data-seed-ph 3.2.2__tar.gz → 3.4.2__tar.gz

{data_seed_ph-3.2.2/data_seed_ph.egg-info → data_seed_ph-3.4.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: data-seed-ph
-Version: 3.2.2
+Version: 3.4.2
 Summary: A Python library for generating realistic, synthetic Philippine-based datasets.
 Author-email: "Christian G. Garcia" <iyaniyan03112003@gmail.com>
 License-Expression: MIT
@@ -40,4 +40,4 @@ Dynamic: license-file
 >[!NOTE]
 > Designed for database seeding, API testing, machine learning dataset generation, academic research, and software development prototyping with support for CSV, JSON and SQL export formats.
 
-View full documentation: [Github](https://github.com/christiangarcia0311/data-seed-ph)
+[View full documentation](https://github.com/christiangarcia0311/data-seed-ph)

{data_seed_ph-3.2.2 → data_seed_ph-3.4.2}/README.md

@@ -15,4 +15,4 @@
 >[!NOTE]
 > Designed for database seeding, API testing, machine learning dataset generation, academic research, and software development prototyping with support for CSV, JSON and SQL export formats.
 
-View full documentation: [Github](https://github.com/christiangarcia0311/data-seed-ph)
+[View full documentation](https://github.com/christiangarcia0311/data-seed-ph)

{data_seed_ph-3.2.2 → data_seed_ph-3.4.2/data_seed_ph.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: data-seed-ph
-Version: 3.2.2
+Version: 3.4.2
 Summary: A Python library for generating realistic, synthetic Philippine-based datasets.
 Author-email: "Christian G. Garcia" <iyaniyan03112003@gmail.com>
 License-Expression: MIT
@@ -40,4 +40,4 @@ Dynamic: license-file
 >[!NOTE]
 > Designed for database seeding, API testing, machine learning dataset generation, academic research, and software development prototyping with support for CSV, JSON and SQL export formats.
 
-View full documentation: [Github](https://github.com/christiangarcia0311/data-seed-ph)
+[View full documentation](https://github.com/christiangarcia0311/data-seed-ph)

{data_seed_ph-3.2.2 → data_seed_ph-3.4.2}/data_seed_ph.egg-info/SOURCES.txt

@@ -21,7 +21,4 @@ seed/providers/data/luzon_locations.json
 seed/providers/data/mindanao_locations.json
 seed/providers/data/preferences.json
 seed/providers/data/source.md
-seed/providers/data/visayas_locations.json
-tests/test_core.py
-tests/test_exporter.py
-tests/test_providers.py
+seed/providers/data/visayas_locations.json

{data_seed_ph-3.2.2 → data_seed_ph-3.4.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "data-seed-ph"
-version = "3.2.2"
+version = "3.4.2"
 description = "A Python library for generating realistic, synthetic Philippine-based datasets."
 readme = "README.md"
 requires-python = ">=3.9"

{data_seed_ph-3.2.2 → data_seed_ph-3.4.2}/seed/__init__.py

@@ -11,5 +11,5 @@ Information
 
 from .core import Dataset
 
-__version__ = '3.2.1'
+__version__ = '3.4.0'
 __all__ = ['Dataset']

{data_seed_ph-3.2.2 → data_seed_ph-3.4.2}/seed/core.py

@@ -34,6 +34,7 @@ from .providers.data_provider import (
 import pandas as pd
 import numpy as np
 import random as rd
+import os
 
     
 
@@ -89,6 +90,12 @@ class Dataset:
             'strand': lambda: self.data.education.strand()
         }
         
+    def _reset_providers(self):
+        self.data.personal = PersonalDataProvider()
+        self.data.address = AddressDataProvider()
+        self.data.contact = ContactDataProvider(self.data.personal)
+        self.data.education = EducationDataProvider()
+        
     def generate(self, rows: int, features: dict) -> pd.DataFrame:
         self.dump = {feature: [] for feature in features}
 
@@ -132,6 +139,33 @@ class Dataset:
 
         return pd.DataFrame(self.dump)
     
+    '''
+    v.3.4.2: [feat #1] Batch Generation with Variations
+    ----------------------------------------------------
+    '''
+    
+    def generate_batch(self, configurations: list) -> dict:
+        batch_results: dict = {}
+        
+        for index, config in enumerate(configurations):
+            
+            if not isinstance(config, dict): raise ValueError(f'Configuration at index {index} must be a dictionary format.')
+            if 'rows' not in config or 'features' not in config: raise ValueError(f'Configuration at index {index} must contain `features` keys and `rows`')
+            
+            batch_name = config.get('name', f'batch_{index + 1}')
+            
+            rows = config['rows']
+            features = config['features']  
+            
+            if not isinstance(rows, int) or rows <= 0: raise ValueError(f'Batch {batch_name} must be a postive integer.')
+            if not isinstance(features, dict) or not features: raise ValueError(f'Batch {batch_name}: features cannot be empty dictionary.')
+            
+            self._reset_providers()
+            
+            batch_results[batch_name] = self.generate(rows=rows, features=features)
+            
+        return batch_results
+    
     def save(
         self, 
         data: dict, 
@@ -182,10 +216,27 @@ class Dataset:
                     raise ValueError(f'Unsupported database type: {database_type}')
             case _:
                 raise ValueError(f'Unsupported format: {format}')
-
-
-'''
-For future implementation
-
-@author: Christian G. Garcia
-'''
+    
+    def save_batch(
+        self, 
+        batch_data: dict, 
+        format: str, 
+        output_dir: str = './output', 
+        **kwargs
+    ):
+        os.makedirs(output_dir, exist_ok=True)
+        
+        for batch_name, data in batch_data.items():
+            filename = os.path.join(output_dir, batch_name)
+            
+            try:
+                self.save(
+                    data=data,
+                    filename=filename,
+                    format=format,
+                    **kwargs
+                )
+                
+                print(f'Batch {batch_name} successfully saved.')
+            except Exception as e:
+                print(f'Error saving batch {batch_name}: {e}')

data_seed_ph-3.2.2/tests/test_core.py

@@ -1,186 +0,0 @@
-import pytest
-import pandas as pd
-import tempfile
-import os
-from seed.core import Dataset, Data
-from seed.providers.data_provider import PersonalDataProvider, AddressDataProvider, ContactDataProvider
-from seed.exports.exporter import SaveData
-
-class TestData:
-    def test_data_initialization(self):
-        data = Data()
-        assert isinstance(data.personal, PersonalDataProvider)
-        assert isinstance(data.address, AddressDataProvider)
-        assert isinstance(data.contact, ContactDataProvider)
-
-class TestDataset:
-    def setup_method(self):
-        self.dataset = Dataset()
-    
-    def test_dataset_initialization(self):
-        assert isinstance(self.dataset.export, SaveData)
-        assert isinstance(self.dataset.data, Data)
-        assert isinstance(self.dataset.dump, dict)
-        assert len(self.dataset.data_map) > 0
-    
-    def test_generate_basic(self):
-        features = {'firstname': 'firstname', 'lastname': 'lastname'}
-        result = self.dataset.generate(5, features)
-        assert isinstance(result, pd.DataFrame)
-        assert len(result) == 5
-        assert 'firstname' in result.columns
-        assert 'lastname' in result.columns
-    
-    def test_generate_with_personal_data(self):
-        features = {
-            'full_name': 'fullname',
-            'birthdate': 'birthdate',
-            'age': 'age'
-        }
-        result = self.dataset.generate(10, features)
-        assert len(result) == 10
-        for col in features.keys():
-            assert col in result.columns
-    
-    def test_generate_with_address_data(self):
-        features = {
-            'region': 'region',
-            'province': 'province',
-            'city': 'city'
-        }
-        result = self.dataset.generate(5, features)
-        assert len(result) == 5
-        assert 'region' in result.columns
-
-    def test_generate_with_all_personal_fields(self):
-        features = {
-            'first': 'firstname',
-            'last': 'lastname',
-            'suffix': 'suffixname',
-            'male': 'malename',
-            'female': 'femalename',
-            'civil': 'civil',
-            'religion': 'religion',
-            'blood': 'bloodtype',
-            'nationality': 'nationality'
-        }
-        result = self.dataset.generate(3, features)
-        assert len(result) == 3
-
-    def test_generate_with_island_groups(self):
-        features = {
-            'luzon_region': 'luzon',
-            'visayas_region': 'visayas',
-            'mindanao_region': 'mindanao'
-        }
-        result = self.dataset.generate(2, features)
-        assert len(result) == 2
-
-    def test_generate_with_address_variants(self):
-        features = {
-            'municipality': 'municipality',
-            'barangay': 'barangay',
-            'full_addr': 'fulladdress',
-            'mobile': 'mobile',
-            'mobile_intl': 'mobile_int',
-            'email': 'email'
-        }
-        result = self.dataset.generate(3, features)
-        assert len(result) == 3
-
-    def test_generate_with_parameterized_province(self):
-        features = {
-            'province': 'province:Surigao Del Norte'
-        }
-        result = self.dataset.generate(2, features)
-        assert len(result) == 2
-
-    def test_generate_with_int_range(self):
-        features = {
-            'score': (10, 100)
-        }
-        result = self.dataset.generate(5, features)
-        assert len(result) == 5
-        assert all(10 <= val <= 100 for val in result['score'])
-
-    def test_generate_with_float_range(self):
-        features = {
-            'price': (10.5, 99.99, 'float')
-        }
-        result = self.dataset.generate(5, features)
-        assert len(result) == 5
-        assert all(isinstance(val, float) for val in result['price'])
-
-    def test_generate_with_list_choices(self):
-        features = {
-            'status': ['active', 'inactive', 'pending']
-        }
-        result = self.dataset.generate(5, features)
-        assert len(result) == 5
-        assert all(val in ['active', 'inactive', 'pending'] for val in result['status'])
-
-    def test_generate_invalid_feature_type(self):
-        features = {
-            'invalid': 12345
-        }
-        with pytest.raises(ValueError):
-            self.dataset.generate(2, features)
-
-    def test_generate_invalid_feature_keyword(self):
-        features = {
-            'bad': 'nonexistent'
-        }
-        with pytest.raises(ValueError):
-            self.dataset.generate(2, features)
-
-    def test_generate_invalid_parameterized_keyword(self):
-        features = {
-            'bad': 'invalid:param'
-        }
-        with pytest.raises(ValueError):
-            self.dataset.generate(2, features)
-
-    def test_save_csv(self):
-        features = {'name': 'firstname'}
-        self.dataset.generate(5, features)
-        temp_dir = tempfile.mkdtemp()
-        try:
-            self.dataset.save(self.dataset.dump, os.path.join(temp_dir, 'test'), 'csv')
-            assert os.path.exists(os.path.join(temp_dir, 'test.csv'))
-        finally:
-            for f in os.listdir(temp_dir):
-                os.remove(os.path.join(temp_dir, f))
-            os.rmdir(temp_dir)
-
-    def test_save_json(self):
-        features = {'name': 'firstname'}
-        self.dataset.generate(3, features)
-        temp_dir = tempfile.mkdtemp()
-        try:
-            self.dataset.save(self.dataset.dump, os.path.join(temp_dir, 'test'), 'json')
-            assert os.path.exists(os.path.join(temp_dir, 'test.json'))
-        finally:
-            for f in os.listdir(temp_dir):
-                os.remove(os.path.join(temp_dir, f))
-            os.rmdir(temp_dir)
-
-    def test_save_sql(self):
-        features = {'name': 'firstname'}
-        self.dataset.generate(2, features)
-        temp_dir = tempfile.mkdtemp()
-        try:
-            db_path = os.path.join(temp_dir, 'test.db')
-            self.dataset.save(
-                self.dataset.dump,
-                db_path,
-                'sql',
-                table_name='test_table',
-                database_name=db_path
-            )
-            assert os.path.exists(db_path)
-        finally:
-            for f in os.listdir(temp_dir):
-                path = os.path.join(temp_dir, f)
-                if os.path.isfile(path):
-                    os.remove(path)
-            os.rmdir(temp_dir)

data_seed_ph-3.2.2/tests/test_exporter.py

@@ -1,75 +0,0 @@
-import pytest
-import os
-import tempfile
-import pandas as pd
-from seed.exports.exporter import SaveData
-
-class TestSaveData:
-    def setup_method(self):
-        self.saver = SaveData()
-        self.temp_dir = tempfile.mkdtemp()
-    
-    def teardown_method(self):
-        for file in os.listdir(self.temp_dir):
-            filepath = os.path.join(self.temp_dir, file)
-            if os.path.isfile(filepath):
-                os.remove(filepath)
-        os.rmdir(self.temp_dir)
-    
-    def test_to_csv_with_dict(self):
-        data = {'name': ['John', 'Jane'], 'age': [25, 30]}
-        filepath = os.path.join(self.temp_dir, 'test.csv')
-        self.saver.to_csv(data, filepath)
-        assert os.path.exists(filepath)
-    
-    def test_to_csv_with_dataframe(self):
-        data = pd.DataFrame({'name': ['John', 'Jane'], 'age': [25, 30]})
-        filepath = os.path.join(self.temp_dir, 'test.csv')
-        self.saver.to_csv(data, filepath)
-        assert os.path.exists(filepath)
-    
-    def test_to_json_with_dict(self):
-        data = {'name': ['John', 'Jane'], 'age': [25, 30]}
-        filepath = os.path.join(self.temp_dir, 'test.json')
-        self.saver.to_json(data, filepath)
-        assert os.path.exists(filepath)
-    
-    def test_to_json_with_dataframe(self):
-        data = pd.DataFrame({'name': ['John', 'Jane'], 'age': [25, 30]})
-        filepath = os.path.join(self.temp_dir, 'test.json')
-        self.saver.to_json(data, filepath)
-        assert os.path.exists(filepath)
-    
-    def test_invalid_data_type(self, capsys):
-        self.saver.to_csv([1, 2, 3], os.path.join(self.temp_dir, 'test.csv'))
-        captured = capsys.readouterr()
-        assert 'Data must be a dictionary or a pandas DataFrame.' in captured.out
-
-    def test_to_sql_sqlite(self):
-        data = {'name': ['John', 'Jane'], 'age': [25, 30]}
-        db_path = os.path.join(self.temp_dir, 'test.db')
-        self.saver.to_sql(data, 'users', 'sqlite', db_path)
-        assert os.path.exists(db_path)
-
-    def test_to_sql_with_dataframe(self):
-        data = pd.DataFrame({'name': ['John', 'Jane'], 'age': [25, 30]})
-        db_path = os.path.join(self.temp_dir, 'test.db')
-        self.saver.to_sql(data, 'users', 'sqlite', db_path)
-        assert os.path.exists(db_path)
-
-    def test_to_sql_custom_primary_key(self):
-        data = {'user_id': [1, 2], 'name': ['John', 'Jane']}
-        db_path = os.path.join(self.temp_dir, 'test.db')
-        self.saver.to_sql(data, 'users', 'sqlite', db_path, primary_key='user_id')
-        assert os.path.exists(db_path)
-
-    def test_to_json_invalid_data(self, capsys):
-        self.saver.to_json([1, 2, 3], os.path.join(self.temp_dir, 'test.json'))
-        captured = capsys.readouterr()
-        assert 'Data must be a dictionary or a pandas DataFrame.' in captured.out
-
-    def test_to_sql_invalid_database_type(self):
-        data = {'name': ['John', 'Jane'], 'age': [25, 30]}
-        db_path = os.path.join(self.temp_dir, 'test.db')
-        with pytest.raises(ValueError):
-            self.saver.to_sql(data, 'users', 'postgresql', db_path)

data_seed_ph-3.2.2/tests/test_providers.py

@@ -1,258 +0,0 @@
-import pytest
-from seed.providers.data_provider import PersonalDataProvider, AddressDataProvider, ContactDataProvider
-
-class TestPersonalDataProvider:
-    def setup_method(self):
-        self.provider = PersonalDataProvider()
-    
-    def test_firstname_generation(self):
-        name = self.provider.firstname()
-        assert isinstance(name, str)
-        assert len(name) > 0
-    
-    def test_firstname_by_gender(self):
-        male_name = self.provider.firstname(gender='male')
-        female_name = self.provider.firstname(gender='female')
-        assert isinstance(male_name, str)
-        assert isinstance(female_name, str)
-    
-    def test_lastname_generation(self):
-        name = self.provider.lastname()
-        assert isinstance(name, str)
-        assert len(name) > 0
-    
-    def test_fullname_generation(self):
-        fullname = self.provider.fullname()
-        assert isinstance(fullname, str)
-        assert len(fullname) > 0
-    
-    def test_gender_generation(self):
-        self.provider.firstname()
-        gender = self.provider.gender()
-        assert gender in ['male', 'female']
-    
-    def test_birthdate_generation(self):
-        date = self.provider.birthdate()
-        assert isinstance(date, str)
-        assert len(date) > 0
-    
-    def test_suffix_generation(self):
-        suffix = self.provider.suffix()
-        assert isinstance(suffix, str)
-
-    def test_age_generation(self):
-        self.provider.birthdate()
-        age = self.provider.age()
-        assert isinstance(age, int)
-        assert age >= 12
-
-    def test_age_without_birthdate(self):
-        provider = PersonalDataProvider()
-        with pytest.raises(ValueError):
-            provider.age()
-
-    def test_civil_generation(self):
-        civil = self.provider.civil()
-        assert isinstance(civil, str)
-
-    def test_religion_generation(self):
-        religion = self.provider.religion()
-        assert isinstance(religion, str)
-
-    def test_blood_type_generation(self):
-        blood = self.provider.blood_type()
-        assert isinstance(blood, str)
-
-    def test_nationality_generation(self):
-        nationality = self.provider.nationality()
-        assert isinstance(nationality, str)
-
-    def test_fullname_male(self):
-        fullname = self.provider.fullname(gender='male')
-        assert isinstance(fullname, str)
-        assert len(fullname.split()) >= 2
-
-    def test_fullname_female(self):
-        fullname = self.provider.fullname(gender='female')
-        assert isinstance(fullname, str)
-        assert len(fullname.split()) >= 2
-
-    def test_birthdate_custom_age_range(self):
-        date = self.provider.birthdate(min_age=18, max_age=65)
-        assert isinstance(date, str)
-
-class TestAddressDataProvider:
-    def setup_method(self):
-        self.provider = AddressDataProvider()
-    
-    def test_region_generation(self):
-        region = self.provider.region()
-        assert isinstance(region, str)
-        assert len(region) > 0
-    
-    def test_province_generation(self):
-        province = self.provider.province()
-        assert isinstance(province, str)
-    
-    def test_city_generation(self):
-        city = self.provider.city()
-        assert isinstance(city, str)
-
-    def test_municipality_generation(self):
-        municipality = self.provider.municipality()
-        assert isinstance(municipality, str) or municipality is None
-
-    def test_barangay_generation(self):
-        barangay = self.provider.barangay()
-        assert isinstance(barangay, str) or barangay is None
-
-    def test_full_address_generation(self):
-        address = self.provider.full_address()
-        assert isinstance(address, str)
-        assert len(address) > 0
-
-    def test_region_with_island_group_luzon(self):
-        region = self.provider.region('luzon')
-        assert isinstance(region, str)
-
-    def test_region_with_island_group_visayas(self):
-        region = self.provider.region('visayas')
-        assert isinstance(region, str)
-
-    def test_region_with_island_group_mindanao(self):
-        region = self.provider.region('mindanao')
-        assert isinstance(region, str)
-
-    def test_province_with_name(self):
-        self.provider.region()
-        province = self.provider.province('Surigao Del Norte')
-        assert province == 'Surigao Del Norte'
-
-    def test_province_with_invalid_name(self):
-        with pytest.raises(ValueError):
-            self.provider.province('InvalidProvince')
-
-    def test_city_with_name(self):
-        self.provider.region()
-        self.provider.province()
-        city = self.provider.city()
-        if city:
-            assert isinstance(city, str)
-
-    def test_municipality_with_name(self):
-        self.provider.region()
-        self.provider.province()
-        municipality = self.provider.municipality()
-        if municipality:
-            assert isinstance(municipality, str)
-            
-    def test_province_with_name_luzon(self):
-        self.provider.region('luzon')
-        province = self.provider.province('Ilocos Norte')
-        assert province == 'Ilocos Norte'
-
-    def test_province_with_name_visayas(self):
-        self.provider.region('visayas')
-        province = self.provider.province()
-        assert isinstance(province, str) or province is None
-
-    def test_province_with_name_mindanao(self):
-        self.provider.region('mindanao')
-        province = self.provider.province()
-        assert isinstance(province, str) or province is None
-
-    def test_municipality_with_invalid_name(self):
-        self.provider.region()
-        self.provider.province()
-        with pytest.raises(ValueError):
-            self.provider.municipality('InvalidMunicipality')
-
-    def test_city_with_invalid_name(self):
-        self.provider.region()
-        self.provider.province()
-        with pytest.raises(ValueError):
-            self.provider.city('InvalidCity')
-
-    def test_barangay_after_city(self):
-        self.provider.region()
-        self.provider.province()
-        self.provider.city()
-        barangay = self.provider.barangay()
-        assert isinstance(barangay, str) or barangay is None
-
-    def test_barangay_after_municipality(self):
-        self.provider.region()
-        self.provider.province()
-        municipality = self.provider.municipality()
-        if municipality:
-            barangay = self.provider.barangay()
-            assert isinstance(barangay, str) or barangay is None
-
-    def test_full_address_with_no_city_or_municipality(self):
-        address = self.provider.full_address()
-        assert isinstance(address, str)
-        assert len(address) > 0
-
-    def test_region_different_island_groups(self):
-        reg1 = self.provider.region('luzon')
-        assert isinstance(reg1, str)
-        reg2 = self.provider.region('visayas')
-        assert isinstance(reg2, str)
-        reg3 = self.provider.region('mindanao')
-        assert isinstance(reg3, str)
-
-    def test_province_resets_location(self):
-        self.provider.region()
-        self.provider.province()
-        self.provider.city()
-        self.provider.province()
-        assert self.provider.current_location is None
-        assert self.provider.current_location_type is None
-
-    def test_email_no_personal_data(self):
-        provider = ContactDataProvider(None)
-        email = provider.email()
-        assert isinstance(email, str)
-        assert '@' in email
-
-class TestContactDataProvider:
-    def setup_method(self):
-        self.personal = PersonalDataProvider()
-        self.provider = ContactDataProvider(self.personal)
-    
-    def test_mobile_generation(self):
-        mobile = self.provider.mobile()
-        assert isinstance(mobile, str)
-        assert mobile.startswith('09')
-
-    def test_mobile_generation_default(self):
-        mobile = self.provider.mobile('default')
-        assert isinstance(mobile, str)
-        assert mobile.startswith('09')
-
-    def test_mobile_international_format(self):
-        mobile = self.provider.mobile('international')
-        assert isinstance(mobile, str)
-        assert mobile.startswith('+639')
-
-    def test_mobile_invalid_format(self):
-        mobile = self.provider.mobile('invalid')
-        assert isinstance(mobile, str)
-        assert mobile.startswith('09')
-
-    def test_email_generation(self):
-        email = self.provider.email()
-        assert isinstance(email, str)
-        assert '@' in email
-
-    def test_email_with_custom_names(self):
-        email = self.provider.email('John', 'Doe')
-        assert isinstance(email, str)
-        assert '@' in email
-
-    def test_email_with_generated_names(self):
-        self.personal.firstname()
-        self.personal.lastname()
-        email = self.provider.email()
-        assert isinstance(email, str)
-        assert '@' in email