Question

Consider a tee export of a MySQL query.

SELECT * FROM mytable;
+----------+-------+----------+-----------+-------+------+--------+
| label1   | lbl2  | label3   | label4    | lbl5  | lbl6 | label7 |
+----------+-------+----------+-----------+-------+------+--------+
| ABCDEFGH | YNNYY | 0.001596 |  0.161152 |     2 |    1 | a      |
| ABCDEFGH | YNNYY | 0.001404 |  0.162774 |     3 |    1 | a      |
     *
     *
     *
| ABCDEFGH | YNNYY | 0.001286 | 10.941642 |  5999 |    1 | a      |
| ABCDEFGH | YNNYY | 0.001315 | 10.942950 |  6000 |    1 | a      |
+----------+-------+----------+-----------+-------+------+--------+
9995 rows in set (0.04 sec)

I would like to process this mysqlqtee.txt file through sed or perl, filtering the actual data rows only.

I can tell sed or perl to: “Comment out every line starting with the static text of ‘| AB’, please!”

sed -i '.old' 's/\(^\| AB.*\)/#\1/g' mysqlqtee.txt
perl -pi.old -e 's/(^\| AB.*)/#$1/g' mysqlqtee.txt

These get me:

SELECT * FROM mytable ORDER BY timecode;
+----------+-------+----------+-----------+-------+------+--------+
| label1   | lbl2  | label3   | label4    | lbl5  | lbl6 | label7 |
+----------+-------+----------+-----------+-------+------+--------+
#| ABCDEFGH | YNNYY | 0.001596 |  0.161152 |     2 |    1 | a      |
#| ABCDEFGH | YNNYY | 0.001404 |  0.162774 |     3 |    1 | a      |
     *
     *
     *
#| ABCDEFGH | YNNYY | 0.001286 | 10.941642 |  5999 |    1 | a      |
#| ABCDEFGH | YNNYY | 0.001315 | 10.942950 |  6000 |    1 | a      |
+----------+-------+----------+-----------+-------+------+--------+
9995 rows in set (0.04 sec)

nicely commenting out all the actual data rows and leave every other line untouched.

What I am so far unable to tell perl or sed is to: “Comment out every line starting with anything OTHER THAN the static text of ‘| AB’, please!”
Which would get me:

#SELECT * FROM mytable ORDER BY timecode;
#+----------+-------+----------+-----------+-------+------+--------+
#| label1   | lbl2  | label3   | label4    | lbl5  | lbl6 | label7 |
#+----------+-------+----------+-----------+-------+------+--------+
| ABCDEFGH | YNNYY | 0.001596 |  0.161152 |     2 |    1 | a      |
| ABCDEFGH | YNNYY | 0.001404 |  0.162774 |     3 |    1 | a      |
     *
     *
     *
| ABCDEFGH | YNNYY | 0.001286 | 10.941642 |  5999 |    1 | a      |
| ABCDEFGH | YNNYY | 0.001315 | 10.942950 |  6000 |    1 | a      |
#+----------+-------+----------+-----------+-------+------+--------+
#9995 rows in set (0.04 sec)

I don’t seem to be able to translate the “IT DOESN’T START WITH” part to regexp.
The dual use of ^ meaning NOT and LINE_START at the same time causes trouble.
I can negate the starting letter with s/^[^\|]/, but that also leaves the header row out.

I managed to do this in perl, using an IF statement.
But it still disturbs me way too much that I couldn’t do it with a single s///g.
Can this be done that way?

How do I translate the “IT DOESN’T START WITH” part to regexp? Either sed or perl solution is fine!

Answer 1

You can negate a regex with sed by putting ! after it like so:

sed -i.old '/^| AB/!s/^/#/' mysqltree.txt

Output

$ sed '/^| AB/!s/^/#/' mysqltree.txt
#SELECT * FROM mytable;
#+----------+-------+----------+-----------+-------+------+--------+
#| label1   | lbl2  | label3   | label4    | lbl5  | lbl6 | label7 |
#+----------+-------+----------+-----------+-------+------+--------+
| ABCDEFGH | YNNYY | 0.001596 |  0.161152 |     2 |    1 | a      |
| ABCDEFGH | YNNYY | 0.001404 |  0.162774 |     3 |    1 | a      |
| ABCDEFGH | YNNYY | 0.001286 | 10.941642 |  5999 |    1 | a      |
| ABCDEFGH | YNNYY | 0.001315 | 10.942950 |  6000 |    1 | a      |
#+----------+-------+----------+-----------+-------+------+--------+
#9995 rows in set (0.04 sec)