Spaces:
Sleeping
Sleeping
File size: 4,693 Bytes
8b9797b d747029 8b9797b d747029 8b9797b d747029 8b9797b d747029 8b9797b d747029 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 |
import streamlit as st
import pandas as pd
# 1. Loading the dataset
df = pd.read_excel("data/table.xlsx")
# 2. Preprocessing the dataset
df["Bionic prototype"] = df["Bionic prototype"].str.strip()
df["Materials"] = df["Materials"].str.split(";").apply(lambda x: [material.strip() for material in x])
df["Method"] = df["Method"].str.split(";").apply(lambda x: [method.strip() for method in x])
df["Multifunction"] = df["Multifunction"].str.split(";").apply(lambda x: [multifunction.strip() for multifunction in x])
# 3. Saving the processed dataset
# df.to_excel("data/filtered_table.xlsx", index=False)
# 4. Extracting a unique list for each column
bionic_prototype_list = df["Bionic prototype"].unique()
method_list = df["Method"].explode().unique()
multifunction_list = df["Multifunction"].explode().unique()
bionic_prototype_list.sort()
method_list.sort()
multifunction_list.sort()
ordered_multifunction_list = ["Antifogging", "Self-cleaning", "Antireflective", "Antibacterial", "Anti-icing",
"Antiwetting", "Large FOV", "Fast motion detection", "Structural color",
"Droplet directional migration", "Anti-drag", "Water collection",
"Self-propelled actuator"]
other_multifunction_list = [x for x in multifunction_list if x not in ordered_multifunction_list]
ordered_multifunction_list.extend(other_multifunction_list)
def show_res_link(row_idx):
res_link = df.iloc[row_idx]['Res link']
st.sidebar.info(f'Resource Link: {res_link}')
def show_bionic_prototype_content(row_idx):
bionic_prototype = df.iloc[row_idx]['Bionic prototype']
st.sidebar.info(f'Bionic prototype is: {bionic_prototype}')
with st.sidebar:
st.slider(
label="Search results limit",
min_value=1,
max_value=50,
value=20,
step=1,
key="limit",
help="Limit the number of search results",
)
st.multiselect(
label="Display columns",
options=["Bionic prototype", "Multifunction", "Method", "Materials", "Res link"],
default=["Bionic prototype", "Multifunction", "Method", "Materials", "Res link"],
help="Select columns to display in the search results",
key="display_columns",
)
st.title("Bionic Path Selection")
st.multiselect(
label="Multifunction",
options=ordered_multifunction_list,
default=[],
help="Select the multifunction to display in the search results",
placeholder="Select the multifunction to display in the search results",
key="multifunction_option"
)
st.session_state.disabled = False if len(st.session_state.multifunction_option) > 0 else True
left_col, right_col = st.columns(2)
with left_col:
st.selectbox(
label="Bionic prototype",
options=bionic_prototype_list,
help="Select the bionic prototype to display in the search results",
placeholder="Select the bionic prototype to display in the search results",
index=None,
key="bionic_prototype_option",
disabled=st.session_state.disabled
)
with right_col:
st.multiselect(
label="Method",
options=method_list,
default=[],
help="Select the method to display in the search results",
placeholder="Select the method to display in the search results",
key="method_option",
disabled=st.session_state.disabled
)
search = st.button("Search")
if search:
multifunction_option = st.session_state.multifunction_option
bionic_prototype_option = st.session_state.bionic_prototype_option
method_option = st.session_state.method_option
# Filter the multifunction column
filtered_df = df[
df["Multifunction"].apply(lambda x: all(multifunction in x for multifunction in multifunction_option))]
# Filter the bionic prototype column
filtered_df = filtered_df[filtered_df["Bionic prototype"] == bionic_prototype_option] \
if (bionic_prototype_option is not None and not st.session_state.disabled and not filtered_df.empty) \
else filtered_df
# Filter the method column
filtered_df = filtered_df[filtered_df["Method"].apply(lambda x: any(method in x for method in method_option))] \
if (len(method_option) > 0 and not st.session_state.disabled and not filtered_df.empty) \
else filtered_df
# Reset the index
filtered_df = filtered_df.reset_index(drop=True)
st.dataframe(
filtered_df[st.session_state.display_columns].head(st.session_state.limit),
column_config={
"Res link": st.column_config.LinkColumn(
"Resource", display_text=None
)
}
)
|