+
+ # Replace URLs with escape string for now, so that the Markov selector
+ # won't see them as different strings. Stash replaced URLs in urls.
+ urls = []
+ url_escape = "\nURL"
+ url_starts = ["http://", "https://", "<http://", "<https://"]
+ for i in range(len(tokens)):
+ for url_start in url_starts:
+ if tokens[i][:len(url_start)] == url_start:
+ length = len(tokens[i])
+ if url_start[0] == "<":
+ try:
+ length = tokens[i].index(">") + 1
+ except ValueError:
+ pass
+ urls += [tokens[i][:length]]
+ tokens[i] = url_escape + tokens[i][length:]
+ break
+
+ # For each snippet of select_length, use markov() to find continuation
+ # token from selections. Replace present users' names with malkovich.
+ # Start snippets with the beginning of a sentence, if possible.