first commit

main.py

@@ -0,0 +1,125 @@
+# HOW TO USE
+# 1. Install Tesseract (https://github.com/tesseract-ocr/tesseract/releases/)
+# 2. Install required libraries (see in the code below)
+# 3. Put all your Opus Magnum-generated gifs in a folder
+# 4. Facultative: if you are importing your gifs automatically from Discord using DiscordChatExporter, also save the CSV data file
+# 5. Facultative: if you want to make an automatic correspondence between the usernames and desired display names, you can create a username.csv file with 2 columns: usename and name
+# 6. Change the settings in the code below according to your needs
+# 7. If everything goes well, the results should be output on the console as well as in a csv file.
+
+
+# Import required packages
+import cv2
+import pytesseract
+from PIL import Image
+import os
+import pandas as pd
+
+
+### SETTINGS
+
+# Mention the installed location of Tesseract-OCR in your system
+# pytesseract.pytesseract.tesseract_cmd = 'C:/Program Files/Tesseract-OCR/tesseract.exe'
+pytesseract.pytesseract.tesseract_cmd = '/usr/bin/tesseract'
+
+# Insert the path of the folder containing the GIFs here
+gifs_path = "gifs_opus_magnum"
+
+# Do you want to include the Discord usernames in the output (requires Discord messages data file)? If False, will use name of file instead
+use_discord_data = False
+# Insert the path of the CSV file containing Discord messages data
+discord_data_path = "EvLan - EvLan 2 - solutions-opus-magnum [1260715259514327070].csv"
+
+# Do you want to include the real names of the participants (requires username correspondence file)?
+use_real_names = False
+# Insert the path of the CSV file containing username -> name correspondence
+usernames_data_path = "usernames.csv"
+
+# Insert here the desired path for the output CSV file
+output_path = "results_opus_magnum.csv"
+
+### END OF SETTINGS
+
+
+results = pd.DataFrame(columns=['username','name','puzzle','cost','cycles','area','notes'])
+
+filenames =  os.listdir(gifs_path)
+
+if use_discord_data:
+    discord_data_df = pd.read_csv(discord_data_path)
+if use_real_names:
+    usernames_data_df = pd.read_csv(usernames_data_path)
+
+for filename in filenames:
+    img_path = os.path.join(gifs_path, filename)
+
+    # Convert GIF to JPG
+    with Image.open(img_path) as img:
+        width, height = img.size
+        img.seek(0)
+        rgb_img = img.convert("RGB")
+        rgb_img.save("temp.jpg", "JPEG")
+
+    # Read image from which text needs to be extracted
+    img = cv2.imread("temp.jpg")
+
+    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
+
+    # Optional: resize for better OCR
+    gray = cv2.resize(gray, None, fx=1, fy=1, interpolation=cv2.INTER_CUBIC)
+
+    # Manually crop regions based on known layout (x, y, w, h)
+    regions = [
+        (15, 600, 330, 28), # PUZZLE NAME
+        (412, 603, 65, 22), # COST
+        (577, 603, 65, 22), # CYCLES
+        (739, 603, 65, 22) # AREA
+    ]
+
+    output = img.copy()
+    
+    # "Username", "name" and "notes" fields are filled in in this section
+    username = filename
+    notes = ""
+    if use_discord_data:
+        for _, row in discord_data_df.iterrows():
+            attachments = row['Attachments'].split(',')
+            for attachment in attachments:
+                if filename == attachment.split('\\')[1]:
+                    username = row['Author']
+                    notes = row['Content']
+    if use_real_names:
+        name = usernames_data_df.loc[username == usernames_data_df['username']].iloc[0]['name']
+    else:
+        name = ""
+
+    def find_text(dims, gray, output, content):
+        x, y, w, h = dims
+        roi = gray[y:y+h, x:x+w]
+        roi = cv2.bitwise_not(roi)
+        if content == 'digits' or content == 'digits_with_6':
+            config = "--oem 3 --psm 7 -c tessedit_char_whitelist=0123456789"
+        else:
+            config = "--oem 3 --psm 7"
+        text = pytesseract.image_to_string(roi, config=config).strip()
+        # Remove the extra 6 (actually the G for Gold) for cost value 
+        if content == 'digits_with_6':
+            text = text[:-1]
+        cv2.rectangle(output, (x, y), (x+w, y+h), (0, 255, 0), 2)
+        return text
+    
+    puzzle = find_text(regions[0], gray, output, 'letters')
+    cost = find_text(regions[1], gray, output, 'digits_with_6')
+    cycles = find_text(regions[2], gray, output, 'digits')
+    area = find_text(regions[3], gray, output, 'digits')
+    
+    results.loc[len(results)] = [username, name, puzzle, cost, cycles, area, notes]
+
+    # Save image with green rectangles around the considered zones, for debug purposes
+    #cv2.imwrite("output_debug.jpg", output)
+    
+    os.remove("temp.jpg")
+
+print("Done.")
+print(results)
+results.to_csv(output_path)