better-git-of-theseus 0.4.2__py3-none-any.whl → 0.4.5__py3-none-any.whl

{better_git_of_theseus-0.4.2.dist-info → better_git_of_theseus-0.4.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: better-git-of-theseus
-Version: 0.4.2
+Version: 0.4.5
 Summary: Plot stats on Git repositories with interactive Plotly charts
 Home-page: https://github.com/onewesong/better-git-of-theseus
 Author: Erik Bernhardsson

{better_git_of_theseus-0.4.2.dist-info → better_git_of_theseus-0.4.5.dist-info}/RECORD

@@ -1,15 +1,15 @@
-better_git_of_theseus-0.4.2.dist-info/licenses/LICENSE,sha256=yNNDAWUe1WLKnuUcRp9X95C-yP2lfGl69m97Ftw-DUw,11345
+better_git_of_theseus-0.4.5.dist-info/licenses/LICENSE,sha256=yNNDAWUe1WLKnuUcRp9X95C-yP2lfGl69m97Ftw-DUw,11345
 git_of_theseus/__init__.py,sha256=LeG5tCOgvZMmKOjmO_HRg54sWF2K3-lTBf8H_vHMFio,273
 git_of_theseus/analyze.py,sha256=78E1G2FdSS9VZd0jKSnO5gpXwzNCjtzkSAxSzadYM3A,21547
-git_of_theseus/app.py,sha256=8aS72pMg4CyEqdYhcsz1QZSxYw8wh_iQLvb-2LOK25A,5192
-git_of_theseus/cmd.py,sha256=4W8C0tb-9Uejq4WRjGyYKXmZgE4HQBk7KFIKrozY4Og,639
+git_of_theseus/app.py,sha256=6GuBWC3WaN0VkwHAmKM_6ZNalf7NcXgXYq3ZP6XMDvY,8463
+git_of_theseus/cmd.py,sha256=kvi3sgC0ICj7PvXuzCQsijU5swS3JkfAtIT5yZRNbFo,550
 git_of_theseus/line_plot.py,sha256=LegoVy0VEFT4sM5fYCES-I_2H9UaerCopDI3J2dyHeU,3117
-git_of_theseus/plotly_plots.py,sha256=c_9rJo3qlOy-TdHPsvuDH-6hVVO0_xYq-DmOnmgqOCE,7414
+git_of_theseus/plotly_plots.py,sha256=lbLA4E0SE7Hl4TSxHcAWjdLee_yWLzURX0qAaPmZxus,8576
 git_of_theseus/stack_plot.py,sha256=q4-YlW3PyiwbIBFeHBA3dsdR1I_XKUQD74hAuSfhIR4,3150
 git_of_theseus/survival_plot.py,sha256=NEITAa0pMD9uJVsPd7JA71ucavnG1RxgC-F6Jk-K5bE,4868
 git_of_theseus/utils.py,sha256=Xw2udch9ixSgFInGhIC4_RJ_9IB3E8MmV1dmznavCWc,1026
-better_git_of_theseus-0.4.2.dist-info/METADATA,sha256=ZqXuZz94OODNr50giDYHD5Iihw5D_38undkBRXIeHDI,3633
-better_git_of_theseus-0.4.2.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
-better_git_of_theseus-0.4.2.dist-info/entry_points.txt,sha256=dEBL6oCDAozY13Y_qxS_6-qkyCA7R2TpjoLH6QJR72g,66
-better_git_of_theseus-0.4.2.dist-info/top_level.txt,sha256=2kpp8WgiBzqVLxua_mBS00Nj4cUORaRbJi121THJ_0o,15
-better_git_of_theseus-0.4.2.dist-info/RECORD,,
+better_git_of_theseus-0.4.5.dist-info/METADATA,sha256=tB4xIc-SpbNWz2ZKN1Kq8kV_Pn9ahcnw6KGj1HdyBwQ,3633
+better_git_of_theseus-0.4.5.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+better_git_of_theseus-0.4.5.dist-info/entry_points.txt,sha256=dEBL6oCDAozY13Y_qxS_6-qkyCA7R2TpjoLH6QJR72g,66
+better_git_of_theseus-0.4.5.dist-info/top_level.txt,sha256=2kpp8WgiBzqVLxua_mBS00Nj4cUORaRbJi121THJ_0o,15
+better_git_of_theseus-0.4.5.dist-info/RECORD,,

git_of_theseus/app.py

@@ -4,13 +4,26 @@ import tempfile
 import shutil
 try:
     from git_of_theseus.analyze import analyze
-    from git_of_theseus.plotly_plots import plotly_stack_plot, plotly_line_plot, plotly_survival_plot
+    from git_of_theseus.plotly_plots import plotly_stack_plot, plotly_line_plot, plotly_survival_plot, plotly_bar_plot
 except ImportError:
     from analyze import analyze
-    from plotly_plots import plotly_stack_plot, plotly_line_plot, plotly_survival_plot
+    from plotly_plots import plotly_stack_plot, plotly_line_plot, plotly_survival_plot, plotly_bar_plot
 
 st.set_page_config(page_title="Git of Theseus Dash", layout="wide")
 
+# GitHub Link in Sidebar
+st.sidebar.markdown(
+    """
+    <div style="display: flex; align-items: center; margin-bottom: 20px;">
+        <img src="https://github.githubassets.com/images/modules/logos_page/GitHub-Mark.png" width="30" style="margin-right: 10px;">
+        <a href="https://github.com/onewesong/better-git-of-theseus" target="_blank" style="text-decoration: none; color: inherit; font-weight: bold;">
+            better-git-of-theseus
+        </a>
+    </div>
+    """,
+    unsafe_allow_html=True
+)
+
 st.title("📊 Git of Theseus - Repository Analysis")
 
 import sys
@@ -18,12 +31,49 @@ import sys
 # Sidebar Configuration
 st.sidebar.header("Configuration")
 
+with st.sidebar.expander("📖 How to use", expanded=False):
+    st.markdown("""
+    **Better Git of Theseus** is a tool to analyze the evolution of Git repositories.
+    
+    ### Plots Explained:
+    - **Stack Plot**: Shows code growth over time, broken down by cohort (when code was added).
+    - **Line Plot**: Shows trends across different dimensions (Author, Extension, etc.).
+    - **Distribution**: Shows the **current** distribution (Who contributed most, which file types are dominant).
+    - **Survival Plot**: Estimates how long a line of code typically lasts before being modified or deleted.
+    
+    ### Tips:
+    - **Cohort Format**: `%Y` (Yearly) and `%Y-%m` (Monthly) are recommended.
+    - **Mailmap**: Use a `.mailmap` file in the repo root to resolve duplicate author names.
+    """)
+
 default_repo = "."
 if len(sys.argv) > 1:
     default_repo = sys.argv[1]
 
-repo_path = st.sidebar.text_input("Git Repository Path", value=default_repo)
-branch = st.sidebar.text_input("Branch", value="master")
+repo_path = default_repo
+# Path display removed as per user request
+
+# Fetch branches for the selectbox
+try:
+    import git
+    repo = git.Repo(repo_path)
+    # Get local branches
+    branches = [h.name for h in repo.heads]
+    
+    # Try to determine the best default branch (active one, or master/main)
+    try:
+        current_active = repo.active_branch.name
+    except:
+        current_active = "master"
+        
+    if current_active in branches:
+        branches.remove(current_active)
+    
+    options = [current_active] + sorted(branches)
+    branch = st.sidebar.selectbox("Branch", options=options)
+except Exception as e:
+    # Fallback if git repo access fails
+    branch = st.sidebar.text_input("Branch", value="master")
 
 with st.sidebar.expander("Analysis Parameters"):
     cohortfm = st.text_input(
@@ -35,9 +85,23 @@ with st.sidebar.expander("Analysis Parameters"):
              "- `%Y-W%W`: Week (e.g., 2023-W01)\n"
              "- `%Y-%m-%d`: Day"
     )
-    interval = st.number_input("Interval (seconds)", value=7 * 24 * 60 * 60)
-    procs = st.number_input("Processes", value=2, min_value=1)
-    ignore = st.text_area("Ignore (comma separated)").split(",")
+    interval = st.number_input(
+        "Analysis Interval (seconds)", 
+        value=7 * 24 * 60 * 60,
+        help="The time step between data points. Default is 604800s (7 days). Larger values are faster; smaller values result in smoother curves."
+    )
+    st.caption(f"Current resolution: {interval / 86400:.1f} days")
+    
+    procs = st.number_input(
+        "Parallel Processes", 
+        value=2, 
+        min_value=1,
+        help="Number of concurrent processes. Increase to speed up analysis on multi-core CPUs, but note it increases RAM usage."
+    )
+    ignore = st.text_area(
+        "Ignore Patterns",
+        help="Glob patterns to ignore (comma separated), e.g.: 'tests/**, *.md'"
+    ).split(",")
     ignore = [i.strip() for i in ignore if i.strip()]
 
 @st.cache_data(show_spinner=False)
@@ -71,7 +135,7 @@ if st.sidebar.button("🚀 Run Analysis") or (len(sys.argv) > 1 and st.session_s
 # Main View
 if st.session_state.analysis_results:
     results = st.session_state.analysis_results
-    tab1, tab2, tab3 = st.tabs(["Stack Plot", "Line Plot", "Survival Plot"])
+    tab1, tab2, tab3, tab4 = st.tabs(["Stack Plot", "Line Plot", "Distribution", "Survival Plot"])
 
     with tab1:
         st.header("Stack Plot")
@@ -115,6 +179,22 @@ if st.session_state.analysis_results:
                 st.warning(f"Data for {data_source_label_line} not found.")
 
     with tab3:
+        st.header("Latest Distribution")
+        col1, col2 = st.columns([1, 3])
+        with col1:
+            data_source_label_bar = st.selectbox("Data Source", list(source_map.keys()), key="bar_source")
+            data_key_bar = source_map[data_source_label_bar]
+            max_n_bar = st.slider("Max Series", 5, 100, 30, key="bar_max_n")
+        with col2:
+            project_name = os.path.basename(os.path.abspath(repo_path))
+            data_bar = results.get(data_key_bar)
+            if data_bar:
+                fig = plotly_bar_plot(data_bar, max_n=max_n_bar, title=f"{project_name} - {data_source_label_bar}")
+                st.plotly_chart(fig, width="stretch")
+            else:
+                st.warning(f"Data for {data_source_label_bar} not found.")
+
+    with tab4:
         st.header("Survival Plot")
         col1, col2 = st.columns([1, 3])
         with col1:

git_of_theseus/cmd.py

@@ -7,9 +7,8 @@ def main():
     cmd_dir = os.path.dirname(os.path.abspath(__file__))
     app_path = os.path.join(cmd_dir, "app.py")
     
-    # The first argument is the repo path, default to current directory
-    repo_path = sys.argv[1] if len(sys.argv) > 1 else os.getcwd()
-    repo_path = os.path.abspath(repo_path)
+    # Always use the current working directory
+    repo_path = os.path.abspath(os.getcwd())
     
     # Run streamlit
     # We pass the repo_path as an argument to the streamlit script

git_of_theseus/plotly_plots.py

@@ -240,4 +240,37 @@ def plotly_survival_plot(commit_history, exp_fit=False, years=5, title=None):
     )
 
 
+    return fig
+
+def plotly_bar_plot(data, max_n=20, title=None):
+    ts, y, labels = _process_stack_line_data(data, max_n, normalize=False)
+    
+    # Get latest data point (current state)
+    latest_values = [row[-1] for row in y]
+    
+    # Sort by value for better bar chart presentation
+    # (Though _process_stack_line_data already does some sorting, we want descending order)
+    indices = sorted(range(len(labels)), key=lambda i: latest_values[i], reverse=True)
+    
+    sorted_labels = [labels[i] for i in indices]
+    sorted_values = [latest_values[i] for i in indices]
+    
+    # Generate colors
+    colors = px.colors.qualitative.Plotly
+    if len(sorted_labels) > len(colors):
+        colors = px.colors.qualitative.Dark24
+
+    fig = go.Figure(go.Bar(
+        x=sorted_labels,
+        y=sorted_values,
+        marker_color=[colors[i % len(colors)] for i in range(len(sorted_labels))]
+    ))
+
+    fig.update_layout(
+        title=dict(text=f"{title} (Current Distribution)" if title else "Current Distribution", x=0.5),
+        yaxis=dict(title="Lines of Code"),
+        xaxis=dict(title=""),
+        margin=dict(l=20, r=20, t=50, b=100),
+    )
+    
     return fig