v/pol
1
0
mirror of https://github.com/taroved/pol synced 2025-05-16 06:10:09 -07:00

feeder time distance fix in progress

This commit is contained in:
Alexandr Nesterenko 2017-07-11 14:37:44 -07:00
parent 576da9ca7a
commit c3b4a6835d
6 changed files with 148 additions and 9 deletions

View File

@ -136,8 +136,8 @@ def downloadError(error, request=None):
request.write('Traceback: ' + error.getTraceback())
else:
request.write('Something wrong. Geek comment: ' + error.getErrorMessage())
sys.stderr.write(datetime.datetime.now())
sys.stderr.write('\n'.join('Downloader error: ' + error.getErrorMessage(), 'Traceback: ' + error.getTraceback()))
sys.stderr.write(str(datetime.now()))
sys.stderr.write('\n'.join(['Downloader error: ' + error.getErrorMessage(), 'Traceback: ' + error.getTraceback()]))
request.finish()

83
feed.py
View File

@ -13,6 +13,67 @@ from settings import DATABASES, DOWNLOADER_USER_AGENT
url_hash_regexp = re.compile('(#.*)?$')
POST_TIME_DISTANCE = 15 # minutes
FIELD_IDS = {'title': 1, 'description': 2, 'title_link': 3}
def save_post(conn, md5sum, created, feed_id, post_fields):
cur = conn.cursor()
try:
#import pdb;pdb.set_trace()
cur.execute("""insert into frontend_post (md5sum, created, feed_id)
values (%s, %s, %s)""", (md5sum.hexdigest(), created, feed_id))
finally:
print(cur._last_executed)
post_id = conn.insert_id()
for key in ['title', 'description', 'title_link']:
if key in post_fields:
try:
cur.execute("""insert into frontend_postfield (field_id, post_id, `text`)
values (%s, %s, %s)""", (FIELD_IDS[key], post_id, post_fields[key].encode('utf-8')))
finally:
print(cur._last_executed)
def fill_dates(feed_id, items):
if not items:
return []
hashes = {}
for item in items:
#create md5
h = md5('')
for key in ['title', 'description', 'title_link']:
if key in item:
h.update(item[key].encode('utf-8'))
hashes[h] = item
#fetch dates from db
fetched_dates = {}
db = get_conn()
with db:
quoted_hashes = ','.join(["'%s'" % (h.hexdigest()) for h in hashes])
cur = db.cursor()
cur.execute("""select p.md5sum, p.created, p.id
from frontend_post p
where p.md5sum in (%s)
and p.id=%s""" % (quoted_hashes, feed_id,))
rows = cur.fetchall()
print(cur._last_executed)
for row in rows:
md5hash = row[0]
created = row[1]
post_id = row[2]
fetched_dates[md5hash] = datetime.datetime.fromtimestamp(int(created))
cur_time = datetime.datetime.now()
new_posts = []
for h in hashes:
if h in fetched_dates:
hashes[h]['time'] = fetched_date[h]
else:
hashes[h]['time'] = cur_time
save_post(db, h, cur_time, feed_id, hashes[h])
cur_time -= datetime.timedelta(minutes=POST_TIME_DISTANCE)
def element_to_string(element):
s = [element.text] if element.text else []
for sub_element in element:
@ -42,11 +103,12 @@ def buildFeed(response, feed_config):
anchor = element[0].xpath('ancestor-or-self::node()[name()="a"]')
if anchor and anchor[0].get('href'):
title_link = _build_link(response.body_as_unicode(), feed_config['uri'], anchor[0].get('href'))
if len(item) == len(feed_config['fields']): # all fields are required
item['title_link'] = title_link
if title_link:
item['title_link'] = title_link
items.append(item)
title = response.selector.xpath('//title/text()').extract()
#build feed
@ -57,10 +119,13 @@ def buildFeed(response, feed_config):
"Source page url: " + feed_config['uri'],
language="en",
)
fill_dates(feed_config['id'], items)
for item in items:
title = item['title'] if 'title' in item else ''
desc = item['description'] if 'description' in item else ''
if item['title_link']:
if 'title_link' in item:
link = item['title_link']
else:
link = url_hash_regexp.sub('#' + md5((title+desc).encode('utf-8')).hexdigest(), feed_config['uri'])
@ -75,9 +140,8 @@ def buildFeed(response, feed_config):
def getFeedData(request, feed_id):
# get url, xpathes
creds = DATABASES['default']
db = MySQLdb.connect(host=creds['HOST'], port=int(creds['PORT']), user=creds['USER'], passwd=creds['PASSWORD'], db=creds['NAME'])
feed = {}
db = get_conn()
with db:
cur = db.cursor()
cur.execute("""select f.uri, f.xpath, fi.name, ff.xpath from frontend_feed f
@ -88,6 +152,7 @@ def getFeedData(request, feed_id):
for row in rows:
if not feed:
feed['id'] = feed_id
feed['uri'] = row[0]
feed['xpath'] = row[1]
feed['fields'] = {}
@ -97,3 +162,9 @@ def getFeedData(request, feed_id):
return [feed['uri'], feed]
else:
return 'Feed generator error: config of feed is empty'
def get_conn():
creds = DATABASES['default']
db = MySQLdb.connect(host=creds['HOST'], port=int(creds['PORT']), user=creds['USER'], passwd=creds['PASSWORD'], db=creds['NAME'])
db.autocommit(True)
return db

View File

@ -12,5 +12,12 @@
},
"model": "frontend.field",
"pk": 2
},
{
"fields": {
"name": "link"
},
"model": "frontend.field",
"pk": 3
}
]

View File

@ -0,0 +1,47 @@
# -*- coding: utf-8 -*-
from __future__ import unicode_literals
from django.db import migrations, models
import datetime
class Migration(migrations.Migration):
dependencies = [
('frontend', '0001_initial'),
]
operations = [
migrations.CreateModel(
name='Post',
fields=[
('id', models.AutoField(verbose_name='ID', serialize=False, auto_created=True, primary_key=True)),
('md5sum', models.CharField(max_length=32)),
('created', models.DateTimeField(auto_now_add=True)),
],
),
migrations.CreateModel(
name='PostField',
fields=[
('id', models.AutoField(verbose_name='ID', serialize=False, auto_created=True, primary_key=True)),
('text', models.TextField(max_length=65535)),
('field', models.ForeignKey(to='frontend.Field')),
('post', models.ForeignKey(to='frontend.Post')),
],
),
migrations.AddField(
model_name='feed',
name='created',
field=models.DateTimeField(default=datetime.datetime(2017, 7, 11, 21, 19, 23, 580569), auto_now_add=True),
preserve_default=False,
),
migrations.AddField(
model_name='post',
name='feed',
field=models.ForeignKey(to='frontend.Feed'),
),
migrations.AlterIndexTogether(
name='post',
index_together=set([('feed', 'md5sum')]),
),
]

View File

@ -1,9 +1,9 @@
from django.db import models
class Feed(models.Model):
uri = models.CharField(max_length=2000)
xpath = models.CharField(max_length=2000)
created = models.DateTimeField(auto_now_add=True)
class Field(models.Model):
name = models.CharField(max_length=200)
@ -12,3 +12,16 @@ class FeedField(models.Model):
feed = models.ForeignKey(Feed, on_delete=models.CASCADE)
field = models.ForeignKey(Field, on_delete=models.CASCADE)
xpath = models.CharField(max_length=2000)
class Post(models.Model):
feed = models.ForeignKey(Feed, on_delete=models.CASCADE)
md5sum = models.CharField(max_length=32)
created = models.DateTimeField(auto_now_add=True)
class Meta:
index_together = ['feed', 'md5sum']
class PostField(models.Model):
post = models.ForeignKey(Post, on_delete=models.CASCADE)
field = models.ForeignKey(Field, on_delete=models.CASCADE)
text = models.TextField(max_length=64*1024-1)

View File

@ -3,6 +3,7 @@ Django==1.8.6
lxml==3.3.3
Scrapy==1.0.3
django-pipeline==1.5.4
django-unixtimestampfield==0.3.9
mysqlclient==1.3.7
w3lib==1.12.0
feedgenerator==1.8