From 7f777b5e3d27d22e2d9f8329930ce88213ea588a Mon Sep 17 00:00:00 2001
From: ineed bots <ineedbots@pbot.org>
Date: Wed, 27 Dec 2023 14:57:53 -0600
Subject: [PATCH] feature(compiler): produce a developer map, mapping
 instruction opcode pos to source line number

---
 include/xsk/arc/assembler.hpp       |  2 ++
 include/xsk/arc/common/assembly.hpp |  1 +
 include/xsk/arc/common/types.hpp    |  2 +-
 include/xsk/arc/compiler.hpp        |  1 +
 include/xsk/gsc/assembler.hpp       |  2 ++
 include/xsk/gsc/common/assembly.hpp |  1 +
 include/xsk/gsc/common/types.hpp    |  2 +-
 include/xsk/gsc/compiler.hpp        |  1 +
 src/arc/assembler.cpp               | 13 +++++++++++++
 src/arc/compiler.cpp                | 19 +++++++++++++++++++
 src/gsc/assembler.cpp               | 13 +++++++++++++
 src/gsc/compiler.cpp                | 19 +++++++++++++++++++
 src/tool/main.cpp                   | 14 ++++++++++++++
 13 files changed, 88 insertions(+), 2 deletions(-)

diff --git a/include/xsk/arc/assembler.hpp b/include/xsk/arc/assembler.hpp
index 792f502d..af215ebb 100644
--- a/include/xsk/arc/assembler.hpp
+++ b/include/xsk/arc/assembler.hpp
@@ -17,6 +17,7 @@ class assembler
     function const* func_;
     assembly const* assembly_;
     utils::writer script_;
+    utils::writer dev_map_;
     std::unordered_map<std::string, u16> strpool_;
     std::vector<export_ref> exports_;
     std::vector<import_ref> imports_;
@@ -26,6 +27,7 @@ class assembler
 public:
     assembler(context const* ctx);
     auto assemble(assembly const& data, std::string const& name = {}) -> buffer;
+    auto get_dev_map() -> buffer;
 
 private:
     auto assemble_function(function& func) -> void;
diff --git a/include/xsk/arc/common/assembly.hpp b/include/xsk/arc/common/assembly.hpp
index a8c1b3a6..8efc27a5 100644
--- a/include/xsk/arc/common/assembly.hpp
+++ b/include/xsk/arc/common/assembly.hpp
@@ -166,6 +166,7 @@ struct instruction
     u32 size;
     opcode opcode;
     std::vector<std::string> data;
+    std::optional<position> pos;
 
     static auto make() -> instruction::ptr
     {
diff --git a/include/xsk/arc/common/types.hpp b/include/xsk/arc/common/types.hpp
index d9a6ccd2..d9f417a6 100644
--- a/include/xsk/arc/common/types.hpp
+++ b/include/xsk/arc/common/types.hpp
@@ -6,9 +6,9 @@
 #pragma once
 
 #include "xsk/arc/common/asset.hpp"
+#include "xsk/arc/common/location.hpp"
 #include "xsk/arc/common/assembly.hpp"
 #include "xsk/arc/common/buffer.hpp"
-#include "xsk/arc/common/location.hpp"
 #include "xsk/arc/common/exception.hpp"
 #include "xsk/arc/common/lookahead.hpp"
 #include "xsk/arc/common/directive.hpp"
diff --git a/include/xsk/arc/compiler.hpp b/include/xsk/arc/compiler.hpp
index 47ad5fd7..ea0d5c06 100644
--- a/include/xsk/arc/compiler.hpp
+++ b/include/xsk/arc/compiler.hpp
@@ -25,6 +25,7 @@ class compiler
     bool can_break_;
     bool can_continue_;
     bool developer_thread_;
+    std::optional<position> debug_pos_;
 
 public:
     compiler(context* ctx);
diff --git a/include/xsk/gsc/assembler.hpp b/include/xsk/gsc/assembler.hpp
index 98fa1613..1cd5e62a 100644
--- a/include/xsk/gsc/assembler.hpp
+++ b/include/xsk/gsc/assembler.hpp
@@ -18,10 +18,12 @@ class assembler
     assembly const* assembly_;
     utils::writer script_;
     utils::writer stack_;
+    utils::writer dev_map_;
 
 public:
     assembler(context const* ctx);
     auto assemble(assembly const& data) -> std::pair<buffer, buffer>;
+    auto get_dev_map() -> buffer;
 
 private:
     auto assemble_function(function const& func) -> void;
diff --git a/include/xsk/gsc/common/assembly.hpp b/include/xsk/gsc/common/assembly.hpp
index e4dd6df1..73f199b4 100644
--- a/include/xsk/gsc/common/assembly.hpp
+++ b/include/xsk/gsc/common/assembly.hpp
@@ -228,6 +228,7 @@ struct instruction
     u32 size;
     opcode opcode;
     std::vector<std::string> data;
+    std::optional<position> pos;
 
     static auto make() -> instruction::ptr
     {
diff --git a/include/xsk/gsc/common/types.hpp b/include/xsk/gsc/common/types.hpp
index 32fe5376..b9dc7f74 100644
--- a/include/xsk/gsc/common/types.hpp
+++ b/include/xsk/gsc/common/types.hpp
@@ -8,8 +8,8 @@
 #include "xsk/gsc/common/asset.hpp"
 #include "xsk/gsc/common/scope.hpp"
 #include "xsk/gsc/common/buffer.hpp"
-#include "xsk/gsc/common/assembly.hpp"
 #include "xsk/gsc/common/location.hpp"
+#include "xsk/gsc/common/assembly.hpp"
 #include "xsk/gsc/common/exception.hpp"
 #include "xsk/gsc/common/lookahead.hpp"
 #include "xsk/gsc/common/directive.hpp"
diff --git a/include/xsk/gsc/compiler.hpp b/include/xsk/gsc/compiler.hpp
index 19eb0726..fe5a897b 100644
--- a/include/xsk/gsc/compiler.hpp
+++ b/include/xsk/gsc/compiler.hpp
@@ -28,6 +28,7 @@ class compiler
     bool can_continue_;
     bool developer_thread_;
     bool animload_;
+    std::optional<position> debug_pos_;
 
 public:
     compiler(context* ctx);
diff --git a/src/arc/assembler.cpp b/src/arc/assembler.cpp
index c7f53f2e..f88d9e67 100644
--- a/src/arc/assembler.cpp
+++ b/src/arc/assembler.cpp
@@ -18,6 +18,7 @@ auto assembler::assemble(assembly const& data, std::string const& name) -> buffe
 {
     assembly_ = &data;
     script_.clear();
+    dev_map_.clear();
     strpool_.clear();
     exports_.clear();
     imports_.clear();
@@ -226,6 +227,11 @@ auto assembler::assemble(assembly const& data, std::string const& name) -> buffe
     return buffer{ script_.data(), script_.pos() };
 }
 
+auto assembler::get_dev_map() -> buffer
+{
+    return buffer{ dev_map_.data(), dev_map_.pos() };
+}
+
 auto assembler::assemble_function(function& func) -> void
 {
     auto labels = std::unordered_map<u32, std::string>();
@@ -270,6 +276,13 @@ auto assembler::assemble_function(function& func) -> void
 
 auto assembler::assemble_instruction(instruction const& inst) -> void
 {
+    if (inst.pos.has_value())
+    {
+        dev_map_.write<u32>(script_.pos());
+        dev_map_.write<position::counter_type>(inst.pos->line);
+        dev_map_.write<position::counter_type>(inst.pos->column);
+    }
+
     script_.write<u8>(static_cast<u8>(ctx_->opcode_id(inst.opcode)));
 
     switch (inst.opcode)
diff --git a/src/arc/compiler.cpp b/src/arc/compiler.cpp
index 8e1b4e92..4aa4b0c9 100644
--- a/src/arc/compiler.cpp
+++ b/src/arc/compiler.cpp
@@ -138,6 +138,8 @@ auto compiler::emit_decl_function(decl_function const& func) -> void
 
 auto compiler::emit_stmt(stmt const& stm) -> void
 {
+    debug_pos_ = stm.loc().begin;
+
     switch (stm.kind())
     {
         case node::stmt_list:
@@ -686,6 +688,8 @@ auto compiler::emit_stmt_prof_end(stmt_prof_end const&) -> void
 
 auto compiler::emit_expr(expr const& exp) -> void
 {
+    debug_pos_ = exp.loc().begin;
+
     switch (exp.kind())
     {
         case node::expr_paren:
@@ -1867,6 +1871,11 @@ auto compiler::emit_opcode(opcode op) -> void
     inst->size = ctx_->opcode_size(op);
     inst->index = index_;
 
+    if (debug_pos_.has_value())
+    {
+        inst->pos = debug_pos_;
+    }
+
     index_ += inst->size;
 }
 
@@ -1880,6 +1889,11 @@ auto compiler::emit_opcode(opcode op, std::string const& data) -> void
     inst->index = index_;
     inst->data.push_back(data);
 
+    if (debug_pos_.has_value())
+    {
+        inst->pos = debug_pos_;
+    }
+
     index_ += inst->size;
 }
 
@@ -1893,6 +1907,11 @@ auto compiler::emit_opcode(opcode op, std::vector<std::string> const& data) -> v
     inst->index = index_;
     inst->data = data;
 
+    if (debug_pos_.has_value())
+    {
+        inst->pos = debug_pos_;
+    }
+
     index_ += inst->size;
 }
 
diff --git a/src/gsc/assembler.cpp b/src/gsc/assembler.cpp
index b7113747..eccba686 100644
--- a/src/gsc/assembler.cpp
+++ b/src/gsc/assembler.cpp
@@ -19,6 +19,7 @@ auto assembler::assemble(assembly const& data) -> std::pair<buffer, buffer>
     assembly_ = &data;
     script_.clear();
     stack_.clear();
+    dev_map_.clear();
 
     script_.write<u8>(ctx_->opcode_id(opcode::OP_End));
 
@@ -30,6 +31,11 @@ auto assembler::assemble(assembly const& data) -> std::pair<buffer, buffer>
     return { buffer{ script_.data(), script_.pos() }, buffer{ stack_.data(), stack_.pos() } };
 }
 
+auto assembler::get_dev_map() -> buffer
+{
+    return buffer{ dev_map_.data(), dev_map_.pos() };
+}
+
 auto assembler::assemble_function(function const& func) -> void
 {
     func_ = &func;
@@ -61,6 +67,13 @@ auto assembler::assemble_function(function const& func) -> void
 
 auto assembler::assemble_instruction(instruction const& inst) -> void
 {
+    if (inst.pos.has_value())
+    {
+        dev_map_.write<u32>(script_.pos());
+        dev_map_.write<position::counter_type>(inst.pos->line);
+        dev_map_.write<position::counter_type>(inst.pos->column);
+    }
+
     script_.write<u8>(ctx_->opcode_id(inst.opcode));
 
     switch (inst.opcode)
diff --git a/src/gsc/compiler.cpp b/src/gsc/compiler.cpp
index 4b37b6df..a0cb753f 100644
--- a/src/gsc/compiler.cpp
+++ b/src/gsc/compiler.cpp
@@ -147,6 +147,8 @@ auto compiler::emit_decl_function(decl_function const& func) -> void
 
 auto compiler::emit_stmt(stmt const& stm, scope& scp, bool last) -> void
 {
+    debug_pos_ = stm.loc().begin;
+
     switch (stm.kind())
     {
         case node::stmt_list:
@@ -925,6 +927,8 @@ auto compiler::emit_stmt_assertmsg(stmt_assertmsg const&, scope&) -> void
 
 auto compiler::emit_expr(expr const& exp, scope& scp) -> void
 {
+    debug_pos_ = exp.loc().begin;
+
     switch (exp.kind())
     {
         case node::expr_paren:
@@ -2198,6 +2202,11 @@ auto compiler::emit_opcode(opcode op) -> void
     inst->size = ctx_->opcode_size(op);
     inst->index = index_;
 
+    if (debug_pos_.has_value())
+    {
+        inst->pos = debug_pos_;
+    }
+
     index_ += inst->size;
 }
 
@@ -2211,6 +2220,11 @@ auto compiler::emit_opcode(opcode op, std::string const& data) -> void
     inst->index = index_;
     inst->data.push_back(data);
 
+    if (debug_pos_.has_value())
+    {
+        inst->pos = debug_pos_;
+    }
+
     index_ += inst->size;
 }
 
@@ -2224,6 +2238,11 @@ auto compiler::emit_opcode(opcode op, std::vector<std::string> const& data) -> v
     inst->index = index_;
     inst->data = data;
 
+    if (debug_pos_.has_value())
+    {
+        inst->pos = debug_pos_;
+    }
+
     index_ += inst->size;
 }
 
diff --git a/src/tool/main.cpp b/src/tool/main.cpp
index fa9163b2..bbe45d5a 100644
--- a/src/tool/main.cpp
+++ b/src/tool/main.cpp
@@ -286,6 +286,13 @@ auto compile_file(game game, mach mach, fs::path file, fs::path rel) -> void
                 auto result = script.serialize();
                 utils::file::save(fs::path{ "compiled" } / rel, result);
                 std::cout << fmt::format("compiled {}\n", rel.generic_string());
+
+                auto dev_map = contexts[game][mach]->assembler().get_dev_map();
+                if (dev_map.size > 0)
+                {
+                    utils::file::save(fs::path{ "compiled" } / fs::path{ "developer_maps" } / rel, dev_map.data, dev_map.size);
+                    std::cout << fmt::format("wrote developer map {}\n", rel.generic_string());
+                }
             }
         }
     }
@@ -753,6 +760,13 @@ void compile_file(game game, mach mach, fs::path const& file, fs::path rel)
 
         utils::file::save(fs::path{ "compiled" } / rel, outbin.data, outbin.size);
         std::cout << fmt::format("compiled {}\n", rel.generic_string());
+
+        auto dev_map = contexts[game][mach]->assembler().get_dev_map();
+        if (dev_map.size > 0)
+        {
+            utils::file::save(fs::path{ "compiled" } / fs::path{ "developer_maps" } / rel, dev_map.data, dev_map.size);
+            std::cout << fmt::format("wrote developer map {}\n", rel.generic_string());
+        }
     }
     catch (std::exception const& e)
     {