#Undisputed Bot Reference

File: und.py · Lines: 1456 · Game: Undisputed Boxing

#Overview

The Undisputed bot is the most complex bot, featuring:

• OCR-based fighter selection using Tesseract
• Fuzzy text matching using RapidFuzz
• Screen template matching for state detection
• Pixel-based damage stats reading during round breaks
• Color-based winner detection after fights
• Auto-focus loop running every 1 second

#Global Variables

#Functions

#print_and_emit(message, level) ACTIVE

Print to console + emit to /logs namespace.

#log_to_file(message) ACTIVE

Write to und.log if USE_LOG_FILE=True, else print.

#focus_window_by_exact_title(title, quiet) ACTIVE

Focus game window via pywinauto. Connects to Undisputed.exe.

#_auto_focus_loop() ACTIVE

Background thread that calls focus_window_by_exact_title('Undisputed', quiet=True) every 1 second.

#_start_auto_focus_if_needed() ACTIVE

Start the auto-focus thread if not already running.

#OCR Functions

#read_first_name(frame, away) ACTIVE

OCR read fighter's first name from screen.

ROI Coordinates: (231, 863) to (455, 890) for home

#read_last_name(frame, away) ACTIVE

OCR read fighter's last name.

ROI: (221, 890) to (530, 933) for home, (1366, 890) to (1706, 930) for away

#read_player_name(frame, away) ACTIVE

OCR read fighter's full player name.

ROI: (206, 941) to (485, 968) for home, (1424, 946) to (1728, 971) for away

#read_p1_stuck(frame) ACTIVE

OCR check if game is stuck at P1 selection screen.

ROI: (1077, 470) to (1202, 573)

#compare_text_fuzzy(text1, text2, threshold) ACTIVE

Compare two texts using multiple RapidFuzz algorithms and return best score.

Algorithms used: fuzz.ratio, fuzz.partial_ratio, fuzz.token_sort_ratio, fuzz.WRatio

Returns: (is_match: bool, best_score: float)

#Fight Setup

#startMatch1() ACTIVE

Complete fighter selection flow:

1. Weight Selection: Maps weight class (1-10) to number of Key.2 presses
2. Red Corner Selection: Loop OCR + fuzzy match → navigate with Key.right
3. Blue Corner Selection: Same process for away corner

Special OCR Corrections:

• JOE LOUIS → matches against 'LoYIg' fallback
• REGIS PROGAIS → matches against 'PROGCRAIS'
• SAUL ALVAREZ → matches against 'SAGL' or 'SAOL'
• ERIC ESCH → matches against 'BRIG'
• MUHAMMAD ALI → excludes '64' variant

#_start_sequence_from_payload(data, source_label) ACTIVE

Unified start sequence handler (used by both on_new_match and on_start_match).

Flow:

1. Clear flags → Stop memory thread
2. Start OBS stream (with retry up to 5x)
3. Parse payload (player names, weight, match ID)
4. Navigate game menus (15s wait → Press F → Press R)
5. P1 stuck detection
6. Call startMatch1() for fighter selection
7. Set difficulty → Confirm fight
8. Start memory_listener_thread()

#Screen State Handlers

#on_screen_changed(screen_name, similarity) ACTIVE

Main screen state callback. Handles:

#Winner Detection

#detect_color_name(x, y) ACTIVE

Detect winner side by pixel color at coordinates.

Uses Euclidean color distance for matching.

#get_pixel_color(x, y) INTERNAL

Get hex color of pixel at screen coordinates.

#hex_to_rgb(hex_color) INTERNAL

Convert hex color to RGB tuple.

#color_distance(c1, c2) INTERNAL

Calculate Euclidean distance between two RGB colors.

#Damage Stats (Break Time)

#get_bar_percent(start_x, end_x, y_pos, avoid_hex, tolerance) ACTIVE

Scan pixel bar to calculate fill percentage.

• Home bars: scan left-to-right
• Away bars: scan right-to-left
• Skips pixels matching background color (#010101)

#get_box_percent_inverse(points, avoid_hex, tolerance) ACTIVE

Calculate percentage of active (non-black) pixels from a set of coordinate points.

#get_all_screen_stats() ACTIVE

Read all damage stats from screen:

#log_stats_to_file(data) ACTIVE

Append stats to match_stats_log.json (JSON Lines format) for debugging.

#Memory Thread

#memory_listener_thread() ACTIVE

Main loop running every 0.5 seconds:

1. Read gameTime and currentRound
2. During break time (gameTime == 0): capture screen stats
3. Read boxer scorecard data from memory
4. Detect match start when gameTime > 0
5. Emit match-data with round scores

#Socket Events (Emitted)

#Socket Events (Received)