Fix clipped table text after HWP conversion

hwp2hwpx/section_converter.py

@@ -1,6 +1,8 @@
 """Convert HWP BodyText sections to HWPX section XML files."""
 
 import struct
+import math
+import re
 from .xml_builder import root_element, sub, make_tag
 from . import value_maps as vm
 
@@ -53,6 +55,24 @@ def _compute_final_dimensions(sc_content):
     return int(round(w)), int(round(h))
 
 
+def _weighted_text_len(text):
+    """Approximate rendered text length for table wrapping.
+
+    HWP cell heights in older binary files can be lower than the height needed
+    by HWPX viewers after line wrapping. Count CJK characters as full width and
+    ASCII as narrower so we only expand rows that are likely to clip.
+    """
+    total = 0.0
+    for ch in text:
+        if ch.isspace():
+            total += 0.25
+        elif ord(ch) < 128:
+            total += 0.55
+        else:
+            total += 1.0
+    return total
+
+
 def _transform_point(x, y, sc_content):
     """Transform a point through the scalerotation chain (scaling only, no translation).
 
@@ -1107,6 +1127,63 @@ def _build_table_cells(self, tbl, start_pos, end_pos, ctrl_level):
 
             self._build_table_cell(tr, cell_content, cell_start, cell_end)
 
+        self._adjust_table_row_heights(tbl)
+
+    def _adjust_table_row_heights(self, tbl):
+        """Expand table row heights when wrapped cell text would be clipped."""
+        for tr in tbl.findall(make_tag("hp", "tr")):
+            required_heights = []
+            cells = []
+            for tc in tr.findall(make_tag("hp", "tc")):
+                csz = tc.find(make_tag("hp", "cellSz"))
+                if csz is None:
+                    continue
+
+                try:
+                    width = int(csz.get("width", "0"))
+                    height = int(csz.get("height", "0"))
+                except ValueError:
+                    continue
+
+                text = " ".join(
+                    t.text or "" for t in tc.iter(make_tag("hp", "t"))
+                    if (t.text or "").strip()
+                )
+                text = re.sub(r"\s+", " ", text).strip()
+                if not text:
+                    continue
+
+                margin = tc.find(make_tag("hp", "cellMargin"))
+                if margin is not None:
+                    left = int(margin.get("left", "141"))
+                    right = int(margin.get("right", "141"))
+                    top = int(margin.get("top", "141"))
+                    bottom = int(margin.get("bottom", "141"))
+                else:
+                    left = right = top = bottom = 141
+
+                # HWP char shape heights are hundredths of a point; 1000 is a
+                # common 10pt body font. Section conversion does not have direct
+                # access to header charPr records here, so use the body default.
+                font_height = 1000
+                available_width = max(1, width - left - right)
+                chars_per_line = max(1.0, available_width / (font_height * 0.92))
+                lines = max(1, math.ceil(_weighted_text_len(text) / chars_per_line))
+                required = int(math.ceil(lines * font_height * 1.32 + top + bottom + 180))
+                required = max(required, int(font_height * 1.55 + top + bottom))
+
+                if height < required:
+                    required_heights.append(required)
+                    cells.append((csz, height))
+
+            if not required_heights:
+                continue
+
+            row_height = max(required_heights)
+            for csz, height in cells:
+                if height < row_height:
+                    csz.set("height", str(row_height))
+
     def _build_table_cell(self, tr, cell_content, children_start, children_end):
         """Build a table cell element."""
         tc = sub(tr, "hp", "tc")