Add Book.ancestor m2m.
[wolnelektury.git] / apps / catalogue / utils.py
index d1cee50..df7e438 100644 (file)
@@ -1,9 +1,353 @@
 # -*- coding: utf-8 -*-
+# This file is part of Wolnelektury, licensed under GNU Affero GPLv3 or later.
+# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
+#
+from __future__ import with_statement
 
+import hashlib
+import random
+import re
+import time
+from base64 import urlsafe_b64encode
+
+from django.http import HttpResponse
+from django.core.files.uploadedfile import UploadedFile
+from django.core.files.storage import DefaultStorage
+from django.utils.encoding import force_unicode
+from django.utils.translation import get_language
+from django.conf import settings
+from os import mkdir, path, unlink
+from errno import EEXIST, ENOENT
+from fcntl import flock, LOCK_EX
+from zipfile import ZipFile
+
+from reporting.utils import read_chunks
+
+# Use the system (hardware-based) random number generator if it exists.
+if hasattr(random, 'SystemRandom'):
+    randrange = random.SystemRandom().randrange
+else:
+    randrange = random.randrange
+MAX_SESSION_KEY = 18446744073709551616L     # 2 << 63
+
+
+def get_random_hash(seed):
+    sha_digest = hashlib.sha1('%s%s%s%s' %
+        (randrange(0, MAX_SESSION_KEY), time.time(), unicode(seed).encode('utf-8', 'replace'),
+        settings.SECRET_KEY)).digest()
+    return urlsafe_b64encode(sha_digest).replace('=', '').replace('_', '-').lower()
+
+
+def split_tags(tags, initial=None):
+    if initial is None:
+        result = {}
+    else:
+        result = initial
 
-def split_tags(tags):
-    result = {}
     for tag in tags:
         result.setdefault(tag.category, []).append(tag)
     return result
 
+
+class ExistingFile(UploadedFile):
+
+    def __init__(self, path, *args, **kwargs):
+        self.path = path
+        super(ExistingFile, self).__init__(*args, **kwargs)
+
+    def temporary_file_path(self):
+        return self.path
+
+    def close(self):
+        pass
+
+
+class LockFile(object):
+    """
+    A file lock monitor class; createas an ${objname}.lock
+    file in directory dir, and locks it exclusively.
+    To be used in 'with' construct.
+    """
+    def __init__(self, dir, objname):
+        self.lockname = path.join(dir, objname + ".lock")
+
+    def __enter__(self):
+        self.lock = open(self.lockname, 'w')
+        flock(self.lock, LOCK_EX)
+
+    def __exit__(self, *err):
+        try:
+            unlink(self.lockname)
+        except OSError as oe:
+            if oe.errno != EEXIST:
+                raise oe
+        self.lock.close()
+
+
+#@task
+def create_zip(paths, zip_slug):
+    """
+    Creates a zip in MEDIA_ROOT/zip directory containing files from path.
+    Resulting archive filename is ${zip_slug}.zip
+    Returns it's path relative to MEDIA_ROOT (no initial slash)
+    """
+    # directory to store zip files
+    zip_path = path.join(settings.MEDIA_ROOT, 'zip')
+
+    try:
+        mkdir(zip_path)
+    except OSError as oe:
+        if oe.errno != EEXIST:
+            raise oe
+    zip_filename = zip_slug + ".zip"
+
+    with LockFile(zip_path, zip_slug):
+        if not path.exists(path.join(zip_path, zip_filename)):
+            zipf = ZipFile(path.join(zip_path, zip_filename), 'w')
+            try:
+                for arcname, p in paths:
+                    if arcname is None:
+                        arcname = path.basename(p)
+                    zipf.write(p, arcname)
+            finally:
+                zipf.close()
+
+        return 'zip/' + zip_filename
+
+
+def remove_zip(zip_slug):
+    """
+    removes the ${zip_slug}.zip file from zip store.
+    """
+    zip_file = path.join(settings.MEDIA_ROOT, 'zip', zip_slug + '.zip')
+    try:
+        unlink(zip_file)
+    except OSError as oe:
+        if oe.errno != ENOENT:
+            raise oe
+
+
+class AttachmentHttpResponse(HttpResponse):
+    """Response serving a file to be downloaded.
+    """
+    def __init__(self, file_path, file_name, mimetype):
+        super(AttachmentHttpResponse, self).__init__(mimetype=mimetype)
+        self['Content-Disposition'] = 'attachment; filename=%s' % file_name
+        self.file_path = file_path
+        self.file_name = file_name
+
+        with open(DefaultStorage().path(self.file_path)) as f:
+            for chunk in read_chunks(f):
+                self.write(chunk)
+
+class MultiQuerySet(object):
+    def __init__(self, *args, **kwargs):
+        self.querysets = args
+        self._count = None
+
+    def count(self):
+        if not self._count:
+            self._count = sum(len(qs) for qs in self.querysets)
+        return self._count
+
+    def __len__(self):
+        return self.count()
+
+    def __getitem__(self, item):
+        try:
+            indices = (offset, stop, step) = item.indices(self.count())
+        except AttributeError:
+            # it's not a slice - make it one
+            return self[item : item + 1][0]
+        items = []
+        total_len = stop - offset
+        for qs in self.querysets:
+            if len(qs) < offset:
+                offset -= len(qs)
+            else:
+                items += list(qs[offset:stop])
+                if len(items) >= total_len:
+                    return items
+                else:
+                    offset = 0
+                    stop = total_len - len(items)
+                    continue
+
+class SortedMultiQuerySet(MultiQuerySet):
+    def __init__(self, *args, **kwargs):
+        self.order_by = kwargs.pop('order_by', None)
+        self.sortfn = kwargs.pop('sortfn', None)
+        if self.order_by is not None:
+            self.sortfn = lambda a, b: cmp((getattr(a, f) for f in self.order_by),
+                                           (getattr(b, f) for f in self.order_by))
+        super(SortedMultiQuerySet, self).__init__(*args, **kwargs)
+
+    def __getitem__(self, item):
+        sort_heads = [0] * len(self.querysets)
+        try:
+            indices = (offset, stop, step) = item.indices(self.count())
+        except AttributeError:
+            # it's not a slice - make it one
+            return self[item : item + 1][0]
+        items = []
+        total_len = stop - offset
+        skipped = 0
+        i_s = range(len(sort_heads))
+
+        while len(items) < total_len:
+            candidate = None
+            candidate_i = None
+            for i in i_s:
+                def get_next():
+                    return self.querysets[i][sort_heads[i]]
+                try:
+                    if candidate is None:
+                        candidate = get_next()
+                        candidate_i = i
+                    else:
+                        competitor = get_next()
+                        if self.sortfn(candidate, competitor) > 0:
+                            candidate = competitor
+                            candidate_i = i
+                except IndexError:
+                    continue # continue next sort_head
+            # we have no more elements:
+            if candidate is None:
+                break
+            sort_heads[candidate_i] += 1
+            if skipped < offset:
+                skipped += 1
+                continue # continue next item
+            items.append(candidate)
+
+        return items
+
+
+def truncate_html_words(s, num, end_text='...'):
+    """Truncates HTML to a certain number of words (not counting tags and
+    comments). Closes opened tags if they were correctly closed in the given
+    html. Takes an optional argument of what should be used to notify that the
+    string has been truncated, defaulting to ellipsis (...).
+
+    Newlines in the HTML are preserved.
+
+    This is just a version of django.utils.text.truncate_html_words with no space before the end_text.
+    """
+    s = force_unicode(s)
+    length = int(num)
+    if length <= 0:
+        return u''
+    html4_singlets = ('br', 'col', 'link', 'base', 'img', 'param', 'area', 'hr', 'input')
+    # Set up regular expressions
+    re_words = re.compile(r'&.*?;|<.*?>|(\w[\w-]*)', re.U)
+    re_tag = re.compile(r'<(/)?([^ ]+?)(?: (/)| .*?)?>')
+    # Count non-HTML words and keep note of open tags
+    pos = 0
+    end_text_pos = 0
+    words = 0
+    open_tags = []
+    while words <= length:
+        m = re_words.search(s, pos)
+        if not m:
+            # Checked through whole string
+            break
+        pos = m.end(0)
+        if m.group(1):
+            # It's an actual non-HTML word
+            words += 1
+            if words == length:
+                end_text_pos = pos
+            continue
+        # Check for tag
+        tag = re_tag.match(m.group(0))
+        if not tag or end_text_pos:
+            # Don't worry about non tags or tags after our truncate point
+            continue
+        closing_tag, tagname, self_closing = tag.groups()
+        tagname = tagname.lower()  # Element names are always case-insensitive
+        if self_closing or tagname in html4_singlets:
+            pass
+        elif closing_tag:
+            # Check for match in open tags list
+            try:
+                i = open_tags.index(tagname)
+            except ValueError:
+                pass
+            else:
+                # SGML: An end tag closes, back to the matching start tag, all unclosed intervening start tags with omitted end tags
+                open_tags = open_tags[i+1:]
+        else:
+            # Add it to the start of the open tags list
+            open_tags.insert(0, tagname)
+    if words <= length:
+        # Don't try to close tags if we don't need to truncate
+        return s
+    out = s[:end_text_pos]
+    if end_text:
+        out += end_text
+    # Close any tags still open
+    for tag in open_tags:
+        out += '</%s>' % tag
+    # Return string
+    return out
+
+
+def customizations_hash(customizations):
+    customizations.sort()
+    return hash(tuple(customizations))
+
+
+def get_customized_pdf_path(book, customizations):
+    """
+    Returns a MEDIA_ROOT relative path for a customized pdf. The name will contain a hash of customization options.
+    """
+    h = customizations_hash(customizations)
+    return 'book/%s/%s-custom-%s.pdf' % (book.slug, book.slug, h)
+
+
+def clear_custom_pdf(book):
+    """
+    Returns a list of paths to generated customized pdf of a book
+    """
+    from waiter.utils import clear_cache
+    clear_cache('book/%s' % book.slug)
+
+
+class AppSettings(object):
+    """Allows specyfying custom settings for an app, with default values.
+
+    Just subclass, set some properties and instantiate with a prefix.
+    Getting a SETTING from an instance will check for prefix_SETTING
+    in project settings if set, else take the default. The value will be
+    then filtered through _more_SETTING method, if there is one.
+
+    """
+    def __init__(self, prefix):
+        self._prefix = prefix
+
+    def __getattribute__(self, name):
+        if name.startswith('_'):
+            return object.__getattribute__(self, name)
+        value = getattr(settings,
+                         "%s_%s" % (self._prefix, name),
+                         object.__getattribute__(self, name))
+        more = "_more_%s" % name
+        if hasattr(self, more):
+            value = getattr(self, more)(value)
+        return value
+
+
+def trim_query_log(trim_to=25):
+    """
+connection.queries includes all SQL statements -- INSERTs, UPDATES, SELECTs, etc. Each time your app hits the database, the query will be recorded.
+This can sometimes occupy lots of memory, so trim it here a bit.
+    """
+    if settings.DEBUG:
+        from django.db import connection
+        connection.queries = trim_to > 0 \
+            and connection.queries[-trim_to:] \
+            or []
+
+
+def delete_from_cache_by_language(cache, key_template):
+    cache.delete_many([key_template % lc for lc, ln in settings.LANGUAGES])