Create FeatureLayer from scratch

I'm trying to create new instance of arcgis.features.FeatureLayer using python arcgis package.

I find it impossible, I'm traversing through documentation and examples and more examples, trying different things and I find myself going round in circles and slowly I start pulling my hear out...

This is what I have:

import arcgis
import json
from arcgis.gis import GIS

g = GIS(username="my_uname", password="my_pwd")
prop = arcgis.features.FeatureCollection(dictdata={'attributes': {'foo': 'bar', 'lorem': 'ipsum'}})
prop_json=json.dumps({"featureCollection": {"layers": [dict(prop.properties)]}})
item_properties={"type": "Feature Collection", "title": "test_feature_collection_01", "text": prop_json}
it = g.content.add(item_properties=item_properties)

At this point - I can't understand why it.layers yields empty results. I believe item_properties is malformed resulting in arcgis ignoring my layers definition... but I have nowhere to check what it should look like. I figured I'd like to use something from arcgis to generate layer definition for me rather than to handcraft JSON myself so it's future-proof.

What I want to do with that item later is this:

lr = arcgis.features.FeatureLayer.fromitem(item=it)

This fails with TypeError: item must be a type of service, not Feature Collection

So I figured I could publish item and use it here.

pit = it.publish()
lr = arcgis.features.FeatureLayer.fromitem(item=pit)

I need FeatureLayer to be able to call append (so I can just throw extra data each time I have anything new I want to push)

But to my surprise I cannot even publish the item and I'm getting Exception: Job failed. (to my bigger surprise item actually gets published as I can see it through content manager website)

I also tried creating "CSV" item type:

import arcgis
import json
from arcgis.gis import GIS


g = GIS(username="my_uname", password="my_pwd")
item_properties={"type": "CSV", "title": "test_feature_collection_01"}
it = g.content.add(item_properties=item_properties, data="/tmp/foo.csv")
pit = it.publish()

lr = arcgis.features.FeatureLayer.fromitem(item=pit)

However this results in layer-less item, which results in unhandled exception IndexError: list index out of range (because method arcgis calls tries to get to layers which is empty...)

Please help...

I managed to get sort-of what I need... however I'm still trying to understand how to make arcgis to upload some data directly from my hard drive rather than from another item. Anyways here is the code:

import arcgis
import json
from arcgis.gis import GIS

g = GIS(username="my_uname", password="my_pwd")

item_properties_1={"type": "CSV", "title": "test_feature_collection_01"}
it_1 = g.content.add(item_properties=item_properties_1, data="/tmp/my_data.csv")
pit_1 = it_1.publish()
table = arcgis.features.Table.fromitem(item=pit_1)

# Now we want to throw some extra data, do we have to create another item to do that??

item_properties_2={"type": "CSV", "title": "test_feature_collection_02"}
it_2 = g.content.add(item_properties=item_properties_2, data="/tmp/more_data.csv")
source_info = g.content.analyze(file_path='/tmp/more_data.csv', file_type='csv', location_type='none')

# Data from `it_2` appears to be appended to `table` container (which is published item, not item itself)

table.append(item_id=it_2.id, upload_format='csv', source_info=source_info["publishParameters"], upsert=False)

So to append some data to existing item I have to create new item...

Edit:

To upload data without having to create interim item use edit_features instead of append.

# create table like shown in sample above

from arcgis import features
f = features.Feature.from_dict({"attributes": {"some_attribute": "value"}})
fs = features.FeatureSet(features=[f])
table.edit_features(adds=fs)