[Sync] + new Album sync

This commit is contained in:
Roman Timashev committed 2017-09-16 17:08:31 +03:00
1 parent 3c8f6418b5
commit 62982a9945
7 files changed
+307 -330

No files matched your search

+1 -1
View File
@@ -138,7 +138,7 @@ dir_awards = f('Awards')
dir_schedules = f('Schedules')
dir_articles = f('Articles')
dir_collective = f('Collective')
dir_gallery = f('Gallery')
dir_albums = f('Gallery')
dir_pages = f('Pages')
page_url_base = 'https://static.shlisselburg.org/art/images/page-{page}-{id}-{size}{ext}'
+6 -2
View File
@@ -1,6 +1,7 @@
import settings
from db import collection
from sync.core import Sync
from sync.core.SyncAlbum import SyncAlbum
from sync.core.SyncArticle import SyncArticle
from sync.core.SyncDocument import SyncDocument
from sync.core.SyncPage import SyncPage
@@ -68,16 +69,19 @@ async def run(run_interval=0):
provider=io(settings.dir_schedules),
collection=c(settings.collection_schedules),
)
albums = SyncAlbum(
provider=io(settings.dir_albums),
collection=c(settings.collection_albums),
)
while True:
# main(io(settings.dir_gallery), sync_albums, settings.collection_albums)
await asyncio.wait([
main(documents),
main(pages),
main(persons),
main(articles),
main(schedules),
main(albums),
])
if run_interval == 0:
+2 -2
View File
@@ -89,12 +89,12 @@ def create_date_and_title_from_folder_name(folder_name, date_formats=None):
return date, title
def create_post(md, image_path_fn):
async def create_post(md, image_path_fn):
html = lxml.html.fromstring(md)
for img in html.cssselect('img'):
src = img.get('src')
path = image_path_fn(src)
path = await image_path_fn(src)
if path:
img.set('src', path)
-169
View File
@@ -1,169 +0,0 @@
import os
import settings
from sync import create_date_and_title_from_folder_name, create_post_from_image_list, create_date, untouched, \
synced_images_ids, synced_image_id
from sync.core.album import SyncAlbum
from sync.data import list_images
from utils.fn import lmap, lmapfn, key_mapper
from utils.image import create_image
from utils.io import read_yaml_md
class ImageCreator:
def __init__(self, provider, cache):
super().__init__()
self.provider = provider
self.cache = cache if cache else set()
def create_image(self, file, sizes, url_fn, output_dir, skip_processing=False):
filehash = self.provider.hash(file)
cached = self.cached(filehash)
return create_image(self.provider.get_abs(file), sizes, url_fn, output_dir,
skip_processing=skip_processing) if not cached else cached
def cached(self, name):
return self.cache[name] if name in self.cache else None
def get_manifest(provider, filepath):
data = provider.read(filepath).read().decode('utf-8')
y, m = read_yaml_md(data)
data = {**y} if y else {}
_ = lambda key, d=None: data[key] if key in data else d
date = create_date(str(_('date')), settings.date_formats_direct)
if date:
data['date'] = date
album_id = _('id')
if album_id:
data['id'] = str(album_id)
post = {'post': m} if m else {}
return {**data, **post}
def album_from_folder(sync, path):
# PARSE FOLDER NAME -> GET OPTIONAL DATE/TITLE
date, title = create_date_and_title_from_folder_name(
os.path.basename(path)
)
images = lmap(
lambda i: os.path.relpath(i, path),
list_images(sync.provider, path)
)
# CREATE BASIC_MANIFEST BASED ON FOLDER NAME AND IMAGE CONTENT
document = {
'folder': path,
'date': date,
'title': title,
'post': create_post_from_image_list(images),
'images': images
}
# TRY TO FILL UP BASIC_MANIFEST WITH MD_MANIFEST
manifest = sync.provider.type_filter(path, '.md')
manifest = manifest[0] if len(manifest) else None
manifest = get_manifest(sync.provider, manifest) if manifest else {}
manifest = manifest if manifest else {}
document = {
**document,
**manifest
}
# CREATE DOCUMENT IDENTITY
document = sync.create_id(document)
# CREATE HASH OF DOCUMENT FILE
document = sync.create_hash(document)
return document
def main(sync, do_update, do_delete):
# GET ALBUMS FOLDERS
documents = lmap(
lambda i: i,
filter(
lambda i: sync.provider.is_dir(i),
sync.provider.scan('.')
)
)
documents = lmap(
lambda folder: album_from_folder(sync, folder),
documents
)
# CREATE SCOPE OF CURRENT SESSION
scope_documents_ids = lmapfn(documents)(
lambda i: i['id']
)
# SKIP UNTOUCHED DOCUMENTS
documents = untouched(documents, sync)
# MAP EVENT MANIFEST -> EVENT_OBJECT
documents = lmap(
sync.create_album,
documents
)
# # FILTER INCORRECT EVENTS
documents = list(filter(None, documents))
# REPLACE IMAGES OBJECTS WITH IT _ID IN MONGODB
documents = lmap(
key_mapper('images', synced_images_ids),
documents
)
documents = lmap(
key_mapper('preview', synced_image_id),
documents
)
# SYNC OBJECT WITH DB
if do_update:
documents = lmap(
sync.update,
documents
)
documents_to_remove = sync.query({'id': {'$nin': scope_documents_ids}})
if do_delete:
documents_to_remove = lmap(
sync.delete,
map(
lambda i: {'_id': i['_id']},
documents_to_remove
)
)
return documents, documents_to_remove
def sync_albums(provider, collection, update=True, delete=True):
"""
:param collection:
:param provider:
:param update:
:param delete:
:return:
"""
return main(
SyncAlbum(
collection,
provider,
image_creator=ImageCreator(provider, cache=None),
images_dir=settings.dir_static_images
),
do_update=update,
do_delete=delete
)
+51
View File
@@ -0,0 +1,51 @@
import logging
import settings
from sync import untouched
from sync.core import Sync
from sync.models.Album import Album
logger = logging.getLogger(settings.name + '.Album')
class SyncAlbum(Sync):
async def run(self):
logger.info('Checking for update')
documents = await Album.scan(self.provider)
documents_id = [doc.id for doc in documents]
# self.validate(documents)
logger.info('Found {} Item(s)'.format(len(documents)))
# SKIP UNTOUCHED DOCUMENTS
documents = await untouched(documents)
logger.info('Changed {} Items(s)'.format(len(documents)))
# UPDATING
if settings.update_enabled:
if len(documents) == 0:
logger.info('No Items to update')
for document in documents:
logger.info('Updating Item {}'.format(document.id))
await document.build()
await document.save()
logger.info('Updated Item {}'.format(document.id))
# DELETING
if settings.delete_enabled:
documents_delete = list(self.query({'id': {'$nin': documents_id}}))
if len(documents_delete) == 0:
logger.info('No Items to delete')
else:
for document in documents_delete:
logger.info('Deleting Item {}'.format(document['id']))
q = {'_id': document['_id']}
self.delete(q)
# await self.clean()
-139
View File
@@ -1,139 +0,0 @@
import os
from lxml import etree
import lxml.html
import settings
from sync import create_post
from sync.core import Sync
from sync.models.product import create_product
from utils.fn import lmap, iterate_iter_over_fns, combine
from utils.hash import hash_file, hash_str
from utils.text.transform import url_encode_text
def create_url(a, i, s, e):
return 'https://static.shlisselburg.org/art/images/product-{album}-{id}-{size}{ext}'.format(
album=a,
id=i,
size=s,
ext=e
)
def url_creator(aid, pid):
return lambda size, ext: create_url(aid, pid, size, ext.lower())
class SyncAlbum(Sync):
@staticmethod
def id_from_file(i):
return url_encode_text(
os.path.splitext(i)[0]
)
def __init__(self, collection, provider, image_creator, images_dir):
super().__init__()
self.collection = collection
self.provider = provider
self.image_creator = image_creator
self.skip_image_processing = False
self.images_dir = images_dir
def create_id(self, document):
if 'id' in document:
return document
y = str(document['date'].year)
document['id'] = url_encode_text('-'.join([y, document['title']]))
return document
def create_hash(self, document):
files = sorted(document['images'])
hashes = lmap(
self.provider.hash,
lmap(
lambda i: os.path.join(document['folder'], i),
files
)
)
document['hash'] = hash_str(
combine([hash_str(document)] + hashes)
)
return document
def create_image(self, file, url_fn):
ss = settings.album_image_sizes
od = self.images_dir
sp = self.skip_image_processing
return self.image_creator.create_image(file, ss, url_fn, od, skip_processing=sp)
def create_album(self, doc):
# url_base = 'https://static.shburg.org/art/image/product-{album}-{id}-{size}{ext}'
image_id = SyncAlbum.id_from_file
image_path = lambda i: os.path.join(doc['folder'], i)
if 'title' not in doc or not doc['title']:
return None
create_album_id = lambda date, title: '{date}-{title}'.format(
date=date.strftime('%Y'),
title=url_encode_text(title)
)
doc['id'] = doc['id'] if 'id' in doc else create_album_id(doc['date'], doc['title'])
if 'album' in doc:
doc['album'] = self.create_products(doc, image_path)
def process_post_image(src):
path = image_path(src)
if self.provider.exists(path):
url = url_creator(
doc['id'],
image_id(src)
)
image = self.create_image(path, url)
if image:
images.append(image)
return image['data']['big']['url']
return None
images = []
post_html = create_post(doc['post'], process_post_image)
if 'preview' in doc:
preview = os.path.join(doc['folder'], doc['preview'])
doc['preview'] = self.create_image(preview, url_creator(
doc['id'],
image_id(doc['preview'])
))
doc['post'] = post_html
doc['images'] = images
return doc
def create_products(self, doc, image_file_fn):
image_id = SyncAlbum.id_from_file
iter_album = iterate_iter_over_fns([
lambda product: {
**product,
'image': self.create_image(
image_file_fn(product['file']),
url_creator(
doc['id'],
image_id(product['file'])
)
)
}
])
return lmap(
create_product,
iter_album(doc['album'])
)
+247 -17
View File
@@ -1,21 +1,251 @@
__author__ = 'Roman Timashev'
import logging
import os
import lxml.html
import kazimir
import settings
from db import collection
from kazimir import kazimir_to_html
from sync import create_date_and_title_from_folder_name, images_from_html, create_post_from_image_list, create_date, \
create_post
from sync.data import Provider, list_images
from sync.models import Model
from sync.models.Image import Image
from utils.fn import swap_ext, first, map_cases, constant, combine, iterate_iter_over_fns
from utils.hash import hash_str
from utils.io import read_yaml_md, parse_yaml_front_matter
from utils.text.transform import url_encode_text
logger = logging.getLogger(settings.name + '.Album')
store = collection(settings.collection_albums)
class Album:
def __init__(self, type, title, uri, course, course_uri, date, teacher, comment):
super().__init__()
self.products = []
self.type = type
self.title = title
self.uri = uri
self.course = course
self.course_uri = course_uri
self.date = date
self.teacher = teacher
self.comment = comment
def find(item_id: str):
q = {'id': item_id}
try:
return store.find_one(q)
except ValueError:
pass
return None
def add_product(self, product):
self.products.append(product)
def __repr__(self, *args, **kwargs):
return '[%s by %s (%d items)]' % (self.title, self.teacher, len(self.products))
class Album(Model):
@staticmethod
async def scan(provider):
documents = [i for i in provider.scan('.') if provider.is_dir(i)]
documents = [Album.read(provider, i) for i in documents]
return documents
@staticmethod
def read(provider, path):
# PARSE FOLDER NAME -> GET OPTIONAL DATE/TITLE
date, title = create_date_and_title_from_folder_name(
os.path.basename(path)
)
images = list_images(provider, path)
images = [os.path.relpath(i, path) for i in images]
# CREATE BASIC_MANIFEST BASED ON FOLDER NAME AND IMAGE CONTENT
document = {
'folder': path,
'date': date,
'title': title,
'post': create_post_from_image_list(images),
'images': images
}
# TRY TO FILL UP BASIC_MANIFEST WITH MD_MANIFEST
manifest_file = provider.type_filter(path, '.md')
manifest_file = manifest_file[0] if len(manifest_file) else None
manifest = get_manifest(provider, manifest_file) if manifest_file else {}
manifest = manifest if manifest else {}
file = manifest_file if manifest_file else path
return Album(
provider,
file,
params={
**document,
**manifest,
},
)
def __init__(self, provider, file, params=None):
# self.__id_template: str = '{category}-{file}'
# self.__url_template: str = settings.document_url_template
# self.__url_preview_template: str = settings.document_url_preview_template
self.sizes = settings.image_sizes
self.version = '1'
self.origin = settings.origin
self.post = None
self.images = None
super().__init__(provider, store, file, params=params)
def init(self):
self.__set_id()
self.__set_hash()
def validate(self):
document = self.params
if 'date' not in document or not document['date']:
return None
if 'title' not in document or not document['title']:
return None
async def build(self, **kwargs):
params = self.params
image_id = id_from_file
images = []
create_album_id = lambda date, title: '{date}-{title}'.format(
date=date.strftime('%Y'),
title=url_encode_text(title)
)
# params['id'] = params['id'] if 'id' in params else create_album_id(params['date'], params['title'])
# if 'album' in params:
# params['album'] = self.create_products(params, image_path)
async def process_post_image(src):
url = url_creator(
self.id,
id_from_file(src)
)
image_path = os.path.join(params['folder'], src)
image = await Image.new(self.provider, image_path, self.sizes, url)
if image:
images.append(image)
return image.get_size('big')['url']
return None
post_html = await create_post(params['post'], process_post_image)
if self.has_param('preview'):
preview = self.get_param('preview')
preview = self._get_relpath(preview)
preview = await Image.new(self.provider, preview, sizes=self.sizes)
else:
# preview = 'https://art.shlisselburg.org/entrance.jpg'
preview = None
self.post = post_html
self.images = images
self.preview = preview
# sizes = kwargs['sizes']
# post, images = await create_post(self.provider, self.id, self.params, sizes)
# self.post = post
# self.images = images
#
# post, images = await create_post(self.provider, self.id, self.params, sizes)
# self.post = kazimir.html_from_tree(post)
# self.images = images
def create_products(self, doc, image_file_fn):
pass
# image_id = id_from_file
# iter_album = iterate_iter_over_fns([
# lambda product: {
# **product,
# 'image': self.create_image(
# image_file_fn(product['file']),
# url_creator(
# doc['id'],
# image_id(product['file'])
# )
# )
# }
# ])
#
# return lmap(
# create_product,
# iter_album(doc['album'])
# )
def bake(self):
preview = None if not self.preview else self.preview.id
return {
**self.params,
'images': [i.id for i in self.images],
'preview': preview,
'post': self.post,
'id': self.id,
'hash': self.hash,
'file': self.file,
'origin': self.origin,
'version': self.version,
}
def __set_id(self):
if 'id' in self.params:
self.id = self.params['id']
else:
year = str(self.params['date'].year)
title = self.params['title']
self.id = url_encode_text('%s-%s' % (year, title))
def __set_hash(self):
files = self.get_param('images')
files = [self._get_relpath(i) for i in files]
if not self.provider.is_dir(self.file):
files.append(self.file)
files = sorted(files)
files = [self.provider.hash(i) for i in files]
self.hash = hash_str(files)
self.hash = hash_str(
combine([hash_str(self.params)] + files)
)
def __str__(self):
return '<Album hash={} id={} file={}>'.format(self.hash, self.id, self.file)
def get_manifest(provider, filepath):
data = provider.read(filepath).read().decode('utf-8')
y, m = read_yaml_md(data)
data = {**y} if y else {}
_ = lambda key, d=None: data[key] if key in data else d
date = create_date(str(_('date')), settings.date_formats_direct)
if date:
data['date'] = date
album_id = _('id')
if album_id:
data['id'] = str(album_id)
post = {'post': m} if m else {}
return {**data, **post}
def create_url(a, i, s, e):
return 'https://static.shlisselburg.org/art/images/product-{album}-{id}-{size}{ext}'.format(
album=a,
id=i,
size=s,
ext=e
)
def url_creator(aid, pid):
return lambda file, size, ext: create_url(aid, pid, size, ext.lower())
def id_from_file(i):
return url_encode_text(
os.path.splitext(i)[0]
)