fix

.vscode/launch.json

@@ -0,0 +1,21 @@
+{
+    "version": "0.2.0",
+    "configurations": [{
+         "name": "Python: Streamlit",
+         "type": "python",
+         "request": "launch",
+         "module": "streamlit",
+         "env": {
+             "STREAMLIT_APP": "app.py",
+             "STREAMLIT_ENV": "development",
+             "AWS_PROFILE": "root",
+             "PYTHONPATH": "${workspaceRoot}/src",
+         },
+         "args": [
+             "run",
+             "D:/Terradot/huggingface/df_align/app.py"
+         ],
+         "jinja": true
+     }
+    ]
+ }

app.py

@@ -32,8 +32,8 @@ st.sidebar.title('Proyecto Crea')
 st.sidebar.write('Solo uso interno')
 # add sidebar with 2 upload buttons
 st.sidebar.header('Upload Files')
-uploaded_file = st.sidebar.file_uploader('Upload your shapefile', type=['shp', 'geojson', 'json'])
-uploaded_file2 = st.sidebar.file_uploader('Upload your csv file', type=['csv'])
+uploaded_file = st.sidebar.file_uploader('Upload your shapefile', type=['shp', 'geojson', 'json'], disabled = True)
+uploaded_file2 = st.sidebar.file_uploader('Upload your csv file', type=['csv'], disabled = True)
 
 if uploaded_file is not None:
         lotes_gdf = convert_to_gdf(uploaded_file)
@@ -118,46 +118,51 @@ if test:
     lotes_gdf['campo_obs'] = lotes_gdf['campo'].map(similar_dict)
 
     cutoff = 0.3
-    for campo , df in lotes_gdf.groupby('campo'):
-        for i, row in df.iterrows():
-            c_obs = obs_df_2023.loc[obs_df_2023.Campo == row.campo_obs]
-            lote_obs = c_obs.Lote.unique()
-            # normalize the stings to lowercase and remove punctuation
-            lote_obs_norm = [str(c).lower() for c in lote_obs]
-            lote_obs_norm = [c.replace('.', ' ') for c in lote_obs_norm]
-            similar_items = get_close_matches(row.lote, lote_obs_norm, N, cutoff)
-            # get the index of the most similar item
-            similar_items_idx = [lote_obs_norm.index(i) for i in similar_items]
-
-            # get the most similar item in the original list
-            similar_items = [lote_obs[i] for i in similar_items_idx]
-
-            if similar_items:
-                lotes_gdf.loc[i, 'lote_obs'] = similar_items[0]
-            else:
-                lotes_gdf.loc[i, 'lote_obs'] = "no match"
+    # for campo , df in lotes_gdf.groupby('campo'):
+    #     for i, row in df.iterrows():
+    #         c_obs = obs_df_2023.loc[obs_df_2023.Campo == row.campo_obs]
+    #         lote_obs = c_obs.Lote.unique()
+    #         # normalize the stings to lowercase and remove punctuation
+    #         lote_obs_norm = [str(c).lower() for c in lote_obs]
+    #         lote_obs_norm = [c.replace('.', ' ') for c in lote_obs_norm]
+    #         similar_items = get_close_matches(row.lote, lote_obs_norm, N, cutoff)
+    #         # get the index of the most similar item
+    #         similar_items_idx = [lote_obs_norm.index(i) for i in similar_items]
+
+    #         # get the most similar item in the original list
+    #         similar_items = [lote_obs[i] for i in similar_items_idx]
+
+    #         if similar_items:
+    #             lotes_gdf.loc[i, 'lote_obs'] = similar_items[0]
+    #         else:
+    #             lotes_gdf.loc[i, 'lote_obs'] = "no match"
 
 
     def show_field(key):
+        key, selected_value = key
+        lote_obs = obs_df_2023[obs_df_2023['Campo'] == selected_value]['Lote'].unique().tolist()
+        lote_obs.insert(0, 'no match')
         with col3:
             # st.header(st.session_state['key'])
             st.header('Lote')
-            df_field = lotes_gdf[lotes_gdf['campo'] == st.session_state['key']]
+            df_field = lotes_gdf[lotes_gdf['campo'] == key]
             fields = df_field['lote'].unique().tolist()
-            fields_obs = df_field['lote_obs'].unique().tolist()
+
             with st.form("fields_form"):
+                
                 for j,field in enumerate(fields):
-                    default = df_field[df_field['lote']==field]['lote_obs'].values[0]
-                    selected_value = st.multiselect(f'{field} (.shp):', fields_obs, default=default, key='field'+str(j))
-
-                if  st.form_submit_button("Submit"):
-                    for j,field in enumerate(fields):
-                        if selected_value:
-                            lotes_gdf.loc[(lotes_gdf['campo'] == st.session_state['key']) & (lotes_gdf['lote'] == field), 'lote_obs'] = selected_value[j]
-                        else:
-                            lotes_gdf.loc[(lotes_gdf['campo'] == st.session_state['key']) & (lotes_gdf['lote'] == field), 'lote_obs'] = 'no match'
+                    similar_items = get_close_matches(field, lote_obs, 3, 0.70)
+                    default = similar_items[0] if similar_items else 'no match'
+                    # selected_value = st.multiselect(f'{field} (.shp):', lote_obs, default=default, key='field'+str(j))
+                    selected_value = st.selectbox(f'{field} (.shp):', lote_obs, index = lote_obs.index(default)  , key='field'+str(j))
+
+                    # if selected_value:
+                    #     lotes_gdf.loc[(lotes_gdf['campo'] == key) & (lotes_gdf['lote'] == field), 'lote_obs'] = selected_value
+                    # else:
+                    lotes_gdf.loc[(lotes_gdf['campo'] == key) & (lotes_gdf['lote'] == field), 'lote_obs'] = default
                 
-                # st.dataframe(df.drop(columns=['geometry','nombre','apellido','campo','campo_obs']))
+                if st.form_submit_button("Submit"):
+                    st.dataframe(lotes_gdf.drop(columns=['geometry','nombre','apellido','campo','campo_obs']))
 
 
     def on_click(key):
@@ -169,14 +174,20 @@ if test:
     with col2:
         st.header('Campo')
         for i, key in enumerate(all_keys):
-            # selected_value = st.selectbox(f'{key}:', campo_obs, index=campo_obs.index(similar_dict_df[similar_dict_df['gdf'] == key]['obs'].values[0]))
-            selected_value = st.multiselect(f'{key} (.shp):', campo_obs, default=similar_dict_df[similar_dict_df['gdf'] == key]['obs'].values[0], key=i)
+            selected_value = st.selectbox(f'{key}:', campo_obs, index=campo_obs.index(similar_dict_df[similar_dict_df['gdf'] == key]['obs'].values[0]))
+            # selected_value = st.multiselect(f'{key} (.shp):', campo_obs, default=similar_dict_df[similar_dict_df['gdf'] == key]['obs'].values[0], key=i)
+            lotes_gdf.loc[(lotes_gdf['campo'] == key) & (lotes_gdf['empresa'] == selected_company), 'campo_obs'] = similar_dict_df[similar_dict_df['gdf'] == key]['obs'].values[0]
+
             if selected_value:
-                similar_dict_df.loc[similar_dict_df['gdf'] == key, 'obs'] = selected_value[0]
+                similar_dict_df.loc[similar_dict_df['gdf'] == key, 'obs'] = selected_value
+                value = selected_value
+                lotes_gdf.loc[(lotes_gdf['campo'] == key) & (lotes_gdf['empresa'] == selected_company), 'campo_obs'] = selected_value
             else:
-                similar_dict_df.loc[similar_dict_df['gdf'] == key, 'obs'] = 'no match'
+                # similar_dict_df.loc[similar_dict_df['gdf'] == key, 'obs'] = 'no match'
+                value = similar_dict_df.loc[similar_dict_df['gdf'] == key, 'obs']
+                lotes_gdf.loc[(lotes_gdf['campo'] == key) & (lotes_gdf['empresa'] == selected_company), 'campo_obs'] = value
 
-            st.button('Show Fields', key=key, on_click=on_click(key))
+            st.button('Show Fields', key=key, on_click=on_click([key,value]))
 
 
         # st.dataframe(similar_dict_df)
@@ -191,7 +202,7 @@ if test:
 st.sidebar.download_button(
         label="Download GeoJSON",
         data=lotes_gdf.to_json().encode('utf-8'),
-        file_name='lotes.geojson',
+        file_name=f'{selected_company}.geojson',
         # mime='text/csv',
         mime = 'application/json',
     )