webcache/cache.py

import web
import cPickle as pickle
from time import gmtime
from datetime import datetime, timedelta
from os import mkdir
from os.path import getmtime, exists, dirname, realpath
from hashlib import md5
from copy import copy


ADMINISTRATION_FILE = 'administration.py'


def to_dir(path):
    """
    Transform a path to an absolute directory, that can be used as prefix
    before a filename.
    """
    return realpath(dirname(path)) + '/'


def assert_file_exists(path):
    if not exists(path):
        raise IOError('File "%s" does not exist.' % path)


def seconds_to_datetime(seconds):
    return datetime(*gmtime(seconds)[:7])


class Cache:
    """
    A Cache instance represents a single file in the cache directory, which is
    a concatenation of the file list.
    """

    def __init__(self, root='', files=[], cached='cached',
            expires={'days': 365}):
        """
        The constructor takes the following arguments (all are optional):
        1. The directory in which the cached files are located (empty by
           default).
        2. An initial list of number of files to include in the cahce file.
        3. The cached files directory (defaults to 'cached').
        4. A dictionary containing arguments to the timedelta constructor, that
           indicates how long the cache object sould live.
        """
        self.root = to_dir(root + '/')
        self.cached = to_dir(cached + '/')
        self.files = map(lambda f: self.root + f, files)
        self.expires = expires

        self.assert_files_exist()

    def assert_files_exist(self):
        map(assert_file_exists, self.files)

    def __str__(self):
        return '<Cache filename=%s files=[%s]>' \
               % (self.filename(), ','.join(self.files))

    def add(self, path, absolute=False):
        """
        Add a file to the cache object. Requires the full path to the file,
        relative to the root directory. If the second argument is True, the
        root directory will not be prepended to the path.
        """
        if not absolute:
            path = self.root + path

        assert_file_exists(path)
        self.files.append(path)

    def remove(self, path, absolute=False):
        """
        Remove a file from the cache object. Requires the full path to the
        file, relative to the root directory. If the second argument is True,
        the root directory will not be prepended to the path.
        """
        if not absolute:
            path = self.root + path

        self.files.remove(path)

    def assert_modification_dates_exist(self):
        """
        Get the latest modification dates fo each file in the cache object.
        """
        if hasattr(self, 'modified'):
            return

        self.modified = {}

        for path in self.files:
            self.modified[path] = getmtime(path)

    def assert_cached_folder_exist(self):
        """
        Assert that the folder for cached files is created.
        """
        if not exists(self.cached):
            mkdir(self.cached)

    def save_administration(self, old_admin={}):
        """
        Generate a Python file containing the modification dates of the cached
        file list. The old_admin paramter can contain additional files which
        are not in this cache object, but do need to be kept in the
        administration for other cache objects.
        """
        self.assert_modification_dates_exist()
        self.assert_cached_folder_exist()

        admin = copy(old_admin)
        admin.update(self.modified)

        f = open(self.cached + ADMINISTRATION_FILE, 'w')
        pickle.dump(self.modified, f)
        f.close()

    def load_administration(self):
        path = self.cached + ADMINISTRATION_FILE

        if not exists(path):
            return {}

        f = open(path, 'r')
        modified = pickle.load(f)
        f.close()

        return modified

    def last_modified(self):
        self.assert_modification_dates_exist()

        return self.modified[max(self.modified)]

    def etag(self):
        """
        Generate an Etag for the cache object, using the names of the files
        included and the latest modification date.
        """
        h = md5()
        h.update(','.join(self.files) + str(self.last_modified()))

        return h.hexdigest()

    def filename(self):
        h = md5()
        h.update(','.join(self.files))

        return self.cached + h.hexdigest()

    def content(self):
        content = ''

        for path in self.files:
            f = open(path, 'r')
            content += f.read()
            f.close()

        return content

    def output(self):
        """"""
        # Update cached file
        last_modified = self.last_modified()
        path = self.filename()
        admin = self.load_administration()

        if not exists(path):
            web.debug('Cached file "%s" does not exist yet, generating it...')
            server_modified = True
        else:
            server_modified = False

            for f_path, f_modified in self.modified.iteritems():
                if f_path not in admin:
                    web.debug('File "%s" has been added.' % f_path)
                    server_modified = True
                elif f_modified > admin[f_path]:
                    web.debug('File "%s" has been updated.' % f_path)
                    server_modified = True

        if server_modified:
            self.save_administration(admin)

            content = self.content()
            f = open(path, 'w')
            f.write(content)
            f.close()

        try:
            web.http.modified(seconds_to_datetime(last_modified), self.etag())
            web.http.expires(timedelta(**self.expires))
            web.header('Cache-Control', 'private')

            if not server_modified:
                # Concatenated content has not been loaded yet, read the cached
                # file
                web.debug('Cached file "%s" already exists, sending content...')
                f = open(path, 'r')
                content = f.read()
                f.close()

            return content
        except web.NotModified as e:
            web.debug('Cached file "%s" not modified, setting 304 header...' \
                      % path)
            raise e