feat(strings): add canonical locales.json (RFC 0003)

CMakeLists.txt

@@ -259,6 +259,7 @@ if(BUILD_CAPI)
         dasher_add_test(dasher_deterministic_tests test_deterministic.cpp)
         dasher_add_test(dasher_training_tests test_training.cpp)
         dasher_add_test(dasher_node_tree_tests test_node_tree.cpp)
+        dasher_add_test(dasher_locales_sync_tests test_locales_sync.cpp)
 
         # Phase B (characterization) — closes coverage gaps noted in the review.
         #   - buffer_lifetime: the C API "valid until next call" contract

Strings/locales.json

@@ -0,0 +1,170 @@
+{
+  "_comment": "Canonical list of locales DasherCore supports. Mirrors the strings_*.json files in this directory — the single source of truth for every frontend's locale picker (RFC 0003). Display ordering is the frontend's concern. Regenerate with the table in RFC 0003; keep in sync with strings_*.json.",
+  "locales": [
+    {
+      "code": "af",
+      "endonym": "Afrikaans",
+      "rtl": false
+    },
+    {
+      "code": "ar",
+      "endonym": "العربية",
+      "rtl": true
+    },
+    {
+      "code": "bn",
+      "endonym": "বাংলা",
+      "rtl": false
+    },
+    {
+      "code": "cs",
+      "endonym": "Čeština",
+      "rtl": false
+    },
+    {
+      "code": "da",
+      "endonym": "Dansk",
+      "rtl": false
+    },
+    {
+      "code": "de",
+      "endonym": "Deutsch",
+      "rtl": false
+    },
+    {
+      "code": "el",
+      "endonym": "Ελληνικά",
+      "rtl": false
+    },
+    {
+      "code": "en",
+      "endonym": "English",
+      "rtl": false
+    },
+    {
+      "code": "es",
+      "endonym": "Español",
+      "rtl": false
+    },
+    {
+      "code": "fa",
+      "endonym": "فارسی",
+      "rtl": true
+    },
+    {
+      "code": "fi",
+      "endonym": "Suomi",
+      "rtl": false
+    },
+    {
+      "code": "fr",
+      "endonym": "Français",
+      "rtl": false
+    },
+    {
+      "code": "gu",
+      "endonym": "ગુજરાતી",
+      "rtl": false
+    },
+    {
+      "code": "hi",
+      "endonym": "हिन्दी",
+      "rtl": false
+    },
+    {
+      "code": "hu",
+      "endonym": "Magyar",
+      "rtl": false
+    },
+    {
+      "code": "it",
+      "endonym": "Italiano",
+      "rtl": false
+    },
+    {
+      "code": "kn",
+      "endonym": "ಕನ್ನಡ",
+      "rtl": false
+    },
+    {
+      "code": "ml",
+      "endonym": "മലയാളം",
+      "rtl": false
+    },
+    {
+      "code": "mr",
+      "endonym": "मराठी",
+      "rtl": false
+    },
+    {
+      "code": "nl",
+      "endonym": "Nederlands",
+      "rtl": false
+    },
+    {
+      "code": "pa",
+      "endonym": "ਪੰਜਾਬੀ",
+      "rtl": false
+    },
+    {
+      "code": "pl",
+      "endonym": "Polski",
+      "rtl": false
+    },
+    {
+      "code": "pt",
+      "endonym": "Português (Brasil)",
+      "rtl": false
+    },
+    {
+      "code": "pt-PT",
+      "endonym": "Português (Portugal)",
+      "rtl": false
+    },
+    {
+      "code": "ru",
+      "endonym": "Русский",
+      "rtl": false
+    },
+    {
+      "code": "sv",
+      "endonym": "Svenska",
+      "rtl": false
+    },
+    {
+      "code": "sw",
+      "endonym": "Kiswahili",
+      "rtl": false
+    },
+    {
+      "code": "ta",
+      "endonym": "தமிழ்",
+      "rtl": false
+    },
+    {
+      "code": "te",
+      "endonym": "తెలుగు",
+      "rtl": false
+    },
+    {
+      "code": "th",
+      "endonym": "ไทย",
+      "rtl": false
+    },
+    {
+      "code": "ur",
+      "endonym": "اردو",
+      "rtl": true
+    },
+    {
+      "code": "zh-CN",
+      "endonym": "中文（简体）",
+      "rtl": false
+    },
+    {
+      "code": "zu",
+      "endonym": "isiZulu",
+      "rtl": false
+    }
+  ]
+}

tests/test_locales_sync.cpp

@@ -0,0 +1,83 @@
+// test_locales_sync.cpp — guard that Strings/locales.json (the canonical locale
+// list, RFC 0003) stays in lock-step with the strings_*.json files actually
+// shipped. Prevents silent drift: add a strings_xx.json but forget locales.json
+// (or vice versa) and this test fails.
+//
+// Self-contained: no JSON dependency. Codes are pulled from the
+// strings_<code>.json filenames and from a focused scan of locales.json (a
+// generated, controlled file whose "code" keys are the only `"code":` matches).
+
+#include "test_common.h"
+
+#include <algorithm>
+#include <filesystem>
+#include <fstream>
+#include <iterator>
+#include <set>
+#include <sstream>
+#include <string>
+
+namespace fs = std::filesystem;
+
+static fs::path strings_dir() {
+    return fs::path(get_test_data_dir()) / "Strings";
+}
+
+// Codes implied by the shipped strings_<code>.json files.
+static std::set<std::string> shipped_codes() {
+    std::set<std::string> codes;
+    for (const auto& entry : fs::directory_iterator(strings_dir())) {
+        const std::string name = entry.path().filename().string();
+        const std::string pfx = "strings_";
+        const std::string ext = ".json";
+        if (name.rfind(pfx, 0) == 0 && name.size() > pfx.size() + ext.size() &&
+            name.compare(name.size() - ext.size(), ext.size(), ext) == 0) {
+            codes.insert(name.substr(pfx.size(), name.size() - pfx.size() - ext.size()));
+        }
+    }
+    return codes;
+}
+
+// Codes listed in locales.json, by scanning for `"code": "<value>"`.
+static std::set<std::string> listed_codes() {
+    const fs::path file = strings_dir() / "locales.json";
+    std::ifstream in(file);
+    INFO("could not open " << file.string());
+    REQUIRE(in.good());
+    std::stringstream ss;
+    ss << in.rdbuf();
+    const std::string text = ss.str();
+    const std::string needle = "\"code\":";
+    std::set<std::string> codes;
+    size_t pos = 0;
+    while ((pos = text.find(needle, pos)) != std::string::npos) {
+        pos += needle.size();
+        const size_t q1 = text.find('"', pos);
+        if (q1 == std::string::npos) break;
+        const size_t q2 = text.find('"', q1 + 1);
+        if (q2 == std::string::npos) break;
+        codes.insert(text.substr(q1 + 1, q2 - q1 - 1));
+        pos = q2 + 1;
+    }
+    return codes;
+}
+
+TEST_CASE("locales.json matches the shipped strings_*.json files") {
+    const std::set<std::string> shipped = shipped_codes();
+    const std::set<std::string> listed = listed_codes();
+
+    REQUIRE_FALSE(shipped.empty());
+    MESSAGE("shipped=" << shipped.size() << " listed=" << listed.size());
+
+    std::set<std::string> missing; // shipped but not listed
+    std::set<std::string> extra;   // listed but not shipped
+    std::set_difference(shipped.begin(), shipped.end(), listed.begin(), listed.end(),
+                        std::inserter(missing, missing.begin()));
+    std::set_difference(listed.begin(), listed.end(), shipped.begin(), shipped.end(),
+                        std::inserter(extra, extra.begin()));
+
+    INFO("locales.json missing codes that have a strings_*.json file (missing=" << missing.size() << ")");
+    REQUIRE(missing.empty());
+    INFO("locales.json lists codes with no strings_*.json file (extra=" << extra.size() << ")");
+    REQUIRE(extra.empty());
+}