src/backend/wiki.py

   1 #!/usr/bin/python3
   2
   3 import difflib
   4 import logging
   5 import os.path
   6 import re
   7 import urllib.parse
   8
   9 from . import misc
  10 from . import util
  11 from .decorators import *
  12
  13 class Wiki(misc.Object):
  14         def _get_pages(self, query, *args):
  15                 res = self.db.query(query, *args)
  16
  17                 for row in res:
  18                         yield Page(self.backend, row.id, data=row)
  19
  20         def _get_page(self, query, *args):
  21                 res = self.db.get(query, *args)
  22
  23                 if res:
  24                         return Page(self.backend, res.id, data=res)
  25
  26         def __iter__(self):
  27                 return self._get_pages(
  28                         "SELECT wiki.* FROM wiki_current current \
  29                                 LEFT JOIN wiki ON current.id = wiki.id \
  30                                 WHERE current.deleted IS FALSE \
  31                                 ORDER BY page",
  32                 )
  33
  34         def make_path(self, page, path):
  35                 # Nothing to do for absolute links
  36                 if path.startswith("/"):
  37                         pass
  38
  39                 # Relative links (one-level down)
  40                 elif path.startswith("./"):
  41                         path = os.path.join(page, path)
  42
  43                 # All other relative links
  44                 else:
  45                         p = os.path.dirname(page)
  46                         path = os.path.join(p, path)
  47
  48                 # Normalise links
  49                 return os.path.normpath(path)
  50
  51         def page_exists(self, path):
  52                 page = self.get_page(path)
  53
  54                 # Page must have been found and not deleted
  55                 return page and not page.was_deleted()
  56
  57         def get_page_title(self, page, default=None):
  58                 doc = self.get_page(page)
  59                 if doc:
  60                         title = doc.title
  61                 else:
  62                         title = os.path.basename(page)
  63
  64                 return title
  65
  66         def get_page(self, page, revision=None):
  67                 page = Page.sanitise_page_name(page)
  68
  69                 # Split the path into parts
  70                 parts = page.split("/")
  71
  72                 # Check if this is an action
  73                 if any((part.startswith("_") for part in parts)):
  74                         return
  75
  76                 if revision:
  77                         return self._get_page("SELECT * FROM wiki WHERE page = %s \
  78                                 AND timestamp = %s", page, revision)
  79                 else:
  80                         return self._get_page("SELECT * FROM wiki WHERE page = %s \
  81                                 ORDER BY timestamp DESC LIMIT 1", page)
  82
  83         def get_recent_changes(self, account, limit=None):
  84                 pages = self._get_pages("SELECT * FROM wiki \
  85                         ORDER BY timestamp DESC")
  86
  87                 for page in pages:
  88                         if not page.check_acl(account):
  89                                 continue
  90
  91                         yield page
  92
  93                         limit -= 1
  94                         if not limit:
  95                                 break
  96
  97         def create_page(self, page, author, content, changes=None, address=None):
  98                 page = Page.sanitise_page_name(page)
  99
 100                 # Write page to the database
 101                 page = self._get_page("INSERT INTO wiki(page, author_uid, markdown, changes, address) \
 102                         VALUES(%s, %s, %s, %s, %s) RETURNING *", page, author.uid, content or None, changes, address)
 103
 104                 # Send email to all watchers
 105                 page._send_watcher_emails(excludes=[author])
 106
 107                 return page
 108
 109         def delete_page(self, page, author, **kwargs):
 110                 # Do nothing if the page does not exist
 111                 if not self.get_page(page):
 112                         return
 113
 114                 # Just creates a blank last version of the page
 115                 self.create_page(page, author=author, content=None, **kwargs)
 116
 117         def make_breadcrumbs(self, url):
 118                 # Split and strip all empty elements (double slashes)
 119                 parts = list(e for e in url.split("/") if e)
 120
 121                 ret = []
 122                 for part in ("/".join(parts[:i]) for i in range(1, len(parts))):
 123                         ret.append(("/%s" % part, self.get_page_title(part, os.path.basename(part))))
 124
 125                 return ret
 126
 127         def search(self, query, account=None, limit=None):
 128                 res = self._get_pages("SELECT wiki.* FROM wiki_search_index search_index \
 129                         LEFT JOIN wiki ON search_index.wiki_id = wiki.id \
 130                         WHERE search_index.document @@ websearch_to_tsquery('english', %s) \
 131                                 ORDER BY ts_rank(search_index.document, websearch_to_tsquery('english', %s)) DESC",
 132                         query, query)
 133
 134                 pages = []
 135                 for page in res:
 136                         # Skip any pages the user doesn't have permission for
 137                         if not page.check_acl(account):
 138                                 continue
 139
 140                         # Return any other pages
 141                         pages.append(page)
 142
 143                         # Break when we have found enough pages
 144                         if limit and len(pages) >= limit:
 145                                 break
 146
 147                 return pages
 148
 149         def refresh(self):
 150                 """
 151                         Needs to be called after a page has been changed
 152                 """
 153                 self.db.execute("REFRESH MATERIALIZED VIEW wiki_search_index")
 154
 155         def get_watchlist(self, account):
 156                 pages = self._get_pages("""
 157                         WITH pages AS (
 158                                 SELECT
 159                                         *
 160                                 FROM
 161                                         wiki_current
 162                                 LEFT JOIN
 163                                         wiki ON wiki_current.id = wiki.id
 164                         )
 165
 166                         SELECT
 167                                 *
 168                         FROM
 169                                 wiki_watchlist watchlist
 170                         JOIN
 171                                 pages ON watchlist.page = pages.page
 172                         WHERE
 173                                 watchlist.uid = %s
 174                         """, account.uid,
 175                 )
 176
 177                 return sorted(pages)
 178
 179         # ACL
 180
 181         def check_acl(self, page, account):
 182                 res = self.db.query("SELECT * FROM wiki_acls \
 183                         WHERE %s ILIKE (path || '%%') ORDER BY LENGTH(path) DESC LIMIT 1", page)
 184
 185                 for row in res:
 186                         # Access not permitted when user is not logged in
 187                         if not account:
 188                                 return False
 189
 190                         # If user is in a matching group, we grant permission
 191                         for group in row.groups:
 192                                 if account.is_member_of_group(group):
 193                                         return True
 194
 195                         # Otherwise access is not permitted
 196                         return False
 197
 198                 # If no ACLs are found, we permit access
 199                 return True
 200
 201         # Files
 202
 203         def _get_files(self, query, *args):
 204                 res = self.db.query(query, *args)
 205
 206                 for row in res:
 207                         yield File(self.backend, row.id, data=row)
 208
 209         def _get_file(self, query, *args):
 210                 res = self.db.get(query, *args)
 211
 212                 if res:
 213                         return File(self.backend, res.id, data=res)
 214
 215         def get_files(self, path):
 216                 files = self._get_files("SELECT * FROM wiki_files \
 217                         WHERE path = %s AND deleted_at IS NULL ORDER BY filename", path)
 218
 219                 return list(files)
 220
 221         def get_file_by_path(self, path, revision=None):
 222                 path, filename = os.path.dirname(path), os.path.basename(path)
 223
 224                 if revision:
 225                         # Fetch a specific revision
 226                         return self._get_file("SELECT * FROM wiki_files \
 227                                 WHERE path = %s AND filename = %s AND created_at <= %s \
 228                                 ORDER BY created_at DESC LIMIT 1", path, filename, revision)
 229
 230                 # Fetch latest version
 231                 return self._get_file("SELECT * FROM wiki_files \
 232                         WHERE path = %s AND filename = %s AND deleted_at IS NULL",
 233                         path, filename)
 234
 235         def get_file_by_path_and_filename(self, path, filename):
 236                 return self._get_file("SELECT * FROM wiki_files \
 237                         WHERE path = %s AND filename = %s AND deleted_at IS NULL",
 238                         path, filename)
 239
 240         def upload(self, path, filename, data, mimetype, author, address):
 241                 # Replace any existing files
 242                 file = self.get_file_by_path_and_filename(path, filename)
 243                 if file:
 244                         file.delete(author)
 245
 246                 # Upload the blob first
 247                 blob = self.db.get("INSERT INTO wiki_blobs(data) VALUES(%s) \
 248                         ON CONFLICT (digest(data, %s)) DO UPDATE SET data = EXCLUDED.data \
 249                         RETURNING id", data, "MD5")
 250
 251                 # Create entry for file
 252                 return self._get_file("INSERT INTO wiki_files(path, filename, author_uid, address, \
 253                         mimetype, blob_id, size) VALUES(%s,  %s, %s, %s, %s, %s, %s) RETURNING *", path,
 254                         filename, author.uid, address, mimetype, blob.id, len(data))
 255
 256         def render(self, path, text):
 257                 r = WikiRenderer(self.backend, path)
 258
 259                 return r.render(text)
 260
 261
 262 class Page(misc.Object):
 263         def init(self, id, data=None):
 264                 self.id = id
 265                 self.data = data
 266
 267         def __repr__(self):
 268                 return "<%s %s %s>" % (self.__class__.__name__, self.page, self.timestamp)
 269
 270         def __eq__(self, other):
 271                 if isinstance(other, self.__class__):
 272                         return self.id == other.id
 273
 274                 return NotImplemented
 275
 276         def __lt__(self, other):
 277                 if isinstance(other, self.__class__):
 278                         if self.page == other.page:
 279                                 return self.timestamp < other.timestamp
 280
 281                         return self.page < other.page
 282
 283                 return NotImplemented
 284
 285         @staticmethod
 286         def sanitise_page_name(page):
 287                 if not page:
 288                         return "/"
 289
 290                 # Make sure that the page name does NOT end with a /
 291                 if page.endswith("/"):
 292                         page = page[:-1]
 293
 294                 # Make sure the page name starts with a /
 295                 if not page.startswith("/"):
 296                         page = "/%s" % page
 297
 298                 # Remove any double slashes
 299                 page = page.replace("//", "/")
 300
 301                 return page
 302
 303         @property
 304         def url(self):
 305                 return "/docs%s" % self.page
 306
 307         @property
 308         def full_url(self):
 309                 return "https://www.ipfire.org%s" % self.url
 310
 311         @property
 312         def page(self):
 313                 return self.data.page
 314
 315         @property
 316         def title(self):
 317                 return self._title or os.path.basename(self.page[1:])
 318
 319         @property
 320         def _title(self):
 321                 if not self.markdown:
 322                         return
 323
 324                 # Find first H1 headline in markdown
 325                 markdown = self.markdown.splitlines()
 326
 327                 m = re.match(r"^#\s*(.*)( #)?$", markdown[0])
 328                 if m:
 329                         return m.group(1)
 330
 331         @lazy_property
 332         def author(self):
 333                 if self.data.author_uid:
 334                         return self.backend.accounts.get_by_uid(self.data.author_uid)
 335
 336         @property
 337         def markdown(self):
 338                 return self.data.markdown or ""
 339
 340         @property
 341         def html(self):
 342                 lines = []
 343
 344                 # Strip off the first line if it contains a heading (as it will be shown separately)
 345                 for i, line in enumerate(self.markdown.splitlines()):
 346                         if i == 0 and line.startswith("#"):
 347                                 continue
 348
 349                         lines.append(line)
 350
 351                 return self.backend.wiki.render(self.page, "\n".join(lines))
 352
 353         @property
 354         def timestamp(self):
 355                 return self.data.timestamp
 356
 357         def was_deleted(self):
 358                 return not self.markdown
 359
 360         @lazy_property
 361         def breadcrumbs(self):
 362                 return self.backend.wiki.make_breadcrumbs(self.page)
 363
 364         def is_latest_revision(self):
 365                 return self.get_latest_revision() == self
 366
 367         def get_latest_revision(self):
 368                 revisions = self.get_revisions()
 369
 370                 # Return first object
 371                 for rev in revisions:
 372                         return rev
 373
 374         def get_revisions(self):
 375                 return self.backend.wiki._get_pages("SELECT * FROM wiki \
 376                         WHERE page = %s ORDER BY timestamp DESC", self.page)
 377
 378         @lazy_property
 379         def previous_revision(self):
 380                 return self.backend.wiki._get_page("SELECT * FROM wiki \
 381                         WHERE page = %s AND timestamp < %s ORDER BY timestamp DESC \
 382                         LIMIT 1", self.page, self.timestamp)
 383
 384         @property
 385         def changes(self):
 386                 return self.data.changes
 387
 388         # ACL
 389
 390         def check_acl(self, account):
 391                 return self.backend.wiki.check_acl(self.page, account)
 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                         # Check permissions
 451                         if not self.backend.wiki.check_acl(self.page, watcher):
 452                                 logging.debug("Watcher %s does not have permissions" % watcher)
 453                                 continue
 454
 455                         logging.debug("Sending watcher email to %s" % watcher)
 456
 457                         # Compose message
 458                         self.backend.messages.send_template("wiki/messages/page-changed",
 459                                 account=watcher, page=self, priority=-10)
 460
 461         def restore(self, author, address, comment=None):
 462                 changes = "Restore to revision from %s" % self.timestamp.isoformat()
 463
 464                 # Append comment
 465                 if comment:
 466                         changes = "%s: %s" % (changes, comment)
 467
 468                 return self.backend.wiki.create_page(self.page,
 469                         author, self.markdown, changes=changes, address=address)
 470
 471
 472 class File(misc.Object):
 473         def init(self, id, data):
 474                 self.id   = id
 475                 self.data = data
 476
 477         def __eq__(self, other):
 478                 if isinstance(other, self.__class__):
 479                         return self.id == other.id
 480
 481         @property
 482         def url(self):
 483                 return os.path.join(self.path, self.filename)
 484
 485         @property
 486         def path(self):
 487                 return self.data.path
 488
 489         @property
 490         def filename(self):
 491                 return self.data.filename
 492
 493         @property
 494         def mimetype(self):
 495                 return self.data.mimetype
 496
 497         @property
 498         def size(self):
 499                 return self.data.size
 500
 501         @lazy_property
 502         def author(self):
 503                 if self.data.author_uid:
 504                         return self.backend.accounts.get_by_uid(self.data.author_uid)
 505
 506         @property
 507         def created_at(self):
 508                 return self.data.created_at
 509
 510         def delete(self, author=None):
 511                 self.db.execute("UPDATE wiki_files SET deleted_at = NOW(), deleted_by = %s \
 512                         WHERE id = %s", author.uid if author else None, self.id)
 513
 514         @property
 515         def deleted_at(self):
 516                 return self.data.deleted_at
 517
 518         def get_latest_revision(self):
 519                 revisions = self.get_revisions()
 520
 521                 # Return first object
 522                 for rev in revisions:
 523                         return rev
 524
 525         def get_revisions(self):
 526                 revisions = self.backend.wiki._get_files("SELECT * FROM wiki_files \
 527                         WHERE path = %s AND filename = %s ORDER BY created_at DESC", self.path, self.filename)
 528
 529                 return list(revisions)
 530
 531         def is_pdf(self):
 532                 return self.mimetype in ("application/pdf", "application/x-pdf")
 533
 534         def is_image(self):
 535                 return self.mimetype.startswith("image/")
 536
 537         def is_vector_image(self):
 538                 return self.mimetype in ("image/svg+xml",)
 539
 540         def is_bitmap_image(self):
 541                 return self.is_image() and not self.is_vector_image()
 542
 543         @lazy_property
 544         def blob(self):
 545                 res = self.db.get("SELECT data FROM wiki_blobs \
 546                         WHERE id = %s", self.data.blob_id)
 547
 548                 if res:
 549                         return bytes(res.data)
 550
 551         async def get_thumbnail(self, size):
 552                 assert self.is_bitmap_image()
 553
 554                 cache_key = "-".join((
 555                         self.path,
 556                         util.normalize(self.filename),
 557                         self.created_at.isoformat(),
 558                         "%spx" % size,
 559                 ))
 560
 561                 # Try to fetch the data from the cache
 562                 thumbnail = await self.backend.cache.get(cache_key)
 563                 if thumbnail:
 564                         return thumbnail
 565
 566                 # Generate the thumbnail
 567                 thumbnail = util.generate_thumbnail(self.blob, size)
 568
 569                 # Put it into the cache for forever
 570                 await self.backend.cache.set(cache_key, thumbnail)
 571
 572                 return thumbnail
 573
 574
 575 class WikiRenderer(misc.Object):
 576         schemas = (
 577                 "ftp://",
 578                 "git://",
 579                 "http://",
 580                 "https://",
 581                 "rsync://",
 582                 "sftp://",
 583                 "ssh://",
 584                 "webcal://",
 585         )
 586
 587         # Links
 588         links = re.compile(r"<a href=\"(.*?)\">(.*?)</a>")
 589
 590         # Images
 591         images = re.compile(r"<img alt(?:=\"(.*?)\")? src=\"(.*?)\" (?:title=\"(.*?)\" )?/>")
 592
 593         def init(self, path):
 594                 self.path = path
 595
 596         def _render_link(self, m):
 597                 url, text = m.groups()
 598
 599                 # External Links
 600                 for schema in self.schemas:
 601                         if url.startswith(schema):
 602                                 return """<a class="link-external" href="%s">%s</a>""" % \
 603                                         (url, text or url)
 604
 605                 # Emails
 606                 if "@" in url:
 607                         # Strip mailto:
 608                         if url.startswith("mailto:"):
 609                                 url = url[7:]
 610
 611                         return """<a class="link-external" href="mailto:%s">%s</a>""" % \
 612                                 (url, text or url)
 613
 614                 # Everything else must be an internal link
 615                 path = self.backend.wiki.make_path(self.path, url)
 616
 617                 return """<a href="/docs%s">%s</a>""" % \
 618                         (path, text or self.backend.wiki.get_page_title(path))
 619
 620         def _render_image(self, m):
 621                 alt_text, url, caption = m.groups()
 622
 623                 html = """
 624                         <div class="columns is-centered">
 625                                 <div class="column is-8">
 626                                         <figure class="image">
 627                                                 <img src="/docs%s" alt="%s">
 628                                                 <figcaption class="figure-caption">%s</figcaption>
 629                                         </figure>
 630                                 </div>
 631                         </div>
 632                 """
 633
 634                 # Skip any absolute and external URLs
 635                 if url.startswith("/") or url.startswith("https://") or url.startswith("http://"):
 636                         return html % (url, alt_text, caption or "")
 637
 638                 # Try to split query string
 639                 url, delimiter, qs = url.partition("?")
 640
 641                 # Parse query arguments
 642                 args = urllib.parse.parse_qs(qs)
 643
 644                 # Build absolute path
 645                 url = self.backend.wiki.make_path(self.path, url)
 646
 647                 # Find image
 648                 file = self.backend.wiki.get_file_by_path(url)
 649                 if not file or not file.is_image():
 650                         return "<!-- Could not find image %s in %s -->" % (url, self.path)
 651
 652                 # Scale down the image if not already done
 653                 if not "s" in args:
 654                         args["s"] = "920"
 655
 656                 # Append arguments to the URL
 657                 if args:
 658                         url = "%s?%s" % (url, urllib.parse.urlencode(args))
 659
 660                 return html % (url, caption, caption or "")
 661
 662         def render(self, text):
 663                 logging.debug("Rendering %s" % self.path)
 664
 665                 # Borrow this from the blog
 666                 text = self.backend.blog._render_text(text, lang="markdown")
 667
 668                 # Postprocess links
 669                 text = self.links.sub(self._render_link, text)
 670
 671                 # Postprocess images to <figure>
 672                 text = self.images.sub(self._render_image, text)
 673
 674                 return text