src/backend/wiki.py

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