github-repo-delta.py

#!/usr/bin/env python3
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from dateutil import parser


# Custom date parser
def custom_date_parser(date_string):
    date_string = date_string.split(" ")[0] + " " + date_string.split(" ")[1]
    return parser.parse(date_string)


# Generate this tsv file using the following command (takes a while):
# gh pr list --limit 9999999 --state all > pr.tsv

df = pd.read_csv(
    "pr.tsv",
    sep="\t",
    header=None,
    names=["pr_number", "pr_title", "branch", "status", "date"],
)
df["date"] = df["date"].apply(custom_date_parser)

df = df.sort_values("date")

df.set_index(pd.DatetimeIndex(df["date"]), inplace=True)

# Resample the DataFrame by day and count the number of PRs
prs_per_day = df.resample("D").size()

# Plotting
plt.figure(figsize=(36, 18))
sns.lineplot(data=prs_per_day, color="lightgrey")
plt.title("Rate of PR creation over time")
plt.xlabel("Date")
plt.ylabel("PRs per day")

# Calculate dates for every 10,000th PR and time to achieve each milestone
milestone_prs = df[df["pr_number"] % 10000 == 0]
milestones = pd.DataFrame(
    {"milestone": milestone_prs["pr_number"], "date": milestone_prs.index}
)
milestones["time_to_achieve"] = milestones["date"].diff()
milestones["time_to_achieve"].fillna(
    milestones.loc[milestones["milestone"] == 10000, "date"] - df.index.min(),
    inplace=True,
)

busiest_day = prs_per_day.idxmax()

milestones = pd.concat(
    [
        milestones,
        pd.DataFrame(
            {"milestone": "Busiest Day", "date": busiest_day, "time_to_achieve": None},
            index=[0],
        ),
    ],
    ignore_index=True,
)

# Draw vertical lines for milestones and annotate
for _, row in milestones.iterrows():
    color = "red"
    if row["milestone"] == "Busiest Day":
        date = str(row["date"]).split(" ")[0]
        annotation_text = f"Busiest Day ({date})-{prs_per_day.loc[busiest_day]} PRs"
        color = "green"
    else:
        annotation_text = f'PR {row["milestone"]}-{row["time_to_achieve"].days} days'
        color = "red"
    plt.axvline(x=row["date"], color=color, linestyle="--")
    plt.annotate(
        annotation_text,
        (row["date"], 0),
        xytext=(5, 0),  # Offset from the line a bit
        textcoords="offset points",
        rotation=90,
        verticalalignment="bottom",
        color="blue",
        fontsize=12,
    )

plt.savefig("pr_rate.svg", format="svg")

print(milestones)