[M5+] generalize nix_cmake_scan config picker

src/resolver/nix_cmake_scan.cpp

@@ -123,41 +123,82 @@ auto scan_imported_targets(std::string_view config_text) -> std::vector<std::str
 
 namespace {
 
-// Score a stem against the queried package name. Lower is better.
+// Normalize a name for cross-shape comparison:
-//   0 — exact match (case-insensitive)
+//   1. lowercase
-//   1 — prefix match (one is a case-insensitive prefix of the other)
+//   2. drop a leading "lib" if the remainder is at least 3 chars long
-//   2 — fallback (any other non-empty target list)
+//      (handles `libllvm` → `llvm`, `libclang` → `clang`)
-auto match_score(std::string_view stem, std::string_view pkg) -> int {
+//   3. drop a trailing version-ish suffix `[-_]?[0-9][0-9._-]*`
-    auto eq_ci = [](std::string_view a, std::string_view b) {
+//      (handles `Boost-1.89.0` → `boost`, `boost_atomic-1.89.0`
-        if (a.size() != b.size()) {
+//      → `boost_atomic`); does nothing when no suffix is present
-            return false;
+//   4. drop every remaining non-alphanumeric character so spelling
-        }
+//      variants like `gRPC`/`grpc` and `nlohmann_json`/`nlohmann-json`
-        for (std::size_t i = 0; i < a.size(); ++i) {
+//      collapse together
-            auto al = std::tolower(static_cast<unsigned char>(a[i]));
+auto normalize(std::string_view s) -> std::string {
-            auto bl = std::tolower(static_cast<unsigned char>(b[i]));
+    std::string out;
-            if (al != bl) {
+    out.reserve(s.size());
-                return false;
+    for (char c : s) {
-            }
+        out += static_cast<char>(std::tolower(static_cast<unsigned char>(c)));
-        }
+    }
-        return true;
+    if (out.size() > 3 && out.starts_with("lib")) {
+        out.erase(0, 3);
+    }
+    // Trim trailing version suffix. Walk back over [0-9._-] then peel
+    // a single optional separator if there's a digit run preceding it.
+    auto is_vchar = [](char c) {
+        return std::isdigit(static_cast<unsigned char>(c)) || c == '.'
+               || c == '-' || c == '_';
     };
-    if (eq_ci(stem, pkg)) {
+    std::size_t end = out.size();
+    while (end > 0 && is_vchar(out[end - 1])) {
+        --end;
+    }
+    bool has_digit = false;
+    for (auto i = end; i < out.size(); ++i) {
+        if (std::isdigit(static_cast<unsigned char>(out[i]))) {
+            has_digit = true;
+            break;
+        }
+    }
+    if (has_digit) {
+        out.erase(end);
+        if (!out.empty() && (out.back() == '-' || out.back() == '_')) {
+            out.pop_back();
+        }
+    }
+    std::string compact;
+    compact.reserve(out.size());
+    for (char c : out) {
+        if (std::isalnum(static_cast<unsigned char>(c))) {
+            compact += c;
+        }
+    }
+    return compact;
+}
+
+// Score a candidate against the queried package name. Lower is better.
+//   0 — parent_dir or stem == pkg after normalization
+//   1 — parent_dir or stem and pkg share a prefix in either direction
+//       (covers boost_<component> losing to a top-level Boost dir,
+//       and glfw matching glfw3)
+//   2 — fallback: candidate has IMPORTED targets but no name overlap
+auto match_score(std::string_view parent_dir, std::string_view stem,
+                 std::string_view pkg) -> int {
+    auto p = normalize(parent_dir);
+    auto s = normalize(stem);
+    auto q = normalize(pkg);
+    if (q.empty()) {
+        return 2;
+    }
+    if (p == q || s == q) {
         return 0;
     }
-    auto starts_with_ci = [](std::string_view longer, std::string_view shorter) {
+    auto prefix_either = [](const std::string& a, const std::string& b) {
-        if (shorter.size() > longer.size()) {
+        if (a.empty() || b.empty()) {
             return false;
         }
-        for (std::size_t i = 0; i < shorter.size(); ++i) {
+        return a.starts_with(b) || b.starts_with(a);
-            auto a = std::tolower(static_cast<unsigned char>(longer[i]));
-            auto b = std::tolower(static_cast<unsigned char>(shorter[i]));
-            if (a != b) {
-                return false;
-            }
-        }
-        return true;
     };
-    if (starts_with_ci(stem, pkg) || starts_with_ci(pkg, stem)) {
+    if (prefix_either(p, q) || prefix_either(s, q)) {
         return 1;
     }
     return 2;
@@ -187,6 +228,8 @@ auto nix_cmake_scan(const fs::path& store_path, const std::string& package_name)
 
     struct Match {
         int score;
+        std::string stem;        // for tiebreak: prefer shorter stems
+        std::string parent_dir;  // for tiebreak: then shorter parent dir
         NixCmakeCandidate cand;
     };
     std::vector<Match> matches;
@@ -226,13 +269,16 @@ auto nix_cmake_scan(const fs::path& store_path, const std::string& package_name)
         }
 
         auto stem = config_stem_to_package(entry.path().filename().string());
+        auto parent_dir_name = pkg_dir.filename().string();
         NixCmakeCandidate cand{
             .find_package = std::format("{} CONFIG REQUIRED", stem),
             .targets = std::move(targets),
             .config_file = entry.path(),
         };
         matches.push_back(Match{
-            .score = match_score(stem, package_name),
+            .score = match_score(parent_dir_name, stem, package_name),
+            .stem = stem,
+            .parent_dir = std::move(parent_dir_name),
             .cand = std::move(cand),
         });
     }
@@ -245,7 +291,25 @@ auto nix_cmake_scan(const fs::path& store_path, const std::string& package_name)
         });
     }
 
-    std::ranges::stable_sort(matches, {}, &Match::score);
+    // Deterministic tiebreak within the same score band: prefer the
+    // candidate with the shorter normalized stem (the "more canonical"
+    // name for find_package), then the shorter parent dir name, then
+    // alphabetical order for full reproducibility.
+    std::ranges::stable_sort(matches, [](const Match& a, const Match& b) {
+        if (a.score != b.score) {
+            return a.score < b.score;
+        }
+        if (a.stem.size() != b.stem.size()) {
+            return a.stem.size() < b.stem.size();
+        }
+        if (a.parent_dir.size() != b.parent_dir.size()) {
+            return a.parent_dir.size() < b.parent_dir.size();
+        }
+        if (a.parent_dir != b.parent_dir) {
+            return a.parent_dir < b.parent_dir;
+        }
+        return a.stem < b.stem;
+    });
     return std::move(matches.front().cand);
 }
 

tests/nix_cmake_scan_parse.cpp

@@ -136,6 +136,69 @@ TEST_CASE("nix_cmake_scan ignores Config files with no IMPORTED targets",
     REQUIRE(r.error().code == ErrorCode::ResolutionUnknownPackage);
 }
 
+TEST_CASE("nix_cmake_scan strips a leading lib prefix when scoring",
+          "[resolver][nix_cmake_scan]") {
+    // Real layout for nixpkgs `libllvm.dev`. `libllvm` should match
+    // `LLVMConfig.cmake`'s parent dir (`llvm`) after lib-stripping
+    // and beat `polly/PollyConfig.cmake`.
+    auto store = fresh_store();
+    touch_config(store, "llvm/LLVMConfig.cmake",
+                 R"(add_library(LLVM SHARED IMPORTED))");
+    touch_config(store, "polly/PollyConfig.cmake",
+                 R"(add_library(Polly SHARED IMPORTED))");
+
+    auto r = nix_cmake_scan(store, "libllvm");
+    REQUIRE(r.has_value());
+    REQUIRE(r->find_package == "LLVM CONFIG REQUIRED");
+    REQUIRE(r->config_file.filename() == "LLVMConfig.cmake");
+}
+
+TEST_CASE("nix_cmake_scan picks the canonical config from a versioned dir",
+          "[resolver][nix_cmake_scan]") {
+    // Boost ships a top-level `Boost-1.89.0/BoostConfig.cmake` plus
+    // 47 modular component configs at sibling versioned dirs.
+    // After version-suffix stripping the top-level Boost should win.
+    auto store = fresh_store();
+    touch_config(store, "Boost-1.89.0/BoostConfig.cmake",
+                 R"(add_library(Boost::headers INTERFACE IMPORTED))");
+    touch_config(store, "boost_atomic-1.89.0/boost_atomic-config.cmake",
+                 R"(add_library(Boost::atomic SHARED IMPORTED))");
+    touch_config(store, "boost_filesystem-1.89.0/boost_filesystem-config.cmake",
+                 R"(add_library(Boost::filesystem SHARED IMPORTED))");
+
+    auto r = nix_cmake_scan(store, "boost");
+    REQUIRE(r.has_value());
+    REQUIRE(r->find_package == "Boost CONFIG REQUIRED");
+    REQUIRE(r->config_file.filename() == "BoostConfig.cmake");
+}
+
+TEST_CASE("nix_cmake_scan ignores satellite configs in multi-config stores",
+          "[resolver][nix_cmake_scan]") {
+    // Mirrors `protobuf` which ships protobuf-config.cmake plus an
+    // unrelated utf8_range-config.cmake under the same prefix tree.
+    auto store = fresh_store();
+    touch_config(store, "protobuf/protobuf-config.cmake",
+                 R"(add_library(protobuf::libprotobuf SHARED IMPORTED))");
+    touch_config(store, "utf8_range/utf8_range-config.cmake",
+                 R"(add_library(utf8_range::utf8_range SHARED IMPORTED))");
+
+    auto r = nix_cmake_scan(store, "protobuf");
+    REQUIRE(r.has_value());
+    REQUIRE(r->find_package == "protobuf CONFIG REQUIRED");
+}
+
+TEST_CASE("nix_cmake_scan normalizes case for spelling variants",
+          "[resolver][nix_cmake_scan]") {
+    // grpc ships `gRPCConfig.cmake` under `lib/cmake/grpc/`.
+    auto store = fresh_store();
+    touch_config(store, "grpc/gRPCConfig.cmake",
+                 R"(add_library(gRPC::grpc++ SHARED IMPORTED))");
+
+    auto r = nix_cmake_scan(store, "grpc");
+    REQUIRE(r.has_value());
+    REQUIRE(r->find_package == "gRPC CONFIG REQUIRED");
+}
+
 TEST_CASE("nix_cmake_scan scans sibling *-targets.cmake files",
           "[resolver][nix_cmake_scan]") {
     // Mirrors fmt's real on-disk layout: the *-config.cmake is empty of