compress_site.py

#!/usr/bin/env python3
import sys
import os
import shutil
import subprocess

class SiteCompressor:
    """
    Compresses a static web site generated by Sphinx by:
    (1) Merging all _static directories into a single _static directory at the
        top level.
    (2) Using gzip to compress the file types below, adding a .gz suffix to
        the filename. 
    """
    # Filename extensions for files that we want to compress
    gzip_extensions = ['.html', '.css', '.js', '.woff', '.svg']
    image_extensions = ['.png', '.pdf', '.svg',] 
    
    def __init__(self, site):
        if not os.path.isdir(site):
            raise RuntimeError(
                'A Brotlifier must be instantiated with a directory')
        self.site = os.path.abspath(site)

    def merge_static_dirs(self):
        static_dirs = []
        for dirpath, dirnames, filenames in os.walk(self.site):
            if dirpath.endswith('_static') and not os.path.islink(dirpath):
                static_dirs.append(dirpath)
        main_static_dir = os.path.join(self.site, '_static')
        try:
            static_dirs.remove(main_static_dir)
        except ValueError:
            pass
        # Copy the contents of each subsidiary static dir into the main one
        # then replace the subsidiary with a symlink.
        for dirpath in static_dirs:
            # This assumes that all symlinks point inside the static dir.
            shutil.copytree(dirpath, main_static_dir, symlinks=True,
                                dirs_exist_ok=True)
            relpath = os.path.relpath(main_static_dir, os.path.dirname(dirpath))
            shutil.rmtree(dirpath)
            os.symlink(relpath, dirpath)

    def clean_images(self):
        """All images should be in _static or _images."""
        # I have no idea why these get copied rather than moved.
        # Apparently this does not happen with a standard build.
        to_delete = []
        for dirpath, dirnames, filenames in os.walk(self.site):
            if dirpath.find('_static') >= 0 or dirpath.find('_images') >= 0:
                continue
            for filename in filenames:
                base, ext = os.path.splitext(filename)
                if ext in self.image_extensions:
                    print('deleting', os.path.join(dirpath, filename))
                    to_delete.append(os.path.join(dirpath, filename))
        for path in to_delete:
            os.unlink(path)

    def compress_files(self):
        """Compress compressible files."""
        compressible = []
        for dirpath, dirnames, filenames in os.walk(self.site):
            for filename in filenames:
                base, ext = os.path.splitext(filename)
                if ext in self.gzip_extensions:
                    compressible.append(os.path.join(dirpath, filename))
        for path in compressible:
            subprocess.run(['gzip', '--best', path])

if __name__ == '__main__':
    try:
        site = sys.argv[1]
    except IndexError:
        print('Please provide the site root directory as an argument.')
        sys.exit(1)
    if not os.path.isdir(site):
        print('The site root must be a directory.')
        sys.exit(1)
    compressor = SiteCompressor(site)
    print('Merging _static directories ..')
    compressor.merge_static_dirs()
    print("Deleting extraneous images ...")
    compressor.clean_images()
    print('Compressing files ...')
    compressor.compress_files()