summaryrefslogtreecommitdiff
path: root/libguile/strings.c
diff options
context:
space:
mode:
authorMichael Gran <spk121@yahoo.com>2009-08-23 06:50:45 -0700
committerMichael Gran <spk121@yahoo.com>2009-08-23 09:29:45 -0700
commit587a33556fdef90025c1b7d4d172af649c8ebba8 (patch)
tree97792a8e7e1a2b9cb05596df2332294a6c316671 /libguile/strings.c
parent27646f414e9350c2bf9f35982082bcabfb475c5d (diff)
downloadguile-587a33556fdef90025c1b7d4d172af649c8ebba8.tar.gz
Modify socket and time functions for wide strings
* libguile/socket.c (scm_recv): receive the message without holding the stringbuf writing lock (scm_send): try to narrow a string before using it * libguile/stime.c (strftime): convert string to UTF-8 so that it can be safely passed to strftime (strptime): convert input string to UTF-8 so that it can be safely passed through strptime * libguile/strings.c (narrow_stringbuf): new function (scm_i_try_narrow_string): new function * libguile/strings.h: new declaration for scm_i_try_narrow_string
Diffstat (limited to 'libguile/strings.c')
-rw-r--r--libguile/strings.c143
1 files changed, 133 insertions, 10 deletions
diff --git a/libguile/strings.c b/libguile/strings.c
index c6464de6b..59487bd06 100644
--- a/libguile/strings.c
+++ b/libguile/strings.c
@@ -239,6 +239,36 @@ widen_stringbuf (SCM buf)
}
}
+/* Convert a stringbuf of 32-bit UCS-4-encoded characters to one
+ containing 8-bit Latin-1-encoded characters, if possible. */
+static void
+narrow_stringbuf (SCM buf)
+{
+ size_t i, len;
+ scm_t_wchar *wmem;
+ char *mem;
+
+ if (!STRINGBUF_WIDE (buf))
+ return;
+
+ len = STRINGBUF_OUTLINE_LENGTH (buf);
+ i = 0;
+ wmem = STRINGBUF_WIDE_CHARS (buf);
+ while (i < len)
+ if (wmem[i++] > 0xFF)
+ return;
+
+ mem = scm_gc_malloc (sizeof (char) * (len + 1), "string");
+ for (i = 0; i < len; i++)
+ mem[i] = (unsigned char) wmem[i];
+
+ scm_gc_free (wmem, sizeof (scm_t_wchar) * (len + 1), "string");
+
+ SCM_SET_CELL_WORD_0 (buf, SCM_CELL_WORD_0 (buf) ^ STRINGBUF_F_WIDE);
+ SCM_SET_CELL_WORD_1 (buf, mem);
+ SCM_SET_CELL_WORD_2 (buf, len);
+}
+
scm_i_pthread_mutex_t stringbuf_write_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER;
/* Copy-on-write strings.
@@ -459,6 +489,18 @@ scm_i_is_narrow_string (SCM str)
return !STRINGBUF_WIDE (STRING_STRINGBUF (str));
}
+/* Try to coerce a string to be narrow. It if is narrow already, do
+ nothing. If it is wide, shrink it to narrow if none of its
+ characters are above 0xFF. Return true if the string is narrow or
+ was made to be narrow. */
+int
+scm_i_try_narrow_string (SCM str)
+{
+ narrow_stringbuf (STRING_STRINGBUF (str));
+
+ return scm_i_is_narrow_string (str);
+}
+
/* Returns a pointer to the 8-bit Latin-1 encoded character array of
STR. */
const char *
@@ -623,7 +665,7 @@ scm_i_string_set_x (SCM str, size_t p, scm_t_wchar chr)
if (scm_i_is_narrow_string (str))
{
char *dst = scm_i_string_writable_chars (str);
- dst[p] = (char) (unsigned char) chr;
+ dst[p] = chr;
}
else
{
@@ -633,7 +675,7 @@ scm_i_string_set_x (SCM str, size_t p, scm_t_wchar chr)
}
/* Symbols.
-
+
Basic symbol creation and accessing is done here, the rest is in
symbols.[hc]. This has been done to keep stringbufs and the
internals of strings and string-like objects confined to this file.
@@ -866,7 +908,7 @@ SCM_DEFINE (scm_sys_string_dump, "%string-dump", 1, 0, 0, (SCM str),
else
e5 = scm_cons (scm_from_locale_symbol ("read-only"),
SCM_BOOL_F);
-
+
/* Stringbuf info */
if (!STRINGBUF_WIDE (buf))
{
@@ -1426,6 +1468,80 @@ scm_from_locale_string (const char *str)
return scm_from_locale_stringn (str, -1);
}
+static SCM
+scm_from_stringn (const char *str, size_t len, const char *encoding,
+ scm_t_string_failed_conversion_handler handler)
+{
+ size_t u32len, i;
+ scm_t_wchar *u32;
+ int wide = 0;
+ SCM res;
+
+ u32len = 0;
+ u32 = (scm_t_wchar *) u32_conv_from_encoding (encoding,
+ (enum iconv_ilseq_handler)
+ handler,
+ str, len,
+ NULL,
+ NULL, &u32len);
+
+ if (u32 == NULL)
+ {
+ if (errno == ENOMEM)
+ scm_memory_error ("locale string conversion");
+ else
+ {
+ /* There are invalid sequences in the input string. Since
+ it is partially nonsense, what is the best strategy for
+ printing it in the error message? */
+ SCM errstr;
+ char *dst;
+ /* We'll just print it unconverted and hope for the best. */
+ errstr = scm_i_make_string (len, &dst);
+ memcpy (dst, str, len);
+ scm_misc_error (NULL, "input locale conversion error from ~s: ~s",
+ scm_list_2 (scm_from_locale_string (encoding),
+ errstr));
+ scm_remember_upto_here_1 (errstr);
+ }
+ }
+
+ i = 0;
+ while (i < u32len)
+ if (u32[i++] > 0xFF)
+ {
+ wide = 1;
+ break;
+ }
+
+ if (!wide)
+ {
+ char *dst;
+ res = scm_i_make_string (u32len, &dst);
+ for (i = 0; i < u32len; i ++)
+ dst[i] = (unsigned char) u32[i];
+ dst[u32len] = '\0';
+ }
+ else
+ {
+ scm_t_wchar *wdst;
+ res = scm_i_make_wide_string (u32len, &wdst);
+ u32_cpy ((scm_t_uint32 *) wdst, (scm_t_uint32 *) u32, u32len);
+ wdst[u32len] = 0;
+ }
+
+ free (u32);
+ return res;
+}
+
+SCM
+scm_i_from_utf8_string (const scm_t_uint8 *str)
+{
+ return scm_from_stringn ((const char *) str,
+ strlen ((char *) str), "UTF-8",
+ SCM_FAILED_CONVERSION_ERROR);
+}
+
/* Create a new scheme string from the C string STR. The memory of
STR may be used directly as storage for the new string. */
SCM
@@ -1519,16 +1635,15 @@ scm_to_locale_stringn (SCM str, size_t * lenp)
/* In the future, enc will hold the port's encoding. */
enc = NULL;
- return scm_to_stringn (str, lenp, enc,
+ return scm_to_stringn (str, lenp, enc,
SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE);
}
/* Low-level scheme to C string conversion function. */
char *
-scm_to_stringn (SCM str, size_t * lenp, const char *encoding,
+scm_to_stringn (SCM str, size_t *lenp, const char *encoding,
scm_t_string_failed_conversion_handler handler)
{
- static const char iso[11] = "ISO-8859-1";
char *buf;
size_t ilen, len, i;
@@ -1544,7 +1659,7 @@ scm_to_stringn (SCM str, size_t * lenp, const char *encoding,
*lenp = 0;
return buf;
}
-
+
if (lenp == NULL)
for (i = 0; i < ilen; i++)
if (scm_i_string_ref (str, i) == '\0')
@@ -1570,16 +1685,16 @@ scm_to_stringn (SCM str, size_t * lenp, const char *encoding,
}
}
-
+
buf = NULL;
len = 0;
- buf = u32_conv_to_encoding (iso,
+ buf = u32_conv_to_encoding (encoding ? encoding : "ISO-8859-1",
(enum iconv_ilseq_handler) handler,
(scm_t_uint32 *) scm_i_string_wide_chars (str),
ilen, NULL, NULL, &len);
if (buf == NULL)
scm_misc_error (NULL, "cannot convert to output locale ~s: \"~s\"",
- scm_list_2 (scm_from_locale_string (iso), str));
+ scm_list_2 (scm_from_locale_string (encoding), str));
if (handler == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
unistring_escapes_to_guile_escapes (&buf, &len);
@@ -1602,6 +1717,14 @@ scm_to_locale_string (SCM str)
return scm_to_locale_stringn (str, NULL);
}
+scm_t_uint8 *
+scm_i_to_utf8_string (SCM str)
+{
+ char *u8str;
+ u8str = scm_to_stringn (str, NULL, "UTF-8", SCM_FAILED_CONVERSION_ERROR);
+ return (scm_t_uint8 *) u8str;
+}
+
size_t
scm_to_locale_stringbuf (SCM str, char *buf, size_t max_len)
{