Update app.py

app.py

@@ -6,7 +6,6 @@ from PIL import Image
 
 import pandas as pd
 import streamlit as st
-from fsspec.implementations.local import LocalFileSystem
 from huggingface_hub import HfFileSystem
 
 import streamlit.components.v1 as components
@@ -24,24 +23,19 @@ class Field:
     other_params: Optional[Dict[str, object]] = field(default_factory=lambda: {})
 
 # Function to get user ID from URL
-def get_param_from_url(param):
-    user_id = st.query_params.get(param, "")
+def get_user_id_from_url():
+    user_id = st.query_params.get("user_id", "")
     return user_id
 
+HF_TOKEN = os.environ.get("HF_TOKEN_WRITE")
+print("is none?", HF_TOKEN is None)
+hf_fs = HfFileSystem(token=HF_TOKEN)
 
 ########################################################################################
 # CHANGE THE FOLLOWING VARIABLES ACCORDING TO YOUR NEEDS
 
-# 'local' or 'hf'. hf is for Hugging Face file system but has limits on the number of access per hour
-filesystem = 'hf'
-# path to repo or local file system TODO rename
 input_repo_path = 'datasets/emvecchi/annotation'
 output_repo_path = 'datasets/emvecchi/annotation'
-#filesystem = 'local'
-# path to repo or local file system
-#input_repo_path = '/data/mod_pred_annotation'
-#output_repo_path = '/data/mod_pred_annotation'
-
 to_annotate_file_name = 'to_annotate.csv'  # CSV file to annotate
 COLS_TO_SAVE = ['comment_id','comment','confidence_score']
 
@@ -194,11 +188,6 @@ fields: List[Field] = [
         Field(name="other_comments", type="text", title="Please provide any additional details or information: *(optional)*", mandatory=False),
     ]),
 ]
-
-url_conditional_fields = [
-    Field(name="skip", type="skip_checkbox",
-          title="I am uncomfortable annotating this text and  voluntarily skip this instance", mandatory=False)
-]
 INPUT_FIELD_DEFAULT_VALUES = {'slider': 0,
                               'text': '',
                               'textarea': '',
@@ -212,27 +201,11 @@ SHOW_HELP_ICON = False
 SHOW_VALIDATION_ERROR_MESSAGE = True
 
 ########################################################################################
-
-if filesystem == 'hf':
-    HF_TOKEN = os.environ.get("HF_TOKEN_WRITE")
-    print("is none?", HF_TOKEN is None)
-    hf_fs = HfFileSystem(token=HF_TOKEN)
-else:
-    hf_fs = LocalFileSystem()
-
-def get_start_index():
-    if hf_fs.exists(output_repo_path + '/' + get_base_path()):
-        files = hf_fs.ls(output_repo_path + '/' + get_base_path())
-        return len(files) - 2
-    return -3
-
-
-def read_data():
-    #assert st.session_state.phase, "Phase not provided"
-    #with hf_fs.open(input_repo_path + '/' + to_annotate_file_name.format(phase=st.session_state.phase)) as f:
-    with hf_fs.open(input_repo_path + '/' + to_annotate_file_name) as f:
+def read_data(_path):
+    with hf_fs.open(input_repo_path + '/' + _path) as f:
         return pd.read_csv(f)
 
+
 def read_saved_data():
     _path = get_path()
     if hf_fs.exists(output_repo_path + '/' + _path):
@@ -243,20 +216,16 @@ def read_saved_data():
                 print(e)
     return None
 
+
 # Write a remote file
 def save_data(data):
-    if not hf_fs.exists(f"{output_repo_path}/{get_base_path()}"):
-        hf_fs.mkdir(f"{output_repo_path}/{get_base_path()}")
+    hf_fs.mkdir(f"{output_repo_path}/{data['user_id']}")
     with hf_fs.open(f"{output_repo_path}/{get_path()}", "w") as f:
         f.write(json.dumps(data))
 
 
-def get_base_path():
-    #return f"{st.session_state.phase}/{st.session_state.batch}/{st.session_state.user_id}"
-    return f"{st.session_state.user_id}"
-
 def get_path():
-    return f"{get_base_path()}/{st.session_state.current_index}.json"
+    return f"{st.session_state.user_id}/{st.session_state.current_index}.json"
 
 
 def display_image(image_path):
@@ -307,6 +276,7 @@ def show_field(f: Field, index: int, data_collected):
         st.session_state.following_mandatory = False
         match f.type:
             case 'input_col':
+                st.write(f.title)
                 if f.name == 'image_name':
                     st.write(f.title)
                     image_name = st.session_state.data.iloc[index][f.name]
@@ -314,10 +284,7 @@ def show_field(f: Field, index: int, data_collected):
                         image_path = os.path.join(input_repo_path, 'images', image_name)
                         display_image(image_path)
                 else:
-                    value = st.session_state.data.iloc[index][f.name]
-                    if value and value is not np.nan:
-                        st.write(f.title)
-                        st.write(value)
+                    st.write(st.session_state.data.iloc[index][f.name])
             case 'markdown':
                 st.markdown(f.title)
             case 'expander' | 'container':
@@ -335,17 +302,14 @@ def show_field(f: Field, index: int, data_collected):
             f.title = f'**{f.title}**\n\n{f.help}' if f.help else f.title
 
         validation_error = False
-
-        # form is not displayed for first time
-        if st.session_state.form_displayed == st.session_state.current_index:
-            if f.mandatory or st.session_state.following_mandatory:
+        if f.mandatory or st.session_state.following_mandatory:
+            # form is not displayed for first time
+            if st.session_state.form_displayed == st.session_state.current_index:
                 if st.session_state[key] == INPUT_FIELD_DEFAULT_VALUES[f.type]:
                     st.session_state.valid = False
                     validation_error = True
                 elif f.following_mandatory_values and st.session_state[key] in f.following_mandatory_values:
                     st.session_state.following_mandatory = True
-
-        if f.mandatory or st.session_state.following_mandatory:
             f.title += " :red[* required!]" if (validation_error and not SHOW_VALIDATION_ERROR_MESSAGE) else' :red[*]'
         f.help = None
 
@@ -407,32 +371,28 @@ def show_fields(fields: List[Field]):
     data_collected = read_saved_data()
     st.session_state.data_inputs_keys = []
     st.session_state.following_mandatory = False
-    
+
     for field in fields:
         show_field(field, index, data_collected)
 
     submitted = st.form_submit_button("Submit")
     if submitted:
-        skip_sample = ('skip' in st.session_state and st.session_state['skip'])
-        if not skip_sample and not st.session_state.valid:
+        if not st.session_state.valid:
             st.error("Please fill in all mandatory fields")
             # st.rerun() # filed-out values are not shown otherwise
         else:
             with st.spinner(text="saving"):
-                prep_and_save_data(index, skip_sample)
+                save_data({
+                    'user_id': st.session_state.user_id,
+                    'index': st.session_state.current_index,
+                    **(st.session_state.data.iloc[index][COLS_TO_SAVE].to_dict() if index >= 0 else {}),
+                    **{k: st.session_state[k+str(index)] for k in st.session_state.data_inputs_keys}
+                })
             st.success("Feedback submitted successfully!")
             navigate(1)
 
     st.session_state.form_displayed = st.session_state.current_index
 
-def prep_and_save_data(index, skip_sample):
-    save_data({
-        'user_id': st.session_state.user_id,
-        'index': st.session_state.current_index,
-        **(st.session_state.data.iloc[index][COLS_TO_SAVE].to_dict() if index >= 0 else {}),
-        **{k: st.session_state[k + str(index)] for k in st.session_state.data_inputs_keys},
-        'skip': skip_sample
-    })
 
 #st.set_page_config(layout='wide')
 # Title of the app
@@ -448,39 +408,15 @@ div[data-testid="stMarkdownContainer"] > p {
     </style>
     """, unsafe_allow_html=True)
 
-
-def add_annotation_guidelines():
-    st.write(f"username is {st.session_state.user_id}")
-    st.markdown(
-        "<details open><summary>Annotation Guidelines</summary>" + guidelines_text + "</details>"
-        , unsafe_allow_html=True)
-
-
-#st.session_state.phase = get_param_from_url("phase")
 # Load the data to annotate
 if 'data' not in st.session_state:
-    #st.session_state.batch = get_param_from_url("batch")
-    #data = read_data()
-    #if st.session_state:
-    #    st.session_state.data = data[data['batch'] == int(st.session_state.batch)]
-    #else:
-    #    raise ValueError("Batch not provided")
-    st.session_state.data = read_data()
-    data = read_data()
-    
-user_id_from_url = get_param_from_url("user_id")
-if user_id_from_url:
-    st.session_state.user_id = user_id_from_url
-
+    st.session_state.data = read_data(to_annotate_file_name)
 
+# Initialize the current index
 if 'current_index' not in st.session_state:
-    start_index = get_start_index()
-    st.session_state.current_index = start_index
+    st.session_state.current_index = -3
     st.session_state.form_displayed = -3
 
-if get_param_from_url('show_extra_fields'):
-    fields += url_conditional_fields
-
 
 def add_validated_submit(fields, message):
     st.session_state.form_displayed = st.session_state.current_index
@@ -494,14 +430,21 @@ def add_checked_submit():
     check = st.checkbox('I agree', key='consent')
     add_validated_submit([check], "Please agree to give your consent to proceed")
 
-    
+
+def add_annotation_guidelines():
+    st.write(f"username is {st.session_state.user_id}")
+    st.markdown(
+        "<details open><summary>Annotation Guidelines</summary>" + guidelines_text + "</details>"
+        , unsafe_allow_html=True)
 if st.session_state.current_index == -3:
     with st.form("data_form"):
         st.markdown(consent_text)
         add_checked_submit()
 
 elif st.session_state.current_index == -2:
-    if st.session_state.get('user_id'):
+    user_id_from_url = get_user_id_from_url()
+    if user_id_from_url:
+        st.session_state.user_id = user_id_from_url
         navigate(1)
     else:
         with st.form("data_form"):
@@ -518,13 +461,12 @@ elif st.session_state.current_index < len(st.session_state.data):
     with st.form("data_form"+str(st.session_state.current_index)):
         show_fields(fields)
 
-
 elif st.session_state.current_index == len(st.session_state.data):
     with st.form("intro_form"):
         show_fields(end_fields)
-    
+        
 else:
-    st.write(f"Thank you, again! You are all done.")
+    st.write(f"Thank you for taking part in this study!")
 
 # Navigation buttons
 if st.session_state.current_index > 0:
@@ -539,4 +481,5 @@ st.markdown(
         visibility: hidden;
     }
     </style>""", unsafe_allow_html=True
-)
+)
+