Changeset 5779

Timestamp:

Dec 13, 2017, 12:06:28 PM (17 months ago)

Author:

cameron

Message:

Remove RE compiler dependency on linebreak/mAny

Location:

icGREP/icgrep-devel/icgrep

Files:

• grep_engine.cpp (modified) (2 diffs)
• re/exclude_CC.cpp (modified) (1 diff)
• re/re_compiler.cpp (modified) (6 diffs)
• re/re_compiler.h (modified) (1 diff)

icGREP/icgrep-devel/icgrep/grep_engine.cpp

@@ -24 +24 @@
 #include <re/re_cc.h>
 #include <re/casing.h>
+#include <re/exclude_CC.h>
 #include <re/re_toolchain.h>
 #include <toolchain/toolchain.h>
@@ -118 +119 @@
         REs[i] = resolveGraphemeMode(REs[i], false /* not in grapheme mode at top level*/);
         REs[i] = re::resolveNames(REs[i]);
+        REs[i] = exclude_CC(REs[i], re::makeCC(re::makeCC(0x0A, 0x0D), re::makeCC(re::makeCC(0x85), re::makeCC(0x2028, 0x2029))));
+
         const auto UnicodeSets = re::collectUnicodeSets(REs[i]);
         std::vector<std::vector<unsigned>> exclusiveSetIDs;

icGREP/icgrep-devel/icgrep/re/exclude_CC.cpp

@@ -58 +58 @@
             case Name::Type::Unicode:
             case Name::Type::UnicodeProperty:
-                return exclude_CC(n->getDefinition(), cc);
+                return makeName(subtractCC(cast<CC>(n->getDefinition()), cc));
             case Name::Type::ZeroWidth:
                 return re;

icGREP/icgrep-devel/icgrep/re/re_compiler.cpp

@@ -87 +87 @@
         CC * one = makeCC(std::move(*i->first));
         CC * two = makeCC(std::move(*i->second));
-        PabloAST * pablo_one = pb.createAnd(mCCCompiler.compileCC(one), mAny);
-        PabloAST * pablo_two = pb.createAnd(mCCCompiler.compileCC(two), mAny);
+        PabloAST * pablo_one = mCCCompiler.compileCC(one);
+        PabloAST * pablo_two = mCCCompiler.compileCC(two);
         PabloAST * one1 = pb.createAdvance(pablo_one, 1, "one1");
         PabloAST * follow = pb.createAnd(pb.createScanThru(pb.createAnd(mInitial, one1), mNonFinal), pablo_two);
@@ -145 +145 @@
 inline MarkerType RE_Compiler::compileAny(const MarkerType m, PabloBuilder & pb) {
     PabloAST * nextFinalByte = markerVar(AdvanceMarker(m, MarkerPosition::FinalPostPositionUnit, pb));
-    PabloAST * lb = mLineBreak;
-    if (UNICODE_LINE_BREAK) {
-        lb = pb.createOr(mLineBreak, mCRLF);
-    }
-    return makeMarker(MarkerPosition::FinalMatchUnit, pb.createAnd(nextFinalByte, pb.createNot(lb), "dot"));
+    return makeMarker(MarkerPosition::FinalMatchUnit, nextFinalByte);
 }
 
@@ -159 +155 @@
         nextPos = AdvanceMarker(marker, MarkerPosition::FinalPostPositionUnit, pb);
     }
-    return makeMarker(MarkerPosition::FinalMatchUnit, pb.createAnd(markerVar(nextPos), pb.createAnd(mCCCompiler.compileCC(cc, pb), mAny)));
+    return makeMarker(MarkerPosition::FinalMatchUnit, pb.createAnd(markerVar(nextPos), mCCCompiler.compileCC(cc, pb)));
 }
 
@@ -271 +267 @@
         if (LLVM_LIKELY(markerPos(cond) == MarkerPosition::FinalMatchUnit)) {
             MarkerType postCond = AdvanceMarker(cond, MarkerPosition::FinalPostPositionUnit, pb);
-            PabloAST * boundaryCond = pb.createAnd(mAny, pb.createXor(markerVar(cond), markerVar(postCond)));
+            PabloAST * boundaryCond = pb.createXor(markerVar(cond), markerVar(postCond));
             if (a->getSense() == Assertion::Sense::Negative) {
                 boundaryCond = pb.createNot(boundaryCond);
@@ -663 +659 @@
 , mLineBreak(nullptr)
 , mCRLF(nullptr)
-, mAny(nullptr)
 , mGraphemeBoundaryRule(nullptr)
 , mInitial(nullptr)
@@ -674 +669 @@
     Var * const linebreak = mKernel->getInputStreamVar("linebreak");
     mLineBreak = mPB.createExtract(linebreak, mPB.getInteger(0));
-    mAny = mPB.createNot(mLineBreak, "any");
     Var * const required = mKernel->getInputStreamVar("required");
     mInitial = mPB.createExtract(required, mPB.getInteger(0));

icGREP/icgrep-devel/icgrep/re/re_compiler.h

@@ -126 +126 @@
     pablo::PabloAST *                               mLineBreak;
     pablo::PabloAST *                               mCRLF;
-    pablo::PabloAST *                               mAny;
     pablo::PabloAST *                               mGraphemeBoundaryRule;
     pablo::PabloAST *                               mInitial;