src/backend/wiki.py

   1 #!/usr/bin/python3
   2
   3 import PIL
   4 import difflib
   5 import io
   6 import logging
   7 import os.path
   8 import re
   9 import tornado.gen
  10 import urllib.parse
  11
  12 from . import misc
  13 from . import util
  14 from .decorators import *
  15
  16 class Wiki(misc.Object):
  17         def _get_pages(self, query, *args):
  18                 res = self.db.query(query, *args)
  19
  20                 for row in res:
  21                         yield Page(self.backend, row.id, data=row)
  22
  23         def _get_page(self, query, *args):
  24                 res = self.db.get(query, *args)
  25
  26                 if res:
  27                         return Page(self.backend, res.id, data=res)
  28
  29         def get_page_title(self, page, default=None):
  30                 doc = self.get_page(page)
  31                 if doc:
  32                         return doc.title
  33
  34                 return default or os.path.basename(page)
  35
  36         def get_page(self, page, revision=None):
  37                 page = Page.sanitise_page_name(page)
  38                 assert page
  39
  40                 if revision:
  41                         return self._get_page("SELECT * FROM wiki WHERE page = %s \
  42                                 AND timestamp = %s", page, revision)
  43                 else:
  44                         return self._get_page("SELECT * FROM wiki WHERE page = %s \
  45                                 ORDER BY timestamp DESC LIMIT 1", page)
  46
  47         def get_recent_changes(self, account, limit=None):
  48                 pages = self._get_pages("SELECT * FROM wiki \
  49                         WHERE timestamp >= NOW() - INTERVAL '4 weeks' \
  50                         ORDER BY timestamp DESC")
  51
  52                 for page in pages:
  53                         if not page.check_acl(account):
  54                                 continue
  55
  56                         yield page
  57
  58                         limit -= 1
  59                         if not limit:
  60                                 break
  61
  62         def create_page(self, page, author, content, changes=None, address=None):
  63                 page = Page.sanitise_page_name(page)
  64
  65                 # Write page to the database
  66                 page = self._get_page("INSERT INTO wiki(page, author_uid, markdown, changes, address) \
  67                         VALUES(%s, %s, %s, %s, %s) RETURNING *", page, author.uid, content or None, changes, address)
  68
  69                 # Render markdown
  70                 page.bake()
  71
  72                 # Send email to all watchers
  73                 page._send_watcher_emails(excludes=[author])
  74
  75                 return page
  76
  77         def delete_page(self, page, author, **kwargs):
  78                 # Do nothing if the page does not exist
  79                 if not self.get_page(page):
  80                         return
  81
  82                 # Just creates a blank last version of the page
  83                 self.create_page(page, author=author, content=None, **kwargs)
  84
  85         @tornado.gen.coroutine
  86         def rebake(self):
  87                 """
  88                         Re-renders all wiki pages
  89                 """
  90                 pages = self._get_pages("SELECT * FROM wiki ORDER BY timestamp DESC")
  91
  92                 for page in pages:
  93                         logging.info("Baking %s from %s" % (page.page, page.timestamp))
  94
  95                         with self.db.transaction():
  96                                 page.bake()
  97
  98         def make_breadcrumbs(self, url):
  99                 # Split and strip all empty elements (double slashes)
 100                 parts = list(e for e in url.split("/") if e)
 101
 102                 ret = []
 103                 for part in ("/".join(parts[:i]) for i in range(1, len(parts))):
 104                         ret.append(("/%s" % part, self.get_page_title(part, os.path.basename(part))))
 105
 106                 return ret
 107
 108         def search(self, query, account=None, limit=None):
 109                 query = util.parse_search_query(query)
 110
 111                 res = self._get_pages("SELECT wiki.* FROM wiki_search_index search_index \
 112                         LEFT JOIN wiki ON search_index.wiki_id = wiki.id \
 113                         WHERE search_index.document @@ to_tsquery('english', %s) \
 114                                 ORDER BY ts_rank(search_index.document, to_tsquery('english', %s)) DESC",
 115                         query, query)
 116
 117                 for page in res:
 118                         # Skip any pages the user doesn't have permission for
 119                         if not page.check_acl(account):
 120                                 continue
 121
 122                         # Return any other pages
 123                         yield page
 124
 125                         limit -= 1
 126                         if not limit:
 127                                 break
 128
 129         def refresh(self):
 130                 """
 131                         Needs to be called after a page has been changed
 132                 """
 133                 self.db.execute("REFRESH MATERIALIZED VIEW wiki_search_index")
 134
 135         # ACL
 136
 137         def check_acl(self, page, account):
 138                 res = self.db.query("SELECT * FROM wiki_acls \
 139                         WHERE %s ILIKE (path || '%%') ORDER BY LENGTH(path) DESC LIMIT 1", page)
 140
 141                 for row in res:
 142                         # Access not permitted when user is not logged in
 143                         if not account:
 144                                 return False
 145
 146                         # If user is in a matching group, we grant permission
 147                         for group in row.groups:
 148                                 if group in account.groups:
 149                                         return True
 150
 151                         # Otherwise access is not permitted
 152                         return False
 153
 154                 # If no ACLs are found, we permit access
 155                 return True
 156
 157         # Files
 158
 159         def _get_files(self, query, *args):
 160                 res = self.db.query(query, *args)
 161
 162                 for row in res:
 163                         yield File(self.backend, row.id, data=row)
 164
 165         def _get_file(self, query, *args):
 166                 res = self.db.get(query, *args)
 167
 168                 if res:
 169                         return File(self.backend, res.id, data=res)
 170
 171         def get_files(self, path):
 172                 files = self._get_files("SELECT * FROM wiki_files \
 173                         WHERE path = %s AND deleted_at IS NULL ORDER BY filename", path)
 174
 175                 return list(files)
 176
 177         def get_file_by_path(self, path):
 178                 path, filename = os.path.dirname(path), os.path.basename(path)
 179
 180                 return self._get_file("SELECT * FROM wiki_files \
 181                         WHERE path = %s AND filename = %s AND deleted_at IS NULL", path, filename)
 182
 183         def upload(self, path, filename, data, mimetype, author, address):
 184                 # Upload the blob first
 185                 blob = self.db.get("INSERT INTO wiki_blobs(data) VALUES(%s) RETURNING id", data)
 186
 187                 # Create entry for file
 188                 return self._get_file("INSERT INTO wiki_files(path, filename, author_uid, address, \
 189                         mimetype, blob_id, size) VALUES(%s,  %s, %s, %s, %s, %s, %s) RETURNING *", path,
 190                         filename, author.uid, address, mimetype, blob.id, len(data))
 191
 192         def find_image(self, path, filename):
 193                 for p in (path, os.path.dirname(path)):
 194                         file = self.get_file_by_path(os.path.join(p, filename))
 195
 196                         if file and file.is_image():
 197                                 return file
 198
 199
 200 class Page(misc.Object):
 201         def init(self, id, data=None):
 202                 self.id = id
 203                 self.data = data
 204
 205         def __repr__(self):
 206                 return "<%s %s %s>" % (self.__class__.__name__, self.page, self.timestamp)
 207
 208         def __eq__(self, other):
 209                 if isinstance(other, self.__class__):
 210                         return self.id == other.id
 211
 212         def __lt__(self, other):
 213                 if isinstance(other, self.__class__):
 214                         if self.page == other.page:
 215                                 return self.timestamp < other.timestamp
 216
 217                         return self.page < other.page
 218
 219         @staticmethod
 220         def sanitise_page_name(page):
 221                 if not page:
 222                         return "/"
 223
 224                 # Make sure that the page name does NOT end with a /
 225                 if page.endswith("/"):
 226                         page = page[:-1]
 227
 228                 # Make sure the page name starts with a /
 229                 if not page.startswith("/"):
 230                         page = "/%s" % page
 231
 232                 # Remove any double slashes
 233                 page = page.replace("//", "/")
 234
 235                 return page
 236
 237         @property
 238         def url(self):
 239                 return self.page
 240
 241         @property
 242         def full_url(self):
 243                 return "https://wiki.ipfire.org%s" % self.url
 244
 245         @property
 246         def page(self):
 247                 return self.data.page
 248
 249         @property
 250         def title(self):
 251                 return self._title or os.path.basename(self.page[1:])
 252
 253         @property
 254         def _title(self):
 255                 if not self.markdown:
 256                         return
 257
 258                 # Find first H1 headline in markdown
 259                 markdown = self.markdown.splitlines()
 260
 261                 m = re.match(r"^# (.*)( #)?$", markdown[0])
 262                 if m:
 263                         return m.group(1)
 264
 265         @lazy_property
 266         def author(self):
 267                 if self.data.author_uid:
 268                         return self.backend.accounts.get_by_uid(self.data.author_uid)
 269
 270         def _render(self, text):
 271                 logging.debug("Rendering %s" % self)
 272
 273                 # Link images
 274                 replacements = []
 275                 for match in re.finditer(r"!\[(.*)\]\((.*)\)", text):
 276                         alt_text, url = match.groups()
 277
 278                         # Skip any absolute and external URLs
 279                         if url.startswith("/") or url.startswith("https://") or url.startswith("http://"):
 280                                 continue
 281
 282                         # Try to split query string
 283                         url, delimiter, qs = url.partition("?")
 284
 285                         # Parse query arguments
 286                         args = urllib.parse.parse_qs(qs)
 287
 288                         # Find image
 289                         file = self.backend.wiki.find_image(self.page, url)
 290                         if not file:
 291                                 continue
 292
 293                         # Scale down the image if not already done
 294                         if not "s" in args:
 295                                 args["s"] = "768"
 296
 297                         # Format URL
 298                         url = "%s?%s" % (file.url, urllib.parse.urlencode(args))
 299
 300                         replacements.append((match.span(), file, alt_text, url))
 301
 302                 # Apply all replacements
 303                 for (start, end), file, alt_text, url in reversed(replacements):
 304                         text = text[:start] + "[![%s](%s)](%s?action=detail)" % (alt_text, url, file.url) + text[end:]
 305
 306                 # Add wiki links
 307                 patterns = (
 308                         (r"\[\[([\w\d\/]+)(?:\|([\w\d\s]+))\]\]", r"/\1", r"\2", None),
 309                         (r"\[\[([\w\d\/\-\.]+)\]\]", r"\1", r"\1", self.backend.wiki.get_page_title),
 310                 )
 311
 312                 for pattern, link, title, repl in patterns:
 313                         replacements = []
 314
 315                         for match in re.finditer(pattern, text):
 316                                 l = match.expand(link)
 317                                 t = match.expand(title)
 318
 319                                 # Allow relative links
 320                                 if not l.startswith("/"):
 321                                         l = os.path.join(self.page, l)
 322
 323                                 # Normalise links
 324                                 l = os.path.normpath(l)
 325
 326                                 if callable(repl):
 327                                         t = repl(l) or t
 328
 329                                 replacements.append((match.span(), t or l, l))
 330
 331                         # Apply all replacements
 332                         for (start, end), t, l in reversed(replacements):
 333                                 text = text[:start] + "[%s](%s)" % (t, l) + text[end:]
 334
 335                 # Borrow this from the blog
 336                 return self.backend.blog._render_text(text, lang="markdown")
 337
 338         @property
 339         def markdown(self):
 340                 return self.data.markdown or ""
 341
 342         @property
 343         def html(self):
 344                 return self.data.html or self._render(self.markdown)
 345
 346         def bake(self):
 347                 """
 348                         Renders the page's markdown to HTML and stores
 349                         it in the database.
 350                 """
 351                 self.db.execute("UPDATE wiki SET html = %s WHERE id = %s",
 352                         self._render(self.markdown), self.id)
 353
 354         @property
 355         def timestamp(self):
 356                 return self.data.timestamp
 357
 358         def was_deleted(self):
 359                 return self.markdown is None
 360
 361         @lazy_property
 362         def breadcrumbs(self):
 363                 return self.backend.wiki.make_breadcrumbs(self.page)
 364
 365         def get_latest_revision(self):
 366                 revisions = self.get_revisions()
 367
 368                 # Return first object
 369                 for rev in revisions:
 370                         return rev
 371
 372         def get_revisions(self):
 373                 return self.backend.wiki._get_pages("SELECT * FROM wiki \
 374                         WHERE page = %s ORDER BY timestamp DESC", self.page)
 375
 376         @lazy_property
 377         def previous_revision(self):
 378                 return self.backend.wiki._get_page("SELECT * FROM wiki \
 379                         WHERE page = %s AND timestamp < %s ORDER BY timestamp DESC \
 380                         LIMIT 1", self.page, self.timestamp)
 381
 382         @property
 383         def changes(self):
 384                 return self.data.changes
 385
 386         # ACL
 387
 388         def check_acl(self, account):
 389                 return self.backend.wiki.check_acl(self.page, account)
 390
 391         # Sidebar
 392
 393         @lazy_property
 394         def sidebar(self):
 395                 parts = self.page.split("/")
 396
 397                 while parts:
 398                         sidebar = self.backend.wiki.get_page("%s/sidebar" % os.path.join(*parts))
 399                         if sidebar:
 400                                 return sidebar
 401
 402                         parts.pop()
 403
 404         # Watchers
 405
 406         @lazy_property
 407         def diff(self):
 408                 if self.previous_revision:
 409                         diff = difflib.unified_diff(
 410                                 self.previous_revision.markdown.splitlines(),
 411                                 self.markdown.splitlines(),
 412                         )
 413
 414                         return "\n".join(diff)
 415
 416         @property
 417         def watchers(self):
 418                 res = self.db.query("SELECT uid FROM wiki_watchlist \
 419                         WHERE page = %s", self.page)
 420
 421                 for row in res:
 422                         # Search for account by UID and skip if none was found
 423                         account = self.backend.accounts.get_by_uid(row.uid)
 424                         if not account:
 425                                 continue
 426
 427                         # Return the account
 428                         yield account
 429
 430         def is_watched_by(self, account):
 431                 res = self.db.get("SELECT 1 FROM wiki_watchlist \
 432                         WHERE page = %s AND uid = %s", self.page, account.uid)
 433
 434                 if res:
 435                         return True
 436
 437                 return False
 438
 439         def add_watcher(self, account):
 440                 if self.is_watched_by(account):
 441                         return
 442
 443                 self.db.execute("INSERT INTO wiki_watchlist(page, uid) \
 444                         VALUES(%s, %s)", self.page, account.uid)
 445
 446         def remove_watcher(self, account):
 447                 self.db.execute("DELETE FROM wiki_watchlist \
 448                         WHERE page = %s AND uid = %s", self.page, account.uid)
 449
 450         def _send_watcher_emails(self, excludes=[]):
 451                 # Nothing to do if there was no previous revision
 452                 if not self.previous_revision:
 453                         return
 454
 455                 for watcher in self.watchers:
 456                         # Skip everyone who is excluded
 457                         if watcher in excludes:
 458                                 logging.debug("Excluding %s" % watcher)
 459                                 continue
 460
 461                         logging.debug("Sending watcher email to %s" % watcher)
 462
 463                         # Compose message
 464                         self.backend.messages.send_template("wiki/messages/page-changed",
 465                                 recipients=[watcher], page=self, priority=-10)
 466
 467
 468 class File(misc.Object):
 469         def init(self, id, data):
 470                 self.id   = id
 471                 self.data = data
 472
 473         @property
 474         def url(self):
 475                 return os.path.join(self.path, self.filename)
 476
 477         @property
 478         def path(self):
 479                 return self.data.path
 480
 481         @property
 482         def filename(self):
 483                 return self.data.filename
 484
 485         @property
 486         def mimetype(self):
 487                 return self.data.mimetype
 488
 489         @property
 490         def size(self):
 491                 return self.data.size
 492
 493         @lazy_property
 494         def author(self):
 495                 if self.data.author_uid:
 496                         return self.backend.accounts.get_by_uid(self.data.author_uid)
 497
 498         @property
 499         def created_at(self):
 500                 return self.data.created_at
 501
 502         def is_pdf(self):
 503                 return self.mimetype in ("application/pdf", "application/x-pdf")
 504
 505         def is_image(self):
 506                 return self.mimetype.startswith("image/")
 507
 508         @lazy_property
 509         def blob(self):
 510                 res = self.db.get("SELECT data FROM wiki_blobs \
 511                         WHERE id = %s", self.data.blob_id)
 512
 513                 if res:
 514                         return bytes(res.data)
 515
 516         def get_thumbnail(self, size):
 517                 cache_key = "-".join((self.path, util.normalize(self.filename), self.created_at.isoformat(), "%spx" % size))
 518
 519                 # Try to fetch the data from the cache
 520                 thumbnail = self.memcache.get(cache_key)
 521                 if thumbnail:
 522                         return thumbnail
 523
 524                 # Generate the thumbnail
 525                 thumbnail = self._generate_thumbnail(size)
 526
 527                 # Put it into the cache for forever
 528                 self.memcache.set(cache_key, thumbnail)
 529
 530                 return thumbnail
 531
 532         def _generate_thumbnail(self, size):
 533                 image = PIL.Image.open(io.BytesIO(self.blob))
 534
 535                 # Resize the image to the desired resolution
 536                 image.thumbnail((size, size), PIL.Image.ANTIALIAS)
 537
 538                 with io.BytesIO() as f:
 539                         # If writing out the image does not work with optimization,
 540                         # we try to write it out without any optimization.
 541                         try:
 542                                 image.save(f, image.format, optimize=True, quality=98)
 543                         except:
 544                                 image.save(f, image.format, quality=98)
 545
 546                         return f.getvalue()