From 33c708225110a2b88de0a167f254138ff47408a2 Mon Sep 17 00:00:00 2001
From: Geir Storli <geirst@yahooinc.com>
Date: Fri, 18 Mar 2022 15:50:08 +0000
Subject: Add function to lowercase an utf8 string while converting it to ucs4
 codepoints.

---
 vespalib/src/tests/text/lowercase/lowercase_test.cpp | 19 ++++++++++++-------
 vespalib/src/vespa/vespalib/text/lowercase.cpp       | 12 ++++++++++++
 vespalib/src/vespa/vespalib/text/lowercase.h         | 10 ++++++++++
 3 files changed, 34 insertions(+), 7 deletions(-)

diff --git a/vespalib/src/tests/text/lowercase/lowercase_test.cpp b/vespalib/src/tests/text/lowercase/lowercase_test.cpp
index 8117dc20666..ffc6dce427c 100644
--- a/vespalib/src/tests/text/lowercase/lowercase_test.cpp
+++ b/vespalib/src/tests/text/lowercase/lowercase_test.cpp
@@ -9,13 +9,8 @@ LOG_SETUP("lowercase_test");
 
 using namespace vespalib;
 
-TEST_SETUP(Test);
-
-int
-Test::Main()
+TEST("test basic lowercase")
 {
-    TEST_INIT("lowercase_test");
-
     EXPECT_EQUAL('a', LowerCase::convert('A'));
     EXPECT_EQUAL((int8_t)'a', LowerCase::convert((int8_t)'A'));
     EXPECT_EQUAL((uint8_t)'a', LowerCase::convert((uint8_t)'A'));
@@ -34,5 +29,15 @@ Test::Main()
         }
         // printf("lowercase( %d )= %d\n", hi, lo);
     }
-    TEST_DONE();
 }
+
+TEST("lowercase utf8 string to ucs4")
+{
+    auto res = LowerCase::convert_to_ucs4(std::string_view("ABC"));
+    EXPECT_EQUAL(3u, res.size());
+    EXPECT_EQUAL((uint32_t)'a', res[0]);
+    EXPECT_EQUAL((uint32_t)'b', res[1]);
+    EXPECT_EQUAL((uint32_t)'c', res[2]);
+}
+
+TEST_MAIN() { TEST_RUN_ALL(); }
diff --git a/vespalib/src/vespa/vespalib/text/lowercase.cpp b/vespalib/src/vespa/vespalib/text/lowercase.cpp
index 604027d2687..de6c5956fcb 100644
--- a/vespalib/src/vespa/vespalib/text/lowercase.cpp
+++ b/vespalib/src/vespa/vespalib/text/lowercase.cpp
@@ -20,6 +20,18 @@ LowerCase::convert(vespalib::stringref input)
     return output;
 }
 
+std::vector<uint32_t>
+LowerCase::convert_to_ucs4(vespalib::stringref input)
+{
+    std::vector<uint32_t> result;
+    result.reserve(input.size());
+    Utf8Reader reader(input.data());
+    while (reader.hasMore()) {
+        result.emplace_back(convert(reader.getChar()));
+    }
+    return result;
+}
+
 
 /*
   NOTE: the tables below are generated from Java code to
diff --git a/vespalib/src/vespa/vespalib/text/lowercase.h b/vespalib/src/vespa/vespalib/text/lowercase.h
index e0ed742c3c5..dc081c6ba2d 100644
--- a/vespalib/src/vespa/vespalib/text/lowercase.h
+++ b/vespalib/src/vespa/vespalib/text/lowercase.h
@@ -5,6 +5,7 @@
 #pragma once
 
 #include <vespa/vespalib/stllike/string.h>
+#include <vector>
 
 namespace vespalib {
 
@@ -103,6 +104,15 @@ public:
      * CHARACTER (U+FFFD).
      **/
     static vespalib::string convert(vespalib::stringref input);
+
+    /**
+     * Lowercase a string in UTF-8 format while converting it to UCS-4 codepoints.
+     */
+    static std::vector<uint32_t> convert_to_ucs4(vespalib::stringref input);
+    static std::vector<uint32_t> convert_to_ucs4(std::string_view input) {
+        return convert_to_ucs4(vespalib::stringref(input.data(), input.size()));
+    }
+
 };
 
 
-- 
cgit v1.2.3