python-package-folder 2.0.2__py3-none-any.whl → 2.0.5__py3-none-any.whl

python_package_folder/manager.py

@@ -86,6 +86,8 @@ class BuildManager:
             self.project_root, self.src_dir, exclude_patterns=exclude_patterns
         )
         self.subfolder_config: SubfolderBuildConfig | None = None
+        # Cache for package name lookups (expensive operation)
+        self._packages_distributions_cache: dict[str, list[str]] | None = None
 
         # Check if it's a valid Python package directory
         if not any(self.src_dir.glob("*.py")) and not (self.src_dir / "__init__.py").exists():
@@ -271,7 +273,9 @@ class BuildManager:
         # For subfolder builds, extract third-party dependencies and add to pyproject.toml
         if self._is_subfolder_build() and self.subfolder_config:
             # Re-analyze all Python files (including copied dependencies) to find third-party imports
+            print("Analyzing Python files for third-party dependencies...")
             all_python_files = analyzer.find_all_python_files(self.src_dir)
+            print(f"Found {len(all_python_files)} Python files to analyze")
             third_party_deps = self._extract_third_party_dependencies(all_python_files, analyzer)
             if third_party_deps:
                 print(
@@ -394,6 +398,111 @@ class BuildManager:
             elif src_item.is_dir():
                 self._copytree_excluding(src_item, dst_item)
 
+    def _get_package_name_from_import(self, module_name: str) -> str | None:
+        """
+        Get the actual PyPI package name from an import module name.
+
+        This handles cases where the import name differs from the package name
+        (e.g., 'import fitz' from 'pymupdf' package).
+
+        Args:
+            module_name: The module name from the import statement
+
+        Returns:
+            The actual package name, or None if not found
+        """
+        root_module = module_name.split(".")[0]
+        try:
+            # Try Python 3.10+ first (has packages_distributions)
+            import importlib.metadata as importlib_metadata
+
+            # Use packages_distributions() if available (Python 3.10+)
+            # Cache the result since it's expensive to call
+            if hasattr(importlib_metadata, "packages_distributions"):
+                if self._packages_distributions_cache is None:
+                    # Cache the packages_distributions() result
+                    self._packages_distributions_cache = importlib_metadata.packages_distributions()
+                packages_map = self._packages_distributions_cache
+                # packages_map is a dict mapping module names to list of distribution names
+                if root_module in packages_map:
+                    # Return the first distribution name (usually there's only one)
+                    dist_names = packages_map[root_module]
+                    if dist_names:
+                        return dist_names[0]
+
+            # Fallback: search all distributions (this can be slow, so limit search)
+            # Only check top-level package matches to speed up search
+            dist_count = 0
+            max_distributions_to_check = 1000  # Limit to prevent excessive searching
+            for dist in importlib_metadata.distributions():
+                dist_count += 1
+                if dist_count > max_distributions_to_check:
+                    # Too many distributions, give up to avoid hanging
+                    break
+                try:
+                    # Check distribution name first (fast check)
+                    dist_name = dist.metadata.get("Name", "")
+                    # If distribution name matches or contains the module name, check files
+                    if dist_name.lower().replace(
+                        "-", "_"
+                    ) == root_module.lower() or root_module.lower() in dist_name.lower().replace(
+                        "-", "_"
+                    ):
+                        # Check if this distribution provides the module by looking at its files
+                        files = dist.files or []
+                        # Limit file checking to first 100 files per distribution
+                        file_count = 0
+                        for file in files:
+                            file_count += 1
+                            if file_count > 100:
+                                break
+                            file_str = str(file)
+                            # Check if file is the module itself or in a package directory
+                            if (
+                                file.suffix == ".py"
+                                and (file.stem == root_module or file.stem == "__init__")
+                            ) or (
+                                "/" in file_str
+                                and (
+                                    file_str.startswith(f"{root_module}/")
+                                    or file_str.startswith(f"{root_module.replace('_', '-')}/")
+                                )
+                            ):
+                                return dist.metadata["Name"]
+                except Exception:
+                    continue
+
+        except ImportError:
+            try:
+                # Fallback for older Python versions
+                import importlib_metadata
+
+                # Search all distributions
+                for dist in importlib_metadata.distributions():
+                    try:
+                        files = dist.files or []
+                        for file in files:
+                            file_str = str(file)
+                            if (
+                                file.suffix == ".py"
+                                and (file.stem == root_module or file.stem == "__init__")
+                            ) or (
+                                "/" in file_str
+                                and (
+                                    file_str.startswith(f"{root_module}/")
+                                    or file_str.startswith(f"{root_module.replace('_', '-')}/")
+                                )
+                            ):
+                                return dist.metadata["Name"]
+                    except Exception:
+                        continue
+            except ImportError:
+                pass
+        except Exception:
+            pass
+
+        return None
+
     def _extract_third_party_dependencies(
         self, python_files: list[Path], analyzer: ImportAnalyzer
     ) -> list[str]:
@@ -401,18 +510,25 @@ class BuildManager:
         Extract third-party package dependencies from Python files.
 
         Analyzes all Python files to find imports classified as "third_party"
-        and returns a list of unique package names.
+        and returns a list of unique package names. Handles cases where the
+        import name differs from the package name (e.g., 'fitz' -> 'pymupdf').
 
         Args:
             python_files: List of Python file paths to analyze
             analyzer: ImportAnalyzer instance to use for classification
 
         Returns:
-            List of unique third-party package names (e.g., ["pypdf", "requests"])
+            List of unique third-party package names (e.g., ["pypdf", "requests", "pymupdf"])
         """
         third_party_packages: set[str] = set()
+        # Cache package name lookups to avoid repeated expensive searches
+        package_name_cache: dict[str, str | None] = {}
+
+        total_files = len(python_files)
+        for idx, file_path in enumerate(python_files):
+            if idx > 0 and idx % 50 == 0:
+                print(f"  Analyzing file {idx}/{total_files}...", end="\r", flush=True)
 
-        for file_path in python_files:
             imports = analyzer.extract_imports(file_path)
             for imp in imports:
                 analyzer.classify_import(imp, self.src_dir)
@@ -425,17 +541,38 @@ class BuildManager:
                 if root_module in stdlib_modules:
                     continue
 
-                # If classified as third_party, add it
+                # If classified as third_party, try to get actual package name
                 if imp.classification == "third_party":
-                    third_party_packages.add(root_module)
+                    # Check cache first
+                    if root_module not in package_name_cache:
+                        package_name_cache[root_module] = self._get_package_name_from_import(
+                            imp.module_name
+                        )
+                    actual_package = package_name_cache[root_module]
+                    if actual_package:
+                        third_party_packages.add(actual_package)
+                    else:
+                        # Fallback to using the import name
+                        third_party_packages.add(root_module)
                 # If it's ambiguous or unresolved, and not stdlib/local/external,
                 # it's likely a third-party package that needs to be declared
                 elif imp.classification == "ambiguous" or imp.classification is None:
                     # Check if it's not a local or external module
                     if not imp.resolved_path:
-                        # This is likely a third-party package that's not installed
-                        # in the build environment but needs to be declared
-                        third_party_packages.add(root_module)
+                        # Check cache first
+                        if root_module not in package_name_cache:
+                            package_name_cache[root_module] = self._get_package_name_from_import(
+                                imp.module_name
+                            )
+                        actual_package = package_name_cache[root_module]
+                        if actual_package:
+                            third_party_packages.add(actual_package)
+                        else:
+                            # Fallback: use import name (will be normalized later)
+                            third_party_packages.add(root_module)
+
+        if total_files > 50:
+            print()  # New line after progress indicator
 
         return sorted(list(third_party_packages))
 

python_package_folder/subfolder_build.py

@@ -590,17 +590,36 @@ class SubfolderBuildConfig:
         updated_content = self._add_dependencies_to_pyproject(content, dependencies)
         self.temp_pyproject.write_text(updated_content, encoding="utf-8")
 
+    def _normalize_package_name(self, package_name: str) -> str:
+        """
+        Normalize package name for PyPI.
+
+        Converts underscores to hyphens, as PyPI package names typically use hyphens
+        while Python import names use underscores (e.g., 'better_enum' -> 'better-enum').
+
+        Args:
+            package_name: Package name from import statement
+
+        Returns:
+            Normalized package name for PyPI
+        """
+        # Convert underscores to hyphens for PyPI package names
+        # This handles the common case where import names use underscores
+        # but PyPI package names use hyphens
+        return package_name.replace("_", "-")
+
     def _add_dependencies_to_pyproject(self, content: str, dependencies: list[str]) -> str:
         """
         Add dependencies to pyproject.toml content.
 
         Adds the specified dependencies to the [project] section's dependencies list.
         If dependencies already exist, merges them. If no dependencies section exists,
-        creates one.
+        creates one. Package names are normalized (underscores -> hyphens) to match
+        PyPI naming conventions.
 
         Args:
             content: Current pyproject.toml content
-            dependencies: List of dependency names to add
+            dependencies: List of dependency names to add (will be normalized)
 
         Returns:
             Updated pyproject.toml content with dependencies added
@@ -608,6 +627,9 @@ class SubfolderBuildConfig:
         if not dependencies:
             return content
 
+        # Normalize package names (convert underscores to hyphens for PyPI)
+        normalized_deps = [self._normalize_package_name(dep) for dep in dependencies]
+
         lines = content.split("\n")
         result = []
         in_project = False
@@ -631,8 +653,8 @@ class SubfolderBuildConfig:
                 if line.strip().endswith("]"):
                     in_dependencies = False
 
-        # Merge with new dependencies
-        all_deps = sorted(existing_deps | set(dependencies))
+        # Merge with new dependencies (normalized)
+        all_deps = sorted(existing_deps | set(normalized_deps))
 
         # Second pass: build result with dependencies
         in_project = False

{python_package_folder-2.0.2.dist-info → python_package_folder-2.0.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: python-package-folder
-Version: 2.0.2
+Version: 2.0.5
 Summary: Python package to automatically package and build a folder, fetching all relevant dependencies.
 Project-URL: Repository, https://github.com/alelom/python-package-folder
 Author-email: Alessio Lombardi <work@alelom.com>

{python_package_folder-2.0.2.dist-info → python_package_folder-2.0.5.dist-info}/RECORD

@@ -2,16 +2,16 @@ python_package_folder/__init__.py,sha256=DQt-uldOEKfh0MUqCvKdeNKOnpuOvpb7blYvXMy
 python_package_folder/__main__.py,sha256=a-__-VLhYw-J7S7CsHdhtEvQr3RiAZxiYDvKhKTgMX4,291
 python_package_folder/analyzer.py,sha256=w7hc2oyOoPK7tvlwcJDXnB3eiJsuGZc4BkOpTfZP7Vo,12257
 python_package_folder/finder.py,sha256=_LvJ9xBVKv41UK5sbwbNyKmuYjAOqUbzvZhK7NCYQF8,9130
-python_package_folder/manager.py,sha256=t3mbTnn42QCLWSOC5P_XRJPi1bL9hyYaCZVdITENHXk,33764
+python_package_folder/manager.py,sha256=NER-sg-KiTshsnfpwLjsFVkFAbqNbH69gRK-4u4Br0s,40495
 python_package_folder/publisher.py,sha256=TSjdOvxvnWLbJCnduTK_xZBRfvsrq9kpEH-sfebeWkU,13507
 python_package_folder/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 python_package_folder/python_package_folder.py,sha256=RPsqRcIy_LjzzTHdp4qdtFJ4-4xhtR_0YLIC0RlUxFo,8841
-python_package_folder/subfolder_build.py,sha256=IfnXW5Xx8-WMuBc8hTdbCImDIuUAiHvVe1k5V2FbwoY,34339
+python_package_folder/subfolder_build.py,sha256=LMiJ4Ck6PW1x3hmiSK-9fDH4Q6RrHqot4drx4lise3E,35310
 python_package_folder/types.py,sha256=3yeSRR5p_3PDKEAaehW_RJ7NwJHexOIeA08bGaT1iSY,2368
 python_package_folder/utils.py,sha256=lIkWsFKeAYAJ9TDUM99T4pUBHJVbUvCdUgkWQN-LUho,3111
 python_package_folder/version.py,sha256=kIDP6S9trEfs9gj7lBYGxrWm4RPssRla24UtlO9Jkh4,9111
-python_package_folder-2.0.2.dist-info/METADATA,sha256=1AFTide6hdpSNzjVFpLai2Mo-UYyfSSlA9hcx66QJBM,33282
-python_package_folder-2.0.2.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-python_package_folder-2.0.2.dist-info/entry_points.txt,sha256=ttu4wAhoYSHGhWQNercLz9IVTTpXxhVlRA9vSTvaLe0,91
-python_package_folder-2.0.2.dist-info/licenses/LICENSE,sha256=vNgRJh8YiecqZoZld7TtwPI5I72HIymKD9g32fiJjCE,1073
-python_package_folder-2.0.2.dist-info/RECORD,,
+python_package_folder-2.0.5.dist-info/METADATA,sha256=k44aRgDTztx-nSRrykq2BVtNDsXe2Mbftgm9cxPqY58,33282
+python_package_folder-2.0.5.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+python_package_folder-2.0.5.dist-info/entry_points.txt,sha256=ttu4wAhoYSHGhWQNercLz9IVTTpXxhVlRA9vSTvaLe0,91
+python_package_folder-2.0.5.dist-info/licenses/LICENSE,sha256=vNgRJh8YiecqZoZld7TtwPI5I72HIymKD9g32fiJjCE,1073
+python_package_folder-2.0.5.dist-info/RECORD,,