How to find out the biggest number in many documents that contains different numbers

Question

For example, there are some temperature data at those folders in different time. temps.txt contains the temperature number. So how can I use bash script to find out the maximum temperature? (the results only show the date,time and number of temperature,e.g. ./2011.10.20/00:00/temps.txt 27C).

$ ls
2011.10.20  2012.01.20  2012.04.16  2012.07.12  2012.10.07
2011.10.21  2012.01.21  2012.04.17  2012.07.13  2012.10.08
2011.10.22  2012.01.22  2012.04.18  2012.07.14  2012.10.09
$ cd 2011.10.20

$ ls    
00:00   02:25   04:50   07:15   09:40   12:05   14:30   16:55   19:20   21:45
00:05   02:30   04:55   07:20   09:45   12:10   14:35   17:00   19:25   21:50
00:10   02:35   05:00   07:25   09:50   12:15   14:40   17:05   19:30   21:55
$ cd 00:00
$ ls
temps.txt
$ cat temps.txt
Sensor   Location              Temp
------   --------              ----
#1        PROCESSOR_ZONE       27C/80F

Answer 1

You can use the combination find, grep and awk command to get the desired result. The below is a oneliner which will print the file which has the maximum temperature recorded.

find . -mindepth 3 -exec echo -n "{} " \; -exec grep "PROCESSOR_ZONE" {} \; |
awk '{
    split($4,val,"/");
    gsub("C","",val[1]);
    if (max<val[1]) {file=$1; max=val[1]}
} END {print(file)}'

Output

./2012.04.16/00:10/temps.txt

Below is the script version of the oneliner.

#!/bin/bash

# The path where temperature directories and files are kept
path="/tmp/tmp.ADntEuTlUT/"

# Temp file
tempfile=$(mktemp)

# Get the list of files name and their corresponding
# temperature data.
find "${path}" -mindepth 3 -exec echo -n "{} " \; -exec grep "PROCESSOR_ZONE" {} \; > "${tempfile}"

# Parse though the temp file to find the maximum 
# temperature based on Celsius
awk '{split($4,val,"/");gsub("C","",val[1]);if(max<val[1]){file=$1;max=val[1]}} END{print(file)}' "${tempfile}"

# Removing the temp file
rm -f "${tempfile}"

Answer 2

With GNU grep, assuming no file paths contain newline characters:

grep -rHPo 'PROCESSOR_ZONE\s+\K\d+C' . | awk -F: '
   0+$NF >= max {max = $NF; text = $0}; END {print text}'

Answer 3

This solution makes used of the split function in awk to split the fields and do the numeric reverse sort to pop the max number to the top.

find . -name "temps.txt" -print|xargs tail -n 1 | awk '{split($NF,temp,"[CF/]");print temp[1]}'|sort -r | head -n 1