natural-pdf 0.1.21__py3-none-any.whl → 0.1.22__py3-none-any.whl

natural_pdf/cli.py

@@ -11,7 +11,9 @@ from packaging.requirements import Requirement
 # ---------------------------------------------------------------------------
 INSTALL_RECIPES: Dict[str, list[str]] = {
     # heavyweight stacks
-    "paddle": ["paddlepaddle>=3.0.0", "paddleocr>=3.0.1", "paddlex>=3.0.2"],
+    "paddle": ["paddlepaddle>=3.0.0", "paddleocr>=3.0.1", "paddlex>=3.0.2", "pandas>=2.2.0"],
+    "numpy-high": ["numpy>=2.0"],
+    "numpy-low": ["numpy<1.27"],
     "surya": ["surya-ocr>=0.13.0"],
     "yolo": ["doclayout_yolo", "huggingface_hub>=0.29.3"],
     "docling": ["docling"],
@@ -24,7 +26,7 @@ INSTALL_RECIPES: Dict[str, list[str]] = {
 
 def _build_pip_install_args(requirements: list[str], upgrade: bool = True):
     """Return the pip command list to install/upgrade the given requirement strings."""
-    cmd = [sys.executable, "-m", "pip", "install"]
+    cmd = [sys.executable, "-m", "pip", "--quiet", "install"]
     if upgrade:
         cmd.append("--upgrade")
     cmd.extend(requirements)
@@ -48,34 +50,13 @@ def cmd_install(args):
 
         requirements = INSTALL_RECIPES[group_key]
 
-        # Skip paddlex upgrade if already satisfied
-        if group_key == "paddle":
-            try:
-                dist = distribution("paddlex")
-                from packaging.version import parse as V
-                if V(dist.version) >= V("3.0.2"):
-                    print("✓ paddlex already ≥ 3.0.2 – nothing to do.")
-                    continue
-            except PackageNotFoundError:
-                pass
-
         # Special handling for paddle stack: install paddlepaddle & paddleocr first
         # each in its own resolver run, then paddlex.
-        if group_key == "paddle":
-            base_reqs = [r for r in requirements if not r.startswith("paddlex")]
-            for req in base_reqs:
-                pip_cmd = _build_pip_install_args([req])
-                _run(pip_cmd)
-
-            # paddlex last to override the strict pin
-            pip_cmd = _build_pip_install_args(["paddlex==3.0.2"])
+        base_reqs = [r for r in requirements]
+        for req in base_reqs:
+            pip_cmd = _build_pip_install_args([req])
             _run(pip_cmd)
-            print("✔ Paddle stack installed (paddlex upgraded to 3.0.2)")
-        else:
-            for req in requirements:
-                pip_cmd = _build_pip_install_args([req])
-                _run(pip_cmd)
-            print("✔ Finished installing extra dependencies for", group_key)
+        print("✔ Finished installing extra dependencies for", group_key)
 
 
 def main():

natural_pdf/core/pdf.py

@@ -275,56 +275,42 @@ class PDF(ExtractionMixin, ExportMixin, ClassificationMixin):
         )
 
     def _initialize_managers(self):
-        """Initialize manager instances based on DEFAULT_MANAGERS."""
-        self._managers = {}
-        for key, manager_class_or_factory in DEFAULT_MANAGERS.items():
-            try:
-                # Resolve the entry in DEFAULT_MANAGERS which can be:
-                #   1. A class  -> instantiate directly
-                #   2. A factory (callable) returning a class -> call then instantiate
-                #   3. A factory returning a **ready instance** -> use as-is
-
-                resolved = manager_class_or_factory
-
-                # If we have a callable that is *not* a class, call it to obtain the real target
-                # (This is the lazy-import factory case.)
-                if not isinstance(resolved, type) and callable(resolved):
-                    resolved = resolved()
-
-                # At this point `resolved` is either a class or an already-created instance
-                if isinstance(resolved, type):
-                    instance = resolved()  # Instantiate class
-                    self._managers[key] = instance
-                    logger.debug(f"Initialized manager for key '{key}': {resolved.__name__}")
-                else:
-                    # Assume factory already returned an instance
-                    self._managers[key] = resolved
-                    logger.debug(
-                        f"Initialized manager instance for key '{key}': {type(resolved).__name__} (factory-provided instance)"
-                    )
-            except Exception as e:
-                logger.error(f"Failed to initialize manager for key '{key}': {e}")
-                self._managers[key] = None
+        """Set up manager factories for lazy instantiation."""
+        # Store factories/classes for each manager key
+        self._manager_factories = dict(DEFAULT_MANAGERS)
+        self._managers = {}  # Will hold instantiated managers
 
     def get_manager(self, key: str) -> Any:
-        """Retrieve a manager instance by its key."""
-        if key not in self._managers:
+        """Retrieve a manager instance by its key, instantiating it lazily if needed."""
+        # Check if already instantiated
+        if key in self._managers:
+            manager_instance = self._managers[key]
+            if manager_instance is None:
+                raise RuntimeError(f"Manager '{key}' failed to initialize previously.")
+            return manager_instance
+
+        # Not instantiated yet: get factory/class
+        if not hasattr(self, "_manager_factories") or key not in self._manager_factories:
             raise KeyError(
-                f"No manager registered for key '{key}'. Available: {list(self._managers.keys())}"
+                f"No manager registered for key '{key}'. Available: {list(getattr(self, '_manager_factories', {}).keys())}"
             )
-
-        manager_instance = self._managers.get(key)
-
-        if manager_instance is None:
-            manager_class = DEFAULT_MANAGERS.get(key)
-            if manager_class:
-                raise RuntimeError(
-                    f"Manager '{key}' ({manager_class.__name__}) failed to initialize previously."
-                )
+        factory_or_class = self._manager_factories[key]
+        try:
+            resolved = factory_or_class
+            # If it's a callable that's not a class, call it to get the class/instance
+            if not isinstance(resolved, type) and callable(resolved):
+                resolved = resolved()
+            # If it's a class, instantiate it
+            if isinstance(resolved, type):
+                instance = resolved()
             else:
-                raise RuntimeError(f"Manager '{key}' failed to initialize (class not found).")
-
-        return manager_instance
+                instance = resolved  # Already an instance
+            self._managers[key] = instance
+            return instance
+        except Exception as e:
+            logger.error(f"Failed to initialize manager for key '{key}': {e}")
+            self._managers[key] = None
+            raise RuntimeError(f"Manager '{key}' failed to initialize: {e}") from e
 
     def _initialize_highlighter(self):
         pass

natural_pdf/describe/base.py

@@ -292,8 +292,8 @@ def _extract_element_value(element: "Element", column: str) -> Any:
     try:
         if column == 'text':
             text = getattr(element, 'text', '')
-            if text and len(text) > 50:
-                return text[:50] + "..."
+            if text and len(text) > 60:
+                return text[:60] + "..."
             return text or ""
         
         elif column == 'page':

natural_pdf/describe/elements.py

@@ -396,7 +396,7 @@ def _analyze_ocr_quality(elements: List["Element"]) -> Dict[str, Any]:
             text = getattr(element, 'text', '').strip()
             if text:
                 # Truncate long text
-                display_text = text[:50] + "..." if len(text) > 50 else text
+                display_text = text[:60] + "..." if len(text) > 60 else text
                 element_confidences.append((confidence, display_text))
     
     if element_confidences:

{natural_pdf-0.1.21.dist-info → natural_pdf-0.1.22.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: natural-pdf
-Version: 0.1.21
+Version: 0.1.22
 Summary: A more intuitive interface for working with PDFs
 Author-email: Jonathan Soma <jonathan.soma@gmail.com>
 License-Expression: MIT
@@ -21,7 +21,6 @@ Requires-Dist: urllib3
 Requires-Dist: tqdm
 Requires-Dist: pydantic
 Requires-Dist: jenkspy
-Requires-Dist: pikepdf
 Requires-Dist: scipy
 Requires-Dist: torch
 Requires-Dist: torchvision

{natural_pdf-0.1.21.dist-info → natural_pdf-0.1.22.dist-info}/RECORD

@@ -1,5 +1,5 @@
 natural_pdf/__init__.py,sha256=qDFJNF8sbEDO-2WSFAxoWEM8updOUP6dB-ckya0kxfs,3275
-natural_pdf/cli.py,sha256=0nAGVO2f_40E3G9c3Q0bfK5mhROyUJH5W25-YJVLMIo,4749
+natural_pdf/cli.py,sha256=-U3vWyI7Qcxr3b5EWY7d2AcsOZJ7D9RnQIMzQsfoDQM,3971
 natural_pdf/analyzers/__init__.py,sha256=dIXjsMqoxKmd9OOnSBzn12wvdIz7D7YNQRAnXslpJSM,142
 natural_pdf/analyzers/shape_detection_mixin.py,sha256=HHefg-v7CJMxYiJHxdGOdqdtbWe9yk4OBoW3a_aRrjM,81798
 natural_pdf/analyzers/text_options.py,sha256=qEkDaYWla0rIM_gszEOsu52q7C_dAfV81P2HLJZM2sw,3333
@@ -27,10 +27,10 @@ natural_pdf/core/__init__.py,sha256=QC8H4M3KbXwMFiQORZ0pdPlzx1Ix6oKKQSS7Ib2KEaA,
 natural_pdf/core/element_manager.py,sha256=_UdXu51sLi6STzc8Pj4k8R721G3yJixXDLuRHn3hmr8,25731
 natural_pdf/core/highlighting_service.py,sha256=_kQUS6_BBvsLBuSZloFrVag6jN90KzHa0ULyGBjufSs,36955
 natural_pdf/core/page.py,sha256=i3DriIQwoO4RuSrkrCXv44Dz8OL9KXPa2y4GhsD1y18,118324
-natural_pdf/core/pdf.py,sha256=yBvb1iGw9gwVPJ3Rm1EBaZ8_g60TuW_Elhg2EOcJMzc,73871
+natural_pdf/core/pdf.py,sha256=0hsIdvKmr6C3WGdD9N9wHGHaZN4th4QwduLwmJ8rWpM,73269
 natural_pdf/describe/__init__.py,sha256=B3zjuHjFI_dFuBLgXR1Q4v7c72fVDyk84d2hs0H4KV8,561
-natural_pdf/describe/base.py,sha256=7USCFIl4mI5b15LTVkwvhAn_mngMwhwxCnVYaZz5Vdc,16842
-natural_pdf/describe/elements.py,sha256=BOkz2wDhGh6P8NOm6pSNxitgmVokLTISztaFhrxMcdw,12717
+natural_pdf/describe/base.py,sha256=LAZLc_thK2u2surgGd0Pk7CN2uVaZK9AbMOE3-1RmQ4,16842
+natural_pdf/describe/elements.py,sha256=xD8wwR1z5IKat7RIwoAwQRUEL6zJTEwcOKorF4F-xPg,12717
 natural_pdf/describe/mixin.py,sha256=U0x6v8r57KQb8qC3VVo64hvhfXQWsti8vdKBM7AXnMo,3116
 natural_pdf/describe/summary.py,sha256=dPtjrn6fQ8nL0F74RITX2vXlDX7ZgaX9JQPnJB-S_XQ,6735
 natural_pdf/elements/__init__.py,sha256=S8XeiNWJ1WcgnyYKdYV1yxQlAxCCO3FfITT8MQwNbyk,41
@@ -91,9 +91,9 @@ natural_pdf/utils/text_extraction.py,sha256=z6Jhy11pakYCsEpkvh8ldw6DkUFsYF1hCL9Y
 natural_pdf/utils/visualization.py,sha256=30pRWQdsRJh2pSObh-brKVsFgC1n8tHmSrta_UDnVPw,8989
 natural_pdf/widgets/__init__.py,sha256=QTVaUmsw__FCweFYZebwPssQxxUFUMd0wpm_cUbGZJY,181
 natural_pdf/widgets/viewer.py,sha256=2VUY1TzWMDe9I-IVNOosKZ2LaqpjLB62ftMAdk-s6_8,24952
-natural_pdf-0.1.21.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
-natural_pdf-0.1.21.dist-info/METADATA,sha256=zo7RfNG44xTbSSdOgGjciqEVMOMCxRX3BY7qtH3roW8,6261
-natural_pdf-0.1.21.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-natural_pdf-0.1.21.dist-info/entry_points.txt,sha256=1R_KMv7g60UBBpRqGfw7bppsMNGdayR-iJlb9ohEk_8,81
-natural_pdf-0.1.21.dist-info/top_level.txt,sha256=Cyw1zmNDlUZfb5moU-WUWGprrwH7ln_8LDGdmMHF1xI,17
-natural_pdf-0.1.21.dist-info/RECORD,,
+natural_pdf-0.1.22.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
+natural_pdf-0.1.22.dist-info/METADATA,sha256=_e2-VmJ4W8_3IZEQc6QqP2LG99f2b-8f1wZRhDSBZsw,6238
+natural_pdf-0.1.22.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+natural_pdf-0.1.22.dist-info/entry_points.txt,sha256=1R_KMv7g60UBBpRqGfw7bppsMNGdayR-iJlb9ohEk_8,81
+natural_pdf-0.1.22.dist-info/top_level.txt,sha256=Cyw1zmNDlUZfb5moU-WUWGprrwH7ln_8LDGdmMHF1xI,17
+natural_pdf-0.1.22.dist-info/RECORD,,