diff --git a/app/logic/document_outputs.py b/app/logic/document_outputs.py
index f7653a8..7d2181c 100644
--- a/app/logic/document_outputs.py
+++ b/app/logic/document_outputs.py
@@ -72,6 +72,10 @@ from sqlalchemy.orm import Session
from app.core.config import FIELD_ENRICHED_ROOT, OCR_CORRECTED_ROOT
from app.models.document import Document
from app.models.document_version import DocumentVersion
+
+from app.models.document_replica_layout_version import DocumentReplicaLayoutVersion
+from app.models.document_replica_output import DocumentReplicaOutput
+from app.models.document_replica_review_state import DocumentReplicaReviewState
from app.models.text_version import TextVersion
@@ -685,19 +689,10 @@ def save_ocr_corrected_pdf_current(db: Session, document: Document, output_path:
except Exception:
share_path_value = None
- document.share_path = share_path_value
- document.current_path = str(out_path)
- document.canonical_filename = out_path.name
- document.sha256_current = file_hash
- db.add(document)
-
+ # Replica outputs are non-destructive exports for now.
+ # Do not replace the primary/current document path.
db.commit()
- keep_paths = {str(out_path)}
- if document.share_path:
- keep_paths.add(str(document.share_path))
- _prune_old_saved_files(db, document, keep_paths)
-
def save_field_enriched_pdf_current(db: Session, document: Document, output_path: Path) -> None:
if not document.current_path:
@@ -736,3 +731,263 @@ def save_field_enriched_pdf_current(db: Session, document: Document, output_path
if document.share_path:
keep_paths.add(str(document.share_path))
_prune_old_saved_files(db, document, keep_paths)
+
+
+def _next_replica_layout_version_number(db: Session, document_id: int) -> int:
+ return (
+ db.query(func.max(DocumentReplicaLayoutVersion.version_number))
+ .filter(DocumentReplicaLayoutVersion.document_id == document_id)
+ .scalar()
+ or 0
+ ) + 1
+
+
+def _get_current_replica_review_state(document: Document) -> DocumentReplicaReviewState | None:
+ rows = getattr(document, "replica_review_states", None) or []
+ return rows[0] if rows else None
+
+
+def _get_replica_source_context(document: Document):
+ if not document.current_path:
+ raise ValueError("Document has no current_path")
+
+ current_file = Path(document.current_path)
+ if not current_file.exists():
+ raise FileNotFoundError(f"Current file not found: {current_file}")
+
+ raw_ocr = _latest_current_text_version(document, "raw_ocr")
+ reviewed = _latest_current_text_version(document, "reviewed")
+
+ if raw_ocr is None:
+ raise ValueError("No current raw OCR version found")
+ if reviewed is None:
+ raise ValueError("No current reviewed text found")
+ if current_file.suffix.lower() != ".pdf":
+ raise ValueError("Replica PDF generation currently supports PDFs only")
+
+ raw_lines = _flatten_layout_lines(raw_ocr.layout_json)
+ reviewed_lines = _flatten_layout_lines(reviewed.layout_json)
+
+ if not raw_lines:
+ raise ValueError("No OCR line boxes found in raw OCR layout data")
+ if reviewed_lines and len(reviewed_lines) != len(raw_lines):
+ raise ValueError("Reviewed line layout does not match raw OCR line layout")
+
+ source_layout = reviewed.layout_json if reviewed.layout_json else raw_ocr.layout_json
+ if not source_layout:
+ raise ValueError("No source layout found")
+
+ return current_file, raw_ocr, reviewed, source_layout
+
+
+def build_replica_layout(document: Document, mode: str = "shared") -> dict:
+ current_file, raw_ocr, reviewed, source_layout = _get_replica_source_context(document)
+ reader = PdfReader(str(current_file))
+
+ pages = []
+ page_layouts = {page["page"]: page for page in source_layout.get("pages", [])}
+
+ for page_num, pdf_page in enumerate(reader.pages, start=1):
+ page_w = float(pdf_page.mediabox.width)
+ page_h = float(pdf_page.mediabox.height)
+ page_layout = page_layouts.get(page_num, {"lines": []})
+ src_w = float(page_layout.get("image_width") or 1.0)
+ src_h = float(page_layout.get("image_height") or 1.0)
+ scale_x = page_w / src_w
+ scale_y = page_h / src_h
+
+ line_entries = []
+ for line in page_layout.get("lines", []):
+ text_line = (line.get("text") or "").strip()
+ if not text_line:
+ continue
+
+ left, top, right, bottom = line["bbox"]
+ pdf_x = left * scale_x
+ pdf_y = page_h - (bottom * scale_y)
+ box_width = max(10.0, (right - left) * scale_x)
+ box_height = max(6.0, (bottom - top) * scale_y)
+ font_size = _fit_font_size(text_line, box_width, box_height)
+
+ line_entries.append(
+ {
+ "text": text_line,
+ "bbox_source": [left, top, right, bottom],
+ "pdf_x": pdf_x,
+ "pdf_y": pdf_y,
+ "box_width": box_width,
+ "box_height": box_height,
+ "font_family_guess": "Helvetica",
+ "font_size_guess": font_size,
+ "text_color_guess": "#000000",
+ "text_render_mode_clean": 0,
+ "text_render_mode_scan_backed": 3,
+ }
+ )
+
+ pages.append(
+ {
+ "page": page_num,
+ "page_width": page_w,
+ "page_height": page_h,
+ "image_width": src_w,
+ "image_height": src_h,
+ "lines": line_entries,
+ }
+ )
+
+ return {
+ "schema_version": 1,
+ "mode_source": mode,
+ "current_path": str(current_file),
+ "text_version_source": {
+ "raw_ocr_version_id": raw_ocr.id if raw_ocr else None,
+ "reviewed_version_id": reviewed.id if reviewed else None,
+ },
+ "pages": pages,
+ }
+
+
+def _save_replica_layout_version(
+ db: Session,
+ document: Document,
+ layout_json: dict,
+ mode: str,
+ created_by: str = "save_replica_pdf",
+) -> DocumentReplicaLayoutVersion:
+ db.query(DocumentReplicaLayoutVersion).filter(
+ DocumentReplicaLayoutVersion.document_id == document.id,
+ DocumentReplicaLayoutVersion.is_current == True, # noqa: E712
+ ).update({"is_current": False}, synchronize_session=False)
+
+ version = DocumentReplicaLayoutVersion(
+ document_id=document.id,
+ version_number=_next_replica_layout_version_number(db, document.id),
+ version_type="heuristic",
+ render_mode_source=mode,
+ is_current=True,
+ created_by=created_by,
+ quality_flags=[],
+ inference_metadata_json={"pipeline": "heuristic_replica_v1", "mode": mode},
+ layout_json=layout_json,
+ )
+ db.add(version)
+ db.flush()
+
+ state = _get_current_replica_review_state(document)
+ if state is None:
+ state = DocumentReplicaReviewState(document_id=document.id)
+ db.add(state)
+
+ state.current_replica_layout_version_id = version.id
+ state.is_reviewed = False
+ state.is_approved = False
+ state.needs_manual_adjustment = False
+ state.needs_model_retry = False
+ db.flush()
+
+ return version
+
+
+def _render_replica_pdf_from_layout(
+ current_file: Path,
+ layout_json: dict,
+ out_path: Path,
+ mode: str,
+) -> None:
+ reader = PdfReader(str(current_file))
+ out_path.parent.mkdir(parents=True, exist_ok=True)
+
+ with tempfile.TemporaryDirectory() as tmpdirname:
+ tmpdir = Path(tmpdirname)
+ images = _render_pdf_page_images(current_file, tmpdir)
+ overlay_pdf_path = tmpdir / "replica.pdf"
+ c = None
+
+ pages = {page["page"]: page for page in layout_json.get("pages", [])}
+
+ for page_num, img_path in enumerate(images, start=1):
+ pdf_page = reader.pages[page_num - 1]
+ page_w = float(pdf_page.mediabox.width)
+ page_h = float(pdf_page.mediabox.height)
+
+ if c is None:
+ c = canvas.Canvas(str(overlay_pdf_path), pagesize=(page_w, page_h))
+ else:
+ c.setPageSize((page_w, page_h))
+
+ if mode == "scan_backed":
+ c.drawImage(ImageReader(str(img_path)), 0, 0, width=page_w, height=page_h)
+
+ page_layout = pages.get(page_num, {"lines": []})
+
+ for line in page_layout.get("lines", []):
+ text_line = (line.get("text") or "").strip()
+ if not text_line:
+ continue
+
+ text_obj = c.beginText()
+ text_obj.setTextRenderMode(3 if mode == "scan_backed" else 0)
+ text_obj.setFont(line.get("font_family_guess") or "Helvetica", float(line.get("font_size_guess") or 10))
+ text_obj.setTextOrigin(float(line["pdf_x"]), float(line["pdf_y"]) + 1)
+ text_obj.textLine(text_line)
+ c.drawText(text_obj)
+
+ c.showPage()
+
+ if c is None:
+ raise ValueError("Failed to build replica PDF")
+
+ c.save()
+ shutil.copy2(overlay_pdf_path, out_path)
+
+ compress_pdf_with_ghostscript(out_path)
+
+
+def save_replica_pdf(db: Session, document: Document, output_path: Path, mode: str) -> None:
+ if mode not in {"clean", "scan_backed"}:
+ raise ValueError(f"Unsupported replica mode: {mode}")
+
+ current_file, _, _, _ = _get_replica_source_context(document)
+ out_path = Path(output_path)
+ out_path = out_path.with_name(re.sub(r"_v\d+(?=\.[^.]+$)", "", out_path.name))
+
+ stem = re.sub(r"(_replica_clean|_replica_scan_backed)$", "", out_path.stem)
+ suffix = out_path.suffix or ".pdf"
+
+ if mode == "clean":
+ out_path = out_path.with_name(f"{stem}_replica_clean{suffix}")
+ else:
+ out_path = out_path.with_name(f"{stem}_replica_scan_backed{suffix}")
+
+ out_path.parent.mkdir(parents=True, exist_ok=True)
+
+ layout_json = build_replica_layout(document, mode=mode)
+ layout_version = _save_replica_layout_version(db, document, layout_json, mode=mode)
+
+ _render_replica_pdf_from_layout(current_file, layout_json, out_path, mode=mode)
+
+ file_hash = sha256_for_file(out_path)
+ file_size = out_path.stat().st_size
+
+ try:
+ mirror_path = _mirror_to_secondary_owner(document, out_path)
+ share_path_value = str(mirror_path) if mirror_path else None
+ except Exception:
+ share_path_value = None
+
+ output = DocumentReplicaOutput(
+ document_id=document.id,
+ replica_layout_version_id=layout_version.id,
+ output_type=mode,
+ file_path=str(out_path),
+ sha256=file_hash,
+ file_size_bytes=file_size,
+ created_by="save_replica_pdf",
+ render_settings_json={"mode": mode},
+ )
+ db.add(output)
+
+ # Replica outputs are non-destructive exports.
+ # Do not replace the primary/current document path or prune sibling files.
+ db.commit()
diff --git a/app/models/__init__.py b/app/models/__init__.py
index 6a09210..fd4dbad 100644
--- a/app/models/__init__.py
+++ b/app/models/__init__.py
@@ -18,3 +18,6 @@ __all__ = [
"DocumentPreset",
]
from app.models.document_naming_field import DocumentNamingField
+from app.models.document_replica_layout_version import DocumentReplicaLayoutVersion
+from app.models.document_replica_output import DocumentReplicaOutput
+from app.models.document_replica_review_state import DocumentReplicaReviewState
diff --git a/app/models/document.py b/app/models/document.py
index 1389ed5..f16db3a 100644
--- a/app/models/document.py
+++ b/app/models/document.py
@@ -105,3 +105,17 @@ class Document(Base):
cascade="all, delete-orphan",
uselist=False,
)
+
+ replica_layout_versions: Mapped[list["DocumentReplicaLayoutVersion"]] = relationship(
+ back_populates="document",
+ cascade="all, delete-orphan",
+ order_by="DocumentReplicaLayoutVersion.version_number",
+ )
+ replica_outputs: Mapped[list["DocumentReplicaOutput"]] = relationship(
+ back_populates="document",
+ cascade="all, delete-orphan",
+ )
+ replica_review_states: Mapped[list["DocumentReplicaReviewState"]] = relationship(
+ back_populates="document",
+ cascade="all, delete-orphan",
+ )
diff --git a/app/models/document_replica_layout_version.py b/app/models/document_replica_layout_version.py
new file mode 100644
index 0000000..083b2a8
--- /dev/null
+++ b/app/models/document_replica_layout_version.py
@@ -0,0 +1,37 @@
+from sqlalchemy import Boolean, Column, DateTime, ForeignKey, Integer, JSON, String, Text
+from sqlalchemy.orm import relationship
+from sqlalchemy.sql import func
+
+from app.db.base import Base
+
+
+class DocumentReplicaLayoutVersion(Base):
+ __tablename__ = "document_replica_layout_versions"
+
+ id = Column(Integer, primary_key=True, index=True)
+ document_id = Column(Integer, ForeignKey("documents.id"), nullable=False, index=True)
+
+ version_number = Column(Integer, nullable=False)
+ version_type = Column(String, nullable=False, default="heuristic")
+ render_mode_source = Column(String, nullable=False, default="shared")
+ is_current = Column(Boolean, nullable=False, default=True)
+
+ created_by = Column(String, nullable=True)
+ derived_from_text_version_id = Column(Integer, ForeignKey("text_versions.id"), nullable=True)
+ derived_from_replica_layout_version_id = Column(Integer, ForeignKey("document_replica_layout_versions.id"), nullable=True)
+
+ model_name = Column(String, nullable=True)
+ model_version = Column(String, nullable=True)
+ prompt_version = Column(String, nullable=True)
+
+ quality_score = Column(String, nullable=True)
+ quality_note = Column(Text, nullable=True)
+ quality_flags = Column(JSON, nullable=True)
+ inference_metadata_json = Column(JSON, nullable=True)
+ layout_json = Column(JSON, nullable=False)
+
+ created_at = Column(DateTime(timezone=True), server_default=func.now(), nullable=False)
+
+ document = relationship("Document", back_populates="replica_layout_versions")
+ outputs = relationship("DocumentReplicaOutput", back_populates="replica_layout_version", cascade="all, delete-orphan")
+ parent_layout_version = relationship("DocumentReplicaLayoutVersion", remote_side=[id])
diff --git a/app/models/document_replica_output.py b/app/models/document_replica_output.py
new file mode 100644
index 0000000..64dee69
--- /dev/null
+++ b/app/models/document_replica_output.py
@@ -0,0 +1,25 @@
+from sqlalchemy import Column, DateTime, ForeignKey, Integer, JSON, String
+from sqlalchemy.orm import relationship
+from sqlalchemy.sql import func
+
+from app.db.base import Base
+
+
+class DocumentReplicaOutput(Base):
+ __tablename__ = "document_replica_outputs"
+
+ id = Column(Integer, primary_key=True, index=True)
+ document_id = Column(Integer, ForeignKey("documents.id"), nullable=False, index=True)
+ replica_layout_version_id = Column(Integer, ForeignKey("document_replica_layout_versions.id"), nullable=False, index=True)
+
+ output_type = Column(String, nullable=False)
+ file_path = Column(String, nullable=False)
+ sha256 = Column(String, nullable=True)
+ file_size_bytes = Column(Integer, nullable=True)
+ created_by = Column(String, nullable=True)
+ render_settings_json = Column(JSON, nullable=True)
+
+ created_at = Column(DateTime(timezone=True), server_default=func.now(), nullable=False)
+
+ document = relationship("Document", back_populates="replica_outputs")
+ replica_layout_version = relationship("DocumentReplicaLayoutVersion", back_populates="outputs")
diff --git a/app/models/document_replica_review_state.py b/app/models/document_replica_review_state.py
new file mode 100644
index 0000000..d2a9414
--- /dev/null
+++ b/app/models/document_replica_review_state.py
@@ -0,0 +1,27 @@
+from sqlalchemy import Boolean, Column, DateTime, ForeignKey, Integer, Text
+from sqlalchemy.orm import relationship
+from sqlalchemy.sql import func
+
+from app.db.base import Base
+
+
+class DocumentReplicaReviewState(Base):
+ __tablename__ = "document_replica_review_states"
+
+ id = Column(Integer, primary_key=True, index=True)
+ document_id = Column(Integer, ForeignKey("documents.id"), nullable=False, index=True)
+ current_replica_layout_version_id = Column(Integer, ForeignKey("document_replica_layout_versions.id"), nullable=True)
+
+ is_reviewed = Column(Boolean, nullable=False, default=False)
+ is_approved = Column(Boolean, nullable=False, default=False)
+ needs_model_retry = Column(Boolean, nullable=False, default=False)
+ needs_manual_adjustment = Column(Boolean, nullable=False, default=False)
+
+ reviewed_by = Column(Text, nullable=True)
+ review_note = Column(Text, nullable=True)
+
+ reviewed_at = Column(DateTime(timezone=True), nullable=True)
+ created_at = Column(DateTime(timezone=True), server_default=func.now(), nullable=False)
+ updated_at = Column(DateTime(timezone=True), server_default=func.now(), onupdate=func.now(), nullable=False)
+
+ document = relationship("Document", back_populates="replica_review_states")
diff --git a/app/routes/documents.py b/app/routes/documents.py
index 9a5a114..74b5a33 100644
--- a/app/routes/documents.py
+++ b/app/routes/documents.py
@@ -22,6 +22,7 @@ from app.db.deps import get_db
from app.logic.document_outputs import (
save_field_enriched_pdf_current,
save_ocr_corrected_pdf_current,
+ save_replica_pdf,
)
from app.logic.storage_paths import build_proposed_storage_path
from app.logic.extraction import (
@@ -1003,6 +1004,38 @@ def move_to_trash(document_id: str, db: Session = Depends(get_db)):
+
+def _resolve_document_output_path(document, output_path: str = "") -> Path:
+ save_root = get_default_save_root()
+ naming_row = document.naming_fields[0] if getattr(document, "naming_fields", None) else None
+
+ default_output_path = Path(
+ build_proposed_storage_path(
+ document=document,
+ save_root=save_root,
+ naming_row=naming_row,
+ )
+ )
+ default_output_path = default_output_path.with_name(
+ re.sub(r"(?:_v\d+|_\d+)(?=\.[^.]+$)", "", default_output_path.name)
+ )
+ if default_output_path.suffix.lower() != ".pdf":
+ default_output_path = default_output_path.with_suffix(".pdf")
+
+ output_path_raw = (output_path or "").strip()
+ output_path_obj = Path(output_path_raw) if output_path_raw else default_output_path
+
+ if output_path_obj.suffix.lower() != ".pdf":
+ output_path_obj = output_path_obj.with_suffix(".pdf")
+
+ allowed_root = Path(save_root).resolve()
+ resolved_parent = output_path_obj.parent.resolve()
+ if allowed_root != resolved_parent and allowed_root not in resolved_parent.parents:
+ raise ValueError("invalid_output_path")
+
+ output_path_obj.parent.mkdir(parents=True, exist_ok=True)
+ return output_path_obj
+
@router.post("/{document_id}/save-pdf", response_class=RedirectResponse)
def save_pdf(document_id: str, output_path: str = Form(""), db: Session = Depends(get_db)):
if not _storage_available():
@@ -1024,41 +1057,14 @@ def save_pdf(document_id: str, output_path: str = Form(""), db: Session = Depend
if document is None:
return RedirectResponse(url="/documents/", status_code=303)
- save_root = get_default_save_root()
- naming_row = document.naming_fields[0] if getattr(document, "naming_fields", None) else None
-
- default_output_path = Path(
- build_proposed_storage_path(
- document=document,
- save_root=save_root,
- naming_row=naming_row,
- )
- )
- default_output_path = default_output_path.with_name(
- re.sub(r"(?:_v\d+|_\d+)(?=\.[^.]+$)", "", default_output_path.name)
- )
- if default_output_path.suffix.lower() != ".pdf":
- default_output_path = default_output_path.with_suffix(".pdf")
-
- output_path_raw = (output_path or "").strip()
- if output_path_raw:
- output_path_obj = Path(output_path_raw)
- else:
- output_path_obj = default_output_path
-
- if output_path_obj.suffix.lower() != ".pdf":
- output_path_obj = output_path_obj.with_suffix(".pdf")
-
- allowed_root = Path(save_root).resolve()
- resolved_parent = output_path_obj.parent.resolve()
- if allowed_root != resolved_parent and allowed_root not in resolved_parent.parents:
+ try:
+ output_path_obj = _resolve_document_output_path(document, output_path)
+ except ValueError:
return RedirectResponse(
url=f"/documents/{document.document_id}?error=invalid_output_path",
status_code=303,
)
- output_path_obj.parent.mkdir(parents=True, exist_ok=True)
-
has_extracted = bool(getattr(document, "extracted_fields", None))
has_additional = bool(getattr(document, "additional_fields", None))
@@ -1079,6 +1085,70 @@ def save_pdf(document_id: str, output_path: str = Form(""), db: Session = Depend
+
+@router.post("/{document_id}/save-replica-pdf", response_class=RedirectResponse)
+def save_replica_pdf_clean(document_id: str, output_path: str = Form(""), db: Session = Depends(get_db)):
+ if not _storage_available():
+ return RedirectResponse(url=f"/documents/{document_id}?error=storage_unavailable", status_code=303)
+
+ document = (
+ db.query(Document)
+ .options(
+ selectinload(Document.text_versions),
+ selectinload(Document.naming_fields),
+ selectinload(Document.replica_review_states),
+ )
+ .filter(Document.document_id == document_id)
+ .first()
+ )
+ if document is None:
+ return RedirectResponse(url="/documents/", status_code=303)
+
+ try:
+ output_path_obj = _resolve_document_output_path(document, output_path)
+ save_replica_pdf(db, document, output_path_obj, mode="clean")
+ except ValueError as e:
+ if "invalid_output_path" in str(e):
+ return RedirectResponse(url=f"/documents/{document.document_id}?error=invalid_output_path", status_code=303)
+ return RedirectResponse(url=f"/documents/{document.document_id}?error=save_replica_pdf_failed&tab=ocr-review", status_code=303)
+ except Exception:
+ traceback.print_exc()
+ return RedirectResponse(url=f"/documents/{document.document_id}?error=save_replica_pdf_failed&tab=ocr-review", status_code=303)
+
+ return RedirectResponse(url=f"/documents/{document.document_id}?success=saved_replica_pdf&tab=ocr-review", status_code=303)
+
+
+@router.post("/{document_id}/save-replica-pdf-scan-backed", response_class=RedirectResponse)
+def save_replica_pdf_scan_backed(document_id: str, output_path: str = Form(""), db: Session = Depends(get_db)):
+ if not _storage_available():
+ return RedirectResponse(url=f"/documents/{document_id}?error=storage_unavailable", status_code=303)
+
+ document = (
+ db.query(Document)
+ .options(
+ selectinload(Document.text_versions),
+ selectinload(Document.naming_fields),
+ selectinload(Document.replica_review_states),
+ )
+ .filter(Document.document_id == document_id)
+ .first()
+ )
+ if document is None:
+ return RedirectResponse(url="/documents/", status_code=303)
+
+ try:
+ output_path_obj = _resolve_document_output_path(document, output_path)
+ save_replica_pdf(db, document, output_path_obj, mode="scan_backed")
+ except ValueError as e:
+ if "invalid_output_path" in str(e):
+ return RedirectResponse(url=f"/documents/{document.document_id}?error=invalid_output_path", status_code=303)
+ return RedirectResponse(url=f"/documents/{document.document_id}?error=save_replica_pdf_scan_backed_failed&tab=ocr-review", status_code=303)
+ except Exception:
+ traceback.print_exc()
+ return RedirectResponse(url=f"/documents/{document.document_id}?error=save_replica_pdf_scan_backed_failed&tab=ocr-review", status_code=303)
+
+ return RedirectResponse(url=f"/documents/{document.document_id}?success=saved_replica_pdf_scan_backed&tab=ocr-review", status_code=303)
+
@router.post("/{document_id}/save-field-enriched-pdf", response_class=RedirectResponse)
def save_field_enriched_pdf(document_id: str, db: Session = Depends(get_db)):
return RedirectResponse(
@@ -1459,12 +1529,13 @@ async def save_line_items(
)
@router.get("/{document_id}/preview-file")
-def document_preview_file(document_id: str, db: Session = Depends(get_db)):
+def document_preview_file(document_id: str, path: str | None = None, db: Session = Depends(get_db)):
document = db.query(Document).filter(Document.document_id == document_id).first()
- if document is None or not document.current_path:
+ resolved_path = path or (document.current_path if document else None)
+ if document is None or not resolved_path:
return HTMLResponse(content="Preview file not found", status_code=404)
- path_obj = Path(document.current_path)
+ path_obj = Path(resolved_path)
if not path_obj.exists() or not path_obj.is_file():
return HTMLResponse(content="Preview file not found", status_code=404)
@@ -1472,8 +1543,26 @@ def document_preview_file(document_id: str, db: Session = Depends(get_db)):
return FileResponse(path=str(path_obj), media_type=media_type, filename=path_obj.name, headers={"Content-Disposition": "inline; filename=\"" + path_obj.name + "\""})
+
+def _get_latest_replica_output(document, output_type: str):
+ outputs = getattr(document, "replica_outputs", None) or []
+ matches = [row for row in outputs if getattr(row, "output_type", None) == output_type]
+ matches.sort(key=lambda x: getattr(x, "created_at", None) or 0, reverse=True)
+ return matches[0] if matches else None
+
+
+def _build_preview_url_for_path(request: Request, document_id: str, path_value: str | None):
+ if not path_value:
+ return None
+ path_obj = Path(path_value)
+ if not path_obj.exists() or not path_obj.is_file():
+ return None
+ from urllib.parse import quote
+ base = str(request.url_for("document_preview_file", document_id=document_id))
+ return f"{base}?path={quote(str(path_obj))}&v={int(path_obj.stat().st_mtime)}"
+
@router.get("/{document_id}", response_class=HTMLResponse)
-def document_detail(document_id: str, request: Request, queue: str | None = None, db: Session = Depends(get_db)):
+def document_detail(document_id: str, request: Request, queue: str | None = None, viewer_source: str = "scan", db: Session = Depends(get_db)):
current_user = getattr(request.state, "current_user", None)
document = (
db.query(Document)
@@ -1511,12 +1600,26 @@ def document_detail(document_id: str, request: Request, queue: str | None = None
actual_line_count = len(review_text_value.splitlines()) if review_text_value else 0
line_numbers = list(range(1, max(actual_line_count, expected_line_count) + 1))
- file_url = None
+ replica_clean_output = _get_latest_replica_output(document, "clean")
+ replica_scan_backed_output = _get_latest_replica_output(document, "scan_backed")
+
+ scan_path = document.current_path
+ replica_path = replica_clean_output.file_path if replica_clean_output and replica_clean_output.file_path else None
+ replica_scan_backed_path = replica_scan_backed_output.file_path if replica_scan_backed_output and replica_scan_backed_output.file_path else None
+
+ effective_viewer_source = viewer_source or "scan"
+ preview_path = scan_path
+
+ if effective_viewer_source == "replica" and replica_path:
+ preview_path = replica_path
+ elif effective_viewer_source == "replica_scan_backed" and replica_scan_backed_path:
+ preview_path = replica_scan_backed_path
+ else:
+ effective_viewer_source = "scan"
+ preview_path = scan_path
+
storage_available = _storage_available()
- if document.current_path:
- current_path = Path(document.current_path)
- if current_path.exists() and current_path.is_file():
- file_url = str(request.url_for("document_preview_file", document_id=document.document_id))
+ file_url = _build_preview_url_for_path(request, document.document_id, preview_path)
app_url = str(request.url_for("document_detail", document_id=document.document_id))
error = request.query_params.get("error")
@@ -1615,6 +1718,9 @@ def document_detail(document_id: str, request: Request, queue: str | None = None
"review_text_value": review_text_value,
"file_url": file_url,
"storage_available": storage_available,
+ "viewer_source": effective_viewer_source,
+ "replica_clean_output": replica_clean_output,
+ "replica_scan_backed_output": replica_scan_backed_output,
"version_rows": version_rows,
"current_line_item_version": current_line_item_version,
"ocr_version_options": ocr_version_options,
diff --git a/app/static/app.css b/app/static/app.css
index 7eb6767..0233e74 100644
--- a/app/static/app.css
+++ b/app/static/app.css
@@ -6231,3 +6231,40 @@ table {
}
}
/* ===== end line item queue card polish ===== */
+
+
+
+.preview-card-header {
+ display: flex;
+ align-items: center;
+ justify-content: space-between;
+ gap: 0.75rem;
+ flex-wrap: wrap;
+}
+
+.preview-source-toggle {
+ display: flex;
+ gap: 0.45rem;
+ flex-wrap: wrap;
+}
+
+.preview-source-link {
+ display: inline-flex;
+ align-items: center;
+ justify-content: center;
+ min-height: 2rem;
+ padding: 0.35rem 0.7rem;
+ border: 1px solid #d7dce5;
+ border-radius: 999px;
+ background: #fff;
+ color: #334155;
+ text-decoration: none;
+ font-size: 0.82rem;
+ line-height: 1;
+}
+
+.preview-source-link.active {
+ background: #0f172a;
+ border-color: #0f172a;
+ color: #fff;
+}
diff --git a/app/templates/base.html b/app/templates/base.html
index 9148c80..3882793 100644
--- a/app/templates/base.html
+++ b/app/templates/base.html
@@ -4,7 +4,7 @@
{% block title %}Document Processor{% endblock %}
-
+
diff --git a/app/templates/documents/detail.html b/app/templates/documents/detail.html
index 91bd3fc..810e9b5 100644
--- a/app/templates/documents/detail.html
+++ b/app/templates/documents/detail.html
@@ -73,14 +73,26 @@ document.addEventListener("DOMContentLoaded", () => {
OCR rerun successfully.
{% elif success == "regenerated_line_items" %}
Line items regenerated successfully.
+{% elif success == "saved_replica_pdf" %}
+ Replica PDF saved.
+{% elif success == "saved_replica_pdf_scan_backed" %}
+ Scan-backed replica PDF saved.
{% elif success == "saved_reviewed_ocr" %}
Reviewed OCR saved.
+{% elif success == "saved_replica_pdf" %}
+ Replica PDF saved.
+{% elif success == "saved_replica_pdf_scan_backed" %}
+ Scan-backed replica PDF saved.
{% elif success == "saved_reviewed_ocr" %}
Reviewed OCR saved.
{% elif error == "rerun_ocr_failed" %}
OCR rerun failed.
{% elif error == "deprecated_pdf_route_disabled" %}
This deprecated PDF save route has been disabled. Use Save Document instead.
+ {% elif error == "save_replica_pdf_failed" %}
+ Could not save replica PDF.
+ {% elif error == "save_replica_pdf_scan_backed_failed" %}
+ Could not save scan-backed replica PDF.
{% elif error == "save_field_enriched_failed" %}
Could not save field-enriched PDF.
{% endif %}
@@ -163,6 +175,14 @@ document.addEventListener("DOMContentLoaded", () => {
+
+
+
+
@@ -200,7 +220,20 @@ document.addEventListener("DOMContentLoaded", () => {
-
Document preview
+
{% if not storage_available %}
Storage mount unavailable. Preview is temporarily unavailable.
{% elif file_url %}
diff --git a/run.sh b/run.sh
old mode 100755
new mode 100644