source: branches/lib/GDE/SINA/builddir/tests/readwrite.test

#!/bin/bash

source tests/test_helper.sh

TEST_ARB=$TEST_ARGS
maketmpdir T

cp $TEST_ARB $T/ref.arb

SINA="./src/sina --prealigned --preserve-order"


begin_test "Copy ARB to FASTA (TEST_ARB->test.fasta)"
capture_stdouterr "$SINA -i $T/ref.arb -o $T/test.fasta --select-step 1000"
assert_exit_success
assert_output_contains "align 12 sequences"
end_test

begin_test "Copy ARB to FASTA.gz (TEST_ARB->test.fasta.gz)"
capture_stdouterr "$SINA -i $T/ref.arb -o $T/test.fasta.gz --select-step 1000"
assert_exit_success
assert_output_contains "align 12 sequences"
capture_stderr "gunzip -c $T/test.fasta.gz | cmp - $T/test.fasta"
assert_exit_success
end_test

begin_test "Copy ARB to FASTA (TEST_ARB->STDOUT)"
capture_stdouterr "$SINA -i $T/ref.arb -o - --select-step 1000 > $T/test2.fasta"
assert_exit_success
assert_output_contains "align 12 sequences"
capture_stderr "cmp $T/test.fasta $T/test2.fasta"
assert_exit_success
end_test

begin_test "Copy ARB to ARB (TEST_ARB->test.arb)"
capture_stdouterr "$SINA -i $T/ref.arb -o $T/test.arb --select-step 1000"
assert_exit_success
assert_output_contains "align 12 sequences"
end_test

begin_test "Copy FASTA to ARB (test.fasta->test2.arb)"
capture_stdouterr "$SINA -i $T/test.fasta -o $T/test2.arb"
assert_exit_success
assert_output_contains "align 12 sequences"
capture_stdouterr "cmp $T/test.arb $T/test2.arb"
assert_exit_success
end_test

begin_test "Copy FASTA.gz to ARB (test.fasta->test2.arb)"
capture_stdouterr "$SINA -i $T/test.fasta.gz -o $T/test2.arb"
assert_exit_success
assert_output_contains "align 12 sequences"
capture_stdouterr "cmp $T/test.arb $T/test2.arb"
assert_exit_success
end_test

begin_test "Copy ARB into ARB (TEST_ARB->test2.arb)"
capture_stdouterr "$SINA -i $T/ref.arb -o $T/test2.arb --select-step 1001"
assert_exit_success
assert_output_contains "align 12 sequences"
capture_stdouterr "cmp $T/test.arb $T/test2.arb"
assert_exit_failure
end_test

begin_test "Copy ARB to NULL (outtype=NONE)"
capture_stdouterr "$SINA -i $T/test2.arb --outtype NONE"
assert_exit_success
assert_output_contains "align 23 sequences"
end_test

begin_test "Copy FASTA to NULL (test.fasta->/dev/null)"
capture_stdouterr "$SINA -i $T/test.fasta -o /dev/null"
assert_exit_success
assert_output_contains "align 12 sequences"
end_test

begin_test "Meta Data in FASTA header"
capture_stdouterr "$SINA -i $T/test.arb -o $T/test.fasta --meta-fmt header -f acc"
assert_exit_success
assert_output_contains "align 12 sequences"
grep "^>" $T/test.fasta | sed 's/.*\[acc=\([^]]*\)\].*/\1/' | sort -u > $T/acc.txt
capture_stdout "wc -l $T/acc.txt"
assert_output_contains "12"
end_test

begin_test "Meta Data in sidecar CSV"
capture_stdouterr "$SINA -i $T/test.arb -o $T/test.fasta --meta-fmt csv -f acc"
assert_exit_success
assert_output_contains "align 12 sequences"
cut -d, -f2 $T/test.csv | tr -d $'\\r' | tail -n+2 | sort  > $T/acc2.txt
assert_exit_success "cmp $T/acc.txt $T/acc2.txt"
end_test

begin_test "Meta Data in FASTA comments"
capture_stdouterr "$SINA -i $T/test.arb -o $T/test.fasta --meta-fmt comment -f acc"
assert_exit_success
assert_output_contains "align 12 sequences"
sed -n 's/^; acc=//p' $T/test.fasta | sort > $T/acc2.txt
assert_exit_success "cmp $T/acc.txt $T/acc2.txt"
end_test

begin_test "CSV writer"
capture_stdouterr "$SINA -i $T/test.arb -o $T/test.fasta -o $T/test.csv -f acc"
assert_exit_success
assert_output_contains "align 12 sequences"
cut -d, -f2 $T/test.csv | tail -n+2 | sort  > $T/acc2.txt
assert_exit_success "cmp $T/acc.txt $T/acc2.txt"
end_test

begin_test "CSV writer - stdout"
capture_stdouterr "$SINA -i $T/test.arb -o $T/test.fasta --outtype CSV -o - -f acc > $T/test.csv"
assert_exit_success
assert_output_contains "align 12 sequences"
cut -d, -f2 $T/test.csv | tail -n+2 | sort  > $T/acc2.txt
assert_exit_success "cmp $T/acc.txt $T/acc2.txt"
end_test

begin_test "CSV writer - gzip"
capture_stdouterr "$SINA -i $T/test.arb  -o $T/test.csv.gz -f acc"
assert_exit_success
assert_output_contains "align 12 sequences"
gzip -dc $T/test.csv.gz | cut -d, -f2 | tail -n+2 | sort  > $T/acc2.txt
assert_exit_success "cmp $T/acc.txt $T/acc2.txt"
end_test

begin_test "CSV writer - CRLF"
capture_stdouterr "$SINA -i $T/test.arb -o $T/test.csv -f acc --csv-crlf"
assert_exit_success
assert_output_contains "align 12 sequences"
cut -d, -f2 $T/test.csv | tail -n+2 | sort  > $T/acc2.txt
assert_exit_failure "cmp $T/acc.txt $T/acc2.txt"
assert_exit_success "tr -d $'\\r' < $T/acc2.txt | cmp $T/acc.txt -"
assert_exit_success "grep -c $'\\r' $T/test.csv"
assert_output_contains 13
end_test