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

#!/usr/bin/python

import datetime
import feedparser
import markdown2
import re
import textile
import unicodedata

from . import misc

# 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"http://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):
		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 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=["footnotes", "link-patterns", "wiki-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

	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

	# XXX needs caching
	@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

	@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

	# XXX needs caching
	@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,
			"tags"  : tags,
		})

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