From 4ace50435248d3ad12c526234074464623862ee9 Mon Sep 17 00:00:00 2001
From: alpertunga-bile <kingdom19online@hotmail.com>
Date: Wed, 28 Aug 2024 10:08:33 +0300
Subject: [PATCH] preprocess is updated

---
 preprocess.py  | 18 ++++++------------
 pyproject.toml |  2 +-
 2 files changed, 7 insertions(+), 13 deletions(-)

diff --git a/preprocess.py b/preprocess.py
index 57cb0f8..e2a0f67 100644
--- a/preprocess.py
+++ b/preprocess.py
@@ -4,7 +4,7 @@
 
 def remove_exact_keywords(line: str) -> list[str]:
     find_empty_parantheses_regex = compile(r"\(\s*\)")
-    remove_nonprompts_regex = compile(r"[^a-zA-Z()\[\]{}]*")
+    remove_nonprompts_regex = compile(r"[^a-zA-Z()_\-\[\]{}]*")
     remove_nonweighters_regex = compile(r"[()\[\]{}]*")
     remove_inside_regex = compile(r"[^()\[\]{}]*")
 
@@ -20,7 +20,7 @@ def get_unique_list(sequence: list[str]) -> list[str]:
     prompts = get_unique_list(line.split(","))
 
     pure_prompts = OrderedDict()  # order matters, it contains prompts' original forms
-    extracted_pure_prompts = {}  # order isn't important, it contains prompt keywords
+    extracted_pure_prompts = set()  # order isn't important, it contains prompt keywords
 
     # remove exact keyword
     for prompt in prompts:
@@ -28,7 +28,7 @@ def get_unique_list(sequence: list[str]) -> list[str]:
             "", prompt
         ).lstrip()  # from -> ((masterpiece:1.2)) | to -> ((masterpiece))
 
-        if tempPrompt == "":
+        if len(tempPrompt) == 0:
             continue
 
         tempPrompt = remove_nonweighters_regex.sub(
@@ -42,8 +42,7 @@ def get_unique_list(sequence: list[str]) -> list[str]:
 
             if (
                 len(find_empty_parantheses_regex.findall(tempPrompt))
-                > 0  # find () count
-                or tempPrompt == ""
+                > 0  # find balanced parantheses count
             ):
                 # check balanced parantheses
                 inner_parant_count = tempPrompt.count("(")
@@ -67,12 +66,8 @@ def get_unique_list(sequence: list[str]) -> list[str]:
 
             continue
 
-        if tempPrompt == "":
-            tempPrompt = remove_nonprompts_regex.sub("", prompt).lstrip()
-            pure_prompts[tempPrompt] = True
-        else:
-            extracted_pure_prompts[tempPrompt] = True
-            pure_prompts[prompt] = True
+        extracted_pure_prompts.add(tempPrompt)
+        pure_prompts[prompt] = True
 
     return pure_prompts.keys()
 
@@ -126,7 +121,6 @@ def preprocess(line: str) -> str:
     temp_line = ", ".join(remove_exact_keywords(temp_line))
 
     temp_line = fix_commas(temp_line)
-
     temp_line = fix_artifacts(temp_line)
 
     temp_line = remove_emptyprompts_regex.sub(
diff --git a/pyproject.toml b/pyproject.toml
index 6860f88..a695ca2 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,7 +1,7 @@
 [project]
 name = "image-caption-comfyui"
 description = "Using image caption models to extract prompts in ComfyUI"
-version = "1.0.8"
+version = "1.0.9"
 license = { text = "MIT License" }
 dependencies = [
   "transformers",