add script to detect dead code

requirements/edx/development.in

@@ -17,6 +17,7 @@ edx-sphinx-theme                    # Documentation theme
 pyinotify                           # More efficient checking for runserver reload trigger events
 snakefood                           # Lists dependencies between Python modules, used in scripts/dependencies/*
 sphinx                              # Developer documentation builder
+vulture                             # Detects possible dead/unused code, used in scripts/find-dead-code.sh
 
 # Performance timer used in modulestore/perf_tests/test_asset_import_export.py
 -e git+https://github.com/doctoryes/code_block_timer.git@f3d0629f086bcc649c3c77f4bc5b9c2c8172c3bf#egg=code_block_timer

requirements/edx/development.txt

@@ -334,6 +334,7 @@ urlobject==2.4.3
 user-util==0.1.3
 virtualenv==16.0.0
 voluptuous==0.11.1
+vulture==0.26
 w3lib==1.19.0
 watchdog==0.8.3
 web-fragments==0.2.2

scripts/vulture/find-dead-code.sh

+#!/usr/bin/env bash
+set -e
+
+############################################################################
+#
+# find-dead-code.sh
+#
+# This script analyzes the edx-platform for dead code instances using
+# vulture, a static analysis tool that ranks potential unused code using
+# the following confidence scores:
+# - 100% = completely unreachable code: code that follows return, continue
+#          or raise statements, or code that cannot logically be run, i.e.
+#          a condition that can never be True
+# - 90%  = unused imports within a file
+# - 60%  = unused code: code that has no reference in the code base, aside
+#          from definition
+#
+# This script will output a list of dead-code instances with their
+# confidence scores, in descending order of the size of the instance
+# (in lines). This in turn can be used to clean up the edx-platform.
+# However, you shouldn't automatically delete code reported back from
+# vulture. It can report false positives. Often it's not the case that
+# code is truly dead, but has been referenced incorrectly (spelling)
+# elsewhere in the code base, most likely in a refactor. Another example
+# of code that will be interpreted as dead by vulture is anything that
+# is used in a template (which is not considered in the analysis).
+#
+# Therefore, the results of this script should be used as a jumping off
+# point for investigating potential dead code removal
+#
+############################################################################
+
+OUTPUT_DIR="reports/vulture"
+mkdir -p ${OUTPUT_DIR}
+OUTPUT_FILE=${OUTPUT_DIR}/vulture-report.txt
+echo '' > $OUTPUT_FILE
+# exclude test code from analysis, as it isn't explicitly called by other
+# code. Additionally, application code that is only called by tests 
+# should be considered dead
+EXCLUSIONS='/test,/acceptance,cms/envs,lms/envs,/terrain,migrations/,signals.py'
+MIN_CONFIDENCE=90
+# paths to the code on which to run the analysis
+CODE_PATHS='cms common lms openedx'
+WHITELIST_PATH="$(dirname "${BASH_SOURCE[0]}")/whitelist.py"
+echo "Checking for dead code in the following paths: $CODE_PATHS"
+echo "Results can be found in $OUTPUT_FILE"
+vulture $CODE_PATHS $WHITELIST_PATH --exclude $EXCLUSIONS \
+--min-confidence $MIN_CONFIDENCE \
+--sort-by-size |tac > $OUTPUT_FILE

scripts/vulture/whitelist.py

+#!/bin/env python
+
+# Vulture often detects false positives when analyzing a code
+# base. If there are particular things you wish to ignore,
+# add them below. This file is consumed by
+# scripts/dead_code/find-dead-code.sh
+
+from vulture.whitelist_utils import Whitelist
+
+view_whitelilst = Whitelist()
+
+# Example:
+# view_whitelist.name_of_function_to_whitelist