import utils.formatters as formatters
import pandas as pd
from pandas import testing as tm
import math

class TestFormatters:
    def test_format_duration(self):
        assert formatters.format_duration(30) == '00:00:30', 'Formats duration in seconds correctly'
        assert formatters.format_duration(305) == '00:05:05', 'Formats duration in minutes correctly'
        assert formatters.format_duration(5405) == '01:30:05', 'Formats duration in hours correctly'
        assert formatters.format_duration(305.9) == '00:05:06', 'Formats duration in hours correctly'
        assert formatters.format_duration(5405.2) == '01:30:05', 'Formats duration in hours correctly'
        assert formatters.format_duration('00:03:20') == '00:03:20', 'Returns already-formatted duration'

    def test_format_date(self):
        mock_date_df = pd.DataFrame(data={'Dates': ['01-02-23', '01-January-21', '05-Nov-20', '07-Mar-2023']})
        expected_output = pd.DataFrame(data={'Dates': ['01-Feb-2023', '01-Jan-2021', '05-Nov-2020', '07-Mar-2023']})
        tm.assert_series_equal(formatters.format_date(mock_date_df.loc[:,'Dates']), expected_output.loc[:,'Dates']), 'Formats date column values correctly'

    def test_format_dataframe(self):
        mock_df = pd.DataFrame(data={
            'Product Type': [math.nan],
            'Sales Start Date': ['01-01-2020'],
            'Collection Ownership Start Date': ['02-Feb-1999'],
            'Collection Ownership End Date': ['02-02-2003'],
            'Duration': [300],
            'Disc Number': [math.nan],
            'Track Number': [math.nan],
            'Contributor Legal Name': ['David Randolph | WILLOW'],
            'Contributor Type': ['Featured'],
            'Contributor Role(s)': ['Piano | Guitar'],
            'Collection Ownership Percentage of Rights': [math.nan],
            'Collection Ownership Country': ['US'],
            'Collection Ownership Country - Exclusion': ['#NO-TERRITORY-REQUIRED'],
            'Country of Label/Individual/Party Which Funded Recording': ['US'],
            'Country of Recording': ['DE'],
            'Explicit': ['EXPLICIT'],
            'Product Primary Artists': ['Elton John | WILLOW'],
            'Product Featuring Artists': [''],
            'Primary Artist': ['Elton John, WILLOW'],
            'Featuring Artist': ['']
        })

        expected_output = pd.DataFrame(data={
            'Product Type': ['Audio'],
            'Sales Start Date': ['01-Jan-2020'],
            'Collection Ownership Start Date': ['02-Feb-1999'],
            'Collection Ownership End Date': ['02-Feb-2003'],
            'Duration': ['00:05:00'],
            'Disc Number': '1',
            'Track Number': '1',
            'Contributor Legal Name': ['Randolph, David|WILLOW'],
            'Contributor Type': ['Main Performer'],
            'Contributor Role(s)': ['Piano|Guitar'],
            'Collection Ownership Percentage of Rights': [100],
            'Collection Ownership Country': ['US'],
            'Collection Ownership Country - Exclusion': [''],
            'Country of Label/Individual/Party Which Funded Recording': ['US'],
            'Country of Recording': ['DE'],
            'Explicit': ['Yes'],
            'Product Primary Artists': ['Elton John|WILLOW'],
            'Product Featuring Artists': [''],
            'Primary Artist': ['Elton John|WILLOW'],
            'Featuring Artist': ['']
        })

        tm.assert_frame_equal(formatters.format_dataframe(mock_df), expected_output, check_dtype=False), 'Formats column values correctly'

    def test_format_contributor_names(self):
        valid_row = 'David Randolph | Elton Michael John'
        single_name = 'WILLOW'
        multi_middle_names = 'David Randolph | WILLOW | Elton Michael Andrew John'
        row_with_pipe = 'David Randolph | Elton Michael John |'
        pipe_row_only = '|'
        multi_pipes = 'David Randolph | | Elton Michael John | |'
        assert formatters.format_contributor_names(valid_row) == 'Randolph, David|John, Elton Michael'
        assert formatters.format_contributor_names(math.nan) == ''
        assert formatters.format_contributor_names('') == ''
        assert formatters.format_contributor_names(single_name) == 'WILLOW'
        assert formatters.format_contributor_names(multi_middle_names) == 'Randolph, David|WILLOW|John, Elton Michael Andrew'
        assert formatters.format_contributor_names(row_with_pipe) == 'Randolph, David|John, Elton Michael'
        assert formatters.format_contributor_names(multi_pipes) == 'Randolph, David|John, Elton Michael'
        assert formatters.format_contributor_names(pipe_row_only) == ''

    def test_format_participant_names(self):
        valid_row = 'David Randolph | Elton Michael John'
        single_name = 'WILLOW'
        comma_row = 'David Randolph, WILLOW,Elton Michael Andrew John'
        row_with_pipe = 'David Randolph | Elton Michael John |'
        pipe_row_only = '|'
        multi_pipes = 'David Randolph | | Elton Michael John | |'
        ampersand_names = 'David Randolph, WILLOW, & Elton Michael John'
        assert formatters.format_participant_names(valid_row) == 'David Randolph|Elton Michael John'
        assert formatters.format_participant_names(math.nan) == ''
        assert formatters.format_participant_names('') == ''
        assert formatters.format_participant_names(single_name) == 'WILLOW'
        assert formatters.format_participant_names(comma_row) == 'David Randolph|WILLOW|Elton Michael Andrew John'
        assert formatters.format_participant_names(row_with_pipe) == 'David Randolph|Elton Michael John'
        assert formatters.format_participant_names(multi_pipes) == 'David Randolph|Elton Michael John'
        assert formatters.format_participant_names(pipe_row_only) == ''
        assert formatters.format_participant_names(ampersand_names) == 'David Randolph|WILLOW|Elton Michael John'

    def test_format_contributor_types(self):
        valid_types = 'Featured | Main Performers | Artist | Other Featured | Non-Featured Artist'
        types_with_pipes = 'Featured | | Featured |'
        assert formatters.format_contributor_types(valid_types) == 'Main Performer|Main Performer|Main Performer|Featuring Performer|Session Musician'
        assert formatters.format_contributor_types(types_with_pipes) == 'Main Performer|Main Performer'
        assert formatters.format_contributor_types(math.nan) == ''
        assert formatters.format_contributor_types('') == ''

    def test_format_piped_values(self):
        assert formatters.format_piped_values('US |') == 'US'
        assert formatters.format_piped_values('Piano     | Bass Guitar') == 'Piano|Bass Guitar'
        assert formatters.format_piped_values('Piano | Guitar | ') == 'Piano|Guitar'
        assert formatters.format_piped_values('Piano | | Vocals') == 'Piano|Vocals'
        assert formatters.format_piped_values(math.nan) == ''
        assert formatters.format_piped_values('') == ''

    def test_strip_all_strings(self):
        mock_df = pd.DataFrame(data={
            'Song Name': ['Some Song Name       '],
            'Primary Artist': ['   A Primary Artist     '],
            'Duration': [300],
            'Explicit': [math.nan]
        })
        mock_expected_result = pd.DataFrame(data={
            'Song Name': ['Some Song Name'],
            'Primary Artist': ['A Primary Artist'],
            'Duration': [300],
            'Explicit': [math.nan]
        })
        tm.assert_frame_equal(formatters.strip_all_strings(mock_df), mock_expected_result), 'All string values are trimmed'

    def test_format_video_dataframe(self):
        mock_df = pd.DataFrame(data={
            'Product Type': [math.nan],
            'Sales Start Date': ['01-01-2020'],
            'Collection Ownership Start Date': ['02-Feb-1999'],
            'Collection Ownership End Date': ['02-02-2003'],
            'Duration': [300],
            'Contributor Legal Name': ['David Randolph | WILLOW'],
            'Contributor Type': ['Featured'],
            'Contributor Role(s)': ['Piano | Guitar'],
            'Collection Ownership Percentage of Rights': [math.nan],
            'Collection Ownership Country': ['US'],
            'Collection Ownership Country - Exclusion': ['#NO-TERRITORY-REQUIRED'],
            'Country of Label/Individual/Party Which Funded Recording': ['US'],
            'Country of Recording': ['DE'],
            'Explicit': ['EXPLICIT'],
            'Product Primary Artists': ['Elton John | WILLOW'],
            'Product Featuring Artists': [''],
            'Primary Artist': ['Elton John, WILLOW'],
            'Featuring Artist': ['']
        })

        expected_output = pd.DataFrame(data={
            'Product Type': ['Video'],
            'Sales Start Date': ['01-Jan-2020'],
            'Collection Ownership Start Date': ['02-Feb-1999'],
            'Collection Ownership End Date': ['02-Feb-2003'],
            'Duration': ['00:05:00'],
            'Contributor Legal Name': ['Randolph, David|WILLOW'],
            'Contributor Type': ['Main Performer'],
            'Contributor Role(s)': ['Piano|Guitar'],
            'Collection Ownership Percentage of Rights': [100],
            'Collection Ownership Country': ['US'],
            'Collection Ownership Country - Exclusion': [''],
            'Country of Label/Individual/Party Which Funded Recording': ['US'],
            'Country of Recording': ['DE'],
            'Explicit': ['Yes'],
            'Product Primary Artists': ['Elton John|WILLOW'],
            'Product Featuring Artists': [''],
            'Primary Artist': ['Elton John|WILLOW'],
            'Featuring Artist': ['']
        })

        tm.assert_frame_equal(formatters.format_video_dataframe(mock_df), expected_output, check_dtype=False), 'Formats column values correctly'

    def test_format_video_no_product_dataframe(self):
        mock_df = pd.DataFrame(data={
            'Product Type': [math.nan],
            'Sales Start Date': [math.nan],
            'Collection Ownership Start Date': ['02-Feb-1999'],
            'Collection Ownership End Date': ['02-02-2003'],
            'Duration': [300],
            'Contributor Legal Name': ['David Randolph | WILLOW'],
            'Contributor Type': ['Featured'],
            'Contributor Role(s)': ['Piano | Guitar'],
            'Collection Ownership Percentage of Rights': [math.nan],
            'Collection Ownership Country': ['US'],
            'Collection Ownership Country - Exclusion': ['#NO-TERRITORY-REQUIRED'],
            'Country of Label/Individual/Party Which Funded Recording': ['US'],
            'Country of Recording': ['DE'],
            'Explicit': ['EXPLICIT'],
            'Product Primary Artists': [''],
            'Product Featuring Artists': [''],
            'Primary Artist': ['Elton John, WILLOW'],
            'Featuring Artist': ['']
        })

        expected_output = pd.DataFrame(data={
            'Product Type': ['Video'],
            'Sales Start Date': [math.nan],
            'Collection Ownership Start Date': ['02-Feb-1999'],
            'Collection Ownership End Date': ['02-Feb-2003'],
            'Duration': ['00:05:00'],
            'Contributor Legal Name': ['Randolph, David|WILLOW'],
            'Contributor Type': ['Main Performer'],
            'Contributor Role(s)': ['Piano|Guitar'],
            'Collection Ownership Percentage of Rights': [100],
            'Collection Ownership Country': ['US'],
            'Collection Ownership Country - Exclusion': [''],
            'Country of Label/Individual/Party Which Funded Recording': ['US'],
            'Country of Recording': ['DE'],
            'Explicit': ['Yes'],
            'Product Primary Artists': [''],
            'Product Featuring Artists': [''],
            'Primary Artist': ['Elton John|WILLOW'],
            'Featuring Artist': ['']
        })

        tm.assert_frame_equal(formatters.format_video_dataframe(mock_df), expected_output, check_dtype=False), 'Formats column values correctly'