improve plot layout

Author: izikeros

git_commits_graph/config.py

@@ -7,3 +7,6 @@
 
 # bmh | fivethirtyeight | seaborn-whitegrid | seaborn-darkgrid | seaborn-ticks
 DEFAULT_STYLE = "fivethirtyeight"
+DEFAULT_BACKEND = "matplotlib"
+DEFAULT_OUTPUT_FILE = "out.html"
+MAX_NUM_BARS = 200

git_commits_graph/main.py

@@ -2,9 +2,13 @@
 import git
 import matplotlib.pyplot as plt
 import pandas as pd
+from git_commits_graph.config import DEFAULT_BACKEND
+from git_commits_graph.config import DEFAULT_OUTPUT_FILE
 from git_commits_graph.config import DEFAULT_STYLE
 from git_commits_graph.plotters import plot_changes
+from git_commits_graph.plotters import plot_changes_px
 from git_commits_graph.plotters import plot_total_lines
+from git_commits_graph.plotters import plot_total_lines_px
 
 
 @click.command()
@@ -35,6 +39,14 @@
     is_flag=True,
     help="list available plot styles and exit.",
 )
+@click.option(
+    "-e",
+    "--engine",
+    is_flag=False,
+    default=DEFAULT_BACKEND,
+    help="plotting engine to use (matplitlib | plotly)",
+)
+@click.option("-o", "--output-file", help="output file name (for plotly backend)")
 def main(
     git_dir,
     branch,
@@ -44,8 +56,10 @@ def main(
     style,
     list_available_plot_styles,
     aggregate_by,
+    engine="matplotlib",
+    output_file=DEFAULT_OUTPUT_FILE,
 ):
-
+    """Plot git commits timeline main function."""
     if list_available_plot_styles:
         print(plt.style.available)
         exit()
@@ -58,6 +72,8 @@ def main(
         log_scale=log_scale,
         style=style,
         aggregate_by=aggregate_by,
+        backend=engine,
+        output_file=output_file,
     )
 
 
@@ -69,35 +85,51 @@ def git_graph(
     log_scale=False,
     style=DEFAULT_STYLE,
     aggregate_by=None,
+    backend=DEFAULT_BACKEND,
+    output_file=None,
 ):
-    # TODO: KS: 2022-06-06: Fetch commits from the github repo without cloning it.
+    """Plot git commits timeline."""
+    # TODO: KS: 2022-06-06: Fetch commits from the GitHub repo without cloning it.
     #       see: https://stackoverflow.com/a/64561416/3247880
 
     git_dir, repo = get_git_repo(
         changes=changes, git_dir=git_dir, total_lines=total_lines
     )
-    commits = fetch_commits(branch, repo)
+    commits = fetch_commits(branch, repo)  # this might take a long time
 
     plt.style.use(style)
+
+    if backend == "plotly":
+        func_total = plot_total_lines_px
+        func_changes = plot_changes_px
+    elif backend == "matplotlib":
+        func_total = plot_total_lines
+        func_changes = plot_changes
+    else:
+        raise ValueError(f"Unknown backend: {backend}")
+
     if total_lines:
-        plot_total_lines(
+        func_total(
             commits=commits,
             git_dir=git_dir,
             log_scale=log_scale,
             aggregate_by=aggregate_by,
+            output_file=output_file,
         )
         plt.show()
     if changes:
-        plot_changes(
+        func_changes(
             commits=commits,
             git_dir=git_dir,
             log_scale=log_scale,
             aggregate_by=aggregate_by,
+            output_file=output_file,
         )
         plt.show()
 
 
 def get_git_repo(changes, git_dir, total_lines):
+    """Get git repository object."""
     git_dir = git_dir.strip('"').strip("'")
     try:
         repo = git.repo.Repo(git_dir)
@@ -120,6 +152,7 @@ def get_git_repo(changes, git_dir, total_lines):
 
 
 def fetch_commits(branch, repo):
+    """Fetch commits from the git repository."""
     commits = []
     try:
         for i in reversed(list(repo.iter_commits(rev=branch))):
@@ -146,6 +179,7 @@ def fetch_commits(branch, repo):
 
 
 def prepare_commits_dataframe(commits):
+    """Make DataFrame from commits list."""
     commits = pd.DataFrame(commits, columns=["date", "added", "removed"])
     commits["delta"] = commits["added"] - commits["removed"]
     commits.date = pd.to_datetime(commits.date)

git_commits_graph/plotters.py

@@ -2,13 +2,56 @@
 import os
 from typing import Optional
 
+import matplotlib.dates as mdates
+import matplotlib.ticker as ticker
 import pandas as pd
+import plotly.graph_objs as go
 from git_commits_graph.config import FIGSIZE
+from git_commits_graph.config import MAX_NUM_BARS
 from git_commits_graph.config import XTICKS_FMT
 from matplotlib import pyplot as plt
 
 
-def plot_changes(commits, git_dir: str, log_scale: bool, aggregate_by: Optional[str]):
+# Define a function to format the y-axis labels
+def y_fmt(x, pos):
+    if x >= 1000000:
+        return f"{x * 1e-6:.0f}M"
+    elif x >= 1000:
+        return f"{x * 1e-3:.0f}k"
+    else:
+        return f"{x:.0f}"
+
+
+# Create a FuncFormatter object from the y_fmt function
+y_formatter = ticker.FuncFormatter(y_fmt)
+
+
+def data_aggregation(aggregate_by, plot_data_add, plot_data_rem):
+    if aggregate_by:
+        plot_data_add = run_aggregation(plot_data_add, col="added", period=aggregate_by)
+        plot_data_rem = run_aggregation(
+            plot_data_rem, col="removed", period=aggregate_by
+        )
+    else:
+        for agg in ["D", "W", "M", "Y"]:
+            if len(plot_data_add) > MAX_NUM_BARS:
+                plot_data_add = run_aggregation(plot_data_add, col="added", period=agg)
+                plot_data_rem = run_aggregation(
+                    plot_data_rem, col="removed", period=agg
+                )
+            if len(plot_data_add) <= MAX_NUM_BARS:
+                break
+    return plot_data_add, plot_data_rem
+
+
+def plot_changes_px(
+    commits,
+    git_dir: str,
+    log_scale: bool,
+    aggregate_by: Optional[str],
+    output_file="out.html",
+):
+    """Plot added/removed lines timeline."""
     plot_data_add = commits.added
     plot_data_rem = commits.removed
 
@@ -18,27 +61,89 @@ def plot_changes(commits, git_dir: str, log_scale: bool, aggregate_by: Optional[
         plot_data_rem = (plot_data_rem + 1).apply(math.log10)
         ylabel = "log number of lines added/removed"
 
-    if aggregate_by:
-        plot_data_add = run_aggregation(plot_data_add, col="added", period=aggregate_by)
-        plot_data_rem = run_aggregation(
-            plot_data_rem, col="removed", period=aggregate_by
+    plot_data_add, plot_data_rem = data_aggregation(
+        aggregate_by=aggregate_by,
+        plot_data_add=plot_data_add,
+        plot_data_rem=plot_data_rem,
+    )
+
+    plot_data_add = pd.DataFrame(plot_data_add)
+    plot_data_rem = pd.DataFrame(plot_data_rem)
+    fig = go.Figure()
+
+    fig.add_trace(
+        go.Bar(
+            x=plot_data_add.index,
+            y=plot_data_add.added,
+            name="added",
+            marker_color="green",
+        )
+    )
+    fig.add_trace(
+        go.Bar(
+            x=plot_data_rem.index,
+            y=-plot_data_rem.removed,
+            name="removed",
+            marker_color="red",
         )
+    )
+
+    fig.update_layout(
+        title=f"Added/Removed Lines in repo {os.path.basename(git_dir)}",
+        xaxis_title="Date",
+        yaxis_title=ylabel,
+    )
+    if log_scale:
+        fig.update_yaxes(type="log")
+    # fig.show()
+    fig.write_html(output_file)
+    print(f"Saved to {output_file}")
+
+
+def plot_changes(
+    commits, git_dir: str, log_scale: bool, aggregate_by: Optional[str], output_file
+):
+    """Plot added/removed lines timeline."""
+    plot_data_add = commits.added
+    plot_data_rem = commits.removed
+
+    ylabel = "number of lines added/removed"
+    if log_scale:
+        plot_data_add = (plot_data_add + 1).apply(math.log10)
+        plot_data_rem = (plot_data_rem + 1).apply(math.log10)
+        ylabel = "log number of lines added/removed"
+
+    plot_data_add, plot_data_rem = data_aggregation(
+        aggregate_by=aggregate_by,
+        plot_data_add=plot_data_add,
+        plot_data_rem=plot_data_rem,
+    )
 
     plot_data_add = pd.DataFrame(plot_data_add)
+    plot_data_rem = pd.DataFrame(-plot_data_rem)
     fig, ax = plt.subplots(1, 1, figsize=FIGSIZE)
 
+    locator = mdates.AutoDateLocator(minticks=3, maxticks=15)
+    x_formatter = mdates.ConciseDateFormatter(locator)
+
     ax = plot_data_add.plot(kind="bar", ax=ax, color="green", label="added")
-    ax = (-plot_data_rem).plot(kind="bar", ax=ax, color="red", label="removed")
+    ax = plot_data_rem.plot(kind="bar", ax=ax, color="red", label="removed")
+    ax.xaxis.set_major_locator(locator)
+    ax.xaxis.set_major_formatter(x_formatter)
+    ax.yaxis.set_major_formatter(y_formatter)
 
-    # plot xticks (dates)
+    plt.xticks(rotation=45)
     format_xticklabels(ax, plot_data_add)
     plt.ylabel(ylabel)
+
     ax.set_title(f"Added/Removed Lines in repo {os.path.basename(git_dir)}")
     fig.tight_layout()
 
 
-def plot_total_lines(commits, git_dir, log_scale, aggregate_by):
-    fig, ax = plt.subplots(1, 1, figsize=[8, 6])
+def plot_total_lines_px(
+    commits, git_dir, log_scale, aggregate_by, output_file="out.html"
+):
+    """Plot total lines timeline."""
 
     _delta = commits.delta
     if aggregate_by:
@@ -55,24 +160,65 @@ def plot_total_lines(commits, git_dir, log_scale, aggregate_by):
         plot_data = (plot_data + 1).apply(math.log10)
         ylabel = "log number of lines"
 
-    ax = plot_data.plot()
+    fig = go.Figure()
+    fig.add_trace(go.Scatter(x=plot_data.index, y=plot_data.values, mode="lines"))
+    fig.update_layout(
+        title=f"Number of Lines Progress in repo {os.path.basename(git_dir)}",
+        xaxis_title="Date",
+        yaxis_title=ylabel,
+    )
+    fig.update_yaxes(range=[0, 1.1 * plot_data.max()])
+    # fig.show()
+    fig.write_html(output_file)
+    print(f"Saved to {output_file}")
+
+
+def plot_total_lines(commits, git_dir, log_scale, aggregate_by, output_file):
+    """Plot total lines timeline."""
+    fig, ax = plt.subplots(1, 1, figsize=[8, 6])
+
+    _delta = commits.delta
+    _delta.index = pd.to_datetime(_delta.index, utc=True)
+    if aggregate_by:
+        _delta = _delta.reset_index()
+        _delta = _delta.groupby(pd.Grouper(key="date", axis=0, freq=aggregate_by))[
+            "delta"
+        ].sum()
+
+    plot_data = _delta.cumsum()
+
+    ylabel = "number of lines"
+    if log_scale:
+        plot_data = (plot_data + 1).apply(math.log10)
+        ylabel = "log number of lines"
+
+    ax = plot_data.plot(ax=ax, kind="line")
+    locator = mdates.AutoDateLocator(minticks=10, maxticks=15)
+    formatter = mdates.ConciseDateFormatter(locator)
+    ax.xaxis.set_major_locator(locator)
+    # ax.xaxis.set_major_formatter(mdates.AutoDateFormatter(ax.xaxis.get_major_locator()))
+    ax.xaxis.set_major_formatter(mdates.AutoDateFormatter(formatter))
+    ax.yaxis.set_major_formatter(y_formatter)
     plt.ylabel(ylabel)
     ax.set_title(f"Number of Lines Progress in repo {os.path.basename(git_dir)}")
     ax.set_ylim([0, 1.1 * plot_data.max()])
     # format_xticklabels(ax, plot_data)
     ax.xaxis_date()
     # Optional. Just rotates x-ticklabels in this case.
     fig.autofmt_xdate()
+    fig.tight_layout()
 
 
 def format_xticklabels(ax, plot_data_add, fmt=XTICKS_FMT):
+    """Format xticks labels."""
     plot_data_add = plot_data_add.reset_index()
     plot_data_add.date = pd.to_datetime(plot_data_add.date, utc=True)
     plot_data_add["xticks"] = plot_data_add.date.dt.strftime(fmt)
     ax.set_xticklabels(plot_data_add.xticks, rotation=90)
 
 
 def run_aggregation(plot_data_add, col, period):
+    """Aggregate data by period."""
     plot_data_add = plot_data_add.reset_index()
     plot_data_add.date = pd.to_datetime(plot_data_add.date, utc=True)
     plot_data_add = plot_data_add.groupby(pd.Grouper(key="date", axis=0, freq=period))[