merged roels unicsss.py code

etherdump

@@ -1,4 +1,7 @@
 #!/usr/bin/env python
+# License: AGPL
+#
+
 from __future__ import print_function
 # stdlib
 import json, sys, os, re
@@ -47,9 +50,10 @@ p.add_argument("--showurls", default=False, action="store_true", help="flag to d
 p.add_argument("--hidepaths", default=False, action="store_true", help="flag to not display paths")
 p.add_argument("--pretend", default=False, action="store_true", help="flag to not actually save")
 p.add_argument("--add-images", default=False, action="store_true", help="flag to add image tags")
+p.add_argument("--authors-css", default="authors.css", help="filename to save collected authorship css (nb: etherdump will overwrite this file!)")
 
 # TODO css from pad --- ie specify a padid for a stylesheet!!!!!!
-p.add_argument("--css", default="styles.css", help="padid of stylesheet")
+# p.add_argument("--css", default="styles.css", help="padid of stylesheet")
 
 
 args = p.parse_args()
@@ -77,6 +81,7 @@ todo = args.padid
 done = set()
 count = 0
 data = {}
+authors_css_rules = {}
 data['apikey'] = info['apikey']
 
 if args.allpads:
@@ -245,7 +250,15 @@ while len(todo) > 0:
         # extract the style tag (with authorship colors)
         style = t.find(".//style")
         if style != None:
-            style = ET.tostring(style, method="html")
+            if args.authors_css:
+                for i in style.text.splitlines(): 
+                    if len(i):
+                        selector, rule = i.split(' ',1)
+                        authors_css_rules[selector] = rule
+                style = '' # strip the individual style tag from each page (only exports to authors-css file)
+                # nb: it's up to the template to refer to the authors-css file
+            else:
+                style = ET.tostring(style, method="html")
         else:
             style = ""
         # and extract the contents of the body
@@ -273,4 +286,10 @@ while len(todo) > 0:
         if args.limit and count >= args.limit:
             break
     except TypeError:
-        print ("ERROR, skipping!", file=sys.stderr)
+        print ("ERROR, skipping!", file=sys.stderr)
+
+# Write the unified CSS with authors
+if args.authors_css:
+    with open(args.authors_css, 'w') as css:
+        for selector, rule in sorted(authors_css_rules.items()):
+            css.write(selector+' '+rule+'\n')