Handle empty Apertium cohort (fixes #149, fixes #151)

GrammarSoft · Sep 19, 2024 · 131d4c0 · 131d4c0
1 parent 0777679
commit 131d4c0
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 15 deletions.
diff --git a/src/ApertiumApplicator.cpp b/src/ApertiumApplicator.cpp
@@ -398,14 +398,15 @@ void ApertiumApplicator::runGrammarOnText(std::istream& input, std::ostream& out
 
 			blank.append(u"\"<");
 			UChar* p = &token[1];
-			for (; *p && *p != '/' && *p != '<'; ++p) {
+			for (; *p && *p != '/' && *p != '<' && *p != '$'; ++p) {
 				if (*p == '\\') {
 					++p;
 				}
 				blank += *p;
 			}
 			blank.append(u">\"");
 			cCohort->wordform = addTag(blank);
+			blank.clear();
 
 			// Handle the static reading of ^estació<n><f><sg>/season<n><sg>/station<n><sg>$
 			// Gobble up all <tags> until the first / or $ and stuff them in the static reading
@@ -807,10 +808,6 @@ void ApertiumApplicator::printReading(Reading* reading, std::ostream& output, Ap
 }
 
 void ApertiumApplicator::printReading(Reading* reading, std::ostream& output) {
-	if (reading->noprint) {
-		return;
-	}
-
 	size_t firstlower = 0;
 	ApertiumCasing casing = ApertiumCasing::Lower;
 
@@ -925,6 +922,9 @@ void ApertiumApplicator::printCohort(Cohort* cohort, std::ostream& output, bool
 	//Tag::printTagRaw(output, grammar->single_tags[cohort->wordform]);
 	std::sort(cohort->readings.begin(), cohort->readings.end(), Reading::cmp_number);
 	for (auto reading : cohort->readings) {
+		if (reading->noprint) {
+			continue;
+		}
 		if (need_slash) {
 			u_fprintf(output, "/");
 		}
@@ -941,6 +941,9 @@ void ApertiumApplicator::printCohort(Cohort* cohort, std::ostream& output, bool
 	if (trace) {
 		std::sort(cohort->delayed.begin(), cohort->delayed.end(), Reading::cmp_number);
 		for (auto reading : cohort->delayed) {
+			if (reading->noprint) {
+				continue;
+			}
 			if (need_slash) {
 				u_fprintf(output, "/%C", not_sign);
 			}
@@ -952,6 +955,9 @@ void ApertiumApplicator::printCohort(Cohort* cohort, std::ostream& output, bool
 		}
 		std::sort(cohort->deleted.begin(), cohort->deleted.end(), Reading::cmp_number);
 		for (auto reading : cohort->deleted) {
+			if (reading->noprint) {
+				continue;
+			}
 			if (need_slash) {
 				u_fprintf(output, "/%C", not_sign);
 			}

diff --git a/src/GrammarApplicator_reflow.cpp b/src/GrammarApplicator_reflow.cpp
@@ -33,7 +33,7 @@ Tag* GrammarApplicator::makeBaseFromWord(uint32_t tag) {
 
 Tag* GrammarApplicator::makeBaseFromWord(Tag* tag) {
 	const size_t len = tag->tag.size();
-	if (len < 5) {
+	if (len < 4) {
 		return tag;
 	}
 	static thread_local UString n;

diff --git a/src/version.hpp b/src/version.hpp
@@ -27,7 +27,7 @@ constexpr auto CG3_COPYRIGHT_STRING = "Copyright (C) 2007-2024 GrammarSoft ApS.
 
 constexpr uint32_t CG3_VERSION_MAJOR = 1;
 constexpr uint32_t CG3_VERSION_MINOR = 4;
-constexpr uint32_t CG3_VERSION_PATCH = 17;
+constexpr uint32_t CG3_VERSION_PATCH = 18;
 constexpr uint32_t CG3_REVISION = 13898;
 constexpr uint32_t CG3_FEATURE_REV = 13898;
 constexpr uint32_t CG3_TOO_OLD = 10373;

diff --git a/test/Apertium/runall.pl b/test/Apertium/runall.pl
@@ -9,30 +9,32 @@
 
 # Search paths for the binary
 my @binlist = (
-	"../../build/VS16/src/Debug/vislcg3",
-	"../../build/VS16/src/Release/vislcg3",
+	"../build/src/vislcg3",
+	"../../build/VS17/src/Debug/vislcg3",
+	"../../build/VS17/src/Release/vislcg3",
+	"../src/Debug/vislcg3",
+	"../src/Release/vislcg3",
 	"../Debug/vislcg3",
-	"../Common/vislcg3",
 	"../Release/vislcg3",
 	"../src/vislcg3",
 	"../vislcg3",
-);
+	);
 my $binary = "vislcg3";
 
 foreach (@binlist) {
-	if (-x $_) {
+	if (-x $_ && int(`$_ --min-binary-revision 2>/dev/null`) >= 10373) {
 		$binary = $_;
 		last;
 	}
-	elsif (-x $_.".exe") {
+	elsif (-x $_.".exe" && int(`$_.exe --min-binary-revision 2>/dev/null`) >= 10373) {
 		$binary = $_.".exe";
 		last;
 	}
-	elsif (-x "../".$_) {
+	elsif (-x "../".$_ && int(`../$_ --min-binary-revision 2>/dev/null`) >= 10373) {
 		$binary = "../".$_;
 		last;
 	}
-	elsif (-x "../".$_.".exe") {
+	elsif (-x "../".$_.".exe" && int(`../$_.exe --min-binary-revision 2>/dev/null`) >= 10373) {
 		$binary = "../".$_.".exe";
 		last;
 	}