Compact JS resource better.

Source/web/scripts/make-file-arrays.py

Index: Source/web/scripts/make-file-arrays.py
diff --git a/Source/web/scripts/make-file-arrays.py b/Source/web/scripts/make-file-arrays.py
index 21ee6fbd26ef2e05e97136c710aae8b1e0e10bdf..1cf0b535fe5d9cd712e132b3095942b5b6165c8e 100755
--- a/Source/web/scripts/make-file-arrays.py
+++ b/Source/web/scripts/make-file-arrays.py
@@ -34,6 +34,8 @@ import re
 import sys
 from optparse import OptionParser
 
+# Should be spam in the build log?
+DEBUG_OUTPUT = False
 
 def make_variable_name_and_read(file_name):
     result = re.match(r'([\w\d_]+)\.([\w\d_]+)', os.path.basename(file_name))
@@ -46,6 +48,76 @@ def make_variable_name_and_read(file_name):
     return variable_name, content
 
 
+def is_identifier_char(c):
+    assert len(c) == 1
+    return c.isalnum() or c in "$_"
+
+
+def strip_js_whitespace_not_in_strings(content):
+    """Removes whitespace we don't need in this javascript
+    program. Whitespace is relevant inside strings and regexps (not
+    supported) and sometimes new-lines are needed to trigger the
+    automatic semi-colon insertion.
+
+    This is _slightly_ worse than Closure at stripping whitespace
+    but so close that there is no urgent reason to improve it."""
+
+    out = ""
+
+    quoted_char = False  # When True the next char is to be output as is.
+    string_delimiter = None  # When set we're inside a string.
+    lazy_space = False
+    lazy_new_line = False
+    for c in content:
+        if quoted_char:
+            quoted_char = False
+            out += c
+        elif string_delimiter:
+            out += c
+            if c == "\\":
+                quoted_char = True
+            elif c == string_delimiter:
+                string_delimiter = None
+        else:
+            if c == "\\":
+                quoted_char = True
+            elif c == '"' or c == "'":
+                string_delimiter = c
+
+            if c in " \t\n\r\f":
+                lazy_space = True
+                if c == "\n":
+                    lazy_new_line = True
+            else:
+                if lazy_space and out:
+                    output_space = False
+                    if is_identifier_char(out[-1]) and is_identifier_char(c):
+                        # Will prevent "var i" -> "vari" or
+                        # "function $" -> "function$"
+                        output_space = True
+                    elif out[-1] == c and c not in "{}()[]":
+                        # Will prevent "a - -b" -> "a--b" or "a--\n-b" -> "a---b"
+                        output_space = True
+                    if output_space:
+                        if lazy_new_line:
+                            out += "\n"
+                        else:
+                            out += " "
+                lazy_space = False
+                lazy_new_line = False
+                out += c
+    return out
+
+# Quick self testing. If this fails, we better not proceed with the build.
+assert strip_js_whitespace_not_in_strings('  var a = \'  \\"  \\"  \' ;') == 'var a=\'  \\"  \\"  \';'
+
+assert strip_js_whitespace_not_in_strings("a--  - \t--b") == "a-- - --b"
+
+assert strip_js_whitespace_not_in_strings("if (a == $) {\n}") == "if(a==$){}"
+assert strip_js_whitespace_not_in_strings("var \n a") == "var\na"
+assert strip_js_whitespace_not_in_strings("{  { a( (1+ 2) / 3);}\n} ") == "{{a((1+2)/3);}}"
+assert strip_js_whitespace_not_in_strings(r"  '  \'  '  ") == r"'  \'  '"
+
 def strip_whitespace_and_comments(file_name, content):
     result = re.match(r'.*\.([^.]+)', file_name)
     if not result:
@@ -60,20 +132,28 @@ def strip_whitespace_and_comments(file_name, content):
     leading_space = re.compile(r'^[ \t]+', re.MULTILINE)
     trailing_space = re.compile(r'[ \t]+$', re.MULTILINE)
     empty_line = re.compile(r'\n+')
+    orig_content = content
     if extension == 'js':
+        content = orig_content
         content = multi_line_comment.sub('', content)
         content = single_line_comment.sub('', content)
         content = trailing_comment.sub(r'\1', content)
-        content = repeating_space.sub(' ', content)
-        content = leading_space.sub('', content)
-        content = trailing_space.sub('', content)
-        content = empty_line.sub('\n', content)
+        content = strip_js_whitespace_not_in_strings(content)
+        optimized_content_content = content
+        content = orig_content
+        if len(optimized_content_content) < len(content):
+            content = optimized_content_content
     elif extension == 'css':
         content = multi_line_comment.sub('', content)
         content = repeating_space.sub(' ', content)
         content = leading_space.sub('', content)
         content = trailing_space.sub('', content)
         content = empty_line.sub('\n', content)
+    if DEBUG_OUTPUT:
+        in_size = len(orig_content)
+        out_size = len(content)
+        if out_size < in_size:
+            print("%s: Compaction saved %.1f%% for a total of %d bytes." % (file_name, float(100 * (in_size - out_size)) / in_size, in_size - out_size))
     return content
 
 
@@ -81,6 +161,10 @@ def process_file(file_name):
     variable_name, content = make_variable_name_and_read(file_name)
     content = strip_whitespace_and_comments(file_name, content)
     size = len(content)
+    if DEBUG_OUTPUT:
+        out_debug_name = file_name + ".generated"
+        with open(out_debug_name, "w") as out_debug_file:
+            out_debug_file.write(content)
     return variable_name, content