added shebang, replaced if-elses with shortcut notation, ported online mean and variance calc to mean.awk

1 files changed, 25 insertions, 7 deletions

diff --git a/diff.awk b/diff.awk
index 1032659..6dc427d 100644
--- a/diff.awk
+++ b/diff.awk
@@ -1,3 +1,5 @@
+#!/usr/bin/awk -f
+
 ### diff.awk
 # numerical diff along columns
 
@@ -12,11 +14,16 @@ BEGIN {
 NR == 1 {
     header_nf = NF
     for (n=1; n<=NF; n++) {
-        if ($n !~ number)
-            header[n] = "d" $n
-        else
-            header[n] = "col_" n
-        printf("%s,", header[n])
+        printf("%s", $n)
+        ($n !~ number) ?  dheader[n] = "d" $n : dheader[n] = "dcol_" n
+        if (n <= NF)
+            printf("%s", OFS)
+    }
+
+    for (n=1; n<=NF; n++) {
+        printf("%s", dheader[n])
+        if (n < NF)
+            printf("%s", OFS)
     }
 }
 
@@ -26,11 +33,22 @@ NF != 0 {
 
     ### iterate over columns
     for (y=1; y<=max_nf; y++) {
+        if (match(dheader[y], $y))
+            continue
+        if ($y ~ number)
+            printf("%g", $y)
+        if (y <= max_nf)
+            printf("%s", OFS)
+    }
+
+    for (y=1; y<=max_nf; y++) {
         if ($y ~ number) {
             data[y] = $y
-            diff[y] = data[y] - data_prev[y]
-            printf("%g,", diff[y])
+            (data_prev[y] ~ number) ?  diff[y] = data[y] - data_prev[y] : diff[y] = "nan"
             data_prev[y] = data[y]
+            printf("%g", diff[y])
+            if (y < max_nf)
+                printf("%s", OFS)
         }
     }
     printf("\n")