app.py

import streamlit as st
import pandas as pd
from github import Github
from wordcloud import WordCloud
import matplotlib.pyplot as plt
import re
import datetime

g = Github(st.secrets["ACCESS_TOKEN"])
repos = st.secrets["REPO_NAME"].split(",")
repos = [g.get_repo(repo) for repo in repos]

@st.cache_data
def fetch_data():

    issues_data = []
    
    for repo in repos:
        issues = repo.get_issues(state="all")
        
        for issue in issues:
            issues_data.append(
                {
                    'Issue': f"{issue.number} - {issue.title}", 
                    'State': issue.state, 
                    'Created at': issue.created_at, 
                    'Closed at': issue.closed_at,
                    'Last update': issue.updated_at,
                    'Labels': [label.name for label in issue.labels],
                    'Reactions': issue.reactions['total_count'],
                    'Comments': issue.comments,
                    'URL': issue.html_url,
                    'Repository': repo.name,
                }
            )
    return pd.DataFrame(issues_data)

def save_data(df):
    df.to_json("issues.json", orient="records", indent=4, index=False)

@st.cache_data
def load_data():
    try:
        df = pd.read_json("issues.json", convert_dates=["Created at", "Closed at", "Last update"], date_unit="ms")
    except:
        df = fetch_data()
        save_data(df)
    return df


st.title(f"GitHub Issues Dashboard")
status = st.status(label="Loading data...", state="running")

df = load_data()

today = datetime.date.today()

# Section 1: Issue activity metrics
st.header("Issue activity metrics")

col1, col2, col3 = st.columns(3)

state_counts = df['State'].value_counts()
open_issues = df.loc[df['State'] == 'open']
closed_issues = df.loc[df['State'] == 'closed']
closed_issues['Time to Close'] = closed_issues['Closed at'] - closed_issues['Created at']

with col1:
    st.metric(label="Open issues", value=state_counts['open'])

with col2:
    st.metric(label="Closed issues", value=state_counts['closed'])

with col3:
    average_time_to_close = closed_issues['Time to Close'].mean().days
    st.metric(label="Avg. days to close", value=average_time_to_close)


# TODO Plot: number of open vs closed issues by date


# st.subheader("Latest bugs 🐞")
# bug_issues = open_issues[open_issues["Labels"].apply(lambda labels: "type: bug" in labels)]
# bug_issues = bug_issues[["Issue","Labels","Created at","URL"]]
# st.dataframe(
#     bug_issues.sort_values(by="Created at", ascending=False),
#     hide_index=True,
#     column_config={
#         "Issue": st.column_config.TextColumn("Issue", width=400),
#         "Labels": st.column_config.TextColumn("Labels"),
#         "Created at": st.column_config.DatetimeColumn("Created at"),
#         "URL": st.column_config.LinkColumn("🔗", display_text="🔗")
#     }
# )

st.subheader("Latest updates 📝")
col1, col2 = st.columns(2)
with col1:
    last_update_date = st.date_input("Last updated after:", value=today - datetime.timedelta(days=7), format="DD-MM-YYYY")
    last_update_date = datetime.datetime.combine(last_update_date, datetime.datetime.min.time())  # Convert to datetime object
with col2:
    updated_issues = open_issues[open_issues["Last update"] > last_update_date]
    st.metric("Results:", updated_issues.shape[0])

st.dataframe(
    updated_issues[["Issue","Labels", "Repository", "Last update","URL"]].sort_values(by="Last update", ascending=False),
    hide_index=True,
    # use_container_width=True,
    column_config={
        "Issue": st.column_config.TextColumn("Issue", width="large"),
        "Labels": st.column_config.ListColumn("Labels", width="large"),
        "Last update": st.column_config.DatetimeColumn("Last update", width="medium"),
        "URL": st.column_config.LinkColumn("🔗", display_text="🔗", width="small")
    }
)

st.subheader("Stale issues? 🕸️")
col1, col2 = st.columns(2)
with col1:
    not_updated_since = st.date_input("Not updated since:", value=today - datetime.timedelta(days=90), format="DD-MM-YYYY")
    not_updated_since = datetime.datetime.combine(not_updated_since, datetime.datetime.min.time())  # Convert to datetime object
with col2:
    stale_issues = open_issues[open_issues["Last update"] < not_updated_since]
    st.metric("Results:", stale_issues.shape[0])
st.dataframe(
    stale_issues[["Issue","Labels", "Repository", "Last update","URL"]].sort_values(by="Last update", ascending=True),
    hide_index=True,
    # use_container_width=True,
    column_config={
        "Issue": st.column_config.TextColumn("Issue", width="large"),
        "Labels": st.column_config.ListColumn("Labels", width="large"),
        "Last update": st.column_config.DatetimeColumn("Last update", width="medium"),
        "URL": st.column_config.LinkColumn("🔗", display_text="🔗", width="small")
    }
)

# Section 2: Issue classification
st.header("Issue classification")
col1, col2 = st.columns(2)

## Dataframe: Number of open issues by label.
with col1:
    st.subheader("Top ten labels 🔖")
    label_counts = open_issues.groupby("Repository").apply(lambda x: x.explode("Labels").value_counts("Labels").to_frame().reset_index()).reset_index()

    def generate_labels_link(labels,repos):
        links = []
        for label,repo in zip(labels,repos):
            label = label.replace(" ", "+")
            links.append(f"https://github.com/argilla-io/{repo}/issues?q=is:open+is:issue+label:%22{label}%22")
        return links

    label_counts['Link'] = generate_labels_link(label_counts['Labels'],label_counts['Repository'])

    st.dataframe(
        label_counts[["Labels","Repository", "count","Link"]].head(10),
        hide_index=True,
        column_config={
            "Labels": st.column_config.TextColumn("Labels"),
            "count": st.column_config.NumberColumn("Count"),
            "Link": st.column_config.LinkColumn("Link", display_text="🔗")
        }
    )
    
## Cloud of words: Issue titles
with col2:
    st.subheader("Cloud of words ☁️")
    titles = " ".join(open_issues["Issue"])
    titles = re.sub(r'\[.*?\]', '', titles)
    wordcloud = WordCloud(width=800, height=400, background_color="black").generate(titles)
    plt.figure(figsize=(10, 5))
    plt.imshow(wordcloud, interpolation="bilinear")
    plt.axis("off")
    st.pyplot(plt, use_container_width=True)

# # Community engagement
st.header("Community engagement")
# ## Dataframe: Latest issues open by the community
# ## Dataframe: issues sorted by number of comments
st.subheader("Top engaging issues 💬")
engagement_df = open_issues[["Issue","Repository","Reactions","Comments","URL"]].sort_values(by=["Reactions", "Comments"], ascending=False).head(10)
st.dataframe(
    engagement_df,
    hide_index=True,
    # use_container_width=True,
    column_config={
        "Issue": st.column_config.TextColumn("Issue", width="large"),
        "Reactions": st.column_config.NumberColumn("Reactions", format="%d 👍", width="small"),
        "Comments": st.column_config.NumberColumn("Comments", format="%d 💬", width="small"),
        "URL": st.column_config.LinkColumn("🔗", display_text="🔗", width="small")
    }
)

# ## Cloud of words: Comments??
# ## Dataframe: Contributor leaderboard.

# # Issue dependencies
# st.header("Issue dependencies")
# ## Map: dependencies between issues. Network of issue mentions.x

status.update(label="Checking for updated data...", state="running")
updated_data = fetch_data()
if df.equals(updated_data):
    status.update(label="Data is up to date!", state="complete")
else:
    save_data(updated_data)
    status.update(label="Refresh for updated data!", state="complete")