Extract temperature from data bash

Question

I'm using the command line tool from Temperature Monitor, the mac software, which looks like this:

$ /Applications/TemperatureMonitor.app/Contents/MacOS/tempmonitor -c -l -a


SMART Disk Hitachi HTS547550A9E384 (J2250050GMBY3C): 30 C
SMART Disk TOSHIBA MK5065GSXF (71SNCPW4T): 28 C
SMC BATTERY: 30 C
SMC BATTERY POSITION 2: 31 C
SMC BATTERY POSITION 3: 28 C
SMC CPU A DIODE: 47 C
SMC CPU A PROXIMITY: 45 C
SMC GPU 1 CHIP: 40 C
SMC LEFT PALM REST: 28 C
SMC MAIN HEAT SINK 2: 38 C
SMC MAIN HEAT SINK 3: 37 C
SMC MAIN LOGIC BOARD: 36 C
SMC PLATFORM CONTROLLER HUB: 49 C
SMC SSD BAY: 36 C

I want to clean this up a bit. So for example, let's say I want to get the average of the three Battery temperature readings. I thought of piping into grep for Battery, then awking through all the fields for the correct data, but that seems really messy.

So I want the three variables $BATTERY_1, $BATTERY_2, and $BATTERY_3 to have the content 30, 31, and 28 respectively.

Any suggestions on the cleanest way to do so?

Answer 1

It will be easier to create an array and then move the values from the array into the plain variables. It is trivial to do the extraction with awk:

TEMPMON="/Applications/TemperatureMonitor.app/Contents/MacOS/tempmonitor"
battery=( $("$TEMPMON" -c -l -a | awk '/BATTERY/ { print $(NF-1) }') )
BATTERY_1=${battery[0]}
BATTERY_2=${battery[1]}
BATTERY_3=${battery[2]}

Answer 2

To complement @Jonathan Leffler's helpful answer:

If you don't actually need the individual values and instead want the average only, try:

... | awk '/ BATTERY/ {sum+=$(NF-1); ++i} END {OFMT="%.2f"; print sum / i}'

• OFMT="%.2f" sets the (printf-style) output number format to 2 decimal places, resulting in 29.67.

---

Update: The OP, in a comment, asks for output in the format <Item name>: <avg temp> (<temp 1>, <temp 2>, <temp 3>):

... | awk -v itm='BATTERY' '
  $0 ~ itm {
    vals = vals (i ? " " : "") $(NF-1)
    sum += $(NF-1); ++i
  } 
  END {
    printf "%s: %.2f (%s)\n", itm, sum / i, vals
  }'

• -v itm='BATTERY' passes the name of the items to match as awk variable itm.
• $0 ~ itm matches (~) the current input line ($0) against itm (interpreted as a regular expression, which in this simple case performs substring matching).
• awk splits input lines into fields $1, $2, ... by whitespace by default, and stores the number of fields in special variable NF. Since the values in the input data are in the next-to-last field, $(NF-1) references each line's value.
• vals = ... builds up a string list of matching values; note how merely placing strings and variables next to each other causes them to be concatenated (as strings).
  • (i ? " " : "") is a C-style ternary conditional that returns a single space if condition i is true (i.e., if variable i has a nonzero value), and an empty string otherwise. In other words: if the value is not the first one, append a space before appending the value to the list of values built up so far. Note that uninitialized variables in awk default to an empty string in a string context, and 0 (false) in a numeric/Boolean context.
• sum += ... sums up the values; ++i keeps the count of values.
• END is a special pattern whose associated action (block) is processed after all input lines.
• printf, for output based on a format (template) string, works like its C counterpart, and in this case outputs the item name (1st %s, instantiated with itm), the average with 2 decimal places (%.2f, instantiated with sum / i) and the list of values (last %s, instantiated with vals).