init commit

app.py

@@ -0,0 +1,277 @@
+import streamlit as st
+import os
+import pathlib
+import pandas as pd
+from collections import defaultdict
+import json
+import ast
+import copy
+import re
+import tqdm
+
+
+import pandas as pd
+from collections import Counter
+import string
+import os
+import streamlit as st
+import difflib
+from html import escape
+
+
+def generate_diff_html_word_level(text1, text2):
+    """
+    Generates word-level difference between text1 and text2 as HTML, correctly handling spaces.
+    """
+    # Splitting texts into words
+    words1 = text1.split()
+    words2 = text2.split()
+    
+    diff = []
+    matcher = difflib.SequenceMatcher(None, words1, words2)
+    
+    for opcode in matcher.get_opcodes():
+        tag, i1, i2, j1, j2 = opcode
+        if tag == 'replace':
+            diff.append('<del style="background-color: #fbb6ce;">' + escape(' '.join(words1[i1:i2])) + '</del>')
+            diff.append('<ins style="background-color: #b7e4c7;">' + escape(' '.join(words2[j1:j2])) + '</ins>')
+        elif tag == 'delete':
+            diff.append('<del style="background-color: #fbb6ce;">' + escape(' '.join(words1[i1:i2])) + '</del>')
+        elif tag == 'insert':
+            diff.append('<ins style="background-color: #b7e4c7;">' + escape(' '.join(words2[j1:j2])) + '</ins>')
+        elif tag == 'equal':
+            diff.append(escape(' '.join(words1[i1:i2])))
+
+    # Construct final HTML string
+    final_html = ' '.join(diff).replace('</del> <ins', '</del>&nbsp;<ins')
+    return f'<pre style="white-space: pre-wrap;">{final_html}</pre>'
+
+
+
+os.environ["PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION"] = "python"
+st.set_page_config(layout="wide")
+
+query_input = None
+
+@st.cache_data
+def convert_df(df):
+    # IMPORTANT: Cache the conversion to prevent computation on every rerun
+    return df.to_csv(path_or_buf=None, index=False, quotechar='"').encode('utf-8')
+
+
+def get_current_data():
+    cur_query_data = []
+    for id_num, checkbox in current_checkboxes:
+        if checkbox:
+            qid, pid = id_num.split("-----")
+            cur_query_data.append({
+                "qid": qid,
+                "pid": pid,
+                "is_relevant": 0
+            })
+    return convert_df(pd.DataFrame(cur_query_data))
+
+@st.cache_data
+def escape_markdown(text):
+    # List of characters to escape
+    # Adding backslash to the list of special characters to escape itself as well
+    text = text.replace("``", "\"")
+    text = text.replace("$", "\$")
+    special_chars = ['\\', '`', '*', '_', '{', '}', '[', ']', '(', ')', '#', '+', '-', '.', '!', '|', "$"]
+    
+    # Escaping each special character
+    escaped_text = "".join(f"\\{char}" if char in special_chars else char for char in text)
+    
+    return escaped_text
+
+
+
+if 'cur_instance_num' not in st.session_state:
+    st.session_state.cur_instance_num = -1
+
+
+with st.sidebar:
+    st.title("Options")
+
+
+@st.cache_data
+def load_data():
+    data = []
+    with open("diffs.jsonl", "r") as f:
+        for line in f:
+            data.append(json.loads(line))
+    df = pd.DataFrame(data)
+    return df
+
+
+df = load_data()
+df["id_combined"] = df.apply(lambda x: str(x["qid"]), axis=1)
+# aggregate the df so that each qid is a row and the rest is a list of instances
+df = df.groupby("id_combined").agg(lambda x: x.tolist()).reset_index()
+
+# print details about the diff type
+# diff_types = df["diff_type"].value_counts()
+# st.write(diff_types)
+
+# # print how many titles differ by looping through each instances
+# diff_titles = 0
+# for index, row in df.iterrows():
+#     if row["old"] is not None and row["new"] is not None and row["old"][0]["title"] != row["new"][0]["title"]:
+#         diff_titles += 1
+# st.write(f" Number of titles that differ: {diff_titles}")
+
+
+original_map = {item["id_combined"]: item for item in df.to_dict(orient="records")}
+  
+
+col1, col2 = st.columns([1, 3], gap="large")
+
+with st.sidebar:
+    st.success("All files uploaded")
+
+with col1:
+    # breakpoint()
+    ids = df["id_combined"].tolist()
+    set_of_cols = set(ids)
+    container_for_nav = st.container()
+    name_of_columns = sorted([item for item in set_of_cols])
+    instances_to_use = name_of_columns
+    st.title("Instances")
+    
+    def sync_from_drop():
+        if st.session_state.selectbox_instance == "Overview":
+            st.session_state.number_of_col = -1
+            st.session_state.cur_instance_num = -1
+        else:
+            index_of_obj = name_of_columns.index(st.session_state.selectbox_instance)
+            # print("Index of obj: ", index_of_obj, type(index_of_obj)) 
+            st.session_state.number_of_col = index_of_obj
+            st.session_state.cur_instance_num = index_of_obj
+
+    def sync_from_number():
+        st.session_state.cur_instance_num = st.session_state.number_of_col
+        # print("Session state number of col: ", st.session_state.number_of_col, type(st.session_state.number_of_col))
+        if st.session_state.number_of_col == -1:
+            st.session_state.selectbox_instance = "Overview"
+        else:
+            st.session_state.selectbox_instance = name_of_columns[st.session_state.number_of_col]
+
+
+    number_of_col = container_for_nav.number_input(min_value=-1, step=1, max_value=len(instances_to_use) - 1, on_change=sync_from_number, label=f"Select instance by index (up to **{len(instances_to_use) - 1}**)", key="number_of_col")
+    selectbox_instance = container_for_nav.selectbox("Select instance by ID", ["Overview"] + name_of_columns, on_change=sync_from_drop, key="selectbox_instance")
+    st.divider()  
+
+
+with col2:
+    # get instance number
+    inst_index = number_of_col
+
+    if inst_index >= 0:
+        inst_num = instances_to_use[inst_index]
+        
+        st.markdown("<h1 style='text-align: center; color: black;text-decoration: underline;'>Editor</h1>", unsafe_allow_html=True)
+
+        current_checkboxes = []
+
+        diff_types = original_map[inst_num]['diff_type']
+        num_new = Counter(diff_types)["new"]
+        instances = original_map[inst_num]
+
+        container = st.container()
+
+        container.subheader(f"Combined ID: {inst_num} with {len(instances['pid'])} docs ({num_new} are new)")
+        # container.markdown(f"Diff Type: **{**")
+
+        container.divider()
+
+        container.subheader(f"Query Info")
+
+        container.markdown(f"Query ID: {instances['query_info'][0]['_id']}")
+        container.markdown(f"**Query**:{instances['query_info'][0]['text']}")
+
+        # container.markdown(f"Query Instruction OG: {original_map[inst_num]['query_info']['instruction_og']}")
+        # container.markdown(f"Query Instruction New: {original_map[inst_num]['query_info']['instruction_changed']}")
+
+
+        # container.subheader("Instruction ")
+        processed_diff = generate_diff_html_word_level(instances['query_info'][0]['instruction_og'], instances['query_info'][0]['instruction_changed'])
+        with container.container():
+            st.markdown("**Instruction**: " + processed_diff, unsafe_allow_html=True)
+
+        container.divider()
+
+        for i in range(len(instances["pid"])):
+
+            container.markdown(f"Doc {instances['pid'][i]}: **{diff_types[i] if diff_types[i] == 'new' else 'changed'}**")
+
+            # previous qrel score was either relevant (>0) or non-relevant
+            qrel_score = instances["qrel_score"][i]
+            # if relevant highlight blue, if non-relevant make orange with html tags
+            if qrel_score > 0:
+                container.markdown(f"<h3>Previous: <span style='color: blue;'>Relevant</span></h3>", unsafe_allow_html=True)
+            else:
+                container.markdown(f"<h3>Previous: <span style='color: orange;'>Not Relevant</span></h3>", unsafe_allow_html=True)
+
+            combined_id = str(instances["qid"][i]) + "-----" + instances["pid"][i]
+            container.subheader(f"Title")
+
+            if instances['old'][i] is not None and instances['new'][i] is not None and instances['old'][i][0]['title'] == instances['new'][i][0]['title']:
+                container.markdown(f"{instances['old'][i][0]['title']}")
+            elif instances['old'][i] is None and instances['new'][i] is None:
+                container.markdown("")
+            else:
+                if instances['old'][i] is not None:
+                    container.markdown(f"{instances['old'][i][0]['title']}")
+                elif instances['new'][i] is not None:
+                    container.markdown(f"{instances['new'][i][0]['title']}")
+                else:
+                    assert False
+
+
+                if instances['old'][i] is not None and instances['new'][i] is not None:
+                    container.subheader("Title Diff")
+                    processed_diff = generate_diff_html_word_level(instances['old'][i][0]['title'], instances['new'][i][0]['title'])
+                    with container.container():
+                        st.markdown(processed_diff, unsafe_allow_html=True)
+                        # if both are none, say that:
+
+            on = container.toggle('Show original text', key="toggle" + combined_id)
+            if on:
+                container.subheader(f"Original Text")
+
+                if instances['old'][i] is not None:
+                    original_input = container.markdown(instances['old'][i][0]['text'])
+                else:
+                    original_input = None
+                container.markdown("")
+
+            container.subheader(f"New Text")
+            # generated_input = container.markdown(instances['new'][i][0]['text'])
+
+            # # Diff
+            # if original_input is not None and generated_input is not None:
+            #     container.subheader("Diff")
+            #     processed_diff = generate_diff_html_word_level(instances['old'][i][0]['text'], instances['new'][i][0]['text'])
+            #     with container.container():
+            #         st.markdown(processed_diff, unsafe_allow_html=True)
+
+            # container.subheader("Full Doc")
+            container.markdown(instances['full_doc'][i])
+
+            current_checkboxes.append((combined_id, container.checkbox(f'{combined_id} is Non-Relevant', key=combined_id)))
+
+        
+            container.divider()
+
+        # download the editable text and venue name
+        if st.checkbox("Download data as CSV"):
+            st.download_button(
+                label="Download data as CSV",
+                data=get_current_data(),
+                file_name=f'annotation_query_{inst_num}_double_check.csv',
+                mime='text/csv',
+            )
+
+    # none checked
+    elif inst_index < 0:
+        st.title("Overview")

packages.txt

@@ -0,0 +1 @@
+default-jre

requirements.txt

@@ -0,0 +1,7 @@
+pandas==2.0.3
+streamlit==1.33.0
+plotly==5.15.0
+protobuf==3.20.0
+beautifulsoup4==4.12.2
+nltk==3.7
+tqdm==4.66.2