Message ID | 20190625062540.88973-2-dimitriy.ryazantcev@gmail.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
Series | [v5] l10n: localizable upload progress messages | expand |
Dimitriy Ryazantcev <dimitriy.ryazantcev@gmail.com> writes: > Currenly the data rate in throughput_string(...) method is > output by simple strbuf_humanise_bytes(...) call and '/s' append. > But for proper translation of such string the translator needs > full context. > > Add strbuf_humanise_rate(...) method to properly print out > localizable version of data rate ('3.5 MiB/s' etc) with full context. Hmph, so idea is that appending translation of "/s" to translation of "1.4MiB" may not be a good enough translation of "1.4MiB/s"? That does sounds like a good idea, but looking at the heavy duplication of implementation, I would have to say "Yuck" to it. I wonder if an approach like the following illustration would work better? I am not sure how well Q_() and N_() would interact with each other, though. -- >8 -- struct human_format { const char *giga; const char *mega; const char *kilo; const char *byte; const char *bytes; }; static void strbuf_humanise(struct strbuf *buf, off_t bytes, struct human_format *fmt) { if (bytes > 1 << 30) { strbuf_addf(buf, _(fmt->giga), (unsigned)(bytes >> 30), (unsigned)(bytes & ((1 << 30) - 1)) / 10737419); } else if (bytes > 1 << 20) { unsigned x = bytes + 5243; /* for rounding */ strbuf_addf(buf, _(fmt->mega), x >> 20, ((x & ((1 << 20) - 1)) * 100) >> 20); strbuf_addstr(buf, _("")); } else if (bytes > 1 << 10) { unsigned x = bytes + 5; /* for rounding */ strbuf_addf(buf, _(fmt->kilo), x >> 10, ((x & ((1 << 10) - 1)) * 100) >> 10); } else { strbuf_addf(buf, Q_(fmt->byte, fmt->bytes, (unsigned)bytes), (unsigned)bytes); } } void strbuf_humanise_bytes(struct strbuf *buf, off_t bytes) { struct human_format bytes_fmt = { .giga = N_("%u.%2.2u GiB"), .mega = N_("%u.%2.2u MiB"), .kilo = N_("%u.%2.2u KiB"), .byte = N_("%u byte"), .bytes = N_("%u bytes"), }; strbuf_humanise(buf, bytes, &bytes_fmt); } void strbuf_humanise_rate(struct strbuf *buf, off_t bytes) { struct human_format rate_fmt = { .giga = N_("%u.%2.2u GiB/s"), .mega = N_("%u.%2.2u MiB/s"), .kilo = N_("%u.%2.2u KiB/s"), .byte = N_("%u byte/s"), .bytes = N_("%u bytes/s"), }; strbuf_humanise(buf, bytes, &rate_fmt); }
Junio C Hamano <gitster@pobox.com> writes: > > Dimitriy Ryazantcev <dimitriy.ryazantcev@gmail.com> writes: > > > Currenly the data rate in throughput_string(...) method is > > output by simple strbuf_humanise_bytes(...) call and '/s' append. > > But for proper translation of such string the translator needs > > full context. > > > > Add strbuf_humanise_rate(...) method to properly print out > > localizable version of data rate ('3.5 MiB/s' etc) with full context. > > Hmph, so idea is that appending translation of "/s" to translation > of "1.4MiB" may not be a good enough translation of "1.4MiB/s"? > > That does sounds like a good idea, but looking at the heavy > duplication of implementation, I would have to say "Yuck" to it. > > I wonder if an approach like the following illustration would work > better? I am not sure how well Q_() and N_() would interact with > each other, though. Good idea! But unfortunately this will not work properly as xgettext will unable to extract byte/bytes to one plural msgid in *.pot file. Seems it can extract plural forms only from Q_ (ngettext() call) and cannot from _N (gettext_noop() call) in current configuration[0]. Maybe in this case we can try to forward already translated string to strbuf_humanise(): static void strbuf_humanise(struct strbuf *buf, off_t bytes, struct human_format *fmt) { .... } else { strbuf_addf(buf, fmt->byte, (unsigned)bytes); .... } void strbuf_humanise_bytes(struct strbuf *buf, off_t bytes) { struct human_format bytes_fmt = { .giga = _("%u.%2.2u GiB"), .mega = _("%u.%2.2u MiB"), .kilo = _("%u.%2.2u KiB"), .byte = Q_("%u byte", "%u bytes", (unsigned)bytes), }; strbuf_humanise(buf, bytes, &bytes_fmt); } What do you think? [0] https://github.com/git/git/blob/master/Makefile#L2498 > > -- >8 -- > > struct human_format { > const char *giga; > const char *mega; > const char *kilo; > const char *byte; > const char *bytes; > }; > > static void strbuf_humanise(struct strbuf *buf, off_t bytes, struct human_format *fmt) > { > if (bytes > 1 << 30) { > strbuf_addf(buf, _(fmt->giga), > (unsigned)(bytes >> 30), > (unsigned)(bytes & ((1 << 30) - 1)) / 10737419); > } else if (bytes > 1 << 20) { > unsigned x = bytes + 5243; /* for rounding */ > strbuf_addf(buf, _(fmt->mega), > x >> 20, ((x & ((1 << 20) - 1)) * 100) >> 20); > strbuf_addstr(buf, _("")); > } else if (bytes > 1 << 10) { > unsigned x = bytes + 5; /* for rounding */ > strbuf_addf(buf, _(fmt->kilo), > x >> 10, ((x & ((1 << 10) - 1)) * 100) >> 10); > } else { > strbuf_addf(buf, Q_(fmt->byte, fmt->bytes, > (unsigned)bytes), (unsigned)bytes); > } > } > > void strbuf_humanise_bytes(struct strbuf *buf, off_t bytes) > { > struct human_format bytes_fmt = { > .giga = N_("%u.%2.2u GiB"), > .mega = N_("%u.%2.2u MiB"), > .kilo = N_("%u.%2.2u KiB"), > .byte = N_("%u byte"), > .bytes = N_("%u bytes"), > }; > strbuf_humanise(buf, bytes, &bytes_fmt); > } > > void strbuf_humanise_rate(struct strbuf *buf, off_t bytes) > { > struct human_format rate_fmt = { > .giga = N_("%u.%2.2u GiB/s"), > .mega = N_("%u.%2.2u MiB/s"), > .kilo = N_("%u.%2.2u KiB/s"), > .byte = N_("%u byte/s"), > .bytes = N_("%u bytes/s"), > }; > strbuf_humanise(buf, bytes, &rate_fmt); > } -- Sincerely, Dimitriy Ryazantcev
Dimitriy <dimitriy.ryazantcev@gmail.com> writes: > Seems it can extract plural forms only from Q_ (ngettext() call) and > cannot from _N (gettext_noop() call) in current configuration[0]. > Maybe in this case we can try to forward already translated string > to strbuf_humanise(): > > static void strbuf_humanise(struct strbuf *buf, off_t bytes, struct > human_format *fmt) > { > .... > } else { > strbuf_addf(buf, fmt->byte, (unsigned)bytes); > .... > } > > void strbuf_humanise_bytes(struct strbuf *buf, off_t bytes) > { > struct human_format bytes_fmt = { > .giga = _("%u.%2.2u GiB"), > .mega = _("%u.%2.2u MiB"), > .kilo = _("%u.%2.2u KiB"), > .byte = Q_("%u byte", "%u bytes", (unsigned)bytes), > }; > strbuf_humanise(buf, bytes, &bytes_fmt); > } > > What do you think? Well, I was trying to avoid making Q_() call when we did not have to, but if we are running with gettext, we cannot avoid calling Q_() every time this helper is called, as .byte field has to depend on the value of bytes, so...
diff --git a/progress.c b/progress.c index a2e8cf64a8..951f7c7461 100644 --- a/progress.c +++ b/progress.c @@ -150,8 +150,7 @@ static void throughput_string(struct strbuf *buf, uint64_t total, strbuf_addstr(buf, ", "); strbuf_humanise_bytes(buf, total); strbuf_addstr(buf, " | "); - strbuf_humanise_bytes(buf, rate * 1024); - strbuf_addstr(buf, "/s"); + strbuf_humanise_rate(buf, rate * 1024); } void display_throughput(struct progress *progress, uint64_t total) diff --git a/strbuf.c b/strbuf.c index 0e18b259ce..785c9e5b55 100644 --- a/strbuf.c +++ b/strbuf.c @@ -814,19 +814,48 @@ void strbuf_addstr_urlencode(struct strbuf *sb, const char *s, void strbuf_humanise_bytes(struct strbuf *buf, off_t bytes) { if (bytes > 1 << 30) { - strbuf_addf(buf, "%u.%2.2u GiB", + /* TRANSLATORS: IEC 80000-13:2008 gibibyte */ + strbuf_addf(buf, _("%u.%2.2u GiB"), (unsigned)(bytes >> 30), (unsigned)(bytes & ((1 << 30) - 1)) / 10737419); } else if (bytes > 1 << 20) { unsigned x = bytes + 5243; /* for rounding */ - strbuf_addf(buf, "%u.%2.2u MiB", + /* TRANSLATORS: IEC 80000-13:2008 mebibyte */ + strbuf_addf(buf, _("%u.%2.2u MiB"), x >> 20, ((x & ((1 << 20) - 1)) * 100) >> 20); + strbuf_addstr(buf, _("")); } else if (bytes > 1 << 10) { unsigned x = bytes + 5; /* for rounding */ - strbuf_addf(buf, "%u.%2.2u KiB", + /* TRANSLATORS: IEC 80000-13:2008 kibibyte */ + strbuf_addf(buf, _("%u.%2.2u KiB"), x >> 10, ((x & ((1 << 10) - 1)) * 100) >> 10); } else { - strbuf_addf(buf, "%u bytes", (unsigned)bytes); + /* TRANSLATORS: IEC 80000-13:2008 byte */ + strbuf_addf(buf, Q_("%u byte", "%u bytes", (unsigned)bytes), (unsigned)bytes); + } +} + +void strbuf_humanise_rate(struct strbuf *buf, off_t bytes) +{ + if (bytes > 1 << 30) { + /* TRANSLATORS: IEC 80000-13:2008 gibibyte/second */ + strbuf_addf(buf, _("%u.%2.2u GiB/s"), + (unsigned)(bytes >> 30), + (unsigned)(bytes & ((1 << 30) - 1)) / 10737419); + } else if (bytes > 1 << 20) { + unsigned x = bytes + 5243; /* for rounding */ + /* TRANSLATORS: IEC 80000-13:2008 mebibyte/second */ + strbuf_addf(buf, _("%u.%2.2u MiB/s"), + x >> 20, ((x & ((1 << 20) - 1)) * 100) >> 20); + strbuf_addstr(buf, _("")); + } else if (bytes > 1 << 10) { + unsigned x = bytes + 5; /* for rounding */ + /* TRANSLATORS: IEC 80000-13:2008 kibibyte/second */ + strbuf_addf(buf, _("%u.%2.2u KiB/s"), + x >> 10, ((x & ((1 << 10) - 1)) * 100) >> 10); + } else { + /* TRANSLATORS: IEC 80000-13:2008 byte/second */ + strbuf_addf(buf, Q_("%u byte/s", "%u bytes/s", (unsigned)bytes), (unsigned)bytes); } } diff --git a/strbuf.h b/strbuf.h index c8d98dfb95..f62278a0be 100644 --- a/strbuf.h +++ b/strbuf.h @@ -372,6 +372,12 @@ void strbuf_addbuf_percentquote(struct strbuf *dst, const struct strbuf *src); */ void strbuf_humanise_bytes(struct strbuf *buf, off_t bytes); +/** + * Append the given byte rate as a human-readable string (i.e. 12.23 KiB/s, + * 3.50 MiB/s). + */ +void strbuf_humanise_rate(struct strbuf *buf, off_t bytes); + /** * Add a formatted string to the buffer. */
Currenly the data rate in throughput_string(...) method is output by simple strbuf_humanise_bytes(...) call and '/s' append. But for proper translation of such string the translator needs full context. Add strbuf_humanise_rate(...) method to properly print out localizable version of data rate ('3.5 MiB/s' etc) with full context. Strings with the units in strbuf_humanise_bytes(...) are marked for translation. Signed-off-by: Dimitriy Ryazantcev <dimitriy.ryazantcev@gmail.com> --- progress.c | 3 +-- strbuf.c | 37 +++++++++++++++++++++++++++++++++---- strbuf.h | 6 ++++++ 3 files changed, 40 insertions(+), 6 deletions(-)