search: Use PostgreSQL's websearch_to_tsquery()

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

#!/usr/bin/python

import datetime
import feedparser
import html2text
import markdown
import markdown.extensions
import markdown.preprocessors
import re
import textile
import unicodedata

from . import misc
from .decorators import *

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 @@ websearch_to_tsquery('english', %s) \
				ORDER BY ts_rank(search_index.document, websearch_to_tsquery('english', %s)) DESC \
			LIMIT %s", query, query, limit)

	def has_had_recent_activity(self, **kwargs):
		t = datetime.timedelta(**kwargs)

		res = self.db.get("SELECT COUNT(*) AS count FROM blog \
			WHERE published_at IS NOT NULL AND published_at BETWEEN NOW() - %s AND NOW()", 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 markdown.markdown(text,
				extensions=[
					PrettyLinksExtension(),
					"codehilite",
					"fenced_code",
					"footnotes",
					"nl2br",
					"sane_lists",
					"tables",
					"toc",
				])

		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

	async def announce(self):
		posts = self._get_posts("SELECT * FROM blog \
			WHERE (published_at IS NOT NULL AND published_at <= NOW()) \
				AND announced_at IS NULL")

		for post in posts:
			await post.announce()

	async 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)

	@lazy_property
	def plaintext(self):
		h = html2text.HTML2Text()
		h.ignore_links = True

		return h.handle(self.html)

	# Excerpt

	@property
	def excerpt(self):
		paragraphs = self.plaintext.split("\n\n")

		excerpt = []

		for paragraph in paragraphs:
			excerpt.append(paragraph)

			# Add another paragraph if we encountered a headline
			if paragraph.startswith("#"):
				continue

			# End if this paragraph was long enough
			if len(paragraph) >= 40:
				break

		return "\n\n".join(excerpt)

	# 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()

	async def announce(self):
		# Get people who should receive this message
		group = self.backend.groups.get_by_gid("promotional-consent")
		if not group:
			return

		with self.db.transaction():
			# Generate an email for everybody in this group
			for account in group:
				self.backend.messages.send_template("blog/messages/announcement",
					account=account, post=self)

			# Mark this post as announced
			self.db.execute("UPDATE blog SET announced_at = CURRENT_TIMESTAMP \
				WHERE id = %s", self.id)


class PrettyLinksExtension(markdown.extensions.Extension):
	def extendMarkdown(self, md):
		md.preprocessors.register(BugzillaLinksPreprocessor(md), "bugzilla", 10)
		md.preprocessors.register(CVELinksPreprocessor(md), "cve", 10)


class BugzillaLinksPreprocessor(markdown.preprocessors.Preprocessor):
	regex = re.compile(r"(?:#(\d{5,}))", re.I)

	def run(self, lines):
		for line in lines:
			yield self.regex.sub(r"[#\1](https://bugzilla.ipfire.org/show_bug.cgi?id=\1)", line)


class CVELinksPreprocessor(markdown.preprocessors.Preprocessor):
	regex = re.compile(r"(?:CVE)[\s\-](\d{4}\-\d+)")

	def run(self, lines):
		for line in lines:
			yield self.regex.sub(r"[CVE-\1](https://cve.mitre.org/cgi-bin/cvename.cgi?name=\1)", line)