grep regex to ignore comment at end of line

Question

I'm trying to grep through a lot of old PowerBASIC source files in search of a variable, but I'm having trouble getting grep to avoid matching references to the variable in the end-of-line comments.

For example:

ANGLE = 40    ' THIS IS A COMMENT ABOUT ANGLE'S VALUE
FOO = 3/ANGLE ' ANGLE CAN APPEAR ON RIGHT SIDE AS WELL
DELTA = 35    ' ANGLE AND DELTA AREN'T FRIENDS

Initially I was using:

# grep "\bANGLE\b.*'" SRC_FILE.BAS

But -- besides ignoring lines with no comments -- it also prints out lines of code which don't use ANGLE at all (such as the DELTA line). This is because those lines happen to also have a single apostrophe (') in their comments (i.e. AREN'T).

I thought of piping the output to remove the lines I don't want with:

# grep "\bANGLE\b" SRC_FILE.BAS | grep -v "'.*\bANGLE\b"

But unfortunately it also removes the lines of code I do want that just happen to have "ANGLE" in their comments.

How do I get grep to match the lines of actual code involving ANGLE and ignore those lines with only matches in the comments?

Answer 1

grep "^[^']*\bANGLE\b" SRC_FILE.BAS

This only matches lines where everything to the left of ANGLE does not include '. It will have a problem similar to jnylen's solution if there's an apostrophe in a string.

Answer 2

You could take out the comments first:

sed "s/'.*$//" SRC_FILE.BAS | grep '\bANGLE\b'

This will still fail in some cases, if a string contains the word ANGLE or a single quote. To solve this problem accurately, you would need to write a parser instead of using regular expressions.