[svn:parrot] r49221 - in branches/string_macros: include/parrot src src/io src/ops src/pmc src/string src/string/encoding t/op
nwellnhof at svn.parrot.org
nwellnhof at svn.parrot.org
Wed Sep 22 01:17:14 UTC 2010
Author: nwellnhof
Date: Wed Sep 22 01:17:13 2010
New Revision: 49221
URL: https://trac.parrot.org/parrot/changeset/49221
Log:
[str] Switch to STRING_ord macro
Move the whole 'ord' logic into the string vtable functions.
Also modifies Parrot_str_indexed to accept negative indices as well. The old
string_ord function can finally be deprecated.
Modified:
branches/string_macros/include/parrot/string.h
branches/string_macros/include/parrot/string_funcs.h
branches/string_macros/src/io/filehandle.c
branches/string_macros/src/library.c
branches/string_macros/src/ops/core_ops.c
branches/string_macros/src/ops/string.ops
branches/string_macros/src/pmc.c
branches/string_macros/src/pmc/codestring.pmc
branches/string_macros/src/pmc/nci.pmc
branches/string_macros/src/pmc/string.pmc
branches/string_macros/src/spf_render.c
branches/string_macros/src/string/api.c
branches/string_macros/src/string/encoding/shared.c
branches/string_macros/src/string/encoding/shared.h
branches/string_macros/src/string/encoding/ucs2.c
branches/string_macros/src/string/encoding/ucs4.c
branches/string_macros/src/string/encoding/utf16.c
branches/string_macros/src/string/encoding/utf8.c
branches/string_macros/src/warnings.c
branches/string_macros/t/op/string.t
Modified: branches/string_macros/include/parrot/string.h
==============================================================================
--- branches/string_macros/include/parrot/string.h Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/include/parrot/string.h Wed Sep 22 01:17:13 2010 (r49221)
@@ -96,7 +96,7 @@
typedef UINTVAL (*str_vtable_validate_t)(PARROT_INTERP, ARGIN(const STRING *src));
typedef UINTVAL (*str_vtable_scan_t)(PARROT_INTERP, ARGIN(const STRING *src));
-typedef UINTVAL (*str_vtable_ord_t)(PARROT_INTERP, ARGIN(const STRING *src), UINTVAL offset);
+typedef UINTVAL (*str_vtable_ord_t)(PARROT_INTERP, ARGIN(const STRING *src), INTVAL offset);
typedef STRING * (*str_vtable_substr_t)(PARROT_INTERP, ARGIN(const STRING *src), UINTVAL offset, UINTVAL count);
/* character classes */
Modified: branches/string_macros/include/parrot/string_funcs.h
==============================================================================
--- branches/string_macros/include/parrot/string_funcs.h Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/include/parrot/string_funcs.h Wed Sep 22 01:17:13 2010 (r49221)
@@ -201,9 +201,7 @@
PARROT_EXPORT
PARROT_WARN_UNUSED_RESULT
-INTVAL Parrot_str_indexed(PARROT_INTERP,
- ARGIN(const STRING *s),
- UINTVAL idx)
+INTVAL Parrot_str_indexed(PARROT_INTERP, ARGIN(const STRING *s), INTVAL idx)
__attribute__nonnull__(1)
__attribute__nonnull__(2);
Modified: branches/string_macros/src/io/filehandle.c
==============================================================================
--- branches/string_macros/src/io/filehandle.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/io/filehandle.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -52,7 +52,7 @@
mode_len = Parrot_str_byte_length(interp, mode_str);
for (i = 0; i < mode_len; ++i) {
- const INTVAL s = Parrot_str_indexed(interp, mode_str, i);
+ const INTVAL s = STRING_ord(interp, mode_str, i);
switch (s) {
case 'r':
flags |= PIO_F_READ;
Modified: branches/string_macros/src/library.c
==============================================================================
--- branches/string_macros/src/library.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/library.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -448,8 +448,8 @@
STRING * const path_separator_string = string_chr(interp, path_separator);
/* make sure the path has a trailing slash before appending the file */
- if (Parrot_str_indexed(interp, path , path->strlen - 1)
- != Parrot_str_indexed(interp, path_separator_string, 0))
+ if (STRING_ord(interp, path, -1)
+ != STRING_ord(interp, path_separator_string, 0))
path = Parrot_str_concat(interp, path , path_separator_string);
return path;
Modified: branches/string_macros/src/ops/core_ops.c
==============================================================================
--- branches/string_macros/src/ops/core_ops.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/ops/core_ops.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -22223,42 +22223,42 @@
opcode_t *
Parrot_ord_i_s(opcode_t *cur_opcode, PARROT_INTERP) {
const Parrot_Context * const CUR_CTX = Parrot_pcc_get_context_struct(interp, interp->ctx);
- IREG(1) = string_ord(interp, SREG(2), 0);
+ IREG(1) = STRING_ord(interp, SREG(2), 0);
return (opcode_t *)cur_opcode + 3;}
opcode_t *
Parrot_ord_i_sc(opcode_t *cur_opcode, PARROT_INTERP) {
const Parrot_Context * const CUR_CTX = Parrot_pcc_get_context_struct(interp, interp->ctx);
- IREG(1) = string_ord(interp, CONST(2).u.string, 0);
+ IREG(1) = STRING_ord(interp, CONST(2).u.string, 0);
return (opcode_t *)cur_opcode + 3;}
opcode_t *
Parrot_ord_i_s_i(opcode_t *cur_opcode, PARROT_INTERP) {
const Parrot_Context * const CUR_CTX = Parrot_pcc_get_context_struct(interp, interp->ctx);
- IREG(1) = string_ord(interp, SREG(2), IREG(3));
+ IREG(1) = STRING_ord(interp, SREG(2), IREG(3));
return (opcode_t *)cur_opcode + 4;}
opcode_t *
Parrot_ord_i_sc_i(opcode_t *cur_opcode, PARROT_INTERP) {
const Parrot_Context * const CUR_CTX = Parrot_pcc_get_context_struct(interp, interp->ctx);
- IREG(1) = string_ord(interp, CONST(2).u.string, IREG(3));
+ IREG(1) = STRING_ord(interp, CONST(2).u.string, IREG(3));
return (opcode_t *)cur_opcode + 4;}
opcode_t *
Parrot_ord_i_s_ic(opcode_t *cur_opcode, PARROT_INTERP) {
const Parrot_Context * const CUR_CTX = Parrot_pcc_get_context_struct(interp, interp->ctx);
- IREG(1) = string_ord(interp, SREG(2), cur_opcode[3]);
+ IREG(1) = STRING_ord(interp, SREG(2), cur_opcode[3]);
return (opcode_t *)cur_opcode + 4;}
opcode_t *
Parrot_ord_i_sc_ic(opcode_t *cur_opcode, PARROT_INTERP) {
const Parrot_Context * const CUR_CTX = Parrot_pcc_get_context_struct(interp, interp->ctx);
- IREG(1) = string_ord(interp, CONST(2).u.string, cur_opcode[3]);
+ IREG(1) = STRING_ord(interp, CONST(2).u.string, cur_opcode[3]);
return (opcode_t *)cur_opcode + 4;}
Modified: branches/string_macros/src/ops/string.ops
==============================================================================
--- branches/string_macros/src/ops/string.ops Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/ops/string.ops Wed Sep 22 01:17:13 2010 (r49221)
@@ -40,11 +40,11 @@
=cut
inline op ord(out INT, in STR) :base_core {
- $1 = string_ord(interp, $2, 0);
+ $1 = STRING_ord(interp, $2, 0);
}
inline op ord(out INT, in STR, in INT) :base_core {
- $1 = string_ord(interp, $2, $3);
+ $1 = STRING_ord(interp, $2, $3);
}
Modified: branches/string_macros/src/pmc.c
==============================================================================
--- branches/string_macros/src/pmc.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/pmc.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -1004,14 +1004,14 @@
pos = idx;
len = Parrot_str_byte_length(interp, role);
- if (pos && (Parrot_str_indexed(interp, what, pos - 1) != 32)) {
+ if (pos && (STRING_ord(interp, what, pos - 1) != 32)) {
pos += len;
continue;
}
if (pos + len < length) {
pos += len;
- if (Parrot_str_indexed(interp, what, pos) != 32)
+ if (STRING_ord(interp, what, pos) != 32)
continue;
}
Modified: branches/string_macros/src/pmc/codestring.pmc
==============================================================================
--- branches/string_macros/src/pmc/codestring.pmc Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/pmc/codestring.pmc Wed Sep 22 01:17:13 2010 (r49221)
@@ -151,7 +151,8 @@
VTABLE_set_string_keyed_int(INTERP, parts, 1, fmt);
/* Add a newline if necessary */
- if ('\n' != Parrot_str_indexed(INTERP, fmt, Parrot_str_length(INTERP, fmt) - 1))
+ if (STRING_length(fmt) > 0
+ && '\n' != STRING_ord(INTERP, fmt, -1))
VTABLE_set_string_keyed_int(INTERP, parts, 2, newline);
S1 = Parrot_str_join(INTERP, STRINGNULL, parts);
@@ -199,8 +200,8 @@
/* treat \r\n as a single newline */
if (jpos < eos
- && string_ord(INTERP, str, jpos - 1) == 13
- && string_ord(INTERP, str, jpos) == 10) {
+ && STRING_ord(INTERP, str, jpos - 1) == 13
+ && STRING_ord(INTERP, str, jpos) == 10) {
++jpos;
}
/* search for the next newline */
Modified: branches/string_macros/src/pmc/nci.pmc
==============================================================================
--- branches/string_macros/src/pmc/nci.pmc Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/pmc/nci.pmc Wed Sep 22 01:17:13 2010 (r49221)
@@ -78,7 +78,7 @@
size_t i;
for (i = 0; i < sig_length; ++i) {
- const INTVAL c = Parrot_str_indexed(interp, sig, i);
+ const INTVAL c = STRING_ord(interp, sig, i);
PARROT_ASSERT(j < buf_length - 1);
Modified: branches/string_macros/src/pmc/string.pmc
==============================================================================
--- branches/string_macros/src/pmc/string.pmc Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/pmc/string.pmc Wed Sep 22 01:17:13 2010 (r49221)
@@ -447,7 +447,7 @@
VTABLE INTVAL get_integer_keyed_int(INTVAL pos) {
STRING *str_val;
GET_ATTR_str_val(INTERP, SELF, str_val);
- return string_ord(INTERP, str_val, pos);
+ return STRING_ord(INTERP, str_val, pos);
}
VTABLE PMC *get_pmc_keyed(PMC *key) {
Modified: branches/string_macros/src/spf_render.c
==============================================================================
--- branches/string_macros/src/spf_render.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/spf_render.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -147,12 +147,12 @@
if (is_int_type) {
if (info->flags & FLAG_PREC && info->prec == 0 &&
len == 1 &&
- string_ord(interp, str, 0) == '0') {
+ STRING_ord(interp, str, 0) == '0') {
str = Parrot_str_chopn(interp, str, len);
len = 0;
}
/* +, space */
- if (!len || string_ord(interp, str, 0) != '-') {
+ if (!len || STRING_ord(interp, str, 0) != '-') {
if (info->flags & FLAG_PLUS) {
STRING * const cs = CONST_STRING(interp, "+");
str = Parrot_str_concat(interp, cs, str);
@@ -197,8 +197,8 @@
else { /* right-align */
/* signed and zero padded */
if (info->flags & FLAG_ZERO
- && (string_ord(interp, str, 0) == '-' ||
- string_ord(interp, str, 0) == '+')) {
+ && (STRING_ord(interp, str, 0) == '-' ||
+ STRING_ord(interp, str, 0) == '+')) {
STRING *temp = NULL;
STRING *ignored;
temp = Parrot_str_substr(interp, str, 1, len-1);
@@ -429,7 +429,7 @@
char tc[PARROT_SPRINTF_BUFFER_SIZE];
for (i = 0; i < pat_len; ++i) {
- if (string_ord(interp, pat, i) == '%') { /* % */
+ if (STRING_ord(interp, pat, i) == '%') { /* % */
if (len) {
substr = Parrot_str_substr(interp, pat, old, len);
/* XXX This shouldn't modify targ the pointer */
@@ -437,7 +437,7 @@
}
len = 0;
old = i;
- if (string_ord(interp, pat, i + 1) == '%') {
+ if (STRING_ord(interp, pat, i + 1) == '%') {
/* skip this one, make next the first char
* of literal sequence, starting at old */
++i;
@@ -548,7 +548,7 @@
*/
for (++i; i < pat_len && info.phase != PHASE_DONE; ++i) {
- const INTVAL ch = string_ord(interp, pat, i);
+ const INTVAL ch = STRING_ord(interp, pat, i);
switch (info.phase) {
/*@fallthrough@ */ case PHASE_FLAGS:
Modified: branches/string_macros/src/string/api.c
==============================================================================
--- branches/string_macros/src/string/api.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/string/api.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -756,14 +756,13 @@
/*
-=item C<INTVAL Parrot_str_indexed(PARROT_INTERP, const STRING *s, UINTVAL idx)>
+=item C<INTVAL Parrot_str_indexed(PARROT_INTERP, const STRING *s, INTVAL idx)>
-Returns the character (or glyph, depending upon the string's encoding). This
-abstracts the process of finding the Nth character in a (possibly Unicode or
-JIS-encoded) string, the idea being that once the encoding functions are
-fleshed out, this function can do the right thing.
+Returns the codepoint at a given index into a string. Negative indexes are
+treated as counting from the end of the string. Throws an exception if C<s>
+is null or C<idx> is out of bounds.
-Note that this is not range-checked.
+Identical to the STRING_ord macro.
=cut
@@ -772,11 +771,15 @@
PARROT_EXPORT
PARROT_WARN_UNUSED_RESULT
INTVAL
-Parrot_str_indexed(PARROT_INTERP, ARGIN(const STRING *s), UINTVAL idx)
+Parrot_str_indexed(PARROT_INTERP, ARGIN(const STRING *s), INTVAL idx)
{
ASSERT_ARGS(Parrot_str_indexed)
- ASSERT_STRING_SANITY(s);
- return (INTVAL)STRING_ord(interp, s, idx);
+
+ if (s == NULL)
+ Parrot_ex_throw_from_c_args(interp, NULL, EXCEPTION_UNEXPECTED_NULL,
+ "Invalid operation on null string");
+
+ return STRING_ord(interp, s, idx);
}
@@ -826,26 +829,10 @@
string_ord(PARROT_INTERP, ARGIN(const STRING *s), INTVAL idx)
{
ASSERT_ARGS(string_ord)
- const UINTVAL len = STRING_length(s);
- if (idx < 0)
- idx += len;
-
- if ((UINTVAL)idx >= len) {
- const char *err_msg;
-
- if (STRING_IS_NULL(s))
- err_msg = "Cannot get character of NULL string";
- else if (!len)
- err_msg = "Cannot get character of empty string";
- else if (idx >= 0)
- err_msg = "Cannot get character past end of string";
- else if (idx < 0)
- err_msg = "Cannot get character before beginning of string";
-
- Parrot_ex_throw_from_c_args(interp, NULL, EXCEPTION_ORD_OUT_OF_STRING,
- err_msg);
- }
+ if (s == NULL)
+ Parrot_ex_throw_from_c_args(interp, NULL, EXCEPTION_UNEXPECTED_NULL,
+ "Invalid operation on null string");
return STRING_ord(interp, s, idx);
}
@@ -2847,7 +2834,7 @@
Parrot_ex_throw_from_c_args(interp, NULL, EXCEPTION_UNIMPLEMENTED,
"increment only for length = 1 done");
- o = (UINTVAL)string_ord(interp, s, 0);
+ o = (UINTVAL)STRING_ord(interp, s, 0);
if ((o >= 'A' && o < 'Z') || (o >= 'a' && o < 'z')) {
++o;
Modified: branches/string_macros/src/string/encoding/shared.c
==============================================================================
--- branches/string_macros/src/string/encoding/shared.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/string/encoding/shared.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -348,6 +348,35 @@
/*
+=item C<void encoding_ord_error(PARROT_INTERP, const STRING *s, INTVAL offset)>
+
+Throws the right exception if STRING_ord was called with a wrong index.
+C<offset> is the wrong offset into the string C<s>.
+
+=cut
+
+*/
+
+void
+encoding_ord_error(PARROT_INTERP, ARGIN(const STRING *s), INTVAL offset)
+{
+ const UINTVAL len = STRING_length(s);
+ const char *err_msg;
+
+ if (!len)
+ err_msg = "Cannot get character of empty string";
+ else if (offset >= 0)
+ err_msg = "Cannot get character past end of string";
+ else if (offset < 0)
+ err_msg = "Cannot get character before beginning of string";
+
+ Parrot_ex_throw_from_c_args(interp, NULL, EXCEPTION_ORD_OUT_OF_STRING,
+ err_msg);
+}
+
+
+/*
+
=item C<STRING * encoding_substr(PARROT_INTERP, const STRING *src, UINTVAL
offset, UINTVAL count)>
@@ -766,7 +795,7 @@
/*
-=item C<UINTVAL fixed8_ord(PARROT_INTERP, const STRING *src, UINTVAL offset)>
+=item C<UINTVAL fixed8_ord(PARROT_INTERP, const STRING *src, INTVAL idx)>
codepoints are bytes, so delegate
@@ -776,20 +805,18 @@
PARROT_WARN_UNUSED_RESULT
UINTVAL
-fixed8_ord(PARROT_INTERP, ARGIN(const STRING *src),
- UINTVAL offset)
+fixed8_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
{
ASSERT_ARGS(fixed8_ord)
- const unsigned char * const buf = (unsigned char *)src->strstart;
+ const UINTVAL len = STRING_length(src);
- if (offset >= src->bufused) {
-/* Parrot_ex_throw_from_c_args(interp, NULL, 0,
- "fixed8_ord past the end of the buffer (%i of %i)",
- offset, src->bufused); */
- return 0;
- }
+ if (idx < 0)
+ idx += len;
+
+ if ((UINTVAL)idx >= len)
+ encoding_ord_error(interp, src, idx);
- return buf[offset];
+ return (unsigned char)src->strstart[idx];
}
Modified: branches/string_macros/src/string/encoding/shared.h
==============================================================================
--- branches/string_macros/src/string/encoding/shared.h Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/string/encoding/shared.h Wed Sep 22 01:17:13 2010 (r49221)
@@ -85,6 +85,12 @@
__attribute__nonnull__(1)
__attribute__nonnull__(3);
+void encoding_ord_error(PARROT_INTERP,
+ ARGIN(const STRING *s),
+ INTVAL offset)
+ __attribute__nonnull__(1)
+ __attribute__nonnull__(2);
+
PARROT_WARN_UNUSED_RESULT
INTVAL encoding_rindex(PARROT_INTERP,
SHIM(const STRING *src),
@@ -181,7 +187,7 @@
FUNC_MODIFIES(*iter);
PARROT_WARN_UNUSED_RESULT
-UINTVAL fixed8_ord(PARROT_INTERP, ARGIN(const STRING *src), UINTVAL offset)
+UINTVAL fixed8_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
__attribute__nonnull__(1)
__attribute__nonnull__(2);
@@ -277,6 +283,9 @@
#define ASSERT_ARGS_encoding_is_cclass __attribute__unused__ int _ASSERT_ARGS_CHECK = (\
PARROT_ASSERT_ARG(interp) \
, PARROT_ASSERT_ARG(src))
+#define ASSERT_ARGS_encoding_ord_error __attribute__unused__ int _ASSERT_ARGS_CHECK = (\
+ PARROT_ASSERT_ARG(interp) \
+ , PARROT_ASSERT_ARG(s))
#define ASSERT_ARGS_encoding_rindex __attribute__unused__ int _ASSERT_ARGS_CHECK = (\
PARROT_ASSERT_ARG(interp))
#define ASSERT_ARGS_encoding_scan __attribute__unused__ int _ASSERT_ARGS_CHECK = (\
Modified: branches/string_macros/src/string/encoding/ucs2.c
==============================================================================
--- branches/string_macros/src/string/encoding/ucs2.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/string/encoding/ucs2.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -86,9 +86,7 @@
__attribute__nonnull__(3)
FUNC_MODIFIES(*i);
-static UINTVAL ucs2_ord(PARROT_INTERP,
- ARGIN(const STRING *src),
- UINTVAL offset)
+static UINTVAL ucs2_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
__attribute__nonnull__(1)
__attribute__nonnull__(2);
@@ -212,8 +210,7 @@
/*
-=item C<static UINTVAL ucs2_ord(PARROT_INTERP, const STRING *src, UINTVAL
-offset)>
+=item C<static UINTVAL ucs2_ord(PARROT_INTERP, const STRING *src, INTVAL idx)>
Returns the codepoint in string C<src> at position C<offset>.
@@ -222,15 +219,24 @@
*/
static UINTVAL
-ucs2_ord(PARROT_INTERP, ARGIN(const STRING *src), UINTVAL offset)
+ucs2_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
{
ASSERT_ARGS(ucs2_ord)
#if PARROT_HAS_ICU
- const UChar * const s = (const UChar*) src->strstart;
- UNUSED(interp);
- return s[offset];
+ const UINTVAL len = STRING_length(src);
+ const UChar *s;
+
+ if (idx < 0)
+ idx += len;
+
+ if ((UINTVAL)idx >= len)
+ encoding_ord_error(interp, src, idx);
+
+ s = (const UChar *)src->strstart;
+
+ return s[idx];
#else
- UNUSED(offset);
+ UNUSED(idx);
UNUSED(src);
no_ICU_lib(interp);
#endif
Modified: branches/string_macros/src/string/encoding/ucs4.c
==============================================================================
--- branches/string_macros/src/string/encoding/ucs4.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/string/encoding/ucs4.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -86,9 +86,7 @@
__attribute__nonnull__(3)
FUNC_MODIFIES(*i);
-static UINTVAL ucs4_ord(PARROT_INTERP,
- ARGIN(const STRING *src),
- UINTVAL offset)
+static UINTVAL ucs4_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
__attribute__nonnull__(1)
__attribute__nonnull__(2);
@@ -225,8 +223,7 @@
/*
-=item C<static UINTVAL ucs4_ord(PARROT_INTERP, const STRING *src, UINTVAL
-offset)>
+=item C<static UINTVAL ucs4_ord(PARROT_INTERP, const STRING *src, INTVAL idx)>
Returns the codepoint in string C<src> at position C<offset>.
@@ -235,15 +232,24 @@
*/
static UINTVAL
-ucs4_ord(PARROT_INTERP, ARGIN(const STRING *src), UINTVAL offset)
+ucs4_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
{
ASSERT_ARGS(ucs4_ord)
#if PARROT_HAS_ICU
- const UChar32 * const s = (const UChar32*) src->strstart;
- UNUSED(interp);
- return s[offset];
+ const UINTVAL len = STRING_length(src);
+ const UChar32 *s;
+
+ if (idx < 0)
+ idx += len;
+
+ if ((UINTVAL)idx >= len)
+ encoding_ord_error(interp, src, idx);
+
+ s = (const UChar32 *)src->strstart;
+
+ return s[idx];
#else
- UNUSED(offset);
+ UNUSED(idx);
UNUSED(src);
no_ICU_lib(interp);
#endif
Modified: branches/string_macros/src/string/encoding/utf16.c
==============================================================================
--- branches/string_macros/src/string/encoding/utf16.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/string/encoding/utf16.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -73,7 +73,7 @@
static UINTVAL utf16_ord(PARROT_INTERP,
ARGIN(const STRING *src),
- UINTVAL offset)
+ INTVAL idx)
__attribute__nonnull__(1)
__attribute__nonnull__(2);
@@ -264,8 +264,7 @@
/*
-=item C<static UINTVAL utf16_ord(PARROT_INTERP, const STRING *src, UINTVAL
-offset)>
+=item C<static UINTVAL utf16_ord(PARROT_INTERP, const STRING *src, INTVAL idx)>
Returns the codepoint in string C<src> at position C<offset>.
@@ -274,21 +273,29 @@
*/
static UINTVAL
-utf16_ord(PARROT_INTERP, ARGIN(const STRING *src), UINTVAL offset)
+utf16_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
{
ASSERT_ARGS(utf16_ord)
#if PARROT_HAS_ICU
- const UChar * const s = (UChar*) src->strstart;
- UINTVAL c, pos;
- UNUSED(interp);
+ const UINTVAL len = STRING_length(src);
+ const UChar *s;
+ UINTVAL c, pos;
+
+ if (idx < 0)
+ idx += len;
+
+ if ((UINTVAL)idx >= len)
+ encoding_ord_error(interp, src, idx);
+ s = (UChar *)src->strstart;
pos = 0;
- U16_FWD_N_UNSAFE(s, pos, offset);
+ U16_FWD_N_UNSAFE(s, pos, idx);
U16_GET_UNSAFE(s, pos, c);
+
return c;
#else
UNUSED(src);
- UNUSED(offset);
+ UNUSED(idx);
Parrot_ex_throw_from_c_args(interp, NULL, EXCEPTION_LIBRARY_ERROR,
"no ICU lib loaded");
Modified: branches/string_macros/src/string/encoding/utf8.c
==============================================================================
--- branches/string_macros/src/string/encoding/utf8.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/string/encoding/utf8.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -78,9 +78,7 @@
__attribute__nonnull__(3)
FUNC_MODIFIES(*i);
-static UINTVAL utf8_ord(PARROT_INTERP,
- ARGIN(const STRING *src),
- UINTVAL offset)
+static UINTVAL utf8_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
__attribute__nonnull__(1)
__attribute__nonnull__(2);
@@ -94,11 +92,11 @@
PARROT_WARN_UNUSED_RESULT
PARROT_CANNOT_RETURN_NULL
-static const void * utf8_skip_backward(ARGIN(const void *ptr), UINTVAL n)
+static const utf8_t * utf8_skip_backward(ARGIN(const void *ptr), UINTVAL n)
__attribute__nonnull__(1);
PARROT_CANNOT_RETURN_NULL
-static const void * utf8_skip_forward(ARGIN(const void *ptr), UINTVAL n)
+static const utf8_t * utf8_skip_forward(ARGIN(const void *ptr), UINTVAL n)
__attribute__nonnull__(1);
PARROT_CAN_RETURN_NULL
@@ -304,8 +302,7 @@
/*
-=item C<static UINTVAL utf8_ord(PARROT_INTERP, const STRING *src, UINTVAL
-offset)>
+=item C<static UINTVAL utf8_ord(PARROT_INTERP, const STRING *src, INTVAL idx)>
Returns the codepoint in string C<src> at position C<offset>.
@@ -314,10 +311,20 @@
*/
static UINTVAL
-utf8_ord(PARROT_INTERP, ARGIN(const STRING *src), UINTVAL offset)
+utf8_ord(PARROT_INTERP, ARGIN(const STRING *src), INTVAL idx)
{
ASSERT_ARGS(utf8_ord)
- const utf8_t * const start = (const utf8_t *)utf8_skip_forward(src->strstart, offset);
+ const UINTVAL len = STRING_length(src);
+ const utf8_t *start;
+
+ if (idx < 0)
+ idx += len;
+
+ if ((UINTVAL)idx >= len)
+ encoding_ord_error(interp, src, idx);
+
+ start = utf8_skip_forward(src->strstart, idx);
+
return utf8_decode(interp, start);
}
@@ -407,7 +414,7 @@
/*
-=item C<static const void * utf8_skip_forward(const void *ptr, UINTVAL n)>
+=item C<static const utf8_t * utf8_skip_forward(const void *ptr, UINTVAL n)>
Moves C<ptr> C<n> characters forward.
@@ -416,7 +423,7 @@
*/
PARROT_CANNOT_RETURN_NULL
-static const void *
+static const utf8_t *
utf8_skip_forward(ARGIN(const void *ptr), UINTVAL n)
{
ASSERT_ARGS(utf8_skip_forward)
@@ -432,7 +439,7 @@
/*
-=item C<static const void * utf8_skip_backward(const void *ptr, UINTVAL n)>
+=item C<static const utf8_t * utf8_skip_backward(const void *ptr, UINTVAL n)>
Moves C<ptr> C<n> characters back.
@@ -444,7 +451,7 @@
PARROT_WARN_UNUSED_RESULT
PARROT_CANNOT_RETURN_NULL
-static const void *
+static const utf8_t *
utf8_skip_backward(ARGIN(const void *ptr), UINTVAL n)
{
ASSERT_ARGS(utf8_skip_backward)
@@ -479,10 +486,10 @@
const utf8_t *u8ptr = (utf8_t *)((char *)str->strstart + i->bytepos);
if (offset > 0) {
- u8ptr = (const utf8_t *)utf8_skip_forward(u8ptr, offset);
+ u8ptr = utf8_skip_forward(u8ptr, offset);
}
else if (offset < 0) {
- u8ptr = (const utf8_t *)utf8_skip_backward(u8ptr, -offset);
+ u8ptr = utf8_skip_backward(u8ptr, -offset);
}
return utf8_decode(interp, u8ptr);
@@ -508,10 +515,10 @@
const utf8_t *u8ptr = (utf8_t *)((char *)str->strstart + i->bytepos);
if (skip > 0) {
- u8ptr = (const utf8_t *)utf8_skip_forward(u8ptr, skip);
+ u8ptr = utf8_skip_forward(u8ptr, skip);
}
else if (skip < 0) {
- u8ptr = (const utf8_t *)utf8_skip_backward(u8ptr, -skip);
+ u8ptr = utf8_skip_backward(u8ptr, -skip);
}
i->charpos += skip;
@@ -629,22 +636,22 @@
if (pos < i->charpos) {
if (pos <= (i->charpos >> 1)) {
/* go forward from start */
- u8ptr = (const utf8_t *)utf8_skip_forward(u8ptr, pos);
+ u8ptr = utf8_skip_forward(u8ptr, pos);
}
else {
/* go backward from current */
- u8ptr = (const utf8_t *)utf8_skip_backward(u8ptr + i->bytepos, i->charpos - pos);
+ u8ptr = utf8_skip_backward(u8ptr + i->bytepos, i->charpos - pos);
}
}
else {
const UINTVAL len = str->strlen;
if (pos <= i->charpos + ((len - i->charpos) >> 1)) {
/* go forward from current */
- u8ptr = (const utf8_t *)utf8_skip_forward(u8ptr + i->bytepos, pos - i->charpos);
+ u8ptr = utf8_skip_forward(u8ptr + i->bytepos, pos - i->charpos);
}
else {
/* go backward from end */
- u8ptr = (const utf8_t *)utf8_skip_backward(u8ptr + str->bufused, len - pos);
+ u8ptr = utf8_skip_backward(u8ptr + str->bufused, len - pos);
}
}
Modified: branches/string_macros/src/warnings.c
==============================================================================
--- branches/string_macros/src/warnings.c Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/src/warnings.c Wed Sep 22 01:17:13 2010 (r49221)
@@ -74,7 +74,7 @@
Parrot_io_puts(interp, Parrot_io_STDERR(interp), "Unknown warning\n");
else {
Parrot_io_putps(interp, Parrot_io_STDERR(interp), msg);
- if (string_ord(interp, msg, -1) != '\n')
+ if (STRING_ord(interp, msg, -1) != '\n')
Parrot_io_eprintf(interp, "%c", '\n');
}
print_pbc_location(interp);
Modified: branches/string_macros/t/op/string.t
==============================================================================
--- branches/string_macros/t/op/string.t Wed Sep 22 01:16:25 2010 (r49220)
+++ branches/string_macros/t/op/string.t Wed Sep 22 01:17:13 2010 (r49221)
@@ -650,7 +650,7 @@
ord $I0,$S0
ok( 0, 'no exception: 2-param ord, empty string register' )
handler:
- .exception_is( 'Cannot get character of NULL string' )
+ .exception_is( 'Invalid operation on null string' )
.end
.sub exception_three_param_ord_empty_string
@@ -666,7 +666,7 @@
ord $I0,$S0,0
ok( 0, 'no exception: 3-param ord, empty string register' )
handler:
- .exception_is( 'Cannot get character of NULL string' )
+ .exception_is( 'Invalid operation on null string' )
.end
.sub two_param_ord_one_character_string
More information about the parrot-commits
mailing list