[tor-bugs] #9380 [BridgeDB]: Increase strictness of descriptor parsers according to spec

[Tor Bug Tracker & Wiki]

#9380: Increase strictness of descriptor parsers according to spec
-------------------------+-------------------------------------------------
     Reporter:  sysrqb   |      Owner:
         Type:           |     Status:  needs_revision
  enhancement            |  Milestone:
     Priority:  normal   |    Version:
    Component:           |   Keywords:  descriptor,tor-
  BridgeDB               |  bridge,stem,bridgedb,refactoring
   Resolution:           |  Parent ID:  #9462
Actual Points:           |
       Points:           |
-------------------------+-------------------------------------------------
Changes (by isis):

 * status:  new => needs_revision
 * cc: isis@… (added)
 * keywords:   => descriptor,tor-bridge,stem,bridgedb,refactoring
 * parent:   => #9462


Comment:

 Arg. When replying to this two days ago, I did a stupid thing where I
 Ctrl-X'd the text I'd written to paste it into my editor, and then I
 highlighted something else and lost it.

 tl;drewrite:
   * BridgeDB should use stem.
   * Most of BridgeDB's descriptor parsing code should be completely
 rewritten. (see the later point on unittests!)
   * Bridge descriptors are not well documented, and in some places not
 clearly spec'd. This should also be fixed.
   * There is not yet, as far as I know, an easy way to generate fake
 unsanitised bridge descriptors which would actually pass as real, current
 unsanitised bridge descriptors.
   * If atagar is amenable to such, we should create classes for bridge
 descriptors, (or possible modify the existing ones with some sort of
 `is_sanitised_version=False` attribute.
   * We need a real unittest framework, I started looking into a library
 which Lunar pointed me at, called [http://falcao.it/sure/reference.html
 sure], and if it plays well with `twisted.trial`. Especially for checking
 that descriptor parsing is going according to plan, `sure` seems really
 nice. (This should possibly be another ticket.)

 Rather than merging, I am going to rename #9462 and set it as this
 ticket's parent. To make it more concrete, should this ticket then be
 about "BridgeDB should use stem for parsing descriptors according to
 torspec"?

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/9380#comment:6>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online