σ σ" "Σ Σ" "Σ Σ")
+        ("όσος" "ΌΣΟΣ" "όσος" "Όσος" "Όσος")
+        ;; If sigma is already lower case, we don’t want to change it.
+        ("όσοσ" "ΌΣΟΣ" "όσοσ" "Όσοσ" "Όσοσ"))))))
 
 (ert-deftest casefiddle-tests-casing-byte8 ()
   (should-not