File: Re-add support to skip CR (\r) in File::get_as_text

This was removed in #63481, and we confirmed that it's better like this,
but we add back the possibility to strip CR as an option, to optionally
restore the previous behavior.

For performance this is done directly in `String::parse_utf8`.

Also fixes Android `FileAccess::get_line()` as this one _should_ strip CR.

Supersedes #63717.

(cherry picked from commit 1418f97c70)
This commit is contained in:
Rémi Verschelde 2022-07-31 18:46:53 +02:00
parent f9db8fa70a
commit 751b16a285
8 changed files with 26 additions and 13 deletions

View file

@ -2088,13 +2088,13 @@ PoolVector<uint8_t> _File::get_buffer(int64_t p_length) const {
return data;
}
String _File::get_as_text() const {
String _File::get_as_text(bool p_skip_cr) const {
ERR_FAIL_COND_V_MSG(!f, String(), "File must be opened before use.");
uint64_t original_pos = f->get_position();
f->seek(0);
String text = f->get_as_utf8_string();
String text = f->get_as_utf8_string(p_skip_cr);
f->seek(original_pos);
@ -2286,7 +2286,7 @@ void _File::_bind_methods() {
ClassDB::bind_method(D_METHOD("get_buffer", "len"), &_File::get_buffer);
ClassDB::bind_method(D_METHOD("get_line"), &_File::get_line);
ClassDB::bind_method(D_METHOD("get_csv_line", "delim"), &_File::get_csv_line, DEFVAL(","));
ClassDB::bind_method(D_METHOD("get_as_text"), &_File::get_as_text);
ClassDB::bind_method(D_METHOD("get_as_text", "skip_cr"), &_File::get_as_text, DEFVAL(true));
ClassDB::bind_method(D_METHOD("get_md5", "path"), &_File::get_md5);
ClassDB::bind_method(D_METHOD("get_sha256", "path"), &_File::get_sha256);
ClassDB::bind_method(D_METHOD("get_endian_swap"), &_File::get_endian_swap);

View file

@ -551,7 +551,7 @@ public:
PoolVector<uint8_t> get_buffer(int64_t p_length) const; // Get an array of bytes.
String get_line() const;
Vector<String> get_csv_line(const String &p_delim = ",") const;
String get_as_text() const;
String get_as_text(bool p_skip_cr = true) const; // Skip CR by default for compat.
String get_md5(const String &p_path) const;
String get_sha256(const String &p_path) const;

View file

@ -378,7 +378,7 @@ uint64_t FileAccess::get_buffer(uint8_t *p_dst, uint64_t p_length) const {
return i;
}
String FileAccess::get_as_utf8_string() const {
String FileAccess::get_as_utf8_string(bool p_skip_cr) const {
PoolVector<uint8_t> sourcef;
uint64_t len = get_len();
sourcef.resize(len + 1);
@ -389,7 +389,7 @@ String FileAccess::get_as_utf8_string() const {
w[len] = 0;
String s;
if (s.parse_utf8((const char *)w.ptr())) {
if (s.parse_utf8((const char *)w.ptr(), -1, p_skip_cr)) {
return String();
}
return s;

View file

@ -114,7 +114,7 @@ public:
virtual String get_line() const;
virtual String get_token() const;
virtual Vector<String> get_csv_line(const String &p_delim = ",") const;
virtual String get_as_utf8_string() const;
virtual String get_as_utf8_string(bool p_skip_cr = true) const; // Skip CR by default for compat.
/**< use this for files WRITTEN in _big_ endian machines (ie, amiga/mac)
* It's not about the current CPU type but file formats.

View file

@ -1430,7 +1430,7 @@ String String::utf8(const char *p_utf8, int p_len) {
return ret;
};
bool String::parse_utf8(const char *p_utf8, int p_len) {
bool String::parse_utf8(const char *p_utf8, int p_len, bool p_skip_cr) {
#define _UNICERROR(m_err) print_line("Unicode error: " + String(m_err));
if (!p_utf8) {
@ -1462,6 +1462,11 @@ bool String::parse_utf8(const char *p_utf8, int p_len) {
if (skip == 0) {
uint8_t c = *ptrtmp >= 0 ? *ptrtmp : uint8_t(256 + *ptrtmp);
if (p_skip_cr && c == '\r') {
ptrtmp++;
continue;
}
/* Determine the number of characters in sequence */
if ((c & 0x80) == 0) {
skip = 0;
@ -1519,6 +1524,11 @@ bool String::parse_utf8(const char *p_utf8, int p_len) {
while (cstr_size) {
int len = 0;
if (p_skip_cr && *p_utf8 == '\r') {
p_utf8++;
continue;
}
/* Determine the number of characters in sequence */
if ((*p_utf8 & 0x80) == 0) {
len = 1;

View file

@ -301,7 +301,7 @@ public:
CharString ascii(bool p_allow_extended = false) const;
CharString utf8() const;
bool parse_utf8(const char *p_utf8, int p_len = -1); //return true on error
bool parse_utf8(const char *p_utf8, int p_len = -1, bool p_skip_cr = false); //return true on error
static String utf8(const char *p_utf8, int p_len = -1);
static uint32_t hash(const CharType *p_cstr, int p_len); /* hash the string */

View file

@ -87,9 +87,10 @@
</method>
<method name="get_as_text" qualifiers="const">
<return type="String" />
<argument index="0" name="skip_cr" type="bool" default="true" />
<description>
Returns the whole file as a [String].
Text is interpreted as being UTF-8 encoded.
Returns the whole file as a [String]. Text is interpreted as being UTF-8 encoded.
If [code]skip_cr[/code] is [code]true[/code], carriage return characters ([code]\r[/code], CR) will be ignored when parsing the UTF-8, so that only line feed characters ([code]\n[/code], LF) represent a new line (Unix convention).
</description>
</method>
<method name="get_buffer" qualifiers="const">

View file

@ -29,8 +29,10 @@
/*************************************************************************/
#include "file_access_filesystem_jandroid.h"
#include "core/os/os.h"
#include "thread_jandroid.h"
#include <unistd.h>
jobject FileAccessFilesystemJAndroid::file_access_handler = nullptr;
@ -197,7 +199,7 @@ String FileAccessFilesystemJAndroid::get_line() const {
if (elem == '\n' || elem == '\0') {
// Found the end of the line
const_cast<FileAccessFilesystemJAndroid *>(this)->seek(start_position + line_buffer_position + 1);
if (result.parse_utf8((const char *)line_buffer.ptr(), line_buffer_position)) {
if (result.parse_utf8((const char *)line_buffer.ptr(), line_buffer_position, true)) {
return String();
}
return result;
@ -205,7 +207,7 @@ String FileAccessFilesystemJAndroid::get_line() const {
}
}
if (result.parse_utf8((const char *)line_buffer.ptr(), line_buffer_position)) {
if (result.parse_utf8((const char *)line_buffer.ptr(), line_buffer_position, true)) {
return String();
}
return result;