vim syntax html creation is now cached (and should be secure)

pylog_action_impl.py

@@ -1,5 +1,5 @@
-import os, os.path, sys
+import os, os.path, stat, sys
-import urllib
+import urllib, hashlib
 from datetime import datetime
 import pylog_config as cfg
 import subprocess,tempfile
@@ -41,8 +41,9 @@ def disp_txt(path):
 class BodyParser(HTMLParser):
     body=False
     src=''
+    blacklist = ['script']
     def handle_starttag(self, tag, attrs):
-        if self.body:
+        if self.body and not(tag in self.blacklist):
             self.src+='<'+tag
             for attr in attrs:
                 self.src+=' '+attr[0]+'="'+attr[1]+'"'
@@ -52,7 +53,7 @@ class BodyParser(HTMLParser):
     def handle_endtag(self, tag): 
         if tag=='body':
             self.body=False
-        if self.body:
+        if self.body and not(tag in self.blacklist):
             self.src+='</'+tag+'>'
     def handle_data(self, data):
         if self.body:
@@ -65,7 +66,26 @@ class BodyParser(HTMLParser):
 
 
 def get_source_html(path):
-    tmpfile=tempfile.mkstemp(prefix="pylog")[1]
+    tmpdir = os.path.join(tempfile.gettempdir(),"pylog")
+    cache=False
+    if not os.path.exists(tmpdir):
+        os.mkdir(tmpdir,0700)
+        
+    if os.path.exists(tmpdir):
+        s=os.stat(tmpdir)
+        if s.st_mode & stat.S_IRWXU == stat.S_IRWXU and \
+          s.st_mode & stat.S_IRWXG == 0 and \
+          s.st_mode & stat.S_IRWXO == 0 and \
+          os.path.isdir(tmpdir):
+            f=open(path,'r')
+            tmpfile=os.path.join(tmpdir,hashlib.sha256(f.read()).hexdigest())
+            f.close()
+            cache=True
+            
+    if not cache:
+        tmpfile=tempfile.mkstemp(prefix="pylog",dir=tmpdir)[1]
+    
+    if not cache or (cache and not os.path.exists(tmpfile)):
         #we use a fork of vim to generate a nice syntax highlighted html file
         #TODO this requires further testing
         p=subprocess.Popen(["/usr/bin/vim",path,\
@@ -79,6 +99,7 @@ def get_source_html(path):
             "+q!",\
             "+q!"],\
             stdout=subprocess.PIPE,stderr=subprocess.PIPE)
+            
         p.stdout.close()
         p.stderr.close()
         p.wait() #wait for child to terminate
@@ -88,7 +109,8 @@ def get_source_html(path):
     bp.feed(f.read())
     src = bp.src
     f.close()   
-    #delete tmp file
+    #delete tmp file but only if it's not a secure cached file
+    if not cache:
         os.remove(tmpfile)
     return src