From cc30df8706a9ee47c0d88342da4de624fc7c68cc Mon Sep 17 00:00:00 2001
From: Alek Ratzloff <alekratz@gmail.com>
Date: Fri, 27 May 2022 18:41:17 -0700
Subject: [PATCH] Move markov nodes to be single strings

Separating strings by spaces is more memory-friendly than using tuples.

Signed-off-by: Alek Ratzloff <alekratz@gmail.com>
---
 plugins/markov.py | 25 ++++++++++++-------------
 1 file changed, 12 insertions(+), 13 deletions(-)

diff --git a/plugins/markov.py b/plugins/markov.py
index ebb7be5..eb707ba 100644
--- a/plugins/markov.py
+++ b/plugins/markov.py
@@ -21,7 +21,7 @@ def chain_inner_default() -> defaultdict[str | None, int]:
     return defaultdict(int)
 
 
-def chain_default() -> defaultdict[tuple[str, ...], defaultdict[str | None, int]]:
+def chain_default() -> defaultdict[str, defaultdict[str | None, int]]:
     return defaultdict(chain_inner_default)
 
 
@@ -51,11 +51,11 @@ class Chain:
             return
         self.__load()
         for fragment in windows(parts + [None], self.order + 1):
-            head = tuple(fragment[0:-1])
+            head = fragment[0:-1]
             tail = fragment[-1]
-            self.__cache[head][tail] += 1
+            self.__cache[" ".join(head)][tail] += 1
 
-    def get(self, key: tuple[str, ...]) -> dict[str | None, int]:
+    def get(self, key: str) -> dict[str | None, int]:
         self.__last_access = asyncio.get_running_loop().time()
         if self.__cache:
             if key in self.__cache:
@@ -67,7 +67,7 @@ class Chain:
             self.__load()
             return self.__cache[key]
 
-    def set(self, key: tuple[str, ...], value: Mapping[str | None, int]):
+    def set(self, key: str, value: Mapping[str | None, int]):
         self.__last_access = asyncio.get_running_loop().time()
         if not self.__cache:
             # Attempt the cache before writing to it
@@ -85,7 +85,7 @@ class Chain:
             self.__cache = defaultdict(
                 chain_inner_default,
                 {
-                    tuple(key.split(" ")): defaultdict(
+                    key: defaultdict(
                         int,
                         {
                             (None if not word else word): weight
@@ -107,7 +107,7 @@ class Chain:
         with open(self.path, "w") as fp:
             json.dump(
                 {
-                    " ".join(key): {
+                    key: {
                         ("" if word is None else word): weight
                         for word, weight in value.items()
                     }
@@ -131,17 +131,16 @@ class Chain:
 
         words: List[str] = []
 
-        node = random.choice(list(self.__cache.keys()))
+        node = random.choice(list(self.__cache.keys())).split(" ")
         words += node
-        next: str | None = self.choose_next(node)
+        next: str | None = self.choose_next(" ".join(node))
         while next:
             words += [next]
-            node = (*node[1:], next)
-            next = self.choose_next(node)
-
+            node = [*node[1:], next]
+            next = self.choose_next(" ".join(node))
         return " ".join(words)
 
-    def choose_next(self, head: tuple[str, ...]) -> str | None:
+    def choose_next(self, head: str) -> str | None:
         self.__load()
         choices = self.__cache[head]
         words = list(choices.keys())