sync update: moving to ./sync.py as entry point

This commit is contained in:
tmshv committed 2016-08-05 17:54:00 +03:00
1 parent 6f144fb6de
commit 13a3a02613
8 files changed
+456 -83

No files matched your search

+16 -15
View File
@@ -22,7 +22,7 @@ class SyncDocument(Sync):
file = document['file']
filename = os.path.basename(file)
document['type'] = self.document_type(document)
document['type'] = document_type(document)
document['preview'] = self.create_preview(file, sizes=self.sizes, preview_dir=self.dir_static_previews)
document['file'] = {
'name': filename,
@@ -31,11 +31,6 @@ class SyncDocument(Sync):
return document
def document_type(self, document):
if document['category'] == 'Награды':
return 'award'
return 'document'
def create_id(self, document):
bn = os.path.basename(document['file'])
document['id'] = url_encode_text('{category}-{file}'.format(
@@ -56,23 +51,29 @@ class SyncDocument(Sync):
return document
def create_preview(self, file, sizes, preview_dir):
temp_preview_path = pdf_to_jpg(self.provider, file)
def url(size, ext):
return self.url_base_preview.format(id=file, size=size, ext=ext)
temp_preview_path = pdf_to_jpg(self.provider, file)
file = os.path.basename(file)
file = url_encode_text(file)
url = lambda size, ext: self.url_base_preview.format(id=file, size=size, ext=ext)
img = create_image(temp_preview_path, sizes, url, preview_dir)
os.remove(temp_preview_path)
return img
def document_type(document):
if document['category'] == 'Награды':
return 'award'
return 'document'
def pdf_to_jpg(provider, pdf):
# cwd = os.getcwd()
# abspdf = os.path.join(cwd, pdf)
abspdf = provider.get_abs(pdf)
_, temp = mkstemp('.jpg')
image_magick_pdf_to_img(abspdf, temp)
return temp
_, temp_in = mkstemp(suffix='.pdf')
_, temp_out = mkstemp(suffix='.jpg')
abspdf = provider.copy(pdf, temp_in)
image_magick_pdf_to_img(abspdf, temp_out)
os.remove(temp_in)
return temp_out
+11 -3
View File
@@ -1,4 +1,5 @@
import os
from io import BytesIO
class Provider:
@@ -6,8 +7,8 @@ class Provider:
super().__init__()
self.root = root
def open(self, path, mode):
return ''
def read(self, path):
return BytesIO(bytes())
def scan(self, path):
return []
@@ -15,6 +16,9 @@ class Provider:
def glob(self, pattr):
return []
def type_filter(self, path, ext):
return []
def copy(self, path, out):
pass
@@ -29,4 +33,8 @@ class Provider:
return os.path.normpath(abs)
def get_rel(self, path):
return os.path.relpath(path, self.root)
rel = os.path.relpath(path, self.root)
return rel
def is_dir(self, path):
return False
+20 -4
View File
@@ -4,18 +4,31 @@ from glob import glob
from shutil import copyfile
from sync.data import Provider
from utils.fn import lmap
from utils.hash import hash_file
class FSProvider(Provider):
def open(self, path, mode):
return open(self.get_abs(path), mode)
def open(self, path):
return open(self.get_abs(path), 'rb')
def scan(self, path):
return os.scandir(self.get_abs(path))
return lmap(
self.get_rel,
map(
lambda i: i.path,
os.scandir(self.get_abs(path))
)
)
def glob(self, pattr):
return glob(self.get_abs(pattr))
return lmap(self.get_rel, glob(self.get_abs(pattr)))
def type_filter(self, path, ext):
return list(filter(
lambda fn: os.path.splitext(fn)[1] == ext,
self.scan(path)
))
def hash(self, path):
return hash_file(self.get_abs(path))
@@ -28,3 +41,6 @@ class FSProvider(Provider):
file = self.get_abs(path)
return os.path.getsize(file)
def is_dir(self, path):
return os.path.isdir(self.get_abs(path))
+64
View File
@@ -0,0 +1,64 @@
import os
import re
import io
import requests
from sync.data import Provider
from utils.fn import lmap
from yandexdisk import YDClient
class YDProvider(Provider):
def __init__(self, root, access_token):
super().__init__(root)
self.api = YDClient(access_token)
def read(self, path):
remote = self.api.get_download_link(self.get_abs(path))
url = remote['href']
req = requests.get(url)
data = req.content
return io.BytesIO(data)
def scan(self, path):
res = self.api.get_content(self.get_abs(path))
files = []
for item in res['_embedded']['items']:
files.append(item['path'])
files = map(
lambda fp: re.sub('disk:', '', fp),
files
)
return lmap(self.get_rel, files)
def glob(self, pattr):
return []
def type_filter(self, path, ext):
return list(filter(
lambda fn: os.path.splitext(fn)[1] == ext,
self.scan(path)
))
def hash(self, path):
res = self.api.get_content(self.get_abs(path))
return res['sha256']
def copy(self, path, out):
data = self.read(path)
f = open(out, 'wb')
f.write(data.read())
f.close()
return out
def size(self, path):
res = self.api.get_content(self.get_abs(path))
return res['size']
def is_dir(self, path):
res = self.api.get_content(self.get_abs(path))
return res['type'] == 'dir'
+50 -58
View File
@@ -5,17 +5,41 @@ import settings
from db import db
from sync import synced_image_id, untouched
from sync.core.document import SyncDocument
from sync.data.fs import FSProvider
from utils.fn import combine, lmap, lprint_json, lprint, key_mapper, lmapfn, kmap
from utils.fn import combine, lmap, key_mapper, lmapfn
from utils.io import read_yaml
def sync_documents(update=True, delete=True):
provider = FSProvider(settings.dir_documents)
def sync_documents(provider, update=True, delete=True):
"""
DOCUMENT_OBJECT SAMPLE
{
"hash": "c6a2aaf9a393c683c2250dd916115b7056ef16f023e0cd348c184a85cde9c9a0",
"_id": {
"$oid": "56d08411ace9573958e5e497"
},
"type": "document",
"id": "dlya-postupayuschih-dogovor-o-pozhertvovanii-pdf",
"category": "Для поступающих",
"file": {
"size": 53088,
"name": "Договор о пожертвовании.pdf"
},
"url": "https://static.shburg.org/art/uploads/dogovor-o-pozhertvovanii.pdf",
"preview": {
"$oid": "56d08007ace9573958e5e48f"
},
"title": "Договор о пожертвовании"
}
:param provider:
:param update:
:param delete:
:return:
"""
static_path = static_path_fn(settings.dir_static_uploads)
u, d = main(
return main(
SyncDocument(
db()[settings.collection_documents],
provider,
@@ -29,23 +53,17 @@ def sync_documents(update=True, delete=True):
delete_documents=delete
)
print('DELETE DOCUMENTS: %s' % ('NO' if not delete else str(len(d))))
lprint_json(d)
print('UPDATE DOCUMENTS: %s' % ('NO' if not update else str(len(u))))
lprint_json(u)
def main(sync, static_path, update_documents, delete_documents):
documents = get_documents(sync)
documents_id = lmap(
lambda i: i['id'],
lambda document: document['id'],
documents
)
# CHECKING
file_names = set(lmap(
lambda i: i['file'],
lambda document: document['file'],
documents
))
if len(file_names) != len(documents):
@@ -70,7 +88,7 @@ def main(sync, static_path, update_documents, delete_documents):
if update_documents:
# REPLACE PREVIEW_OBJECT WITH IT _ID IN MONGODB
documents = lmap(
key_mapper('preview', lambda i: synced_image_id(i)),
key_mapper('preview', lambda img: synced_image_id(img)),
documents
)
@@ -86,7 +104,7 @@ def main(sync, static_path, update_documents, delete_documents):
documents_delete = lmap(
sync.delete,
map(
lambda i: {'_id': i['_id']},
lambda document: {'_id': document['_id']},
documents_delete
)
)
@@ -107,12 +125,12 @@ def get_documents(sync):
)
# CHOICE BETTER TITLE
documents = lmapfn(documents)(
lambda i: {
**i,
'title': until_none([i['title'], get_pdf_title(sync.provider, i['file'])])
}
)
# documents = lmapfn(documents)(
# lambda i: {
# **i,
# 'title': until_none([i['title'], get_pdf_title(sync.provider, i['file'])])
# }
# )
documents = lmap(sync.create_id, documents)
documents = lmap(sync.create_hash, documents)
@@ -121,8 +139,8 @@ def get_documents(sync):
return documents
def static_path_fn(dir):
return lambda document: os.path.join(dir, os.path.basename(document['url']))
def static_path_fn(dir_path):
return lambda document: os.path.join(dir_path, os.path.basename(document['url']))
def categorize_files_list(files, title):
@@ -177,21 +195,20 @@ def documents_from_yaml(dirpath):
def documents_from_subdirs(provider):
documents = lmap(
lambda i: i.path,
filter(
lambda i: i.is_dir(),
provider.scan('.')
)
)
documents = provider.scan('.')
documents = list(filter(
lambda i: provider.is_dir(i),
documents
))
documents = lmap(
lambda folder: lmap(
lambda path: {
'file': provider.get_rel(path),
# 'file': provider.get_rel(path),
'file': path,
'category': os.path.basename(folder)
},
provider.glob(folder + '/*.pdf')
provider.type_filter(folder, '.pdf')
),
documents
)
@@ -205,7 +222,7 @@ def get_pdf_title(provider, file):
from PyPDF2.generic import TextStringObject
from PyPDF2.generic import IndirectObject
pdf = PdfFileReader(provider.open(file, 'rb'))
pdf = PdfFileReader(provider.read(file))
info = pdf.getDocumentInfo()
if info:
@@ -224,28 +241,3 @@ def until_none(ls):
if (i is not None) and i != '':
better = i
return better
if __name__ == '__main__':
sync_documents()
# sync_documents(delete=False, update=False)
# DOCUMENT_OBJECT SAMPLE
# {
# "hash": "c6a2aaf9a393c683c2250dd916115b7056ef16f023e0cd348c184a85cde9c9a0",
# "_id": {
# "$oid": "56d08411ace9573958e5e497"
# },
# "type": "document",
# "id": "dlya-postupayuschih-dogovor-o-pozhertvovanii-pdf",
# "category": "Для поступающих",
# "file": {
# "size": 53088,
# "name": "Договор о пожертвовании.pdf"
# },
# "url": "https://static.shburg.org/art/uploads/dogovor-o-pozhertvovanii.pdf",
# "preview": {
# "$oid": "56d08007ace9573958e5e48f"
# },
# "title": "Договор о пожертвовании"
# }