351 Commits

Author SHA1 Message Date
Justin Viiret
9b7eca5400 rose: dump leftfix/suffix queue indices 2016-07-08 10:44:56 +10:00
Boris Nagaev
6d87533ef0 fix add_to_engine_blob for iterator=pointer 2016-07-06 19:46:41 +03:00
Justin Viiret
614ca0accf rose: always push CHECK_BOUNDS onto end of program 2016-06-01 10:56:57 +10:00
Justin Viiret
9826522e34 rose: fix CHECK_NOT_HANDLED placement bug
The CHECK_NOT_HANDLED instruction was being inserted into an already
partially-flattened program, which would result in jump offsets becoming
incorrect.

This change places it as part of the normal flow of program
construction, which avoids this issue.
2016-06-01 10:56:52 +10:00
Alex Coyte
11896dcf42 restore ability to ignore common preds for merges
Check for common preds was broken during refactoring in b9c5d65:
    Rework literal overlap checks for merging engines
2016-05-18 16:26:22 +10:00
Justin Viiret
9e0ec02ac9 rose: assert that program offset is sane 2016-05-18 16:22:01 +10:00
Justin Viiret
ee7f31ac39 mpv: native report remapping 2016-05-18 16:22:01 +10:00
Justin Viiret
c101beb541 castle, lbr: native report remap 2016-05-18 16:21:36 +10:00
Justin Viiret
ec985a62f8 castle: add nfa kind to CastleProto 2016-05-18 16:21:03 +10:00
Justin Viiret
1f41a921f2 mcclellan, gough: native report remapping 2016-05-18 16:20:45 +10:00
Justin Viiret
611579511c rose: remap reports to program offsets 2016-05-18 16:20:42 +10:00
Justin Viiret
f40aa6bd40 report_manager: store report -> program mapping 2016-05-18 16:20:38 +10:00
Justin Viiret
fd7f30a5b5 Merge functions no longer need transfer function 2016-04-20 13:34:57 +10:00
Justin Viiret
319d47ae4f Remove OutfixInfo::chained (which meant "is MPV") 2016-04-20 13:34:57 +10:00
Justin Viiret
16985ab6e5 Use a visitor for outfix engine construction 2016-04-20 13:34:57 +10:00
Justin Viiret
32c866a8f9 OutfixInfo: use boost::variant for engines 2016-04-20 13:34:57 +10:00
Justin Viiret
fa27025bcb Wrap MPV puffettes in a struct 2016-04-20 13:34:57 +10:00
Justin Viiret
1d85987d96 FINAL_REPORT: Add specialised instruction
Specialisation of the REPORT instruction that also terminates execution
of the program. Improves performance on programs that generate many
reports.
2016-04-20 13:34:57 +10:00
Justin Viiret
36150bbc19 Rose: replace internal_report with program
Replace the use of the internal_report structure (for reports from
engines, MPV etc) with the Rose program interpreter.

SOM processing was reworked to use a new som_operation structure that is
embedded in the appropriate instructions.
2016-04-20 13:34:57 +10:00
Justin Viiret
50885f210a exhaust: Update interface
- Only use functions in exhaust.h for valid ekeys
 - Use INVALID_EKEY everywhere (remove dupe END_EXHAUST sentinel)
2016-04-20 13:34:55 +10:00
Justin Viiret
f2c0a66b6f Rose: use a multibit for the exhaustion vector
Previously, the exhaustion vector was a standard bitvector, which
required an expensive memset() call at init for databases with a large
number of exhaustion keys.
2016-04-20 13:34:55 +10:00
Alex Coyte
4a482b07ea UE-2919: add termination checks around handleReportInternally 2016-04-20 13:34:54 +10:00
Justin Viiret
9c9eb2ca9a makeReport: remove unused param 2016-04-20 13:34:54 +10:00
Justin Viiret
ff7e101757 Rose: be more precise with CATCHUP instruction
- Fix bugs introduced by recent addition of the boundary program. It's
  not safe to do catchup there.
- Only do catchup once per report set, when necessary.
2016-04-20 13:34:54 +10:00
Justin Viiret
93a262e54c Rose: use program for boundary reports
Use the program to handle report lists at boundaries, rather than the
previous list-of-reports approach.
2016-04-20 13:34:54 +10:00
Justin Viiret
24ccefa3d3 Rose: make all NFA callbacks take scratch as ctx 2016-04-20 13:34:54 +10:00
Justin Viiret
6af1ead0aa Rose: make HWLM callbacks take scratch as ctx 2016-04-20 13:34:54 +10:00
Justin Viiret
f7a40b200d Rose: remove dead function roseHandleMatch 2016-04-20 13:34:54 +10:00
Justin Viiret
b093616aff Rose build: move HWLM build code to own file
To reduce the size of rose_build_bytecode.cpp a little, move the code
that deals with HWLM literal tables into its own new file.
2016-04-20 13:34:54 +10:00
Justin Viiret
67b9784dae Rose: use program for all literal matches
Unifies all literal match paths so that the Rose program is used for all
of them. This removes the previous specialised "direct report" and
"multi direct report" paths. Some additional REPORT instruction work was
necessary for this.

Reworked literal construction path at compile time in prep for using
program offsets as literal IDs.

Completely removed the anchored log runtime, which is no longer worth
the extra complexity.
2016-04-20 13:34:54 +10:00
Justin Viiret
6bcccb4c5d Rose: further generalise literal dedupe work 2016-03-01 11:36:10 +11:00
Justin Viiret
f519fd9bcd Rose: don't assume roles with >1 lit need dedupe
We only require dedupe for such roles when they have literals that can
arrive simultaneously (i.e. one literal overlaps with the suffix of
another).
2016-03-01 11:36:10 +11:00
Justin Viiret
0d9422eed7 Rose: make DR runtime path external report only 2016-03-01 11:36:10 +11:00
Justin Viiret
b62711a18f Rose: only use direct reports for external reports 2016-03-01 11:36:10 +11:00
Justin Viiret
cc5db61686 Rose: allow DR literals to share vertices 2016-03-01 11:36:09 +11:00
Justin Viiret
31dea6c275 Rose: allow direct reports to be aliased 2016-03-01 11:36:09 +11:00
Justin Viiret
6294b8a89b Rose: assign DRs in allocateFinalLiteralId
Previously, direct reports were allocated earlier; now all final IDs are
assigned in the same place.
2016-03-01 11:36:09 +11:00
Justin Viiret
6e8f394d8d Make comparison signed (fix warning) 2016-03-01 11:36:09 +11:00
Justin Viiret
c055f81398 Dump: give abs jump targets in output 2016-03-01 11:35:09 +11:00
Justin Viiret
129578f970 Rose program: Improvements to debug/assertions
- Add current pc to debug printf.
- Assert that pc doesn't escape the RoseEngine structure.
2016-03-01 11:35:09 +11:00
Justin Viiret
030cc2a487 Update Rose state layout description
Many of the things in this list no longer existed.
2016-03-01 11:35:09 +11:00
Justin Viiret
5a1dd54049 Split CHECK_LEFTFIX into CHECK_{INFIX,PREFIX} 2016-03-01 11:35:08 +11:00
Justin Viiret
7a775e5fab reduceInfixQueue: factor out any_work check 2016-03-01 11:34:57 +11:00
Justin Viiret
4d5710a84a Rename rosePrefixCheckMiracles to roseLeftfix... 2016-03-01 11:34:57 +11:00
Justin Viiret
8f3471dd4e Rename reduceQueue to reduceInfixQueue 2016-03-01 11:34:57 +11:00
Justin Viiret
bba8b62635 Use rose_inline rather than really_inline 2016-03-01 11:34:57 +11:00
Justin Viiret
fb8747295e roseTestLeftfix: unify common "nfa is dead" code 2016-03-01 11:34:57 +11:00
Justin Viiret
996eba9686 Add CATCH_UP to report_block, not "parent" program
Also ensure that exhaustion check happens after catch up, as catch up
may fire reports (which could exhaust).
2016-03-01 11:34:57 +11:00
Justin Viiret
7b54856642 Rose: allow block-mode merge of small prefixes
Previously, we disallowed the merging of all Rose prefixes in block mode
where the literal sets are not identical.

This change allows merging if the prefix graphs to be merged are very
small, as a small performance improvement for cases with lots of tiny
prefixes.

This check is deliberately conservative: graphs must have some common
vertices, and the result of the merge must not give up any
accelerability.
2016-03-01 11:34:26 +11:00
Justin Viiret
42d34f19d1 Dump: don't call dumpNfaNotes for SOM reverse NFAs
These NFAs have no queue index.
2016-03-01 11:34:26 +11:00