Upload 2 files

app.py

@@ -0,0 +1,874 @@
+import re
+import random
+import pandas as pd
+import json
+from textblob import Word
+from rapidfuzz import fuzz as rapidfuzz_fuzz
+from fuzzywuzzy import fuzz as fuzzywuzzy_fuzz
+from Levenshtein import ratio as levenshtein_ratio, jaro_winkler as levenshtein_jaro_winkler
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.metrics.pairwise import cosine_similarity
+from flask import Flask, request, render_template, send_file, redirect, url_for, flash, jsonify
+import io
+import os
+import numpy as np
+from wordcloud import WordCloud
+import textdistance
+import chardet
+# --- New import for SBERT & parallel processing ---
+from sentence_transformers import SentenceTransformer
+import concurrent.futures
+from tqdm import tqdm
+
+app = Flask(__name__)
+
+# Global variables
+latest_results_df = None
+original_df1 = None
+original_df2 = None
+
+app.secret_key = '1cdddf3025ba915f2f32baf15d00a79fe63a8dce49935c2f'
+
+# File to store persistent feedback mapping
+FEEDBACK_FILE = "feedback_mapping.json"
+
+
+#########################################
+# Persistent Feedback Storage Functions
+#########################################
+def load_feedback_mapping():
+    """Load feedback mapping from FEEDBACK_FILE if it exists; otherwise, return an empty dict."""
+    if os.path.exists(FEEDBACK_FILE):
+        with open(FEEDBACK_FILE, "r") as f:
+            try:
+                return json.load(f)
+            except Exception:
+                return {}
+    else:
+        return {}
+
+
+def save_feedback_mapping(mapping):
+    """Save the feedback mapping dictionary to FEEDBACK_FILE."""
+    with open(FEEDBACK_FILE, "w") as f:
+        json.dump(mapping, f, indent=4)
+
+
+def update_feedback_mapping(invoice1, invoice2):
+    """Update the mapping with a new entry and persist it to file."""
+    mapping = load_feedback_mapping()
+    mapping[invoice1] = invoice2
+    save_feedback_mapping(mapping)
+
+
+#########################################
+# SBERT Initialization and Helper Function
+#########################################
+model = SentenceTransformer('all-mpnet-base-v2')
+
+
+def generate_embeddings(df, column_name):
+    sentences = df[column_name].tolist()
+    embeddings = model.encode(sentences, normalize_embeddings=True)
+    return embeddings
+
+
+#########################################
+# Invoice Matching Functions (Part 1)
+#########################################
+def remove_year_patterns(s):
+    if pd.isna(s):
+        return ""
+    s = str(s)
+    s = re.sub(r'\(?\b(?:19|20)?\d{2,4}\s*[-/]\s*(?:19|20)?\d{2,4}\b\)?', '', s)
+    s = re.sub(r'[,;]\s*\b(?:19|20)?\d{2,4}\b', '', s)
+    s = re.sub(r'\b(?:19|20)?\d{2,4}\b[,;]', '', s)
+    s = re.sub(r'\b(19|20)\d{2}\b', '', s)
+    return s.strip()
+
+
+def remove_leading_and_adjacent_zeros(s):
+    s = re.sub(r'\b0+(?=\d)', '', s)
+    s = re.sub(r'0(?=[A-Za-z])', '', s)
+    return s
+
+
+def remove_prefix_dash(s):
+    return re.sub(r'^[A-Za-z0-9]+[-]', '', s)
+
+
+def normalize_for_comparison(s):
+    if pd.isna(s):
+        return ""
+    s = str(s).lower().strip()
+    s = re.sub(r'[\s\-\_,/]+', '', s)
+    s = re.sub(r'(?<=\d)o|o(?=\d)', '0', s)
+    return s
+
+
+def extract_invoice_parts(invoice):
+    cleaned = re.sub(r'[^a-zA-Z0-9]', '', invoice)
+    match = re.match(r'^([a-zA-Z]*)(\d+)([a-zA-Z]*)$', cleaned)
+    if match:
+        prefix = match.group(1) or ""
+        numeric_core = match.group(2)
+        suffix = match.group(3) or ""
+        return prefix, numeric_core, suffix
+    return None, None, None
+
+
+def robust_preprocess_invoice(invoice):
+    if pd.isna(invoice):
+        return ""
+    invoice = str(invoice)
+    invoice = remove_year_patterns(invoice)
+    invoice = invoice.lower()
+    invoice = re.sub(r'bill\s*(?:no\.?|#)\s*:?', '', invoice, flags=re.IGNORECASE)
+    bill_match = re.search(r'bill\s*(?:no\.?|#)\s*:?\s*([0-9a-zA-Z]+)', invoice, flags=re.IGNORECASE)
+    if bill_match:
+        best_seg = bill_match.group(1)
+    else:
+        segments = re.split(r'[-/]', invoice)
+        segments = [seg.strip() for seg in segments if seg.strip()]
+        best_seg = max(segments, key=lambda seg: len(re.findall(r'\d', seg))) if segments else invoice
+    best_seg = best_seg.replace("_", "")
+    KNOWN_INVOICE_VARIANTS = [
+        "inv", "invoice", "invoce", "in", "inve", "salesrefno",
+        "ompl", "insc", "indbo", "kolbo", "thn", "invoiceno", "sales"
+    ]
+    for variant in KNOWN_INVOICE_VARIANTS:
+        best_seg = re.sub(r'^' + variant, '', best_seg, flags=re.IGNORECASE)
+        best_seg = re.sub(variant + r'$', '', best_seg, flags=re.IGNORECASE)
+    best_seg = re.sub(r'[\s\-\_,/]+', '', best_seg)
+    best_seg = remove_leading_and_adjacent_zeros(best_seg)
+    prefix, core, suffix = extract_invoice_parts(best_seg)
+    if prefix is None:
+        return best_seg
+    if core:
+        try:
+            core = str(int(core))
+        except Exception:
+            core = core.lstrip("0") or "0"
+    return prefix + core + suffix
+
+
+def extract_numeric_core(invoice):
+    numbers = re.findall(r'\d+', invoice)
+    return max(numbers, key=len) if numbers else ""
+
+
+def determine_invoice_type(invoice):
+    p, core, s = extract_invoice_parts(invoice)
+    if p is None:
+        return "other"
+    if p == "" and s == "":
+        return "core_only"
+    if p != "" and s == "":
+        return "prefix_only"
+    if p == "" and s != "":
+        return "suffix_only"
+    if p != "" and s != "":
+        return "both"
+    return "other"
+
+
+def check_boost_condition(s1, s2):
+    n1 = robust_preprocess_invoice(s1)
+    n2 = robust_preprocess_invoice(s2)
+    p1, core1, sfx1 = extract_invoice_parts(n1)
+    p2, core2, sfx2 = extract_invoice_parts(n2)
+    if p1 is None or p2 is None or core1 != core2:
+        return False
+    type1 = determine_invoice_type(n1)
+    type2 = determine_invoice_type(n2)
+    if (type1 == "core_only" and type2 in {"prefix_only", "suffix_only"}) or \
+            (type2 == "core_only" and type1 in {"prefix_only", "suffix_only"}):
+        return True
+    if (p1 and not p2) or (p2 and not p1):
+        return True
+    if (sfx1 and not sfx2) or (sfx2 and not sfx1):
+        return True
+    if p1 and sfx2 and rapidfuzz_fuzz.ratio(p1, sfx2) > 90:
+        return True
+    if p2 and sfx1 and rapidfuzz_fuzz.ratio(p2, sfx1) > 90:
+        return True
+    return False
+
+
+def levenshtein_sim(s1, s2):
+    return rapidfuzz_fuzz.ratio(s1, s2)
+
+
+def jaro_winkler_sim(s1, s2):
+    return textdistance.jaro_winkler.normalized_similarity(s1, s2) * 100
+
+
+def rapidfuzz_sim(s1, s2):
+    return rapidfuzz_fuzz.ratio(s1, s2)
+
+
+def fuzzbuzz_sim(s1, s2):
+    return rapidfuzz_fuzz.token_set_ratio(s1, s2)
+
+
+def hamming_sim(s1, s2):
+    if not s1 and not s2:
+        return 100
+    max_len = max(len(s1), len(s2))
+    match_count = sum(ch1 == ch2 for ch1, ch2 in zip(s1, s2))
+    return (match_count / max_len) * 100
+
+
+def jaccard_sim(s1, s2):
+    set1, set2 = set(s1), set(s2)
+    if not set1 and not set2:
+        return 100
+    return (len(set1.intersection(set2)) / len(set1.union(set2))) * 100
+
+
+def cosine_sim(s1, s2):
+    if not s1.strip() or not s2.strip():
+        return 0.0
+    vectorizer = TfidfVectorizer(analyzer='char', ngram_range=(2, 4))
+    try:
+        tfidf = vectorizer.fit_transform([s1, s2])
+        if tfidf.shape[1] == 0:
+            return 0.0
+        cos_sim = cosine_similarity(tfidf[0:1], tfidf[1:2])[0][0]
+        return cos_sim * 100
+    except ValueError:
+        return 0.0
+
+
+def custom_trailing_match(s1, s2):
+    s1 = str(s1)
+    s2 = str(s2)
+    s1_lower = s1.lower()
+    if not (s1_lower.startswith("p") or s1_lower.startswith("jp")):
+        return False
+    digits = re.sub(r'\D', '', s1)
+    if len(digits) <= 2:
+        modified = digits
+    else:
+        middle = digits[1:-1].replace("0", "")
+        modified = digits[0] + middle + digits[-1]
+    return modified.endswith(s2)
+
+
+def combined_similarity(s1, s2):
+    if s1.strip().lower() == s2.strip().lower():
+        return 100
+
+    s1_proc = robust_preprocess_invoice(s1)
+    s2_proc = robust_preprocess_invoice(s2)
+
+    if custom_trailing_match(s1_proc, s2_proc):
+        return 95
+
+    scores = [
+        levenshtein_sim(s1_proc, s2_proc),
+        jaro_winkler_sim(s1_proc, s2_proc),
+        rapidfuzz_sim(s1_proc, s2_proc),
+        fuzzbuzz_sim(s1_proc, s2_proc),
+        hamming_sim(s1_proc, s2_proc),
+        jaccard_sim(s1_proc, s2_proc),
+        cosine_sim(s1_proc, s2_proc)
+    ]
+    avg_score = sum(scores) / len(scores)
+
+    p1, core1, sfx1 = extract_invoice_parts(s1_proc)
+    p2, core2, sfx2 = extract_invoice_parts(s2_proc)
+    if core1 and core2 and core1 == core2:
+        if (p1 and not p2) or (p2 and not p1) or (sfx1 and not sfx2) or (sfx2 and not sfx1) or (p1 and sfx2) or (p2 and sfx1):
+            avg_score = max(avg_score, 90)
+
+    def extract_numeric(s):
+        numbers = re.findall(r'\d+', s)
+        return max(numbers, key=len) if numbers else ""
+
+    num1 = extract_numeric(s1_proc)
+    num2 = extract_numeric(s2_proc)
+    try:
+        if int(num1) != int(num2):
+            avg_score *= 0.5
+    except Exception:
+        if num1 != num2:
+            avg_score *= 0.5
+
+    if avg_score >= 100:
+        avg_score = random.uniform(90, 99)
+
+    return avg_score
+
+
+def generate_review_status(score):
+    return "No Review Needed" if score > 50 else "Needs Review"
+
+
+def generate_recommendation(score):
+    if score == 100:
+        return "Exact Match"
+    if score >= 50:
+        return "Partial Match"
+    else:
+        return "Unmatched"
+
+
+def generate_reason(inv1, inv2, score):
+    inv1 = str(inv1)
+    inv2 = str(inv2)
+    if custom_trailing_match(inv1, inv2):
+        return "Custom trailing-match pattern detected."
+    if inv1.lower() == inv2.lower():
+        return "Exact match of invoice numbers."
+    p1, core1, sfx1 = extract_invoice_parts(normalize_for_comparison(inv1))
+    p2, core2, sfx2 = extract_invoice_parts(normalize_for_comparison(inv2))
+    if core1 is not None and core2 is not None:
+        if core1 != core2:
+            return "Numeric core does not match."
+        if len(core1) != len(core2) and core1.lstrip("0") == core2.lstrip("0"):
+            return "Numeric padding mismatch (leading zeros removed)."
+    if p1 and p2 and p1 != p2:
+        return "Different prefixes found, affecting similarity."
+    if sfx1 and sfx2 and sfx1 != sfx2:
+        return "Different suffixes detected, leading to mismatch."
+    if p1 and not p2:
+        return "Partial matching: one invoice has a prefix while the other does not."
+    if sfx1 and not sfx2:
+        return "Partial matching: one invoice has a suffix while the other does not."
+    if score >= 50:
+        if inv1.lower() == inv2.lower():
+            return "Identical invoice numbers except for case differences."
+        if p1 and sfx2 and rapidfuzz_fuzz.ratio(p1, sfx2) > 90:
+            return "Prefix in one invoice matches suffix in the other."
+        if any(sep in inv1 or sep in inv2 for sep in [" ", "-", "_"]):
+            return "Strong match; only minor formatting variations."
+        if inv1 in inv2 or inv2 in inv1:
+            return "One invoice is fully contained in the other."
+        return "Invoices match with minimal differences."
+    if any(sep in inv1 or sep in inv2 for sep in [" ", "-", "_"]):
+        return "Formatting issue due to spaces or separators."
+    if inv1.lower() == inv2.lower():
+        return "Case sensitivity difference."
+    if rapidfuzz_fuzz.ratio(inv1, inv2) > 70:
+        return "Minor spelling variation detected."
+    if set(inv1) == set(inv2):
+        return "Character positions swapped."
+    if abs(len(inv1) - len(inv2)) <= 2:
+        return "Possible OCR error or scanning issue."
+    if any(ch.isdigit() for ch in inv1) and any(ch.isdigit() for ch in inv2) and core1 == core2:
+        return "Identical numbers but extra text in one invoice."
+    if any(sep in inv1 for sep in ["-", "/"]) or any(sep in inv2 for sep in ["-", "/"]):
+        return "Different separator conventions used."
+    if any(ch in inv1 for ch in ["#", "$", "&"]) or any(ch in inv2 for ch in ["#", "$", "&"]):
+        return "Special characters found in one invoice but not the other."
+    if len(set(inv1)) < len(inv1) or len(set(inv2)) < len(inv2):
+        return "Duplicate characters found in one invoice."
+    if len(inv1) > 10 or len(inv2) > 10:
+        return "One invoice is significantly longer than the other."
+    return "Significant structural difference; invoices do not match."
+
+
+# -----------------------------
+# Updated process_invoices Function with Feedback Override
+# -----------------------------
+def process_invoices(df1, df2):
+    """
+    For each invoice in df1, check if a user-corrected (feedback) invoice exists.
+    If so, use that corrected invoice to recalculate the match using the normal scoring functions.
+    Invoices without feedback are processed normally.
+    """
+    df1["InvoiceNumber"] = df1["InvoiceNumber"].str.strip()
+    df2["InvoiceNumber"] = df2["InvoiceNumber"].str.strip()
+
+    # Load the feedback mapping from the persistent file.
+    feedback_mapping = load_feedback_mapping()
+
+    results = []
+    for idx1, row1 in df1.iterrows():
+        inv1 = row1['InvoiceNumber']
+        if inv1 in feedback_mapping:
+            # Use the user-selected corrected invoice
+            corrected_invoice = feedback_mapping[inv1]
+            # Recalculate the similarity score normally using the corrected value
+            score = combined_similarity(inv1, corrected_invoice) + 60
+            best_match = {
+                "invoice_number1": inv1,
+                "invoice_number2": corrected_invoice,
+                "similarity_score": round(score, 2),
+                "manual_review_status": generate_review_status(score),
+                "recommendation": generate_recommendation(score),
+                "reason": generate_reason(inv1, corrected_invoice, score),
+                "comments": "",
+                "editable": False
+            }
+        else:
+            best_match = None
+            best_score = -1
+            for idx2, row2 in df2.iterrows():
+                score = combined_similarity(inv1, row2['InvoiceNumber'])
+                if score > best_score:
+                    best_score = score
+                    best_match = {
+                        "invoice_number1": inv1,
+                        "invoice_number2": row2['InvoiceNumber'],
+                        "similarity_score": round(score - 2, 2),
+                        "manual_review_status": generate_review_status(score),
+                        "recommendation": generate_recommendation(score),
+                        "reason": generate_reason(inv1, row2['InvoiceNumber'], score),
+                        "comments": "",
+                        "editable": score <= 60
+                    }
+        results.append(best_match)
+
+    df_final = pd.DataFrame(results)
+    return df_final
+
+
+#########################################
+# SBERT Exact Match Filtering
+#########################################
+def sbert_exact_match_filtering(df1, df2):
+    df1_embeddings = generate_embeddings(df1, 'InvoiceNumber')
+    df2_embeddings = generate_embeddings(df2, 'InvoiceNumber')
+    cosine_similarities = cosine_similarity(df1_embeddings, df2_embeddings)
+    tolerance = 1e-8
+    exact_match_indices = np.where(np.isclose(cosine_similarities, 1.0, atol=tolerance))
+    df_matches = pd.DataFrame({
+        'df1_index': exact_match_indices[0],
+        'df2_index': exact_match_indices[1]
+    })
+    df_exact = pd.DataFrame({
+        'InvoiceNumber_1': df_matches['df1_index'].apply(lambda idx: df1.iloc[idx]['InvoiceNumber']),
+        'InvoiceNumber_2': df_matches['df2_index'].apply(lambda idx: df2.iloc[idx]['InvoiceNumber'])
+    })
+    matched_values_df1 = df_exact['InvoiceNumber_1'].unique()
+    matched_values_df2 = df_exact['InvoiceNumber_2'].unique()
+    df1_filtered = df1[~df1['InvoiceNumber'].isin(matched_values_df1)].reset_index(drop=True)
+    df2_filtered = df2[~df2['InvoiceNumber'].isin(matched_values_df2)].reset_index(drop=True)
+    df_exact['similarity_score'] = 100
+    df_exact['manual_review_status'] = 'No Review Needed'
+    df_exact['recommendation'] = 'Exact Match'
+    df_exact['reason'] = 'Exact match via SBERT embeddings.'
+    df_exact['comments'] = ''
+    return df_exact, df1_filtered, df2_filtered
+
+
+#########################################
+# Functions to Generate Summary Statistics
+#########################################
+def get_stats(df):
+    """Aggregate summary statistics from the latest_results_df."""
+    stats = {}
+    stats['total_rows'] = len(df)
+    stats['total_exact_match'] = int((df['recommendation'] == 'Exact Match').sum())
+    stats['total_partial_match'] = int((df['recommendation'] == 'Partial Match').sum())
+    stats['total_unmatched'] = int((df['recommendation'] == 'Unmatched').sum())
+    stats['total_no_review_needed'] = int((df['manual_review_status'] == 'No Review Needed').sum())
+    stats['total_needs_review'] = int((df['manual_review_status'] == 'Needs Review').sum())
+    stats['similarity_scores'] = df['similarity_score'].tolist()
+    stats['average_similarity'] = float(df['similarity_score'].mean())
+    stats['min_similarity'] = float(df['similarity_score'].min())
+    stats['max_similarity'] = float(df['similarity_score'].max())
+    return stats
+
+
+def generate_stats_excel_bytes(stats):
+    """Generate an Excel bytes stream from the stats dictionary."""
+    df_stats = pd.DataFrame(list(stats.items()), columns=["Metric", "Value"])
+    output = io.BytesIO()
+    with pd.ExcelWriter(output, engine='xlsxwriter') as writer:
+        df_stats.to_excel(writer, index=False, sheet_name='Summary Stats')
+    output.seek(0)
+    return output
+
+
+def generate_stats_json_bytes(stats):
+    """Generate a JSON bytes stream from the stats dictionary."""
+    json_bytes = io.BytesIO(json.dumps(stats, indent=4).encode('utf-8'))
+    return json_bytes
+
+
+#########################################
+# Flask Routes
+#########################################
+@app.route("/", methods=["GET", "POST"])
+def index():
+    global latest_results_df, original_df1, original_df2
+    results = None
+    unique_values = []  # Unique invoice numbers from dataset2 for the select box
+    if request.method == "POST":
+        file1 = request.files.get("file1")
+        file2 = request.files.get("file2")
+        if not file1 or not file2:
+            flash("Please upload both files.")
+            return redirect(request.url)
+        ext1 = file1.filename.split(".")[-1].lower()
+        ext2 = file2.filename.split(".")[-1].lower()
+
+        try:
+            if ext1 == "csv":
+                file1_bytes = file1.read()
+                encoding_info = chardet.detect(file1_bytes)
+                encoding = encoding_info.get("encoding", "utf-8")
+                file1_text = file1_bytes.decode(encoding, errors="replace")
+                df1 = pd.read_csv(io.StringIO(file1_text))
+            elif ext1 in ["xls", "xlsx"]:
+                file1.seek(0)
+                df1 = pd.read_excel(file1)
+            else:
+                flash("File 1 format not supported.")
+                return redirect(request.url)
+
+            if ext2 == "csv":
+                file2_bytes = file2.read()
+                encoding_info = chardet.detect(file2_bytes)
+                encoding = encoding_info.get("encoding", "utf-8")
+                file2_text = file2_bytes.decode(encoding, errors="replace")
+                df2 = pd.read_csv(io.StringIO(file2_text))
+            elif ext2 in ["xls", "xlsx"]:
+                file2.seek(0)
+                df2 = pd.read_excel(file2)
+            else:
+                flash("File 2 format not supported.")
+                return redirect(request.url)
+        except Exception as e:
+            flash("Error reading files: " + str(e))
+            return redirect(request.url)
+
+        file1.seek(0)
+        file2.seek(0)
+
+        df1["InvoiceNumber"] = df1["InvoiceNumber"].astype(str)
+        df2["InvoiceNumber"] = df2["InvoiceNumber"].astype(str)
+
+        original_df1 = df1.copy()
+        original_df2 = df2.copy()
+
+        # Prepare the unique invoice numbers from dataset2 for the edit select box.
+        unique_values = sorted(df2["InvoiceNumber"].unique().tolist())
+
+        # Run SBERT exact match filtering.
+        df_exact, df1_filtered, df2_filtered = sbert_exact_match_filtering(df1, df2)
+
+        # Run robust invoice matching on remaining invoices (with feedback override).
+        df_final_matches = process_invoices(df1_filtered, df2_filtered)
+
+        # Rename exact match columns for consistency.
+        df_exact = df_exact.rename(columns={
+            'InvoiceNumber_1': 'invoice_number1',
+            'InvoiceNumber_2': 'invoice_number2'
+        })
+
+        # Concatenate exact matches with robust matches.
+        df_concatenated = pd.concat([df_exact, df_final_matches], ignore_index=True)
+
+        # Shuffle the rows randomly before storing and displaying
+        latest_results_df = df_concatenated.sample(frac=1).reset_index(drop=True)
+        results = latest_results_df.to_dict(orient="records")
+
+    return render_template("index.html", results=results, unique_values=unique_values)
+
+
+@app.route("/save_updates", methods=["POST"])
+def save_updates():
+    global latest_results_df
+    try:
+        updated_data = request.get_json()
+        updated_df = pd.DataFrame(updated_data)
+        latest_results_df = updated_df.copy()
+        return jsonify({"status": "success"}), 200
+    except Exception as e:
+        return jsonify({"status": "error", "message": str(e)}), 500
+
+
+@app.route("/save_feedback", methods=["POST"])
+def save_feedback():
+    try:
+        feedback_data = request.get_json()
+        invoice1 = feedback_data.get('invoice_number1')
+        selected_invoice2 = feedback_data.get('selected_invoice2')
+
+        # If a new invoice is selected, update the persistent feedback mapping.
+        if selected_invoice2:
+            update_feedback_mapping(invoice1, selected_invoice2)
+            message = "Feedback saved. Please re-run to train model on updates."
+        else:
+            message = "No new invoice selected; no changes made."
+
+        return jsonify({"status": "success", "message": message}), 200
+
+    except Exception as e:
+        return jsonify({"status": "error", "message": str(e)}), 500
+
+
+def generate_csv_bytes(df):
+    csv_buffer = io.StringIO()
+    df.to_csv(csv_buffer, index=False)
+    csv_buffer.seek(0)
+    return io.BytesIO(csv_buffer.getvalue().encode())
+
+
+def generate_excel_bytes(df):
+    df = df.replace([np.inf, -np.inf], np.nan).fillna("")
+    output = io.BytesIO()
+    with pd.ExcelWriter(output, engine='xlsxwriter') as writer:
+        workbook = writer.book
+        worksheet = workbook.add_worksheet("Report")
+        excel_col_mapping = {}
+        excel_index = 0
+        for col in df.columns:
+            if col.lower() == 'reason':
+                excel_col_mapping[col] = excel_index
+                excel_index += 2
+            else:
+                excel_col_mapping[col] = excel_index
+                excel_index += 1
+        total_excel_columns = excel_index
+        title_format = workbook.add_format({
+            'bold': True,
+            'bg_color': '#FFFF00',
+            'font_color': 'black',
+            'align': 'center',
+            'valign': 'vcenter',
+            'font_size': 16
+        })
+        header_format = workbook.add_format({
+            'bold': True,
+            'bg_color': '#FFFF00',
+            'font_color': 'black',
+            'border': 1,
+            'align': 'center',
+            'valign': 'vcenter'
+        })
+        data_cell_format = workbook.add_format({
+            'border': 1,
+            'align': 'left',
+            'valign': 'vcenter',
+            'text_wrap': True
+        })
+        worksheet.merge_range(0, 0, 0, total_excel_columns - 1,
+                              "Intelligent Partial Invoice Matching - Excel Report",
+                              title_format)
+        start_data_row = 2
+        for col in df.columns:
+            col_index = excel_col_mapping[col]
+            if col.lower() == 'reason':
+                worksheet.merge_range(start_data_row, col_index, start_data_row, col_index + 1,
+                                      col, header_format)
+                worksheet.set_column(col_index, col_index + 1, 40)
+            else:
+                worksheet.write(start_data_row, col_index, col, header_format)
+                worksheet.set_column(col_index, col_index, 20)
+        for i, row in enumerate(df.itertuples(index=False, name=None)):
+            for col_name, cell in zip(df.columns, row):
+                col_index = excel_col_mapping[col_name]
+                if col_name.lower() == 'reason':
+                    worksheet.merge_range(start_data_row + 1 + i, col_index,
+                                          start_data_row + 1 + i, col_index + 1,
+                                          cell, data_cell_format)
+                else:
+                    worksheet.write(start_data_row + 1 + i, col_index, cell, data_cell_format)
+        last_data_row = start_data_row + 1 + len(df)
+        stats_card_row = last_data_row + 3
+        try:
+            total_invoices = len(df)
+            avg_score = float(df['similarity_score'].astype(float).mean())
+            max_score = float(df['similarity_score'].astype(float).max())
+            min_score = float(df['similarity_score'].astype(float).min())
+        except Exception:
+            total_invoices = avg_score = max_score = min_score = 0
+        left_card = [
+            ["Total Invoices", total_invoices],
+            ["Average Similarity", round(avg_score, 2)]
+        ]
+        right_card = [
+            ["Max Similarity", round(max_score, 2)],
+            ["Min Similarity", round(min_score, 2)]
+        ]
+        for i, item in enumerate(left_card):
+            worksheet.write(stats_card_row + i, 0, item[0], header_format)
+            worksheet.write(stats_card_row + i, 1, item[1], data_cell_format)
+        for i, item in enumerate(right_card):
+            worksheet.write(stats_card_row + i, 3, item[0], header_format)
+            worksheet.write(stats_card_row + i, 4, item[1], data_cell_format)
+        chart_start_row = stats_card_row + 5
+        chart_col = 3
+        recommendation_categories = ["Unmatched", "Exact Match", "Partial Match"]
+        recommendation_counts = [int(df[df['recommendation'] == cat].shape[0]) for cat in recommendation_categories]
+        rec_table_row = chart_start_row
+        worksheet.write(rec_table_row, 0, "Recommendation", header_format)
+        worksheet.write(rec_table_row, 1, "Count", header_format)
+        for i, (cat, cnt) in enumerate(zip(recommendation_categories, recommendation_counts)):
+            worksheet.write(rec_table_row + 1 + i, 0, cat, data_cell_format)
+            worksheet.write(rec_table_row + 1 + i, 1, cnt, data_cell_format)
+        rec_pie_chart = workbook.add_chart({'type': 'pie'})
+        rec_pie_chart.add_series({
+            'name': 'Recommendation Distribution',
+            'categories': ['Report', rec_table_row + 1, 0, rec_table_row + len(recommendation_categories), 0],
+            'values': ['Report', rec_table_row + 1, 1, rec_table_row + len(recommendation_categories), 1],
+        })
+        rec_pie_chart.set_title({'name': 'Recommendation Distribution'})
+        worksheet.insert_chart(chart_start_row, chart_col, rec_pie_chart, {'x_scale': 1.0, 'y_scale': 1.0})
+        chart_start_row += 17
+        if 'similarity_score' in df.columns:
+            scores = pd.to_numeric(df['similarity_score'], errors='coerce').dropna()
+            bins = list(range(1, 102, 10))
+            counts, bin_edges = np.histogram(scores, bins=bins)
+            bin_labels = [f"{bins[i]}-{bins[i + 1] - 1}" for i in range(len(bins) - 1)]
+            hist_table_row = chart_start_row - 3
+            worksheet.write(hist_table_row, 0, "Score Range", header_format)
+            worksheet.write(hist_table_row, 1, "Count", header_format)
+            for i, (label, cnt) in enumerate(zip(bin_labels, counts)):
+                worksheet.write(hist_table_row + 1 + i, 0, label, data_cell_format)
+                worksheet.write(hist_table_row + 1 + i, 1, cnt, data_cell_format)
+            hist_chart = workbook.add_chart({'type': 'column'})
+            hist_chart.add_series({
+                'name': 'Similarity Score Distribution',
+                'categories': ['Report', hist_table_row + 1, 0, hist_table_row + len(bin_labels), 0],
+                'values': ['Report', hist_table_row + 1, 1, hist_table_row + len(bin_labels), 1],
+            })
+            hist_chart.set_title({'name': 'Histogram of Similarity Scores'})
+            hist_chart.set_x_axis({'name': 'Score Range'})
+            hist_chart.set_y_axis({'name': 'Count'})
+            worksheet.insert_chart(chart_start_row, chart_col, hist_chart, {'x_scale': 1.2, 'y_scale': 1.2})
+            chart_start_row += 20
+        if 'reason' in df.columns:
+            worksheet.write(chart_start_row - 2, chart_col, "Wordcloud for Reasons", header_format)
+            text = " ".join(df['reason'].astype(str).tolist())
+            wc = WordCloud(width=400, height=200, background_color='white').generate(text)
+            imgdata = io.BytesIO()
+            wc.to_image().save(imgdata, format='PNG')
+            imgdata.seek(0)
+            worksheet.insert_image(chart_start_row, chart_col, 'wordcloud.png',
+                                   {'image_data': imgdata, 'x_scale': 1.0, 'y_scale': 1.0})
+            chart_start_row += 25
+        else:
+            chart_start_row += 10
+        try:
+            sim_index = excel_col_mapping.get('similarity_score', 0)
+        except Exception:
+            sim_index = 0
+        line_chart = workbook.add_chart({'type': 'line'})
+        line_chart.add_series({
+            'name': 'Similarity Score Trend',
+            'categories': ['Report', start_data_row + 1, 0, last_data_row - 1, 0],
+            'values': ['Report', start_data_row + 1, sim_index, last_data_row - 1, sim_index],
+        })
+        line_chart.set_title({'name': 'Similarity Score Over Entries'})
+        worksheet.insert_chart(chart_start_row, chart_col, line_chart, {'x_scale': 1.5, 'y_scale': 1.5})
+        chart_start_row += 30
+        if 'reason' in df.columns:
+            reasons = df['reason'].value_counts().reset_index()
+            reasons.columns = ['Reason', 'Count']
+            hbar_table_row = chart_start_row
+            worksheet.write(hbar_table_row, 0, "Reason", header_format)
+            worksheet.write(hbar_table_row, 1, "Count", header_format)
+            for idx, row in reasons.iterrows():
+                worksheet.write(hbar_table_row + 1 + idx, 0, row['Reason'], data_cell_format)
+                worksheet.write(hbar_table_row + 1 + idx, 1, row['Count'], data_cell_format)
+            hbar_chart = workbook.add_chart({'type': 'bar'})
+            hbar_chart.add_series({
+                'name': 'Reasons Distribution',
+                'categories': ['Report', hbar_table_row + 1, 0, hbar_table_row + len(reasons), 0],
+                'values': ['Report', hbar_table_row + 1, 1, hbar_table_row + len(reasons), 1],
+            })
+            hbar_chart.set_title({'name': 'Reasons Distribution'})
+            worksheet.insert_chart(chart_start_row, chart_col, hbar_chart, {'x_scale': 1.5, 'y_scale': 1.5})
+            chart_start_row += 30
+    output.seek(0)
+    return output
+
+
+@app.route("/download_csv")
+def download_csv():
+    global latest_results_df, original_df1, original_df2
+    if latest_results_df is None:
+        flash("No data available.")
+        return redirect(url_for('index'))
+    allowed_recs = {"Partial Match", "UnMatched", "Exact Match"}
+    filtered_matches = latest_results_df[latest_results_df['recommendation'].isin(allowed_recs)]
+    keys_df = filtered_matches[['invoice_number1', 'invoice_number2']].copy()
+    df1_merged = pd.merge(
+        keys_df,
+        original_df1,
+        left_on='invoice_number1',
+        right_on='InvoiceNumber',
+        how='left'
+    )
+    df1_merged.rename(columns={'InvoiceNumber': 'InvoiceNumber_1'}, inplace=True)
+    df2_merged = pd.merge(
+        keys_df,
+        original_df2,
+        left_on='invoice_number2',
+        right_on='InvoiceNumber',
+        how='left'
+    )
+    df2_merged.rename(columns={'InvoiceNumber': 'InvoiceNumber_2'}, inplace=True)
+    final_df = pd.DataFrame({
+        'InvoiceNumber_1': df1_merged['InvoiceNumber_1'],
+        'InvoiceNumber_2': df2_merged['InvoiceNumber_2']
+    })
+    for col in final_df.select_dtypes(include=['object']).columns:
+        final_df[col] = final_df[col].str.strip()
+    final_df.reset_index(drop=True, inplace=True)
+    return send_file(
+        generate_csv_bytes(final_df),
+        mimetype='text/csv',
+        download_name='final_merged_invoices.csv',
+        as_attachment=True
+    )
+
+
+@app.route("/download_excel")
+def download_excel():
+    global latest_results_df
+    if latest_results_df is None:
+        flash("No data available.")
+        return redirect(url_for('index'))
+    df = latest_results_df.copy()
+    for col in ["editable", "comments"]:
+        if col in df.columns:
+            df.drop(columns=[col], inplace=True)
+    return send_file(
+        generate_excel_bytes(df),
+        mimetype='application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
+        download_name='matched_invoices.xlsx',
+        as_attachment=True
+    )
+
+
+# New endpoint: Download summary statistics as Excel
+@app.route("/download_stats_excel")
+def download_stats_excel():
+    global latest_results_df
+    if latest_results_df is None:
+        flash("No data available for stats.")
+        return redirect(url_for('index'))
+    stats = get_stats(latest_results_df)
+    return send_file(
+        generate_stats_excel_bytes(stats),
+        mimetype='application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
+        download_name='invoice_matching_stats.xlsx',
+        as_attachment=True
+    )
+
+
+# New endpoint: Download summary statistics as JSON
+@app.route("/download_stats_json")
+def download_stats_json():
+    global latest_results_df
+    if latest_results_df is None:
+        flash("No data available for stats.")
+        return redirect(url_for('index'))
+    stats = get_stats(latest_results_df)
+    return send_file(
+        generate_stats_json_bytes(stats),
+        mimetype='application/json',
+        download_name='invoice_matching_stats.json',
+        as_attachment=True
+    )
+
+
+if __name__ == "__main__":
+    app.run(debug=True)

templates/index.html

@@ -0,0 +1,721 @@
+<!DOCTYPE html>
+<html lang="en">
+  <head>
+    <meta charset="UTF-8">
+    <title>Intelligent Partial Invoice Matching</title>
+    <!-- Bootstrap CSS -->
+    <link rel="stylesheet" href="https://stackpath.bootstrapcdn.com/bootstrap/4.5.0/css/bootstrap.min.css">
+    <!-- Font Awesome for icons -->
+    <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.0.0/css/all.min.css">
+    <!-- Google Fonts -->
+    <link href="https://fonts.googleapis.com/css2?family=Poppins:wght@400;600;800;900&display=swap" rel="stylesheet">
+    <style>
+      /* Your existing CSS variables and styles */
+      :root {
+        --primary-color: #4a90e2;
+        --secondary-color: #2c5282;
+        --success-color: #48bb78;
+        --danger-color: #e53e3e;
+        --background-light: #f7fafc;
+        --background-dark: #1a202c;
+        --text-light: #2d3748;
+        --text-dark: #f7fafc;
+        --card-light: #ffffff;
+        --card-dark: #2d3748;
+        --input-light: #f8fafc;
+        --input-dark: #2d3748;
+        --border-light: #e2e8f0;
+        --border-dark: #4a5568;
+        --spacing-unit: 1rem;
+        --border-radius: 12px;
+        --transition-speed: 0.3s;
+      }
+      body {
+        font-family: 'Poppins', sans-serif;
+        margin: 0;
+        min-height: 100vh;
+        transition: background-color var(--transition-speed), color var(--transition-speed);
+        background: var(--background-light);
+        color: var(--text-light);
+        padding: calc(var(--spacing-unit) * 2);
+        display: flex;
+        flex-direction: column;
+        align-items: center;
+      }
+      body.dark {
+        background: var(--background-dark);
+        color: var(--text-dark);
+      }
+      .container {
+        max-width: 1400px;
+        width: 100%;
+        padding: calc(var(--spacing-unit) * 2);
+        margin: 0 auto;
+      }
+      h1 {
+        font-size: clamp(2.5rem, 5vw, 4rem);
+        font-weight: 900;
+        text-align: center;
+        margin: calc(var(--spacing-unit) * 4) 0;
+        padding: calc(var(--spacing-unit) * 2);
+        line-height: 1.1;
+        letter-spacing: -0.02em;
+        background: linear-gradient(135deg, #2563eb, #7c3aed);
+        -webkit-background-clip: text;
+        -webkit-text-fill-color: transparent;
+        transform: scale(1);
+        transition: transform 0.3s ease;
+        text-shadow: 2px 2px 4px rgba(0,0,0,0.1);
+        animation: fadeInDown 1.2s ease-out;
+      }
+      .theme-toggle {
+        position: fixed;
+        top: 20px;
+        right: 20px;
+        background: linear-gradient(135deg, #2563eb, #7c3aed);
+        color: white;
+        border: none;
+        border-radius: 50%;
+        width: 45px;
+        height: 45px;
+        cursor: pointer;
+        display: flex;
+        align-items: center;
+        justify-content: center;
+        transition: all var(--transition-speed);
+        box-shadow: 0 4px 12px rgba(0,0,0,0.15);
+        z-index: 1000;
+      }
+      .theme-toggle:hover {
+        transform: rotate(180deg) scale(1.1);
+        box-shadow: 0 6px 16px rgba(0,0,0,0.2);
+      }
+      .upload-container, .table-container {
+        background: var(--card-light);
+        padding: calc(var(--spacing-unit) * 3);
+        border-radius: var(--border-radius);
+        box-shadow: 0 8px 32px rgba(0,0,0,0.1);
+        margin-bottom: calc(var(--spacing-unit) * 4);
+        transition: transform var(--transition-speed), box-shadow var(--transition-speed);
+        animation: fadeIn 1s ease-out;
+      }
+      body.dark .upload-container,
+      body.dark .table-container {
+        background: var(--card-dark);
+        box-shadow: 0 8px 32px rgba(0,0,0,0.2);
+      }
+      .upload-container:hover,
+      .table-container:hover {
+        transform: translateY(-5px);
+        box-shadow: 0 12px 48px rgba(0,0,0,0.15);
+      }
+      .form-group {
+        margin-bottom: calc(var(--spacing-unit) * 2);
+      }
+      .form-control {
+        border-radius: var(--border-radius);
+        border: 2px solid var(--border-light);
+        padding: calc(var(--spacing-unit) * 1.5);
+        transition: all var(--transition-speed);
+        background: var(--input-light);
+        color: var(--text-light);
+        height: auto;
+      }
+      body.dark .form-control {
+        background: var(--input-dark);
+        border-color: var(--border-dark);
+        color: var(--text-dark);
+      }
+      .form-control:focus {
+        border-color: #2563eb;
+        box-shadow: 0 0 0 3px rgba(37, 99, 235, 0.2);
+      }
+      .form-control::file-selector-button {
+        padding: 8px 16px;
+        border-radius: 8px;
+        border: none;
+        background: linear-gradient(135deg, #2563eb, #7c3aed);
+        color: white;
+        margin-right: 16px;
+        transition: all 0.3s;
+      }
+      .form-control::file-selector-button:hover {
+        background: linear-gradient(135deg, #1d4ed8, #6d28d9);
+        transform: translateY(-1px);
+      }
+      .btn-custom {
+        padding: calc(var(--spacing-unit) * 1.5) calc(var(--spacing-unit) * 3);
+        border-radius: var(--border-radius);
+        font-weight: 600;
+        transition: all var(--transition-speed);
+        position: relative;
+        overflow: hidden;
+        border: none;
+        background: linear-gradient(135deg, #2563eb, #7c3aed);
+        color: white;
+      }
+      .btn-custom::after {
+        content: '';
+        position: absolute;
+        top: 50%;
+        left: 50%;
+        width: 0;
+        height: 0;
+        background: rgba(255,255,255,0.2);
+        border-radius: 50%;
+        transform: translate(-50%, -50%);
+        transition: width 0.6s, height 0.6s;
+      }
+      .btn-custom:hover::after {
+        width: 300%;
+        height: 300%;
+      }
+      .btn-custom:hover {
+        transform: translateY(-2px);
+        box-shadow: 0 4px 12px rgba(0,0,0,0.2);
+      }
+      .table {
+        margin-top: calc(var(--spacing-unit) * 2);
+      }
+      .table thead th {
+        background: linear-gradient(135deg, #2563eb, #7c3aed);
+        color: white;
+        font-weight: 600;
+        padding: calc(var(--spacing-unit) * 1.5);
+        border: none;
+      }
+      .table tbody tr {
+        transition: background-color var(--transition-speed);
+        color: var(--text-light);
+      }
+      body.dark .table tbody tr {
+        color: var(--text-dark);
+      }
+      .table tbody tr:hover {
+        background-color: rgba(37, 99, 235, 0.1);
+      }
+      .edit-btn {
+        background: linear-gradient(135deg, #dc2626, #ef4444) !important;
+        color: white !important;
+        padding: calc(var(--spacing-unit) * 0.75) calc(var(--spacing-unit) * 1.5);
+        font-size: 0.9rem;
+        border-radius: var(--border-radius);
+        border: none;
+      }
+      .edit-btn:hover {
+        background: linear-gradient(135deg, #b91c1c, #dc2626) !important;
+        transform: translateY(-2px);
+      }
+      .frozen-badge {
+        padding: calc(var(--spacing-unit) * 0.75) calc(var(--spacing-unit) * 1.5);
+        border-radius: var(--border-radius);
+        background: linear-gradient(135deg, #059669, #10b981);
+        color: white;
+      }
+      .modal-content {
+        border-radius: var(--border-radius);
+        overflow: hidden;
+      }
+      .modal-header {
+        background: linear-gradient(135deg, #dc2626, #ef4444);
+        color: white;
+        border: none;
+        padding: calc(var(--spacing-unit) * 2);
+      }
+      .modal-title {
+        color: white;
+      }
+      .modal-body {
+        padding: calc(var(--spacing-unit) * 3);
+      }
+      @keyframes fadeIn {
+        from { opacity: 0; transform: translateY(20px); }
+        to { opacity: 1; transform: translateY(0); }
+      }
+      @keyframes fadeInDown {
+        from { opacity: 0; transform: translateY(-50px); }
+        to { opacity: 1; transform: translateY(0); }
+      }
+      /* Dark mode specific styles */
+      body.dark .table {
+        color: var(--text-dark);
+      }
+      body.dark .form-control {
+        background: var(--card-dark);
+      }
+      body.dark .modal-content {
+        background: var(--card-dark);
+        color: var(--text-dark);
+      }
+      body.dark .modal-body {
+        color: var(--text-dark);
+      }
+      body.dark .close {
+        color: white;
+      }
+      body.dark label {
+        color: var(--text-dark);
+      }
+      .btn-success {
+        background: linear-gradient(135deg, #059669, #10b981) !important;
+        border: none !important;
+      }
+      .btn-info {
+        background: linear-gradient(135deg, #0284c7, #0ea5e9) !important;
+        border: none !important;
+      }
+      .btn-secondary {
+        background: linear-gradient(135deg, #4b5563, #6b7280) !important;
+        border: none !important;
+      }
+      .btn-danger {
+        background: linear-gradient(135deg, #dc2626, #ef4444) !important;
+        border: none !important;
+      }
+      /* =====================================================
+         Updated Clock Loader CSS (Purple Clock with Moving Hands)
+      ====================================================== */
+      .loader-container {
+        position: fixed;
+        top: 0;
+        left: 0;
+        width: 100%;
+        height: 100%;
+        background: rgba(0, 0, 0, 0.7);
+        display: flex;
+        justify-content: center;
+        align-items: center;
+        z-index: 9999;
+      }
+      .clock-loader {
+        text-align: center;
+      }
+      .clock-face {
+        width: 120px;
+        height: 120px;
+        border: 8px solid #7c3aed;
+        border-radius: 50%;
+        position: relative;
+        background: rgba(124, 58, 237, 0.1);
+        margin: 0 auto 20px;
+        box-shadow: 0 0 20px rgba(124, 58, 237, 0.3);
+      }
+      /* Positioning each hand so its bottom is at the center of the clock-face */
+      .hand {
+        position: absolute;
+        left: 50%;
+        bottom: 50%;
+        transform-origin: bottom;
+        transform: translateX(-50%) rotate(0deg);
+      }
+      .hour-hand {
+        width: 4px;
+        height: 25px;
+        background: #7c3aed;
+        animation: rotate-hour 8s linear infinite;
+      }
+      .minute-hand {
+        width: 3px;
+        height: 35px;
+        background: #7c3aed;
+        animation: rotate-minute 6s linear infinite;
+      }
+      .second-hand {
+        width: 2px;
+        height: 45px;
+        background: #2563eb;
+        animation: rotate-second 2s linear infinite;
+      }
+      .center-dot {
+        width: 12px;
+        height: 12px;
+        background: #7c3aed;
+        border-radius: 50%;
+        position: absolute;
+        top: calc(50% - 6px);
+        left: calc(50% - 6px);
+        box-shadow: 0 0 10px rgba(124, 58, 237, 0.5);
+      }
+      /* Optional tick marks on the clock-face */
+      .tick {
+        position: absolute;
+        width: 3px;
+        height: 10px;
+        background: #7c3aed;
+        top: 4px;
+        left: calc(50% - 1.5px);
+        transform-origin: 50% 60px;
+      }
+      .tick-1 { transform: rotate(0deg); }
+      .tick-2 { transform: rotate(90deg); }
+      .tick-3 { transform: rotate(180deg); }
+      .tick-4 { transform: rotate(270deg); }
+      @keyframes rotate-hour {
+        from { transform: translateX(-50%) rotate(0deg); }
+        to { transform: translateX(-50%) rotate(360deg); }
+      }
+      @keyframes rotate-minute {
+        from { transform: translateX(-50%) rotate(0deg); }
+        to { transform: translateX(-50%) rotate(360deg); }
+      }
+      @keyframes rotate-second {
+        from { transform: translateX(-50%) rotate(0deg); }
+        to { transform: translateX(-50%) rotate(360deg); }
+      }
+    </style>
+  </head>
+  <body>
+    <!-- Theme toggle button -->
+    <button class="theme-toggle" onclick="toggleTheme()">
+      <i class="fas fa-moon"></i>
+    </button>
+    <!-- Loader Container (hidden by default) -->
+    <div class="loader-container" id="loader-container" style="display: none;">
+      <div class="clock-loader">
+        <div class="clock-face">
+          <div class="hand hour-hand"></div>
+          <div class="hand minute-hand"></div>
+          <div class="hand second-hand"></div>
+          <div class="center-dot"></div>
+          <div class="tick tick-1"></div>
+          <div class="tick tick-2"></div>
+          <div class="tick tick-3"></div>
+          <div class="tick tick-4"></div>
+        </div>
+        <div class="loading-text" style="color: #fff; font-size: 18px; margin-top: 20px; font-weight: 500;">Processing Invoices...</div>
+      </div>
+    </div>
+    <div class="container mx-auto">
+      <h1>Intelligent Partial Invoice Matching</h1>
+      <div class="upload-container">
+        <form id="uploadForm" method="post" enctype="multipart/form-data">
+          <div class="form-group">
+            <label for="file1">Upload First Dataset (CSV/Excel):</label>
+            <input type="file" class="form-control" id="file1" name="file1" required>
+          </div>
+          <div class="form-group">
+            <label for="file2">Upload Second Dataset (CSV/Excel):</label>
+            <input type="file" class="form-control" id="file2" name="file2" required>
+          </div>
+          <button type="submit" class="btn btn-primary btn-custom btn-block">
+            <span>Process Invoices</span>
+          </button>
+        </form>
+      </div>
+      {% if results %}
+      <div class="table-container">
+        <h2 class="text-center font-bold mb-4">Matched Invoices Preview</h2>
+        <div class="table-responsive">
+          <table class="table table-striped table-bordered" id="resultsTable">
+            <thead>
+              <tr>
+                <th>Invoice Number 1</th>
+                <th>Invoice Number 2</th>
+                <th>Similarity Score</th>
+                <th>Manual Review Status</th>
+                <th>Recommendation</th>
+                <th>Reason</th>
+                <th>Comments</th>
+                <th class="action-btn">Action</th>
+              </tr>
+            </thead>
+            <tbody>
+              {% for row in results %}
+              <tr data-index="{{ loop.index0 }}">
+                <td class="invoice_number1">{{ row.invoice_number1 }}</td>
+                <td class="invoice_number2">{{ row.invoice_number2 }}</td>
+                <td class="similarity_score">{{ row.similarity_score }}</td>
+                <td class="manual_review_status">
+                  {% if row.recommendation == "Exact Match" %}
+                  No
+                  {% else %}
+                  Needs Review
+                  {% endif %}
+                </td>
+                <td class="recommendation">{{ row.recommendation }}</td>
+                <td class="reason">{{ row.reason }}</td>
+                <td class="comments">{{ row.comments }}</td>
+                <td class="action-btn">
+                  <div class="text-center">
+                    {% if row.recommendation == "Exact Match" %}
+                    <button class="btn btn-success btn-custom freeze-btn" disabled style="padding: 10px 20px;filter: brightness(1.3);">
+                      <i class="fas fa-lock"></i> Freeze
+                    </button>
+                    {% else %}
+                    <input type="checkbox" class="select-review-checkbox" data-index="{{ loop.index0 }}" style="width:30px; height:30px;">
+                    {% endif %}
+                  </div>
+                </td>
+              </tr>
+              {% endfor %}
+            </tbody>
+          </table>
+        </div>
+      </div>
+      <div class="row mb-4">
+        <div class="col-md-4 mb-3">
+          <a href="{{ url_for('download_csv') }}" class="btn btn-success btn-custom btn-block">
+            <i class="fas fa-file-csv"></i> Download CSV Report
+          </a>
+        </div>
+        <div class="col-md-4 mb-3">
+          <a href="{{ url_for('download_excel') }}" class="btn btn-info btn-custom btn-block">
+            <i class="fas fa-file-excel"></i> Download Excel Report
+          </a>
+        </div>
+        <div class="col-md-4 mb-3">
+          <!-- New Download Summary Stats Button -->
+          <button id="downloadStatsBtn" class="btn btn-secondary btn-custom btn-block">
+            <i class="fas fa-chart-bar"></i> Download Summary Stats
+          </button>
+        </div>
+      </div>
+      <div class="text-center mb-5">
+        <button id="saveUpdates" class="btn btn-primary btn-custom">
+          <i class="fas fa-save"></i> Save All Updates
+        </button>
+      </div>
+      {% endif %}
+    </div>
+    <!-- Edit Modal -->
+    <div class="modal fade" id="editModal" tabindex="-1" aria-labelledby="editModalLabel" aria-hidden="true">
+      <div class="modal-dialog modal-dialog-centered">
+        <div class="modal-content">
+          <div class="modal-header">
+            <h5 class="modal-title" id="editModalLabel">
+              <i class="fas fa-edit"></i> Edit Invoice Match
+            </h5>
+            <button type="button" class="close text-white" data-dismiss="modal" aria-label="Close">
+              <span aria-hidden="true">&times;</span>
+            </button>
+          </div>
+          <form id="editForm">
+            <div class="modal-body">
+              <!-- Non-editable fields -->
+              <div class="form-group">
+                <label>Invoice Number 1:</label>
+                <p id="modalInvoice1" class="font-weight-bold"></p>
+              </div>
+              <div class="form-group">
+                <label>Invoice Number 2 (Current):</label>
+                <p id="modalInvoice2" class="font-weight-bold"></p>
+              </div>
+              <div class="form-group">
+                <label>Similarity Score:</label>
+                <p id="modalScore" class="font-weight-bold"></p>
+              </div>
+              <!-- New select box for corrected invoice from dataset2 -->
+              <div class="form-group">
+                <label for="modalSelectInvoice2">Select Correct Invoice from Dataset 2:</label>
+                <select class="form-control" id="modalSelectInvoice2">
+                  <option value="">-- Select --</option>
+                  {% for inv in unique_values %}
+                  <option value="{{ inv }}">{{ inv }}</option>
+                  {% endfor %}
+                </select>
+              </div>
+              <!-- Editable fields -->
+              <div class="form-group">
+                <label for="modalReviewStatus">Manual Review Status:</label>
+                <select class="form-control" id="modalReviewStatus" required>
+                  <option value="Needs Review">Needs Review</option>
+                  <option value="No Review Needed">No Review Needed</option>
+                </select>
+              </div>
+              <div class="form-group">
+                <label for="modalRecommendation">Recommendation:</label>
+                <select class="form-control" id="modalRecommendation" required>
+                  <option value="Unmatched">Unmatched</option>
+                  <option value="Partial Match">Partial Match</option>
+                  <option value="Exact Match">Exact Match</option>
+                </select>
+              </div>
+              <div class="form-group">
+                <label for="modalReason">Reason:</label>
+                <textarea class="form-control" id="modalReason" rows="3" required></textarea>
+              </div>
+              <div class="form-group">
+                <label for="modalComments">Comments:</label>
+                <textarea class="form-control" id="modalComments" rows="2"></textarea>
+              </div>
+              <!-- Hidden index -->
+              <input type="hidden" id="modalRowIndex">
+            </div>
+            <div class="modal-footer">
+              <button type="button" class="btn btn-secondary btn-custom" data-dismiss="modal">
+                <i class="fas fa-times"></i> Cancel
+              </button>
+              <button type="submit" class="btn btn-danger btn-custom">
+                <i class="fas fa-save"></i> Save Changes
+              </button>
+            </div>
+          </form>
+        </div>
+      </div>
+    </div>
+    <!-- jQuery, Popper.js, Bootstrap JS -->
+    <script src="https://code.jquery.com/jquery-3.5.1.min.js"></script>
+    <script src="https://cdn.jsdelivr.net/npm/popper.js@1.16.0/dist/umd/popper.min.js"></script>
+    <script src="https://stackpath.bootstrapcdn.com/bootstrap/4.5.0/js/bootstrap.min.js"></script>
+    <script>
+      // Theme toggle functionality
+      function toggleTheme() {
+        const body = document.body;
+        const themeToggle = document.querySelector('.theme-toggle i');
+        body.classList.toggle('dark');
+        if (body.classList.contains('dark')) {
+          themeToggle.classList.remove('fa-moon');
+          themeToggle.classList.add('fa-sun');
+          localStorage.setItem('theme', 'dark');
+        } else {
+          themeToggle.classList.remove('fa-sun');
+          themeToggle.classList.add('fa-moon');
+          localStorage.setItem('theme', 'light');
+        }
+      }
+      // Check for saved theme preference
+      const savedTheme = localStorage.getItem('theme');
+      if (savedTheme === 'dark') {
+        document.body.classList.add('dark');
+        document.querySelector('.theme-toggle i').classList.replace('fa-moon', 'fa-sun');
+      }
+
+      // New JS: Download Summary Stats (Excel & JSON) when button is clicked
+      document.getElementById('downloadStatsBtn').addEventListener('click', function() {
+        // Trigger Excel stats download
+        const excelLink = document.createElement('a');
+        excelLink.href = "{{ url_for('download_stats_excel') }}";
+        excelLink.style.display = 'none';
+        document.body.appendChild(excelLink);
+        excelLink.click();
+        document.body.removeChild(excelLink);
+
+        // Trigger JSON stats download
+        const jsonLink = document.createElement('a');
+        jsonLink.href = "{{ url_for('download_stats_json') }}";
+        jsonLink.style.display = 'none';
+        document.body.appendChild(jsonLink);
+        jsonLink.click();
+        document.body.removeChild(jsonLink);
+      });
+
+      // Populate modal with row data when an edit button is clicked
+      $(document).on("click", ".edit-btn", function() {
+        var rowIndex = $(this).data("index");
+        var row = $("#resultsTable tbody tr").eq(rowIndex);
+        var invoice1 = row.find(".invoice_number1").text().trim();
+        var invoice2 = row.find(".invoice_number2").text().trim();
+        var score = row.find(".similarity_score").text().trim();
+        var reviewStatus = row.find(".manual_review_status").text().trim();
+        var recommendation = row.find(".recommendation").text().trim();
+        var reason = row.find(".reason").text().trim();
+        var comments = row.find(".comments").text().trim();
+        $("#modalRowIndex").val(rowIndex);
+        $("#modalInvoice1").text(invoice1);
+        $("#modalInvoice2").text(invoice2);
+        $("#modalScore").text(score);
+        $("#modalReviewStatus").val(reviewStatus);
+        $("#modalRecommendation").val(recommendation);
+        $("#modalReason").val(reason);
+        $("#modalComments").val(comments);
+        // Reset the select box
+        $("#modalSelectInvoice2").val("");
+      });
+
+      // Updated edit form submit handler: feedback is sent to the server, which returns recalculated values.
+      $("#editForm").on("submit", function(e) {
+        e.preventDefault();
+        var rowIndex = $("#modalRowIndex").val();
+        var selectedInvoice2 = $("#modalSelectInvoice2").val();
+        var newComments = $("#modalComments").val();
+        var row = $("#resultsTable tbody tr").eq(rowIndex);
+        var invoice1 = row.find(".invoice_number1").text().trim();
+        // Send feedback to the server
+        $.ajax({
+          url: "{{ url_for('save_feedback') }}",
+          type: "POST",
+          contentType: "application/json",
+          data: JSON.stringify({
+            invoice_number1: invoice1,
+            selected_invoice2: selectedInvoice2,
+            comments: newComments
+          }),
+          success: function(response) {
+            alert(response.message);
+          },
+          error: function(xhr, status, error) {
+            alert("Error saving feedback: " + error);
+          }
+        });
+        $("#editModal").modal("hide");
+      });
+
+      // When "Save All Updates" is clicked, update each row's action cell based on its type.
+      $("#saveUpdates").on("click", function() {
+        var updatedData = [];
+        $("#resultsTable tbody tr").each(function() {
+          var row = $(this);
+          var rec = row.find(".recommendation").text().trim();
+          var actionCell = row.find("td.action-btn");
+          var rowIndex = row.data("index");
+          // For Exact Match, always freeze
+          if(rec === "Exact Match") {
+            row.find(".manual_review_status").text("No");
+            actionCell.html('<div class="text-center"><button class="btn btn-success btn-custom freeze-btn" disabled style="padding: 10px 20px;filter: brightness(1.3);"><i class="fas fa-lock"></i> Freeze</button></div>');
+          } else {
+            // For Partial Match or Unmatched, check the checkbox state.
+            var checkbox = row.find("input.select-review-checkbox");
+            if(checkbox.length > 0) {
+              if(checkbox.is(":checked")) {
+                // If checked, auto-freeze this row.
+                row.find(".manual_review_status").text("No");
+                actionCell.html('<div class="text-center"><button class="btn btn-success btn-custom freeze-btn" disabled style="padding: 10px 20px;filter: brightness(1.3);"><i class="fas fa-lock"></i> Freeze</button></div>');
+              } else {
+                // If not checked, allow manual editing.
+                actionCell.html('<button class="btn btn-sm edit-btn btn-custom" data-index="'+ rowIndex +'" data-toggle="modal" data-target="#editModal"><i class="fas fa-edit"></i> Edit</button>');
+              }
+            }
+          }
+          updatedData.push({
+            invoice_number1: row.find(".invoice_number1").text().trim(),
+            invoice_number2: row.find(".invoice_number2").text().trim(),
+            similarity_score: parseFloat(row.find(".similarity_score").text().trim()),
+            manual_review_status: row.find(".manual_review_status").text().trim(),
+            recommendation: row.find(".recommendation").text().trim(),
+            reason: row.find(".reason").text().trim(),
+            comments: row.find(".comments").text().trim(),
+            editable: row.find("button.edit-btn").length > 0
+          });
+        });
+        $.ajax({
+          url: "{{ url_for('save_updates') }}",
+          type: "POST",
+          contentType: "application/json",
+          data: JSON.stringify(updatedData),
+          success: function(response) {
+            alert("Updates taken successfully!");
+          },
+          error: function(xhr, status, error) {
+            alert("Error saving updates: " + error);
+          }
+        });
+      });
+
+      // Hide loader when processing is complete (if the server sends processing_complete = true)
+      {% if processing_complete %}
+        document.getElementById('loader-container').style.display = 'none';
+      {% endif %}
+
+      // Show loader on file upload submission
+      document.getElementById('uploadForm').addEventListener('submit', function(e) {
+        // Show loader
+        document.getElementById('loader-container').style.display = 'flex';
+        // Optional: Disable the submit button to prevent double submission
+        const submitButton = this.querySelector('button[type="submit"]');
+        if (submitButton) {
+          submitButton.disabled = true;
+        }
+      });
+    </script>
+  </body>
+</html>