On 10.09.18 21:49, Heinrich Schuchardt wrote:
> Up to now the EFI_TEXT_INPUT_PROTOCOL only supported ASCII characters.
> With the patch it can consume UTF-8 from the console.
>
> Currently only the serial console and the console can deliver UTF-8.
> Local consoles are restricted to ASCII.
>
> Signed-off-by: Heinrich Schuchardt <xypron.g...@gmx.de>
> ---
> v2:
> drop support for German keyboard
> move reading of Unicode code to charset.c
> ---
> include/charset.h | 10 +++
> lib/charset.c | 137 +++++++++++++++++++++++------------
> lib/efi_loader/efi_console.c | 11 ++-
> test/unicode_ut.c | 8 +-
> 4 files changed, 108 insertions(+), 58 deletions(-)
>
> diff --git a/include/charset.h b/include/charset.h
> index 686db5a1fe..d1dc326c9b 100644
> --- a/include/charset.h
> +++ b/include/charset.h
> @@ -8,11 +8,21 @@
> #ifndef __CHARSET_H_
> #define __CHARSET_H_
>
> +#include <efi.h>
> #include <linux/kernel.h>
> #include <linux/types.h>
>
> #define MAX_UTF8_PER_UTF16 3
>
> +/**
> + * console_read_unicode() - read Unicode code point from console
> + *
> + * @code: code point
> + * Return: status code:
> + * EFI_SUCCESS, EFI_NOT_READY, or EFI_INVALID_PARAMETER
> + */
> +efi_status_t console_read_unicode(s32 *code);
> +
> /**
> * utf8_get() - get next UTF-8 code point from buffer
> *
> diff --git a/lib/charset.c b/lib/charset.c
> index 72c808ce64..2e582a51f7 100644
> --- a/lib/charset.c
> +++ b/lib/charset.c
> @@ -5,6 +5,7 @@
> * Copyright (c) 2017 Rob Clark
> */
>
> +#include <common.h>
> #include <charset.h>
> #include <capitalization.h>
> #include <malloc.h>
> @@ -18,67 +19,107 @@ static struct capitalization_table
> capitalization_table[] =
> CP437_CAPITALIZATION_TABLE;
> #endif
>
> -s32 utf8_get(const char **src)
> +/**
> + * get_code() - read Unicode code point from UTF-8 stream
> + *
> + * @next: - stream reader
> + * @src: - string buffer passed to stream reader, optional
> + * Return: - Unicode code point
> + */
> +static int get_code(s32 (*next)(const char **src), const char **src)
The stream reader should probably return a u8. Also, I think it makes
sense to declare the argument as opaque so that a real stream reader can
implement reads from a non-string later.
With that I think it should look like this:
static int get_code(u8 (*next)(void *opaque), void *opaque)
> {
> - s32 code = 0;
> - unsigned char c;
> + s32 ch = 0;
>
> - if (!src || !*src)
> - return -1;
> - if (!**src)
> + ch = next(src);
> + if (!ch)
> return 0;
> - c = **src;
> - if (c >= 0x80) {
> - ++*src;
> - if (!**src)
> - return -1;
> - /*
> - * We do not expect a continuation byte (0x80 - 0xbf).
> - * 0x80 is coded as 0xc2 0x80, so we cannot have less then 0xc2
> - * here.
> - * The highest code point is 0x10ffff which is coded as
> - * 0xf4 0x8f 0xbf 0xbf. So we cannot have a byte above 0xf4.
> - */
> - if (c < 0xc2 || code > 0xf4)
> - return -1;
> - if (c >= 0xe0) {
> - if (c >= 0xf0) {
> + if (ch >= 0xc2 && ch <= 0xf4) {
> + int code = 0;
> +
> + if (ch >= 0xe0) {
> + if (ch >= 0xf0) {
> /* 0xf0 - 0xf4 */
> - c &= 0x07;
> - code = c << 18;
> - c = **src;
> - ++*src;
> - if (!**src)
> - return -1;
> - if (c < 0x80 || c > 0xbf)
> - return -1;
> - c &= 0x3f;
> + ch &= 0x07;
> + code = ch << 18;
> + ch = next(src);
> + if (ch < 0x80 || ch > 0xbf)
> + goto error;
> + ch &= 0x3f;
> } else {
> /* 0xe0 - 0xef */
> - c &= 0x0f;
> + ch &= 0x0f;
> }
> - code += c << 12;
> + code += ch << 12;
> if ((code >= 0xD800 && code <= 0xDFFF) ||
> code >= 0x110000)
> - return -1;
> - c = **src;
> - ++*src;
> - if (!**src)
> - return -1;
> - if (c < 0x80 || c > 0xbf)
> - return -1;
> + goto error;
> + ch = next(src);
> + if (ch < 0x80 || ch > 0xbf)
> + goto error;
> }
> /* 0xc0 - 0xdf or continuation byte (0x80 - 0xbf) */
> - c &= 0x3f;
> - code += c << 6;
> - c = **src;
> - if (c < 0x80 || c > 0xbf)
> - return -1;
> - c &= 0x3f;
> + ch &= 0x3f;
> + code += ch << 6;
> + ch = next(src);
> + if (ch < 0x80 || ch > 0xbf)
> + goto error;
> + ch &= 0x3f;
> + ch += code;
> + } else if (ch >= 0x80) {
> + goto error;
> }
> - code += c;
> + return ch;
> +error:
> + return '?';
> +}
> +
> +/**
> + * read_string() - read byte from character string
> + *
> + * @src: - pointer to string
> + * Return: - byte read
> + *
> + * The string pointer is incremented if it does not point to '\0'.
> + */
> +static s32 read_string(const char **src)
Please be consistent in your naming. Either the stream reader is called
"next" or "read" :). So either this function would be
next_string/next_char or the function pointer above would be
"read_char"/"read_u8".
> +{
> + s32 c;
> +
> + if (!src || !*src || !**src)
> + return 0;
> + c = (unsigned char)**src;
If you make c a u8 you can save yourself this cast too.
> ++*src;
> - return code;
> + return c;
> +}
> +
> +/**
> + * read_console() - read byte from console
> + *
> + * @src - not used, needed to match interface
> + * Return: - byte read
> + */
> +static s32 read_console(const char **src)
> +{
> + return getc();
> +}
> +
> +efi_status_t console_read_unicode(s32 *code)
> +{
> + /* Check parameter */
> + if (!code)
> + return EFI_INVALID_PARAMETER;
We should not have any efi dependencies in charset.c. I think it's
perfectly fine to just have this function and read_console() in
efi_console.c.
> + /* Check if input available */
> + if (!tstc())
> + return EFI_NOT_READY;
> + /* Read Unicode code */
> + *code = get_code(read_console, NULL);
> +
> + return EFI_SUCCESS;
> +}
> +
> +s32 utf8_get(const char **src)
> +{
> + return get_code(read_string, src);
> }
>
> int utf8_put(s32 code, char **dst)
> diff --git a/lib/efi_loader/efi_console.c b/lib/efi_loader/efi_console.c
> index 3ca6fe536c..a96eda1b22 100644
> --- a/lib/efi_loader/efi_console.c
> +++ b/lib/efi_loader/efi_console.c
> @@ -449,23 +449,22 @@ static efi_status_t EFIAPI efi_cin_read_key_stroke(
> struct efi_simple_text_input_protocol *this,
> struct efi_input_key *key)
> {
> + efi_status_t ret;
> struct efi_input_key pressed_key = {
> .scan_code = 0,
> .unicode_char = 0,
> };
> - char ch;
> + s32 ch;
>
> EFI_ENTRY("%p, %p", this, key);
>
> /* We don't do interrupts, so check for timers cooperatively */
> efi_timer_check();
>
> - if (!tstc()) {
> - /* No key pressed */
> + ret = console_read_unicode(&ch);
> + /* We do not support multi-word codes */
> + if (ret != EFI_SUCCESS || ch >= 0x10000)
For ch > U16_MAX I'd just declare ch as '?'.
/*
* We can not handle multi-char UTF-16 yet, so ignore input
* that would show up as multiple words.
*/
if (ch > U16_MAX)
ch = '=';
Alex
> return EFI_EXIT(EFI_NOT_READY);
> - }
> -
> - ch = getc();
> if (ch == cESC) {
> /*
> * Xterm Control Sequences
> diff --git a/test/unicode_ut.c b/test/unicode_ut.c
> index b94b4a651f..b115d18afd 100644
> --- a/test/unicode_ut.c
> +++ b/test/unicode_ut.c
> @@ -178,7 +178,7 @@ static int ut_utf8_utf16_strlen(struct unit_test_state
> *uts)
>
> /* illegal utf-8 sequences */
> ut_asserteq(4, utf8_utf16_strlen(j1));
> - ut_asserteq(5, utf8_utf16_strlen(j2));
> + ut_asserteq(4, utf8_utf16_strlen(j2));
> ut_asserteq(3, utf8_utf16_strlen(j3));
>
> return 0;
> @@ -196,7 +196,7 @@ static int ut_utf8_utf16_strnlen(struct unit_test_state
> *uts)
>
> /* illegal utf-8 sequences */
> ut_asserteq(4, utf8_utf16_strnlen(j1, 16));
> - ut_asserteq(5, utf8_utf16_strnlen(j2, 16));
> + ut_asserteq(4, utf8_utf16_strnlen(j2, 16));
> ut_asserteq(3, utf8_utf16_strnlen(j3, 16));
>
> return 0;
> @@ -255,8 +255,8 @@ static int ut_utf8_utf16_strcpy(struct unit_test_state
> *uts)
>
> pos = buf;
> utf8_utf16_strcpy(&pos, j2);
> - ut_asserteq(5, pos - buf);
> - ut_assert(!ut_u16_strcmp(buf, L"j2??l", SIZE_MAX));
> + ut_asserteq(4, pos - buf);
> + ut_assert(!ut_u16_strcmp(buf, L"j2?l", SIZE_MAX));
>
> pos = buf;
> utf8_utf16_strcpy(&pos, j3);
>
_______________________________________________
U-Boot mailing list
U-Boot@lists.denx.de
https://lists.denx.de/listinfo/u-boot
