lang
dir.lang_add.txt
and run lang-add.shlang-export.sh
lang-import.sh
There are 2 modes of operation. If LANG_MODE==0
, only one language is being used (the default compilation approach from plain Arduino IDE).
The reset of this explanation is devoted to LANG_MODE==1
:
language.h
:
// section .loc_sec (originaly .progmem0) will be used for localized translated strings
#define PROGMEM_I2 __attribute__((section(".loc_sec")))
// section .loc_pri (originaly .progmem1) will be used for localized strings in english
#define PROGMEM_I1 __attribute__((section(".loc_pri")))
// section .noloc (originaly progmem2) will be used for not localized strings in english
#define PROGMEM_N1 __attribute__((section(".noloc")))
#define _I(s) (__extension__({static const char __c[] PROGMEM_I1 = "\xff\xff" s; &__c[0];}))
#define ISTR(s) "\xff\xff" s
#define _i(s) lang_get_translation(_I(s))
#define _T(s) lang_get_translation(s)
That explains the macros:
_i
expands into lang_get_translation((__extension__({static const char __c[] PROGMEM_I1 = "\xff\xff" s; &__c[0];})))
. Note the two 0xff's in the beginning of the string._T
expands into lang_get_translation(s)
without the two 0xff's at the beginning.The two 0xff's are somehow magically replaced by real string ID's where the translations are available (still don't know where).
const char* lang_get_translation(const char* s){
if (lang_selected == 0) return s + 2; //primary language selected, return orig. str.
if (lang_table == 0) return s + 2; //sec. lang table not found, return orig. str.
uint16_t ui = pgm_read_word(((uint16_t*)s)); //read string id
if (ui == 0xffff) return s + 2; //translation not found, return orig. str.
ui = pgm_read_word(((uint16_t*)(((char*)lang_table + 16 + ui*2)))); //read relative offset
if (pgm_read_byte(((uint8_t*)((char*)lang_table + ui))) == 0) //read first character
return s + 2;//zero length string == not translated, return orig. str.
return (const char*)((char*)lang_table + ui); //return calculated pointer
}
lang_en.txt
#MSG_CRASH_DET_ONLY_IN_NORMAL c=20 r=4
"Crash detection can\x0abe turned on only in\x0aNormal mode"
lang_en_*.txt
#MSG_CRASH_DET_ONLY_IN_NORMAL c=20 r=4
"Crash detection can\x0abe turned on only in\x0aNormal mode"
"Crash detekce muze\x0abyt zapnuta pouze v\x0aNormal modu"
not_tran.txt
A simple list of strings that are not translated yet.
not_used.txt
A list os strings not currently used in the firmware. May be removed from lang_en_*.txt
files by hand.
config.sh
ARDUINO
. If not found/empty, a default C:/arduino-1.8.5
is used.CONFIG_OK=1
when all good, otherwise sets CONFIG_OK=0
fw-build.sh
Joins firmware HEX and language binaries into one file.
fw-clean.sh
lang-add.sh
Adds new messages into the dictionary regardless of whether there have been any older versions.
lang-build.sh
Generates lang_xx.bin (language binary files) for the whole firmware build.
Arguments:
$1
: language code (en
, cz
, de
, es
, fr
, it
, pl
) or all
all
Input: lang_en.txt
or lang_en_xx.txt
Output: lang_xx.bin
Temporary files: lang_xx.tmp
and lang_xx.dat
Description of the process:
The script first runs lang-check.py $1
and removes empty lines and comments (and non-translated texts) into lang_$1.tmp
.
The tmp file now contains all translated texts (some of them empty, i.e. "").
The tmp file is then transformed into lang_$1.dat
, which is a simple dump of all texts together, each terminated with a \x00
.
Format of the bin
file:
A5 5A
B4 4B
en
converted into ne
, i.e. characters swapped. Only cz
is changed into sc
(old cs
ISO code).\x00
The signature is composed of 2B number of strings and 2B checksum in lang_en.bin. Signature in lang_en.bin is zero.
lang-check.sh
and lang-check.py
Both do the same, only lang-check.py is newer, i.e. lang-check.sh is not used anymore. lang-check.py makes a binary comparison between what's in the dictionary and what's in the binary.
lang-clean.sh
Removes all language output files from lang folder. That means deleting:
lang-export.sh
Exports PO (gettext) for external translators.
lang-import.sh
Import from PO
progmem.sh
Examine content of progmem sections (default is progmem1).
Input:
Outputs:
Description of process:
\x0a
textaddr.sh
Compiles progmem1.var
and lang_en.txt
files to textaddr.txt
file (mapping of progmem addreses to text idenifiers).
Description of process:
Input:
Output:
update_lang.sh