Lichess / app.py
Kaveh
Update app.py
e36519b unverified
# -*- coding: utf-8 -*-
# =============================================
# Gradio App for Chess Game Analysis - Lichess API Version
# v18: Fixed empty chart visibility and chart sizing for mobile
# =============================================
import gradio as gr
import pandas as pd
import plotly.express as px
import plotly.graph_objects as go
import requests
import json
from datetime import datetime, timedelta, timezone
import time
import re
import traceback
# --- Configuration ---
# No Streamlit config needed
# --- Constants & Defaults ---
TIME_PERIOD_OPTIONS = {"Last Month": timedelta(days=30), "Last 3 Months": timedelta(days=90), "Last Year": timedelta(days=365), "Last 3 Years": timedelta(days=3*365)}
DEFAULT_TIME_PERIOD = "Last Year"
PERF_TYPE_OPTIONS_SINGLE = ['Bullet', 'Blitz', 'Rapid']
DEFAULT_PERF_TYPE = 'Bullet'
DEFAULT_RATED_ONLY = True
ECO_CSV_PATH = "eco_to_opening.csv"
TITLES_TO_ANALYZE = ['GM', 'IM', 'FM', 'CM', 'WGM', 'WIM', 'WFM', 'WCM', 'NM']
# =============================================
# Helper Function: Categorize Time Control
# =============================================
def categorize_time_control(tc_str, speed_info):
"""Categorizes time control based on speed info or parsed string."""
if isinstance(speed_info, str) and speed_info in ['bullet', 'blitz', 'rapid', 'classical', 'correspondence']:
return speed_info.capitalize()
if not isinstance(tc_str, str) or tc_str in ['-', '?', 'Unknown']:
return 'Unknown'
if tc_str == 'Correspondence':
return 'Correspondence'
if '+' in tc_str:
parts = tc_str.split('+')
if len(parts) == 2:
base_str, increment_str = parts[0], parts[1]
try:
base = int(base_str)
increment = int(increment_str)
total = base + 40 * increment
if total >= 1500: return 'Classical'
if total >= 480: return 'Rapid'
if total >= 180: return 'Blitz'
if total > 0: return 'Bullet'
return 'Unknown'
except ValueError:
return 'Unknown'
else:
return 'Unknown'
else:
try:
base = int(tc_str)
if base >= 1500: return 'Classical'
if base >= 480: return 'Rapid'
if base >= 180: return 'Blitz'
if base > 0: return 'Bullet'
return 'Unknown'
except ValueError:
tc_lower = tc_str.lower()
if 'classical' in tc_lower: return 'Classical'
if 'rapid' in tc_lower: return 'Rapid'
if 'blitz' in tc_lower: return 'Blitz'
if 'bullet' in tc_lower: return 'Bullet'
return 'Unknown'
# =============================================
# Helper Function: Load ECO Mapping
# =============================================
ECO_MAPPING = {}
try:
df_eco_global = pd.read_csv(ECO_CSV_PATH)
if "ECO Code" in df_eco_global.columns and "Opening Name" in df_eco_global.columns:
ECO_MAPPING = df_eco_global.drop_duplicates(subset=['ECO Code']).set_index('ECO Code')['Opening Name'].to_dict()
print(f"OK: Loaded {len(ECO_MAPPING)} ECO mappings.")
else:
print(f"WARN: ECO file '{ECO_CSV_PATH}' missing columns.")
except FileNotFoundError:
print(f"WARN: ECO file '{ECO_CSV_PATH}' not found.")
except Exception as e:
print(f"WARN: Error loading ECO file: {e}")
# =============================================
# API Data Loading and Processing Function
# =============================================
def load_from_lichess_api(username: str, time_period_key: str, perf_type: str, rated: bool, eco_map: dict, progress=None):
if not username: return pd.DataFrame(), "⚠️ Enter username."
if not perf_type: return pd.DataFrame(), "⚠️ Select game type."
if progress is not None:
try:
progress(0, desc="Initializing...")
except Exception:
pass
username_lower = username.lower()
status_message = f"Fetching {perf_type} games..."
if progress is not None:
progress(0.1, desc=status_message)
since_timestamp_ms = None
time_delta = TIME_PERIOD_OPTIONS.get(time_period_key)
if time_delta:
start_date = datetime.now(timezone.utc) - time_delta
since_timestamp_ms = int(start_date.timestamp() * 1000)
api_params = {"rated": str(rated).lower(), "perfType": perf_type.lower(), "opening": "true", "moves": "false", "tags": "false", "pgnInJson": "false"}
if since_timestamp_ms: api_params["since"] = since_timestamp_ms
api_url = f"https://lichess.org/api/games/user/{username}"
headers = {"Accept": "application/x-ndjson"}
all_games_data = []
error_counter = 0
lines_processed = 0
try:
response = requests.get(api_url, params=api_params, headers=headers, stream=True)
response.raise_for_status()
if progress is not None:
progress(0.3, desc="Processing stream...")
for line in response.iter_lines():
if line:
lines_processed += 1
game_data_raw = line.decode('utf-8')
if progress is not None and lines_processed % 100 == 0:
progress(0.3 + (lines_processed % 1000 / 2000), desc=f"Processing game {lines_processed}...")
try:
game_data = json.loads(game_data_raw)
white_info = game_data.get('players', {}).get('white', {})
black_info = game_data.get('players', {}).get('black', {})
white_user = white_info.get('user', {})
black_user = black_info.get('user', {})
opening_info = game_data.get('opening', {})
clock_info = game_data.get('clock')
game_id = game_data.get('id', 'N/A')
created_at_ms = game_data.get('createdAt')
game_date = pd.to_datetime(created_at_ms, unit='ms', utc=True, errors='coerce')
if pd.isna(game_date): continue
variant = game_data.get('variant', 'standard')
speed = game_data.get('speed', 'unknown')
perf = game_data.get('perf', 'unknown')
status = game_data.get('status', 'unknown')
winner = game_data.get('winner')
white_name = white_user.get('name', 'Unknown')
black_name = black_user.get('name', 'Unknown')
white_title = white_user.get('title')
black_title = black_user.get('title')
white_rating = pd.to_numeric(white_info.get('rating'), errors='coerce')
black_rating = pd.to_numeric(black_info.get('rating'), errors='coerce')
player_color, player_elo, opp_name_raw, opp_title_raw, opp_elo = (None, None, 'Unknown', None, None)
if username_lower == white_name.lower():
player_color, player_elo, opp_name_raw, opp_title_raw, opp_elo = ('White', white_rating, black_name, black_title, black_rating)
elif username_lower == black_name.lower():
player_color, player_elo, opp_name_raw, opp_title_raw, opp_elo = ('Black', black_rating, white_name, white_title, white_rating)
else:
continue
if player_color is None or pd.isna(player_elo) or pd.isna(opp_elo): continue
res_num, res_str = (0.5, "Draw")
if status not in ['draw', 'stalemate']:
if winner == player_color.lower(): res_num, res_str = (1, "Win")
elif winner is not None: res_num, res_str = (0, "Loss")
tc_str = "Unknown"
if clock_info:
init = clock_info.get('initial')
incr = clock_info.get('increment')
if init is not None and incr is not None: tc_str = f"{init}+{incr}"
elif speed == 'correspondence': tc_str = "Correspondence"
eco = opening_info.get('eco', 'Unknown')
op_name_api = opening_info.get('name', 'Unknown Opening').replace('?', '').split(':')[0].strip()
op_name_custom = eco_map.get(eco, f"ECO: {eco}" if eco != 'Unknown' else 'Unknown Opening')
term_map = {"mate": "Normal", "resign": "Normal", "stalemate": "Normal", "timeout": "Time forfeit", "draw": "Normal", "outoftime": "Time forfeit", "cheat": "Cheat", "noStart": "Aborted", "unknownFinish": "Unknown", "variantEnd": "Variant End"}
term = term_map.get(status, "Unknown")
opp_title_final = 'Unknown'
if opp_title_raw and opp_title_raw.strip():
opp_title_clean = opp_title_raw.replace(' ', '').strip().upper()
if opp_title_clean and opp_title_clean != '?': opp_title_final = opp_title_clean
def clean_name(n): return re.sub(r'^(GM|IM|FM|WGM|WIM|WFM|CM|WCM|NM)\s+', '', n).strip()
opp_name_clean = clean_name(opp_name_raw)
all_games_data.append({
'Date': game_date, 'Event': perf, 'White': white_name, 'Black': black_name,
'Result': "1-0" if winner == 'white' else ("0-1" if winner == 'black' else "1/2-1/2"),
'WhiteElo': int(white_rating) if not pd.isna(white_rating) else 0,
'BlackElo': int(black_rating) if not pd.isna(black_rating) else 0,
'ECO': eco, 'OpeningName_API': op_name_api, 'OpeningName_Custom': op_name_custom,
'TimeControl': tc_str, 'Termination': term, 'PlyCount': game_data.get('turns', 0),
'LichessID': game_id, 'PlayerID': username, 'PlayerColor': player_color,
'PlayerElo': int(player_elo), 'OpponentName': opp_name_clean, 'OpponentNameRaw': opp_name_raw,
'OpponentElo': int(opp_elo), 'OpponentTitle': opp_title_final, 'PlayerResultNumeric': res_num,
'PlayerResultString': res_str, 'Variant': variant, 'Speed': speed, 'Status': status, 'PerfType': perf
})
except json.JSONDecodeError:
error_counter += 1
except Exception:
error_counter += 1
except requests.exceptions.RequestException as e:
return pd.DataFrame(), f"🚨 API Error: {e}"
except Exception as e:
return pd.DataFrame(), f"🚨 Error: {e}\n{traceback.format_exc()}"
status_message = f"Processed {len(all_games_data)} games."
if error_counter > 0: status_message += f" Skipped {error_counter} errors."
if not all_games_data: return pd.DataFrame(), f"⚠️ No games found matching criteria."
if progress is not None:
progress(0.8, desc="Finalizing...")
df = pd.DataFrame(all_games_data)
if not df.empty:
df['Date'] = pd.to_datetime(df['Date'], errors='coerce')
df = df.dropna(subset=['Date'])
if df.empty: return df, "⚠️ No games with valid dates."
df['Year'] = df['Date'].dt.year
df['Month'] = df['Date'].dt.month
df['Day'] = df['Date'].dt.day
df['Hour'] = df['Date'].dt.hour
df['DayOfWeekNum'] = df['Date'].dt.dayofweek
df['DayOfWeekName'] = df['Date'].dt.day_name()
df['PlayerElo'] = df['PlayerElo'].astype(int)
df['OpponentElo'] = df['OpponentElo'].astype(int)
df['EloDiff'] = df['PlayerElo'] - df['OpponentElo']
df['TimeControl_Category'] = df.apply(lambda r: categorize_time_control(r['TimeControl'], r['Speed']), axis=1)
df = df.sort_values(by='Date').reset_index(drop=True)
if progress is not None:
progress(1, desc="Complete!")
return df, status_message
# =============================================
# Plotting Functions
# =============================================
def plot_win_loss_pie(df, display_name):
if 'PlayerResultString' not in df.columns: return go.Figure()
result_counts = df['PlayerResultString'].value_counts()
fig = px.pie(values=result_counts.values, names=result_counts.index, title=f'Overall Results for {display_name}',
color=result_counts.index, color_discrete_map={'Win': '#4CAF50', 'Draw': '#B0BEC5', 'Loss': '#F44336'}, hole=0.3)
fig.update_traces(textposition='inside', textinfo='percent+label', pull=[0.05 if x == 'Win' else 0 for x in result_counts.index])
fig.update_layout(dragmode=False, autosize=True, height=400, width=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_win_loss_by_color(df):
if not all(col in df.columns for col in ['PlayerColor', 'PlayerResultString']): return go.Figure()
try:
color_results = df.groupby(['PlayerColor', 'PlayerResultString']).size().unstack(fill_value=0)
except KeyError:
return go.Figure().update_layout(title="Error: Missing Columns")
for res in ['Win', 'Draw', 'Loss']: color_results[res] = color_results.get(res, 0)
color_results = color_results[['Win', 'Draw', 'Loss']]
total = color_results.sum(axis=1)
color_results_pct = color_results.apply(lambda x: x * 100 / total[x.name] if total[x.name] > 0 else 0, axis=1)
fig = px.bar(color_results_pct, barmode='stack', title='Results by Color', labels={'value': '%', 'PlayerColor': 'Played As'},
color='PlayerResultString', color_discrete_map={'Win': '#4CAF50', 'Draw': '#B0BEC5', 'Loss': '#F44336'},
text_auto='.1f', category_orders={"PlayerColor": ["White", "Black"]})
fig.update_layout(yaxis_title="Percentage (%)", xaxis_title="Color Played", dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
fig.update_traces(textangle=0)
return fig
def plot_rating_trend(df, display_name):
if not all(col in df.columns for col in ['Date', 'PlayerElo']): return go.Figure()
df_plot = df.copy()
df_plot['PlayerElo'] = pd.to_numeric(df_plot['PlayerElo'], errors='coerce')
df_sorted = df_plot[df_plot['PlayerElo'].notna() & (df_plot['PlayerElo'] > 0)].sort_values('Date')
if df_sorted.empty: return go.Figure().update_layout(title=f"No Elo data")
fig = go.Figure()
fig.add_trace(go.Scatter(x=df_sorted['Date'], y=df_sorted['PlayerElo'], mode='lines+markers', name='Elo',
line=dict(color='#1E88E5', width=2), marker=dict(size=5, opacity=0.7)))
fig.update_layout(title=f'{display_name}\'s Rating Trend', xaxis_title='Date', yaxis_title='Elo Rating',
hovermode="x unified", xaxis_rangeslider_visible=True, dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_performance_vs_opponent_elo(df):
if not all(col in df.columns for col in ['PlayerResultString', 'EloDiff']): return go.Figure()
fig = px.box(df, x='PlayerResultString', y='EloDiff', title='Elo Advantage vs. Result',
labels={'PlayerResultString': 'Result', 'EloDiff': 'Your Elo - Opponent Elo'},
category_orders={"PlayerResultString": ["Win", "Draw", "Loss"]}, color='PlayerResultString',
color_discrete_map={'Win': '#4CAF50', 'Draw': '#B0BEC5', 'Loss': '#F44336'}, points='outliers')
fig.add_hline(y=0, line_dash="dash", line_color="grey")
fig.update_traces(marker=dict(opacity=0.8))
fig.update_layout(dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_games_by_dow(df):
if 'DayOfWeekName' not in df.columns: return go.Figure()
dow_order = ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"]
games_by_dow = df['DayOfWeekName'].value_counts().reindex(dow_order, fill_value=0)
fig = px.bar(games_by_dow, x=games_by_dow.index, y=games_by_dow.values, title="Games by Day of Week",
labels={'x': 'Day', 'y': 'Games'}, text=games_by_dow.values)
fig.update_traces(marker_color='#9C27B0', textposition='outside')
fig.update_layout(dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_winrate_by_dow(df):
if not all(col in df.columns for col in ['DayOfWeekName', 'PlayerResultNumeric']): return go.Figure()
dow_order = ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"]
wins_by_dow = df[df['PlayerResultNumeric'] == 1].groupby('DayOfWeekName').size()
total_by_dow = df.groupby('DayOfWeekName').size()
win_rate = (wins_by_dow.reindex(total_by_dow.index, fill_value=0) / total_by_dow).fillna(0) * 100
win_rate = win_rate.reindex(dow_order, fill_value=0)
fig = px.bar(win_rate, x=win_rate.index, y=win_rate.values, title="Win Rate (%) by Day",
labels={'x': 'Day', 'y': 'Win Rate (%)'}, text=win_rate.values)
fig.update_traces(marker_color='#FF9800', texttemplate='%{text:.1f}%', textposition='outside')
fig.update_layout(yaxis_range=[0, 100], dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_games_by_hour(df):
if 'Hour' not in df.columns: return go.Figure()
games_by_hour = df['Hour'].value_counts().sort_index().reindex(range(24), fill_value=0)
fig = px.bar(games_by_hour, x=games_by_hour.index, y=games_by_hour.values, title="Games by Hour (UTC)",
labels={'x': 'Hour', 'y': 'Games'}, text=games_by_hour.values)
fig.update_traces(marker_color='#03A9F4', textposition='outside')
fig.update_layout(xaxis=dict(tickmode='linear'), dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_winrate_by_hour(df):
if not all(col in df.columns for col in ['Hour', 'PlayerResultNumeric']): return go.Figure()
wins_by_hour = df[df['PlayerResultNumeric'] == 1].groupby('Hour').size()
total_by_hour = df.groupby('Hour').size()
win_rate = (wins_by_hour.reindex(total_by_hour.index, fill_value=0) / total_by_hour).fillna(0) * 100
win_rate = win_rate.reindex(range(24), fill_value=0)
fig = px.line(win_rate, x=win_rate.index, y=win_rate.values, markers=True, title="Win Rate (%) by Hour (UTC)",
labels={'x': 'Hour', 'y': 'Win Rate (%)'})
fig.update_traces(line_color='#8BC34A')
fig.update_layout(yaxis_range=[0, 100], xaxis=dict(tickmode='linear'), dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_games_per_year(df):
if 'Year' not in df.columns: return go.Figure()
games_per_year = df['Year'].value_counts().sort_index()
fig = px.bar(games_per_year, x=games_per_year.index, y=games_per_year.values, title='Games Per Year',
labels={'x': 'Year', 'y': 'Games'}, text=games_per_year.values)
fig.update_traces(marker_color='#2196F3', textposition='outside')
fig.update_layout(xaxis_title="Year", yaxis_title="Number of Games", xaxis={'type': 'category'}, dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_win_rate_per_year(df):
if not all(col in df.columns for col in ['Year', 'PlayerResultNumeric']): return go.Figure()
wins_per_year = df[df['PlayerResultNumeric'] == 1].groupby('Year').size()
total_per_year = df.groupby('Year').size()
win_rate = (wins_per_year.reindex(total_per_year.index, fill_value=0) / total_per_year).fillna(0) * 100
win_rate.index = win_rate.index.astype(str)
fig = px.line(win_rate, x=win_rate.index, y=win_rate.values, title='Win Rate (%) Per Year', markers=True,
labels={'x': 'Year', 'y': 'Win Rate (%)'})
fig.update_traces(line_color='#FFC107', line_width=2.5)
fig.update_layout(yaxis_range=[0, 100], dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_performance_by_time_control(df):
if not all(col in df.columns for col in ['TimeControl_Category', 'PlayerResultString']): return go.Figure()
try:
tc_results = df.groupby(['TimeControl_Category', 'PlayerResultString']).size().unstack(fill_value=0)
for res in ['Win', 'Draw', 'Loss']: tc_results[res] = tc_results.get(res, 0)
tc_results = tc_results[['Win', 'Draw', 'Loss']]
total = tc_results.sum(axis=1)
tc_results_pct = tc_results.apply(lambda x: x * 100 / total[x.name] if total[x.name] > 0 else 0, axis=1)
found = df['TimeControl_Category'].unique()
pref = ['Bullet', 'Blitz', 'Rapid', 'Classical', 'Correspondence', 'Unknown']
order = [c for c in pref if c in found] + [c for c in found if c not in pref]
tc_results_pct = tc_results_pct.reindex(index=order).dropna(axis=0, how='all')
fig = px.bar(tc_results_pct, title='Performance by Time Control', labels={'value': '%', 'TimeControl_Category': 'Category'},
color='PlayerResultString', color_discrete_map={'Win': '#4CAF50', 'Draw': '#B0BEC5', 'Loss': '#F44336'},
barmode='group', text_auto='.1f')
fig.update_layout(xaxis_title="Time Control Category", yaxis_title="Percentage (%)", dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
fig.update_traces(textangle=0)
return fig
except Exception:
return go.Figure().update_layout(title="Error")
def plot_opening_frequency(df, top_n=20, opening_col='OpeningName_API'):
if opening_col not in df.columns: return go.Figure()
source_label = "Lichess API" if opening_col == 'OpeningName_API' else "Custom Mapping"
opening_counts = df[df[opening_col] != 'Unknown Opening'][opening_col].value_counts().nlargest(top_n)
fig = px.bar(opening_counts, y=opening_counts.index, x=opening_counts.values, orientation='h',
title=f'Top {top_n} Openings ({source_label})', labels={'y': 'Opening', 'x': 'Games'}, text=opening_counts.values)
fig.update_layout(yaxis={'categoryorder': 'total ascending'}, dragmode=False, autosize=True, height=500, margin=dict(l=20, r=20, t=50, b=20))
fig.update_traces(marker_color='#673AB7', textposition='outside')
return fig
def plot_win_rate_by_opening(df, min_games=5, top_n=20, opening_col='OpeningName_API'):
if not all(col in df.columns for col in [opening_col, 'PlayerResultNumeric']): return go.Figure()
source_label = "Lichess API" if opening_col == 'OpeningName_API' else "Custom Mapping"
opening_stats = df.groupby(opening_col).agg(total_games=('PlayerResultNumeric', 'count'),
wins=('PlayerResultNumeric', lambda x: (x == 1).sum()))
opening_stats = opening_stats[(opening_stats['total_games'] >= min_games) & (opening_stats.index != 'Unknown Opening')].copy()
if opening_stats.empty: return go.Figure().update_layout(title=f"No openings >= {min_games} games ({source_label})")
opening_stats['win_rate'] = (opening_stats['wins'] / opening_stats['total_games']) * 100
opening_stats_plot = opening_stats.nlargest(top_n, 'win_rate')
fig = px.bar(opening_stats_plot, y=opening_stats_plot.index, x='win_rate', orientation='h',
title=f'Top {top_n} Openings by Win Rate (Min {min_games} games, {source_label})',
labels={'win_rate': 'Win Rate (%)', opening_col: 'Opening'}, text='win_rate')
fig.update_traces(texttemplate='%{text:.1f}%', textposition='inside', marker_color='#009688')
fig.update_layout(yaxis={'categoryorder': 'total ascending'}, xaxis_title="Win Rate (%)", dragmode=False, autosize=True, height=500, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_most_frequent_opponents(df, top_n=20):
if 'OpponentName' not in df.columns: return go.Figure()
opp_counts = df[df['OpponentName'] != 'Unknown']['OpponentName'].value_counts().nlargest(top_n)
fig = px.bar(opp_counts, y=opp_counts.index, x=opp_counts.values, orientation='h',
title=f'Top {top_n} Opponents', labels={'y': 'Opponent', 'x': 'Games'}, text=opp_counts.values)
fig.update_layout(yaxis={'categoryorder': 'total ascending'}, dragmode=False, autosize=True, height=500, margin=dict(l=20, r=20, t=50, b=20))
fig.update_traces(marker_color='#FF5722', textposition='outside')
return fig
def plot_games_by_dom(df):
if 'Day' not in df.columns: return go.Figure()
games_by_dom = df['Day'].value_counts().sort_index().reindex(range(1, 32), fill_value=0)
fig = px.bar(games_by_dom, x=games_by_dom.index, y=games_by_dom.values, title="Games Played per Day of Month",
labels={'x': 'Day of Month', 'y': 'Number of Games'}, text=games_by_dom.values)
fig.update_traces(marker_color='#E91E63', textposition='outside')
fig.update_layout(xaxis=dict(tickmode='linear'), dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_winrate_by_dom(df):
if not all(col in df.columns for col in ['Day', 'PlayerResultNumeric']): return go.Figure()
wins_by_dom = df[df['PlayerResultNumeric'] == 1].groupby('Day').size()
total_by_dom = df.groupby('Day').size()
win_rate = (wins_by_dom.reindex(total_by_dom.index, fill_value=0) / total_by_dom).fillna(0) * 100
win_rate = win_rate.reindex(range(1, 32), fill_value=0)
fig = px.line(win_rate, x=win_rate.index, y=win_rate.values, markers=True, title="Win Rate (%) per Day of Month",
labels={'x': 'Day of Month', 'y': 'Win Rate (%)'})
fig.update_traces(line_color='#FF5722')
fig.update_layout(yaxis_range=[0, 100], xaxis=dict(tickmode='linear'), dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
return fig
def plot_time_forfeit_summary(wins_tf, losses_tf):
data = {'Outcome': ['Won on Time', 'Lost on Time'], 'Count': [wins_tf, losses_tf]}
df_tf = pd.DataFrame(data)
fig = px.bar(df_tf, x='Outcome', y='Count', title="Time Forfeit Summary", color='Outcome',
color_discrete_map={'Won on Time': '#4CAF50', 'Lost on Time': '#F44336'}, text='Count')
fig.update_layout(showlegend=False, dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
fig.update_traces(textposition='outside')
return fig
def plot_time_forfeit_by_tc(tf_games_df):
if 'TimeControl_Category' not in tf_games_df.columns or tf_games_df.empty: return go.Figure().update_layout(title="No TF Data by Category")
tf_by_tc = tf_games_df['TimeControl_Category'].value_counts()
fig = px.bar(tf_by_tc, x=tf_by_tc.index, y=tf_by_tc.values, title="Time Forfeits by Time Control",
labels={'x': 'Category', 'y': 'Forfeits'}, text=tf_by_tc.values)
fig.update_layout(dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
fig.update_traces(marker_color='#795548', textposition='outside')
return fig
# =============================================
# Helper Functions
# =============================================
def filter_and_analyze_titled(df, titles):
if 'OpponentTitle' not in df.columns: return pd.DataFrame()
titled_games = df[df['OpponentTitle'].isin(titles)].copy()
return titled_games
def filter_and_analyze_time_forfeits(df):
if 'Termination' not in df.columns: return pd.DataFrame(), 0, 0
tf_games = df[df['Termination'].str.contains("Time forfeit", na=False, case=False)].copy()
if tf_games.empty: return tf_games, 0, 0
wins_tf = len(tf_games[tf_games['PlayerResultNumeric'] == 1])
losses_tf = len(tf_games[tf_games['PlayerResultNumeric'] == 0])
return tf_games, wins_tf, losses_tf
# =============================================
# Gradio Main Analysis Function
# =============================================
def perform_full_analysis(username, time_period_key, perf_type, selected_titles_list, progress=gr.Progress(track_tqdm=True)):
df, status_msg = load_from_lichess_api(username, time_period_key, perf_type, DEFAULT_RATED_ONLY, ECO_MAPPING, progress)
num_outputs = 34 # Adjusted for visibility state
if not isinstance(df, pd.DataFrame) or df.empty:
return status_msg, pd.DataFrame(), False, *([None] * (num_outputs - 3))
try:
fig_pie = plot_win_loss_pie(df, username)
fig_color = plot_win_loss_by_color(df)
fig_rating = plot_rating_trend(df, username)
fig_elo_diff = plot_performance_vs_opponent_elo(df)
total_g = len(df)
w = len(df[df['PlayerResultNumeric'] == 1])
l = len(df[df['PlayerResultNumeric'] == 0])
d = len(df[df['PlayerResultNumeric'] == 0.5])
wr = (w / total_g * 100) if total_g > 0 else 0
avg_opp = df['OpponentElo'].mean()
avg_opp_display = f"{avg_opp:.0f}" if not pd.isna(avg_opp) else 'N/A'
overview_stats_md = f"**Total:** {total_g:,} | **WR:** {wr:.1f}% | **W/L/D:** {w}/{l}/{d} | **Avg Opp:** {avg_opp_display}"
fig_games_yr = plot_games_per_year(df)
fig_wr_yr = plot_win_rate_per_year(df)
fig_perf_tc = plot_performance_by_time_control(df)
fig_games_dow = plot_games_by_dow(df)
fig_wr_dow = plot_winrate_by_dow(df)
fig_games_hod = plot_games_by_hour(df)
fig_wr_hod = plot_winrate_by_hour(df)
fig_games_dom = plot_games_by_dom(df)
fig_wr_dom = plot_winrate_by_dom(df)
fig_open_freq_api = plot_opening_frequency(df, top_n=15, opening_col='OpeningName_API')
fig_open_wr_api = plot_win_rate_by_opening(df, min_games=5, top_n=15, opening_col='OpeningName_API')
fig_open_freq_cust = plot_opening_frequency(df, top_n=15, opening_col='OpeningName_Custom') if ECO_MAPPING else go.Figure().update_layout(title="Custom Map Unavailable")
fig_open_wr_cust = plot_win_rate_by_opening(df, min_games=5, top_n=15, opening_col='OpeningName_Custom') if ECO_MAPPING else go.Figure().update_layout(title="Custom Map Unavailable")
fig_opp_freq = plot_most_frequent_opponents(df, top_n=20)
df_opp_list = df[df['OpponentName'] != 'Unknown']['OpponentName'].value_counts().reset_index(name='Games').head(20) if 'OpponentName' in df else pd.DataFrame()
fig_opp_elo = plot_performance_vs_opponent_elo(df)
tf_games, wins_tf, losses_tf = filter_and_analyze_time_forfeits(df)
fig_tf_summary = plot_time_forfeit_summary(wins_tf, losses_tf) if not tf_games.empty else go.Figure().update_layout(title="No Time Forfeit Data")
fig_tf_tc = plot_time_forfeit_by_tc(tf_games) if not tf_games.empty else go.Figure().update_layout(title="No TF Data by Category")
df_tf_list = tf_games[['Date', 'OpponentName', 'PlayerColor', 'PlayerResultString', 'TimeControl', 'PlyCount', 'Termination']].sort_values('Date', ascending=False).head(20) if not tf_games.empty else pd.DataFrame()
term_counts = df['Termination'].value_counts()
fig_term_all = px.bar(term_counts, x=term_counts.index, y=term_counts.values, title="Overall Termination Reasons",
labels={'x': 'Reason', 'y': 'Count'}, text=term_counts.values)
fig_term_all.update_layout(dragmode=False, autosize=True, height=400, margin=dict(l=20, r=20, t=50, b=20))
fig_term_all.update_traces(textposition='outside')
titled_status_msg = ""
fig_titled_pie, fig_titled_color, fig_titled_rating, df_titled_h2h = go.Figure(), go.Figure(), go.Figure(), pd.DataFrame()
if selected_titles_list:
titled_games = filter_and_analyze_titled(df, selected_titles_list)
if not titled_games.empty:
titled_status_msg = f"✅ Found {len(titled_games)} games vs {', '.join(selected_titles_list)}."
fig_titled_pie = plot_win_loss_pie(titled_games, f"{username} vs Titles")
fig_titled_color = plot_win_loss_by_color(titled_games)
fig_titled_rating = plot_rating_trend(titled_games, f"{username} (vs Titles)")
h2h = titled_games.groupby('OpponentNameRaw')['PlayerResultString'].value_counts().unstack(fill_value=0)
for res in ['Win', 'Loss', 'Draw']: h2h[res] = h2h.get(res, 0)
h2h = h2h[['Win', 'Loss', 'Draw']]
h2h['Total'] = h2h.sum(axis=1)
h2h['Score'] = h2h['Win'] + 0.5 * h2h['Draw']
df_titled_h2h = h2h.sort_values('Total', ascending=False).reset_index()
else:
titled_status_msg = f"ℹ️ No games found vs selected titles ({', '.join(selected_titles_list)})."
else:
titled_status_msg = "ℹ️ Select titles from the sidebar to analyze."
return (status_msg, df, True, fig_pie, overview_stats_md, fig_color, fig_rating, fig_elo_diff, fig_games_yr, fig_wr_yr,
"(Results by color shown in Overview)", fig_games_dow, fig_wr_dow, fig_games_hod, fig_wr_hod, fig_games_dom,
fig_wr_dom, fig_perf_tc, fig_open_freq_api, fig_open_wr_api, fig_open_freq_cust, fig_open_wr_cust,
fig_opp_freq, df_opp_list, fig_opp_elo, titled_status_msg, fig_titled_pie, fig_titled_color, fig_titled_rating,
df_titled_h2h, fig_tf_summary, fig_tf_tc, df_tf_list, fig_term_all)
except Exception as e:
error_msg = f"🚨 Error generating results: {e}\n{traceback.format_exc()}"
return error_msg, pd.DataFrame(), False, *([None] * (num_outputs - 3))
# =============================================
# Gradio Interface Definition
# =============================================
css = """
.gradio-container { font-family: 'IBM Plex Sans', sans-serif; }
footer { display: none !important; }
/* Responsive adjustments for plots */
.gr-plot { min-width: 100% !important; }
@media (max-width: 768px) {
.gr-row { flex-direction: column !important; }
.gr-plot { height: 350px !important; margin-bottom: 20px !important; }
}
@media (min-width: 769px) {
.gr-plot { height: 400px !important; }
}
"""
with gr.Blocks(theme=gr.themes.Soft(), css=css) as demo:
gr.Markdown("# ♟️ Lichess Insights\nAnalyze rated game statistics from Lichess API.")
df_state = gr.State(pd.DataFrame())
has_data = gr.State(False) # State to track if data is available
with gr.Row():
with gr.Column(scale=1, min_width=250): # Sidebar
gr.Markdown("## ⚙️ Settings")
username_input = gr.Textbox(label="Lichess Username", placeholder="e.g., DrNykterstein", elem_id="username_box")
time_period_input = gr.Dropdown(label="Time Period", choices=list(TIME_PERIOD_OPTIONS.keys()), value=DEFAULT_TIME_PERIOD)
perf_type_input = gr.Dropdown(label="Game Type", choices=PERF_TYPE_OPTIONS_SINGLE, value=DEFAULT_PERF_TYPE)
analyze_btn = gr.Button("Analyze Games", variant="primary")
status_output = gr.Markdown("")
gr.Markdown("---")
gr.Markdown("### Analyze vs Titled Players")
titled_player_select = gr.CheckboxGroup(label="Select Opponent Titles", choices=TITLES_TO_ANALYZE, value=['GM', 'IM'], elem_id="titled_select")
gr.Markdown("*(Analysis updates on 'Analyze Games' click)*")
with gr.Column(scale=4): # Main Content
with gr.Tabs() as tabs:
with gr.TabItem("1. Overview", id=0):
overview_stats_md_out = gr.Markdown(visible=False)
with gr.Row(visible=False) as overview_row:
overview_plot_pie = gr.Plot(label="Overall Results")
overview_plot_color = gr.Plot(label="Results by Color")
overview_plot_rating = gr.Plot(label="Rating Trend")
overview_plot_elo_diff = gr.Plot(label="Elo Advantage vs. Result")
with gr.TabItem("2. Perf. Over Time", id=1):
with gr.Row(visible=False) as perf_time_row:
time_plot_games_yr = gr.Plot(label="Games per Year")
time_plot_wr_yr = gr.Plot(label="Win Rate per Year")
with gr.TabItem("3. Perf. by Color", id=2):
color_plot_placeholder = gr.Markdown("(Results by color shown in Overview)", visible=False)
with gr.TabItem("4. Time & Date", id=3):
gr.Markdown("### Day of Week")
with gr.Row(visible=False) as dow_row:
time_plot_games_dow = gr.Plot(label="Games by Day of Week")
time_plot_wr_dow = gr.Plot(label="Win Rate by Day of Week")
gr.Markdown("### Hour of Day (UTC)")
with gr.Row(visible=False) as hod_row:
time_plot_games_hod = gr.Plot(label="Games by Hour (UTC)")
time_plot_wr_hod = gr.Plot(label="Win Rate by Hour (UTC)")
gr.Markdown("### Day of Month")
with gr.Row(visible=False) as dom_row:
time_plot_games_dom = gr.Plot(label="Games by Day of Month")
time_plot_wr_dom = gr.Plot(label="Win Rate by Day of Month")
gr.Markdown("### Time Control Category")
time_plot_perf_tc = gr.Plot(label="Performance by Time Control", visible=False)
with gr.TabItem("5. ECO & Openings", id=4):
gr.Markdown("#### API Names")
eco_plot_freq_api = gr.Plot(label="Opening Frequency (API)", visible=False)
eco_plot_wr_api = gr.Plot(label="Opening Win Rate (API)", visible=False)
gr.Markdown("---")
gr.Markdown("#### Custom Map")
if not ECO_MAPPING:
gr.Markdown("⚠️ Custom map not loaded.")
eco_plot_freq_cust = gr.Plot(label="Opening Frequency (Custom)", visible=False)
eco_plot_wr_cust = gr.Plot(label="Opening Win Rate (Custom)", visible=False)
with gr.TabItem("6. Opponents", id=5):
opp_plot_freq = gr.Plot(label="Frequent Opponents", visible=False)
opp_df_list = gr.DataFrame(label="Top Opponents List", wrap=True, visible=False)
opp_plot_elo = gr.Plot(label="Elo Advantage vs Result", visible=False)
with gr.TabItem("7. vs Titled", id=6):
gr.Markdown("Analysis based on sidebar selection.")
titled_status = gr.Markdown(visible=False)
with gr.Row(visible=False) as titled_row:
titled_plot_pie = gr.Plot(label="Results vs Selected Titles")
titled_plot_color = gr.Plot(label="Results by Color vs Selected Titles")
titled_plot_rating = gr.Plot(label="Rating Trend vs Selected Titles")
titled_df_h2h_comp = gr.DataFrame(label="Head-to-Head vs Selected Titles", wrap=True, visible=False)
with gr.TabItem("8. Termination", id=7):
gr.Markdown("### Time Forfeit")
term_plot_tf_summary = gr.Plot(label="Time Forfeit Summary", visible=False)
term_plot_tf_tc = gr.Plot(label="Time Forfeits by Time Control", visible=False)
with gr.Accordion("View Recent TF Games", open=False):
term_df_tf_list = gr.DataFrame(label="Recent TF Games", wrap=True, visible=False)
gr.Markdown("### Overall Termination")
term_plot_all = gr.Plot(label="Overall Termination", visible=False)
# Define visibility updates based on has_data
def update_visibility(has_data_value, *args):
visibility = has_data_value
return (
gr.update(visible=visibility), # overview_stats_md_out
gr.update(visible=visibility), # overview_row
gr.update(visible=visibility), # perf_time_row
gr.update(visible=visibility), # color_plot_placeholder
gr.update(visible=visibility), # dow_row
gr.update(visible=visibility), # hod_row
gr.update(visible=visibility), # dom_row
gr.update(visible=visibility), # time_plot_perf_tc
gr.update(visible=visibility), # eco_plot_freq_api
gr.update(visible=visibility), # eco_plot_wr_api
gr.update(visible=visibility), # eco_plot_freq_cust
gr.update(visible=visibility), # eco_plot_wr_cust
gr.update(visible=visibility), # opp_plot_freq
gr.update(visible=visibility), # opp_df_list
gr.update(visible=visibility), # opp_plot_elo
gr.update(visible=visibility), # titled_status
gr.update(visible=visibility), # titled_row
gr.update(visible=visibility), # titled_df_h2h_comp
gr.update(visible=visibility), # term_plot_tf_summary
gr.update(visible=visibility), # term_plot_tf_tc
gr.update(visible=visibility), # term_df_tf_list
gr.update(visible=visibility), # term_plot_all
)
outputs_list = [
status_output, df_state, has_data,
overview_plot_pie, overview_stats_md_out, overview_plot_color, overview_plot_rating, overview_plot_elo_diff,
time_plot_games_yr, time_plot_wr_yr, color_plot_placeholder, time_plot_games_dow, time_plot_wr_dow,
time_plot_games_hod, time_plot_wr_hod, time_plot_games_dom, time_plot_wr_dom, time_plot_perf_tc,
eco_plot_freq_api, eco_plot_wr_api, eco_plot_freq_cust, eco_plot_wr_cust, opp_plot_freq, opp_df_list,
opp_plot_elo, titled_status, titled_plot_pie, titled_plot_color, titled_plot_rating, titled_df_h2h_comp,
term_plot_tf_summary, term_plot_tf_tc, term_df_tf_list, term_plot_all
]
visibility_outputs = [
overview_stats_md_out, overview_row, perf_time_row, color_plot_placeholder, dow_row, hod_row, dom_row,
time_plot_perf_tc, eco_plot_freq_api, eco_plot_wr_api, eco_plot_freq_cust, eco_plot_wr_cust, opp_plot_freq,
opp_df_list, opp_plot_elo, titled_status, titled_row, titled_df_h2h_comp, term_plot_tf_summary, term_plot_tf_tc,
term_df_tf_list, term_plot_all
]
analyze_btn.click(fn=perform_full_analysis, inputs=[username_input, time_period_input, perf_type_input, titled_player_select], outputs=outputs_list).then(
fn=update_visibility, inputs=[has_data] + outputs_list[3:], outputs=visibility_outputs
)
# --- Launch the Gradio App ---
if __name__ == "__main__":
demo.launch(debug=True)