Checking_Geodatabases_Shapefiles

'''
Title: Checking required items, defined projection, topology and geometry, attributes, field names and file size using arcpy-python for both feature classes
in the geodatabase, and also shapefiles. 
Software requirements: ArcGIS Desktop (version 10.8.2) & Python 2.7
Created by: Shahriar Rahman
Email: shahriar.env12@gmail.com
'''

import arcpy
import os

arcpy.env.overwriteOutput = True

def check_shapefile_issues(shapefile_path, report_path):
    with open(report_path, 'w') as report:
        report.write("Checking: {}\n".format(os.path.basename(shapefile_path)))
        report.write("-" * 50 + "\n")

        # Check all the associated files for the Shapefile(s)
        mandatory_extensions = ['.shp', '.shx', '.dbf']
        for ext in mandatory_extensions:
            if not arcpy.Exists(shapefile_path.replace('.shp', ext)):
                report.write("Missing mandatory file: {}\n".format(shapefile_path.replace('.shp', ext)))

        # Check the projection system
        if not arcpy.Exists(shapefile_path.replace('.shp', '.prj')):
            report.write("Warning: No .prj file found. The shapefile might not have a defined projection.\n")

        # Checking topology
        check_geometry_result = arcpy.CheckGeometry_management(shapefile_path, "in_memory/checkgeom_result")
        
        # Checking geometry issues
        fields = [f.name for f in arcpy.ListFields(check_geometry_result)]
        if "Shape" in fields:
            for row in arcpy.da.SearchCursor(check_geometry_result, ["Shape"]):
                report.write("Geometry issue detected at {}\n".format(row[0].centroid))
        else:
            report.write("No geometry issues detected.\n")

        # Check attributes
        feature_count = int(arcpy.GetCount_management(shapefile_path).getOutput(0))
        attribute_count = sum(1 for row in arcpy.SearchCursor(shapefile_path))
        if feature_count != attribute_count:
            report.write("Mismatch between features and attributes detected.\n")

        # Check for field name length (only for shapefiles)
        fields = arcpy.ListFields(shapefile_path)
        for field in fields:
            if len(field.name) > 10:
                report.write("Field name '{}' exceeds 10 characters.\n".format(field.name))

        # Check for file size (only for the shapefiles)
        if os.path.getsize(shapefile_path) > 2 * (10**9):
            report.write("Warning: Shapefile size exceeds 2GB.\n")

        report.write("\n")

def check_directory_for_shapefiles(directory_path):
    for dirpath, dirnames, filenames in os.walk(directory_path):
        for filename in filenames:
            if filename.endswith('.shp'):
                shapefile_path = os.path.join(dirpath, filename)
                report_path = os.path.join(dirpath, "{}_check_report.txt".format(filename))
                check_shapefile_issues(shapefile_path, report_path)

def check_geodatabase_for_feature_classes(gdb_path):
    arcpy.env.workspace = gdb_path
    feature_classes = arcpy.ListFeatureClasses()
    for fc in feature_classes:
        report_path = os.path.join(gdb_path, "{}_check_report.txt".format(fc))
        check_shapefile_issues(fc, report_path)

def check_directory(directory_path):
    for dirpath, dirnames, filenames in os.walk(directory_path):
        for dirname in dirnames:
            if dirname.endswith('.gdb'):
                check_geodatabase_for_feature_classes(os.path.join(dirpath, dirname))
        check_directory_for_shapefiles(dirpath)

root_directory_path = r"Your Directory" # Replace with your directory path
check_directory(root_directory_path)

print "Reports generated in the respective folders."