🚑 Allow tags after "?" for unknowns, fix #48

divvun · Dec 7, 2021 · 2d4086b · 2d4086b
1 parent e1ee7b2
commit 2d4086b
Show file tree

Hide file tree

Showing 4 changed files with 24 additions and 7 deletions.
diff --git a/src/cgspell.cpp b/src/cgspell.cpp
@@ -20,7 +20,7 @@
 namespace divvun {
 
 static const string subreading_separator = "#";
-static const string unknown_analysis = " ?";
+static const string tag_unknown = "?";
 
 /**
  * Return the size in bytes of the first complete UTF-8 codepoint in c,
@@ -230,11 +230,18 @@ void run_cgspell(std::istream& is,
 		}
 		else if (!result.empty() && result[5].length() != 0)
 		{
-			c.unknown = (result[5] == unknown_analysis);
-			if(c.unknown) {
-				sent.n_unknowns += 1;
-			}
-			c.lines.push_back(line);
+		  std::stringstream ana(result[5]);
+		  std::string tag;
+                  c.unknown = false;
+                  while (ana >> tag) {
+			  if(tag == tag_unknown) {
+				  c.unknown = true;
+			  }
+		  }
+		  if (c.unknown) {
+			  sent.n_unknowns += 1;
+		  }
+		  c.lines.push_back(line);
 		}
 		else if(!result.empty() && result[7].length() != 0) {
 			// TODO: Can we ever get a flush in the middle of readings?

diff --git a/src/util.hpp b/src/util.hpp
@@ -41,7 +41,7 @@ namespace divvun {
 
 const std::basic_regex<char> CG_LINE ("^"
 				      "(\"<(.*)>\".*" // wordform, group 2
-				      "|(\t+)(\"[^\"]*\"\\S*)(\\s+\\S+)*\\s*" // reading, group 3, 4, 5
+				      "|(\t+)(\"[^\"]*\"\\S*)((?:\\s+\\S+)*)\\s*" // reading, group 3, 4, 5
 				      "|:(.*)" // blank, group 6
 				      "|(<STREAMCMD:FLUSH>)" // flush, group 7
 				      "|(;\t+.*)" // traced reading, group 8

diff --git a/test/cgspell/expected.default b/test/cgspell/expected.default
@@ -13,3 +13,9 @@
 "<ballat>"
 	"ballat" V TV Inf
 	"ballat" V TV Ind Prs Pl1
+: 
+"<skuvlabufse>"
+	"skuvllabufse" ? <LastCohort> <NotFirstCohort>
+	"busse" N Sg <W:1> <WA:0> <spelled> "skuvlabusse"S
+		"skuvla" N Sg
+: 
diff --git a/test/cgspell/input.default b/test/cgspell/input.default
@@ -7,3 +7,7 @@
 "<ballat>"
 	"ballat" V TV Inf
 	"ballat" V TV Ind Prs Pl1
+: 
+"<skuvlabufse>"
+	"skuvllabufse" ? <LastCohort> <NotFirstCohort>
+: