Issue 596 - Add Markdown presentation

[Marc-Andre-Rivet]

Fixes#596

• check virtualized performance

• virtualized behavior (varying row height -- breaks assumptions, this will be dealt outside the scope of this PR as this behavior is not caused by MD)

• varying row height with fixed columns (Using DataTable row_selectable and row_deletable with fixed columns results in mismatched row heights #649)

• navigation through and from markdown cell

• define copy/paste behavior (?!)

• define filter behavior -- for now filters will behave the same way for markdown cells as for any other text cells (the value filtered on is the actual value in the df) -- expect at least one visual/sanity test for the filter behavior (guard rail)

• define sort behavior (supporting/using displayed values is out of scope) -- for now sort will behave the same way for markdown cells as for any other text cells (the value sorted on is the actual value in the df) -- expect at least one visual/sanity test for the sort behavior (guard rail)

• visual tests -- cover as much of md syntax as possible

• behavior on non-text input value + test (e.g. number, boolean values, how does ReMarkable behave)

• markdown image resize (test, docs)

• more tests

• docs (will be done in https://github.com/plotly/dash-docs/pull/733)

• Test: Default highlight.js included with table does not assign itself to window.hljs

• Test: Loading external highlight.js will take precedence on the table's version

[Marc-Andre-Rivet]

1. Navigation works well when the user single clicks on the cell, on a double click the focus goes back to the page's body and navigation inside the table becomes impossible

2. Seeing quite a few percy diffs for fixed columns / rows test cases

Very nice visual tests 😄

[Marc-Andre-Rivet]

I've been thinking that maybe we can include the configuration for highlight.js usage, externalize the resource and expect special code block users to include the highlight.js they want as a resource from https://highlightjs.org/download/ ?! Either that or (1) not support it at all or (2) asyncing the resource -- in which case React.Suspense won't do and we'll have to handle it through state on CellMarkdown. This is mostly about completeness.. we should not load JS by default as it will involve a very special use case.

 private static md: Remarkable = new Remarkable({
highlight: (value: string, language: string) => {
const hljs: any = (window as any).hljs;
if (!hljs) {
return '';
}
if (language && hljs.getLanguage(language)) {
try {
return hljs.highlight(language, value).value;
} catch (_) { }
}
try {
return hljs.highlightAuto(value).value;
} catch (_) { }
return '';
}
});

I don't think we should flat out not support highlighting and I don't think this will ever be enough of a priority on its own to be fixed later, if possible I think I'd rather get it in w/ async and wrap it up.

[Marc-Andre-Rivet]

Since this is now our own cherry-picked version of Highlight.js would rather see this somewhere inside of /src, even if it's just /src/third-party/Highlight.js or some variation.

[Marc-Andre-Rivet]

Same as https://github.com/plotly/dash-table/pull/606/files#r358530234

[Marc-Andre-Rivet]

I think that with @types/Remarkable added, all of the remarkable definitions becomes redundant.

[Marc-Andre-Rivet]

This is now redundant

[Marc-Andre-Rivet]

@shammamah Testing navigation with:

import dash
from dash_html_components import Div
from dash_table import DataTable
app = dash.Dash(__name__)
data=[
dict(a=1, b='''```javascript
return 20;```
'''),
dict(a=1, b='''```js
console.log(\'pouet\');
'''),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```')
]
app.layout = Div([
DataTable(
columns=[
dict(name='a', id='a', editable=True),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
DataTable(
columns=[
dict(name='a', id='a'),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
])
app.run_server(debug=True)

Notice that for both tables it's possible to click/select a markdown cell and keyboard/navigate within the markdown column as expected. The top table, with editable=True, does not behave correctly after keyboard/navigate from column a to b. Notice that it's now impossible to keyboard/navigate. Does not apply to the read-only table.

[muzamalrana]

hi Can anyone help me here ? I am totally stuck. Trying to embedd image from link into dash table cell. Following is the code snippet.
html.Div([dt.DataTable(id='my-data',
columns=[
{"id": 'artist_image', "name": "artist_image", "presentation": "markdown"},
{"id": 'song_artist', "name": "song_artist"},
{"id": 'monthly_listeners', "name": "monthly_listeners"},
{"id": 'artist_followers', "name": "artist_followers"},
{"id": 'artist_popularity', "name": "artist_popularity"}]
,style_as_list_view=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
style_data_conditional=[
{
'if': {'row_index': 'odd'},
'backgroundColor': '#000000'
}
],
style_cell={'textAlign': 'center','padding': '5px','backgroundColor':'#00FF00',
'color':'white'},
style_header={
'backgroundColor': '#202020',
'fontWeight': 'bold','color':'white'
})],className='ten columns')])],className='row')], style=tab_style, selected_style=tab_selected_style)
@app.callback(
dash.dependencies.Output('my-data','data'),
[dash.dependencies.Input('my-graph1', 'clickData'),
dash.dependencies.Input('my_dropdown1', 'value'),
dash.dependencies.Input('radio_buttons', 'value'),
dash.dependencies.Input('my_dropdown3', 'value')]
)
def update_graph(click_data,dropdown,radio_button,dropdown3):
try:
genre1=click_data['points'][0]['customdata'][1]
except:
genre1='pop'
artist_data1=sm.artist_dataFrame(songs,artist,click_data,dropdown,radio_button)
print(artist_data1.head(1))
artist_data1=artist_data1[['artist_image','song_artist','monthly_listeners','artist_followers','artist_popularity','date_x']]
artist_data1['date_x']=pd.to_datetime(artist_data1['date_x'])
artist_data1=artist_data1[artist_data1['date_x']==artist_data1['date_x'].max()]
artist_data1=artist_data1.sort_values(by=radio_button,ascending=False).reset_index(drop=True)
artist_data1=artist_data1.drop(['date_x'],axis=1)
if dropdown3==0 or dropdown3=='All':
artist_data1=artist_data1
else:
artist_data1=artist_data1.head(dropdown3)
artist_data1['artist_image']=[html.Img(src=i) for i in artist_data1['artist_image']]
return artist_data1.to_dict('records')

[Marc-Andre-Rivet]

Fixes#596

• check virtualized performance

• virtualized behavior (varying row height -- breaks assumptions, this will be dealt outside the scope of this PR as this behavior is not caused by MD)

• varying row height with fixed columns (Using DataTable row_selectable and row_deletable with fixed columns results in mismatched row heights #649)

• navigation through and from markdown cell

• define copy/paste behavior (?!)

• define filter behavior -- for now filters will behave the same way for markdown cells as for any other text cells (the value filtered on is the actual value in the df) -- expect at least one visual/sanity test for the filter behavior (guard rail)

• define sort behavior (supporting/using displayed values is out of scope) -- for now sort will behave the same way for markdown cells as for any other text cells (the value sorted on is the actual value in the df) -- expect at least one visual/sanity test for the sort behavior (guard rail)

• visual tests -- cover as much of md syntax as possible

• behavior on non-text input value + test (e.g. number, boolean values, how does ReMarkable behave)

• markdown image resize (test, docs)

• more tests

• docs (will be done in https://github.com/plotly/dash-docs/pull/733)

• Test: Default highlight.js included with table does not assign itself to window.hljs

• Test: Loading external highlight.js will take precedence on the table's version

[Marc-Andre-Rivet]

1. Navigation works well when the user single clicks on the cell, on a double click the focus goes back to the page's body and navigation inside the table becomes impossible

2. Seeing quite a few percy diffs for fixed columns / rows test cases

Very nice visual tests 😄

[Marc-Andre-Rivet]

I've been thinking that maybe we can include the configuration for highlight.js usage, externalize the resource and expect special code block users to include the highlight.js they want as a resource from https://highlightjs.org/download/ ?! Either that or (1) not support it at all or (2) asyncing the resource -- in which case React.Suspense won't do and we'll have to handle it through state on CellMarkdown. This is mostly about completeness.. we should not load JS by default as it will involve a very special use case.

 private static md: Remarkable = new Remarkable({
highlight: (value: string, language: string) => {
const hljs: any = (window as any).hljs;
if (!hljs) {
return '';
}
if (language && hljs.getLanguage(language)) {
try {
return hljs.highlight(language, value).value;
} catch (_) { }
}
try {
return hljs.highlightAuto(value).value;
} catch (_) { }
return '';
}
});

I don't think we should flat out not support highlighting and I don't think this will ever be enough of a priority on its own to be fixed later, if possible I think I'd rather get it in w/ async and wrap it up.

[Marc-Andre-Rivet]

Since this is now our own cherry-picked version of Highlight.js would rather see this somewhere inside of /src, even if it's just /src/third-party/Highlight.js or some variation.

[Marc-Andre-Rivet]

Same as https://github.com/plotly/dash-table/pull/606/files#r358530234

[Marc-Andre-Rivet]

I think that with @types/Remarkable added, all of the remarkable definitions becomes redundant.

[Marc-Andre-Rivet]

This is now redundant

[Marc-Andre-Rivet]

@shammamah Testing navigation with:

import dash
from dash_html_components import Div
from dash_table import DataTable
app = dash.Dash(__name__)
data=[
dict(a=1, b='''```javascript
return 20;```
'''),
dict(a=1, b='''```js
console.log(\'pouet\');
'''),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```')
]
app.layout = Div([
DataTable(
columns=[
dict(name='a', id='a', editable=True),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
DataTable(
columns=[
dict(name='a', id='a'),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
])
app.run_server(debug=True)

Notice that for both tables it's possible to click/select a markdown cell and keyboard/navigate within the markdown column as expected. The top table, with editable=True, does not behave correctly after keyboard/navigate from column a to b. Notice that it's now impossible to keyboard/navigate. Does not apply to the read-only table.

[muzamalrana]

hi Can anyone help me here ? I am totally stuck. Trying to embedd image from link into dash table cell. Following is the code snippet.
html.Div([dt.DataTable(id='my-data',
columns=[
{"id": 'artist_image', "name": "artist_image", "presentation": "markdown"},
{"id": 'song_artist', "name": "song_artist"},
{"id": 'monthly_listeners', "name": "monthly_listeners"},
{"id": 'artist_followers', "name": "artist_followers"},
{"id": 'artist_popularity', "name": "artist_popularity"}]
,style_as_list_view=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
style_data_conditional=[
{
'if': {'row_index': 'odd'},
'backgroundColor': '#000000'
}
],
style_cell={'textAlign': 'center','padding': '5px','backgroundColor':'#00FF00',
'color':'white'},
style_header={
'backgroundColor': '#202020',
'fontWeight': 'bold','color':'white'
})],className='ten columns')])],className='row')], style=tab_style, selected_style=tab_selected_style)
@app.callback(
dash.dependencies.Output('my-data','data'),
[dash.dependencies.Input('my-graph1', 'clickData'),
dash.dependencies.Input('my_dropdown1', 'value'),
dash.dependencies.Input('radio_buttons', 'value'),
dash.dependencies.Input('my_dropdown3', 'value')]
)
def update_graph(click_data,dropdown,radio_button,dropdown3):
try:
genre1=click_data['points'][0]['customdata'][1]
except:
genre1='pop'
artist_data1=sm.artist_dataFrame(songs,artist,click_data,dropdown,radio_button)
print(artist_data1.head(1))
artist_data1=artist_data1[['artist_image','song_artist','monthly_listeners','artist_followers','artist_popularity','date_x']]
artist_data1['date_x']=pd.to_datetime(artist_data1['date_x'])
artist_data1=artist_data1[artist_data1['date_x']==artist_data1['date_x'].max()]
artist_data1=artist_data1.sort_values(by=radio_button,ascending=False).reset_index(drop=True)
artist_data1=artist_data1.drop(['date_x'],axis=1)
if dropdown3==0 or dropdown3=='All':
artist_data1=artist_data1
else:
artist_data1=artist_data1.head(dropdown3)
artist_data1['artist_image']=[html.Img(src=i) for i in artist_data1['artist_image']]
return artist_data1.to_dict('records')

[Marc-Andre-Rivet]

Fixes#596

• check virtualized performance

• virtualized behavior (varying row height -- breaks assumptions, this will be dealt outside the scope of this PR as this behavior is not caused by MD)

• varying row height with fixed columns (Using DataTable row_selectable and row_deletable with fixed columns results in mismatched row heights #649)

• navigation through and from markdown cell

• define copy/paste behavior (?!)

• define filter behavior -- for now filters will behave the same way for markdown cells as for any other text cells (the value filtered on is the actual value in the df) -- expect at least one visual/sanity test for the filter behavior (guard rail)

• define sort behavior (supporting/using displayed values is out of scope) -- for now sort will behave the same way for markdown cells as for any other text cells (the value sorted on is the actual value in the df) -- expect at least one visual/sanity test for the sort behavior (guard rail)

• visual tests -- cover as much of md syntax as possible

• behavior on non-text input value + test (e.g. number, boolean values, how does ReMarkable behave)

• markdown image resize (test, docs)

• more tests

• docs (will be done in https://github.com/plotly/dash-docs/pull/733)

• Test: Default highlight.js included with table does not assign itself to window.hljs

• Test: Loading external highlight.js will take precedence on the table's version

[Marc-Andre-Rivet]

1. Navigation works well when the user single clicks on the cell, on a double click the focus goes back to the page's body and navigation inside the table becomes impossible

2. Seeing quite a few percy diffs for fixed columns / rows test cases

Very nice visual tests 😄

[Marc-Andre-Rivet]

I've been thinking that maybe we can include the configuration for highlight.js usage, externalize the resource and expect special code block users to include the highlight.js they want as a resource from https://highlightjs.org/download/ ?! Either that or (1) not support it at all or (2) asyncing the resource -- in which case React.Suspense won't do and we'll have to handle it through state on CellMarkdown. This is mostly about completeness.. we should not load JS by default as it will involve a very special use case.

 private static md: Remarkable = new Remarkable({
highlight: (value: string, language: string) => {
const hljs: any = (window as any).hljs;
if (!hljs) {
return '';
}
if (language && hljs.getLanguage(language)) {
try {
return hljs.highlight(language, value).value;
} catch (_) { }
}
try {
return hljs.highlightAuto(value).value;
} catch (_) { }
return '';
}
});

I don't think we should flat out not support highlighting and I don't think this will ever be enough of a priority on its own to be fixed later, if possible I think I'd rather get it in w/ async and wrap it up.

[Marc-Andre-Rivet]

Since this is now our own cherry-picked version of Highlight.js would rather see this somewhere inside of /src, even if it's just /src/third-party/Highlight.js or some variation.

[Marc-Andre-Rivet]

Same as https://github.com/plotly/dash-table/pull/606/files#r358530234

[Marc-Andre-Rivet]

I think that with @types/Remarkable added, all of the remarkable definitions becomes redundant.

[Marc-Andre-Rivet]

This is now redundant

[Marc-Andre-Rivet]

@shammamah Testing navigation with:

import dash
from dash_html_components import Div
from dash_table import DataTable
app = dash.Dash(__name__)
data=[
dict(a=1, b='''```javascript
return 20;```
'''),
dict(a=1, b='''```js
console.log(\'pouet\');
'''),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```')
]
app.layout = Div([
DataTable(
columns=[
dict(name='a', id='a', editable=True),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
DataTable(
columns=[
dict(name='a', id='a'),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
])
app.run_server(debug=True)

Notice that for both tables it's possible to click/select a markdown cell and keyboard/navigate within the markdown column as expected. The top table, with editable=True, does not behave correctly after keyboard/navigate from column a to b. Notice that it's now impossible to keyboard/navigate. Does not apply to the read-only table.

[muzamalrana]

hi Can anyone help me here ? I am totally stuck. Trying to embedd image from link into dash table cell. Following is the code snippet.
html.Div([dt.DataTable(id='my-data',
columns=[
{"id": 'artist_image', "name": "artist_image", "presentation": "markdown"},
{"id": 'song_artist', "name": "song_artist"},
{"id": 'monthly_listeners', "name": "monthly_listeners"},
{"id": 'artist_followers', "name": "artist_followers"},
{"id": 'artist_popularity', "name": "artist_popularity"}]
,style_as_list_view=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
style_data_conditional=[
{
'if': {'row_index': 'odd'},
'backgroundColor': '#000000'
}
],
style_cell={'textAlign': 'center','padding': '5px','backgroundColor':'#00FF00',
'color':'white'},
style_header={
'backgroundColor': '#202020',
'fontWeight': 'bold','color':'white'
})],className='ten columns')])],className='row')], style=tab_style, selected_style=tab_selected_style)
@app.callback(
dash.dependencies.Output('my-data','data'),
[dash.dependencies.Input('my-graph1', 'clickData'),
dash.dependencies.Input('my_dropdown1', 'value'),
dash.dependencies.Input('radio_buttons', 'value'),
dash.dependencies.Input('my_dropdown3', 'value')]
)
def update_graph(click_data,dropdown,radio_button,dropdown3):
try:
genre1=click_data['points'][0]['customdata'][1]
except:
genre1='pop'
artist_data1=sm.artist_dataFrame(songs,artist,click_data,dropdown,radio_button)
print(artist_data1.head(1))
artist_data1=artist_data1[['artist_image','song_artist','monthly_listeners','artist_followers','artist_popularity','date_x']]
artist_data1['date_x']=pd.to_datetime(artist_data1['date_x'])
artist_data1=artist_data1[artist_data1['date_x']==artist_data1['date_x'].max()]
artist_data1=artist_data1.sort_values(by=radio_button,ascending=False).reset_index(drop=True)
artist_data1=artist_data1.drop(['date_x'],axis=1)
if dropdown3==0 or dropdown3=='All':
artist_data1=artist_data1
else:
artist_data1=artist_data1.head(dropdown3)
artist_data1['artist_image']=[html.Img(src=i) for i in artist_data1['artist_image']]
return artist_data1.to_dict('records')

[Marc-Andre-Rivet]

Fixes#596

• check virtualized performance

• virtualized behavior (varying row height -- breaks assumptions, this will be dealt outside the scope of this PR as this behavior is not caused by MD)

• varying row height with fixed columns (Using DataTable row_selectable and row_deletable with fixed columns results in mismatched row heights #649)

• navigation through and from markdown cell

• define copy/paste behavior (?!)

• define filter behavior -- for now filters will behave the same way for markdown cells as for any other text cells (the value filtered on is the actual value in the df) -- expect at least one visual/sanity test for the filter behavior (guard rail)

• define sort behavior (supporting/using displayed values is out of scope) -- for now sort will behave the same way for markdown cells as for any other text cells (the value sorted on is the actual value in the df) -- expect at least one visual/sanity test for the sort behavior (guard rail)

• visual tests -- cover as much of md syntax as possible

• behavior on non-text input value + test (e.g. number, boolean values, how does ReMarkable behave)

• markdown image resize (test, docs)

• more tests

• docs (will be done in https://github.com/plotly/dash-docs/pull/733)

• Test: Default highlight.js included with table does not assign itself to window.hljs

• Test: Loading external highlight.js will take precedence on the table's version

[Marc-Andre-Rivet]

1. Navigation works well when the user single clicks on the cell, on a double click the focus goes back to the page's body and navigation inside the table becomes impossible

2. Seeing quite a few percy diffs for fixed columns / rows test cases

Very nice visual tests 😄

[Marc-Andre-Rivet]

I've been thinking that maybe we can include the configuration for highlight.js usage, externalize the resource and expect special code block users to include the highlight.js they want as a resource from https://highlightjs.org/download/ ?! Either that or (1) not support it at all or (2) asyncing the resource -- in which case React.Suspense won't do and we'll have to handle it through state on CellMarkdown. This is mostly about completeness.. we should not load JS by default as it will involve a very special use case.

 private static md: Remarkable = new Remarkable({
highlight: (value: string, language: string) => {
const hljs: any = (window as any).hljs;
if (!hljs) {
return '';
}
if (language && hljs.getLanguage(language)) {
try {
return hljs.highlight(language, value).value;
} catch (_) { }
}
try {
return hljs.highlightAuto(value).value;
} catch (_) { }
return '';
}
});

I don't think we should flat out not support highlighting and I don't think this will ever be enough of a priority on its own to be fixed later, if possible I think I'd rather get it in w/ async and wrap it up.

[Marc-Andre-Rivet]

Since this is now our own cherry-picked version of Highlight.js would rather see this somewhere inside of /src, even if it's just /src/third-party/Highlight.js or some variation.

[Marc-Andre-Rivet]

Same as https://github.com/plotly/dash-table/pull/606/files#r358530234

[Marc-Andre-Rivet]

I think that with @types/Remarkable added, all of the remarkable definitions becomes redundant.

[Marc-Andre-Rivet]

This is now redundant

[Marc-Andre-Rivet]

@shammamah Testing navigation with:

import dash
from dash_html_components import Div
from dash_table import DataTable
app = dash.Dash(__name__)
data=[
dict(a=1, b='''```javascript
return 20;```
'''),
dict(a=1, b='''```js
console.log(\'pouet\');
'''),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```')
]
app.layout = Div([
DataTable(
columns=[
dict(name='a', id='a', editable=True),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
DataTable(
columns=[
dict(name='a', id='a'),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
])
app.run_server(debug=True)

Notice that for both tables it's possible to click/select a markdown cell and keyboard/navigate within the markdown column as expected. The top table, with editable=True, does not behave correctly after keyboard/navigate from column a to b. Notice that it's now impossible to keyboard/navigate. Does not apply to the read-only table.

[muzamalrana]

hi Can anyone help me here ? I am totally stuck. Trying to embedd image from link into dash table cell. Following is the code snippet.
html.Div([dt.DataTable(id='my-data',
columns=[
{"id": 'artist_image', "name": "artist_image", "presentation": "markdown"},
{"id": 'song_artist', "name": "song_artist"},
{"id": 'monthly_listeners', "name": "monthly_listeners"},
{"id": 'artist_followers', "name": "artist_followers"},
{"id": 'artist_popularity', "name": "artist_popularity"}]
,style_as_list_view=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
style_data_conditional=[
{
'if': {'row_index': 'odd'},
'backgroundColor': '#000000'
}
],
style_cell={'textAlign': 'center','padding': '5px','backgroundColor':'#00FF00',
'color':'white'},
style_header={
'backgroundColor': '#202020',
'fontWeight': 'bold','color':'white'
})],className='ten columns')])],className='row')], style=tab_style, selected_style=tab_selected_style)
@app.callback(
dash.dependencies.Output('my-data','data'),
[dash.dependencies.Input('my-graph1', 'clickData'),
dash.dependencies.Input('my_dropdown1', 'value'),
dash.dependencies.Input('radio_buttons', 'value'),
dash.dependencies.Input('my_dropdown3', 'value')]
)
def update_graph(click_data,dropdown,radio_button,dropdown3):
try:
genre1=click_data['points'][0]['customdata'][1]
except:
genre1='pop'
artist_data1=sm.artist_dataFrame(songs,artist,click_data,dropdown,radio_button)
print(artist_data1.head(1))
artist_data1=artist_data1[['artist_image','song_artist','monthly_listeners','artist_followers','artist_popularity','date_x']]
artist_data1['date_x']=pd.to_datetime(artist_data1['date_x'])
artist_data1=artist_data1[artist_data1['date_x']==artist_data1['date_x'].max()]
artist_data1=artist_data1.sort_values(by=radio_button,ascending=False).reset_index(drop=True)
artist_data1=artist_data1.drop(['date_x'],axis=1)
if dropdown3==0 or dropdown3=='All':
artist_data1=artist_data1
else:
artist_data1=artist_data1.head(dropdown3)
artist_data1['artist_image']=[html.Img(src=i) for i in artist_data1['artist_image']]
return artist_data1.to_dict('records')

[Marc-Andre-Rivet]

Fixes#596

• check virtualized performance

• virtualized behavior (varying row height -- breaks assumptions, this will be dealt outside the scope of this PR as this behavior is not caused by MD)

• varying row height with fixed columns (Using DataTable row_selectable and row_deletable with fixed columns results in mismatched row heights #649)

• navigation through and from markdown cell

• define copy/paste behavior (?!)

• define filter behavior -- for now filters will behave the same way for markdown cells as for any other text cells (the value filtered on is the actual value in the df) -- expect at least one visual/sanity test for the filter behavior (guard rail)

• define sort behavior (supporting/using displayed values is out of scope) -- for now sort will behave the same way for markdown cells as for any other text cells (the value sorted on is the actual value in the df) -- expect at least one visual/sanity test for the sort behavior (guard rail)

• visual tests -- cover as much of md syntax as possible

• behavior on non-text input value + test (e.g. number, boolean values, how does ReMarkable behave)

• markdown image resize (test, docs)

• more tests

• docs (will be done in https://github.com/plotly/dash-docs/pull/733)

• Test: Default highlight.js included with table does not assign itself to window.hljs

• Test: Loading external highlight.js will take precedence on the table's version

[Marc-Andre-Rivet]

1. Navigation works well when the user single clicks on the cell, on a double click the focus goes back to the page's body and navigation inside the table becomes impossible

2. Seeing quite a few percy diffs for fixed columns / rows test cases

Very nice visual tests 😄

[Marc-Andre-Rivet]

I've been thinking that maybe we can include the configuration for highlight.js usage, externalize the resource and expect special code block users to include the highlight.js they want as a resource from https://highlightjs.org/download/ ?! Either that or (1) not support it at all or (2) asyncing the resource -- in which case React.Suspense won't do and we'll have to handle it through state on CellMarkdown. This is mostly about completeness.. we should not load JS by default as it will involve a very special use case.

 private static md: Remarkable = new Remarkable({
highlight: (value: string, language: string) => {
const hljs: any = (window as any).hljs;
if (!hljs) {
return '';
}
if (language && hljs.getLanguage(language)) {
try {
return hljs.highlight(language, value).value;
} catch (_) { }
}
try {
return hljs.highlightAuto(value).value;
} catch (_) { }
return '';
}
});

I don't think we should flat out not support highlighting and I don't think this will ever be enough of a priority on its own to be fixed later, if possible I think I'd rather get it in w/ async and wrap it up.

[Marc-Andre-Rivet]

Since this is now our own cherry-picked version of Highlight.js would rather see this somewhere inside of /src, even if it's just /src/third-party/Highlight.js or some variation.

[Marc-Andre-Rivet]

Same as https://github.com/plotly/dash-table/pull/606/files#r358530234

[Marc-Andre-Rivet]

I think that with @types/Remarkable added, all of the remarkable definitions becomes redundant.

[Marc-Andre-Rivet]

This is now redundant

[Marc-Andre-Rivet]

@shammamah Testing navigation with:

import dash
from dash_html_components import Div
from dash_table import DataTable
app = dash.Dash(__name__)
data=[
dict(a=1, b='''```javascript
return 20;```
'''),
dict(a=1, b='''```js
console.log(\'pouet\');
'''),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```')
]
app.layout = Div([
DataTable(
columns=[
dict(name='a', id='a', editable=True),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
DataTable(
columns=[
dict(name='a', id='a'),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
])
app.run_server(debug=True)

Notice that for both tables it's possible to click/select a markdown cell and keyboard/navigate within the markdown column as expected. The top table, with editable=True, does not behave correctly after keyboard/navigate from column a to b. Notice that it's now impossible to keyboard/navigate. Does not apply to the read-only table.

[muzamalrana]

hi Can anyone help me here ? I am totally stuck. Trying to embedd image from link into dash table cell. Following is the code snippet.
html.Div([dt.DataTable(id='my-data',
columns=[
{"id": 'artist_image', "name": "artist_image", "presentation": "markdown"},
{"id": 'song_artist', "name": "song_artist"},
{"id": 'monthly_listeners', "name": "monthly_listeners"},
{"id": 'artist_followers', "name": "artist_followers"},
{"id": 'artist_popularity', "name": "artist_popularity"}]
,style_as_list_view=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
style_data_conditional=[
{
'if': {'row_index': 'odd'},
'backgroundColor': '#000000'
}
],
style_cell={'textAlign': 'center','padding': '5px','backgroundColor':'#00FF00',
'color':'white'},
style_header={
'backgroundColor': '#202020',
'fontWeight': 'bold','color':'white'
})],className='ten columns')])],className='row')], style=tab_style, selected_style=tab_selected_style)
@app.callback(
dash.dependencies.Output('my-data','data'),
[dash.dependencies.Input('my-graph1', 'clickData'),
dash.dependencies.Input('my_dropdown1', 'value'),
dash.dependencies.Input('radio_buttons', 'value'),
dash.dependencies.Input('my_dropdown3', 'value')]
)
def update_graph(click_data,dropdown,radio_button,dropdown3):
try:
genre1=click_data['points'][0]['customdata'][1]
except:
genre1='pop'
artist_data1=sm.artist_dataFrame(songs,artist,click_data,dropdown,radio_button)
print(artist_data1.head(1))
artist_data1=artist_data1[['artist_image','song_artist','monthly_listeners','artist_followers','artist_popularity','date_x']]
artist_data1['date_x']=pd.to_datetime(artist_data1['date_x'])
artist_data1=artist_data1[artist_data1['date_x']==artist_data1['date_x'].max()]
artist_data1=artist_data1.sort_values(by=radio_button,ascending=False).reset_index(drop=True)
artist_data1=artist_data1.drop(['date_x'],axis=1)
if dropdown3==0 or dropdown3=='All':
artist_data1=artist_data1
else:
artist_data1=artist_data1.head(dropdown3)
artist_data1['artist_image']=[html.Img(src=i) for i in artist_data1['artist_image']]
return artist_data1.to_dict('records')

[Marc-Andre-Rivet]

Fixes#596

• check virtualized performance

• virtualized behavior (varying row height -- breaks assumptions, this will be dealt outside the scope of this PR as this behavior is not caused by MD)

• varying row height with fixed columns (Using DataTable row_selectable and row_deletable with fixed columns results in mismatched row heights #649)

• navigation through and from markdown cell

• define copy/paste behavior (?!)

• define filter behavior -- for now filters will behave the same way for markdown cells as for any other text cells (the value filtered on is the actual value in the df) -- expect at least one visual/sanity test for the filter behavior (guard rail)

• define sort behavior (supporting/using displayed values is out of scope) -- for now sort will behave the same way for markdown cells as for any other text cells (the value sorted on is the actual value in the df) -- expect at least one visual/sanity test for the sort behavior (guard rail)

• visual tests -- cover as much of md syntax as possible

• behavior on non-text input value + test (e.g. number, boolean values, how does ReMarkable behave)

• markdown image resize (test, docs)

• more tests

• docs (will be done in https://github.com/plotly/dash-docs/pull/733)

• Test: Default highlight.js included with table does not assign itself to window.hljs

• Test: Loading external highlight.js will take precedence on the table's version

[Marc-Andre-Rivet]

1. Navigation works well when the user single clicks on the cell, on a double click the focus goes back to the page's body and navigation inside the table becomes impossible

2. Seeing quite a few percy diffs for fixed columns / rows test cases

Very nice visual tests 😄

[Marc-Andre-Rivet]

I've been thinking that maybe we can include the configuration for highlight.js usage, externalize the resource and expect special code block users to include the highlight.js they want as a resource from https://highlightjs.org/download/ ?! Either that or (1) not support it at all or (2) asyncing the resource -- in which case React.Suspense won't do and we'll have to handle it through state on CellMarkdown. This is mostly about completeness.. we should not load JS by default as it will involve a very special use case.

 private static md: Remarkable = new Remarkable({
highlight: (value: string, language: string) => {
const hljs: any = (window as any).hljs;
if (!hljs) {
return '';
}
if (language && hljs.getLanguage(language)) {
try {
return hljs.highlight(language, value).value;
} catch (_) { }
}
try {
return hljs.highlightAuto(value).value;
} catch (_) { }
return '';
}
});

I don't think we should flat out not support highlighting and I don't think this will ever be enough of a priority on its own to be fixed later, if possible I think I'd rather get it in w/ async and wrap it up.

[Marc-Andre-Rivet]

Since this is now our own cherry-picked version of Highlight.js would rather see this somewhere inside of /src, even if it's just /src/third-party/Highlight.js or some variation.

[Marc-Andre-Rivet]

Same as https://github.com/plotly/dash-table/pull/606/files#r358530234

[Marc-Andre-Rivet]

I think that with @types/Remarkable added, all of the remarkable definitions becomes redundant.

[Marc-Andre-Rivet]

This is now redundant

[Marc-Andre-Rivet]

@shammamah Testing navigation with:

import dash
from dash_html_components import Div
from dash_table import DataTable
app = dash.Dash(__name__)
data=[
dict(a=1, b='''```javascript
return 20;```
'''),
dict(a=1, b='''```js
console.log(\'pouet\');
'''),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```')
]
app.layout = Div([
DataTable(
columns=[
dict(name='a', id='a', editable=True),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
DataTable(
columns=[
dict(name='a', id='a'),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
])
app.run_server(debug=True)

Notice that for both tables it's possible to click/select a markdown cell and keyboard/navigate within the markdown column as expected. The top table, with editable=True, does not behave correctly after keyboard/navigate from column a to b. Notice that it's now impossible to keyboard/navigate. Does not apply to the read-only table.

[muzamalrana]

hi Can anyone help me here ? I am totally stuck. Trying to embedd image from link into dash table cell. Following is the code snippet.
html.Div([dt.DataTable(id='my-data',
columns=[
{"id": 'artist_image', "name": "artist_image", "presentation": "markdown"},
{"id": 'song_artist', "name": "song_artist"},
{"id": 'monthly_listeners', "name": "monthly_listeners"},
{"id": 'artist_followers', "name": "artist_followers"},
{"id": 'artist_popularity', "name": "artist_popularity"}]
,style_as_list_view=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
style_data_conditional=[
{
'if': {'row_index': 'odd'},
'backgroundColor': '#000000'
}
],
style_cell={'textAlign': 'center','padding': '5px','backgroundColor':'#00FF00',
'color':'white'},
style_header={
'backgroundColor': '#202020',
'fontWeight': 'bold','color':'white'
})],className='ten columns')])],className='row')], style=tab_style, selected_style=tab_selected_style)
@app.callback(
dash.dependencies.Output('my-data','data'),
[dash.dependencies.Input('my-graph1', 'clickData'),
dash.dependencies.Input('my_dropdown1', 'value'),
dash.dependencies.Input('radio_buttons', 'value'),
dash.dependencies.Input('my_dropdown3', 'value')]
)
def update_graph(click_data,dropdown,radio_button,dropdown3):
try:
genre1=click_data['points'][0]['customdata'][1]
except:
genre1='pop'
artist_data1=sm.artist_dataFrame(songs,artist,click_data,dropdown,radio_button)
print(artist_data1.head(1))
artist_data1=artist_data1[['artist_image','song_artist','monthly_listeners','artist_followers','artist_popularity','date_x']]
artist_data1['date_x']=pd.to_datetime(artist_data1['date_x'])
artist_data1=artist_data1[artist_data1['date_x']==artist_data1['date_x'].max()]
artist_data1=artist_data1.sort_values(by=radio_button,ascending=False).reset_index(drop=True)
artist_data1=artist_data1.drop(['date_x'],axis=1)
if dropdown3==0 or dropdown3=='All':
artist_data1=artist_data1
else:
artist_data1=artist_data1.head(dropdown3)
artist_data1['artist_image']=[html.Img(src=i) for i in artist_data1['artist_image']]
return artist_data1.to_dict('records')

[Marc-Andre-Rivet]

Fixes#596

• check virtualized performance

• virtualized behavior (varying row height -- breaks assumptions, this will be dealt outside the scope of this PR as this behavior is not caused by MD)

• varying row height with fixed columns (Using DataTable row_selectable and row_deletable with fixed columns results in mismatched row heights #649)

• navigation through and from markdown cell

• define copy/paste behavior (?!)

• define filter behavior -- for now filters will behave the same way for markdown cells as for any other text cells (the value filtered on is the actual value in the df) -- expect at least one visual/sanity test for the filter behavior (guard rail)

• define sort behavior (supporting/using displayed values is out of scope) -- for now sort will behave the same way for markdown cells as for any other text cells (the value sorted on is the actual value in the df) -- expect at least one visual/sanity test for the sort behavior (guard rail)

• visual tests -- cover as much of md syntax as possible

• behavior on non-text input value + test (e.g. number, boolean values, how does ReMarkable behave)

• markdown image resize (test, docs)

• more tests

• docs (will be done in https://github.com/plotly/dash-docs/pull/733)

• Test: Default highlight.js included with table does not assign itself to window.hljs

• Test: Loading external highlight.js will take precedence on the table's version

[Marc-Andre-Rivet]

1. Navigation works well when the user single clicks on the cell, on a double click the focus goes back to the page's body and navigation inside the table becomes impossible

2. Seeing quite a few percy diffs for fixed columns / rows test cases

Very nice visual tests 😄

[Marc-Andre-Rivet]

I've been thinking that maybe we can include the configuration for highlight.js usage, externalize the resource and expect special code block users to include the highlight.js they want as a resource from https://highlightjs.org/download/ ?! Either that or (1) not support it at all or (2) asyncing the resource -- in which case React.Suspense won't do and we'll have to handle it through state on CellMarkdown. This is mostly about completeness.. we should not load JS by default as it will involve a very special use case.

 private static md: Remarkable = new Remarkable({
highlight: (value: string, language: string) => {
const hljs: any = (window as any).hljs;
if (!hljs) {
return '';
}
if (language && hljs.getLanguage(language)) {
try {
return hljs.highlight(language, value).value;
} catch (_) { }
}
try {
return hljs.highlightAuto(value).value;
} catch (_) { }
return '';
}
});

I don't think we should flat out not support highlighting and I don't think this will ever be enough of a priority on its own to be fixed later, if possible I think I'd rather get it in w/ async and wrap it up.

[Marc-Andre-Rivet]

Since this is now our own cherry-picked version of Highlight.js would rather see this somewhere inside of /src, even if it's just /src/third-party/Highlight.js or some variation.

[Marc-Andre-Rivet]

Same as https://github.com/plotly/dash-table/pull/606/files#r358530234

[Marc-Andre-Rivet]

I think that with @types/Remarkable added, all of the remarkable definitions becomes redundant.

[Marc-Andre-Rivet]

This is now redundant

[Marc-Andre-Rivet]

@shammamah Testing navigation with:

import dash
from dash_html_components import Div
from dash_table import DataTable
app = dash.Dash(__name__)
data=[
dict(a=1, b='''```javascript
return 20;```
'''),
dict(a=1, b='''```js
console.log(\'pouet\');
'''),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```')
]
app.layout = Div([
DataTable(
columns=[
dict(name='a', id='a', editable=True),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
DataTable(
columns=[
dict(name='a', id='a'),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
])
app.run_server(debug=True)

Notice that for both tables it's possible to click/select a markdown cell and keyboard/navigate within the markdown column as expected. The top table, with editable=True, does not behave correctly after keyboard/navigate from column a to b. Notice that it's now impossible to keyboard/navigate. Does not apply to the read-only table.

[muzamalrana]

hi Can anyone help me here ? I am totally stuck. Trying to embedd image from link into dash table cell. Following is the code snippet.
html.Div([dt.DataTable(id='my-data',
columns=[
{"id": 'artist_image', "name": "artist_image", "presentation": "markdown"},
{"id": 'song_artist', "name": "song_artist"},
{"id": 'monthly_listeners', "name": "monthly_listeners"},
{"id": 'artist_followers', "name": "artist_followers"},
{"id": 'artist_popularity', "name": "artist_popularity"}]
,style_as_list_view=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
style_data_conditional=[
{
'if': {'row_index': 'odd'},
'backgroundColor': '#000000'
}
],
style_cell={'textAlign': 'center','padding': '5px','backgroundColor':'#00FF00',
'color':'white'},
style_header={
'backgroundColor': '#202020',
'fontWeight': 'bold','color':'white'
})],className='ten columns')])],className='row')], style=tab_style, selected_style=tab_selected_style)
@app.callback(
dash.dependencies.Output('my-data','data'),
[dash.dependencies.Input('my-graph1', 'clickData'),
dash.dependencies.Input('my_dropdown1', 'value'),
dash.dependencies.Input('radio_buttons', 'value'),
dash.dependencies.Input('my_dropdown3', 'value')]
)
def update_graph(click_data,dropdown,radio_button,dropdown3):
try:
genre1=click_data['points'][0]['customdata'][1]
except:
genre1='pop'
artist_data1=sm.artist_dataFrame(songs,artist,click_data,dropdown,radio_button)
print(artist_data1.head(1))
artist_data1=artist_data1[['artist_image','song_artist','monthly_listeners','artist_followers','artist_popularity','date_x']]
artist_data1['date_x']=pd.to_datetime(artist_data1['date_x'])
artist_data1=artist_data1[artist_data1['date_x']==artist_data1['date_x'].max()]
artist_data1=artist_data1.sort_values(by=radio_button,ascending=False).reset_index(drop=True)
artist_data1=artist_data1.drop(['date_x'],axis=1)
if dropdown3==0 or dropdown3=='All':
artist_data1=artist_data1
else:
artist_data1=artist_data1.head(dropdown3)
artist_data1['artist_image']=[html.Img(src=i) for i in artist_data1['artist_image']]
return artist_data1.to_dict('records')

[Marc-Andre-Rivet]

Fixes#596

• check virtualized performance

• virtualized behavior (varying row height -- breaks assumptions, this will be dealt outside the scope of this PR as this behavior is not caused by MD)

• varying row height with fixed columns (Using DataTable row_selectable and row_deletable with fixed columns results in mismatched row heights #649)

• navigation through and from markdown cell

• define copy/paste behavior (?!)

• define filter behavior -- for now filters will behave the same way for markdown cells as for any other text cells (the value filtered on is the actual value in the df) -- expect at least one visual/sanity test for the filter behavior (guard rail)

• define sort behavior (supporting/using displayed values is out of scope) -- for now sort will behave the same way for markdown cells as for any other text cells (the value sorted on is the actual value in the df) -- expect at least one visual/sanity test for the sort behavior (guard rail)

• visual tests -- cover as much of md syntax as possible

• behavior on non-text input value + test (e.g. number, boolean values, how does ReMarkable behave)

• markdown image resize (test, docs)

• more tests

• docs (will be done in https://github.com/plotly/dash-docs/pull/733)

• Test: Default highlight.js included with table does not assign itself to window.hljs

• Test: Loading external highlight.js will take precedence on the table's version

[Marc-Andre-Rivet]

1. Navigation works well when the user single clicks on the cell, on a double click the focus goes back to the page's body and navigation inside the table becomes impossible

2. Seeing quite a few percy diffs for fixed columns / rows test cases

Very nice visual tests 😄

[Marc-Andre-Rivet]

I've been thinking that maybe we can include the configuration for highlight.js usage, externalize the resource and expect special code block users to include the highlight.js they want as a resource from https://highlightjs.org/download/ ?! Either that or (1) not support it at all or (2) asyncing the resource -- in which case React.Suspense won't do and we'll have to handle it through state on CellMarkdown. This is mostly about completeness.. we should not load JS by default as it will involve a very special use case.

 private static md: Remarkable = new Remarkable({
highlight: (value: string, language: string) => {
const hljs: any = (window as any).hljs;
if (!hljs) {
return '';
}
if (language && hljs.getLanguage(language)) {
try {
return hljs.highlight(language, value).value;
} catch (_) { }
}
try {
return hljs.highlightAuto(value).value;
} catch (_) { }
return '';
}
});

I don't think we should flat out not support highlighting and I don't think this will ever be enough of a priority on its own to be fixed later, if possible I think I'd rather get it in w/ async and wrap it up.

[Marc-Andre-Rivet]

Since this is now our own cherry-picked version of Highlight.js would rather see this somewhere inside of /src, even if it's just /src/third-party/Highlight.js or some variation.

[Marc-Andre-Rivet]

Same as https://github.com/plotly/dash-table/pull/606/files#r358530234

[Marc-Andre-Rivet]

I think that with @types/Remarkable added, all of the remarkable definitions becomes redundant.

[Marc-Andre-Rivet]

This is now redundant

[Marc-Andre-Rivet]

@shammamah Testing navigation with:

import dash
from dash_html_components import Div
from dash_table import DataTable
app = dash.Dash(__name__)
data=[
dict(a=1, b='''```javascript
return 20;```
'''),
dict(a=1, b='''```js
console.log(\'pouet\');
'''),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```'),
dict(a=1, b='```js console.log(\'pouet\'); ```')
]
app.layout = Div([
DataTable(
columns=[
dict(name='a', id='a', editable=True),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
DataTable(
columns=[
dict(name='a', id='a'),
dict(name='b', id='b', type='text', presentation='markdown')
],
data=data,
editable=False
),
])
app.run_server(debug=True)

Notice that for both tables it's possible to click/select a markdown cell and keyboard/navigate within the markdown column as expected. The top table, with editable=True, does not behave correctly after keyboard/navigate from column a to b. Notice that it's now impossible to keyboard/navigate. Does not apply to the read-only table.

[muzamalrana]

hi Can anyone help me here ? I am totally stuck. Trying to embedd image from link into dash table cell. Following is the code snippet.
html.Div([dt.DataTable(id='my-data',
columns=[
{"id": 'artist_image', "name": "artist_image", "presentation": "markdown"},
{"id": 'song_artist', "name": "song_artist"},
{"id": 'monthly_listeners', "name": "monthly_listeners"},
{"id": 'artist_followers', "name": "artist_followers"},
{"id": 'artist_popularity', "name": "artist_popularity"}]
,style_as_list_view=True,
filter_action="native",
sort_action="native",
sort_mode="multi",
column_selectable="single",
style_data_conditional=[
{
'if': {'row_index': 'odd'},
'backgroundColor': '#000000'
}
],
style_cell={'textAlign': 'center','padding': '5px','backgroundColor':'#00FF00',
'color':'white'},
style_header={
'backgroundColor': '#202020',
'fontWeight': 'bold','color':'white'
})],className='ten columns')])],className='row')], style=tab_style, selected_style=tab_selected_style)
@app.callback(
dash.dependencies.Output('my-data','data'),
[dash.dependencies.Input('my-graph1', 'clickData'),
dash.dependencies.Input('my_dropdown1', 'value'),
dash.dependencies.Input('radio_buttons', 'value'),
dash.dependencies.Input('my_dropdown3', 'value')]
)
def update_graph(click_data,dropdown,radio_button,dropdown3):
try:
genre1=click_data['points'][0]['customdata'][1]
except:
genre1='pop'
artist_data1=sm.artist_dataFrame(songs,artist,click_data,dropdown,radio_button)
print(artist_data1.head(1))
artist_data1=artist_data1[['artist_image','song_artist','monthly_listeners','artist_followers','artist_popularity','date_x']]
artist_data1['date_x']=pd.to_datetime(artist_data1['date_x'])
artist_data1=artist_data1[artist_data1['date_x']==artist_data1['date_x'].max()]
artist_data1=artist_data1.sort_values(by=radio_button,ascending=False).reset_index(drop=True)
artist_data1=artist_data1.drop(['date_x'],axis=1)
if dropdown3==0 or dropdown3=='All':
artist_data1=artist_data1
else:
artist_data1=artist_data1.head(dropdown3)
artist_data1['artist_image']=[html.Img(src=i) for i in artist_data1['artist_image']]
return artist_data1.to_dict('records')