✨ cgspell limit/beam/max-weight configurable from pipespec.xml

E.g. <cgspell limit="5" beam="20" max-weight="2000">
divvun · May 13, 2020 · a0533a7 · a0533a7
1 parent 2e7768d
commit a0533a7
Show file tree

Hide file tree

Showing 5 changed files with 23 additions and 8 deletions.
diff --git a/src/pipeline.cpp b/src/pipeline.cpp
@@ -87,7 +87,7 @@ void CGCmd::run(stringstream& input, stringstream& output) const
 }
 
 #ifdef HAVE_CGSPELL
-CGSpellCmd::CGSpellCmd (hfst_ospell::Transducer* errmodel, hfst_ospell::Transducer* acceptor, float max_sent_unknown_rate, bool verbose)
+CGSpellCmd::CGSpellCmd (hfst_ospell::Transducer* errmodel, hfst_ospell::Transducer* acceptor, int limit, float beam, float max_weight, float max_sent_unknown_rate, bool verbose)
 	: speller(new Speller(errmodel, acceptor, verbose, max_analysis_weight, max_weight, real_word, limit, beam, time_cutoff, max_sent_unknown_rate))
 {
 	if (!acceptor) {
@@ -97,7 +97,7 @@ CGSpellCmd::CGSpellCmd (hfst_ospell::Transducer* errmodel, hfst_ospell::Transduc
 		throw std::runtime_error("libdivvun: ERROR: CGSpell command couldn't read errmodel");
 	}
 }
-CGSpellCmd::CGSpellCmd (const string& err_path, const string& lex_path, float max_sent_unknown_rate, bool verbose)
+CGSpellCmd::CGSpellCmd (const string& err_path, const string& lex_path, int limit, float beam, float max_weight, float max_sent_unknown_rate, bool verbose)
 	: speller(new Speller(err_path, lex_path, verbose, max_analysis_weight, max_weight, real_word, limit, beam, time_cutoff, max_sent_unknown_rate))
 {
 }
@@ -211,6 +211,9 @@ Pipeline Pipeline::mkPipeline(const unique_ptr<ArPipeSpec>& ar_spec, const u16st
 			};
 			auto* s = new CGSpellCmd(readArchiveExtract(ar_spec->ar_path, args["errmodel"], f),
 						 readArchiveExtract(ar_spec->ar_path, args["lexicon"], f),
+						 cmd.attribute("limit").as_int(10),
+						 cmd.attribute("beam").as_float(15.0),
+						 cmd.attribute("max-weight").as_float(5000.0),
 						 cmd.attribute("max-unknown-rate").as_float(0.4),
 						 verbose);
 			cmds.emplace_back(s);
@@ -292,6 +295,9 @@ Pipeline Pipeline::mkPipeline(const unique_ptr<PipeSpec>& spec, const u16string&
 #ifdef HAVE_CGSPELL
 			cmds.emplace_back(new CGSpellCmd(args["errmodel"],
 							 args["lexicon"],
+							 cmd.attribute("limit").as_int(10),
+							 cmd.attribute("beam").as_float(15.0),
+							 cmd.attribute("max-weight").as_float(5000.0),
 							 cmd.attribute("max-unknown-rate").as_float(0.4),
 							 verbose));
 #else

diff --git a/src/pipeline.hpp b/src/pipeline.hpp
@@ -167,17 +167,14 @@ class CGCmd: public PipeCmd {
 #ifdef HAVE_CGSPELL
 class CGSpellCmd: public PipeCmd {
 	public:
-		CGSpellCmd (hfst_ospell::Transducer* errmodel, hfst_ospell::Transducer* acceptor, float max_sent_unknown_rate, bool verbose);
-		CGSpellCmd (const string& err_path, const string& lex_path, float max_sent_unknown_rate, bool verbose);
+		CGSpellCmd (hfst_ospell::Transducer* errmodel, hfst_ospell::Transducer* acceptor, int limit, float beam, float max_weight, float max_sent_unknown_rate, bool verbose);
+		CGSpellCmd (const string& err_path,            const string& lex_path,            int limit, float beam, float max_weight, float max_sent_unknown_rate, bool verbose);
 		void run(stringstream& input, stringstream& output) const override;
 		~CGSpellCmd() override = default;
 		// Some sane defaults for the speller
 		// TODO: Do we want any of this configurable from pipespec.xml, or from the Checker API?
 		static constexpr Weight max_analysis_weight = -1.0;
-		static constexpr Weight max_weight = 5000.0;
 		static constexpr bool real_word = false;
-		static constexpr unsigned long limit = 10;
-		static constexpr hfst_ospell::Weight beam = 15.0;
 		static constexpr float time_cutoff = 0.0;
 	private:
 		unique_ptr<Speller> speller;

diff --git a/src/pipespec.cpp b/src/pipespec.cpp
@@ -198,8 +198,14 @@ vector<std::pair<string,string>> toPipeSpecShVector(const string& dir, const Pip
 			prog += " -g" + argprepare(dir, args["grammar"]);
 		}
 		else if(name == "cgspell") {
+			int limit = cmd.attribute("limit").as_int(10);
+			float beam = cmd.attribute("beam").as_float(15);
+			float max_weight = cmd.attribute("max-weight").as_float(5000);
 			float max_sent_unknown_rate = cmd.attribute("max-unknown-rate").as_float(0.4);
-			prog = "divvun-cgspell -n 10 -b 15 -w 5000";
+			prog = "divvun-cgspell";
+			prog += " -n " + std::to_string(limit);
+			prog += " -b " + std::to_string(beam);
+			prog += " -w " + std::to_string(max_weight);
 			prog += " -u " + std::to_string(max_sent_unknown_rate);
 			prog += " -l" + argprepare(dir, args["lexicon"]);
 			prog += " -m" +  argprepare(dir, args["errmodel"]);

diff --git a/src/pipespec.dtd b/src/pipespec.dtd
@@ -86,6 +86,9 @@
 <!ELEMENT cg (grammar)>           <!-- arg: grammar.cg3 -->
 <!ELEMENT cgspell ((lexicon, errmodel)|(errmodel, lexicon))> <!-- arg1: acceptor.hfstol, arg2: errmodel.hfst -->
 <!ATTLIST cgspell
+          limit CDATA "10"
+          beam CDATA "15.0"
+          max-weight CDATA "5000.0"
           max-unknown-rate CDATA "0.4">
 <!ELEMENT tokenize (tokenizer)>     <!-- arg: tokeniser.pmhfst -->
 <!ELEMENT tokenise (tokenizer)>     <!-- en_GB alias of the above -->

diff --git a/src/pipespec.rnc b/src/pipespec.rnc
@@ -102,6 +102,9 @@ cgspell =
 # arg1: acceptor.hfstol, arg2: errmodel.hfst
 attlist.cgspell &=
 
+  [ a:defaultValue = "10" ] attribute limit { text }?,
+  [ a:defaultValue = "15.0" ] attribute beam { text }?,
+  [ a:defaultValue = "5000.0" ] attribute max-weight { text }?,
   [ a:defaultValue = "0.4" ] attribute max-unknown-rate { text }?
 tokenize = element tokenize { attlist.tokenize, tokenizer }
 # arg: tokeniser.pmhfst