Pasting your sample data into the following 1-liner (wrapped for posting only) produced the following output. (Season to taste:)

perl -0777pe " 
s[[^0-9() -]+][\n]g; 
s[\s{3,}][\n]g; 
s[---][]g; 
for my$n(1..4){ s[\n.{1,7}\n][\n]msg;}  
print" -
[PASTE SNIPPED]
^Z

(440) 248-2700
(440) 975-0537
(330) 836-9300
(330) 864-5741
 (313)884-6400
(313) 884-0600
(313) 590-9253

(440) 248-2700
(440) 975-0537
(330) 836-9300
(330) 864-5741
 (313)884-6400
(313) 884-0600
(313) 590-9253
[download]

If your data file is very large you would need to drop the slurp (-0777) which would mean that as-is, the filtering wouldn't be as effective. But the principle of first throwing away as much as possible (safetly--replacing with spaces or newlines so that you don't run good data together) is a useful first pass at extracting small pieces from large volumes.