Merge remote-tracking branch 'origin/31-pb-attribute-vocabulary' into staging

Author: steven-varga

src/consumer_bson.hpp

@@ -248,7 +248,7 @@ class BsonTemplateCallback : public clang::ast_matchers::MatchFinder::MatchCallb
 
 	static bool is_chrono_time_point_(const clang::QualType& qt) {
 		std::string name = qt.getAsString();
-		return name.find("std::chrono::time_point") == 0 || name.find("time_point<") != std::string::npos;
+		return name.find("std::chrono::") != std::string::npos && name.find("time_point") != std::string::npos;
 	}
 
 	// ── type name for emitted code ──

src/consumer_pb.hpp

@@ -40,16 +40,9 @@ inline bool is_h5_attr_name(llvm::StringRef name) {
         || name == "serialize_full";
 }
 
-inline bool is_rlp_attr_name(llvm::StringRef name) {
-    return name == "name"      || name == "ignore"      || name == "doc"
-        || name == "alias"     || name == "required"    || name == "timestamp";
-}
-
 inline bool is_json_attr_name(llvm::StringRef name) {
     return name == "name"      || name == "ignore"      || name == "doc"
-        || name == "alias"     || name == "version"     || name == "name_all"
-        || name == "required"  || name == "format"      || name == "pattern"
-        || name == "min"       || name == "max"         || name == "tool_format";
+        || name == "alias"     || name == "required"    || name == "type";
 }
 
 inline bool is_msgpack_attr_name(llvm::StringRef name) {
@@ -75,6 +68,11 @@ inline bool is_avro_attr_name(llvm::StringRef name) {
         || name == "uuid"      || name == "date"        || name == "time";
 }
 
+inline bool is_rlp_attr_name(llvm::StringRef name) {
+    return name == "name"      || name == "ignore"      || name == "doc"
+        || name == "alias"     || name == "required"    || name == "timestamp";
+}
+
 // Skip whitespace and comments at position `i` in `src`. Returns the new
 // position. Comments are emitted to `out` verbatim.
 inline std::size_t skip_ws(llvm::StringRef src, std::size_t i, std::string& out) {
@@ -173,58 +171,45 @@ inline std::vector<llvm::StringRef> split_attrs(llvm::StringRef block) {
     return out;
 }
 
-// Rewrite one attribute spec. If it starts with `h5::<recognized-name>`,
-// convert to `clang::annotate("h5::<name>", <args>)`. Otherwise leave verbatim.
+// Rewrite one attribute spec. If it starts with a recognized namespace,
+// convert to `clang::annotate("ns::<name>", <args>)`. Otherwise leave verbatim.
 inline std::string rewrite_one_attr(llvm::StringRef spec) {
     std::size_t start = 0;
     while (start < spec.size() && std::isspace(static_cast<unsigned char>(spec[start]))) ++start;
     llvm::StringRef leading = spec.substr(0, start);
     llvm::StringRef body    = spec.substr(start);
 
-    bool is_h5  = body.starts_with("h5::");
-    bool is_rlp = body.starts_with("rlp::");
-    if (!is_h5 && !is_rlp) return spec.str();
-    llvm::StringRef ns = is_h5 ? "h5::" : "rlp::";
-    bool is_h5  = body.starts_with("h5::");
-    bool is_json = body.starts_with("json::");
-    if (!is_h5 && !is_json) return spec.str();
-    llvm::StringRef ns = is_h5 ? "h5::" : "json::";
-    bool is_h5  = body.starts_with("h5::");
+    bool is_h5      = body.starts_with("h5::");
+    bool is_json    = body.starts_with("json::");
     bool is_msgpack = body.starts_with("msgpack::");
-    if (!is_h5 && !is_msgpack) return spec.str();
-    llvm::StringRef ns = is_h5 ? "h5::" : "msgpack::";
-    if (!body.starts_with(ns)) return spec.str();
-    bool is_h5  = body.starts_with("h5::");
-    bool is_cbor = body.starts_with("cbor::");
-    if (!is_h5 && !is_cbor) return spec.str();
-    llvm::StringRef ns = is_h5 ? "h5::" : "cbor::";
-    bool is_h5   = body.starts_with("h5::");
-    bool is_bson = body.starts_with("bson::");
-    if (!is_h5 && !is_bson) return spec.str();
-    llvm::StringRef ns = is_h5 ? "h5::" : "bson::";
-    bool is_h5   = body.starts_with("h5::");
-    bool is_avro = body.starts_with("avro::");
-    if (!is_h5 && !is_avro) return spec.str();
-    llvm::StringRef ns = is_h5 ? "h5::" : "avro::";
+    bool is_cbor    = body.starts_with("cbor::");
+    bool is_bson    = body.starts_with("bson::");
+    bool is_avro    = body.starts_with("avro::");
+    bool is_rlp     = body.starts_with("rlp::");
+    if (!is_h5 && !is_json && !is_msgpack && !is_cbor && !is_bson && !is_avro && !is_rlp) return spec.str();
+
+    llvm::StringRef ns;
+    if (is_h5)      ns = "h5::";
+    else if (is_json)    ns = "json::";
+    else if (is_msgpack) ns = "msgpack::";
+    else if (is_cbor)    ns = "cbor::";
+    else if (is_bson)    ns = "bson::";
+    else if (is_avro)    ns = "avro::";
+    else                 ns = "rlp::";
 
     body = body.drop_front(ns.size());
     std::size_t i = 0;
     while (i < body.size()
            && (std::isalnum(static_cast<unsigned char>(body[i])) || body[i] == '_')) ++i;
     if (i == 0) return spec.str();
     llvm::StringRef name = body.substr(0, i);
-    if (is_h5 && !is_h5_attr_name(name)) return spec.str();
-    if (is_rlp && !is_rlp_attr_name(name)) return spec.str();
-    if (is_h5 && !is_h5_attr_name(name)) return spec.str();
-    if (is_json && !is_json_attr_name(name)) return spec.str();
-    if (is_h5 && !is_h5_attr_name(name)) return spec.str();
+    if (is_h5      && !is_h5_attr_name(name))      return spec.str();
+    if (is_json    && !is_json_attr_name(name))    return spec.str();
     if (is_msgpack && !is_msgpack_attr_name(name)) return spec.str();
-    if (is_h5 && !is_h5_attr_name(name)) return spec.str();
-    if (is_cbor && !is_cbor_attr_name(name)) return spec.str();
-    if (is_h5 && !is_h5_attr_name(name)) return spec.str();
-    if (is_bson && !is_bson_attr_name(name)) return spec.str();
-    if (is_h5 && !is_h5_attr_name(name)) return spec.str();
-    if (is_avro && !is_avro_attr_name(name)) return spec.str();
+    if (is_cbor    && !is_cbor_attr_name(name))    return spec.str();
+    if (is_bson    && !is_bson_attr_name(name))    return spec.str();
+    if (is_avro    && !is_avro_attr_name(name))    return spec.str();
+    if (is_rlp     && !is_rlp_attr_name(name))     return spec.str();
 
     while (i < body.size() && std::isspace(static_cast<unsigned char>(body[i]))) ++i;
 
@@ -294,12 +279,9 @@ inline std::string rewrite(llvm::StringRef src) {
                 continue;
             }
             llvm::StringRef block = src.substr(i + 2, close - i - 2);
-            if (block.contains("h5::") || block.contains("rlp::")) {
-            if (block.contains("h5::") || block.contains("json::")) {
-            if (block.contains("h5::") || block.contains("msgpack::")) {
-            if (block.contains("h5::") || block.contains("cbor::")) {
-            if (block.contains("h5::") || block.contains("bson::")) {
-            if (block.contains("h5::") || block.contains("avro::")) {
+            if (block.contains("h5::") || block.contains("json::") || block.contains("msgpack::")
+                || block.contains("cbor::") || block.contains("bson::") || block.contains("avro::")
+                || block.contains("rlp::")) {
                 auto attrs = split_attrs(block);
                 out.append("[[");
                 for (std::size_t k = 0; k < attrs.size(); ++k) {
@@ -332,12 +314,13 @@ inline void install_virtual_files(clang::tooling::ClangTool& Tool,
     for (const auto& p : paths) {
         std::string content = read_file(p);
         if (content.empty()) continue;
-        if (content.find("h5::") == std::string::npos && content.find("rlp::") == std::string::npos) continue;
-        if (content.find("h5::") == std::string::npos && content.find("json::") == std::string::npos) continue;
-        if (content.find("h5::") == std::string::npos && content.find("msgpack::") == std::string::npos) continue;
-        if (content.find("h5::") == std::string::npos && content.find("cbor::") == std::string::npos) continue;
-        if (content.find("h5::") == std::string::npos && content.find("bson::") == std::string::npos) continue;
-        if (content.find("h5::") == std::string::npos && content.find("avro::") == std::string::npos) continue;
+        if (content.find("h5::") == std::string::npos
+            && content.find("json::") == std::string::npos
+            && content.find("msgpack::") == std::string::npos
+            && content.find("cbor::") == std::string::npos
+            && content.find("bson::") == std::string::npos
+            && content.find("avro::") == std::string::npos
+            && content.find("rlp::") == std::string::npos) continue;
         storage.push_back(rewrite(content));
         Tool.mapVirtualFile(p, storage.back());
     }

src/consumer_proto.hpp

@@ -18,15 +18,17 @@
 #include "producer_h5.hpp"
 #include "producer_sql.hpp"
 #include "consumer.hpp"
-#include "consumer_rlp.hpp"
+#include "consumer_json.hpp"
+#include "consumer_msgpack.hpp"
+#include "consumer_cbor.hpp"
 #include "consumer_bson.hpp"
 #include "consumer_avro.hpp"
+#include "consumer_rlp.hpp"
 #include "producer_pb.hpp"
 #include "consumer_pb.hpp"
+#include "consumer_proto.hpp"
+#include "pb_attr_translator.hpp"
 #include "h5_attr_translator.hpp"
-#include "consumer_json.hpp"
-#include "consumer_msgpack.hpp"
-#include "consumer_cbor.hpp"
 
 clang::ast_matchers::StatementMatcher h5templateMatcher = clang::ast_matchers::callExpr( clang::ast_matchers::allOf(
 	clang::ast_matchers::hasDescendant( clang::ast_matchers::declRefExpr( clang::ast_matchers::to( clang::ast_matchers::varDecl().bind("variableDecl")  ) ) ),
@@ -52,8 +54,6 @@ clang::ast_matchers::StatementMatcher h5templateMatcher = clang::ast_matchers::c
 	))  )))
 ));
 
-enum class OutputFormat { hdf5, protobuf, json, msgpack, cbor, bson, avro, rlp,
-                          sql_postgres, sql_mysql, sql_lite3 };
 // pbTemplateMatcher: same shape as h5templateMatcher, but triggers on the
 // pb.hpp public surface. Covers pb::encode, pb::decode, and pb::encode_into.
 clang::ast_matchers::StatementMatcher pbTemplateMatcher = clang::ast_matchers::callExpr( clang::ast_matchers::allOf(
@@ -71,7 +71,8 @@ clang::ast_matchers::StatementMatcher pbTemplateMatcher = clang::ast_matchers::c
 	))  )))
 ));
 
-enum class OutputFormat { hdf5, protobuf, json, msgpack, cbor, bson, avro, rlp };
+enum class OutputFormat { hdf5, protobuf, json, msgpack, cbor, bson, avro, rlp,
+                          sql_postgres, sql_mysql, sql_lite3 };
 
 static llvm::cl::OptionCategory MyToolCategory("h5cpp options");
 static llvm::cl::extrahelp CommonHelp(clang::tooling::CommonOptionsParser::HelpMessage);
@@ -103,6 +104,11 @@ static llvm::cl::opt<OutputFormat> Format(llvm::cl::desc("Output format:"),
     llvm::cl::init(OutputFormat::hdf5),
     llvm::cl::cat(MyToolCategory));
 
+static llvm::cl::opt<std::string> ProtoOutputFile("proto-out",
+    llvm::cl::desc("Phase 3: .proto schema output (used with --protobuf)"),
+    llvm::cl::value_desc("file"),
+    llvm::cl::cat(MyToolCategory));
+
 static llvm::cl::opt<bool> CheckMode("check",
     llvm::cl::desc("Verify that the existing generated file is up to date (exit 1 if stale)"),
     llvm::cl::cat(MyToolCategory),
@@ -128,6 +134,12 @@ int main(int argc, const char **argv) {
 	h5_attr_translator::install_virtual_files(
 		Tool, OptionsParser.getSourcePathList(), _h5_attr_storage);
 
+	// Issue #31: rewrite [[pb::xxx(...)]] → [[clang::annotate("pb::xxx", ...)]]
+	// for each source path before Clang sees it.
+	std::vector<std::string> _pb_attr_storage;
+	pb_attr_translator::install_virtual_files(
+		Tool, OptionsParser.getSourcePathList(), _pb_attr_storage);
+
 	std::string work_path = OutputFile;
 	if (CheckMode) {
 		work_path = OutputFile + ".h5cpp-check";
@@ -145,64 +157,52 @@ int main(int argc, const char **argv) {
 			}
 			case OutputFormat::protobuf: {
 				PbTemplateCallback<PbProducer> callback(work_path);
-				clang::ast_matchers::MatchFinder Finder;
 				Finder.addMatcher(pbTemplateMatcher, &callback);
+				std::optional<ProtoTemplateCallback> proto_cb;
+				if (!ProtoOutputFile.empty()) {
+					proto_cb.emplace(ProtoOutputFile);
+					Finder.addMatcher(pbTemplateMatcher, &*proto_cb);
+				}
 				rc = Tool.run(clang::tooling::newFrontendActionFactory(&Finder).get());
 				if (rc == 0 && callback.error()) rc = 1;
 				break;
+			}
 			case OutputFormat::json: {
 				JsonTemplateCallback callback(work_path);
 				Finder.addMatcher(h5templateMatcher, &callback);
 				rc = Tool.run(clang::tooling::newFrontendActionFactory(&Finder).get());
-			}
-			case OutputFormat::json:
-				llvm::errs() << "h5cpp-compiler: --format json not yet implemented\n";
-				rc = 1;
 				break;
 			}
-			case OutputFormat::msgpack:
-				llvm::errs() << "h5cpp-compiler: --format msgpack not yet implemented\n";
-				rc = 1;
 			case OutputFormat::msgpack: {
 				MsgpackTemplateCallback callback(work_path);
 				Finder.addMatcher(h5templateMatcher, &callback);
 				rc = Tool.run(clang::tooling::newFrontendActionFactory(&Finder).get());
 				break;
 			}
-				llvm::errs() << "h5cpp-compiler: --format cbor not yet implemented\n";
-				rc = 1;
 			case OutputFormat::cbor: {
 				CborTemplateCallback callback(work_path);
 				Finder.addMatcher(h5templateMatcher, &callback);
 				rc = Tool.run(clang::tooling::newFrontendActionFactory(&Finder).get());
 				break;
 			}
-			case OutputFormat::bson:
-				llvm::errs() << "h5cpp-compiler: --format bson not yet implemented\n";
-				rc = 1;
 			case OutputFormat::bson: {
 				BsonTemplateCallback callback(work_path);
 				Finder.addMatcher(h5templateMatcher, &callback);
 				rc = Tool.run(clang::tooling::newFrontendActionFactory(&Finder).get());
 				break;
 			}
-			case OutputFormat::avro:
-				llvm::errs() << "h5cpp-compiler: --format avro not yet implemented\n";
-				rc = 1;
 			case OutputFormat::avro: {
 				AvroTemplateCallback callback(work_path);
 				Finder.addMatcher(h5templateMatcher, &callback);
 				rc = Tool.run(clang::tooling::newFrontendActionFactory(&Finder).get());
 				break;
+			}
 			case OutputFormat::rlp: {
 				RlpTemplateCallback callback(work_path);
 				Finder.addMatcher(h5templateMatcher, &callback);
 				rc = Tool.run(clang::tooling::newFrontendActionFactory(&Finder).get());
-			}
-			case OutputFormat::rlp:
-				llvm::errs() << "h5cpp-compiler: --format rlp not yet implemented\n";
-				rc = 1;
 				break;
+			}
 			case OutputFormat::sql_postgres: {
 				H5TemplateCallback<SqlProducer<SqlDialect::postgres>> callback(work_path);
 				Finder.addMatcher(h5templateMatcher, &callback);
@@ -221,7 +221,6 @@ int main(int argc, const char **argv) {
 				rc = Tool.run(clang::tooling::newFrontendActionFactory(&Finder).get());
 				break;
 			}
-			}
 		}
 	}