fix(libreoffice): address import review comments

libreoffice/agent-harness/cli_anything/libreoffice/core/importer.py

@@ -156,7 +156,11 @@ def _apply_metadata(project: Dict[str, Any], meta_xml: str, source_path: str) ->
     if not meta_xml:
         return
 
-    root = ET.fromstring(meta_xml)
+    try:
+        root = ET.fromstring(meta_xml)
+    except ET.ParseError as e:
+        raise ValueError(f"Invalid ODF meta.xml in: {source_path}") from e
+
     mappings = {
         "title": ("dc", "title"),
         "author": ("dc", "creator"),
@@ -295,7 +299,7 @@ def _parse_calc_row(row_elem: ET.Element, row_index: int) -> Dict[str, Dict[str,
 
 def _cell_data(cell_elem: ET.Element) -> Optional[Dict[str, Any]]:
     value_type = _attr(cell_elem, "office", "value-type") or "string"
-    formula = _attr(cell_elem, "table", "formula")
+    formula = _normalize_formula(_attr(cell_elem, "table", "formula"))
     text = _text_content(cell_elem)
     numeric_value = _attr(cell_elem, "office", "value")
 
@@ -322,6 +326,16 @@ def _cell_data(cell_elem: ET.Element) -> Optional[Dict[str, Any]]:
     return data
 
 
+def _normalize_formula(formula: Optional[str]) -> Optional[str]:
+    """Strip ODF formula namespace prefixes before storing project state."""
+    if formula is None:
+        return None
+    for prefix in ("of:", "oooc:"):
+        if formula.startswith(prefix):
+            return formula[len(prefix):]
+    return formula
+
+
 def _parse_impress_content(root: ET.Element) -> List[Dict[str, Any]]:
     body = root.find(_q("office", "body"))
     presentation = body.find(_q("office", "presentation")) if body is not None else None

libreoffice/agent-harness/cli_anything/libreoffice/tests/TEST.md

@@ -4,9 +4,9 @@
 
 | File | Test Classes | Test Count | Focus |
 |------|-------------|------------|-------|
-| `test_core.py` | 7 | 97 | Unit tests for document, import, Writer, Calc, Impress, styles, session |
+| `test_core.py` | 7 | 99 | Unit tests for document, import, Writer, Calc, Impress, styles, session |
 | `test_full_e2e.py` | 17 | 73 | E2E workflows: ODF ZIP structure, XML validity, import, export formats, CLI subprocess |
-| **Total** | **24** | **170** | |
+| **Total** | **24** | **172** | |
 
 ## Unit Tests (`test_core.py`)
 
@@ -23,14 +23,16 @@ All unit tests use synthetic/in-memory data only. No LibreOffice installation re
 - List available profiles
 - Metadata populated on creation (title, created date)
 
-### TestImport (7 tests)
+### TestImport (9 tests)
 - List supported import formats, including ODF and Microsoft Office extensions
 - Import generated ODT into Writer project content
 - Import generated ODS into Calc sheets and cells
+- Normalize imported Calc formulas before re-export
 - Import generated ODP into Impress slides
 - Route DOCX import through LibreOffice conversion without requiring LibreOffice in unit tests
 - Reject unsupported import formats
 - Reject invalid ODF files with a clean error
+- Reject malformed ODF meta.xml with a clean error
 
 ### TestWriter (18 tests)
 - Add paragraph with default and custom style
@@ -171,6 +173,6 @@ E2E tests produce real ODF files (ODT/ODS/ODP) and validate ZIP structure, XML c
 ## Test Results
 
 ```
-test_core.py: 97 passed in 0.15s
+test_core.py: 99 passed in 0.15s
 test_full_e2e.py: 73 passed in 58.33s
 ```

libreoffice/agent-harness/cli_anything/libreoffice/tests/test_core.py

@@ -8,6 +8,7 @@ import os
 import sys
 import tempfile
 import shutil
+import zipfile
 import pytest
 
 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__)))))
@@ -35,6 +36,7 @@ from cli_anything.libreoffice.core.styles import (
 from cli_anything.libreoffice.core.session import Session
 from cli_anything.libreoffice.core.export import to_odt, to_ods, to_odp
 from cli_anything.libreoffice.core import importer as importer_mod
+from cli_anything.libreoffice.utils.odf_utils import parse_odf
 
 
 # ── Document Tests ───────────────────────────────────────────────
@@ -186,6 +188,25 @@ class TestImport:
         assert imported["sheets"][0]["cells"]["B1"]["value"] == 42.0
         assert imported["sheets"][0]["cells"]["B1"]["type"] == "float"
 
+    def test_import_calc_formula_normalizes_odf_prefix(self):
+        proj = create_document(doc_type="calc", name="formula_calc")
+        set_cell(proj, "A1", "1", cell_type="float")
+        set_cell(proj, "A2", "2", cell_type="float")
+        set_cell(proj, "A3", "0", cell_type="float", formula="=A1+A2")
+
+        with tempfile.TemporaryDirectory() as tmp:
+            source = os.path.join(tmp, "formula.ods")
+            roundtrip = os.path.join(tmp, "roundtrip.ods")
+            to_ods(proj, source)
+            imported = importer_mod.import_document(source)
+            to_ods(imported, roundtrip)
+            content_xml = parse_odf(roundtrip)["content_xml"]
+
+        formula = imported["sheets"][0]["cells"]["A3"]["formula"]
+        assert formula == "=A1+A2"
+        assert 'table:formula="of:=A1+A2"' in content_xml
+        assert "of:of:=" not in content_xml
+
     def test_import_impress_odp(self):
         proj = create_document(doc_type="impress", name="import_impress")
         add_slide(proj, title="Intro", content="Welcome")
@@ -246,6 +267,25 @@ class TestImport:
         finally:
             os.unlink(path)
 
+    def test_reject_malformed_odf_meta_xml(self):
+        proj = create_document(doc_type="writer")
+        add_paragraph(proj, text="body")
+
+        with tempfile.TemporaryDirectory() as tmp:
+            source = os.path.join(tmp, "source.odt")
+            malformed = os.path.join(tmp, "malformed.odt")
+            to_odt(proj, source)
+
+            with zipfile.ZipFile(source, "r") as zin, zipfile.ZipFile(malformed, "w") as zout:
+                for info in zin.infolist():
+                    data = zin.read(info.filename)
+                    if info.filename == "meta.xml":
+                        data = b"<broken>"
+                    zout.writestr(info, data)
+
+            with pytest.raises(ValueError, match="Invalid ODF meta.xml"):
+                importer_mod.import_document(malformed)
+
 
 # ── Writer Tests ─────────────────────────────────────────────────
 

libreoffice/agent-harness/cli_anything/libreoffice/utils/lo_backend.py

@@ -14,6 +14,7 @@ import os
 import shutil
 import subprocess
 import tempfile
+from pathlib import Path
 from typing import Optional
 
 
@@ -112,20 +113,26 @@ def convert(
             tempfile.TemporaryDirectory(prefix="lo-runtime-") as runtime_dir, \
             tempfile.TemporaryDirectory(prefix="lo-config-") as config_dir, \
             tempfile.TemporaryDirectory(prefix="lo-cache-") as cache_dir:
-        os.chmod(runtime_dir, 0o700)
         env = os.environ.copy()
-        env.update({
-            "XDG_RUNTIME_DIR": runtime_dir,
-            "XDG_CONFIG_HOME": config_dir,
-            "XDG_CACHE_HOME": cache_dir,
-        })
+        if os.name == "posix":
+            try:
+                os.chmod(runtime_dir, 0o700)
+            except OSError:
+                pass
+            env.update({
+                "XDG_RUNTIME_DIR": runtime_dir,
+                "XDG_CONFIG_HOME": config_dir,
+                "XDG_CACHE_HOME": cache_dir,
+            })
+
+        profile_uri = Path(profile_dir).resolve().as_uri()
 
         cmd = [
             lo,
             "--headless",
             "--nologo",
             "--nofirststartwizard",
-            f"-env:UserInstallation=file://{profile_dir}",
+            f"-env:UserInstallation={profile_uri}",
             "--convert-to", output_format,
             "--outdir", output_dir,
             input_path,