Tree shaking POC for Elm

shake_elm_tree.py

import os
import sys
import re
import string


expressions = []
# elm produces nice output where each complete top level expression starts with
# a line with zero indentation and continues up until the next line with zero
# indentation

current_expression = { "type": "other", "body": [] }
inside_var = False

# collect the entire program into a single string for full text search
body = ""
# identify each top level variable name
global_bindings = []

for line in sys.stdin.readlines():
    body += line
    if re.match("^var", line):
        expressions.append(current_expression)
        match = re.match("^var\s([^=\s]+)\s=", line)
        var = None
        if match:
            var = match.group(1)
            global_bindings.append(var) 
        current_expression = { "type": "var", "binding": var, "body": [ line ] }
        inside_var = True
    else:
        if inside_var:
            # check for termination of the var
            if re.match("^\s+", line) or re.match(".*;\s*$", line):
                current_expression["body"].append(line)
            else:
                inside_var = False
                expressions.append(current_expression)
                current_expression = { "type": "other", "body": [ line ] }
                
        else:
            current_expression["body"].append(line)

# the last expression ends at the end of the document
expressions.append(current_expression)


shake_off = set()
for var in global_bindings: 
    if string.count(body, var) == 1:
        shake_off.add(var)


for expr in expressions:
    if expr["type"] == "var":
        if expr["binding"] and expr["binding"] in shake_off:
            for line in expr["body"]:
                sys.stdout.write("//" + line)
        else: 
            for line in expr["body"]:
                sys.stdout.write(line)

    else:
        for line in expr["body"]:
            sys.stdout.write(line)

Note that the script as written only comments out the code it shakes off so you can visually inspect what it thinks is not needed. Don't be surprised if the size of the output increases if you run this script unmodified.