From c90ec1a9352898213af737932e61730a0b7d6f23 Mon Sep 17 00:00:00 2001
From: falsycat <me@falsy.cat>
Date: Tue, 6 Sep 2022 18:15:15 +0900
Subject: [PATCH] add fprob_calc.sh

---
 exp/bidx_random.sh | 15 +++++++++++++
 exp/fprob_calc.sh  | 55 ++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 70 insertions(+)
 create mode 100755 exp/bidx_random.sh
 create mode 100755 exp/fprob_calc.sh

diff --git a/exp/bidx_random.sh b/exp/bidx_random.sh
new file mode 100755
index 0000000..3ce45b7
--- /dev/null
+++ b/exp/bidx_random.sh
@@ -0,0 +1,15 @@
+#!/bin/bash
+
+# generate random block indices
+
+row=$1
+col=$2
+max=$3
+
+for ((y = 0; y < row; ++y)); do
+  n=$((RANDOM%col + 1))
+  for ((x = 0; x < n; ++x)); do
+    printf "$((RANDOM%max)) "
+  done
+  printf "\n"
+done
diff --git a/exp/fprob_calc.sh b/exp/fprob_calc.sh
new file mode 100755
index 0000000..35a2e2f
--- /dev/null
+++ b/exp/fprob_calc.sh
@@ -0,0 +1,55 @@
+#!/bin/bash
+
+# calculate fprob from indices and fprob matrix
+# output: average feature's feature probability and average non-feature's feature probability
+
+bidx=$1
+fprob=$2
+n=$3
+
+script=$(cat - <<EOS
+BEGIN {
+  type = 0;  # 0: bidx, 1: fprob
+
+  nprob     = 0;
+  nprob_max = 0;
+  pprob     = 0;
+  pprob_max = 0;
+}
+
+{
+  if (type == 0) {
+    split("", feats);
+
+    for (i = 1; i <= NF; ++i) {
+      feats[i] = \$i;
+    }
+
+    type = 1;
+  } else if (type == 1) {
+    sum = 0;
+    for (i = 1; i <= NF; ++i) {
+      sum += \$i;
+    }
+
+    lprob = 0;
+    for (i = 1; i <= length(feats); ++i) {
+      lprob += \$(feats[i]+1);
+    }
+
+    nprob += sum-lprob;
+    pprob += lprob;
+
+    nprob_max += NF-length(feats)
+    pprob_max += length(feats)
+
+    type  = 0;
+  }
+}
+
+END {
+  print pprob/pprob_max, nprob/nprob_max;
+}
+EOS
+)
+paste -d "\n" $bidx $fprob | head -n$((n*2)) | awk "$script"