feat(pdf): probe bundled Tesseract first when running frozen

Adds runtime support for the bundled Tesseract that ships inside the
DataTools installer / portable / AppImage artifacts. When DataTools
is launched from a PyInstaller frozen bundle the OCR engine now
resolves automatically — no end-user install required.

New helpers in src/pdf_extract.py:
- _bundled_tesseract_path() → Path | None — returns
  <sys._MEIPASS>/tesseract/tesseract[.exe] when getattr(sys,
  "frozen", False) AND sys._MEIPASS are present; None in dev.
- _bundled_tessdata_dir() → Path | None — same gating, returns
  <sys._MEIPASS>/tesseract/tessdata.
- _apply_bundled_tessdata_prefix() — sets TESSDATA_PREFIX to the
  bundled tessdata dir before any pytesseract call; only if frozen,
  dir exists, and the user hasn't already overridden the env var.

Discovery order in ocr_available() / _autodetect_tesseract_path():
1. DATATOOLS_TESSERACT_PATH env override (existing)
2. Bundled binary (NEW — frozen-only)
3. System PATH (existing)
4. Windows well-known install dirs (existing legacy fallback)

In dev (not frozen) every new probe is a no-op so the developer
experience is unchanged.

12 new tests cover frozen vs. non-frozen detection on each platform,
the user-override respect for TESSDATA_PREFIX, autodetect priority
ordering, and the no-bundled-dir graceful path.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

tests/test_pdf_extract.py

@@ -12,9 +12,16 @@ a fixture statement at test time.
 
 from __future__ import annotations
 
+import os
+from pathlib import Path
+
+from src import pdf_extract
 from src.pdf_extract import (
     Page,
     WordBox,
+    _apply_bundled_tessdata_prefix,
+    _bundled_tessdata_dir,
+    _bundled_tesseract_path,
     _extract_account_number,
     _extract_statement_period,
     _find_amount_tokens,
@@ -456,3 +463,131 @@ class TestYearFromFilename:
     def test_empty_filename(self):
         assert year_from_filename("") is None
         assert year_from_filename(None) is None
+
+
+class TestBundledTesseractPath:
+    """Frozen-bundle Tesseract discovery for installer / portable builds.
+
+    The build agent packages Tesseract at
+    ``<sys._MEIPASS>/tesseract/tesseract[.exe]`` with language data
+    at ``<sys._MEIPASS>/tesseract/tessdata``. These tests pin that
+    contract on the runtime side."""
+
+    def test_returns_none_when_not_frozen(self, monkeypatch):
+        # Default dev environment: ``sys.frozen`` is unset.
+        monkeypatch.delattr("sys.frozen", raising=False)
+        monkeypatch.delattr("sys._MEIPASS", raising=False)
+        assert _bundled_tesseract_path() is None
+        assert _bundled_tessdata_dir() is None
+
+    def test_returns_none_when_frozen_but_no_meipass(self, monkeypatch):
+        # Defensive: ``sys.frozen`` true but ``_MEIPASS`` missing
+        # (shouldn't happen in real PyInstaller bundles but guard
+        # the helper so it can't NoneType-explode).
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.delattr("sys._MEIPASS", raising=False)
+        assert _bundled_tesseract_path() is None
+        assert _bundled_tessdata_dir() is None
+
+    def test_frozen_linux_returns_unsuffixed_binary(
+        self, monkeypatch, tmp_path,
+    ):
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.setattr("sys._MEIPASS", str(tmp_path), raising=False)
+        monkeypatch.setattr("platform.system", lambda: "Linux")
+        expected = tmp_path / "tesseract" / "tesseract"
+        assert _bundled_tesseract_path() == expected
+
+    def test_frozen_macos_returns_unsuffixed_binary(
+        self, monkeypatch, tmp_path,
+    ):
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.setattr("sys._MEIPASS", str(tmp_path), raising=False)
+        monkeypatch.setattr("platform.system", lambda: "Darwin")
+        expected = tmp_path / "tesseract" / "tesseract"
+        assert _bundled_tesseract_path() == expected
+
+    def test_frozen_windows_returns_exe_binary(self, monkeypatch, tmp_path):
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.setattr("sys._MEIPASS", str(tmp_path), raising=False)
+        monkeypatch.setattr("platform.system", lambda: "Windows")
+        expected = tmp_path / "tesseract" / "tesseract.exe"
+        assert _bundled_tesseract_path() == expected
+
+    def test_frozen_returns_tessdata_dir(self, monkeypatch, tmp_path):
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.setattr("sys._MEIPASS", str(tmp_path), raising=False)
+        expected = tmp_path / "tesseract" / "tessdata"
+        assert _bundled_tessdata_dir() == expected
+
+
+class TestAutodetectFavoursBundled:
+    """When a bundled binary exists, ``_autodetect_tesseract_path``
+    should return it BEFORE falling through to Windows install
+    locations — frozen builds shouldn't depend on the user's
+    system tesseract even on Windows."""
+
+    def test_bundled_wins_over_windows_program_files(
+        self, monkeypatch, tmp_path,
+    ):
+        # Simulate frozen Windows build with a bundled binary on disk.
+        bundle_root = tmp_path / "bundle"
+        bundled_bin = bundle_root / "tesseract" / "tesseract.exe"
+        bundled_bin.parent.mkdir(parents=True)
+        bundled_bin.write_bytes(b"")
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.setattr(
+            "sys._MEIPASS", str(bundle_root), raising=False,
+        )
+        monkeypatch.setattr("platform.system", lambda: "Windows")
+        # Pretend the Program Files install also exists — bundled
+        # should still win because we probe it first.
+        monkeypatch.setattr(Path, "exists", lambda self: True)
+        assert pdf_extract._autodetect_tesseract_path() == str(bundled_bin)
+
+    def test_falls_through_when_not_frozen(self, monkeypatch):
+        # Dev: not frozen, not Windows → no candidate at all.
+        monkeypatch.delattr("sys.frozen", raising=False)
+        monkeypatch.delattr("sys._MEIPASS", raising=False)
+        monkeypatch.setattr("platform.system", lambda: "Linux")
+        assert pdf_extract._autodetect_tesseract_path() is None
+
+
+class TestApplyBundledTessdataPrefix:
+    """``TESSDATA_PREFIX`` env var handling — bundled data should be
+    pointed at without clobbering a user override."""
+
+    def test_no_op_when_not_frozen(self, monkeypatch):
+        monkeypatch.delenv("TESSDATA_PREFIX", raising=False)
+        monkeypatch.delattr("sys.frozen", raising=False)
+        monkeypatch.delattr("sys._MEIPASS", raising=False)
+        _apply_bundled_tessdata_prefix()
+        assert "TESSDATA_PREFIX" not in os.environ
+
+    def test_sets_when_frozen_and_bundled_exists(
+        self, monkeypatch, tmp_path,
+    ):
+        tessdata = tmp_path / "tesseract" / "tessdata"
+        tessdata.mkdir(parents=True)
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.setattr("sys._MEIPASS", str(tmp_path), raising=False)
+        monkeypatch.delenv("TESSDATA_PREFIX", raising=False)
+        _apply_bundled_tessdata_prefix()
+        assert os.environ.get("TESSDATA_PREFIX") == str(tessdata)
+
+    def test_does_not_clobber_user_override(self, monkeypatch, tmp_path):
+        tessdata = tmp_path / "tesseract" / "tessdata"
+        tessdata.mkdir(parents=True)
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.setattr("sys._MEIPASS", str(tmp_path), raising=False)
+        monkeypatch.setenv("TESSDATA_PREFIX", "/user/picked/this")
+        _apply_bundled_tessdata_prefix()
+        assert os.environ["TESSDATA_PREFIX"] == "/user/picked/this"
+
+    def test_no_op_when_bundled_dir_missing(self, monkeypatch, tmp_path):
+        # Frozen, but the build didn't ship a tessdata dir.
+        monkeypatch.setattr("sys.frozen", True, raising=False)
+        monkeypatch.setattr("sys._MEIPASS", str(tmp_path), raising=False)
+        monkeypatch.delenv("TESSDATA_PREFIX", raising=False)
+        _apply_bundled_tessdata_prefix()
+        assert "TESSDATA_PREFIX" not in os.environ