summaryrefslogtreecommitdiff
path: root/transcode.c
diff options
context:
space:
mode:
Diffstat (limited to 'transcode.c')
-rw-r--r--transcode.c130
1 files changed, 63 insertions, 67 deletions
diff --git a/transcode.c b/transcode.c
index a72afdc44b..84c579eb2e 100644
--- a/transcode.c
+++ b/transcode.c
@@ -9,31 +9,20 @@
**********************************************************************/
-#include "ruby/internal/config.h"
-
-#include <ctype.h>
-
#include "internal.h"
-#include "internal/array.h"
-#include "internal/inits.h"
-#include "internal/object.h"
-#include "internal/string.h"
-#include "internal/transcode.h"
-#include "ruby/encoding.h"
-
#include "transcode_data.h"
-#include "id.h"
+#include <ctype.h>
#define ENABLE_ECONV_NEWLINE_OPTION 1
/* VALUE rb_cEncoding = rb_define_class("Encoding", rb_cObject); */
-static VALUE rb_eUndefinedConversionError;
-static VALUE rb_eInvalidByteSequenceError;
-static VALUE rb_eConverterNotFoundError;
+VALUE rb_eUndefinedConversionError;
+VALUE rb_eInvalidByteSequenceError;
+VALUE rb_eConverterNotFoundError;
VALUE rb_cEncodingConverter;
-static VALUE sym_invalid, sym_undef, sym_replace, sym_fallback;
+static VALUE sym_invalid, sym_undef, sym_replace, sym_fallback, sym_aref;
static VALUE sym_xml, sym_text, sym_attr;
static VALUE sym_universal_newline;
static VALUE sym_crlf_newline;
@@ -379,12 +368,14 @@ load_transcoder_entry(transcoder_entry_t *entry)
const size_t total_len = sizeof(transcoder_lib_prefix) - 1 + len;
const VALUE fn = rb_str_new(0, total_len);
char *const path = RSTRING_PTR(fn);
+ const int safe = rb_safe_level();
memcpy(path, transcoder_lib_prefix, sizeof(transcoder_lib_prefix) - 1);
memcpy(path + sizeof(transcoder_lib_prefix) - 1, lib, len);
rb_str_set_len(fn, total_len);
+ FL_UNSET(fn, FL_TAINT);
OBJ_FREEZE(fn);
- rb_require_string(fn);
+ rb_require_safe(fn, safe > 3 ? 3 : safe);
}
if (entry->transcoder)
@@ -983,10 +974,21 @@ rb_econv_open0(const char *sname, const char *dname, int ecflags)
int num_trans;
rb_econv_t *ec;
- /* Just check if sname and dname are defined */
- /* (This check is needed?) */
- if (*sname) rb_enc_find_index(sname);
- if (*dname) rb_enc_find_index(dname);
+ int sidx, didx;
+
+ if (*sname) {
+ sidx = rb_enc_find_index(sname);
+ if (0 <= sidx) {
+ rb_enc_from_index(sidx);
+ }
+ }
+
+ if (*dname) {
+ didx = rb_enc_find_index(dname);
+ if (0 <= didx) {
+ rb_enc_from_index(didx);
+ }
+ }
if (*sname == '\0' && *dname == '\0') {
num_trans = 0;
@@ -1192,6 +1194,7 @@ rb_trans_conv(rb_econv_t *ec,
if (ec->elems[0].last_result == econv_after_output)
ec->elems[0].last_result = econv_source_buffer_empty;
+ needreport_index = -1;
for (i = ec->num_trans-1; 0 <= i; i--) {
switch (ec->elems[i].last_result) {
case econv_invalid_byte_sequence:
@@ -1200,6 +1203,7 @@ rb_trans_conv(rb_econv_t *ec,
case econv_after_output:
case econv_finished:
sweep_start = i+1;
+ needreport_index = i;
goto found_needreport;
case econv_destination_buffer_full:
@@ -1850,6 +1854,7 @@ rb_econv_substr_append(rb_econv_t *ec, VALUE src, long off, long len, VALUE dst,
src = rb_str_new_frozen(src);
dst = rb_econv_append(ec, RSTRING_PTR(src) + off, len, dst, flags);
RB_GC_GUARD(src);
+ OBJ_INFECT_RAW(dst, src);
return dst;
}
@@ -2048,6 +2053,7 @@ make_econv_exception(rb_econv_t *ec)
size_t readagain_len = ec->last_error.readagain_len;
VALUE bytes2 = Qnil;
VALUE dumped2;
+ int idx;
if (ec->last_error.result == econv_incomplete_input) {
mesg = rb_sprintf("incomplete %s on %s",
StringValueCStr(dumped),
@@ -2071,7 +2077,17 @@ make_econv_exception(rb_econv_t *ec)
rb_ivar_set(exc, rb_intern("error_bytes"), bytes);
rb_ivar_set(exc, rb_intern("readagain_bytes"), bytes2);
rb_ivar_set(exc, rb_intern("incomplete_input"), ec->last_error.result == econv_incomplete_input ? Qtrue : Qfalse);
- goto set_encs;
+
+ set_encs:
+ rb_ivar_set(exc, rb_intern("source_encoding_name"), rb_str_new2(ec->last_error.source_encoding));
+ rb_ivar_set(exc, rb_intern("destination_encoding_name"), rb_str_new2(ec->last_error.destination_encoding));
+ idx = rb_enc_find_index(ec->last_error.source_encoding);
+ if (0 <= idx)
+ rb_ivar_set(exc, rb_intern("source_encoding"), rb_enc_from_encoding(rb_enc_from_index(idx)));
+ idx = rb_enc_find_index(ec->last_error.destination_encoding);
+ if (0 <= idx)
+ rb_ivar_set(exc, rb_intern("destination_encoding"), rb_enc_from_encoding(rb_enc_from_index(idx)));
+ return exc;
}
if (ec->last_error.result == econv_undefined_conversion) {
VALUE bytes = rb_str_new((const char *)ec->last_error.error_bytes_start,
@@ -2123,17 +2139,6 @@ make_econv_exception(rb_econv_t *ec)
goto set_encs;
}
return Qnil;
-
- set_encs:
- rb_ivar_set(exc, rb_intern("source_encoding_name"), rb_str_new2(ec->last_error.source_encoding));
- rb_ivar_set(exc, rb_intern("destination_encoding_name"), rb_str_new2(ec->last_error.destination_encoding));
- int idx = rb_enc_find_index(ec->last_error.source_encoding);
- if (0 <= idx)
- rb_ivar_set(exc, rb_intern("source_encoding"), rb_enc_from_encoding(rb_enc_from_index(idx)));
- idx = rb_enc_find_index(ec->last_error.destination_encoding);
- if (0 <= idx)
- rb_ivar_set(exc, rb_intern("destination_encoding"), rb_enc_from_encoding(rb_enc_from_index(idx)));
- return exc;
}
static void
@@ -2251,7 +2256,7 @@ method_fallback(VALUE fallback, VALUE c)
static VALUE
aref_fallback(VALUE fallback, VALUE c)
{
- return rb_funcallv_public(fallback, idAREF, 1, &c);
+ return rb_funcall3(fallback, sym_aref, 1, &c);
}
static void
@@ -2420,7 +2425,6 @@ static int
econv_opts(VALUE opt, int ecflags)
{
VALUE v;
- int newlineflag = 0;
v = rb_hash_aref(opt, sym_invalid);
if (NIL_P(v)) {
@@ -2466,7 +2470,6 @@ econv_opts(VALUE opt, int ecflags)
#ifdef ENABLE_ECONV_NEWLINE_OPTION
v = rb_hash_aref(opt, sym_newline);
if (!NIL_P(v)) {
- newlineflag = 2;
ecflags &= ~ECONV_NEWLINE_DECORATOR_MASK;
if (v == sym_universal) {
ecflags |= ECONV_UNIVERSAL_NEWLINE_DECORATOR;
@@ -2488,9 +2491,10 @@ econv_opts(VALUE opt, int ecflags)
rb_raise(rb_eArgError, "unexpected value for newline option");
}
}
+ else
#endif
{
- int setflags = 0;
+ int setflags = 0, newlineflag = 0;
v = rb_hash_aref(opt, sym_universal_newline);
if (RTEST(v))
@@ -2507,15 +2511,9 @@ econv_opts(VALUE opt, int ecflags)
setflags |= ECONV_CR_NEWLINE_DECORATOR;
newlineflag |= !NIL_P(v);
- switch (newlineflag) {
- case 1:
+ if (newlineflag) {
ecflags &= ~ECONV_NEWLINE_DECORATOR_MASK;
ecflags |= setflags;
- break;
-
- case 3:
- rb_warning(":newline option preceds other newline options");
- break;
}
}
@@ -2552,7 +2550,7 @@ rb_econv_prepare_options(VALUE opthash, VALUE *opts, int ecflags)
if (!NIL_P(v)) {
VALUE h = rb_check_hash_type(v);
if (NIL_P(h)
- ? (rb_obj_is_proc(v) || rb_obj_is_method(v) || rb_respond_to(v, idAREF))
+ ? (rb_obj_is_proc(v) || rb_obj_is_method(v) || rb_respond_to(v, sym_aref))
: (v = h, 1)) {
if (NIL_P(newhash))
newhash = rb_hash_new();
@@ -2784,14 +2782,14 @@ str_encode_associate(VALUE str, int encidx)
/*
* call-seq:
- * str.encode!(encoding, **options) -> str
- * str.encode!(dst_encoding, src_encoding, **options) -> str
+ * str.encode!(encoding [, options] ) -> str
+ * str.encode!(dst_encoding, src_encoding [, options] ) -> str
*
* The first form transcodes the contents of <i>str</i> from
* str.encoding to +encoding+.
* The second form transcodes the contents of <i>str</i> from
* src_encoding to dst_encoding.
- * The +options+ keyword arguments give details for conversion. See String#encode
+ * The options Hash gives details for conversion. See String#encode
* for details.
* Returns the string even if no changes were made.
*/
@@ -2820,9 +2818,9 @@ static VALUE encoded_dup(VALUE newstr, VALUE str, int encidx);
/*
* call-seq:
- * str.encode(encoding, **options) -> str
- * str.encode(dst_encoding, src_encoding, **options) -> str
- * str.encode(**options) -> str
+ * str.encode(encoding [, options] ) -> str
+ * str.encode(dst_encoding, src_encoding [, options] ) -> str
+ * str.encode([options]) -> str
*
* The first form returns a copy of +str+ transcoded
* to encoding +encoding+.
@@ -2838,8 +2836,8 @@ static VALUE encoded_dup(VALUE newstr, VALUE str, int encidx);
* in the source encoding. The last form by default does not raise
* exceptions but uses replacement strings.
*
- * The +options+ keyword arguments give details for conversion.
- * The arguments are:
+ * The +options+ Hash gives details for conversion and can have the following
+ * keys:
*
* :invalid ::
* If the value is +:replace+, #encode replaces invalid byte sequences in
@@ -2907,11 +2905,6 @@ encoded_dup(VALUE newstr, VALUE str, int encidx)
return str_encode_associate(newstr, encidx);
}
-/*
- * Document-class: Encoding::Converter
- *
- * Encoding conversion class.
- */
static void
econv_free(void *ptr)
{
@@ -2927,7 +2920,7 @@ econv_memsize(const void *ptr)
static const rb_data_type_t econv_data_type = {
"econv",
- {0, econv_free, econv_memsize,},
+ {NULL, econv_free, econv_memsize,},
0, 0, RUBY_TYPED_FREE_IMMEDIATELY
};
@@ -3158,10 +3151,10 @@ econv_s_search_convpath(int argc, VALUE *argv, VALUE klass)
transcode_search_path(sname, dname, search_convpath_i, &convpath);
if (NIL_P(convpath)) {
- VALUE exc = rb_econv_open_exc(sname, dname, ecflags);
- RB_GC_GUARD(snamev);
- RB_GC_GUARD(dnamev);
- rb_exc_raise(exc);
+ VALUE exc = rb_econv_open_exc(sname, dname, ecflags);
+ RB_GC_GUARD(snamev);
+ RB_GC_GUARD(dnamev);
+ rb_exc_raise(exc);
}
if (decorate_convpath(convpath, ecflags) == -1) {
@@ -3795,6 +3788,7 @@ econv_primitive_convert(int argc, VALUE *argv, VALUE self)
res = rb_econv_convert(ec, &ip, is, &op, os, flags);
rb_str_set_len(output, op-(unsigned char *)RSTRING_PTR(output));
if (!NIL_P(input)) {
+ OBJ_INFECT_RAW(output, input);
rb_str_drop_bytes(input, ip - (unsigned char *)RSTRING_PTR(input));
}
@@ -4085,7 +4079,7 @@ econv_insert_output(VALUE self, VALUE string)
}
/*
- * call-seq:
+ * call-seq
* ec.putback -> string
* ec.putback(max_numbytes) -> string
*
@@ -4116,9 +4110,10 @@ econv_putback(int argc, VALUE *argv, VALUE self)
int putbackable;
VALUE str, max;
- if (!rb_check_arity(argc, 0, 1) || NIL_P(max = argv[0])) {
+ rb_scan_args(argc, argv, "01", &max);
+
+ if (NIL_P(max))
n = rb_econv_putbackable(ec);
- }
else {
n = NUM2INT(max);
putbackable = rb_econv_putbackable(ec);
@@ -4425,6 +4420,7 @@ Init_transcode(void)
sym_undef = ID2SYM(rb_intern("undef"));
sym_replace = ID2SYM(rb_intern("replace"));
sym_fallback = ID2SYM(rb_intern("fallback"));
+ sym_aref = ID2SYM(rb_intern("[]"));
sym_xml = ID2SYM(rb_intern("xml"));
sym_text = ID2SYM(rb_intern("text"));
sym_attr = ID2SYM(rb_intern("attr"));
@@ -4462,7 +4458,7 @@ InitVM_transcode(void)
rb_define_method(rb_cString, "encode", str_encode, -1);
rb_define_method(rb_cString, "encode!", str_encode_bang, -1);
- rb_cEncodingConverter = rb_define_class_under(rb_cEncoding, "Converter", rb_cObject);
+ rb_cEncodingConverter = rb_define_class_under(rb_cEncoding, "Converter", rb_cData);
rb_define_alloc_func(rb_cEncodingConverter, econv_s_allocate);
rb_define_singleton_method(rb_cEncodingConverter, "asciicompat_encoding", econv_s_asciicompat_encoding, 1);
rb_define_singleton_method(rb_cEncodingConverter, "search_convpath", econv_s_search_convpath, -1);
generated by cgit v1.2.3 (git 2.25.1) at 2026-06-06 19:28:58 +0000