Optimized compilation of string patterns - #540

Draft

chengluyu wants to merge 34 commits into

hkust-taco:hkmc2from

chengluyu:compiled-string-patterns

Commits on Jul 3, 2026

Compile string patterns with automata

chengluyu
and
claude
committed
105fa99

Commits on Jul 4, 2026

Shrink compiled string automata tables

chengluyu
and
claude
committed
592c813
Merge branch 'hkmc2' into compiled-string-patterns
LPTK
authored
e2fa0f7

Commits on Jul 11, 2026

Add opt-in size measurement for compiled string pattern tables

chengluyu
and
claude
committed
a4f702f
Fuse op-carrying single-edge states in string automata

chengluyu
and
claude
committed
eb7a728
Minimize the reverse DFA in recognition-only string tables

chengluyu
and
claude
committed
4ab837a
Pack string-table integer sections as base64 varints

chengluyu
and
claude
committed
cc60472
Support astral code points in character range patterns

chengluyu
and
claude
committed
8aefa77
Warn when string patterns fall back to backtracking matching

chengluyu
and
claude
committed
f6026af
Compile negations of pure patterns in string regions

chengluyu
and
claude
committed
10dfb09

Commits on Jul 12, 2026

Compile conjunctions with one impure branch in string regions

chengluyu
and
claude
committed
8caf76d

Commits on Jul 13, 2026

Merge upstream in
chengluyu
committed
2967a33

Commits on Jul 24, 2026

Merge branch 'hkmc2' into compiled-string-patterns
LPTK
authored
df8ba26
Restore indented blank lines stripped by the string-pattern commits
claude
committed
838ec0d
Add regression tests for compiled string pattern bugs
claude
committed
71e1715
Add review notes for the compiled string pattern PR
claude
committed
5cbc7fd
Fix three miscompilations in compiled string patterns
claude
committed
42a1829
Reorganize the string pattern tests
claude
committed
330a095
Keep constructs the string automaton rejects on the legacy path
claude
committed
c83320a
Drop the empty alphabet class and an unused helper
claude
committed
3c80103
Stop swallowing internal errors when compiling `unapplyStringPrefix`
claude
committed
05920e9

Commits on Jul 25, 2026

Correct the `unapplyStringPrefix` doc: it is always generated
claude
committed
c58e2b8
Collapse empty int ranges to `Never` instead of the wildcard
claude
committed
6095982
Reject guards and chained patterns within string patterns
claude
committed
98995d9
Merge branch 'hkmc2' into compiled-string-patterns

LPTK
and
claude
committed
994f9a8
Drop the empty-int-range guard, subsumed by the encoding fix
claude
committed
476705f
Record the post-review fix status in the PR540 findings
claude
committed
df3eed0
Give range patterns their intended typed, single-character semantics
claude
committed
8f66ead
Run transforms under `@compile` even in match-only mode
claude
committed
90e213e
Fix four more string-pattern miscompilations from the review
claude
committed
0173355
Harden diagnostics, invariants, and coverage of the string compiler
claude
committed
95343ed
Pin the poisoned-definition semantics and record the follow-up fix round
claude
committed
07b5987
Pin the mixed string/class definition the whole-body-region gate protects
claude
committed
21cc53c

Commits on Jul 28, 2026

Merge the PR #540 review round into the string-compiler work

chengluyu
and
claude
committed
0db6ba6