For completeness' sake, here is an EDL that comskip generated last evening:
0.00 16.97 0
372.40 534.00 0
1018.77 1185.20 0
1746.77 2007.53 0
2700.23 2983.27 0
3495.80 3779.77 0
Similarly, here is the ffmeta file that comskip generated for the same program (same input, 2 different output formats):
;FFMETADATA1
[CHAPTER]
TIMEBASE=1/100
START=0
END=1696
title=Commercial Segment
[CHAPTER]
TIMEBASE=1/100
START=1700
END=37240
title=Show Segment
[CHAPTER]
TIMEBASE=1/100
START=37240
END=53400
title=Commercial Segment
[CHAPTER]
TIMEBASE=1/100
START=53403
END=101876
title=Show Segment
[CHAPTER]
TIMEBASE=1/100
START=101876
END=118520
title=Commercial Segment
[CHAPTER]
TIMEBASE=1/100
START=118523
END=174676
title=Show Segment
[CHAPTER]
TIMEBASE=1/100
START=174676
END=200753
title=Commercial Segment
[CHAPTER]
TIMEBASE=1/100
START=200756
END=270023
title=Show Segment
[CHAPTER]
TIMEBASE=1/100
START=270023
END=298326
title=Commercial Segment
[CHAPTER]
TIMEBASE=1/100
START=298330
END=349580
title=Show Segment
[CHAPTER]
TIMEBASE=1/100
START=349580
END=377976
title=Commercial Segment
As you can see, it would be pretty simple to use a simple AWK script to parse the EDL and generate a complementary ffmeta.
However, if you are interested in completely removing the commercial segments, here is the ffsplit file that comskip generated to output several segments, that could then be concatenated into a single without the commercials.
-c copy -ss 17.000 -t 355.400 segment001.ts
-c copy -ss 534.033 -t 484.733 segment002.ts
-c copy -ss 1185.233 -t 561.533 segment003.ts
-c copy -ss 2007.567 -t 692.667 segment004.ts
-c copy -ss 2983.300 -t 512.500 segment005.ts