app.py

import gradio as gr
import pandas as pd
import numpy as np
import json
from io import StringIO
from collections import OrderedDict





#   define variables

#   distance matrices
branch_name_dm = "graph_geometry/distance_matrix"
commit_id_dm = "cfde6f4ba4"   # ebcfc50abe/commits/cfde6f4ba4
dm_activityNodes = "activity_node+distance_matrix_ped_mm_noEntr"
dm_transportStops = "an_stations+distance_matrix_ped_mm_art_noEntr"

#   land use attributes
branch_name_lu = "graph_geometry/activity_nodes_with_land_use"
commit_id_lu = "13ae6cdd30"

#   livability attributes
notion_lu_domains = "407c2fce664f4dde8940bb416780a86d"
notion_domain_attributes = "01401b78420f4296a2449f587d4ed9c9"






def test(input_json):
    print("Received input")
    # Parse the input JSON string
    try:
        inputs = json.loads(input_json)
    except json.JSONDecodeError:
        inputs = json.loads(input_json.replace("'", '"'))


    
    # Accessing input data from Grasshopper
    
    matrix = inputs['input']["matrix"]
    landuses = inputs['input']["landuse_areas"]
    attributeMapperDict = inputs['input']["attributeMapperDict"]
    landuseMapperDict = inputs['input']["landuseMapperDict"]
    
    alpha = inputs['input']["alpha"]
    alpha = float(alpha)
    threshold = inputs['input']["threshold"]
    threshold = float(threshold)
    
    df_matrix = pd.DataFrame(matrix).T
    df_landuses = pd.DataFrame(landuses).T
    df_matrix = df_matrix.round(0).astype(int)
    df_landuses = df_landuses.round(0).astype(int)




    
    # create a mask based on the matrix size and ids, crop activity nodes to the mask
    mask_connected = df_matrix.index.tolist()

    valid_indexes = [idx for idx in mask_connected if idx in df_landuses.index]
    # Identify and report missing indexes
    missing_indexes = set(mask_connected) - set(valid_indexes)
    if missing_indexes:
        print(f"Error: The following indexes were not found in the DataFrame: {missing_indexes}, length: {len(missing_indexes)}")
    
    # Apply the filtered mask
    df_landuses_filtered = df_landuses.loc[valid_indexes]


    # find a set of unique domains, to which subdomains are aggregated    
    temp = []
    for key, values in attributeMapperDict.items():
      domain = attributeMapperDict[key]['domain']
      for item in domain:
        if ',' in item:
          domain_list = item.split(',')
          attributeMapperDict[key]['domain'] = domain_list
          for domain in domain_list:
            temp.append(domain) 
        else:
          if item != 0: 
              temp.append(item)  
    
    domainsUnique = list(set(temp))


    # find a list of unique subdomains, to which land uses are aggregated
    temp = []    
    for key, values in landuseMapperDict.items():
      subdomain = str(landuseMapperDict[key])
      if subdomain != 0: 
        temp.append(subdomain) 
        
    subdomainsUnique = list(set(temp))
    

    
    
    def landusesToSubdomains(DistanceMatrix, LanduseDf, LanduseToSubdomainDict, UniqueSubdomainsList):
        df_LivabilitySubdomainsArea = pd.DataFrame(0, index=DistanceMatrix.index, columns=UniqueSubdomainsList)
    
        for subdomain in UniqueSubdomainsList:
            for lu, lu_subdomain in LanduseToSubdomainDict.items():
                if lu_subdomain == subdomain:
                    if lu in LanduseDf.columns:
                        df_LivabilitySubdomainsArea[subdomain] = df_LivabilitySubdomainsArea[subdomain].add(LanduseDf[lu], fill_value=0)
                    else:
                        print(f"Warning: Column '{lu}' not found in landuse database")
    
        return df_LivabilitySubdomainsArea

    

    LivabilitySubdomainsWeights = landusesToSubdomains(df_matrix,df_landuses_filtered,landuseMapperDict,subdomainsUnique)
    



    
    def FindWorkplaces (DistanceMatrix,SubdomainAttributeDict,destinationWeights,UniqueSubdomainsList ):
        
        df_LivabilitySubdomainsWorkplaces = pd.DataFrame(0, index=DistanceMatrix.index, columns=['jobs'])
    
        for domain in UniqueSubdomainsList:
          for key, value_list in SubdomainAttributeDict.items():
            sqm_per_empl = float(SubdomainAttributeDict[domain]['sqmPerEmpl'][0])
            if key in destinationWeights.columns and key == domain:
              if sqm_per_empl > 0:
                df_LivabilitySubdomainsWorkplaces['jobs'] += (round(destinationWeights[key] / sqm_per_empl,2)).fillna(0)
              else:
                df_LivabilitySubdomainsWorkplaces['jobs'] += 0
    
        return df_LivabilitySubdomainsWorkplaces


    WorkplacesNumber = FindWorkplaces(df_matrix,attributeMapperDict,LivabilitySubdomainsWeights,subdomainsUnique)
    
    # prepare an input weights dataframe for the parameter LivabilitySubdomainsInputs
    LivabilitySubdomainsInputs =pd.concat([LivabilitySubdomainsWeights, WorkplacesNumber], axis=1)

    
    
    def computeAccessibility (DistanceMatrix, destinationWeights=None,alpha = 0.0038, threshold = 600):
    
        decay_factors = np.exp(-alpha * DistanceMatrix) * (DistanceMatrix <= threshold)
        subdomainsAccessibility = pd.DataFrame(index=DistanceMatrix.index, columns=destinationWeights.columns)
        # for weighted accessibility (e. g. areas)
        if not destinationWeights.empty:
            for col in destinationWeights.columns:
                subdomainsAccessibility[col] = (decay_factors * destinationWeights[col].values).sum(axis=1)
        # for unweighted accessibility (e. g. points of interest)
        else:
            for col in DistanceMatrix.columns:
                subdomainsAccessibility[col] = (decay_factors * 1).sum(axis=1)
        
        return subdomainsAccessibility
    
    subdomainsAccessibility = computeAccessibility(df_matrix,LivabilitySubdomainsInputs,alpha,threshold)




    def remap(value, B_min, B_max, C_min, C_max):
        return C_min + (((value - B_min) / (B_max - B_min))* (C_max - C_min))    


    if 'jobs' not in subdomainsAccessibility.columns:
        print("Error: Column 'jobs' does not exist in the subdomainsAccessibility.")

    
    
    def accessibilityToLivability (DistanceMatrix,subdomainsAccessibility, SubdomainAttributeDict,UniqueDomainsList):
    
        livability = pd.DataFrame(index=DistanceMatrix.index, columns=subdomainsAccessibility.columns)
        livability.drop(columns='jobs', inplace=True)
        livability["Workplaces"] = 0
        livability.fillna(0, inplace=True)
        
             
        for domain in UniqueDomainsList:
            livability[domain] = 0
    
    
        # remap accessibility to livability points
        
        for key, values in SubdomainAttributeDict.items():
            if key == 'commercial':
                threshold = float(SubdomainAttributeDict[key]['thresholds'])
                max_livability = float(SubdomainAttributeDict[key]['max_points'])
                livability_score = remap(subdomainsAccessibility['jobs'], 0, threshold, 0, max_livability)
                livability.loc[subdomainsAccessibility['jobs'] >= threshold, 'Workplaces'] = max_livability
                livability.loc[subdomainsAccessibility['jobs'] < threshold, 'Workplaces'] = livability_score
            elif key in subdomainsAccessibility.columns and key != 'commercial':
                domain = [str(item) for item in SubdomainAttributeDict[key]['domain']]
                threshold = float(SubdomainAttributeDict[key]['thresholds'])
                max_livability = float(SubdomainAttributeDict[key]['max_points'])
                sqm_per_employee = SubdomainAttributeDict[key]['sqmPerEmpl']
                
                livability_score = remap(subdomainsAccessibility[key], 0, threshold, 0, max_livability)
                livability.loc[subdomainsAccessibility[key] >= threshold, key] = max_livability
                livability.loc[subdomainsAccessibility[key] < threshold, key] = livability_score
                if any(domain):
                    for item in domain:
                        if domain != 'Workplaces':
                            livability.loc[subdomainsAccessibility[key] >= threshold, item] += max_livability
                            livability.loc[subdomainsAccessibility[key] < threshold, item] += livability_score
            

        return livability
    
    
    

    livability = accessibilityToLivability(df_matrix,subdomainsAccessibility,attributeMapperDict,domainsUnique)
    

    livability_dictionary = livability.to_dict('index')
    LivabilitySubdomainsInputs_dictionary = LivabilitySubdomainsInputs.to_dict('index')
    subdomainsAccessibility_dictionary = subdomainsAccessibility.to_dict('index')
    
    # Prepare the output
    output = {
        "subdomainsAccessibility_dictionary": subdomainsAccessibility_dictionary,
        "livability_dictionary": livability_dictionary,
        "subdomainsWeights_dictionary": LivabilitySubdomainsInputs_dictionary
    }


    
    return json.dumps(output)

    # Define the Gradio interface with a single JSON input
iface = gr.Interface(
    fn=test,
    inputs=gr.Textbox(label="Input JSON", lines=20, placeholder="Enter JSON with all parameters here..."),
    outputs=gr.JSON(label="Output JSON"),
    title="testspace"
)

iface.launch()