src/catalogue/management/__init__.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # This file is part of FNP-Redakcja, licensed under GNU Affero GPLv3 or later.
   4 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   5 #
   6 from collections import defaultdict
   7 from django.db import transaction
   8 from lxml import etree
   9
  10
  11 class XmlUpdater(object):
  12     """A base class for massive XML updates.
  13
  14     In a subclass, override `fix_tree` and/or use `fixes_field` decorator.
  15     Attributes:
  16     * commit_desc: commits description
  17     * retain_publishable: set publishable if head is (default: True)
  18     * only_first_chunk: process only first chunks of books (default: False)
  19     """
  20     commit_desc = "auto-update"
  21     retain_publishable = True
  22     only_first_chunk = False
  23
  24     _element_fixers = defaultdict(list)
  25
  26     def __init__(self):
  27         self.counters = defaultdict(lambda: 0)
  28
  29     @classmethod
  30     def fixes_elements(cls, xpath):
  31         """Decorator, registering a function as a fixer for given field type.
  32
  33         Any decorated function will be called like
  34             f(element, change=..., verbose=...)
  35         providing changeset as context.
  36
  37         :param xpath: element lookup, e.g. ".//{namespace-uri}tag-name"
  38         :returns: True if anything changed
  39         """
  40         def wrapper(fixer):
  41             cls._element_fixers[xpath].append(fixer)
  42             return fixer
  43         return wrapper
  44
  45     def fix_tree(self, tree, verbose):
  46         """Override to provide general tree-fixing mechanism.
  47
  48         :param tree: the parsed XML tree
  49         :param verbose: verbosity level
  50         :returns: True if anythig changed
  51         """
  52         return False
  53
  54     def fix_chunk(self, chunk, user, verbose=0, dry_run=False):
  55         """Runs the update for a single chunk."""
  56         if verbose >= 2:
  57             print(chunk.get_absolute_url())
  58         old_head = chunk.head
  59         src = old_head.materialize()
  60         try:
  61             tree = etree.fromstring(src)
  62         except:
  63             if verbose:
  64                 print("%s: invalid XML" % chunk.get_absolute_url())
  65             self.counters['Bad XML'] += 1
  66             return
  67
  68         dirty = False
  69         # Call the general fixing function.
  70         if self.fix_tree(tree, verbose=verbose):
  71             dirty = True
  72         # Call the registered fixers.
  73         for xpath, fixers in self._element_fixers.items():
  74             for elem in tree.findall(xpath):
  75                 for fixer in fixers:
  76                     if fixer(elem, change=old_head, verbose=verbose):
  77                         dirty = True
  78
  79         if not dirty:
  80             self.counters['Clean'] += 1
  81             return
  82
  83         if not dry_run:
  84             new_head = chunk.commit(
  85                 etree.tostring(tree, encoding='unicode'),
  86                 author=user,
  87                 description=self.commit_desc
  88             )
  89             if self.retain_publishable:
  90                 if old_head.publishable:
  91                     new_head.set_publishable(True)
  92         if verbose >= 2:
  93             print("done")
  94         self.counters['Updated chunks'] += 1
  95
  96     def run(self, user, verbose=0, dry_run=False, books=None):
  97         """Runs the actual update."""
  98         if books is None:
  99             from catalogue.models import Book
 100             books = Book.objects.all()
 101
 102         # Start transaction management.
 103         with transaction.atomic():
 104             for book in books:
 105                 self.counters['All books'] += 1
 106                 chunks = book.chunk_set.all()
 107                 if self.only_first_chunk:
 108                     chunks = chunks[:1]
 109                 for chunk in chunks:
 110                     self.counters['All chunks'] += 1
 111                     self.fix_chunk(chunk, user, verbose, dry_run)
 112
 113     def print_results(self):
 114         """Prints the counters."""
 115         for item in sorted(self.counters.items()):
 116             print("%s: %d" % item)