Add files via upload

Linyoux · web-flow · commit d3d577e741c7 · 2024-11-21T23:24:49.000+08:00
diff --git a/README.md b/README.md
@@ -0,0 +1,26 @@
+# VGal
+#### 介绍：
+本项目主要用于实现静态视频播放的体验优化。通过视频分段播放，获得近似视觉小说的体验。
+
+#### 注意：
+请勿用于非法用途，
+请勿用于已正常移植的游戏，
+请勿未经汉化组允许而录屏其译的游戏，
+请勿未经录屏者允许而用其录屏。
+
+#### 建议：
+通过magpie缩放，让录屏更高清。
+
+#### 原理：
+PC（供给端）：负责录屏，处理分割点。通过OCR识别视频各位置文本，进而获取视频分割点。
+
+安卓（使用端）：通过分割点，实现视频分段播放（点击屏幕，跳转到下一段视频，也就是下一段话），获得类似移植版的体验。
+
+#### 安装：
+
+> pip uninstall opencv-python opencv-python-headless  
+> pip install opencv-contrib-python easyocr Pillow difflib json torch torchvision torchaudio
+
+在C:\Users\你的用户名\.EasyOCR\model中，安装craft_mlt_25k，english_g2，zh_sim_g2的PTH文件
+
+安卓源码：https://github.com/Linyoux/VGal/tree/main/Android/app/src/main/java/client/vgal
diff --git a/confirmRegion.py b/confirmRegion.py
@@ -0,0 +1,89 @@
+import time
+import tkinter as tk
+from PIL import Image, ImageTk
+
+import videoprocess
+
+file_path = ""
+region = None
+root = None
+
+intervalEntry = None
+similarityEntry = None
+
+def button1_action():
+    global intervalEntry,similarityEntry,root
+
+    interval = float(intervalEntry.get())
+    similarityEntry = float(similarityEntry.get())
+
+    root.destroy()
+    start_time = time.time()  # 获取当前时间戳
+    text_groups = videoprocess.process_video(file_path,region,interval,similarityEntry)
+
+
+    if "/" in file_path:
+        file_name = file_path[file_path.rindex("/") + 1:]
+    else:
+        file_name = file_path
+    videoprocess.write_to_script(file_name,text_groups,"start.vgs")
+
+    end_time = time.time()  # 获取当前时间戳
+    elapsed_time = end_time - start_time  # 计算耗时
+    print("处理完成")
+    print(f"耗时: {elapsed_time:.4f} 秒")
+
+def reset_action():
+    exit()
+
+
+def show_confirmWindow(tknode,crop_area,file):
+    global file_path,region,root,similarityEntry,intervalEntry
+    region = crop_area
+    file_path = file
+    root = tk.Toplevel(tknode)
+    root.title("区域确认")
+
+    # 加载图片对象
+
+    image_path = "current_frame.jpg"  # 替换为你的图片路径
+    pil_image = Image.open(image_path)
+    cropped_image = pil_image.crop(crop_area)
+    tk_image = ImageTk.PhotoImage(cropped_image)
+    #
+    # # 创建标签显示图片
+    image_label = tk.Label(root, image=tk_image)
+    image_label.grid(row=0, column=0, columnspan=2, padx=20, pady=20)
+
+    label = tk.Label(root, text="处理间隔（单位：秒）:")
+    label.grid(row=1, column=0, pady=10 )  # 添加一些垂直间距
+
+    # 创建一个输入框
+    default_value = tk.StringVar()
+    default_value.set("0.5")  # 设置默认值
+    intervalEntry = tk.Entry(root,textvariable=default_value)
+    intervalEntry.grid(row=1, column=1, pady=10)
+
+
+    label2 = tk.Label(root, text="文本相似度:")
+    label2.grid(row=2, column=0, pady=10 )  # 添加一些垂直间距
+
+    default_value = tk.StringVar()
+    default_value.set("0.6")  # 设置默认值
+    # 创建一个输入框
+    similarityEntry = tk.Entry(root,textvariable=default_value)
+    similarityEntry.grid(row=2, column=1, pady=10 )
+
+    # 创建按钮
+    button1 = tk.Button(root, text="确定",command=button1_action)
+    button1.grid(row=3, column=0, pady=10 )
+
+    button2 = tk.Button(root, text="取消", command=reset_action)
+    button2.grid(row=3, column=1, pady=10)
+
+    # 调整行和列的权重，使组件能够在水平和垂直方向上平均分布
+    root.rowconfigure(0, weight=1)
+    root.columnconfigure(0, weight=1)
+    root.columnconfigure(1, weight=1)
+
+    root.mainloop()
diff --git a/python依赖安装.bat b/python依赖安装.bat
@@ -0,0 +1,2 @@
+pip uninstall opencv-python opencv-python-headless
+pip install opencv-contrib-python easyocr Pillow difflib json torch torchvision torchaudio
diff --git a/videoprocess.py b/videoprocess.py
@@ -0,0 +1,147 @@
+# encoding:utf-8
+import re
+
+import cv2
+import easyocr
+from difflib import SequenceMatcher
+import json
+
+class TextGroup:
+    def __init__(self):
+        self.texts = []  # 文本列表
+        self.timestamps = []  # 时间戳列表
+
+
+def calculate_similarity(str1, str2):
+    """计算两个字符串的相似度"""
+    return SequenceMatcher(None, str1, str2).ratio()
+
+def filter(text):
+    japanese_pattern = r'[\u3040-\u309F\u30A0-\u30FF\u4E00-\u9FA5]'
+    chinese_pattern = r'[\u4e00-\u9fa5]'
+
+    newtext = re.sub(japanese_pattern, "", text)
+    newtext = re.sub(chinese_pattern, "", newtext)
+
+    return text != newtext
+
+def format_timestamp(seconds):
+    """将秒数转换为 HH:MM:SS.milliseconds 格式"""
+    hours = int(seconds // 3600)
+    minutes = int((seconds % 3600) // 60)
+    secs = int(seconds % 60)
+    millis = int((seconds - int(seconds)) * 1000)  # 获取毫秒部分
+    return f"{hours:02d}:{minutes:02d}:{secs:02d}.{millis:03d}"
+
+
+def process_video(video_path, region, interval=0.5, similarity_threshold=0.6):
+    """处理视频并提取文本"""
+    reader = easyocr.Reader(['ch_sim', 'en'],download_enabled=False)
+
+    cap = cv2.VideoCapture(video_path)
+    fps = cap.get(cv2.CAP_PROP_FPS)
+    total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
+
+    text_groups = []
+    last_text = ""
+
+    interval_frames = int(fps * interval)
+
+    for frame_num in range(0, total_frames, interval_frames):
+        current_timestamp = frame_num / fps
+        cap.set(cv2.CAP_PROP_POS_FRAMES, frame_num)
+        print(f"正在处理第{frame_num}/{total_frames}帧")
+
+
+
+        ret, frame = cap.read()
+        if not ret:
+            break
+
+        roi = frame[region[1]:region[3],
+              region[0]:region[2]]
+
+        results = reader.readtext(roi)
+        current_text = " ".join([result[1] for result in results])
+
+        if not current_text or not filter(current_text):
+            continue
+
+        if not text_groups:
+            new_group = TextGroup()
+            new_group.texts.append(current_text)
+            new_group.timestamps.append(current_timestamp)
+            text_groups.append(new_group)
+            last_text = current_text
+            continue
+
+        similarity = calculate_similarity(last_text, current_text)
+
+        if similarity >= similarity_threshold:
+            text_groups[-1].texts.append(current_text)
+            text_groups[-1].timestamps.append(current_timestamp)
+        else:
+            new_group = TextGroup()
+            new_group.texts.append(current_text)
+            new_group.timestamps.append(current_timestamp)
+            text_groups.append(new_group)
+
+        last_text = current_text
+        #
+        # if frame_num > 50:
+        #     break
+
+    cap.release()
+
+    return text_groups
+
+def write_to_script(videoName,text_groups, output_file):
+    with open(output_file,"w",encoding="utf-8") as f:
+        f.write("play " + videoName)
+        f.write("\n")
+        f.write("proc")
+        f.write("\n")
+        f.write("\n")
+
+        for i, group in enumerate(text_groups, 1):
+            time = round(group.timestamps[-1],2)
+            text = group.texts[-1]
+
+            f.write("text " + text)
+            f.write("\n")
+            f.write("time " + str(time))
+            f.write("\n")
+            f.write("proc\n\n")
+
+
+
+def write_to_file(text_groups, output_file):
+    dats = []
+    """将文本分组写入文件"""
+    with open(output_file, 'w', encoding='utf-8') as f:
+        for i, group in enumerate(text_groups, 1):
+            f.write(f"Group {i}:\n")
+            for j, (text, timestamp) in enumerate(zip(group.texts, group.timestamps)):
+                f.write(f"  Timestamp: {format_timestamp(timestamp)} - Text: {text}\n")
+            f.write("\n")
+
+            dats.append({
+                "time": round(group.timestamps[-1],2) -0.05,
+                "text": group.texts[-1]
+            })
+        with open("script.json","w",encoding="utf-16") as f:
+            json.dump(dats,f,indent=4,ensure_ascii=False)
+
+def main():
+    video_path = 'b.mp4'  # 替换为你的视频路径
+    output_file = 'output.txt'  # 输出文件名
+    region = [00, 0, 2560, 496]  # 示例区域，根据实际需求调整
+
+    text_groups = process_video(video_path, region,0.5)
+    write_to_file(text_groups, output_file)
+
+    print(f"结果已写入 {output_file}")
+
+
+if __name__ == "__main__":
+    main()
diff --git a/videoselect.py b/videoselect.py
@@ -0,0 +1,112 @@
+import cv2
+import tkinter as tk
+from tkinter import filedialog
+import confirmRegion
+import sys
+rect_start = None
+rect_end = None
+drawing = False
+paused = False
+current_frame = None
+root = None
+running = False
+file_path = None
+
+def select_video():
+    global root,file_path
+    root = tk.Tk()
+    root.withdraw()  # 隐藏主窗口
+    file_path = filedialog.askopenfilename(filetypes=[("Video Files", "*.mp4 *.avi *.mov *.mkv")])
+    if file_path:
+        preview_video(file_path)
+    root.destroy()
+
+
+def draw_rectangle(event, x, y, flags, param):
+    global rect_start, rect_end, drawing, current_frame, running, file_path
+
+    if event == cv2.EVENT_LBUTTONDOWN:
+        drawing = True
+        rect_start = (x, y)
+        rect_end = rect_start
+
+    elif event == cv2.EVENT_MOUSEMOVE:
+        if drawing:
+            # 使用当前帧的副本来绘制矩形
+            temp_frame = current_frame.copy()
+            rect_end = (x, y)
+            cv2.rectangle(temp_frame, rect_start, rect_end, (255, 0, 0), 2)
+            cv2.imshow("Video Preview", temp_frame)
+
+    elif event == cv2.EVENT_LBUTTONUP:
+        drawing = False
+        rect_end = (x, y)
+
+        # 排序坐标
+        x1, y1 = rect_start
+        x2, y2 = rect_end
+        left = min(x1, x2)
+        right = max(x1, x2)
+        top = min(y1, y2)
+        bottom = max(y1, y2)
+
+        # 更新为排序后的区域
+        rect_start = (left, top)
+        rect_end = (right, bottom)
+
+        cv2.rectangle(current_frame, rect_start, rect_end, (255, 0, 0), 2)
+        cv2.imshow("Video Preview", current_frame)
+
+        # 选择完区域后直接退出程序
+        if rect_start and rect_end:
+            print("选择的区域坐标：", rect_start, rect_end)
+            cv2.imwrite('current_frame.jpg', current_frame)
+            cv2.destroyAllWindows()  # 关闭窗口
+            running = False
+            confirmRegion.show_confirmWindow(root, (left, top, right, bottom), file_path)
+            exit()  # 退出程序
+
+
+def preview_video(file_path):
+    global current_frame, paused, running
+    cap = cv2.VideoCapture(file_path)
+
+    if not cap.isOpened():
+        print("Error opening video file")
+        return
+
+    # 创建一个全屏窗口
+    cv2.namedWindow("Video Preview", cv2.WND_PROP_FULLSCREEN)
+    cv2.setWindowProperty("Video Preview", cv2.WND_PROP_FULLSCREEN, cv2.WINDOW_FULLSCREEN)
+    cv2.setMouseCallback("Video Preview", draw_rectangle)
+    running = True
+
+    while running:
+        if not paused:
+            ret, current_frame = cap.read()
+            if not ret:
+                cap.set(cv2.CAP_PROP_POS_FRAMES, 0)  # 重置视频到开头
+                continue
+
+        # 显示视频帧
+        cv2.imshow("Video Preview", current_frame)
+
+        # 处理键盘输入
+        key = cv2.waitKey(25) & 0xFF
+        if key == 27:  # ESC键
+            break
+        elif key == 32:  # 空格键
+            paused = not paused
+            print("暂停状态：", "已暂停" if paused else "已恢复")
+
+        # 检查窗口是否关闭
+        if cv2.getWindowProperty("Video Preview", cv2.WND_PROP_VISIBLE) < 1:
+            break
+
+    cap.release()
+    cv2.destroyAllWindows()
+
+
+
+if __name__ == "__main__":
+    select_video()
diff --git a/开始处理.bat b/开始处理.bat
@@ -0,0 +1 @@
+python videoselect.py

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+pip uninstall opencv-python opencv-python-headless`
	`2`	`+pip install opencv-contrib-python easyocr Pillow difflib json torch torchvision torchaudio`