[ipfire.org.git] / src / backend / blog.py

#!/usr/bin/python

import datetime
import feedparser
import markdown2
import re
import textile
import tornado.gen
import unicodedata

from . import misc
from . import util
from .decorators import *

# Used to automatically link some things
link_patterns = (
	# Find bug reports
	(re.compile(r"(?:#(\d+))", re.I), r"https://bugzilla.ipfire.org/show_bug.cgi?id=\1"),

	# Email Addresses
	(re.compile(r"([a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+)"), r"mailto:\1"),

	# CVE Numbers
	(re.compile(r"(?:CVE)[\s\-](\d{4}\-\d+)"), r"https://cve.mitre.org/cgi-bin/cvename.cgi?name=\1"),
)

class Blog(misc.Object):
	def _get_post(self, query, *args):
		res = self.db.get(query, *args)

		if res:
			return Post(self.backend, res.id, data=res)

	def _get_posts(self, query, *args):
		res = self.db.query(query, *args)

		for row in res:
			yield Post(self.backend, row.id, data=row)

	def get_by_id(self, id):
		return self._get_post("SELECT * FROM blog \
			WHERE id = %s", id)

	def get_by_slug(self, slug, published=True):
		if published:
			return self._get_post("SELECT * FROM blog \
				WHERE slug = %s AND published_at <= NOW()", slug)

		return self._get_post("SELECT * FROM blog \
			WHERE slug = %s", slug)

	def get_newest(self, limit=None):
		return self._get_posts("SELECT * FROM blog \
			WHERE published_at IS NOT NULL \
				AND published_at <= NOW() \
			ORDER BY published_at DESC LIMIT %s", limit)

	def get_by_tag(self, tag, limit=None):
		return self._get_posts("SELECT * FROM blog \
			WHERE published_at IS NOT NULL \
				AND published_at <= NOW() \
				AND %s = ANY(tags) \
			ORDER BY published_at DESC LIMIT %s", tag, limit)

	def get_by_author(self, author, limit=None):
		return self._get_posts("SELECT * FROM blog \
			WHERE (author = %s OR author_uid = %s) \
				AND published_at IS NOT NULL \
				AND published_at <= NOW() \
			ORDER BY published_at DESC LIMIT %s",
			author.name, author.uid, limit)

	def get_by_year(self, year):
		return self._get_posts("SELECT * FROM blog \
			WHERE EXTRACT(year FROM published_at) = %s \
				AND published_at IS NOT NULL \
				AND published_at <= NOW() \
			ORDER BY published_at DESC", year)

	def get_drafts(self, author=None, limit=None):
		if author:
			return self._get_posts("SELECT * FROM blog \
				WHERE author_uid = %s \
					AND (published_at IS NULL OR published_at > NOW()) \
				ORDER BY COALESCE(updated_at, created_at) DESC LIMIT %s",
				author.uid, limit)

		return self._get_posts("SELECT * FROM blog \
			WHERE (published_at IS NULL OR published_at > NOW()) \
			ORDER BY COALESCE(updated_at, created_at) DESC LIMIT %s", limit)

	def search(self, query, limit=None):
		query = util.parse_search_query(query)

		return self._get_posts("SELECT blog.* FROM blog \
			LEFT JOIN blog_search_index search_index ON blog.id = search_index.post_id \
			WHERE search_index.document @@ to_tsquery('english', %s) \
				ORDER BY ts_rank(search_index.document, to_tsquery('english', %s)) DESC \
			LIMIT %s", query, query, limit)

	def has_had_recent_activity(self, t=None):
		if t is None:
			t = datetime.timedelta(hours=24)

		res = self.db.get("SELECT COUNT(*) AS count FROM blog \
			WHERE published_at IS NOT NULL AND published_at >= NOW() - %s", t)

		if res and res.count > 0:
			return True

		return False

	def create_post(self, title, text, author, tags=[], lang="markdown"):
		"""
			Creates a new post and returns the resulting Post object
		"""
		# Pre-render HTML
		html = self._render_text(text, lang=lang)

		return self._get_post("INSERT INTO blog(title, slug, text, html, lang, author_uid, tags) \
			VALUES(%s, %s, %s, %s, %s, %s, %s) RETURNING *", title, self._make_slug(title), text,
			html, lang, author.uid, list(tags))

	def _make_slug(self, s):
		# Remove any non-ASCII characters
		try:
			s = unicodedata.normalize("NFKD", s)
		except TypeError:
			pass

		# Remove excessive whitespace
		s = re.sub(r"[^\w]+", " ", s)

		slug = "-".join(s.split()).lower()

		while True:
			e = self.db.get("SELECT 1 FROM blog WHERE slug = %s", slug)
			if not e:
				break

			slug += "-"

		return slug

	def _render_text(self, text, lang="markdown"):
		if lang == "markdown":
			return markdown2.markdown(text, link_patterns=link_patterns,
				extras=[
					"code-friendly",
					"cuddled-lists",
					"fenced-code-blocks",
					"footnotes",
					"header-ids",
					"link-patterns",
					"tables",
				])

		elif lang == "textile":
			return textile.textile(text)

		return text

	def refresh(self):
		"""
			Needs to be called after a post has been changed
			and updates the search index.
		"""
		self.db.execute("REFRESH MATERIALIZED VIEW blog_search_index")

	@property
	def years(self):
		res = self.db.query("SELECT DISTINCT EXTRACT(year FROM published_at)::integer AS year \
			FROM blog WHERE published_at IS NOT NULL AND published_at <= NOW() \
			ORDER BY year DESC")

		for row in res:
			yield row.year

	@tornado.gen.coroutine
	def update_feeds(self):
		"""
			Updates all enabled feeds
		"""
		for feed in self.db.query("SELECT * FROM blog_feeds WHERE enabled IS TRUE"):
			try:
				f = feedparser.parse(feed.url)
			except Exception as e:
				raise e

			with self.db.transaction():
				# Update name
				self.db.execute("UPDATE blog_feeds SET name = %s \
					WHERE id = %s", f.feed.title, feed.id)

				# Walk through all entries
				for entry in f.entries:
					# Skip everything without the "blog.ipfire.org" tag
					try:
						tags = list((t.term for t in entry.tags))

						if not "blog.ipfire.org" in tags:
							continue
					except AttributeError:
						continue

					# Get link to the posting site
					link = entry.links[0].href

					# Check if the entry has already been imported
					res = self.db.get("SELECT id, (updated_at < %s) AS needs_update \
							FROM blog WHERE feed_id = %s AND foreign_id = %s",
							entry.updated, feed.id, entry.id)
					if res:
						# If the post needs to be updated, we do so
						if res.needs_update:
							self.db.execute("UPDATE blog SET title = %s, author = %s, \
								published_at = %s, updated_at = %s, html = %s, link = %s, \
								tags = %s WHERE id = %s", entry.title, entry.author,
								entry.published, entry.updated, entry.summary, link,
								feed.tags + tags, res.id)

						# Done here
						continue

					# Insert the new post
					self.db.execute("INSERT INTO blog(title, slug, author, \
						published_at, html, link, tags, updated_at, feed_id, foreign_id) \
						VALUES(%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)",
						entry.title, self._make_slug(entry.title), entry.author,
						entry.published, entry.summary, link, feed.tags + tags,
						entry.updated, feed.id, entry.id)

				# Mark feed as updated
				self.db.execute("UPDATE blog_feeds SET last_updated_at = CURRENT_TIMESTAMP \
					WHERE id = %s" % feed.id)

		# Refresh the search index
		with self.db.transaction():
			self.refresh()


class Post(misc.Object):
	def init(self, id, data=None):
		self.id   = id
		self.data = data

	# Title

	@property
	def title(self):
		return self.data.title

	@property
	def slug(self):
		return self.data.slug

	@lazy_property
	def author(self):
		if self.data.author_uid:
			return self.backend.accounts.get_by_uid(self.data.author_uid)

		return self.data.author

	@property
	def created_at(self):
		return self.data.created_at

	@property
	def lang(self):
		return self.data.lang

	# Published?

	@property
	def published_at(self):
		return self.data.published_at

	def is_published(self):
		"""
			Returns True if the post is already published
		"""
		return self.published_at and self.published_at <= datetime.datetime.now()

	def publish(self, when=None):
		if self.is_published():
			return

		self.db.execute("UPDATE blog SET published_at = COALESCE(%s, CURRENT_TIMESTAMP) \
			WHERE id = %s", when, self.id)

		# Update search indices
		self.backend.blog.refresh()

	# Updated?

	@property
	def updated_at(self):
		return self.data.updated_at

	# Text

	@property
	def text(self):
		return self.data.text

	# HTML

	@lazy_property
	def html(self):
		"""
			Returns this post as rendered HTML
		"""
		return self.data.html or self.backend.blog._render_text(self.text, lang=self.lang)

	# Tags

	@property
	def tags(self):
		return self.data.tags

	# Link

	@property
	def link(self):
		return self.data.link

	@lazy_property
	def release(self):
		return self.backend.releases._get_release("SELECT * FROM releases \
			WHERE published IS NOT NULL AND published <= NOW() AND blog_id = %s", self.id)

	def is_editable(self, editor):
		# Authors can edit their own posts
		return self.author == editor

	def update(self, title, text, tags=[]):
		"""
			Called to update the content of this post
		"""
		# Update slug when post isn't published yet
		slug = self.backend.blog._make_slug(title) \
			if not self.is_published() and not self.title == title else self.slug

		# Render and cache HTML
		html = self.backend.blog._render_text(text, lang=self.lang)

		self.db.execute("UPDATE blog SET title = %s, slug = %s, text = %s, html = %s, \
			tags = %s, updated_at = CURRENT_TIMESTAMP WHERE id = %s",
			title, slug, text, html, list(tags), self.id)

		# Update cache
		self.data.update({
			"title" : title,
			"slug"  : slug,
			"text"  : text,
			"html"  : html,
			"tags"  : tags,
		})

		# Update search index if post is published
		if self.is_published():
			self.backend.blog.refresh()

	def delete(self):
		self.db.execute("DELETE FROM blog WHERE id = %s", self.id)

		# Update search indices
		self.backend.blog.refresh()
Commit	Line	Data
0a6875dc MT	1	#!/usr/bin/python
0a6875dc MT	2
541c952b	3	import datetime
c70a7c29	4	import feedparser
2de5ad8a	5	import markdown2
c70a7c29	6	import re
7e64f6a3	7	import textile
d1a77356	8	import tornado.gen
c70a7c29	9	import unicodedata
7e64f6a3	10
0a6875dc	11	from . import misc
9523790a	12	from . import util
a3a850a4	13	from .decorators import *
0a6875dc	14
2de5ad8a MT	15	# Used to automatically link some things
	16	link_patterns = (
	17	# Find bug reports
	18	(re.compile(r"(?:#(\d+))", re.I), r"https://bugzilla.ipfire.org/show_bug.cgi?id=\1"),
	19
	20	# Email Addresses
	21	(re.compile(r"([a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+)"), r"mailto:\1"),
	22
	23	# CVE Numbers
5c6aa47c	24	(re.compile(r"(?:CVE)[\s\-](\d{4}\-\d+)"), r"https://cve.mitre.org/cgi-bin/cvename.cgi?name=\1"),
2de5ad8a MT	25	)
2de5ad8a MT	26
0a6875dc MT	27	class Blog(misc.Object):
	28	def _get_post(self, query, *args):
	29	res = self.db.get(query, *args)
	30
	31	if res:
	32	return Post(self.backend, res.id, data=res)
	33
	34	def _get_posts(self, query, *args):
	35	res = self.db.query(query, *args)
	36
	37	for row in res:
	38	yield Post(self.backend, row.id, data=row)
	39
487417ad MT	40	def get_by_id(self, id):
	41	return self._get_post("SELECT * FROM blog \
	42	WHERE id = %s", id)
	43
df157ede MT	44	def get_by_slug(self, slug, published=True):
	45	if published:
	46	return self._get_post("SELECT * FROM blog \
	47	WHERE slug = %s AND published_at <= NOW()", slug)
	48
0a6875dc	49	return self._get_post("SELECT * FROM blog \
df157ede MT	50	WHERE slug = %s", slug)
df157ede MT	51
0a6875dc MT	52	def get_newest(self, limit=None):
	53	return self._get_posts("SELECT * FROM blog \
	54	WHERE published_at IS NOT NULL \
	55	AND published_at <= NOW() \
	56	ORDER BY published_at DESC LIMIT %s", limit)
	57
	58	def get_by_tag(self, tag, limit=None):
	59	return self._get_posts("SELECT * FROM blog \
	60	WHERE published_at IS NOT NULL \
	61	AND published_at <= NOW() \
	62	AND %s = ANY(tags) \
4bde7f18	63	ORDER BY published_at DESC LIMIT %s", tag, limit)
0a6875dc	64
cdf85ee7	65	def get_by_author(self, author, limit=None):
0a6875dc	66	return self._get_posts("SELECT * FROM blog \
cdf85ee7	67	WHERE (author = %s OR author_uid = %s) \
0a6875dc MT	68	AND published_at IS NOT NULL \
0a6875dc MT	69	AND published_at <= NOW() \
cdf85ee7 MT	70	ORDER BY published_at DESC LIMIT %s",
cdf85ee7 MT	71	author.name, author.uid, limit)
0a6875dc	72
7e64f6a3 MT	73	def get_by_year(self, year):
	74	return self._get_posts("SELECT * FROM blog \
	75	WHERE EXTRACT(year FROM published_at) = %s \
	76	AND published_at IS NOT NULL \
	77	AND published_at <= NOW() \
	78	ORDER BY published_at DESC", year)
	79
0b342a05 MT	80	def get_drafts(self, author=None, limit=None):
	81	if author:
	82	return self._get_posts("SELECT * FROM blog \
	83	WHERE author_uid = %s \
	84	AND (published_at IS NULL OR published_at > NOW()) \
	85	ORDER BY COALESCE(updated_at, created_at) DESC LIMIT %s",
	86	author.uid, limit)
	87
	88	return self._get_posts("SELECT * FROM blog \
	89	WHERE (published_at IS NULL OR published_at > NOW()) \
	90	ORDER BY COALESCE(updated_at, created_at) DESC LIMIT %s", limit)
	91
0a6875dc	92	def search(self, query, limit=None):
9523790a	93	query = util.parse_search_query(query)
a47ddc3d	94
0a6875dc MT	95	return self._get_posts("SELECT blog.* FROM blog \
	96	LEFT JOIN blog_search_index search_index ON blog.id = search_index.post_id \
	97	WHERE search_index.document @@ to_tsquery('english', %s) \
	98	ORDER BY ts_rank(search_index.document, to_tsquery('english', %s)) DESC \
	99	LIMIT %s", query, query, limit)
	100
611adbfb MT	101	def has_had_recent_activity(self, t=None):
	102	if t is None:
	103	t = datetime.timedelta(hours=24)
	104
	105	res = self.db.get("SELECT COUNT(*) AS count FROM blog \
	106	WHERE published_at IS NOT NULL AND published_at >= NOW() - %s", t)
	107
	108	if res and res.count > 0:
	109	return True
	110
	111	return False
	112
694c4f08	113	def create_post(self, title, text, author, tags=[], lang="markdown"):
541c952b MT	114	"""
	115	Creates a new post and returns the resulting Post object
	116	"""
694c4f08 MT	117	# Pre-render HTML
	118	html = self._render_text(text, lang=lang)
	119
	120	return self._get_post("INSERT INTO blog(title, slug, text, html, lang, author_uid, tags) \
	121	VALUES(%s, %s, %s, %s, %s, %s, %s) RETURNING *", title, self._make_slug(title), text,
	122	html, lang, author.uid, list(tags))
541c952b	123
c70a7c29 MT	124	def _make_slug(self, s):
	125	# Remove any non-ASCII characters
	126	try:
	127	s = unicodedata.normalize("NFKD", s)
	128	except TypeError:
	129	pass
	130
	131	# Remove excessive whitespace
	132	s = re.sub(r"[^\w]+", " ", s)
	133
	134	slug = "-".join(s.split()).lower()
	135
	136	while True:
	137	e = self.db.get("SELECT 1 FROM blog WHERE slug = %s", slug)
	138	if not e:
	139	break
	140
	141	slug += "-"
	142
	143	return slug
	144
2de5ad8a MT	145	def _render_text(self, text, lang="markdown"):
	146	if lang == "markdown":
	147	return markdown2.markdown(text, link_patterns=link_patterns,
e0ef6d39 MT	148	extras=[
	149	"code-friendly",
	150	"cuddled-lists",
	151	"fenced-code-blocks",
	152	"footnotes",
	153	"header-ids",
	154	"link-patterns",
	155	"tables",
	156	])
2de5ad8a MT	157
	158	elif lang == "textile":
	159	return textile.textile(text)
	160
	161	return text
	162
0a6875dc MT	163	def refresh(self):
	164	"""
	165	Needs to be called after a post has been changed
	166	and updates the search index.
	167	"""
	168	self.db.execute("REFRESH MATERIALIZED VIEW blog_search_index")
	169
7e64f6a3 MT	170	@property
	171	def years(self):
	172	res = self.db.query("SELECT DISTINCT EXTRACT(year FROM published_at)::integer AS year \
	173	FROM blog WHERE published_at IS NOT NULL AND published_at <= NOW() \
	174	ORDER BY year DESC")
	175
	176	for row in res:
	177	yield row.year
	178
d1a77356	179	@tornado.gen.coroutine
c70a7c29 MT	180	def update_feeds(self):
	181	"""
	182	Updates all enabled feeds
	183	"""
	184	for feed in self.db.query("SELECT * FROM blog_feeds WHERE enabled IS TRUE"):
	185	try:
	186	f = feedparser.parse(feed.url)
	187	except Exception as e:
	188	raise e
	189
	190	with self.db.transaction():
	191	# Update name
	192	self.db.execute("UPDATE blog_feeds SET name = %s \
	193	WHERE id = %s", f.feed.title, feed.id)
	194
	195	# Walk through all entries
	196	for entry in f.entries:
	197	# Skip everything without the "blog.ipfire.org" tag
	198	try:
	199	tags = list((t.term for t in entry.tags))
	200
	201	if not "blog.ipfire.org" in tags:
	202	continue
	203	except AttributeError:
	204	continue
	205
	206	# Get link to the posting site
	207	link = entry.links[0].href
	208
	209	# Check if the entry has already been imported
	210	res = self.db.get("SELECT id, (updated_at < %s) AS needs_update \
	211	FROM blog WHERE feed_id = %s AND foreign_id = %s",
	212	entry.updated, feed.id, entry.id)
	213	if res:
	214	# If the post needs to be updated, we do so
	215	if res.needs_update:
	216	self.db.execute("UPDATE blog SET title = %s, author = %s, \
	217	published_at = %s, updated_at = %s, html = %s, link = %s, \
	218	tags = %s WHERE id = %s", entry.title, entry.author,
	219	entry.published, entry.updated, entry.summary, link,
	220	feed.tags + tags, res.id)
	221
	222	# Done here
	223	continue
	224
	225	# Insert the new post
	226	self.db.execute("INSERT INTO blog(title, slug, author, \
	227	published_at, html, link, tags, updated_at, feed_id, foreign_id) \
	228	VALUES(%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)",
	229	entry.title, self._make_slug(entry.title), entry.author,
	230	entry.published, entry.summary, link, feed.tags + tags,
	231	entry.updated, feed.id, entry.id)
	232
20277bf5 MT	233	# Mark feed as updated
	234	self.db.execute("UPDATE blog_feeds SET last_updated_at = CURRENT_TIMESTAMP \
	235	WHERE id = %s" % feed.id)
	236
c70a7c29 MT	237	# Refresh the search index
	238	with self.db.transaction():
	239	self.refresh()
	240
0a6875dc MT	241
	242	class Post(misc.Object):
	243	def init(self, id, data=None):
	244	self.id = id
	245	self.data = data
	246
541c952b MT	247	# Title
541c952b MT	248
93725180 MT	249	@property
93725180 MT	250	def title(self):
0a6875dc MT	251	return self.data.title
	252
	253	@property
	254	def slug(self):
	255	return self.data.slug
	256
a3a850a4	257	@lazy_property
0a6875dc MT	258	def author(self):
	259	if self.data.author_uid:
	260	return self.backend.accounts.get_by_uid(self.data.author_uid)
	261
cdf85ee7 MT	262	return self.data.author
cdf85ee7 MT	263
0a6875dc MT	264	@property
	265	def created_at(self):
	266	return self.data.created_at
	267
2de5ad8a MT	268	@property
	269	def lang(self):
	270	return self.data.lang
	271
541c952b MT	272	# Published?
541c952b MT	273
0a6875dc MT	274	@property
	275	def published_at(self):
	276	return self.data.published_at
	277
541c952b MT	278	def is_published(self):
	279	"""
	280	Returns True if the post is already published
	281	"""
	282	return self.published_at and self.published_at <= datetime.datetime.now()
	283
9ea64cef	284	def publish(self, when=None):
541c952b MT	285	if self.is_published():
	286	return
	287
9ea64cef MT	288	self.db.execute("UPDATE blog SET published_at = COALESCE(%s, CURRENT_TIMESTAMP) \
9ea64cef MT	289	WHERE id = %s", when, self.id)
541c952b MT	290
	291	# Update search indices
	292	self.backend.blog.refresh()
	293
	294	# Updated?
	295
7e64f6a3	296	@property
541c952b MT	297	def updated_at(self):
	298	return self.data.updated_at
	299
541c952b MT	300	# Text
541c952b MT	301
93725180 MT	302	@property
93725180 MT	303	def text(self):
541c952b MT	304	return self.data.text
541c952b MT	305
541c952b	306	# HTML
7e64f6a3	307
a3a850a4	308	@lazy_property
0a6875dc MT	309	def html(self):
	310	"""
	311	Returns this post as rendered HTML
	312	"""
2de5ad8a	313	return self.data.html or self.backend.blog._render_text(self.text, lang=self.lang)
8ebc98d4	314
541c952b MT	315	# Tags
541c952b MT	316
93725180 MT	317	@property
93725180 MT	318	def tags(self):
8ebc98d4	319	return self.data.tags
1e76fec4	320
93725180	321	# Link
541c952b	322
1e76fec4 MT	323	@property
	324	def link(self):
	325	return self.data.link
984e4e7b	326
a3a850a4	327	@lazy_property
984e4e7b MT	328	def release(self):
	329	return self.backend.releases._get_release("SELECT * FROM releases \
	330	WHERE published IS NOT NULL AND published <= NOW() AND blog_id = %s", self.id)
e8a81a70 MT	331
	332	def is_editable(self, editor):
	333	# Authors can edit their own posts
	334	return self.author == editor
93725180 MT	335
	336	def update(self, title, text, tags=[]):
	337	"""
	338	Called to update the content of this post
	339	"""
	340	# Update slug when post isn't published yet
baa294fb MT	341	slug = self.backend.blog._make_slug(title) \
baa294fb MT	342	if not self.is_published() and not self.title == title else self.slug
93725180	343
694c4f08 MT	344	# Render and cache HTML
694c4f08 MT	345	html = self.backend.blog._render_text(text, lang=self.lang)
93725180	346
694c4f08	347	self.db.execute("UPDATE blog SET title = %s, slug = %s, text = %s, html = %s, \
93725180	348	tags = %s, updated_at = CURRENT_TIMESTAMP WHERE id = %s",
694c4f08	349	title, slug, text, html, list(tags), self.id)
93725180 MT	350
	351	# Update cache
	352	self.data.update({
	353	"title" : title,
	354	"slug" : slug,
	355	"text" : text,
d73edae7	356	"html" : html,
93725180 MT	357	"tags" : tags,
	358	})
	359
	360	# Update search index if post is published
	361	if self.is_published():
	362	self.backend.blog.refresh()
914238a5 MT	363
	364	def delete(self):
	365	self.db.execute("DELETE FROM blog WHERE id = %s", self.id)
	366
	367	# Update search indices
	368	self.backend.blog.refresh()