feat: updated font handling in older pdfs, along with tables

gateway/app/routers/documents.py

@@ -294,6 +294,31 @@ def get_font_bytes(document_id: str, internal_font_id: str) -> Response:
         headers={"Cache-Control": "public, max-age=31536000, immutable", "ETag": etag},
     )
 
+@router.get("/{document_id}/font-reconstructed")
+def get_reconstructed_font_bytes(document_id: str, internal_font_id: str) -> Response:
+    """Tier-2: a cmap-augmented copy of an embedded font (original glyph program + synthesized
+    Unicode cmap) so the WASM live preview can reuse the document's real glyphs and match the
+    saved result. 204 when reconstruction isn't possible -> frontend falls back to Tier-1."""
+    if not engine.is_available():
+        return Response(status_code=status.HTTP_204_NO_CONTENT)
+    if not internal_font_id or len(internal_font_id) > 256:
+        return Response(status_code=status.HTTP_204_NO_CONTENT)
+    doc_info = document_store.get_document(document_id)
+    if not doc_info:
+        return Response(status_code=status.HTTP_204_NO_CONTENT)
+    try:
+        data = bytes(doc_info["doc_instance"].get_reconstructed_font_data(internal_font_id))
+    except Exception:
+        data = b""
+    if not data or data[:4] not in _SFNT_TTF_MAGIC:
+        return Response(status_code=status.HTTP_204_NO_CONTENT)
+    etag = '"' + hashlib.sha256(data).hexdigest()[:32] + '"'
+    return Response(
+        content=data,
+        media_type="font/ttf",
+        headers={"Cache-Control": "public, max-age=31536000, immutable", "ETag": etag},
+    )
+
 
 @router.get("/{document_id}/raw")
 def get_document_raw(document_id: str) -> Response:
@@ -463,6 +488,7 @@ class TextRunModel(BaseModel):
     h: float
     object_indices: list[int] = []
     color: str = "#000000"
+    font_fidelity: str = "exact"
 
 
 class TextLineModel(BaseModel):
@@ -547,6 +573,7 @@ def get_page_model(document_id: str, page_index: int) -> PageModelResponse:
                             h=r.h,
                             object_indices=r.object_indices,
                             color=getattr(r, "fill_color", "#000000") or "#000000",
+                            font_fidelity=getattr(r, "font_fidelity", "exact") or "exact",
                         )
                     )
                 lines.append(

gateway/app/routers/edits.py

@@ -214,6 +214,7 @@ class ReplaceTextData(BaseModel):
     text: str
     internalFontId: str
     fontSize: float
+    disableJustify: bool = False
 
 
 class ReplaceTextOperation(BaseModel):