PiratesIRC/Stream-Mapparr 1.26.2491549

on GitHub

1.26.2491549 (2026-09-06)

Added

• New action, Scan for Placeholder Patterns, which finds the numbered stream
  name families your Placeholder Name Patterns do not cover. Requested as
  issue #43. The setting only ever helped with the naming schemes you already
  thought to write down, and nothing in the interface told apart "this
  installation has no placeholder families" from "the patterns you wrote match
  none of them". The reporter found two whole uncovered families, one of them
  their largest, only by pulling every stream name through the API by hand.

  The scan replaces the numbers in every stream name with a slot, so MAX 100 and
  MAX 101 become the one family MAX #, and reports the families that no
  configured pattern covers, each with an anchored regular expression to paste.
  It reads one database column that matching already loads, opens no provider
  connection, changes no setting and writes nothing to the database. The full
  readout goes to /config/stream-mapparr/placeholder-name-scan.txt, because a
  notification shows only about 280 characters.

  Three decisions in it were measured rather than assumed, against 25,068 live
  stream names:

  A digit immediately followed by K is left alone, because 4K and 8K are
  resolution tags rather than slot numbers. With that rule off, 15 further
  templates covering 418 streams are grouped by their resolution tag instead of
  by a slot number.

  A family needs at least three different numbers in its slot, not merely three
  streams. Five rows of HBO 1 from five sources are five sources for one name,
  not five slots. A separate stream-count threshold was written and then removed,
  because three distinct numbers already implies three streams, so the second
  rule could never refuse anything the first admitted.

  Families whose streams carry EPG data are reported first and separately from
  those carrying none, because a placeholder can only ever be resolved when
  there is guide data to resolve it from. This matters more than expected: on
  this installation 131 families are uncovered and only 16 hold a stream with an
  EPG identifier, so a single undifferentiated list would report a problem eight
  times larger than the one worth acting on. The detailed list is capped, and
  says how many it left out rather than cutting silently.

  The readout is plain ASCII, the same rule as the CSV export preamble, with any
  other character written as a backslash-u escape. Provider names really do carry
  such characters here. Python regular expressions accept that form, so every
  suggested pattern still matches the name it came from; all 132 suggestions
  generated from live data were checked against their own example name.

  This reports only. Nothing is ever added to your pattern list, and not every
  numbered family is a placeholder: a numbered channel family whose names are
  already informative matches better as it is.

  Two code reviews of the first version found four defects, all reproduced
  before they were changed and all now covered by a named test that fails when
  the fix is reverted:

  A character above the basic plane, such as an emoji or a flag, was escaped in
  the four-digit form. That form carries a minimum of four digits, not exactly
  four, so an emoji produced five and Python read only the first four. The
  pattern compiled, matched nothing, and the family kept reporting as uncovered
  with nothing saying why. The earlier test used a character inside the basic
  plane, the one class that already worked.

  A backslash followed by a digit produced the same two characters as an escaped
  literal hash, so two different names grouped into one family and the suggested
  pattern matched neither.

  The scan ran inside the request with no yield, no input cap and no time
  budget, while the pattern safety gate deliberately admits patterns that can
  backtrack polynomially on the promise that the runtime bounds them. It now
  hands the worker back every 500 names, skips a name over 500 characters and
  stops at a five second budget, using the same limits as the regex
  pre-processing path. A stopped scan is reported as partial rather than passed
  off as finished, because a family the walk never reached is missing, not
  covered.

  A failure to write the readout returned a plain success. The full readout
  exists only in that file, so the notification now says the file is missing,
  and says it first, because the notification drops lines from the end.

  The ranking was reversed after four independent reviews and a measurement
  of the live database. The first version ranked uncovered families by how
  many of their streams carried an EPG identifier, and split the readout into
  families that carried one and families that did not, suggesting a pattern only
  for the first group. Measured against the live installation, that was wrong in
  both directions.

  Reading the code settles which direction the risk runs. A stream with no EPG
  identifier returns from the resolver immediately, so adding a pattern for such
  a family cannot change matching at all. A stream that does resolve has its
  matching name replaced by the programme currently airing. So the old ranking
  promoted the families where a pattern changes behaviour and demoted the ones
  where it cannot.

  Measured on the database rather than argued: 3,700 streams carry an EPG
  identifier, only 218 of those identifiers match a guide row, and 28 streams
  could resolve a programme at the moment of measurement. Of the 16 families the
  old ranking promoted, one contained a stream whose identifier matched a guide
  row, and none could resolve a programme. That one was a numbered channel
  lineup, which is the case a pattern harms. Meanwhile the largest family on the
  installation, at 273 streams, was reduced to a single line with no suggested
  pattern.

  Families are now ranked by size, largest first, which is what the reporter of
  the issue did by hand. Every uncovered family gets a suggested pattern. The
  EPG identifier count stays as a note on each family, saying how many of its
  streams carry one out of how many, and a family where at least half of them do
  is marked with a caution explaining that a pattern there replaces a working
  name with whatever is airing. On the live data that caution fires on 13
  families, every one a recognisable channel lineup such as ITV or BeIN Sports,
  and those families moved from the top of the list to ranks 31 to 60.

  Also from those reviews: distinct slot numbers are counted within one slot
  rather than across slots, a resolution tag groups regardless of the case of
  its K, a suggestion too long for the setting to accept is marked as such
  instead of being offered as though it worked, a database row that is not a
  dictionary is skipped rather than raising, and the second listing is capped
  and says how many it left out.