src/backend/wiki.py

   1 #!/usr/bin/python3
   2
   3 import PIL
   4 import PIL.ImageFilter
   5 import difflib
   6 import io
   7 import logging
   8 import os.path
   9 import re
  10 import tornado.gen
  11 import urllib.parse
  12
  13 from . import misc
  14 from . import util
  15 from .decorators import *
  16
  17 class Wiki(misc.Object):
  18         def _get_pages(self, query, *args):
  19                 res = self.db.query(query, *args)
  20
  21                 for row in res:
  22                         yield Page(self.backend, row.id, data=row)
  23
  24         def _get_page(self, query, *args):
  25                 res = self.db.get(query, *args)
  26
  27                 if res:
  28                         return Page(self.backend, res.id, data=res)
  29
  30         def get_page_title(self, page, default=None):
  31                 doc = self.get_page(page)
  32                 if doc:
  33                         return doc.title
  34
  35                 return default or os.path.basename(page)
  36
  37         def get_page(self, page, revision=None):
  38                 page = Page.sanitise_page_name(page)
  39                 assert page
  40
  41                 if revision:
  42                         return self._get_page("SELECT * FROM wiki WHERE page = %s \
  43                                 AND timestamp = %s", page, revision)
  44                 else:
  45                         return self._get_page("SELECT * FROM wiki WHERE page = %s \
  46                                 ORDER BY timestamp DESC LIMIT 1", page)
  47
  48         def get_recent_changes(self, account, limit=None):
  49                 pages = self._get_pages("SELECT * FROM wiki \
  50                         WHERE timestamp >= NOW() - INTERVAL '4 weeks' \
  51                         ORDER BY timestamp DESC")
  52
  53                 for page in pages:
  54                         if not page.check_acl(account):
  55                                 continue
  56
  57                         yield page
  58
  59                         limit -= 1
  60                         if not limit:
  61                                 break
  62
  63         def create_page(self, page, author, content, changes=None, address=None):
  64                 page = Page.sanitise_page_name(page)
  65
  66                 # Write page to the database
  67                 page = self._get_page("INSERT INTO wiki(page, author_uid, markdown, changes, address) \
  68                         VALUES(%s, %s, %s, %s, %s) RETURNING *", page, author.uid, content or None, changes, address)
  69
  70                 # Send email to all watchers
  71                 page._send_watcher_emails(excludes=[author])
  72
  73                 return page
  74
  75         def delete_page(self, page, author, **kwargs):
  76                 # Do nothing if the page does not exist
  77                 if not self.get_page(page):
  78                         return
  79
  80                 # Just creates a blank last version of the page
  81                 self.create_page(page, author=author, content=None, **kwargs)
  82
  83         def make_breadcrumbs(self, url):
  84                 # Split and strip all empty elements (double slashes)
  85                 parts = list(e for e in url.split("/") if e)
  86
  87                 ret = []
  88                 for part in ("/".join(parts[:i]) for i in range(1, len(parts))):
  89                         ret.append(("/%s" % part, self.get_page_title(part, os.path.basename(part))))
  90
  91                 return ret
  92
  93         def search(self, query, account=None, limit=None):
  94                 query = util.parse_search_query(query)
  95
  96                 res = self._get_pages("SELECT wiki.* FROM wiki_search_index search_index \
  97                         LEFT JOIN wiki ON search_index.wiki_id = wiki.id \
  98                         WHERE search_index.document @@ to_tsquery('english', %s) \
  99                                 ORDER BY ts_rank(search_index.document, to_tsquery('english', %s)) DESC",
 100                         query, query)
 101
 102                 pages = []
 103                 for page in res:
 104                         # Skip any pages the user doesn't have permission for
 105                         if not page.check_acl(account):
 106                                 continue
 107
 108                         # Return any other pages
 109                         pages.append(page)
 110
 111                         # Break when we have found enough pages
 112                         if limit and len(pages) >= limit:
 113                                 break
 114
 115                 return pages
 116
 117         def refresh(self):
 118                 """
 119                         Needs to be called after a page has been changed
 120                 """
 121                 self.db.execute("REFRESH MATERIALIZED VIEW wiki_search_index")
 122
 123         # ACL
 124
 125         def check_acl(self, page, account):
 126                 res = self.db.query("SELECT * FROM wiki_acls \
 127                         WHERE %s ILIKE (path || '%%') ORDER BY LENGTH(path) DESC LIMIT 1", page)
 128
 129                 for row in res:
 130                         # Access not permitted when user is not logged in
 131                         if not account:
 132                                 return False
 133
 134                         # If user is in a matching group, we grant permission
 135                         for group in row.groups:
 136                                 if group in account.groups:
 137                                         return True
 138
 139                         # Otherwise access is not permitted
 140                         return False
 141
 142                 # If no ACLs are found, we permit access
 143                 return True
 144
 145         # Files
 146
 147         def _get_files(self, query, *args):
 148                 res = self.db.query(query, *args)
 149
 150                 for row in res:
 151                         yield File(self.backend, row.id, data=row)
 152
 153         def _get_file(self, query, *args):
 154                 res = self.db.get(query, *args)
 155
 156                 if res:
 157                         return File(self.backend, res.id, data=res)
 158
 159         def get_files(self, path):
 160                 files = self._get_files("SELECT * FROM wiki_files \
 161                         WHERE path = %s AND deleted_at IS NULL ORDER BY filename", path)
 162
 163                 return list(files)
 164
 165         def get_file_by_path(self, path):
 166                 path, filename = os.path.dirname(path), os.path.basename(path)
 167
 168                 return self._get_file("SELECT * FROM wiki_files \
 169                         WHERE path = %s AND filename = %s AND deleted_at IS NULL", path, filename)
 170
 171         def upload(self, path, filename, data, mimetype, author, address):
 172                 # Upload the blob first
 173                 blob = self.db.get("INSERT INTO wiki_blobs(data) VALUES(%s) RETURNING id", data)
 174
 175                 # Create entry for file
 176                 return self._get_file("INSERT INTO wiki_files(path, filename, author_uid, address, \
 177                         mimetype, blob_id, size) VALUES(%s,  %s, %s, %s, %s, %s, %s) RETURNING *", path,
 178                         filename, author.uid, address, mimetype, blob.id, len(data))
 179
 180         def find_image(self, path, filename):
 181                 for p in (path, os.path.dirname(path)):
 182                         file = self.get_file_by_path(os.path.join(p, filename))
 183
 184                         if file and file.is_image():
 185                                 return file
 186
 187
 188 class Page(misc.Object):
 189         def init(self, id, data=None):
 190                 self.id = id
 191                 self.data = data
 192
 193         def __repr__(self):
 194                 return "<%s %s %s>" % (self.__class__.__name__, self.page, self.timestamp)
 195
 196         def __eq__(self, other):
 197                 if isinstance(other, self.__class__):
 198                         return self.id == other.id
 199
 200         def __lt__(self, other):
 201                 if isinstance(other, self.__class__):
 202                         if self.page == other.page:
 203                                 return self.timestamp < other.timestamp
 204
 205                         return self.page < other.page
 206
 207         @staticmethod
 208         def sanitise_page_name(page):
 209                 if not page:
 210                         return "/"
 211
 212                 # Make sure that the page name does NOT end with a /
 213                 if page.endswith("/"):
 214                         page = page[:-1]
 215
 216                 # Make sure the page name starts with a /
 217                 if not page.startswith("/"):
 218                         page = "/%s" % page
 219
 220                 # Remove any double slashes
 221                 page = page.replace("//", "/")
 222
 223                 return page
 224
 225         @property
 226         def url(self):
 227                 return self.page
 228
 229         @property
 230         def full_url(self):
 231                 return "https://wiki.ipfire.org%s" % self.url
 232
 233         @property
 234         def page(self):
 235                 return self.data.page
 236
 237         @property
 238         def title(self):
 239                 return self._title or os.path.basename(self.page[1:])
 240
 241         @property
 242         def _title(self):
 243                 if not self.markdown:
 244                         return
 245
 246                 # Find first H1 headline in markdown
 247                 markdown = self.markdown.splitlines()
 248
 249                 m = re.match(r"^# (.*)( #)?$", markdown[0])
 250                 if m:
 251                         return m.group(1)
 252
 253         @lazy_property
 254         def author(self):
 255                 if self.data.author_uid:
 256                         return self.backend.accounts.get_by_uid(self.data.author_uid)
 257
 258         def _render(self, text):
 259                 logging.debug("Rendering %s" % self)
 260
 261                 # Link images
 262                 replacements = []
 263                 for match in re.finditer(r"!\[(.*?)\]\((.*?)\)", text):
 264                         alt_text, url = match.groups()
 265
 266                         # Skip any absolute and external URLs
 267                         if url.startswith("/") or url.startswith("https://") or url.startswith("http://"):
 268                                 continue
 269
 270                         # Try to split query string
 271                         url, delimiter, qs = url.partition("?")
 272
 273                         # Parse query arguments
 274                         args = urllib.parse.parse_qs(qs)
 275
 276                         # Find image
 277                         file = self.backend.wiki.find_image(self.page, url)
 278                         if not file:
 279                                 continue
 280
 281                         # Scale down the image if not already done
 282                         if not "s" in args:
 283                                 args["s"] = "768"
 284
 285                         # Format URL
 286                         url = "%s?%s" % (file.url, urllib.parse.urlencode(args))
 287
 288                         replacements.append((match.span(), file, alt_text, url))
 289
 290                 # Apply all replacements
 291                 for (start, end), file, alt_text, url in reversed(replacements):
 292                         text = text[:start] + "[![%s](%s)](%s?action=detail)" % (alt_text, url, file.url) + text[end:]
 293
 294                 # Add wiki links
 295                 patterns = (
 296                         (r"\[\[([\w\d\/\-\.]+)(?:\|(.+?))\]\]", r"\1", r"\2", None, True),
 297                         (r"\[\[([\w\d\/\-\.]+)\]\]", r"\1", r"\1", self.backend.wiki.get_page_title, True),
 298
 299                         # External links
 300                         (r"\[\[((?:ftp|git|https?|rsync|sftp|ssh|webcal)\:\/\/.+?)(?:\|(.+?))\]\]",
 301                                 r"\1", r"\2", None, False),
 302
 303                         # Mail
 304                         (r"\[\[([a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+)\]\]",
 305                                 r"\1", r"\1", None, False),
 306                 )
 307
 308                 for pattern, link, title, repl, internal in patterns:
 309                         replacements = []
 310
 311                         for match in re.finditer(pattern, text):
 312                                 l = match.expand(link)
 313                                 t = match.expand(title)
 314
 315                                 if internal:
 316                                         # Allow relative links
 317                                         if not l.startswith("/"):
 318                                                 l = os.path.join(self.page, l)
 319
 320                                         # Normalise links
 321                                         l = os.path.normpath(l)
 322
 323                                 if callable(repl):
 324                                         t = repl(l) or t
 325
 326                                 replacements.append((match.span(), t or l, l))
 327
 328                         # Apply all replacements
 329                         for (start, end), t, l in reversed(replacements):
 330                                 text = text[:start] + "[%s](%s)" % (t, l) + text[end:]
 331
 332                 # Borrow this from the blog
 333                 return self.backend.blog._render_text(text, lang="markdown")
 334
 335         @property
 336         def markdown(self):
 337                 return self.data.markdown or ""
 338
 339         @property
 340         def html(self):
 341                 return self._render(self.markdown)
 342
 343         @property
 344         def timestamp(self):
 345                 return self.data.timestamp
 346
 347         def was_deleted(self):
 348                 return self.markdown is None
 349
 350         @lazy_property
 351         def breadcrumbs(self):
 352                 return self.backend.wiki.make_breadcrumbs(self.page)
 353
 354         def get_latest_revision(self):
 355                 revisions = self.get_revisions()
 356
 357                 # Return first object
 358                 for rev in revisions:
 359                         return rev
 360
 361         def get_revisions(self):
 362                 return self.backend.wiki._get_pages("SELECT * FROM wiki \
 363                         WHERE page = %s ORDER BY timestamp DESC", self.page)
 364
 365         @lazy_property
 366         def previous_revision(self):
 367                 return self.backend.wiki._get_page("SELECT * FROM wiki \
 368                         WHERE page = %s AND timestamp < %s ORDER BY timestamp DESC \
 369                         LIMIT 1", self.page, self.timestamp)
 370
 371         @property
 372         def changes(self):
 373                 return self.data.changes
 374
 375         # ACL
 376
 377         def check_acl(self, account):
 378                 return self.backend.wiki.check_acl(self.page, account)
 379
 380         # Sidebar
 381
 382         @lazy_property
 383         def sidebar(self):
 384                 parts = self.page.split("/")
 385
 386                 while parts:
 387                         sidebar = self.backend.wiki.get_page("%s/sidebar" % os.path.join(*parts))
 388                         if sidebar:
 389                                 return sidebar
 390
 391                         parts.pop()
 392
 393         # Watchers
 394
 395         @lazy_property
 396         def diff(self):
 397                 if self.previous_revision:
 398                         diff = difflib.unified_diff(
 399                                 self.previous_revision.markdown.splitlines(),
 400                                 self.markdown.splitlines(),
 401                         )
 402
 403                         return "\n".join(diff)
 404
 405         @property
 406         def watchers(self):
 407                 res = self.db.query("SELECT uid FROM wiki_watchlist \
 408                         WHERE page = %s", self.page)
 409
 410                 for row in res:
 411                         # Search for account by UID and skip if none was found
 412                         account = self.backend.accounts.get_by_uid(row.uid)
 413                         if not account:
 414                                 continue
 415
 416                         # Return the account
 417                         yield account
 418
 419         def is_watched_by(self, account):
 420                 res = self.db.get("SELECT 1 FROM wiki_watchlist \
 421                         WHERE page = %s AND uid = %s", self.page, account.uid)
 422
 423                 if res:
 424                         return True
 425
 426                 return False
 427
 428         def add_watcher(self, account):
 429                 if self.is_watched_by(account):
 430                         return
 431
 432                 self.db.execute("INSERT INTO wiki_watchlist(page, uid) \
 433                         VALUES(%s, %s)", self.page, account.uid)
 434
 435         def remove_watcher(self, account):
 436                 self.db.execute("DELETE FROM wiki_watchlist \
 437                         WHERE page = %s AND uid = %s", self.page, account.uid)
 438
 439         def _send_watcher_emails(self, excludes=[]):
 440                 # Nothing to do if there was no previous revision
 441                 if not self.previous_revision:
 442                         return
 443
 444                 for watcher in self.watchers:
 445                         # Skip everyone who is excluded
 446                         if watcher in excludes:
 447                                 logging.debug("Excluding %s" % watcher)
 448                                 continue
 449
 450                         logging.debug("Sending watcher email to %s" % watcher)
 451
 452                         # Compose message
 453                         self.backend.messages.send_template("wiki/messages/page-changed",
 454                                 recipients=[watcher], page=self, priority=-10)
 455
 456
 457 class File(misc.Object):
 458         def init(self, id, data):
 459                 self.id   = id
 460                 self.data = data
 461
 462         @property
 463         def url(self):
 464                 return os.path.join(self.path, self.filename)
 465
 466         @property
 467         def path(self):
 468                 return self.data.path
 469
 470         @property
 471         def filename(self):
 472                 return self.data.filename
 473
 474         @property
 475         def mimetype(self):
 476                 return self.data.mimetype
 477
 478         @property
 479         def size(self):
 480                 return self.data.size
 481
 482         @lazy_property
 483         def author(self):
 484                 if self.data.author_uid:
 485                         return self.backend.accounts.get_by_uid(self.data.author_uid)
 486
 487         @property
 488         def created_at(self):
 489                 return self.data.created_at
 490
 491         def is_pdf(self):
 492                 return self.mimetype in ("application/pdf", "application/x-pdf")
 493
 494         def is_image(self):
 495                 return self.mimetype.startswith("image/")
 496
 497         @lazy_property
 498         def blob(self):
 499                 res = self.db.get("SELECT data FROM wiki_blobs \
 500                         WHERE id = %s", self.data.blob_id)
 501
 502                 if res:
 503                         return bytes(res.data)
 504
 505         def get_thumbnail(self, size):
 506                 cache_key = "-".join((self.path, util.normalize(self.filename), self.created_at.isoformat(), "%spx" % size))
 507
 508                 # Try to fetch the data from the cache
 509                 thumbnail = self.memcache.get(cache_key)
 510                 if thumbnail:
 511                         return thumbnail
 512
 513                 # Generate the thumbnail
 514                 thumbnail = self._generate_thumbnail(size)
 515
 516                 # Put it into the cache for forever
 517                 self.memcache.set(cache_key, thumbnail)
 518
 519                 return thumbnail
 520
 521         def _generate_thumbnail(self, size, **args):
 522                 image = PIL.Image.open(io.BytesIO(self.blob))
 523
 524                 # Remove any alpha-channels
 525                 if image.format == "JPEG" and not image.mode == "RGB":
 526                         # Make a white background
 527                         background = PIL.Image.new("RGBA", image.size, (255,255,255))
 528
 529                         # Flatten both images together
 530                         flattened_image = PIL.Image.alpha_composite(background, image)
 531
 532                         # Remove the alpha channel
 533                         image = flattened_image.convert("RGB")
 534
 535                 # Resize the image to the desired resolution
 536                 image.thumbnail((size, size), PIL.Image.LANCZOS)
 537
 538                 if image.format == "JPEG":
 539                         # Apply a gaussian blur to make compression easier
 540                         image = image.filter(PIL.ImageFilter.GaussianBlur(radius=0.05))
 541
 542                         # Arguments to optimise the compression
 543                         args.update({
 544                                 "subsampling" : "4:2:0",
 545                                 "quality"     : 70,
 546                         })
 547
 548                 with io.BytesIO() as f:
 549                         # If writing out the image does not work with optimization,
 550                         # we try to write it out without any optimization.
 551                         try:
 552                                 image.save(f, image.format, optimize=True, **args)
 553                         except:
 554                                 image.save(f, image.format, **args)
 555
 556                         return f.getvalue()