Support for using DOCS to switch to and from UTF-8 mode.

[sgt/charset] / iso2022s.c

diff --git a/iso2022s.c b/iso2022s.c
index 53b5d44..3cf8fa4 100644 (file)
--- a/iso2022s.c
+++ b/iso2022s.c
@@ -1,9 +1,5 @@
 /*
  * iso2022s.c - support for ISO-2022 subset encodings.
- * 
- * (The `s' suffix on the filename is there to leave `iso2022.c'
- * free for the unlikely event that I ever attempt to implement
- * _full_ ISO-2022 in this library!)
  */
 
 #ifndef ENUM_CHARSETS
@@ -42,7 +38,7 @@ struct iso2022 {
      * in ASCII order, so that we can narrow down the list as
      * necessary.
      */
-    struct iso2022_escape *escapes;    /* must be sorted in ASCII order! */
+    const struct iso2022_escape *escapes;/* must be sorted in ASCII order! */
     int nescapes;
 
     /*
@@ -482,13 +478,13 @@ static int iso2022jp_from_ucs(long int ucs, int *subcharset,
        return 0;
     }
 }
-static struct iso2022_escape iso2022jp_escapes[] = {
+static const struct iso2022_escape iso2022jp_escapes[] = {
     {"\033$@", 0xFFFFFFC0, 0x00000002, -1, -1},   /* we ignore this one */
     {"\033$B", 0xFFFFFFC0, 0x00000002, 0, 2},
     {"\033(B", 0xFFFFFFC0, 0x00000000, 0, 0},
     {"\033(J", 0xFFFFFFC0, 0x00000001, 0, 1},
 };
-static struct iso2022 iso2022jp = {
+static const struct iso2022 iso2022jp = {
     iso2022jp_escapes, lenof(iso2022jp_escapes),
     "\1\1\2", "\3", 0x80000000, NULL, FALSE,
     iso2022jp_to_ucs, iso2022jp_from_ucs
@@ -525,12 +521,12 @@ static int iso2022kr_from_ucs(long int ucs, int *subcharset,
        return 0;
     }
 }
-static struct iso2022_escape iso2022kr_escapes[] = {
+static const struct iso2022_escape iso2022kr_escapes[] = {
     {"\016", 0x8FFFFFFF, 0x10000000, -1, -1},
     {"\017", 0x8FFFFFFF, 0x00000000, 0, 0},
     {"\033$)C", 0xFFFFF03F, 0x00000040, 1, 1},   /* bits[11:6] <- 1 */
 };
-static struct iso2022 iso2022kr = {
+static const struct iso2022 iso2022kr = {
     iso2022kr_escapes, lenof(iso2022kr_escapes),
     "\1\2", "\2", 0x80000040, "\033$)C", FALSE,
     iso2022kr_to_ucs, iso2022kr_from_ucs
@@ -692,7 +688,7 @@ static int ctext_from_ucs(long int ucs, int *subcharset, unsigned long *bytes)
  *    in either; we prefer GR where possible since this leads to a
  *    more compact EUC-like encoding.
  */
-static struct iso2022_escape ctext_escapes[] = {
+static const struct iso2022_escape ctext_escapes[] = {
     SEQ("\033$(A", 0|RO, CTEXT_GB2312),
     SEQ("\033$(B", 0|RO, CTEXT_JISX0208),
     SEQ("\033$(C", 0|RO, CTEXT_KSC5601),
@@ -713,8 +709,35 @@ static struct iso2022_escape ctext_escapes[] = {
     SEQ("\033-H", 1, CTEXT_ISO8859_8),
     SEQ("\033-L", 1, CTEXT_ISO8859_5),
     SEQ("\033-M", 1, CTEXT_ISO8859_9),
+
+    /*
+     * Cross-testing against Xutf8TextListToTextProperty() turns up
+     * some additional character sets and ISO 2022 features
+     * supported by that and not by us:
+     * 
+     *         - Single-byte right-hand-half character sets `ESC - f',
+     *           `ESC - T' and `ESC - Y'.
+     * 
+     *         - A really horrifying mechanism used to escape completely
+     *           from the ISO 2022 framework: ESC % / <length>
+     *           <charset-name> <text>. Xutf8* uses this to encode
+     *           "iso8859-14", "iso8859-15" and "big5-0".
+     *            * This mechanism is particularly nasty because we can't
+     *              efficiently encode it on the fly! It requires that the
+     *              length of the text encoded in the foreign charset is
+     *              given _before_ the text in question, so if we're
+     *              receiving one character at a time we simply can't look
+     *              ahead and so we would have to encode each individual
+     *              character in a separate one of these sequences.
+     * 
+     *         - ESC % G and ESC % @ to shift to and from UTF-8 mode, as a
+     *           last resort for anything we still don't support.
+     *            * Interestingly, ctext.ps actually _disallows_ this: it
+     *              says that the above extension mechanism is the only
+     *              one permitted. Ho hum.
+     */
 };
-static struct iso2022 ctext = {
+static const struct iso2022 ctext = {
     ctext_escapes, lenof(ctext_escapes),
     "\1\1\1\1\1\1\1\1\1\1\1\1\2\2\2\2",  /* must match the enum above */
     "", 0x80000000 | (CTEXT_ASCII<<0) | (CTEXT_ISO8859_1<<6), "", TRUE,