Add script to compute number of files per file type per package

kgryte · kgryte · commit 7dfe517a17a3 · 2017-10-09T21:04:56.000-07:00
diff --git a/tools/git/scripts/num_files_per_file_type_per_pkg b/tools/git/scripts/num_files_per_file_type_per_pkg
@@ -0,0 +1,57 @@
+#!/usr/bin/env bash
+#
+# Prints the number of files in a package.
+#
+# <package_path> <percentage> <total> <file_type>
+#
+# Notes:
+#
+# * For files using alternative filename extensions, no attempt is made to infer the file type. Unrecognized filename extensions are categorized as "other".
+# * `README.md` files are included in Markdown totals.
+
+# Determine root directory:
+root="$(git rev-parse --show-toplevel)"
+
+# Define the path to a utility to list packages:
+find_pkgs="${root}/tools/scripts/find_packages"
+
+# Define the path to a utility to list files:
+find_files="${root}/tools/scripts/find_files"
+
+# Define the path to a utility to annotate a file statistic stream:
+annotate_file_type="${root}/tools/git/scripts/annotate_file_type.awk"
+
+# Get a list of packages:
+pkgs="$(PACKAGES_FILTER=${PACKAGES_FILTER} ${find_pkgs})"
+
+# For each package, determine the number of lines...
+for pkg in ${pkgs}; do
+	# * `find_files`
+	#   - Find package files.
+	# * `awk '{}'`
+	#   - Prefix each file to match annotation interface.
+	# * `annotate_file_type`
+	#   - Prefix each result with a file type.
+	# * `awk '{}'`
+	#   - Compute statistics.
+	# * `sort -k2nr`
+	#   - Sort in reverse numeric order based on the computed percentage.
+	FILES_PATTERN="${FILES_PATTERN}" FILES_FILTER="$pkg/.*" "${find_files}" | awk '{print 1 OFS $0}' | awk -f "${annotate_file_type}" | awk -v pkg="$pkg" '
+$3 ~ /README\.md$/ {
+	lines["README"] += 1
+}
+{
+	lines[$1] += 1
+	total += 1
+}
+
+END {
+	print pkg OFS 1 OFS total OFS "total"
+	for (k in lines) {
+		n = lines[k]
+		pct = int(n/total*10000)/10000
+		print pkg OFS pct OFS n OFS k
+	}
+}
+' | sort -k2nr
+done
