Bhuvi13 commited on
Commit
82f069e
·
verified ·
1 Parent(s): c755a11

Update src/streamlit_app.py

Browse files
Files changed (1) hide show
  1. src/streamlit_app.py +274 -475
src/streamlit_app.py CHANGED
@@ -1,6 +1,12 @@
1
  import os
2
  # --- Fix: ensure HOME is writable before Streamlit initializes ---
3
  from pathlib import Path
 
 
 
 
 
 
4
 
5
  _home = os.environ.get("HOME", "")
6
  if _home in ("", "/", None):
@@ -32,45 +38,10 @@ import pandas as pd
32
  from PIL import Image
33
  from huggingface_hub import login
34
 
35
-
36
- # ---------------------------
37
- # Helper: Update session state on widget change
38
- # ---------------------------
39
- def _update_flat_field(field_name: str, widget_key: str):
40
- """Update a top-level field in edited_data."""
41
- current_hash = st.session_state.get("current_file_hash")
42
- if not current_hash or current_hash not in st.session_state.batch_results:
43
- return
44
- new_value = st.session_state[widget_key]
45
- st.session_state.batch_results[current_hash]["edited_data"][field_name] = new_value
46
-
47
-
48
- def _update_nested_field(field_path: str, widget_key: str):
49
- """Update a nested field like 'Sender.Name' or 'Bank Details.bank_name'."""
50
- current_hash = st.session_state.get("current_file_hash")
51
- if not current_hash or current_hash not in st.session_state.batch_results:
52
- return
53
- new_value = st.session_state[widget_key]
54
- data_ref = st.session_state.batch_results[current_hash]["edited_data"]
55
-
56
- keys = field_path.split(".")
57
- d = data_ref
58
- for k in keys[:-1]:
59
- if k not in d or not isinstance(d[k], dict):
60
- d[k] = {}
61
- d = d[k]
62
- d[keys[-1]] = new_value
63
-
64
-
65
- def _update_line_items(widget_key: str):
66
- """Auto-update line items when data_editor changes."""
67
- current_hash = st.session_state.get("current_file_hash")
68
- if not current_hash or current_hash not in st.session_state.batch_results:
69
- return
70
- df = st.session_state[widget_key]
71
- records = df.to_dict('records')
72
- st.session_state.batch_results[current_hash]["edited_data"]["Itemized Data"] = records
73
-
74
 
75
  # ---------------------------
76
  # UI: main
@@ -236,44 +207,25 @@ def run_inference_on_image(image: Image.Image, processor, model, device, decoder
236
  # ---------------------------
237
  def map_prediction_to_ui(pred):
238
  import json, re
239
- from collections import defaultdict
240
 
241
- # --- parse raw string payloads that embed JSON ---
242
  def safe_json_load(s):
243
  if s is None:
244
  return None
245
  if isinstance(s, (dict, list)):
246
  return s
247
  if isinstance(s, str):
248
- s = s.strip()
249
- if s == "":
250
- return None
251
  try:
252
  return json.loads(s)
253
  except Exception:
254
- # try to extract balanced-brace substrings (simple approach)
255
- subs = []
256
- stack = []
257
- start = None
258
- for i, ch in enumerate(s):
259
- if ch == "{":
260
- if not stack:
261
- start = i
262
- stack.append("{")
263
- elif ch == "}":
264
- if stack:
265
- stack.pop()
266
- if not stack and start is not None:
267
- subs.append(s[start:i+1])
268
- start = None
269
- for sub in subs:
270
- try:
271
- return json.loads(sub)
272
- except Exception:
273
- continue
274
  return None
275
 
276
- # --- normalize numeric strings like "1,800.00" -> float ---
277
  def clean_number(x):
278
  if x is None:
279
  return 0.0
@@ -282,7 +234,6 @@ def map_prediction_to_ui(pred):
282
  s = str(x).strip()
283
  if s == "":
284
  return 0.0
285
- # remove commas and non-number chars except dot and minus
286
  s = re.sub(r"[,\s]", "", s)
287
  s = re.sub(r"[^\d\.\-]", "", s)
288
  if s in ("", ".", "-", "-."):
@@ -292,69 +243,28 @@ def map_prediction_to_ui(pred):
292
  except Exception:
293
  return 0.0
294
 
295
- # --- collect all keys -> list of values from pred, recursively ---
296
- def collect_keys(obj, out):
297
- if isinstance(obj, dict):
298
- for k, v in obj.items():
299
- lk = str(k).strip().lower()
300
- out[lk].append(v)
301
- collect_keys(v, out)
302
- elif isinstance(obj, list):
303
- for it in obj:
304
- collect_keys(it, out)
305
- else:
306
- # primitive: handled via parent key
307
- pass
308
-
309
- # --- find list-of-dicts candidates for items (recursively) ---
310
- def collect_lists_of_dicts(obj, out_lists):
311
- if isinstance(obj, dict):
312
- for v in obj.values():
313
- if isinstance(v, list) and v and isinstance(v[0], dict):
314
- out_lists.append(v)
315
- else:
316
- collect_lists_of_dicts(v, out_lists)
317
- elif isinstance(obj, list):
318
- for it in obj:
319
- if isinstance(it, list) and it and isinstance(it[0], dict):
320
- out_lists.append(it)
321
- else:
322
- collect_lists_of_dicts(it, out_lists)
323
-
324
- # --- map item dict -> UI item row using the keys you specified in example ---
325
- def map_item_dict(it):
326
- if not isinstance(it, dict):
327
- return None
328
- # lowered keys mapping
329
- lower = {str(k).strip().lower(): v for k, v in it.items()}
330
- desc = (lower.get("descriptions") or lower.get("description") or lower.get("desc") or lower.get("item") or "")
331
- qty = lower.get("quantity") or lower.get("qty") or lower.get("count") or ""
332
- unit_price = lower.get("unit_price") or lower.get("price") or ""
333
- amount = lower.get("amount") or lower.get("line_total") or lower.get("line total") or lower.get("total") or ""
334
- tax = lower.get("tax") or lower.get("tax_amount") or ""
335
- line_total = lower.get("line_total") or lower.get("line_total".lower()) or lower.get("line total") or amount
336
-
337
- return {
338
- "Description": str(desc).strip(),
339
- "Quantity": float(clean_number(qty)),
340
- "Unit Price": float(clean_number(unit_price)),
341
- "Amount": float(clean_number(amount)),
342
- "Tax": float(clean_number(tax)),
343
- "Line Total": float(clean_number(line_total))
344
- }
345
-
346
- # ----------------- Start mapping -----------------
347
- # Try parse if pred is a JSON-like string
348
- parsed = safe_json_load(pred) if isinstance(pred, str) else pred
349
- if parsed is None and isinstance(pred, str):
350
- # not parseable -> fallback to empty UI
351
- parsed = None
352
-
353
- if parsed is None and not isinstance(pred, dict):
354
- # nothing we can map
355
- parsed = pred # will still allow collect_keys if it's dict; else produce empty ui
356
 
357
- # create empty UI template
358
  ui = {
359
  "Invoice Number": "",
360
  "Invoice Date": "",
@@ -374,122 +284,138 @@ def map_prediction_to_ui(pred):
374
  "Itemized Data": []
375
  }
376
 
377
- # If parsed is a dict, collect all keys and list-of-dict candidates
378
- key_map = defaultdict(list) # lowercase-key -> list of values
379
- list_candidates = [] # list of list-of-dicts found
380
- if isinstance(parsed, dict):
381
- collect_keys(parsed, key_map)
382
- collect_lists_of_dicts(parsed, list_candidates)
383
- elif isinstance(pred, dict):
384
- # if parsing failed but original pred is dict, use that
385
- collect_keys(pred, key_map)
386
- collect_lists_of_dicts(pred, list_candidates)
387
-
388
- # Helper to pick first non-empty value from candidate keys
389
- def pick_first(*candidate_keys):
390
- for k in candidate_keys:
391
- lk = k.strip().lower()
392
- if lk in key_map:
393
- # pick first non-empty
394
- for v in key_map[lk]:
395
- if v is None:
396
- continue
397
- # return primitive or string immediately; if dict/list, return as-is
398
- if isinstance(v, (dict, list)):
399
- return v
400
- s = str(v).strip()
401
- if s != "":
402
- return s
403
- return None
404
 
405
- # Map simple scalar fields using the exact keys you provided (plus common close variants)
406
- ui["Invoice Number"] = pick_first("invoice_no", "invoice_number", "invoiceid", "invoice id") or ""
407
- ui["Invoice Date"] = pick_first("invoice_date", "date", "invoice date") or ""
408
- ui["Due Date"] = pick_first("due_date", "due_date", "due") or ""
409
- ui["Sender Name"] = pick_first("sender_name", "sender") or ""
410
- ui["Sender Address"] = pick_first("sender_addr", "sender_address", "sender addr") or ""
411
- ui["Recipient Name"] = pick_first("rcpt_name", "recipient_name", "recipient", "rcpt") or ""
412
- ui["Recipient Address"] = pick_first("rcpt_addr", "recipient_address", "recipient addr") or ""
413
 
414
- # bank details: gather keys that start with 'bank_' or exact matches
415
- bank = {}
416
- for bk in ("bank_name", "bank_acc_no", "bank_account_number", "bank_acc_name", "bank_iban", "bank_swift", "bank_routing", "bank_branch", "iban"):
417
- val = pick_first(bk, bk.replace("bank_", "")) # allow both 'iban' and 'bank_iban'
418
- if val:
419
- # normalize key name to bank_* form
420
- if bk == "iban":
421
- bank["bank_iban"] = str(val)
422
- else:
423
- bank[bk] = str(val)
424
- ui["Bank Details"] = bank
 
 
 
 
 
 
 
 
 
 
 
 
425
 
426
- # summary / totals
427
- ui["Subtotal"] = clean_number(pick_first("subtotal", "sub_total", "sub total") or 0.0)
428
- ui["Tax Percentage"] = clean_number(pick_first("tax_rate", "tax_percentage", "tax pct", "tax percentage") or 0.0)
429
- ui["Total Tax"] = clean_number(pick_first("tax_amount", "tax", "total_tax") or 0.0)
430
- ui["Total Amount"] = clean_number(pick_first("total_amount", "grand_total", "total", "amount") or 0.0)
431
- ui["Currency"] = (pick_first("currency") or "").strip()
432
-
433
- # Item extraction:
434
- items_rows = []
435
-
436
- # 1) If we found an explicit list-of-dicts candidate, use the first one that looks like invoice items
437
- def list_looks_like_items(lst):
438
- if not isinstance(lst, list) or not lst:
439
- return False
440
- if not isinstance(lst[0], dict):
441
- return False
442
- # check if any expected item key present in first element
443
- expected = {"descriptions", "description", "desc", "item", "quantity", "qty", "amount", "unit_price", "line_total", "line_total".lower(), "line_total"}
444
- keys0 = {str(k).strip().lower() for k in lst[0].keys()}
445
- return bool(expected.intersection(keys0))
446
-
447
- for cand in list_candidates:
448
- if list_looks_like_items(cand):
449
- for it in cand:
450
- row = map_item_dict(it)
451
- if row is not None:
452
- items_rows.append(row)
453
- # prefer first plausible list
454
- if items_rows:
455
- break
456
-
457
- # 2) If no list-of-dicts found, try to find a single dict anywhere that looks like an item (e.g., 'items': {...} as dict)
458
- if not items_rows:
459
- # search key_map values for dicts that have item-like keys
460
- for k, vals in key_map.items():
461
- for v in vals:
462
- if isinstance(v, dict):
463
- # does this dict have an item-like key?
464
- lower_keys = {str(x).strip().lower() for x in v.keys()}
465
- if lower_keys.intersection({"descriptions", "description", "desc", "amount", "line_total", "quantity", "qty", "unit_price"}):
466
- row = map_item_dict(v)
467
- if row is not None:
468
- items_rows.append(row)
469
- # we don't break because there might be multiple item-like dicts at different keys,
470
- # but continue scanning to collect all.
471
- # 3) Last resort: if key_map contains 'descriptions' or 'amount' as scalar but no dict, build a single-item row
472
- if not items_rows:
473
- desc = pick_first("descriptions", "description")
474
- amt = pick_first("amount", "line_total")
475
- qty = pick_first("quantity", "qty")
476
- unit_price = pick_first("unit_price", "price")
477
- if desc or amt or qty or unit_price:
478
- items_rows.append({
479
- "Description": str(desc or ""),
480
- "Quantity": float(clean_number(qty)),
481
- "Unit Price": float(clean_number(unit_price)),
482
- "Amount": float(clean_number(amt)),
483
- "Tax": float(clean_number(pick_first("tax", "tax_amount") or 0.0)),
484
- "Line Total": float(clean_number(amt or 0.0))
485
- })
486
-
487
- ui["Itemized Data"] = items_rows
488
-
489
- # Also set Sender/Recipient convenience fields
490
  ui["Sender"] = {"Name": ui["Sender Name"], "Address": ui["Sender Address"]}
491
  ui["Recipient"] = {"Name": ui["Recipient Name"], "Address": ui["Recipient Address"]}
492
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
493
  return ui
494
 
495
  # ---------------------------
@@ -499,53 +425,35 @@ def flatten_invoice_to_rows(invoice_data) -> list:
499
  """
500
  Converts nested invoice data into a flat list of rows (one per line item),
501
  with invoice-level and sender/recipient/bank fields repeated in each row.
502
- Ensures all expected bank fields are present, even if empty.
503
  """
504
- # Define all expected bank fields (must match UI keys)
505
- EXPECTED_BANK_FIELDS = [
506
- "bank_name",
507
- "bank_account_number",
508
- "bank_acc_name",
509
- "bank_iban",
510
- "bank_swift",
511
- "bank_routing",
512
- "bank_branch"
513
- ]
514
-
515
  rows = []
516
  line_items = invoice_data.get("Itemized Data", [])
517
 
518
- # Prepare base invoice info
519
- base_row = {
520
- "Invoice Number": invoice_data.get("Invoice Number", ""),
521
- "Invoice Date": invoice_data.get("Invoice Date", ""),
522
- "Due Date": invoice_data.get("Due Date", ""),
523
- "Currency": invoice_data.get("Currency", ""),
524
- "Subtotal": invoice_data.get("Subtotal", 0.0),
525
- "Tax Percentage": invoice_data.get("Tax Percentage", 0.0),
526
- "Total Tax": invoice_data.get("Total Tax", 0.0),
527
- "Total Amount": invoice_data.get("Total Amount", 0.0),
528
- "Sender Name": invoice_data.get("Sender", {}).get("Name", ""),
529
- "Sender Address": invoice_data.get("Sender", {}).get("Address", ""),
530
- "Recipient Name": invoice_data.get("Recipient", {}).get("Name", ""),
531
- "Recipient Address": invoice_data.get("Recipient", {}).get("Address", ""),
532
- }
533
-
534
- # Initialize all expected bank fields to empty string
535
- bank_row = {field: "" for field in EXPECTED_BANK_FIELDS}
536
-
537
- # Override with actual values if present
538
- bank_details = invoice_data.get("Bank Details", {}) or {}
539
- for k, v in bank_details.items():
540
- key_name = k if str(k).startswith("bank_") else f"bank_{k}"
541
- if key_name in bank_row:
542
- bank_row[key_name] = str(v).strip() if v is not None else ""
543
-
544
- base_row.update(bank_row)
545
-
546
  if not line_items:
547
  # If no line items, create one row with invoice info only
548
- row = base_row.copy()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
549
  row.update({
550
  "Item Description": "",
551
  "Item Quantity": 0,
@@ -559,7 +467,29 @@ def flatten_invoice_to_rows(invoice_data) -> list:
559
 
560
  # For each line item, create a row with all invoice context
561
  for item in line_items:
562
- row = base_row.copy()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
563
  row.update({
564
  "Item Description": item.get("Description", ""),
565
  "Item Quantity": item.get("Quantity", 0),
@@ -568,6 +498,7 @@ def flatten_invoice_to_rows(invoice_data) -> list:
568
  "Item Tax": item.get("Tax", 0.0),
569
  "Item Line Total": item.get("Line Total", item.get("Amount", 0.0)),
570
  })
 
571
  rows.append(row)
572
 
573
  return rows
@@ -678,15 +609,15 @@ if not st.session_state.is_processing_batch and len(st.session_state.batch_resul
678
  # RESULTS VIEW — Show selector + editable form
679
  # ---------------------------
680
  elif len(st.session_state.batch_results) > 0:
681
- # ---------------------------
682
- # Global Download All — produce a single Excel file (concatenated rows) and trigger direct download
683
- # ---------------------------
684
  if st.button("📦 Download All Results (Excel)", key="download_all"):
685
- # Collect rows from all invoices and concatenate into one DataFrame
686
  all_rows = []
687
  for file_hash, result in st.session_state.batch_results.items():
688
  rows = flatten_invoice_to_rows(result["edited_data"])
689
- # Annotate rows with source file name so user can identify which invoice each row came from
690
  for r in rows:
691
  r["Source File"] = result.get("file_name", file_hash)
692
  all_rows.extend(rows)
@@ -696,13 +627,13 @@ elif len(st.session_state.batch_results) > 0:
696
  else:
697
  full_df = pd.DataFrame(all_rows)
698
 
699
- # Reorder columns to put Source File first
700
  cols = list(full_df.columns)
701
  if "Source File" in cols:
702
  cols = ["Source File"] + [c for c in cols if c != "Source File"]
703
  full_df = full_df[cols]
704
 
705
- # Try to write XLSX (preferred). If engine not available, fall back to CSV.
706
  buffer = BytesIO()
707
  dl_filename = "all_extracted_invoices.xlsx"
708
  tried_xlsx = False
@@ -723,7 +654,7 @@ elif len(st.session_state.batch_results) > 0:
723
  dl_filename = "all_extracted_invoices.csv"
724
  mime = "text/csv"
725
 
726
- # Trigger immediate download via a data URI and small HTML snippet
727
  import base64
728
  import streamlit.components.v1 as components
729
  b64 = base64.b64encode(file_bytes).decode()
@@ -765,7 +696,8 @@ elif len(st.session_state.batch_results) > 0:
765
  # Get current file data
766
  current = st.session_state.batch_results[selected_hash]
767
  image = current["image"]
768
- data = current["edited_data"] # user edits go here
 
769
 
770
  # Layout
771
  left_col, right_col = st.columns([1, 1])
@@ -802,7 +734,6 @@ elif len(st.session_state.batch_results) > 0:
802
  st.rerun()
803
  except Exception as e:
804
  st.error(f"Re-run failed: {e}")
805
-
806
  tabs = st.tabs(["Invoice Details", "Sender/Recipient info", "Bank Details", "Line Items"])
807
 
808
  st.markdown(
@@ -831,225 +762,88 @@ elif len(st.session_state.batch_results) > 0:
831
  # ---------- Invoice Details ----------
832
  with tabs[0]:
833
  with st.container():
834
- widget_key = f"invoice_number_{selected_hash}"
835
- st.text_input(
836
- "Invoice Number",
837
- value=data.get('Invoice Number', ''),
838
- key=widget_key,
839
- on_change=_update_flat_field,
840
- args=("Invoice Number", widget_key)
841
- )
842
-
843
- widget_key = f"invoice_date_text_{selected_hash}"
844
- st.text_input(
845
- "Invoice Date",
846
- value=str(data.get('Invoice Date', '')).strip(),
847
- key=widget_key,
848
- on_change=_update_flat_field,
849
- args=("Invoice Date", widget_key)
850
- )
851
-
852
- widget_key = f"due_date_text_{selected_hash}"
853
- st.text_input(
854
- "Due Date",
855
- value=str(data.get('Due Date', '')).strip(),
856
- key=widget_key,
857
- on_change=_update_flat_field,
858
- args=("Due Date", widget_key)
859
- )
860
-
861
  curr_options = ['USD', 'EUR', 'GBP', 'INR', 'Other']
862
- curr_value = data.get('Currency', 'USD')
863
  curr_index = curr_options.index(curr_value) if curr_value in curr_options else (len(curr_options) - 1)
864
- new_curr = st.selectbox(
865
- "Currency",
866
- options=curr_options,
867
- index=curr_index,
868
- key=f"currency_select_{selected_hash}",
869
- on_change=_update_flat_field,
870
- args=("Currency", f"currency_select_{selected_hash}")
871
- )
872
  if new_curr == 'Other':
873
- widget_key = f"custom_currency_{selected_hash}"
874
- st.text_input(
875
- "Specify Currency",
876
- value=data.get('Currency', ''),
877
- key=widget_key,
878
- on_change=_update_flat_field,
879
- args=("Currency", widget_key)
880
- )
881
-
882
- def safe_number_input(label, value, key, field_name):
883
- try:
884
- v = float(value)
885
- except Exception:
886
- v = 0.0
887
- st.number_input(
888
- label,
889
- value=v,
890
- key=key,
891
- on_change=_update_flat_field,
892
- args=(field_name, key)
893
- )
894
-
895
- safe_number_input("Subtotal", data.get('Subtotal', 0.0), f"subtotal_{selected_hash}", "Subtotal")
896
- safe_number_input("Tax Percentage", data.get('Tax Percentage', 0.0), f"tax_pct_{selected_hash}", "Tax Percentage")
897
- safe_number_input("Total Tax", data.get('Total Tax', 0.0), f"total_tax_{selected_hash}", "Total Tax")
898
- safe_number_input("Total Amount", data.get('Total Amount', 0.0), f"total_amount_{selected_hash}", "Total Amount")
899
 
900
  # ---------- Sender / Recipient ----------
901
  with tabs[1]:
902
- if 'Sender' not in data:
903
- data['Sender'] = {'Name': '', 'Address': ''}
904
- if 'Recipient' not in data:
905
- data['Recipient'] = {'Name': '', 'Address': ''}
906
-
907
- sender_info = data['Sender']
908
- recipient_info = data['Recipient']
909
-
910
  with st.container():
911
- widget_key = f"sender_name_{selected_hash}"
912
- st.text_input(
913
- "Sender Name*",
914
- value=sender_info.get('Name', ''),
915
- key=widget_key,
916
- on_change=_update_nested_field,
917
- args=("Sender.Name", widget_key)
918
- )
919
-
920
- widget_key = f"sender_address_{selected_hash}"
921
- st.text_area(
922
- "Sender Address*",
923
- value=sender_info.get('Address', ''),
924
- key=widget_key,
925
- on_change=_update_nested_field,
926
- args=("Sender.Address", widget_key)
927
- )
928
-
929
- widget_key = f"recipient_name_{selected_hash}"
930
- st.text_input(
931
- "Recipient Name*",
932
- value=recipient_info.get('Name', ''),
933
- key=widget_key,
934
- on_change=_update_nested_field,
935
- args=("Recipient.Name", widget_key)
936
- )
937
-
938
- widget_key = f"recipient_address_{selected_hash}"
939
- st.text_area(
940
- "Recipient Address*",
941
- value=recipient_info.get('Address', ''),
942
- key=widget_key,
943
- on_change=_update_nested_field,
944
- args=("Recipient.Address", widget_key)
945
- )
946
-
947
  if st.button("⇄ Swap", help="Swap sender and recipient information", key=f"swap_{selected_hash}"):
948
- data['Sender'], data['Recipient'] = data['Recipient'], data['Sender']
949
- st.session_state.batch_results[selected_hash]["edited_data"] = data
950
  st.rerun()
951
 
952
  # ---------- Bank Details ----------
953
  with tabs[2]:
954
- bank_info = data.get('Bank Details', {}) or {}
955
  with st.container():
956
- widget_key = f"bank_name_{selected_hash}"
957
- st.text_input(
958
- "Bank Name",
959
- value=bank_info.get('bank_name', ''),
960
- key=widget_key,
961
- on_change=_update_nested_field,
962
- args=("Bank Details.bank_name", widget_key)
963
- )
964
-
965
- widget_key = f"bank_account_{selected_hash}"
966
- st.text_input(
967
- "Account Number",
968
- value=bank_info.get('bank_account_number', '') or bank_info.get('bank_acc_no',''),
969
- key=widget_key,
970
- on_change=_update_nested_field,
971
- args=("Bank Details.bank_account_number", widget_key)
972
- )
973
-
974
- widget_key = f"bank_acc_name_{selected_hash}"
975
- st.text_input(
976
- "Bank Account Name",
977
- value=bank_info.get('bank_acc_name', '') or bank_info.get('bank_acc_name', ''),
978
- key=widget_key,
979
- on_change=_update_nested_field,
980
- args=("Bank Details.bank_acc_name", widget_key)
981
- )
982
-
983
- widget_key = f"iban_{selected_hash}"
984
- st.text_input(
985
- "IBAN",
986
- value=bank_info.get('bank_iban', ''),
987
- key=widget_key,
988
- on_change=_update_nested_field,
989
- args=("Bank Details.bank_iban", widget_key)
990
- )
991
-
992
- widget_key = f"swift_code_{selected_hash}"
993
- st.text_input(
994
- "SWIFT Code",
995
- value=bank_info.get('bank_swift', ''),
996
- key=widget_key,
997
- on_change=_update_nested_field,
998
- args=("Bank Details.bank_swift", widget_key)
999
- )
1000
-
1001
- widget_key = f"routing_{selected_hash}"
1002
- st.text_input(
1003
- "Routing Number",
1004
- value=bank_info.get('bank_routing', ''),
1005
- key=widget_key,
1006
- on_change=_update_nested_field,
1007
- args=("Bank Details.bank_routing", widget_key)
1008
- )
1009
-
1010
- widget_key = f"branch_{selected_hash}"
1011
- st.text_input(
1012
- "Branch",
1013
- value=bank_info.get('bank_branch', ''),
1014
- key=widget_key,
1015
- on_change=_update_nested_field,
1016
- args=("Bank Details.bank_branch", widget_key)
1017
- )
1018
 
1019
  # ---------- Line Items ----------
1020
  with tabs[3]:
1021
  editor_key = f"item_editor_{selected_hash}"
1022
-
1023
- item_rows = data.get('Itemized Data', [])
1024
  df = pd.DataFrame(item_rows)
1025
  for col in ["Description", "Quantity", "Unit Price", "Amount", "Tax", "Line Total"]:
1026
  if col not in df.columns:
1027
  df[col] = ""
1028
-
1029
- st.write("✏️ Edit line items below. Changes are saved automatically.")
1030
- st.data_editor(
1031
  df,
1032
  num_rows="dynamic",
1033
  key=editor_key,
1034
  use_container_width=True,
1035
- on_change=_update_line_items,
1036
- args=(editor_key,)
1037
  )
1038
-
1039
- if len(df) == 0:
1040
  st.info("No line items found in the invoice.")
1041
 
1042
- # Save button (per file) — now optional since edits auto-save
1043
- if st.button("💾 Save Edits for This File (Optional)", key=f"save_{selected_hash}"):
1044
- # This is now redundant but kept for user reassurance
1045
- st.success(f"✅ Edits already auto-saved for {current['file_name']}")
 
 
 
1046
 
1047
  # Download buttons (per file)
1048
  st.markdown("---")
1049
  col_a, col_b, col_c = st.columns([1, 1, 1])
 
 
 
 
 
 
 
 
 
1050
  with col_b:
1051
  # ✅ Flatten entire invoice into rows (one per line item)
1052
- rows = flatten_invoice_to_rows(data)
1053
  full_df = pd.DataFrame(rows)
1054
 
1055
  # Optional: Reorder columns for better readability
@@ -1076,6 +870,11 @@ elif len(st.session_state.batch_results) > 0:
1076
  mime="text/csv",
1077
  key=f"dl_csv_{selected_hash}"
1078
  )
 
 
 
 
 
1079
 
1080
  # ---------------------------
1081
  # PROCESSING STATE — Show progress
 
1
  import os
2
  # --- Fix: ensure HOME is writable before Streamlit initializes ---
3
  from pathlib import Path
4
+ def safe_number_input(label, value, key):
5
+ try:
6
+ v = float(value)
7
+ except Exception:
8
+ v = 0.0
9
+ return st.number_input(label, value=v, key=key)
10
 
11
  _home = os.environ.get("HOME", "")
12
  if _home in ("", "/", None):
 
38
  from PIL import Image
39
  from huggingface_hub import login
40
 
41
+ # 👇 ADD CALLBACK FUNCTION AT TOP LEVEL
42
+ def trigger_rerun():
43
+ # Forces immediate rerun after any edit — ensures edited_df is fresh
44
+ pass
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
45
 
46
  # ---------------------------
47
  # UI: main
 
207
  # ---------------------------
208
  def map_prediction_to_ui(pred):
209
  import json, re
210
+ from datetime import datetime
211
 
 
212
  def safe_json_load(s):
213
  if s is None:
214
  return None
215
  if isinstance(s, (dict, list)):
216
  return s
217
  if isinstance(s, str):
 
 
 
218
  try:
219
  return json.loads(s)
220
  except Exception:
221
+ try:
222
+ t = s.strip()
223
+ t = t.replace("\\'", "'").replace('\"{', '{').replace('}\"', '}')
224
+ return json.loads(t)
225
+ except Exception:
226
+ return None
 
 
 
 
 
 
 
 
 
 
 
 
 
 
227
  return None
228
 
 
229
  def clean_number(x):
230
  if x is None:
231
  return 0.0
 
234
  s = str(x).strip()
235
  if s == "":
236
  return 0.0
 
237
  s = re.sub(r"[,\s]", "", s)
238
  s = re.sub(r"[^\d\.\-]", "", s)
239
  if s in ("", ".", "-", "-."):
 
243
  except Exception:
244
  return 0.0
245
 
246
+ def parse_date(s):
247
+ if not s:
248
+ return ""
249
+ s = str(s).strip()
250
+ for fmt in ("%Y-%m-%d", "%d-%m-%Y", "%d/%m/%Y", "%m/%d/%Y", "%d.%m.%Y"):
251
+ try:
252
+ return datetime.strptime(s, fmt).strftime("%Y-%m-%d")
253
+ except Exception:
254
+ pass
255
+ m = re.match(r"^(\d{1,2})/(\d{1,2})/(\d{4})$", s)
256
+ if m:
257
+ a, b, y = int(m.group(1)), int(m.group(2)), int(m.group(3))
258
+ if a > 12:
259
+ d, mo = a, b
260
+ else:
261
+ mo, d = a, b
262
+ try:
263
+ return datetime(year=y, month=mo, day=d).strftime("%Y-%m-%d")
264
+ except Exception:
265
+ return s
266
+ return s
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
267
 
 
268
  ui = {
269
  "Invoice Number": "",
270
  "Invoice Date": "",
 
284
  "Itemized Data": []
285
  }
286
 
287
+ if pred is None:
288
+ return ui
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
289
 
290
+ if isinstance(pred, str):
291
+ parsed = safe_json_load(pred)
292
+ if parsed is not None:
293
+ pred = parsed
 
 
 
 
294
 
295
+ gt = None
296
+ if isinstance(pred, dict):
297
+ if "gt_parse" in pred:
298
+ gp = pred["gt_parse"]
299
+ gp_parsed = safe_json_load(gp)
300
+ gt = gp_parsed if gp_parsed is not None else (gp if isinstance(gp, dict) else {})
301
+ else:
302
+ gt = pred
303
+ else:
304
+ return ui
305
+
306
+ header = gt.get("header") or {}
307
+ items = gt.get("items") or []
308
+ summary = gt.get("summary") or {}
309
+
310
+ ui["Invoice Number"] = header.get("invoice_no") or header.get("invoice_number") or ui["Invoice Number"]
311
+ ui["Invoice Date"] = str(header.get("invoice_date") or header.get("inv_date") or "")
312
+ ui["Due Date"] = str(header.get("due_date") or header.get("due") or "")
313
+
314
+ ui["Sender Name"] = header.get("sender_name") or header.get("seller_name") or header.get("from_name") or ui["Sender Name"]
315
+ ui["Sender Address"] = header.get("sender_addr") or header.get("sender_address") or header.get("seller_addr") or ui["Sender Address"]
316
+ ui["Recipient Name"] = header.get("rcpt_name") or header.get("recipient_name") or header.get("to_name") or ui["Recipient Name"]
317
+ ui["Recipient Address"] = header.get("rcpt_addr") or header.get("rcpt_address") or header.get("recipient_address") or ui["Recipient Address"]
318
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
319
  ui["Sender"] = {"Name": ui["Sender Name"], "Address": ui["Sender Address"]}
320
  ui["Recipient"] = {"Name": ui["Recipient Name"], "Address": ui["Recipient Address"]}
321
 
322
+ bank = {}
323
+ if header.get("bank_name"):
324
+ bank["bank_name"] = str(header.get("bank_name")).strip()
325
+ if header.get("bank_acc_no"):
326
+ bank["bank_account_number"] = str(header.get("bank_acc_no")).strip()
327
+ if header.get("bank_account_number"):
328
+ bank["bank_account_number"] = bank.get("bank_account_number") or str(header.get("bank_account_number")).strip()
329
+ if header.get("bank_iban"):
330
+ bank["bank_iban"] = str(header.get("bank_iban")).strip()
331
+ if header.get("bank_routing"):
332
+ bank["bank_routing"] = str(header.get("bank_routing")).strip()
333
+ if header.get("bank_swift"):
334
+ bank["bank_swift"] = str(header.get("bank_swift")).strip()
335
+ if header.get("bank_branch"):
336
+ bank["bank_branch"] = str(header.get("bank_branch")).strip()
337
+ if header.get("bank_acc_name"):
338
+ bank["bank_acc_name"] = str(header.get("bank_acc_name")).strip()
339
+ hb = header.get("bank")
340
+ if isinstance(hb, dict):
341
+ for k, v in hb.items():
342
+ if not v:
343
+ continue
344
+ lk = k.lower()
345
+ if "iban" in lk:
346
+ bank["bank_iban"] = bank.get("bank_iban") or str(v).strip()
347
+ elif "swift" in lk:
348
+ bank["bank_swift"] = bank.get("bank_swift") or str(v).strip()
349
+ elif "acc" in lk or "account" in lk:
350
+ bank["bank_account_number"] = bank.get("bank_account_number") or str(v).strip()
351
+ elif "name" in lk and "bank" in lk:
352
+ bank["bank_name"] = bank.get("bank_name") or str(v).strip()
353
+ elif "branch" in lk:
354
+ bank["bank_branch"] = bank.get("bank_branch") or str(v).strip()
355
+ elif "acc_name" in lk or "account_name" in lk:
356
+ bank["bank_acc_name"] = bank.get("bank_acc_name") or str(v).strip()
357
+
358
+ ui["Bank Details"] = bank
359
+
360
+ ui["Subtotal"] = clean_number(summary.get("subtotal") or summary.get("sub_total") or summary.get("subTotal"))
361
+ ui["Tax Percentage"] = clean_number(summary.get("tax_rate") or summary.get("taxRate") or summary.get("tax_percentage"))
362
+ ui["Total Tax"] = clean_number(summary.get("tax_amount") or summary.get("tax") or summary.get("taxAmount"))
363
+ ui["Total Amount"] = clean_number(summary.get("total_amount") or summary.get("grand_total") or summary.get("total") or summary.get("amount_total"))
364
+ ui["Currency"] = summary.get("currency") or header.get("currency") or ui["Currency"] or ""
365
+
366
+ normalized_items = []
367
+
368
+ if isinstance(items, str):
369
+ parsed_items = safe_json_load(items)
370
+ if parsed_items is not None:
371
+ items = parsed_items
372
+
373
+ if isinstance(items, dict):
374
+ if any(isinstance(v, list) for v in items.values()):
375
+ list_cols = {k: v for k, v in items.items() if isinstance(v, list)}
376
+ max_len = max((len(v) for v in list_cols.values()), default=0)
377
+ for i in range(max_len):
378
+ row = {}
379
+ for k, v in items.items():
380
+ if isinstance(v, list):
381
+ row[k] = v[i] if i < len(v) else ""
382
+ else:
383
+ row[k] = v
384
+ normalized_items.append(row)
385
+ else:
386
+ normalized_items.append(items)
387
+ elif isinstance(items, list):
388
+ normalized_items = items
389
+ else:
390
+ normalized_items = []
391
+
392
+ item_rows = []
393
+ for it in normalized_items:
394
+ if not isinstance(it, dict):
395
+ item_rows.append({"Description": str(it), "Quantity": 1, "Unit Price": 0.0, "Amount": 0.0, "Tax": 0.0, "Line Total": 0.0})
396
+ continue
397
+ desc = it.get("descriptions") or it.get("description") or it.get("desc") or it.get("item") or it.get("name") or ""
398
+ qty = it.get("quantity") or it.get("qty") or it.get("Quantity") or ""
399
+ unit = it.get("unit_price") or it.get("unitPrice") or it.get("price") or ""
400
+ amt = it.get("amount") or it.get("Line_total") or it.get("line_total") or it.get("total") or ""
401
+
402
+ # Extract item-level tax if available under common keys
403
+ tax_val = it.get("tax") or it.get("tax_amount") or it.get("line_tax") or it.get("item_tax") or it.get("taxAmount") or ""
404
+
405
+ # Extract explicit line total if present; otherwise fall back to amount
406
+ line_total_val = it.get("Line_total") or it.get("line_total") or it.get("lineTotal") or amt
407
+
408
+ item_rows.append({
409
+ "Description": str(desc).strip(),
410
+ "Quantity": float(clean_number(qty)),
411
+ "Unit Price": float(clean_number(unit)),
412
+ "Amount": float(clean_number(amt)),
413
+ "Tax": float(clean_number(tax_val)),
414
+ "Line Total": float(clean_number(line_total_val))
415
+ })
416
+
417
+ ui["Itemized Data"] = item_rows
418
+
419
  return ui
420
 
421
  # ---------------------------
 
425
  """
426
  Converts nested invoice data into a flat list of rows (one per line item),
427
  with invoice-level and sender/recipient/bank fields repeated in each row.
 
428
  """
 
 
 
 
 
 
 
 
 
 
 
429
  rows = []
430
  line_items = invoice_data.get("Itemized Data", [])
431
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
432
  if not line_items:
433
  # If no line items, create one row with invoice info only
434
+ row = {
435
+ "Invoice Number": invoice_data.get("Invoice Number", ""),
436
+ "Invoice Date": invoice_data.get("Invoice Date", ""),
437
+ "Due Date": invoice_data.get("Due Date", ""),
438
+ "Currency": invoice_data.get("Currency", ""),
439
+ "Subtotal": invoice_data.get("Subtotal", 0.0),
440
+ "Tax Percentage": invoice_data.get("Tax Percentage", 0.0),
441
+ "Total Tax": invoice_data.get("Total Tax", 0.0),
442
+ "Total Amount": invoice_data.get("Total Amount", 0.0),
443
+ "Sender Name": invoice_data.get("Sender", {}).get("Name", ""),
444
+ "Sender Address": invoice_data.get("Sender", {}).get("Address", ""),
445
+ "Recipient Name": invoice_data.get("Recipient", {}).get("Name", ""),
446
+ "Recipient Address": invoice_data.get("Recipient", {}).get("Address", ""),
447
+ }
448
+
449
+ # Flatten bank details
450
+ bank = invoice_data.get("Bank Details", {})
451
+ for k, v in bank.items():
452
+ # Avoid double-prefixing if key already contains 'bank_'
453
+ key_name = k if str(k).startswith("bank_") else f"bank_{k}"
454
+ row[key_name] = v
455
+
456
+ # Add empty line item fields
457
  row.update({
458
  "Item Description": "",
459
  "Item Quantity": 0,
 
467
 
468
  # For each line item, create a row with all invoice context
469
  for item in line_items:
470
+ row = {
471
+ "Invoice Number": invoice_data.get("Invoice Number", ""),
472
+ "Invoice Date": invoice_data.get("Invoice Date", ""),
473
+ "Due Date": invoice_data.get("Due Date", ""),
474
+ "Currency": invoice_data.get("Currency", ""),
475
+ "Subtotal": invoice_data.get("Subtotal", 0.0),
476
+ "Tax Percentage": invoice_data.get("Tax Percentage", 0.0),
477
+ "Total Tax": invoice_data.get("Total Tax", 0.0),
478
+ "Total Amount": invoice_data.get("Total Amount", 0.0),
479
+ "Sender Name": invoice_data.get("Sender", {}).get("Name", ""),
480
+ "Sender Address": invoice_data.get("Sender", {}).get("Address", ""),
481
+ "Recipient Name": invoice_data.get("Recipient", {}).get("Name", ""),
482
+ "Recipient Address": invoice_data.get("Recipient", {}).get("Address", ""),
483
+ }
484
+
485
+ # Flatten bank details
486
+ bank = invoice_data.get("Bank Details", {})
487
+ for k, v in bank.items():
488
+ # Avoid double-prefixing if key already contains 'bank_'
489
+ key_name = k if str(k).startswith("bank_") else f"bank_{k}"
490
+ row[key_name] = v
491
+
492
+ # Add line item fields
493
  row.update({
494
  "Item Description": item.get("Description", ""),
495
  "Item Quantity": item.get("Quantity", 0),
 
498
  "Item Tax": item.get("Tax", 0.0),
499
  "Item Line Total": item.get("Line Total", item.get("Amount", 0.0)),
500
  })
501
+
502
  rows.append(row)
503
 
504
  return rows
 
609
  # RESULTS VIEW — Show selector + editable form
610
  # ---------------------------
611
  elif len(st.session_state.batch_results) > 0:
612
+ # ---------------------------
613
+ # Global Download All — produce a single Excel file (concatenated rows) and trigger direct download
614
+ # ---------------------------
615
  if st.button("📦 Download All Results (Excel)", key="download_all"):
616
+ # Collect rows from all invoices and concatenate into one DataFrame
617
  all_rows = []
618
  for file_hash, result in st.session_state.batch_results.items():
619
  rows = flatten_invoice_to_rows(result["edited_data"])
620
+ # Annotate rows with source file name so user can identify which invoice each row came from
621
  for r in rows:
622
  r["Source File"] = result.get("file_name", file_hash)
623
  all_rows.extend(rows)
 
627
  else:
628
  full_df = pd.DataFrame(all_rows)
629
 
630
+ # Reorder columns to put Source File first
631
  cols = list(full_df.columns)
632
  if "Source File" in cols:
633
  cols = ["Source File"] + [c for c in cols if c != "Source File"]
634
  full_df = full_df[cols]
635
 
636
+ # Try to write XLSX (preferred). If engine not available, fall back to CSV.
637
  buffer = BytesIO()
638
  dl_filename = "all_extracted_invoices.xlsx"
639
  tried_xlsx = False
 
654
  dl_filename = "all_extracted_invoices.csv"
655
  mime = "text/csv"
656
 
657
+ # Trigger immediate download via a data URI and small HTML snippet
658
  import base64
659
  import streamlit.components.v1 as components
660
  b64 = base64.b64encode(file_bytes).decode()
 
696
  # Get current file data
697
  current = st.session_state.batch_results[selected_hash]
698
  image = current["image"]
699
+ import copy
700
+ form_data = copy.deepcopy(current["edited_data"])
701
 
702
  # Layout
703
  left_col, right_col = st.columns([1, 1])
 
734
  st.rerun()
735
  except Exception as e:
736
  st.error(f"Re-run failed: {e}")
 
737
  tabs = st.tabs(["Invoice Details", "Sender/Recipient info", "Bank Details", "Line Items"])
738
 
739
  st.markdown(
 
762
  # ---------- Invoice Details ----------
763
  with tabs[0]:
764
  with st.container():
765
+ form_data['Invoice Number'] = st.text_input("Invoice Number", value=form_data.get('Invoice Number', ''), key=f"invoice_number_{selected_hash}")
766
+ form_data['Invoice Date'] = st.text_input("Invoice Date", value=str(form_data.get('Invoice Date', '')).strip(), key=f"invoice_date_text_{selected_hash}")
767
+ form_data['Due Date'] = st.text_input("Due Date", value=str(form_data.get('Due Date', '')).strip(), key=f"due_date_text_{selected_hash}")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
768
  curr_options = ['USD', 'EUR', 'GBP', 'INR', 'Other']
769
+ curr_value = form_data.get('Currency', 'USD')
770
  curr_index = curr_options.index(curr_value) if curr_value in curr_options else (len(curr_options) - 1)
771
+ new_curr = st.selectbox("Currency", options=curr_options, index=curr_index, key=f"currency_select_{selected_hash}")
 
 
 
 
 
 
 
772
  if new_curr == 'Other':
773
+ new_curr = st.text_input("Specify Currency", value=form_data.get('Currency', ''), key=f"custom_currency_{selected_hash}")
774
+ form_data['Currency'] = new_curr
775
+ form_data['Subtotal'] = safe_number_input("Subtotal", form_data.get('Subtotal', 0.0), f"subtotal_{selected_hash}")
776
+ form_data['Tax Percentage'] = safe_number_input("Tax Percentage", form_data.get('Tax Percentage', 0.0), f"tax_pct_{selected_hash}")
777
+ form_data['Total Tax'] = safe_number_input("Total Tax", form_data.get('Total Tax', 0.0), f"total_tax_{selected_hash}")
778
+ form_data['Total Amount'] = safe_number_input("Total Amount", form_data.get('Total Amount', 0.0), f"total_amount_{selected_hash}")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
779
 
780
  # ---------- Sender / Recipient ----------
781
  with tabs[1]:
782
+ sender_info = form_data.get('Sender', {'Name': '', 'Address': ''})
783
+ recipient_info = form_data.get('Recipient', {'Name': '', 'Address': ''})
 
 
 
 
 
 
784
  with st.container():
785
+ sender_info['Name'] = st.text_input("Sender Name*", value=sender_info.get('Name', ''), key=f"sender_name_{selected_hash}")
786
+ sender_info['Address'] = st.text_area("Sender Address*", value=sender_info.get('Address', ''), key=f"sender_address_{selected_hash}")
787
+ recipient_info['Name'] = st.text_input("Recipient Name*", value=recipient_info.get('Name', ''), key=f"recipient_name_{selected_hash}")
788
+ recipient_info['Address'] = st.text_area("Recipient Address*", value=recipient_info.get('Address', ''), key=f"recipient_address_{selected_hash}")
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
789
  if st.button("⇄ Swap", help="Swap sender and recipient information", key=f"swap_{selected_hash}"):
790
+ form_data['Sender'], form_data['Recipient'] = form_data['Recipient'], form_data['Sender']
 
791
  st.rerun()
792
 
793
  # ---------- Bank Details ----------
794
  with tabs[2]:
795
+ bank_info = form_data.get('Bank Details', {}) or {}
796
  with st.container():
797
+ bank_info['bank_name'] = st.text_input("Bank Name", value=bank_info.get('bank_name', ''), key=f"bank_name_{selected_hash}")
798
+ bank_info['bank_account_number'] = st.text_input("Account Number", value=bank_info.get('bank_account_number', '') or bank_info.get('bank_acc_no',''), key=f"bank_account_{selected_hash}")
799
+ bank_info['bank_acc_name'] = st.text_input("Bank Account Name", value=bank_info.get('bank_acc_name', '') or bank_info.get('bank_acc_name', ''), key=f"bank_acc_name_{selected_hash}")
800
+ bank_info['bank_iban'] = st.text_input("IBAN", value=bank_info.get('bank_iban', ''), key=f"iban_{selected_hash}")
801
+ bank_info['bank_swift'] = st.text_input("SWIFT Code", value=bank_info.get('bank_swift', ''), key=f"swift_code_{selected_hash}")
802
+ bank_info['bank_routing'] = st.text_input("Routing Number", value=bank_info.get('bank_routing', ''), key=f"routing_{selected_hash}")
803
+ bank_info['bank_branch'] = st.text_input("Branch", value=bank_info.get('bank_branch', ''), key=f"branch_{selected_hash}")
804
+ form_data['Bank Details'] = bank_info
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
805
 
806
  # ---------- Line Items ----------
807
  with tabs[3]:
808
  editor_key = f"item_editor_{selected_hash}"
809
+ item_rows = form_data.get('Itemized Data', [])
 
810
  df = pd.DataFrame(item_rows)
811
  for col in ["Description", "Quantity", "Unit Price", "Amount", "Tax", "Line Total"]:
812
  if col not in df.columns:
813
  df[col] = ""
814
+ st.write("✏️ Edit line items below. Press Enter or click outside a cell to confirm each edit.")
815
+ edited_df = st.data_editor(
 
816
  df,
817
  num_rows="dynamic",
818
  key=editor_key,
819
  use_container_width=True,
 
 
820
  )
821
+ if len(edited_df) == 0:
 
822
  st.info("No line items found in the invoice.")
823
 
824
+ # Save button (per file)
825
+ if st.button("💾 Save Edits for This File", key=f"save_{selected_hash}"):
826
+ # Update line items from data editor
827
+ form_data['Itemized Data'] = edited_df.to_dict('records')
828
+ # Update session state with complete form data
829
+ st.session_state.batch_results[selected_hash]["edited_data"] = form_data
830
+ st.success(f"✅ Edits saved for {current['file_name']}")
831
 
832
  # Download buttons (per file)
833
  st.markdown("---")
834
  col_a, col_b, col_c = st.columns([1, 1, 1])
835
+ #with col_a:
836
+ #jsonl_str = json.dumps(data, ensure_ascii=False, indent=2)
837
+ #st.download_button(
838
+ # "📥 Download JSON",
839
+ #jsonl_str.encode("utf-8"),
840
+ #file_name=f"{Path(current['file_name']).stem}_extracted.json",
841
+ #mime="application/json",
842
+ #key=f"dl_json_{selected_hash}"
843
+ #)
844
  with col_b:
845
  # ✅ Flatten entire invoice into rows (one per line item)
846
+ rows = flatten_invoice_to_rows(form_data)
847
  full_df = pd.DataFrame(rows)
848
 
849
  # Optional: Reorder columns for better readability
 
870
  mime="text/csv",
871
  key=f"dl_csv_{selected_hash}"
872
  )
873
+ # Global Download All — produce a single Excel file (concatenated rows) and trigger direct download
874
+
875
+
876
+ # ---------------------------
877
+ # PROCESSING STATE
878
 
879
  # ---------------------------
880
  # PROCESSING STATE — Show progress