Restructuring, linting

Author: dannyadair

src/pdfbaker/__init__.py

@@ -1,23 +1,22 @@
 """Core functionality for document generation."""
 
 from .common import (
-    load_pages,
-    compress_pdf,
     combine_pdfs,
+    compress_pdf,
     convert_svg_to_pdf,
+    load_pages,
 )
-
 from .render import (
-    process_template_data,
     create_env,
+    encode_image,
+    encode_images,
     highlight,
-    process_style,
-    process_text_with_jinja,
     process_list_item_texts,
     process_list_items,
+    process_style,
+    process_template_data,
+    process_text_with_jinja,
     space_bullets,
-    encode_image,
-    encode_images,
 )
 
 __all__ = [

src/pdfbaker/__main__.py

@@ -27,109 +27,153 @@ def _deep_merge(base, update):
     return merged
 
 
-def main(config_path=None):
-    """Main function for the document generator."""
-    # Get config path from arguments if not provided
+def _get_config_path(config_path=None):
+    """Get and validate the configuration file path."""
     if config_path is None:
         if len(sys.argv) < 2:
             print("Error: Config file path is required")
             print("Usage: python -m pdfbaker <config_file_path>")
-            return 1
+            return None
         config_path = sys.argv[1]
-    
-    # Load main configuration
-    config_path = Path(config_path).resolve()  # Get absolute path
+
+    config_path = Path(config_path).resolve()
     if not config_path.exists():
         print(f"Error: Configuration file not found: {config_path}")
-        return 1
-        
+        return None
+
+    return config_path
+
+
+def _load_config(config_path):
+    """Load configuration from a YAML file."""
     with open(config_path, encoding="utf-8") as f:
-        config = yaml.safe_load(f)
-        
-    # Create root output directories relative to config file
-    base_dir = config_path.parent
+        return yaml.safe_load(f)
+
+
+def _setup_output_directories(base_dir):
+    """Create and return build and dist directories."""
     build_dir = base_dir / "build"
     dist_dir = base_dir / "dist"
     os.makedirs(build_dir, exist_ok=True)
     os.makedirs(dist_dir, exist_ok=True)
+    return build_dir, dist_dir
 
-    # Allow document_paths to be absolute or relative to config file
-    documents = config.get("documents", [])
+
+def _resolve_document_paths(base_dir, documents):
+    """Resolve document paths to absolute paths."""
     document_paths = {}
-    
+
     for doc_name in documents:
-        # Check if document path is specified
         if isinstance(doc_name, dict):
             # Format: {"name": "doc_name", "path": "/absolute/path/to/doc"}
             doc_path = Path(doc_name["path"])
             doc_name = doc_name["name"]
         else:
             # Default: document in subdirectory with same name as doc_name
             doc_path = base_dir / doc_name
-        
-        # Store the absolute path
+
         document_paths[doc_name] = doc_path.resolve()
 
-    # Process each document
-    for doc_name, doc_path in document_paths.items():
-        if not doc_path.is_dir():
-            print(f'Warning: Directory missing for document "{doc_name}" at {doc_path} - skipping')
-            continue
-        
-        bake_path = doc_path / "bake.py"
-        if not bake_path.exists():
-            print(f'Warning: bake.py missing for document "{doc_name}" - skipping')
-            continue
-            
-        config_yml_path = doc_path / "config.yml"
-        if not config_yml_path.exists():
-            print(f'Warning: config.yml missing for document "{doc_name}" - skipping')
-            continue
-        
-        print(f'Processing document "{doc_name}" from {doc_path}...')
-        
-        # Load document-specific bake
-        doc_bake = importlib.util.spec_from_file_location(
-            f"documents.{doc_name}.bake",
-            bake_path
+    return document_paths
+
+
+def _validate_document(doc_name, doc_path):
+    """Validate that a document has all required files."""
+    if not doc_path.is_dir():
+        print(
+            f'Warning: Directory missing for document "{doc_name}" '
+            f"at {doc_path} - skipping"
         )
-        module = importlib.util.module_from_spec(doc_bake)
-        doc_bake.loader.exec_module(module)
-        
-        # Merge document configuration
-        with open(config_yml_path, encoding="utf-8") as f:
-            doc_config = yaml.safe_load(f)
-        merged_config = _deep_merge(config, doc_config)
-
-        # Create document-specific output directories
-        doc_build_dir = build_dir / doc_name
-        doc_dist_dir = dist_dir / doc_name
-        os.makedirs(doc_build_dir, exist_ok=True)
-        os.makedirs(doc_dist_dir, exist_ok=True)
-        
-        # Clean document-specific output directories
-        for dir_path in [doc_build_dir, doc_dist_dir]:
-            for file in os.listdir(dir_path):
-                file_path = dir_path / file
-                if os.path.isfile(file_path):
-                    os.remove(file_path)
-
-        # Prepare paths for the document processor
-        paths = {
-            "doc_dir": doc_path,
-            "templates_dir": doc_path / "templates",
-            "pages_dir": doc_path / "pages",
-            "images_dir": doc_path / "images",
-            "build_dir": doc_build_dir,
-            "dist_dir": doc_dist_dir,
-        }
-
-        # Prepare Jinja environment
-        jinja_env = create_env(paths["templates_dir"])
-
-        # Process the document
-        module.process_document(paths, merged_config, jinja_env)
-        
+        return False
+
+    bake_path = doc_path / "bake.py"
+    if not bake_path.exists():
+        print(f'Warning: bake.py missing for document "{doc_name}" - skipping')
+        return False
+
+    config_yml_path = doc_path / "config.yml"
+    if not config_yml_path.exists():
+        print(f'Warning: config.yml missing for document "{doc_name}" - skipping')
+        return False
+
+    return bake_path, config_yml_path
+
+
+def _load_document_bake_module(doc_name, bake_path):
+    """Load the document's bake.py module."""
+    doc_bake = importlib.util.spec_from_file_location(
+        f"documents.{doc_name}.bake", bake_path
+    )
+    module = importlib.util.module_from_spec(doc_bake)
+    doc_bake.loader.exec_module(module)
+    return module
+
+
+def _setup_document_directories(build_dir, dist_dir, doc_name):
+    """Set up and clean document-specific build and dist directories."""
+    doc_build_dir = build_dir / doc_name
+    doc_dist_dir = dist_dir / doc_name
+    os.makedirs(doc_build_dir, exist_ok=True)
+    os.makedirs(doc_dist_dir, exist_ok=True)
+
+    # Clean document-specific output directories
+    for dir_path in [doc_build_dir, doc_dist_dir]:
+        for file in os.listdir(dir_path):
+            file_path = dir_path / file
+            if os.path.isfile(file_path):
+                os.remove(file_path)
+
+    return doc_build_dir, doc_dist_dir
+
+
+def _process_document(doc_name, doc_path, config, build_dir, dist_dir):
+    """Process an individual document."""
+    validation_result = _validate_document(doc_name, doc_path)
+    if not validation_result:
+        return
+
+    print(f'Processing document "{doc_name}" from {doc_path}...')
+    bake_path, config_yml_path = validation_result
+    bake_module = _load_document_bake_module(doc_name, bake_path)
+    with open(config_yml_path, encoding="utf-8") as f:
+        doc_config = yaml.safe_load(f)
+
+    doc_build_dir, doc_dist_dir = _setup_document_directories(
+        build_dir, dist_dir, doc_name
+    )
+    paths = {
+        "doc_dir": doc_path,
+        "templates_dir": doc_path / "templates",
+        "pages_dir": doc_path / "pages",
+        "images_dir": doc_path / "images",
+        "build_dir": doc_build_dir,
+        "dist_dir": doc_dist_dir,
+    }
+
+    bake_module.process_document(
+        paths=paths,
+        config=_deep_merge(config, doc_config),
+        jinja_env=create_env(paths["templates_dir"]),
+    )
+
+
+def main(config_path=None):
+    """Main function for the document generator."""
+    config_path = _get_config_path(config_path)
+    if config_path is None:
+        return 1
+
+    config = _load_config(config_path)
+
+    base_dir = config_path.parent
+    build_dir, dist_dir = _setup_output_directories(base_dir)
+
+    documents = config.get("documents", [])
+    document_paths = _resolve_document_paths(base_dir, documents)
+
+    for doc_name, doc_path in document_paths.items():
+        _process_document(doc_name, doc_path, config, build_dir, dist_dir)
+
     return 0
 
 

src/pdfbaker/common.py

@@ -1,14 +1,17 @@
 """Common functionality for document generation."""
 
 import subprocess
-from pathlib import Path
 
-# TODO: Switch to CairoSVG (Ubuntu installs 2.7.1)
-# as soon as the SVG templates are tidied up (update README)
-# from cairosvg import svg2pdf
 import pypdf
 import yaml
 
+try:
+    from cairosvg import svg2pdf
+
+    CAIROSVG_AVAILABLE = True
+except ImportError:
+    CAIROSVG_AVAILABLE = False
+
 
 def load_pages(pages_dir):
     """Load page configurations from a specific subdirectory."""
@@ -70,26 +73,22 @@ def convert_svg_to_pdf(svg_path, pdf_path, backend="cairosvg"):
     """
     if backend == "inkscape":
         return _convert_with_inkscape(svg_path, pdf_path)
-    else:  # Default to cairosvg
-        return _convert_with_cairosvg(svg_path, pdf_path)
+    return _convert_with_cairosvg(svg_path, pdf_path)
 
 
 def _convert_with_cairosvg(svg_path, pdf_path):
     """Convert SVG to PDF using CairoSVG."""
-    try:
-        # Only import cairosvg when needed
-        from cairosvg import svg2pdf
-        
-        with open(svg_path, 'rb') as svg_file:
-            svg2pdf(file_obj=svg_file, write_to=pdf_path)
-        
-        return pdf_path
-    except ImportError:
+    if not CAIROSVG_AVAILABLE:
         raise ImportError(
             "CairoSVG is not installed. Please install it with 'pip install cairosvg' "
             "or set svg2pdf_backend to 'inkscape' in your config."
         )
 
+    with open(svg_path, "rb") as svg_file:
+        svg2pdf(file_obj=svg_file, write_to=pdf_path)
+
+    return pdf_path
+
 
 def _convert_with_inkscape(svg_path, pdf_path):
     """Convert SVG to PDF using Inkscape."""
@@ -103,8 +102,8 @@ def _convert_with_inkscape(svg_path, pdf_path):
             check=True,
         )
         return pdf_path
-    except (subprocess.SubprocessError, FileNotFoundError):
+    except (subprocess.SubprocessError, FileNotFoundError) as exc:
         raise RuntimeError(
-            "Inkscape command failed. Please ensure Inkscape is installed and in your PATH "
-            "or set svg2pdf_backend to 'cairosvg' in your config."
-        )
+            "Inkscape command failed. Please ensure Inkscape is installed "
+            "and in your PATH or set svg2pdf_backend to 'cairosvg' in your config."
+        ) from exc

src/pdfbaker/render.py

@@ -152,7 +152,7 @@ def encode_image(filename, images_dir):
     image_path = os.path.join(images_dir, filename)
     if not os.path.exists(image_path):
         raise FileNotFoundError(f"Image not found: {image_path}")
-        
+
     with open(image_path, "rb") as f:
         binary_fc = f.read()
         base64_utf8_str = base64.b64encode(binary_fc).decode("utf-8")