🎨 simplify awk program

facebookresearch · Mar 27, 2020 · b09eb28 · b09eb28
1 parent ec524ce
commit b09eb28
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 36 deletions.
diff --git a/scripts/measure_scores.e2e.replfail.sh b/scripts/measure_scores.e2e.replfail.sh
@@ -23,8 +23,6 @@ fi
 SCORER=e2e-metrics/measure_scores.py
 org=data/E2E_compositional/testset_w_refs.csv
 tmp=scripts/tmp
-baserepl=$tmp/baserepl
-hyprepl=$tmp/hyprepl
 hyp=$tmp/hyp
 ref=$tmp/ref
 src=$tmp/src
@@ -36,18 +34,11 @@ rmtreeinfo () {
 }
 
 repl=$(grep H- $gen | awk -F '\t' '$2=="-inf" {print $1}' | cut -d '-' -f 2 | awk '{print $1+1}')
-
-awk 'NR==FNR{l[$0];next;} (FNR in l)' <(echo "$repl") \
-  <(grep H- $base | sort -n -k 2 -t -) > $baserepl
-
-awk 'NR==FNR{l[$0];next;} !(FNR in l)' <(echo "$repl") \
-  <(grep H- $gen | sort -n -k 2 -t -) > $hyprepl
-
-cat $baserepl >> $hyprepl
-cat $hyprepl | sort -n -k 2 -t - | awk -F '\t' '{print $3}' | rmtreeinfo > $hyp
+awk -F '\t' 'NR==FNR {l[$0];next;} !(FNR in l) {print $3} (FNR in l) {print $6}' \
+  <(echo "$repl") <(paste <(grep H- $gen | sort -n -k 2 -t -) <(grep H- $base | sort -n -k 2 -t -)) | \
+  rmtreeinfo > $hyp
 
 cat $ref_tree | rmtreeinfo > $ref
-
 tail -n +2 $org | cut -d '"' -f 2 > $src
 
 python scripts/_eval_e2e_helper.py

diff --git a/scripts/measure_scores.replfail.sh b/scripts/measure_scores.replfail.sh
@@ -22,8 +22,6 @@ fi
 
 SCORER=e2e-metrics/measure_scores.py
 tmp=scripts/tmp
-baserepl=$tmp/baserepl
-hyprepl=$tmp/hyprepl
 hyp=$tmp/hyp
 ref=$tmp/ref
 
@@ -34,15 +32,9 @@ rmtreeinfo () {
 }
 
 repl=$(grep H- $gen | awk -F '\t' '$2=="-inf" {print $1}' | cut -d '-' -f 2 | awk '{print $1+1}')
-
-awk 'NR==FNR{l[$0];next;} (FNR in l)' <(echo "$repl") \
-  <(grep H- $base | sort -n -k 2 -t -) > $baserepl
-
-awk 'NR==FNR{l[$0];next;} !(FNR in l)' <(echo "$repl") \
-  <(grep H- $gen | sort -n -k 2 -t -) > $hyprepl
-
-cat $baserepl >> $hyprepl
-cat $hyprepl | sort -n -k 2 -t - | awk -F '\t' '{print $3}' | rmtreeinfo > $hyp
+awk -F '\t' 'NR==FNR {l[$0];next;} !(FNR in l) {print $3} (FNR in l) {print $6}' \
+  <(echo "$repl") <(paste <(grep H- $gen | sort -n -k 2 -t -) <(grep H- $base | sort -n -k 2 -t -)) | \
+  rmtreeinfo > $hyp
 
 cat $ref_tree | rmtreeinfo > $ref
 

diff --git a/scripts/tree_acc.replfail.sh b/scripts/tree_acc.replfail.sh
@@ -14,8 +14,6 @@ tmp=scripts/tmp
 tsv=$tmp/tsv
 id=$tmp/id
 src=$tmp/src
-baserepl=$tmp/baserepl
-hyprepl=$tmp/hyprepl
 hyp=$tmp/hyp
 
 mkdir -p $tmp
@@ -24,20 +22,13 @@ tree_acc () {
   python compute_tree_acc.py -tsv $1
 }
 
-repl=$(grep H- $gen | awk -F '\t' '$2=="-inf" {print $1}' | cut -d '-' -f 2 | awk '{print $1+1}')
-
 grep S- $gen | sort -n -k 2 -t - | awk -F '\t' '{print $1}' | sed 's/^S-//' > $id
-
 grep S- $gen | sort -n -k 2 -t - | awk -F '\t' '{print $2}' > $src
 
-awk 'NR==FNR{l[$0];next;} (FNR in l)' <(echo "$repl") \
-  <(grep H- $base | sort -n -k 2 -t -) > $baserepl
-
-awk 'NR==FNR{l[$0];next;} !(FNR in l)' <(echo "$repl") \
-  <(grep H- $gen | sort -n -k 2 -t -) > $hyprepl
-
-cat $baserepl >> $hyprepl
-cat $hyprepl | sort -n -k 2 -t - | awk -F '\t' '{print $3}' > $hyp
+repl=$(grep H- $gen | awk -F '\t' '$2=="-inf" {print $1}' | cut -d '-' -f 2 | awk '{print $1+1}')
+awk -F '\t' 'NR==FNR {l[$0];next;} !(FNR in l) {print $3} (FNR in l) {print $6}' \
+  <(echo "$repl") <(paste <(grep H- $gen | sort -n -k 2 -t -) <(grep H- $base | sort -n -k 2 -t -)) \
+  > $hyp
 
 paste $id $src $hyp > $tsv