import os
import typing
import streamlit as st
from kiara import Kiara
from kiara_modules.playground.markus.streamlit import (
find_all_aliases_of_type,
init_session,
onboard_file,
)
# we'll use a pipeline description here, instead of a Python kiara module name
pipeline_file = os.path.join(os.path.dirname(__file__), "..", "..",
"pipelines", "create_graph_from_tables.json")
kiara_obj, workflow = init_session(st, module_type=pipeline_file)
st.title("Kiara/streamlit experiment - create a network graph")
def table_mgmt(kiara: Kiara):
fp = st.sidebar.file_uploader("Import table(s) from csv file(s)",
type=["csv"],
accept_multiple_files=True)
if fp:
onboard_file(kiara=kiara, st=st, uploaded_file=fp)
st.sidebar.write("## All your tables:")
all_table_aliases = find_all_aliases_of_type(kiara, value_type="table")
if not all_table_aliases:
st.title("Kiara/streamlit experiment - Topic modelling")
# This is the workflow we are going to use, check out the json file for details.
workflow_file = os.path.join(os.path.dirname(__file__), "..", "..",
"pipelines", "topic_modeling_end_to_end.json")
# this is a preparation step that:
# - creates the kiara session
# - creates a workflow from our json file
# - saves the workflow object in the streamlit session state object
#
# The last step is useful, because a kiara workflow handles the workflow state itself (e.g. keeping already processed
# results in memory, only re-compute certain steps if relevant inputs have changed, etc.
# This means that we don't need (or want) streamlit to get in the way with its own caching (although in some cases we'll
# take advantage of that too.
kiara, workflow = init_session(st, module_type=workflow_file)
# the variable that holds the user input that points to the folder that contains the txt files/corpus.
path = None
with st.form(key="create_graph"):
path = st.text_input("Path to corpus")
load_path = st.form_submit_button(label="Load corpus")
# setting up some input controls on the sidebar
# compute coherence and num_topic are not used at the moment because of some bug with spacy (or gensim -- not sure)
languages = st.sidebar.multiselect("Languages",
["german", "italian", "english"])
# compute_coherence = st.sidebar.checkbox("Compute coherence", value=False)
# num_topics = st.sidebar.slider("Num topic", min_value=1, value=7, max_value=20)
# if no path was specified, it makes no sense to continue
import streamlit as st
from kiara_modules.playground.markus.streamlit import (
find_all_aliases_of_type,
init_session,
onboard_file,
set_workflow_input,
)
st.title("Kiara/streamlit experiment: simple sql query")
# initialize kiar workflow and store it in the streamlit session state
module_type = "table.query.sql"
kiara, workflow = init_session(st, module_type=module_type)
# add a file import widget, so users can import files if their kiara data store is empty
fp = st.file_uploader(
label="Onboard table data (.csv)", type=["csv"], accept_multiple_files=True
)
if fp:
onboard_file(kiara=kiara, st=st, uploaded_file=fp)
selected_alias = st.selectbox(
"Select dataset", find_all_aliases_of_type(kiara, value_type="table")
)
if not selected_alias:
st.stop()