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