From: Ulya Trofimovich <skvadrik@gmail.com>
Date: Mon, 17 Jul 2017 21:36:02 +0000 (+0100)
Subject: Forbid merging of tag variables for tags with and without history.
X-Git-Tag: 1.0~39^2~26
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=ddb3d8507872f9e98f8df8e4929631b88ff31862;p=re2c

Forbid merging of tag variables for tags with and without history.
---

diff --git a/re2c/src/adfa/adfa.cc b/re2c/src/adfa/adfa.cc
index e5267700..48981f42 100644
--- a/re2c/src/adfa/adfa.cc
+++ b/re2c/src/adfa/adfa.cc
@@ -37,6 +37,7 @@ DFA::DFA
 	, charset(dfa.charset)
 	, rules(dfa.rules)
 	, tags(dfa.tags)
+	, listvers(dfa.listvers)
 	, finvers(dfa.finvers)
 	, tcpool(dfa.tcpool)
 	, max_fill (0)
@@ -105,6 +106,7 @@ DFA::~DFA()
 	delete &charset;
 	delete &rules;
 	delete &tags;
+	delete &listvers;
 	delete[] finvers;
 	delete &tcpool;
 }
diff --git a/re2c/src/adfa/adfa.h b/re2c/src/adfa/adfa.h
index 05c3b524..50e3c1d3 100644
--- a/re2c/src/adfa/adfa.h
+++ b/re2c/src/adfa/adfa.h
@@ -70,6 +70,7 @@ struct DFA
 	std::vector<uint32_t> &charset;
 	std::valarray<Rule> &rules;
 	std::vector<Tag> &tags;
+	std::set<tagver_t> &listvers;
 	const tagver_t *finvers;
 	tcpool_t &tcpool;
 	size_t max_fill;
diff --git a/re2c/src/code/emit_dfa.cc b/re2c/src/code/emit_dfa.cc
index 00ccca3b..93600359 100644
--- a/re2c/src/code/emit_dfa.cc
+++ b/re2c/src/code/emit_dfa.cc
@@ -121,50 +121,6 @@ void DFA::emit_dot(OutputFile &o, bool last_cond) const
 	}
 }
 
-static void find_list_tags_cmd(const tcmd_t *p, bool *list)
-{
-	for (; p; p = p->next) {
-		if (tcmd_t::isadd(p)) {
-			list[p->lhs] = list[p->rhs] = true;
-		}
-	}
-}
-
-static void prop_list_tags_cmd(const tcmd_t *p, bool *list)
-{
-	for (; p; p = p->next) {
-		if (tcmd_t::iscopy(p)) {
-			const tagver_t l = p->lhs, r = p->rhs;
-			if (list[l]) list[r] = true;
-			if (list[r]) list[l] = true;
-		}
-	}
-}
-
-static void find_list_tags(const DFA &dfa, bool *list)
-{
-	find_list_tags_cmd(dfa.tcpool[dfa.tags0], list);
-	for (State *s = dfa.head; s; s = s->next) {
-		const Go &go = s->go;
-		find_list_tags_cmd(dfa.tcpool[go.tags], list);
-		for (uint32_t i = 0; i < go.nSpans; ++i) {
-			find_list_tags_cmd(dfa.tcpool[go.span[i].tags], list);
-		}
-		find_list_tags_cmd(dfa.tcpool[s->rule_tags], list);
-		find_list_tags_cmd(dfa.tcpool[s->fall_tags], list);
-	}
-	prop_list_tags_cmd(dfa.tcpool[dfa.tags0], list);
-	for (State *s = dfa.head; s; s = s->next) {
-		const Go &go = s->go;
-		prop_list_tags_cmd(dfa.tcpool[go.tags], list);
-		for (uint32_t i = 0; i < go.nSpans; ++i) {
-			prop_list_tags_cmd(dfa.tcpool[go.span[i].tags], list);
-		}
-		prop_list_tags_cmd(dfa.tcpool[s->rule_tags], list);
-		prop_list_tags_cmd(dfa.tcpool[s->fall_tags], list);
-	}
-}
-
 void DFA::emit(Output & output, uint32_t& ind, bool isLastCond, bool& bPrologBrace)
 {
 	OutputFile &o = output.source;
@@ -181,17 +137,14 @@ void DFA::emit(Output & output, uint32_t& ind, bool isLastCond, bool& bPrologBra
 				tagvars.insert(*tag.name);
 			}
 		}
-		bool *list = new bool[maxtagver + 1]();
-		find_list_tags(*this, list);
 		for (tagver_t v = 1; v <= maxtagver; ++v) {
 			const std::string name = vartag_name(v, opts->tags_prefix);
-			if (list[v]) {
+			if (listvers.find(v) != listvers.end()) {
 				taglistnames.insert(name);
 			} else {
 				tagnames.insert(name);
 			}
 		}
-		delete[] list;
 		ob.tags.insert(tagnames.begin(), tagnames.end());
 		ob.taglists.insert(taglistnames.begin(), taglistnames.end());
 	}
diff --git a/re2c/src/dfa/cfg/interfere.cc b/re2c/src/dfa/cfg/interfere.cc
index 8562e76e..e2d91b92 100644
--- a/re2c/src/dfa/cfg/interfere.cc
+++ b/re2c/src/dfa/cfg/interfere.cc
@@ -10,7 +10,8 @@ static void interfere(const tcmd_t *cmd, const bool *live, bool *interf, bool *b
 
 void cfg_t::interference(const cfg_t &cfg, const bool *live, bool *interf)
 {
-	const size_t nver = static_cast<size_t>(cfg.dfa.maxtagver) + 1;
+	const tagver_t maxver = cfg.dfa.maxtagver + 1;
+	const size_t nver = static_cast<size_t>(maxver);
 	bool *buf = new bool[nver];
 	vals_t *vals = new vals_t[nver]();
 	const cfg_bb_t *b = cfg.bblocks, *e = b + cfg.nbbfin;
@@ -20,6 +21,16 @@ void cfg_t::interference(const cfg_t &cfg, const bool *live, bool *interf)
 		interfere(b->cmd, live, interf, buf, vals, nver);
 	}
 
+	// versions of tags with/without history interfere
+	std::set<tagver_t> &lvs = cfg.dfa.listvers;
+	for (std::set<tagver_t>::iterator i = lvs.begin(); i != lvs.end(); ++i) {
+		for (tagver_t u = *i, v = 0; v < maxver; ++v) {
+			if (lvs.find(v) == lvs.end()) {
+				interf[v * maxver + u] = interf[u * maxver + v] = true;
+			}
+		}
+	}
+
 	delete[] buf;
 	delete[] vals;
 }
diff --git a/re2c/src/dfa/cfg/rename.cc b/re2c/src/dfa/cfg/rename.cc
index 64f5fc4b..5b660ecd 100644
--- a/re2c/src/dfa/cfg/rename.cc
+++ b/re2c/src/dfa/cfg/rename.cc
@@ -35,6 +35,13 @@ void cfg_t::renaming(cfg_t &cfg, const tagver_t *ver2new, tagver_t maxver)
 			f = ver2new[f];
 		}
 	}
+
+	// versions of tags with history
+	std::set<tagver_t> newlvs, &oldlvs = cfg.dfa.listvers;
+	for (std::set<tagver_t>::iterator i = oldlvs.begin(); i != oldlvs.end(); ++i) {
+		newlvs.insert(ver2new[*i]);
+	}
+	oldlvs.swap(newlvs);
 }
 
 } // namespace re2c
diff --git a/re2c/src/dfa/determinization.cc b/re2c/src/dfa/determinization.cc
index 70a1f83a..6efbf292 100644
--- a/re2c/src/dfa/determinization.cc
+++ b/re2c/src/dfa/determinization.cc
@@ -56,6 +56,7 @@ dfa_t::dfa_t(const nfa_t &nfa, const opt_t *opts,
 	, charset(nfa.charset)
 	, rules(nfa.rules)
 	, tags(nfa.tags)
+	, listvers(*new std::set<tagver_t>)
 	, finvers(NULL)
 	, tcpool(*new tcpool_t)
 	, maxtagver(0)
@@ -84,6 +85,15 @@ dfa_t::dfa_t(const nfa_t &nfa, const opt_t *opts,
 		finvers[i] = fixed(tags[i]) ? TAGVER_ZERO : ++maxtagver;
 	}
 
+	// mark tags with history (initial and final)
+	for (size_t i = 0; i < ntag; ++i) {
+		if (history(tags[i])) {
+			tagver_t v = static_cast<tagver_t>(i) + 1, f = finvers[i];
+			if (f != TAGVER_ZERO) listvers.insert(f);
+			listvers.insert(v);
+		}
+	}
+
 	// iterate while new kernels are added: for each alphabet symbol,
 	// build tagged epsilon-closure of all reachable NFA states,
 	// then find identical or mappable DFA state or add a new one
@@ -100,6 +110,10 @@ dfa_t::dfa_t(const nfa_t &nfa, const opt_t *opts,
 			acts = closure(clos1, clos2, tagpool, tcpool, rules, maxtagver, newvers, dump.shadow, tags);
 			find_state(*this, i, c, kernels, clos2, acts, dump);
 		}
+		// mark tags with history
+		for (newvers_t::iterator j = newvers.begin(); j != newvers.end(); ++j) {
+			if (history(tags[j->first.tag])) listvers.insert(abs(j->second));
+		}
 	}
 
 	if (!opts->posix_captures) {
diff --git a/re2c/src/dfa/dfa.h b/re2c/src/dfa/dfa.h
index 85db8d6c..78e04f72 100644
--- a/re2c/src/dfa/dfa.h
+++ b/re2c/src/dfa/dfa.h
@@ -54,6 +54,7 @@ struct dfa_t
 	std::vector<uint32_t> &charset;
 	std::valarray<Rule> &rules;
 	std::vector<Tag> &tags;
+	std::set<tagver_t> &listvers;
 	tagver_t *finvers;
 	tcpool_t &tcpool;
 	tagver_t maxtagver;