Hi,
I'm looking at adding a "java_quoting_style" to quotearg for
use in Bison. After consulting
<URI:http://java.sun.com/docs/books/jls/third_edition/html/lexical.html>,
I've come up with the attached patch which isn't complete
yet.
Unfortunately, my knowledge of C's (gnulib's?) handling of
locales, coding systems, multibyte sequences and wide char-
acters is non-existent, and so I don't know what the result
of `quotearg ("äöü")' should look like and what it should
depend on.
Any comments and suggestions are highly appreciated :-).
TIA,
Tim
diff --git a/lib/quotearg.c b/lib/quotearg.c
index da8ba1e..1c9a873 100644
--- a/lib/quotearg.c
+++ b/lib/quotearg.c
@@ -72,6 +72,7 @@ char const *const quoting_style_args[] =
"escape",
"locale",
"clocale",
+ "java",
0
};
@@ -85,7 +86,8 @@ enum quoting_style const quoting_style_vals[] =
c_maybe_quoting_style,
escape_quoting_style,
locale_quoting_style,
- clocale_quoting_style
+ clocale_quoting_style,
+ java_quoting_style
};
/* The default quoting options. */
@@ -295,6 +297,14 @@ quotearg_buffer_restyled (char *buffer, size_t buffersize,
elide_outer_quotes = false;
break;
+ case java_quoting_style:
+ if (!elide_outer_quotes)
+ STORE ('"');
+ backslash_escapes = true;
+ quote_string = "\"";
+ quote_string_len = 1;
+ break;
+
default:
abort ();
}
@@ -382,13 +392,24 @@ quotearg_buffer_restyled (char *buffer, size_t buffersize,
}
break;
- case '\a': esc = 'a'; goto c_escape;
+ case '\a': esc = quoting_style == java_quoting_style ? '7' : 'a'; goto c_escape;
case '\b': esc = 'b'; goto c_escape;
case '\f': esc = 'f'; goto c_escape;
case '\n': esc = 'n'; goto c_and_shell_escape;
case '\r': esc = 'r'; goto c_and_shell_escape;
case '\t': esc = 't'; goto c_and_shell_escape;
- case '\v': esc = 'v'; goto c_escape;
+
+ case '\v':
+ if (quoting_style != java_quoting_style)
+ {
+ esc = 'v';
+ goto c_escape;
+ }
+ STORE ('\\');
+ STORE ('1');
+ c = '3';
+ break;
+
case '\\': esc = c;
/* No need to escape the escape if we are trying to elide
outer quotes and nothing else is problematic. */
diff --git a/lib/quotearg.h b/lib/quotearg.h
index 2756d76..f2a2064 100644
--- a/lib/quotearg.h
+++ b/lib/quotearg.h
@@ -161,6 +161,17 @@ enum quoting_style
*/
clocale_quoting_style,
+ /* Quote names as for a Java language string.
+
+ quotearg_buffer:
+ "\"simple\"", "\"\\0 \\t\\n'\\\"\\033??/\\\\\"", "\"a:b\""
+ quotearg:
+ "\"simple\"", "\"\\0 \\t\\n'\\\"\\033??/\\\\\"", "\"a:b\""
+ quotearg_colon:
+ "\"simple\"", "\"\\0 \\t\\n'\\\"\\033??/\\\\\"", "\"a\\:b\""
+ */
+ java_quoting_style,
+
/* Like clocale_quoting_style except use the custom quotation marks
set by set_custom_quoting. If custom quotation marks are not
set, the behavior is undefined.
diff --git a/tests/test-quotearg-simple.c b/tests/test-quotearg-simple.c
index 4d9c87c..2524046 100644
--- a/tests/test-quotearg-simple.c
+++ b/tests/test-quotearg-simple.c
@@ -101,7 +101,15 @@ static struct result_groups results_g[] = {
"\"" LQ_ENC RQ_ENC "\"", "\"" LQ RQ "\"" },
{ "\"\"", "\"\\0001\\0\"", 9, "\"simple\"",
"\" \\t\\n'\\\"\\033?""?/\\\\\"", "\"a\\:b\"", "\"a\\\\b\"",
- "\"" LQ_ENC RQ_ENC "\"", "\"" LQ RQ "\"" } }
+ "\"" LQ_ENC RQ_ENC "\"", "\"" LQ RQ "\"" } },
+
+ /* java_quoting_style */
+ { { "\"\"", "\"\\0001\\0\"", 9, "\"simple\"", "\" \\t\\n'\\\"\\033?""?/\\\\\"", "\"a:b\"",
+ "\"a\\\\b\"", "\"" LQ_ENC RQ_ENC "\"", "FIXME" },
+ { "\"\"", "\"\\0001\\0\"", 9, "\"simple\"", "\" \\t\\n'\\\"\\033?""?/\\\\\"", "\"a:b\"",
+ "\"a\\\\b\"", "\"" LQ_ENC RQ_ENC "\"", "FIXME" },
+ { "\"\"", "\"\\0001\\0\"", 9, "\"simple\"", "\" \\t\\n'\\\"\\033?""?/\\\\\"", "\"a\\:b\"",
+ "\"a\\\\b\"", "\"" LQ_ENC RQ_ENC "\"", "FIXME" } }
};
static struct result_groups flag_results[] = {
@@ -242,7 +250,7 @@ main (int argc _GL_UNUSED, char *argv[])
testing for quoting of str7, use the ascii_only flag to decide
what to expect for the 8-bit data being quoted. */
ASSERT (!isprint ('\033'));
- for (i = literal_quoting_style; i <= clocale_quoting_style; i++)
+ for (i = literal_quoting_style; i <= java_quoting_style; i++)
{
set_quoting_style (NULL, (enum quoting_style) i);
compare_strings (use_quotearg_buffer, &results_g[i].group1, ascii_only);
