Break down text file in bash - regex

Question

I have a text file in the following format:
variableStep chrom=chr1 span=10
10161 1
10171 1
10181 2
10191 2
10201 2
10211 2
10221 2
10231 2
10241 2
10251 1
variableStep chrom=chr10 span=10
70711 1
70721 2
70731 2
70741 2
70751 2
70761 2
70771 2
70781 2
70791 1
71161 1
71171 1
71181 1
variableStep chrom=chr11 span=10
104731 1
104741 1
104751 1
104761 1
104771 1
104781 1
104791 1
104801 1
128711 1
128721 1
128731 1
I need a way to break this down into several files named for example "chr1.txt", "chr10.txt and "chr11.txt". How would I go about doing this?
I about the the following way:
cat file.txt | \
while IFS=$'\t' read -r -a rowArray; do
echo -e "${rowArray[0]}\t${rowArray[1]}\t${rowArray[2]}"
done > $file.mod.txt
That reads line by line and then saves line by line. However, I need something a little more elaborate that spans rows. "chr1.txt" would include everything from the row 10161 1 to row 10251 1, "chr10.txt" would include everything from the row 70711 1 to row 71181 1, etc. It's also specific in that I have to read in the actual chr# from each line as well, and save that as the file name.
The help is really appreciated.

Answer 1

awk -F'[ =]' '
$1 == "variableStep" {file = $3 ".txt"; next}
file != "" {print > file}' < input.txt

Answer 2

This worked for me:
IFS=$'\n'
curfile=""
content=($(< file.txt))
for ((idx = 0; idx < ${#content[#]}; idx++)); do
if [[ ${content[idx]} =~ ^.*chrom=(\\b.*?\\b)\ .*$ ]]; then
curfile="${BASH_REMATCH[1]}.txt"
rm -rf ${curfile}
elif [ -n "${curfile}" ]; then
echo ${content[idx]} >> ${curfile}
fi
done

Answer 3

Awk is appropriate for this problem domain because the text file is already (more or less) organized into columns. Here's what I would use:
awk 'NF == 3 && index($2, "=") { filename = substr($2, index($2, "=") + 1) }
NF == 2 && filename { print $0 > (filename ".txt") }' < input.txt
Explanation:
Think of the lines starting with variableStep as "three columns" and the other lines as "two columns". The above script says, "Parse the text file line-by-line; if a line has three columns and the second column contains an '=' character, assign 'all of the characters in the second column that occur after the '=' character' to a variable called filename. If a line has two columns and the filename variable's been assigned, write the entire line to the file that's constructed by concatenating the string in the filename variable with '.txt'".
Notes:
NF is a built-in variable in Awk that represents the "number of fields", where a "field" (in this case) can be thought of as a column of data.
$0 and $2 are built-in variables that represent the entire line and the second column of data, respectively. ($1 represents the first column, $3 represents the third column, etc...)
substr and index are built-in functions described here: http://www.gnu.org/software/gawk/manual/gawk.html#String-Functions
The redirection operator (>) acts differently in Awk than it does in a shell script; subsequent writes to the same file are appended.
String concatenation is performed simply by writing expressions next to each other. The parenthesis ensure the concatenation happens before the file gets written to.
More details can be found here: http://www.gnu.org/software/gawk/manual/gawk.html#Two-Rules

Answer 4

i used sed to filter ....
code part :
Kaizen ~/so_test $ cat zsplit.sh
cntr=1;
prev=1;
for curr in `cat ztmpfile2.txt | nl | grep variableStep | tr -s " " | cut -d" " -f2 | sed -n 's/variableStep//p'`
do
sed -n "$prev,$(( ${curr} - 1))p" ztmpfile2.txt > zchap$cntr.txt ;
#echo "displaying : : zchap$cntr.txt " ;
#cat zchap$cntr.txt ;
prev=$curr; cntr=$(( $cntr + 1 ));
done
sed -n "$prev,$ p" ztmpfile2.txt > zchap$cntr.txt ;
#echo "displaying : : zchap$cntr.txt " ;
#cat zchap$cntr.txt ;
output :
Kaizen ~/so_test $ ./zsplit.sh
+ ./zsplit.sh
zchap1.txt :: 1 :: 1
displaying : : zchap1.txt
variableStep chrom=chr1 span=10
zchap2.txt :: 1 :: 12
displaying : : zchap2.txt
variableStep chrom=chr1 span=10
10161 1
10171 1
10181 2
10191 2
10201 2
10211 2
10221 2
10231 2
10241 2
10251 1
zchap3.txt :: 12 :: 25
displaying : : zchap3.txt
variableStep chrom=chr10 span=10
70711 1
70721 2
70731 2
70741 2
70751 2
70761 2
70771 2
70781 2
70791 1
71161 1
71171 1
71181 1
displaying : : zchap4.txt
variableStep chrom=chr11 span=10
104731 1
104741 1
104751 1
104761 1
104771 1
104781 1
104791 1
104801 1
128711 1
128721 1
128731 1
from the result zchap* files , iff you want you can remove the line : variableStep chrom=chr11 span=10 by using sed -- sed -i '/variableStep/d' zchap*
does this help ?