cat Japanese text file

[ntzrmtthihu777]

Hmm... yes I just ran my example on a dupe oto.ini so as not to ruin the real thing if it goes horridly wrong >.< and it would have, lol.
There are a number of Japanese characters whose romaji equivalent would match a few search strings, such as

sha.wav=,0,0,0,0,0

matches all three of these search strings:

s/a.wav=,/a.wav=あ,/g
s/ha.wav=,/ha.wav=は,/g
s/sha.wav=,/sha.wav=しゃ,/g

So how could confusion be prevented in this matter?

[NOTE]
The reverse is not totally true, the same example (but instead from hiragana to romaji) would not have many issues:

しゃ.wav=,0,0,0,0,0

does not match

s/あ.wav=,/あ.wav=a,/g
s/は.wav=,/は.wav=ha,/g
s/しゃ.wav=,/しゃ.wav=sha,/g

Though they are the same in reverse.

[Vaphell]

add ^ to your regex to signal you want to match from start, not anywhere

did you try my script? finetuning s/// expressions by hand for few hundred pairs would be ungodly tedious.

[ntzrmtthihu777]

No, I have not tried it yet, in the middle of a couple of things, >.<
I actually intend to use your original awk script to generate the s/// expressions for me, as it can easily be done with:

$ cat syn.txt
a あ
i い
u う
e え
o お
$ awk '{ printf("s/%s.wav=,/%s.wav=%s,/g\n", $1, $1, $2); }' syn.txt
$ awk '{ printf("s/%s.wav=,/%s.wav=%s,/g\n", $2, $2, $1); }' syn.txt 

If my grasp of awk is correct this should give me the expressions I need, correct?
Also, where exactly would the ^ go? still a bit of a linux n00b, but loving every thing I learn, lol.

[Vaphell]

oh ok, my script uses pure bash to generate but the result will be pretty much identical.

printf("s/^%s.wav=,/%s.wav=%s,/g\n", $1, $1, $2);

[ntzrmtthihu777]

oh ok, my script uses pure bash to generate but the result will be pretty much identical.

printf("s/^%s.wav=,/%s.wav=%s,/g\n", $1, $1, $2);

Ah, thank you so much! I basically want it the way I was describing (two files you use for sed) to mirror an existing Windows alias tool in order to be more familiar to UTAU users. Many thanks, my friend.

[ntzrmtthihu777]

Having an issue with my script and yours, Vaphell.
The iconv is returning an error message, yours at position 6 and mine at 78** or some such number. Any tips?

[Vaphell]

i guess the problem is that the original file is in SHIFT_JIS and the script doesn't take that into account

try this:

iconv -f SHIFT_JIS -t UTF-8 oto.txt | sed -f sed1.txt | iconv -f UTF-8 -t SHIFT_JIS > output1.txt

[ntzrmtthihu777]

i guess the problem is that the original file is in SHIFT_JIS and the script doesn't take that into account

try this:

iconv -f SHIFT_JIS -t UTF-8 oto.txt | sed -f sed1.txt | iconv -f UTF-8 -t SHIFT_JIS > output1.txt

Ah, just figured part of it out, lol. Mine stopped later because I was running it against "test", an oto.ini I removed all (or so I thought) the aliases from (apparently I missed one or two) but yours was targeting the raw oto.ini which still had aliases and stopped short on the first one due to its あ alias. Also the $a and $b orders were wrong, so even when I cleaned out all the aliases it did nothing, lol. Should have been:

while read -r a b
do
  echo "s/^$b[.]wav=[^,]*,/$b.wav=$a,/"
done < syn.txt > hira_roma

while read -r a b
do
  echo "s/^$a[.]wav=[^,]*,/$a.wav=$b,/"
done < syn.txt > roma_hira

[ntzrmtthihu777]

Aha, absolutely done! I can now do as needed, thank you so much for your help Vaphell!