app.py

__copyright__ = "Copyright (C) 2023 Ali Mustapha"
__license__ = "GPL-3.0-or-later"

import gradio as gr
from get_gender import GenderPredictor
from GitScraping import CommitInfo
from get_region import RegionPredictor
import pandas as pd
import utils.plot as plot 


class GenderPredictorApp:
    def __init__(self, modelpath):
        self.gender_predictor = GenderPredictor(modelpath)
        self.Region_predictor=RegionPredictor(models_directory="saved_model/Regions")
        
        self.setup_ui()

    def setup_ui(self):
        name = gr.inputs.Textbox(label="Name")
        output = gr.outputs.Textbox(label="Predicted Gender")
        interface1_fn = gr.Interface(fn=self.predict_name, inputs=name, outputs=output, title="GitGender: Exploring Global Gender Disparities in Public Code Contributions",cache_examples=True )
        name = gr.inputs.Textbox(label="Git-url")
        pie_chart_output = gr.Plot(label="Authors by gender")
        histo_chart = gr.Plot(label="Known commits by gender")
        region_commits = gr.Plot(label="Known commits by gender")
        data_output =gr.Dataframe(headers=None,label="Contributers Details")
        interface2_fn = gr.Interface(self.predict_github_url, inputs=name, outputs=[pie_chart_output, histo_chart,region_commits,data_output], title="Determining the Geographic Origin of Public Code Contributors" )
        demo = gr.TabbedInterface([interface1_fn, interface2_fn], ["Test Model", "Exploring Diversity in GitHub Repositories"])
        self.demo = demo
                
        

    def predict_name(self, name):
        prediction, proba = self.gender_predictor.predict_gender(name)
        if prediction == 0:
            prediction = "Male with probability: " + str(proba) + "%"
        elif prediction == 1:
            prediction = "Female with probability: " + str(proba) + "%"
        else:
            prediction = "Unknown or not a name"

        prediction = name + " is " + prediction
        return prediction
    
    def predict_github_url(self, url):
        commit_info = CommitInfo(url)
        df,first_commit_dates = commit_info.get_first_commit_dates()
        first_commit_dates[['Predicted_Gender', 'Confidence']] = first_commit_dates['Author'].apply(lambda name: pd.Series(self.gender_predictor.predict_gender(name)))
        first_commit_dates['Predicted_Gender'] = first_commit_dates['Predicted_Gender'].replace({0: "Male", 1: "Female", 2: "Unknown"})
        Gender_Percentage=plot.get_gender_percentage(first_commit_dates)
        Results=first_commit_dates[first_commit_dates["Predicted_Gender"]!="Unknown"]
        Results=self.Region_predictor.get_region(Results)

        merged_df = df.merge(Results[["Author","sub-region-prediction","Predicted_Gender","Confidence"]], on=["Author"])
        # Group by Year and Predicted_Gender, then count the occurrences
        commit_per_gender_counts = merged_df.groupby(['Year', 'Predicted_Gender']).size().reset_index(name='Count')
        commits_per_gender=plot.get_commits_per_gender(commit_per_gender_counts)
        commits_per_region=plot.get_commits_per_region(merged_df,url)
        
        return Gender_Percentage,commits_per_gender,commits_per_region,Results[["Author","sub-region-prediction","Predicted_Gender","First_Commit_Date"]]
    def launch(self):
        self.demo.launch()


if __name__ == "__main__":
    modelpath = "saved_model/gender_model.tf"
    app = GenderPredictorApp(modelpath)
    app.launch()