Integrate combinable DFA capture resolution for our PCRE dialect regexes, part 1.

.github/workflows/ci.yml

@@ -374,7 +374,7 @@ jobs:
         cc: [ clang ]
         make: [ bmake ]
         debug: [ DEBUG, RELEASE ] # RELEASE=1 is a no-op
-        mode: [ m, p, d ]
+        mode: [ r, s, m, i, M, p ]
         exclude:
           - os: macos
             cc: gcc # it's clang anyway
@@ -409,8 +409,15 @@ jobs:
         path: ${{ env.build }}
         key: build-${{ matrix.make }}-${{ matrix.os }}-${{ matrix.cc }}-${{ matrix.debug }}-${{ matrix.san }}-${{ github.sha }}
 
-    # note we do the fuzzing unconditionally; each run adds to the corpus
+    # note we do the fuzzing unconditionally; each run adds to the corpus.
+    #
+    # We only run fuzzing for PRs in the base repo, this prevents attempting
+    # to purge the seed cache from a PR syncing a forked repo, which fails
+    # due to a permissions error (I'm unsure why, I think PRs from clones can't
+    # purge a cache in CI presumably for security/DoS reasons). PRs from clones
+    # still run fuzzing, just from empty, and do not save their seeds.
     - name: Restore seeds (mode ${{ matrix.mode }})
+      if: github.repository == 'katef/libfsm'
       uses: actions/cache/restore@v3
       id: cache-seeds
       with:

Makefile

@@ -108,6 +108,7 @@ SUBDIR += src
 SUBDIR += tests/capture
 SUBDIR += tests/complement
 SUBDIR += tests/gen
+SUBDIR += tests/idmap
 SUBDIR += tests/intersect
 #SUBDIR += tests/ir # XXX: fragile due to state numbering
 SUBDIR += tests/eclosure

fuzz/Makefile

@@ -7,6 +7,15 @@ ${BUILD}/fuzz/: ${BUILD}
 
 DIR += ${BUILD}/fuzz
 
+# Uncomment to enable capture fuzzing using PCRE as a test oracle.
+#PCRE_CMP=1
+
+.if PCRE_CMP
+PKG += libpcre2-8
+LFLAGS.fuzzer += ${LIBS.libpcre2-8}
+CFLAGS.${SRC:Mfuzz/target.c} += -DCMP_PCRE=1
+.endif
+
 .for src in ${SRC:Mfuzz/*.c}
 CFLAGS.${src} += -std=c99
 .endfor
@@ -15,7 +24,7 @@ CFLAGS.${src} += -std=c99
 fuzz:: ${BUILD}/fuzz/fuzzer
 
 ${BUILD}/fuzz/fuzzer: mkdir
-	${CC} -o $@ ${LFLAGS} ${.ALLSRC:M*.o} ${.ALLSRC:M*.a}
+	${CC} -o $@ ${LFLAGS} ${LFLAGS.fuzzer} ${.ALLSRC:M*.o} ${.ALLSRC:M*.a}
 
 .for lib in ${LIB:Mlibfsm} ${LIB:Mlibre}
 ${BUILD}/fuzz/fuzzer: ${BUILD}/lib/${lib:R}.a

fuzz/run_fuzzer

@@ -4,6 +4,8 @@ BUILD=../build
 FUZZER=${BUILD}/fuzz/fuzzer
 SEEDS=${BUILD}/fuzz/fuzzer_seeds
 
+ARG=$1
+
 SECONDS=${SECONDS:-60}
 WORKERS=${WORKERS:-4}
 SEEDS=${SEEDS:-seeds}
@@ -25,5 +27,9 @@ if [ ! -d "${SEEDS}" ]; then
    mkdir -p "${SEEDS}"
 fi
 
-echo "\n==== ${FUZZER}"
-${FUZZER} -jobs=${WORKERS} -workers=${WORKERS} -max_total_time=${SECONDS} ${SEEDS}
+if [ -z "${ARG}" ]; then
+    echo "\n==== ${FUZZER}"
+    exec ${FUZZER} -jobs=${WORKERS} -workers=${WORKERS} -max_total_time=${SECONDS} ${SEEDS}
+else
+    exec ${FUZZER} ${ARG}
+fi