Extract numbers from a text file and place them in another file

Question

I have a text file whose contents looks like this:

{"result":"true","last":0.02094,"high":0.0213,"low":0.01682,"avg":0.01871,"sell":0.02094,"buy":0.02054,"vol_ltc":5145.8262,"vol_btc":96.29887,"rate_change_percentage":"23.47"}`

(This is data from an API)

I want it formatted so my program can read it, but I need to extract the numbers because some of the exchanges are formatted differently by the API, for example, this:

{"result":"true","last":"0.00000070","high":"0.00000073","low":"0.00000069","avg":"0.00000071","sell":"0.00000071","buy":"0.00000069","vol_doge":71072621.657,"vol_btc":50.64272,"rate_change_percentage":"0.00"}

(Difference: quotes around price values)
And I need something that looks like this:
(Using the first one)

0.02094
0.0213
0.01682
0.01871
0.02094
0.02054
5145.8262
96.29887
23.47

Alternatively something like this (again, using the first one)

last
0.02094
high
0.0213
low
0.01682
avg
0.01871
sell
0.02094
buy
0.02054
vol_ltc
5145.8262
vol_btc
96.29887
rate_change_percentage
23.47

If you need more examples of strings provided from the API, go to http://data.bter.com/api#tickers

Answer 1

As the API returns JSON objects, I'd use a JSON parser for it, combined with sed and grep.

I've decided for jq, which is contained in the Ubuntu repo's, so it could be installed using sudo apt-get install jq; I don't know what distro you are using, so you'll have to find that by yourself if you want to use this solution.

jq ".[]" <filename> | sed s/\"//g | grep -P '^\d+(\.\d+)$'

Now, let me split this up:

1. jq ".[]" <filename> will print the value for each key in the map. This probably doesn't work with nested maps, though.
2. sed s/\"//g will strip the list of any quotation marks.
3. grep -P '^\d+(\.\d+)?$' will return those lines that consist of integer or decimal numbers only.

This command chain will print the result to stdout. To get it into a file, simply pipe the output into it:

jq ".[]" <filename> | sed s/\"//g | grep -P '^\d+(\.\d+)$' > <other_filename>

Answer 2

here is a perl one-liner:

perl -pe 's/.*?(\d+\.\d+)/$1 . "\n"/eg' file1 > file2