Try replacing sed with python, compare results

diracdeltas · diracdeltas · commit b9b5631ed6c5 · 2013-09-17T14:10:51.000-07:00
diff --git a/utils/merge-rulesets.py b/utils/merge-rulesets.py
@@ -38,6 +38,20 @@
 if os.path.isfile(rulesets_fn):
   os.system("chmod u+w " + rulesets_fn)
 
+def rulesize():
+  return len(open(rulesets_fn).read())
+
+def clean_up(rulefile):
+    """Remove extra whitespace and comments from ruleset library"""
+    comment_and_newline_pattern = re.compile(r"<!--.*?-->|\n|\r", flags=re.DOTALL)
+    rulefile = comment_and_newline_pattern.sub('', rulefile)
+    to_and_from_pattern = re.compile(r'\s*(to=|from=)')
+    rulefile = to_and_from_pattern.sub(r' \1', rulefile)
+    rulefile = re.sub(r">\s*<", r"><", rulefile)
+    rulefile = re.sub(r" />", r"/>", rulefile)
+    rulefile = re.sub(r"</ruleset>", r"</ruleset>\n", rulefile)
+    return rulefile
+
 library = open(rulesets_fn,"w")
 
 # XXX TODO replace all sed commands with native Python
@@ -51,29 +65,23 @@
   library.write('<rulesetlibrary>')
 
 # Include the filename.xml as the "f" attribute
+print("Removing whitespaces and comments...")
+
+crush = rulesize()
+
 for rfile in sorted(xml_ruleset_files):
   ruleset = open(rfile).read()
   fn=os.path.basename(rfile)
   ruleset = ruleset.replace("<ruleset", '<ruleset f="%s"' % fn, 1)
-  library.write(ruleset)
+  library.write(clean_up(ruleset))
 library.write("</rulesetlibrary>\n")
 library.close()
 
-print("Removing whitespaces and comments...")
-
-def rulesize():
-  return len(open(rulesets_fn).read())
+#sedcmd = ["sed", "-i", "-e", ":a", "-re"]
+#call(sedcmd + [r"s/<!--.*?-->//g;/<!--/N;//ba", rulesets_fn])
+#call(["sed", "-i", r":a;N;$!ba;s/\n//g;s/>[ 	]*</></g;s/[ 	]*to=/ to=/g;s/[ 	]*from=/ from=/g;s/ \/>/\/>/g", rulesets_fn])
+#call(["sed", "-i", r"s/<\/ruleset>/<\/ruleset>\n/g", rulesets_fn])
 
-def clean_up(rulefile):
-    """Remove extra whitespace and comments from ruleset library"""
-    comment_and_newline_pattern = re.compile(r"<!--.*?-->|\n|\r", flags=re.DOTALL)
-    rulefile = comment_and_newline_pattern.sub('', rulefile)
-
-crush = rulesize()
-sedcmd = ["sed", "-i", "-e", ":a", "-re"]
-call(sedcmd + [r"s/<!--.*?-->//g;/<!--/N;//ba", rulesets_fn])
-call(["sed", "-i", r":a;N;$!ba;s/\n//g;s/>[ 	]*</></g;s/[ 	]*to=/ to=/g;s/[ 	]*from=/ from=/g;s/ \/>/\/>/g", rulesets_fn])
-call(["sed", "-i", r"s/<\/ruleset>/<\/ruleset>\n/g", rulesets_fn])
 print("Crushed", crush, "bytes of rulesets into", rulesize())
 
 try:
