docslicer 0.2.4__tar.gz → 0.2.5__tar.gz

{docslicer-0.2.4 → docslicer-0.2.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docslicer
-Version: 0.2.4
+Version: 0.2.5
 Summary: Answer questions from documents too large to fit in context, reading only the sections you need
 Author-email: "Market Framer Inc." <jelle@docslicer.ai>
 License-Expression: AGPL-3.0-only

{docslicer-0.2.4 → docslicer-0.2.5}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "docslicer"
-version = "0.2.4"
+version = "0.2.5"
 description = "Answer questions from documents too large to fit in context, reading only the sections you need"
 readme = "README.md"
 license = "AGPL-3.0-only"

{docslicer-0.2.4 → docslicer-0.2.5}/src/docslicer/__init__.py

@@ -20,7 +20,7 @@ try:
     from importlib.metadata import version as _pkg_version
     __version__ = _pkg_version("docslicer")
 except Exception:
-    __version__ = "0.2.4"
+    __version__ = "0.2.5"
 
 from ._config import ParseConfig
 from ._result import ParseResult, Chunk, Block, Table, TableCell, Chart, ChartPoint, BBox, HierarchyNode, HierarchyTree

{docslicer-0.2.4 → docslicer-0.2.5}/src/docslicer/_orchestrator.py

@@ -601,7 +601,7 @@ def _run_pipeline(
             ),
         )
 
-    df_blocks, df_chunks = _run_shared_pipeline(
+    df_lines, df_blocks, df_chunks = _run_shared_pipeline(
         lines_df=df_lines,
         df_table_cells=df_table_cells,
         chart_points_df=df_chart_points,
@@ -609,6 +609,12 @@ def _run_pipeline(
         on_stage=on_stage,
     )
 
+    # The shared pipeline works on a copy, so the frame the type-specific
+    # orchestrator registered under "lines" predates every shared
+    # classification. Rebind to the enriched one consumers actually want.
+    if config.debug:
+        early_steps["lines"] = df_lines
+
     return _build_result(
         discovered_metadata,
         df_blocks,

{docslicer-0.2.4 → docslicer-0.2.5}/src/docslicer/shared/shared_orchestrator.py

@@ -50,7 +50,7 @@ def run_pipeline(
     on_stage: Optional[Callable[[str], None]] = None,
     config: ParseConfig = None,
     metadata: Dict[str, Any] = None,
-) -> tuple[pd.DataFrame, pd.DataFrame]:
+) -> tuple[pd.DataFrame, pd.DataFrame, pd.DataFrame]:
     """
     Run shared document processing steps.
     
@@ -69,12 +69,15 @@ def run_pipeline(
         metadata: Document metadata to attach to output
     
     Returns:
-        DataFrame with chunks
+        (df_lines, df_blocks, df_chunks) — df_lines is the *enriched* line frame
+        carrying the shared classifications (navigation, TOC, exhibit, doc region,
+        hierarchy), not the input frame. Callers that expose lines for debugging
+        must rebind to this one; the input frame is left untouched.
     """
     config = config or DEFAULT_CONFIG
     
     if lines_df.empty:
-        return lines_df
+        return lines_df, lines_df, lines_df
     
     df = lines_df.copy()
     
@@ -144,7 +147,7 @@ def run_pipeline(
 
     # Step 07 - Chunk Building (blocks -> chunks)
     if not config.chunking:
-        return df_blocks, pd.DataFrame()
+        return df, df_blocks, pd.DataFrame()
 
     with timed_step("chunk_building", logger=logger):
         df_chunks = build_chunks(
@@ -165,4 +168,4 @@ def run_pipeline(
             if key not in df_chunks.columns:
                 df_chunks[key] = value
 
-    return df_blocks, df_chunks
+    return df, df_blocks, df_chunks

{docslicer-0.2.4 → docslicer-0.2.5}/src/docslicer.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docslicer
-Version: 0.2.4
+Version: 0.2.5
 Summary: Answer questions from documents too large to fit in context, reading only the sections you need
 Author-email: "Market Framer Inc." <jelle@docslicer.ai>
 License-Expression: AGPL-3.0-only