[Lightweight] Add dbtool fold

dbtool fold --output FILE  emits a self-contained baseline (.cpp plugin
or .sql script) that reproduces the post-migration state from an empty
DB. .sql output requires --dialect (sqlite, postgres, mssql, mysql);
.cpp output is dialect-agnostic. Runs without any DB connection - loads
plugins, walks migrations in memory, writes a file.

Built on a new pure plan-walk primitive
MigrationManager::FoldRegisteredMigrations(formatter, upToInclusive)
that folds every registered migration into a per-table view of the
final shape plus a chronological list of data steps, indexes, and
releases.

The fold module (src/Lightweight/MigrationFold/{Folder,CppEmitter,
SqlEmitter}.{hpp,cpp}) emits via the existing ToSql() formatter path so
each dialect's CREATE TABLE / CREATE INDEX / INSERT codegen stays the
single source of truth. The .cpp emitter wraps the body in
LIGHTWEIGHT_SQL_MIGRATION; the .sql emitter additionally emits CREATE
TABLE schema_migrations and a stamping INSERT for every folded
timestamp so the post-fold DB looks identical to a real apply-all run.

Also pulls in CodeGen/SplitFileWriter shared codegen helper used by the
.cpp emitter to bin-pack large baselines across multiple files.

Tests: fold unit tests cover create/altercolumn/drop-table cleanup,
data-step chronological order, --up-to truncation, RawSql passthrough,
column rename FK propagation, release-range filtering, ResolveUpTo
parsing. SqlEmitter/CppEmitter round-trip tests verify the emitted
artifacts match the expected shape. SplitFileWriter tests cover bin-
packing, single-chunk, zero-budget, and oversize-block boundaries.

All [Fold] and [SplitFileWriter] tests pass against sqlite3,
mssql2022, and postgres. Full SqlMigration suite (44 cases / 210
assertions) green on all three.

Signed-off-by: Christian Parpart <christian@parpart.family>

Author: christianparpart

src/Lightweight/CMakeLists.txt

@@ -63,6 +63,12 @@ set(HEADER_FILES
     SqlQuery/Select.hpp
     SqlQuery/Update.hpp
 
+    CodeGen/SplitFileWriter.hpp
+
+    MigrationFold/Folder.hpp
+    MigrationFold/CppEmitter.hpp
+    MigrationFold/SqlEmitter.hpp
+
     DataMapper/BelongsTo.hpp
     DataMapper/DataMapper.hpp
     DataMapper/Error.hpp
@@ -125,6 +131,12 @@ set(SOURCE_FILES
     SqlQuery/Migrate.cpp
     SqlQuery/MigrationPlan.cpp
     SqlQuery/Select.cpp
+
+    CodeGen/SplitFileWriter.cpp
+
+    MigrationFold/Folder.cpp
+    MigrationFold/CppEmitter.cpp
+    MigrationFold/SqlEmitter.cpp
     SqlQueryFormatter.cpp
     SqlSchema.cpp
     SqlStatement.cpp

src/Lightweight/CodeGen/SplitFileWriter.cpp

@@ -0,0 +1,176 @@
+// SPDX-License-Identifier: Apache-2.0
+
+#include "SplitFileWriter.hpp"
+
+#include <format>
+#include <fstream>
+#include <stdexcept>
+#include <system_error>
+
+namespace Lightweight::CodeGen
+{
+
+std::vector<std::vector<CodeBlock>> GroupBlocksByLineBudget(std::vector<CodeBlock> const& blocks,
+                                                            std::size_t maxLinesPerFile)
+{
+    std::vector<std::vector<CodeBlock>> chunks;
+    if (blocks.empty())
+        return chunks;
+
+    if (maxLinesPerFile == 0)
+    {
+        chunks.emplace_back(blocks);
+        return chunks;
+    }
+
+    chunks.emplace_back();
+    std::size_t currentLines = 0;
+    for (auto const& block: blocks)
+    {
+        if (!chunks.back().empty() && currentLines + block.lineCount > maxLinesPerFile)
+        {
+            chunks.emplace_back();
+            currentLines = 0;
+        }
+        chunks.back().push_back(block);
+        currentLines += block.lineCount;
+    }
+    if (chunks.back().empty())
+        chunks.pop_back();
+    return chunks;
+}
+
+namespace
+{
+    /// @brief Ensures the parent directory exists so subsequent `ofstream` opens succeed.
+    void EnsureParentDirectoryExists(std::filesystem::path const& filePath)
+    {
+        auto const parent = filePath.parent_path();
+        if (parent.empty())
+            return;
+        std::error_code ec;
+        std::filesystem::create_directories(parent, ec);
+        if (ec)
+            throw std::runtime_error(std::format("Failed to create directory {}: {}", parent.string(), ec.message()));
+    }
+
+    /// @brief Writes one chunk file with optional header/footer.
+    void WriteChunkFile(std::filesystem::path const& path,
+                        std::vector<CodeBlock> const& chunk,
+                        std::string_view header,
+                        std::string_view footer)
+    {
+        EnsureParentDirectoryExists(path);
+        std::ofstream out(path);
+        if (!out.is_open())
+            throw std::runtime_error(std::format("Failed to open output file: {}", path.string()));
+        if (!header.empty())
+            out << header;
+        for (auto const& block: chunk)
+            out << block.content;
+        if (!footer.empty())
+            out << footer;
+    }
+
+    /// @brief Total line count of all blocks; cached on the `CodeBlock` so this loop
+    /// is just a sum, not a per-call newline scan.
+    [[nodiscard]] std::size_t TotalLines(std::vector<CodeBlock> const& blocks)
+    {
+        std::size_t total = 0;
+        for (auto const& b: blocks)
+            total += b.lineCount;
+        return total;
+    }
+} // namespace
+
+WriteResult EmitChunked(std::filesystem::path const& outputPath,
+                        std::vector<CodeBlock> const& blocks,
+                        std::size_t maxLinesPerFile,
+                        std::string_view fileHeader,
+                        std::string_view fileFooter)
+{
+    WriteResult result;
+
+    if (maxLinesPerFile == 0 || TotalLines(blocks) <= maxLinesPerFile)
+    {
+        WriteChunkFile(outputPath, blocks, fileHeader, fileFooter);
+        result.writtenFiles.push_back(outputPath);
+        return result;
+    }
+
+    auto const chunks = GroupBlocksByLineBudget(blocks, maxLinesPerFile);
+    auto const stem = outputPath.parent_path() / outputPath.stem();
+    auto const ext = outputPath.extension().string();
+
+    for (std::size_t i = 0; i < chunks.size(); ++i)
+    {
+        auto partPath = std::filesystem::path { std::format("{}_part{:02}{}", stem.string(), i + 1, ext) };
+        WriteChunkFile(partPath, chunks[i], fileHeader, fileFooter);
+        result.writtenFiles.push_back(std::move(partPath));
+    }
+
+    return result;
+}
+
+void EmitPluginCmake(std::filesystem::path const& outputDir, std::string_view pluginName, std::string_view sourceGlob)
+{
+    std::error_code ec;
+    std::filesystem::create_directories(outputDir, ec);
+    if (ec)
+        throw std::runtime_error(std::format("Failed to create directory {}: {}", outputDir.string(), ec.message()));
+
+    auto const cmakePath = outputDir / "CMakeLists.txt";
+    {
+        std::ofstream out(cmakePath);
+        if (!out.is_open())
+            throw std::runtime_error(std::format("Failed to open output file: {}", cmakePath.string()));
+        out << "# SPDX-License-Identifier: Apache-2.0\n"
+            << "# Auto-generated. DO NOT EDIT.\n"
+            << "\n"
+            << "cmake_minimum_required(VERSION 3.25)\n"
+            << "\n"
+            << "# Pick up every generated migration source. CONFIGURE_DEPENDS makes CMake re-glob\n"
+            << "# when the generator regenerates the directory, so new sources enter the build\n"
+            << "# without a manual reconfigure.\n"
+            << "file(GLOB " << pluginName << "_MIGRATIONS CONFIGURE_DEPENDS\n"
+            << "    \"${CMAKE_CURRENT_SOURCE_DIR}/" << sourceGlob << "\"\n"
+            << ")\n"
+            << "\n"
+            << "add_library(" << pluginName << " MODULE\n"
+            << "    Plugin.cpp\n"
+            << "    ${" << pluginName << "_MIGRATIONS}\n"
+            << ")\n"
+            << "\n"
+            << "target_link_libraries(" << pluginName << " PRIVATE Lightweight::Lightweight)\n"
+            << "\n"
+            << "set_target_properties(" << pluginName << " PROPERTIES\n"
+            << "    LIBRARY_OUTPUT_DIRECTORY \"${CMAKE_BINARY_DIR}/plugins\"\n"
+            << "    RUNTIME_OUTPUT_DIRECTORY \"${CMAKE_BINARY_DIR}/plugins\"\n"
+            << "    PREFIX \"\"\n"
+            << ")\n"
+            << "\n"
+            << "# Generated migration sources are large and intentionally literal; skip clang-tidy\n"
+            << "# so lint thresholds (function size, cognitive complexity) don't trip on them.\n"
+            << "set_target_properties(" << pluginName << " PROPERTIES CXX_CLANG_TIDY \"\")\n"
+            << "set_source_files_properties(${" << pluginName << "_MIGRATIONS} PROPERTIES SKIP_LINTING TRUE)\n";
+    }
+
+    auto const pluginPath = outputDir / "Plugin.cpp";
+    {
+        std::ofstream out(pluginPath);
+        if (!out.is_open())
+            throw std::runtime_error(std::format("Failed to open output file: {}", pluginPath.string()));
+        out << "// SPDX-License-Identifier: Apache-2.0\n"
+            << "// Auto-generated. DO NOT EDIT.\n"
+            << "//\n"
+            << "// Migration plugin entry point. Individual migrations self-register with the\n"
+            << "// MigrationManager via static initialization; this file only exposes the\n"
+            << "// plugin ABI that dbtool expects when dlopen'ing the shared module.\n"
+            << "\n"
+            << "#include <Lightweight/SqlMigration.hpp>\n"
+            << "\n"
+            << "LIGHTWEIGHT_MIGRATION_PLUGIN()\n";
+    }
+}
+
+} // namespace Lightweight::CodeGen

src/Lightweight/CodeGen/SplitFileWriter.hpp

@@ -0,0 +1,78 @@
+// SPDX-License-Identifier: Apache-2.0
+
+#pragma once
+
+#include "../Api.hpp"
+
+#include <cstddef>
+#include <filesystem>
+#include <string>
+#include <string_view>
+#include <vector>
+
+namespace Lightweight::CodeGen
+{
+
+/// @brief A pre-rendered text block plus its line count, ready to be packed into
+/// one or more output files by `EmitChunked`.
+///
+/// The block is treated as opaque — `EmitChunked` never splits a block in two.
+/// Callers compute `lineCount` once and reuse the same value across multiple
+/// passes, since counting newlines on every visit would be wasteful for large
+/// migrations.
+struct CodeBlock
+{
+    /// Pre-rendered text of the block. Treated as opaque by `EmitChunked` —
+    /// never split across files.
+    std::string content;
+
+    /// Number of newlines in `content`, computed once by the caller and reused
+    /// across packing passes.
+    std::size_t lineCount = 0;
+};
+
+/// @brief Greedy bin-packing of `blocks` across files of at most `maxLinesPerFile`
+/// lines.
+///
+/// One block always lands wholly in one chunk — even when its line count exceeds
+/// the budget — because builder DSL chains cannot be broken across translation
+/// units. The returned outer vector has no empty inner vectors. When
+/// `maxLinesPerFile == 0`, every block lands in a single chunk.
+[[nodiscard]] LIGHTWEIGHT_API std::vector<std::vector<CodeBlock>> GroupBlocksByLineBudget(
+    std::vector<CodeBlock> const& blocks, std::size_t maxLinesPerFile);
+
+/// @brief Result of an `EmitChunked` call.
+struct WriteResult
+{
+    /// All files actually written, in apply order. Either `[outputPath]` (single
+    /// file) or `[<stem>_part01.<ext>, <stem>_part02.<ext>, ...]`.
+    std::vector<std::filesystem::path> writtenFiles;
+};
+
+/// @brief Writes one or more output files, splitting `blocks` across `<stem>_partNN.<ext>`
+/// siblings when their combined line count exceeds `maxLinesPerFile`. When the
+/// total fits, a single `outputPath` is written and split is skipped.
+///
+/// `fileHeader` and `fileFooter` are emitted at the top/bottom of every produced
+/// file (so all chunks remain self-contained translation units when used for
+/// `.cpp` outputs). Pass empty strings to skip them.
+///
+/// @return `WriteResult` listing the paths actually written.
+/// @throws `std::runtime_error` if any output file cannot be opened.
+[[nodiscard]] LIGHTWEIGHT_API WriteResult EmitChunked(std::filesystem::path const& outputPath,
+                                                      std::vector<CodeBlock> const& blocks,
+                                                      std::size_t maxLinesPerFile,
+                                                      std::string_view fileHeader = {},
+                                                      std::string_view fileFooter = {});
+
+/// @brief Writes a `CMakeLists.txt` and a `Plugin.cpp` next to the generated
+/// migration sources so the output directory becomes a drop-in plugin.
+///
+/// Mirrors the layout of `src/tools/LupMigrationsPlugin/` so consumers can
+/// `add_subdirectory()` the generated dir and pick up a self-registering plugin
+/// without further glue.
+LIGHTWEIGHT_API void EmitPluginCmake(std::filesystem::path const& outputDir,
+                                     std::string_view pluginName,
+                                     std::string_view sourceGlob = "lup_*.cpp");
+
+} // namespace Lightweight::CodeGen

src/Lightweight/DataBinder/SqlGuid.cpp

@@ -92,8 +92,8 @@ std::optional<SqlGuid> SqlGuid::TryParse(std::string_view const& text) noexcept
     // Variant nibble at position 19 must be a valid hex digit
     // We accept all variants (RFC 4122: 8-B, Microsoft: C-D, etc.)
     auto const variant = text[19];
-    auto const isHexDigit = (variant >= '0' && variant <= '9') || (variant >= 'A' && variant <= 'F')
-                            || (variant >= 'a' && variant <= 'f');
+    auto const isHexDigit =
+        (variant >= '0' && variant <= '9') || (variant >= 'A' && variant <= 'F') || (variant >= 'a' && variant <= 'f');
     if (!isHexDigit)
         return std::nullopt;
 

src/Lightweight/DataBinder/SqlGuid.hpp

@@ -81,8 +81,8 @@ constexpr SqlGuid SqlGuid::UnsafeParse(std::string_view const& text) noexcept
     // Variant nibble at position 19 must be a valid hex digit
     // We accept all variants (RFC 4122: 8-B, Microsoft: C-D, etc.)
     auto const variant = text[19];
-    auto const isHexDigit = (variant >= '0' && variant <= '9') || (variant >= 'A' && variant <= 'F')
-                            || (variant >= 'a' && variant <= 'f');
+    auto const isHexDigit =
+        (variant >= '0' && variant <= '9') || (variant >= 'A' && variant <= 'F') || (variant >= 'a' && variant <= 'f');
     if (!isHexDigit)
         return { "\x04" };
 

src/Lightweight/DataBinder/SqlRawColumn.hpp

@@ -48,8 +48,9 @@ struct SqlDataBinder<SqlRawColumn>
         // requires the ColumnSize to be the buffer length when not using data-at-execution.
         // A size of 0 causes HY104 "Invalid precision value" error.
         SQLULEN columnSize = value.metadata.size;
-        if (columnSize == 0 && (value.metadata.sqlType == SQL_LONGVARCHAR || value.metadata.sqlType == SQL_LONGVARBINARY
-                                || value.metadata.sqlType == SQL_WLONGVARCHAR))
+        if (columnSize == 0
+            && (value.metadata.sqlType == SQL_LONGVARCHAR || value.metadata.sqlType == SQL_LONGVARBINARY
+                || value.metadata.sqlType == SQL_WLONGVARCHAR))
         {
             columnSize = value.metadata.bufferLength;
         }