Index: channels/bookmarks.py
==================================================================
--- channels/bookmarks.py
+++ channels/bookmarks.py
@@ -40,10 +40,11 @@
     listformat = "any"
     categories = ["favourite", ]  # timer, links, search, and links show up as needed
     finder_song = { "genre": "Youtube ", "format": "video/youtube", "playing": "current_", "title": "The Finder song", "url": "http://youtube.com/v/omyZy4H8y9M", "homepage": "http://youtu.be/omyZy4H8y9M" }
     streams = {"favourite":[finder_song], "search":[], "scripts":[], "timer":[], "history":[], }
     default = "favourite"
+    fixed_size = [32,24]
 
 
     # cache list, to determine if a PLS url is bookmarked
     urls = []
 

ADDED   contrib/scripts.py
Index: contrib/scripts.py
==================================================================
--- contrib/scripts.py
+++ contrib/scripts.py
@@ -0,0 +1,128 @@
+# encoding: UTF-8
+# api: streamtuner2
+# title: Script stations
+# description: User scripts for individual stations
+# type: feature
+# category: bookmark
+# version: 0.1
+# priority: theoretical
+#
+# This plugin provides for a simpler alternative to channel plugins.
+# It reads the ~./config/streamtuner2/script/ directory for script
+# files, and shows them in the bookmarks channel. Each script may
+# scan/uncover a station url at runtime.
+# Which obviously isn't meant for easily parseable stations, but for
+# the more difficult cases.
+#
+# There's support for python scripts obviously, but any executable
+# file (scripting language) can be run and queried for urls.
+#
+# Each is supposed to contain a meta comment block (much like this
+# plugin), except it's using more stream-oriented descriptors:
+#
+#   #!/bin/sh
+#   # title: Station title
+#   # description: Fetching a live stream
+#   # genre: jazz
+#   # homepage: http://example.org/
+#   #
+#   # Now normally, it would not just print something static.
+#   
+#   echo "http://example.org/.mp3"
+#
+# Obviously the purpose is more complicated extractions. Which is
+# why a .py script had acces to all ST2s parsing tools already.
+# 
+# This is implemented using the action.handler hooks for urn:
+# modules. But ensures the resolver script is run each time - by
+# not caching the final stream url.
+# Conversly this plugin prevents editing of script station entries.
+#
+
+
+import os, shutil, copy, subprocess, sys, StringIO
+import csv, zipfile
+import re, json, pq
+import ahttp
+import config
+from config import *
+import uikit
+from compat2and3 import *
+import action
+from channels import *
+
+
+# dynamic station extractors from ~/.config/streamtuner2/scripts/
+class scripts (object):
+
+    # plugin info
+    module = "scripts"
+    meta = plugin_meta()
+    parent = None
+    dir = conf.dir + "/scripts"
+
+    # register hooks
+    def __init__(self, parent):
+        if not os.path.exists(self.dir):
+            os.mkdir(self.dir)
+        self.parent = parent
+        self.bm = parent.bookmarks
+        action.handler["urn:script"] = self.urn_resolve
+        self.bm.add_category("scripts")
+        self.bm.category_plugins["scripts"] = self
+
+    # find script files and compile station dicts
+    def update_streams(self, cat):
+        r = []
+        for fn in os.listdir(self.dir):
+            meta = config.plugin_meta(fn=self.dir+"/"+fn)
+            r.append(dict(
+                genre = meta.get("genre", meta.get("type", "script")),
+                title = meta.get("title", fn),
+                playing = meta.get("description", meta.get("playing", "")),
+                homepage = meta.get("homepage", ""),
+                listeners = to_int(meta.get("listeners", "1")),
+                bitrate = to_int(meta.get("bitrate", "64")),
+                format = meta.get("format", "audio/mp3"),
+                listformat = "href",
+                file = fn,
+                url = "urn:script:" + fn
+            ))
+        return r
+
+    # run'em
+    def urn_resolve(self, row, *x):
+        if not row.get("file"):
+            return
+
+        # prepare
+        fn = "%s/%s" % (self.dir, row["file"])
+        row = copy.copy(row)
+        row["url"] = ""
+        output = None
+
+        # executable
+        if os.path.isfile(fn) and os.access(fn, os.X_OK):
+            f = subprocess.Popen([fn], stdout=subprocess.PIPE)
+            output, err = f.communicate()
+        # plain python script
+        elif re.match("^[\w+-]\.py$"):
+            real_stdout, sys.stdout = sys.stdout, StringIO.StringIO()
+            execfile(fn)
+            output, sys.stdout = sys.stdout.getvalue(), real_stdout
+        # none
+        else:
+            return
+
+        # extract urls
+        if output:
+            urls = re.findall("(\w+://\S+)", output)
+            if urls:
+                row["url"] = urls[0] # action module does not currently support multi-urls
+        return row
+        # Unlike other .resolve_urn() handlers this one returns a copy,
+        # does not modify the passed dict. And this is only handled in
+        # action.run_fmt_url(), but not by GenericChannel.play(). Why
+        # OTOH this incudes a *double script invocation*.
+
+