Jan Mühlnikel
commited on
Commit
·
ad2cbdd
1
Parent(s):
bbb4e78
data update
Browse files- functions/filter_projects.py +8 -3
- similarity_page.py +9 -3
functions/filter_projects.py
CHANGED
|
@@ -5,9 +5,13 @@ def contains_code(crs_codes, code_list):
|
|
| 5 |
codes = str(crs_codes).split(';')
|
| 6 |
return any(code in code_list for code in codes)
|
| 7 |
|
| 8 |
-
def filter_projects(df, crs3_list, crs5_list, sdg_str, country_code_list, orga_code_list,
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
# Check if filters where not all should be selected are empty
|
| 10 |
-
if crs3_list != [] or crs5_list != [] or sdg_str != "" or query != "":
|
| 11 |
|
| 12 |
# FILTER CRS
|
| 13 |
if crs3_list and not crs5_list:
|
|
@@ -35,11 +39,12 @@ def filter_projects(df, crs3_list, crs5_list, sdg_str, country_code_list, orga_c
|
|
| 35 |
df = df[df['orga_abbreviation'].isin(orga_code_list)]
|
| 36 |
|
| 37 |
# FILTER QUERY
|
|
|
|
| 38 |
if query != "" and len(df) > 0:
|
| 39 |
if len(df) < TOP_X_PROJECTS:
|
| 40 |
TOP_X_PROJECTS = len(df)
|
| 41 |
df = search(query, model, embeddings, df, TOP_X_PROJECTS)
|
| 42 |
-
|
| 43 |
|
| 44 |
|
| 45 |
return df
|
|
|
|
| 5 |
codes = str(crs_codes).split(';')
|
| 6 |
return any(code in code_list for code in codes)
|
| 7 |
|
| 8 |
+
def filter_projects(df, crs3_list, crs5_list, sdg_str, country_code_list, orga_code_list,
|
| 9 |
+
#query,
|
| 10 |
+
model,
|
| 11 |
+
#embeddings,
|
| 12 |
+
TOP_X_PROJECTS=30):
|
| 13 |
# Check if filters where not all should be selected are empty
|
| 14 |
+
if crs3_list != [] or crs5_list != [] or sdg_str != "": #or query != "":
|
| 15 |
|
| 16 |
# FILTER CRS
|
| 17 |
if crs3_list and not crs5_list:
|
|
|
|
| 39 |
df = df[df['orga_abbreviation'].isin(orga_code_list)]
|
| 40 |
|
| 41 |
# FILTER QUERY
|
| 42 |
+
"""
|
| 43 |
if query != "" and len(df) > 0:
|
| 44 |
if len(df) < TOP_X_PROJECTS:
|
| 45 |
TOP_X_PROJECTS = len(df)
|
| 46 |
df = search(query, model, embeddings, df, TOP_X_PROJECTS)
|
| 47 |
+
"""
|
| 48 |
|
| 49 |
|
| 50 |
return df
|
similarity_page.py
CHANGED
|
@@ -111,6 +111,7 @@ def load_model():
|
|
| 111 |
return model
|
| 112 |
|
| 113 |
# Load Embeddings
|
|
|
|
| 114 |
@st.cache_data
|
| 115 |
def load_embeddings_and_index():
|
| 116 |
# Load embeddings
|
|
@@ -119,7 +120,7 @@ def load_embeddings_and_index():
|
|
| 119 |
embeddings = stored_data["embeddings"]
|
| 120 |
|
| 121 |
return embeddings
|
| 122 |
-
|
| 123 |
|
| 124 |
# USE CACHE FUNCTIONS
|
| 125 |
sim_matrix = load_sim_matrix()
|
|
@@ -198,7 +199,7 @@ def show_multi_matching_page():
|
|
| 198 |
)
|
| 199 |
|
| 200 |
# SEARCH BOX
|
| 201 |
-
query = st.text_input("Search Query")
|
| 202 |
|
| 203 |
with col3:
|
| 204 |
# COUNTRY SELECTION
|
|
@@ -242,7 +243,11 @@ def show_multi_matching_page():
|
|
| 242 |
|
| 243 |
# FILTER DF WITH SELECTED FILTER OPTIONS
|
| 244 |
TOP_X_PROJECTS = 30
|
| 245 |
-
filtered_df = filter_projects(projects_df, crs3_list, crs5_list, sdg_str, country_code_list, orga_code_list,
|
|
|
|
|
|
|
|
|
|
|
|
|
| 246 |
if isinstance(filtered_df, pd.DataFrame) and len(filtered_df) != 0:
|
| 247 |
# FIND MATCHES
|
| 248 |
## If only same country checkbox i sactivated
|
|
@@ -312,6 +317,7 @@ def show_single_matching_page():
|
|
| 312 |
else:
|
| 313 |
search_list = title_search_list
|
| 314 |
|
|
|
|
| 315 |
project_option = st.selectbox(
|
| 316 |
label = 'Search for a project',
|
| 317 |
index = None,
|
|
|
|
| 111 |
return model
|
| 112 |
|
| 113 |
# Load Embeddings
|
| 114 |
+
"""
|
| 115 |
@st.cache_data
|
| 116 |
def load_embeddings_and_index():
|
| 117 |
# Load embeddings
|
|
|
|
| 120 |
embeddings = stored_data["embeddings"]
|
| 121 |
|
| 122 |
return embeddings
|
| 123 |
+
"""
|
| 124 |
|
| 125 |
# USE CACHE FUNCTIONS
|
| 126 |
sim_matrix = load_sim_matrix()
|
|
|
|
| 199 |
)
|
| 200 |
|
| 201 |
# SEARCH BOX
|
| 202 |
+
#query = st.text_input("Search Query")
|
| 203 |
|
| 204 |
with col3:
|
| 205 |
# COUNTRY SELECTION
|
|
|
|
| 243 |
|
| 244 |
# FILTER DF WITH SELECTED FILTER OPTIONS
|
| 245 |
TOP_X_PROJECTS = 30
|
| 246 |
+
filtered_df = filter_projects(projects_df, crs3_list, crs5_list, sdg_str, country_code_list, orga_code_list,
|
| 247 |
+
#query,
|
| 248 |
+
model,
|
| 249 |
+
#embeddings,
|
| 250 |
+
TOP_X_PROJECTS)
|
| 251 |
if isinstance(filtered_df, pd.DataFrame) and len(filtered_df) != 0:
|
| 252 |
# FIND MATCHES
|
| 253 |
## If only same country checkbox i sactivated
|
|
|
|
| 317 |
else:
|
| 318 |
search_list = title_search_list
|
| 319 |
|
| 320 |
+
|
| 321 |
project_option = st.selectbox(
|
| 322 |
label = 'Search for a project',
|
| 323 |
index = None,
|