Split out size formatting

format-size

@@ -0,0 +1,37 @@
+#!/bin/bash
+# Format size in bytes into a readable string of the form "1.23 KiB"; reads from stdin and arguments and can take multiple numbers separated by any amount of whitespace including newlines
+{
+	if [ ! -t 0 ]; then cat; fi
+	echo "$@"
+} |
+	tr '\t ' '\n' |
+	awk '
+		BEGIN {
+			units[0] = "B";
+			units[1] = "KiB";
+			units[2] = "MiB";
+			units[3] = "GiB";
+			units[4] = "TiB";
+			units[5] = "PiB";
+		}
+
+		{
+			size += $1;
+		}
+
+		END {
+			if (size > 0) {
+				magnitude = int(log(size) / log(1024));
+				if (magnitude > 5) {
+					magnitude = 5;
+				}
+			} else {
+				magnitude = 0;
+			}
+			if (magnitude > 0) {
+				sizeformat = "%.2f";
+			} else {
+				sizeformat = "%d";
+			}
+			printf sizeformat " %s\n", size / (1024 ^ magnitude), units[magnitude];
+		}'

warc-size

@@ -1,3 +1,3 @@
 #!/bin/bash
 # Total size of all WARCs in the current directory (or subdirectories)
-find -name '*.warc.gz' -printf '%s\n' | awk 'BEGIN { units[0] = "B"; units[1] = "KiB"; units[2] = "MiB"; units[3] = "GiB"; units[4] = "TiB"; units[5] = "PiB"; }  { size += $1 }  END { if (size > 0) { magnitude = int(log(size) / log(1024)); if (magnitude > 5) { magnitude = 5; } } else { magnitude = 0; } if (magnitude > 0) { sizeformat = "%.2f"; } else { sizeformat = "%d"; }  printf sizeformat " %s\n", size / (1024 ^ magnitude), units[magnitude]; }'
+find -name '*.warc.gz' -printf '%s\n' | "$(cd "$(dirname "$0")"; pwd -P)/format-size"