Merge pull request #137 from elong0527/refactor-render-strategies

Refactor render strategies

Author: elong0527

.python-version

@@ -270,7 +270,7 @@ def _render_column_headers(self, document: Any, page: PageContext) -> list[str]:
         return header_elements
 
     def _render_body(self, document: Any, page: PageContext) -> list[str]:
-        page_attrs = page.final_body_attrs or document.rtf_body
+        page_attrs = page.final_body_attrs or page.table_attrs or document.rtf_body
         page_df = page.data
         col_widths = page.col_widths
 

src/rtflite/encoding/renderer.py

@@ -125,6 +125,7 @@ def encode(self, document: Any) -> str:
                     is_last_page=True,
                     col_widths=col_widths,
                     needs_header=True,
+                    table_attrs=processed_attrs,
                 )
             ]
 

src/rtflite/encoding/unified_encoder.py

@@ -29,8 +29,10 @@ def _should_show_element(self, element_location: str, page: PageContext) -> bool
     def _apply_pagination_borders(self, document, page: PageContext) -> Any:
         """Apply proper borders for paginated context following r2rtf design."""
 
-        # Start with a deep copy of the document's body attributes
-        page_attrs = deepcopy(document.rtf_body)
+        # Start with a deep copy of the page's table attributes (processed/sliced)
+        # or document's body attributes if not available
+        base_attrs = page.table_attrs or document.rtf_body
+        page_attrs = deepcopy(base_attrs)
 
         page_df_height = page.data.height
         page_df_width = page.data.width

src/rtflite/pagination/processor.py

@@ -28,6 +28,9 @@ class PageContext(BaseModel):
     # Content Flags
     needs_header: bool = True
 
+    # Base attributes for the table body (sliced/processed)
+    table_attrs: TableAttributes | None = None
+
     # Feature-specific Metadata (populated by strategies or processors)
     subline_header: dict[str, Any] | None = None
     pageby_header_info: dict[str, Any] | None = None

src/rtflite/pagination/strategies/base.py

@@ -40,6 +40,7 @@ def paginate(self, context: PaginationContext) -> list[PageContext]:
                     is_last_page=(page_num == len(page_breaks) - 1),
                     col_widths=context.col_widths,
                     needs_header=(context.rtf_body.pageby_header or page_num == 0),
+                    table_attrs=context.table_attrs,
                 )
             )
 

src/rtflite/pagination/strategies/defaults.py

@@ -50,6 +50,7 @@ def paginate(self, context: PaginationContext) -> list[PageContext]:
                 is_last_page=(page_num == len(page_breaks) - 1),
                 col_widths=context.col_widths,
                 needs_header=needs_header,
+                table_attrs=context.table_attrs,
             )
 
             # Add page_by header info
@@ -155,6 +156,7 @@ def paginate(self, context: PaginationContext) -> list[PageContext]:
                 is_last_page=(page_num == len(page_breaks) - 1),
                 col_widths=context.col_widths,
                 needs_header=is_first or context.rtf_body.pageby_header,
+                table_attrs=context.table_attrs,
             )
 
             if subline_by:

src/rtflite/pagination/strategies/grouping.py

@@ -0,0 +1,65 @@
+import re
+
+import polars as pl
+
+import rtflite as rtf
+
+
+def test_page_by_column_alignment():
+    """Test that column properties are correctly aligned when page_by is used."""
+    df = pl.DataFrame(
+        {
+            "section": ["-----", "Age", "Age"],
+            "item": ["Participant in Population", "    <60", "    >=60"],
+            "value": [55, 25, 30],
+        }
+    )
+
+    doc = rtf.RTFDocument(
+        df=df,
+        rtf_body=rtf.RTFBody(
+            page_by="section",
+            col_rel_width=[1],
+            text_justification=["l", "l", "c"],
+            border_top=["single", "", ""],
+            border_bottom=["single", "", ""],
+        ),
+    )
+
+    rtf_content = doc.rtf_encode()
+
+    # We look for the row containing "55" (value column) and "Participant in Population"
+    rows = rtf_content.split("\\row")
+    row_with_55 = None
+    for row in rows:
+        if "55" in row and "Participant in Population" in row:
+            row_with_55 = row
+            break
+
+    assert row_with_55 is not None, "Could not find the row with data"
+
+    # Check alignment for "Participant in Population"
+    idx_part = row_with_55.find("Participant in Population")
+    pre_part = row_with_55[:idx_part]
+    aligns_part = re.findall(r"\\q[lcrj]", pre_part)
+
+    # Note: The table definition (\trowd...) also contains \trqc etc. but that's row
+    # alignment. Cell content alignment is \ql, \qc etc. inside \pard.
+    # Usually \pard resets, so we should see \ql inside the \pard block.
+    # \trowd... contains \clvertalt etc. but not \ql/\qc usually (unless \clq...?)
+    # The output shows: \pard\hyphpar0... \ql ... content.
+
+    assert len(aligns_part) > 0, "No alignment found for Participant"
+    last_align_part = aligns_part[-1]
+    assert last_align_part == "\\ql", (
+        f"Expected \\ql for Participant, got {last_align_part}"
+    )
+
+    # Check alignment for "55"
+    idx_55 = row_with_55.find("55")
+    pre_55 = row_with_55[:idx_55]
+    aligns_55 = re.findall(r"\\q[lcrj]", pre_55)
+
+    assert len(aligns_55) > 0, "No alignment found for 55"
+    last_align_55 = aligns_55[-1]
+    assert last_align_55 == "\\qc", f"Expected \\qc for 55, got {last_align_55}"