Form to provide metadata : Headquarters, unit , currency , company name,

app/menu.py

@@ -17,6 +17,7 @@ def display_pages_menu() -> None:
         st.markdown("# Pipeline steps")
         st.page_link("pages/0_Import_File.py", label="Upload PDF")
         st.page_link("pages/1_Selected_Pages.py", label="Pages selection")
-        st.page_link("pages/2_Merge_Tables.py", label="Merge tables")
-        st.page_link("pages/3_Clean_Headers.py", label="Headers setup")
-        st.page_link("pages/4_Clean_Tables.py", label="Tables customization")
+        st.page_link("pages/2_Metadata.py", label="Metadata")
+        st.page_link("pages/3_Merge_Tables.py", label="Merge tables")
+        st.page_link("pages/4_Clean_Headers.py", label="Headers setup")
+        st.page_link("pages/5_Clean_Tables.py", label="Tables customization")

app/pages/1_Selected_Pages.py

@@ -70,4 +70,4 @@ def set_validate() -> None:
             st.session_state["working_file_pdf"].name,
             [i - 1 for i in selected_pages],
         )
-        st.switch_page("pages/2_Merge_Tables.py")
+        st.switch_page("pages/2_Metadata.py")

app/pages/2_Metadata.py

@@ -0,0 +1,101 @@
+import streamlit as st
+from utils import set_algorithm_name, get_pdf_iframe
+from menu import display_pages_menu
+from country_by_country.utils.constants import (
+    JURIDICTIONS,
+    CURRENCIES,
+    SECTORS,
+    COMPANIES,
+)
+from Levenshtein import distance
+import sys
+import logging
+import pandas as pd
+import numpy as np
+import re
+
+logging.basicConfig(stream=sys.stdout, level=logging.INFO, format="%(message)s")
+
+
+st.set_page_config(layout="wide", page_title="Report metadata")
+st.title("Country by Country Tax Reporting analysis : Metadata")
+st.subheader(
+    "This page will allow you to fill in metadata about the report : company name, headquarter, currency, unit, ...",
+)
+display_pages_menu()
+
+
+if "pdf_after_page_validation" in st.session_state:
+    col1, col2 = st.columns(2)
+    with col1:
+        st.markdown(
+            get_pdf_iframe(st.session_state["pdf_after_page_validation"]),
+            unsafe_allow_html=True,
+        )
+    with col2:
+        with st.form("metadata_form"):
+            if "metadata" in st.session_state:
+                company_name = st.session_state["metadata"]["company_name"]
+                sector = st.session_state["metadata"]["sector"]
+                year = st.session_state["metadata"]["year"]
+                currency = st.session_state["metadata"]["currency"]
+                unit = st.session_state["metadata"]["unit"]
+                headquarter = st.session_state["metadata"]["headquarter"]
+            else:
+                company_name = None
+                sector = None
+                year = ""
+                currency = None
+                unit = None
+                headquarter = ""
+            companies = list(COMPANIES.keys())
+            company_name = st.selectbox(
+                "Company name",
+                companies,
+                index=companies.index(company_name) if company_name else 0,
+            )
+
+            sector = st.selectbox(
+                "Sector", SECTORS, index=SECTORS.index(sector) if sector else 0
+            )
+
+            year = st.text_input("Year", value=year)
+
+            currencies = {
+                (
+                    CURRENCIES[currency]["AlphabeticCode"],
+                    CURRENCIES[currency]["Currency"],
+                )
+                for currency in CURRENCIES
+            }
+            currencies = sorted(currencies, key=lambda x: x[0])
+            currencies = [f"{currency[0]} - {currency[1]}" for currency in currencies]
+            currency = st.selectbox(
+                "Currency",
+                currencies,
+                index=currencies.index(currency) if currency else 0,
+            )
+
+            units = ["thousands", "millions", "10 millions", "100 millions", "billions"]
+            unit = st.selectbox("Unit", units, index=units.index(unit) if unit else 0)
+
+            headquarters = list(JURIDICTIONS.keys())
+            headquarter = st.selectbox(
+                "Headquarter location",
+                headquarters,
+                index=headquarters.index(headquarter) if headquarter else 0,
+            )
+
+            submitted = st.form_submit_button(
+                label="Submit",
+            )
+            if submitted:
+                st.session_state["metadata"] = {
+                    "company_name": company_name,
+                    "sector": sector,
+                    "year": year,
+                    "currency": currency,
+                    "unit": unit,
+                    "headquarter": headquarter,
+                }
+                st.switch_page("pages/3_Merge_Tables.py")

app/pages/2_Merge_Tables.py → app/pages/3_Merge_Tables.py

@@ -83,7 +83,7 @@ def remove_table(key: str) -> None:
 
         if "first_time_merge" not in st.session_state:
             st.session_state["first_time_merge"] = False
-            st.switch_page("pages/3_Clean_Headers.py")
+            st.switch_page("pages/4_Clean_Headers.py")
 
     col1, col2, col3 = st.columns([3, 1, 3])
     with col1:
@@ -127,7 +127,7 @@ def remove_table(key: str) -> None:
             args=(table_extractor,),
         )
         if validated:
-            st.switch_page("pages/3_Clean_Headers.py")
+            st.switch_page("pages/4_Clean_Headers.py")
 
     with col3:
         if merged is True:

app/pages/3_Clean_Headers.py → app/pages/4_Clean_Headers.py

@@ -109,7 +109,7 @@ def set_headers(algorithm_name: str) -> None:
             )
 
             if submitted:
-                st.switch_page("pages/4_Clean_Tables.py")
+                st.switch_page("pages/5_Clean_Tables.py")
 
     st.markdown("# Current extraction")
     st.markdown("The extracted table is displaye below")

app/utils.py

@@ -20,4 +20,22 @@ def set_algorithm_name(my_key: str) -> None:
 
 @st.cache_data
 def to_csv_file(df: pd.DataFrame) -> bytes:
+    # Populate the columns with the metadata, if available
+    # They may not be available if the user skipped the metadata page
+    # by not clicking on Submit
+    if "metadata" in st.session_state:
+        df = df.assign(company=st.session_state["metadata"]["company_name"])
+        df = df.assign(sector=st.session_state["metadata"]["sector"])
+        df = df.assign(year=st.session_state["metadata"]["year"])
+        df = df.assign(currency=st.session_state["metadata"]["currency"])
+        df = df.assign(unit=st.session_state["metadata"]["unit"])
+        df = df.assign(headquarter=st.session_state["metadata"]["headquarter"])
+    else:
+        df = df.assign(company="")
+        df = df.assign(sector="")
+        df = df.assign(year="")
+        df = df.assign(currency="")
+        df = df.assign(unit="")
+        df = df.assign(headquarter="")
+
     return df.to_csv(index=False).encode("utf-8")