From 78aa3ee250ac94be8a92bedbc2a937cf95f64f96 Mon Sep 17 00:00:00 2001
From: Kevin Brubeck Unhammer <unhammer@fsfe.org>
Date: Thu, 19 Sep 2024 19:30:36 +0200
Subject: [PATCH] Put casing of &ADDED word on first non-added to the right
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Also changes withCasing to apply lowering where casing==lower – should
have no effect unless we're changing casing like this, since getCasing
only returns lower if there are no uppers.

Closes https://github.com/divvun/libdivvun/issues/77
---
 src/suggest.cpp                       | 30 ++++++++++++++++-----------
 src/suggest.hpp                       | 13 +++++++++---
 test/suggest/expected.move-after.json |  2 +-
 test/suggest/expected.move.json       |  2 +-
 4 files changed, 30 insertions(+), 17 deletions(-)

diff --git a/src/suggest.cpp b/src/suggest.cpp
index 9dae47c..174c733 100644
--- a/src/suggest.cpp
+++ b/src/suggest.cpp
@@ -576,22 +576,11 @@ if(verbose)	std::cerr << "\033[1;33mright=\t" << i_right << "\033[0m" << std::en
 	UStringVector reps = {u""};
 	UStringVector reps_suggestwf = {}; // If we're doing SUGGESTWF, we ignore reps
 	string prev_added_before_blank = "";
+	std::optional<Casing> addedcasing = std::nullopt;
 	for (size_t i = i_left; i <= i_right; ++i) {
 		const auto& trg = sentence.cohorts[i];
 		Casing casing = getCasing(toUtf8(trg.form));
 
-		// std::cerr << "\033[0;35mtrg.added=\t" << trg.added << " i=" << i << "i_left" << i_left<< "\033[0m" << std::endl;
-		if(trg.added) {
-			for(size_t j = i; j <= i_right; j++) {
-				const auto& right_of_trg = sentence.cohorts[j];
-				if(!right_of_trg.added) {
-					// std::cerr << "\033[1;35mright_of_added=\t" << toUtf8(right_of_trg.form) << " j=" << j << "\033[0m" << std::endl;
-					casing = getCasing(toUtf8(right_of_trg.form));
-					break;
-				}
-			}
-		}
-
 if(verbose)		std::cerr << "\033[1;34mi=\t" << i << "\033[0m" << std::endl;
 if(verbose)		std::cerr << "\033[1;34mtrg.form=\t'" << toUtf8(trg.form) << "'\033[0m" << std::endl;
 if(verbose)		std::cerr << "\033[1;34mtrg.id=\t" << trg.id << "\033[0m" << std::endl;
@@ -604,6 +593,23 @@ if(verbose)		std::cerr << "\033[1;35mtrg.raw_pre_blank=\t'" << trg.raw_pre_blank
 if(verbose)			std::cerr << "\t\t\033[1;36mdelete=\t" << toUtf8(trg.form) << "\033[0m" << std::endl;
 		}
 
+		if(trg.added) {
+			// This word was added, get casing from a non-added word to the right:
+			for(size_t j = i; j <= i_right; j++) {
+				const auto& right_of_trg = sentence.cohorts[j];
+				if(!right_of_trg.added) {
+					addedcasing = casing;
+					casing = getCasing(toUtf8(right_of_trg.form));
+					break;
+				}
+			}
+		}
+		else if(addedcasing.has_value() && !del) {
+			// This word was not &ADDED, but is preceded by an added word:
+			casing = addedcasing.value();
+			addedcasing = std::nullopt;
+		}
+
 		bool added_before_blank = false;
 		bool fixedcase = false;
 		bool applies_deletion = trg.id == src.id && src_applies_deletion;
diff --git a/src/suggest.hpp b/src/suggest.hpp
index e9778c4..b40ec56 100644
--- a/src/suggest.hpp
+++ b/src/suggest.hpp
@@ -46,6 +46,7 @@
 namespace divvun {
 
 using std::variant;
+using std::optional;
 using std::pair;
 using std::string;
 using std::stringstream;
@@ -144,10 +145,16 @@ inline std::string totitle(const string& input) {
 	std::transform(w.begin(), w.begin() + 1, w.begin(), std::towupper);
 	return wideToUtf8(w);
 }
+
+inline std::string tolower(const string& input) {
+	std::wstring w = wideFromUtf8(input);
+	setlocale(LC_ALL, "");
+	std::transform(w.begin(), w.begin() + 1, w.begin(), std::towlower);
+	return wideToUtf8(w);
+}
 // #endif
 
-inline std::string withCasing(
-  bool fixedcase, const Casing& inputCasing, const string& input) {
+inline std::string withCasing(bool fixedcase, const Casing& inputCasing, const string& input) {
 	if (fixedcase) {
 		return input;
 	}
@@ -159,7 +166,7 @@ inline std::string withCasing(
 	case mIxed:
 		return input;
 	case lower:
-		return input;
+		return tolower(input);
 	}
 	// should never get to this point
 	return input;
diff --git a/test/suggest/expected.move-after.json b/test/suggest/expected.move-after.json
index a252a82..dd13e0b 100644
--- a/test/suggest/expected.move-after.json
+++ b/test/suggest/expected.move-after.json
@@ -1 +1 @@
-{"errs":[["Nrel mellom nabs",4,20,"syn-abs-wordorder","syn-abs-wordorder",["Nabs Nrel mellom"],"syn-abs-wordorder"]],"text":"før Nrel mellom nabs"}
\ No newline at end of file
+{"errs":[["Nrel mellom nabs",4,20,"syn-abs-wordorder","syn-abs-wordorder",["Nabs nrel mellom"],"syn-abs-wordorder"]],"text":"før Nrel mellom nabs"}
\ No newline at end of file
diff --git a/test/suggest/expected.move.json b/test/suggest/expected.move.json
index a239e62..8a2714f 100644
--- a/test/suggest/expected.move.json
+++ b/test/suggest/expected.move.json
@@ -1 +1 @@
-{"errs":[["Nrel mellom nabs",0,16,"syn-abs-wordorder","syn-abs-wordorder",["Nabs Nrel mellom"],"syn-abs-wordorder"]],"text":"Nrel mellom nabs"}
\ No newline at end of file
+{"errs":[["Nrel mellom nabs",0,16,"syn-abs-wordorder","syn-abs-wordorder",["Nabs nrel mellom"],"syn-abs-wordorder"]],"text":"Nrel mellom nabs"}
\ No newline at end of file