From 5ddf1e4b4d0187c1b99491daa328cae2c16f03f1 Mon Sep 17 00:00:00 2001
From: Tyler Wilding <xTVaser@users.noreply.github.com>
Date: Sun, 21 Jan 2024 01:05:41 -0500
Subject: [PATCH] lsp: flag unnamed variables in IR files (#3325)

![image](https://github.com/open-goal/jak-project/assets/13153231/a020fbf6-65b9-4317-8a9d-5991f2d15e32)
---
 .../config/jak3/ntsc_v1/var_names.jsonc       | 281 ++++--------------
 lsp/state/workspace.cpp                       |  32 +-
 lsp/state/workspace.h                         |   4 +-
 3 files changed, 91 insertions(+), 226 deletions(-)

diff --git a/decompiler/config/jak3/ntsc_v1/var_names.jsonc b/decompiler/config/jak3/ntsc_v1/var_names.jsonc
index 98bfe546d5..290dc00ec9 100644
--- a/decompiler/config/jak3/ntsc_v1/var_names.jsonc
+++ b/decompiler/config/jak3/ntsc_v1/var_names.jsonc
@@ -1,313 +1,162 @@
 {
   "(method 0 inline-array-class)": {
-    "args": [
-      "allocation",
-      "type-to-make",
-      "count"
-    ]
+    "args": ["allocation", "type-to-make", "count"]
   },
   "(method 9 inline-array-class)": {
-    "args": [
-      "object-to-insert"
-    ]
+    "args": ["object-to-insert"]
   },
   "(method 3 inline-array-class)": {
-    "args": [
-      "idx-to-remove"
-    ]
+    "args": ["idx-to-remove"]
   },
   "identity": {
-    "args": [
-      "obj"
-    ]
+    "args": ["obj"]
   },
   "1/": {
-    "args": [
-      "x"
-    ]
+    "args": ["x"]
   },
   "+": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "-": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "*": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "/": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "ash": {
-    "args": [
-      "x",
-      "shift-amount"
-    ]
+    "args": ["x", "shift-amount"]
   },
   "mod": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "rem": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "abs": {
-    "args": [
-      "x"
-    ]
+    "args": ["x"]
   },
   "min": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "max": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "logior": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "logand": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "lognor": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "logxor": {
-    "args": [
-      "a",
-      "b"
-    ]
+    "args": ["a", "b"]
   },
   "lognot": {
-    "args": [
-      "x"
-    ]
+    "args": ["x"]
   },
   "basic-type?": {
-    "args": [
-      "obj",
-      "typ"
-    ]
+    "args": ["obj", "typ"]
   },
   "type-type?": {
-    "args": [
-      "check-type",
-      "parent-type"
-    ]
+    "args": ["check-type", "parent-type"]
   },
   "type?": {
-    "args": [
-      "obj",
-      "desired-type"
-    ]
+    "args": ["obj", "desired-type"]
   },
   "find-parent-method": {
-    "args": [
-      "typ",
-      "method-id"
-    ]
+    "args": ["typ", "method-id"]
   },
   "ref&": {
-    "args": [
-      "list",
-      "idx"
-    ]
+    "args": ["list", "idx"]
   },
   "ref": {
-    "args": [
-      "list",
-      "idx"
-    ]
+    "args": ["list", "idx"]
   },
   "last": {
-    "args": [
-      "list"
-    ]
+    "args": ["list"]
   },
   "member": {
-    "args": [
-      "obj-to-find",
-      "list"
-    ]
+    "args": ["obj-to-find", "list"]
   },
   "nmember": {
-    "args": [
-      "obj-to-find",
-      "list"
-    ]
+    "args": ["obj-to-find", "list"]
   },
   "assoc": {
-    "args": [
-      "key",
-      "assoc-list"
-    ]
+    "args": ["key", "assoc-list"]
   },
   "assoce": {
-    "args": [
-      "key",
-      "assoc-list"
-    ]
+    "args": ["key", "assoc-list"]
   },
   "nassoc": {
-    "args": [
-      "key",
-      "assoc-list"
-    ]
+    "args": ["key", "assoc-list"]
   },
   "nassoce": {
-    "args": [
-      "key",
-      "assoc-list"
-    ]
+    "args": ["key", "assoc-list"]
   },
   "append!": {
-    "args": [
-      "list",
-      "new-obj"
-    ]
+    "args": ["list", "new-obj"]
   },
   "delete!": {
-    "args": [
-      "obj",
-      "list"
-    ]
+    "args": ["obj", "list"]
   },
   "delete-car!": {
-    "args": [
-      "car-to-match",
-      "list"
-    ]
+    "args": ["car-to-match", "list"]
   },
   "insert-cons!": {
-    "args": [
-      "new-obj",
-      "list"
-    ]
+    "args": ["new-obj", "list"]
   },
   "sort": {
-    "args": [
-      "list",
-      "compare-func"
-    ]
+    "args": ["list", "compare-func"]
   },
   "string->symbol-debug": {
-    "args": [
-      "str"
-    ]
+    "args": ["str"]
   },
   "symbol->string-debug": {
-    "args": [
-      "sym"
-    ]
+    "args": ["sym"]
   },
   "symbol->hash": {
-    "args": [
-      "sym"
-    ]
+    "args": ["sym"]
   },
   "mem-copy!": {
-    "args": [
-      "dst",
-      "src",
-      "bytes"
-    ]
+    "args": ["dst", "src", "bytes"]
   },
   "qmem-copy<-!": {
-    "args": [
-      "dst",
-      "src",
-      "qwc"
-    ]
+    "args": ["dst", "src", "qwc"]
   },
   "qmem-copy->!": {
-    "args": [
-      "dst",
-      "src",
-      "qwc"
-    ]
+    "args": ["dst", "src", "qwc"]
   },
   "qmem-clear!": {
-    "args": [
-      "dst",
-      "qwc"
-    ]
+    "args": ["dst", "qwc"]
   },
   "mem-set32!": {
-    "args": [
-      "dst",
-      "word-count",
-      "value"
-    ]
+    "args": ["dst", "word-count", "value"]
   },
   "mem-or!": {
-    "args": [
-      "dst",
-      "src",
-      "bytes"
-    ]
+    "args": ["dst", "src", "bytes"]
   },
   "fact": {
-    "args": [
-      "x"
-    ]
+    "args": ["x"]
   },
   "print": {
-    "args": [
-      "obj"
-    ]
+    "args": ["obj"]
   },
   "printl": {
-    "args": [
-      "obj"
-    ]
+    "args": ["obj"]
   },
   "inspect": {
-    "args": [
-      "obj"
-    ]
+    "args": ["obj"]
   },
   "mem-print": {
-    "args": [
-      "ptr",
-      "word-count"
-    ]
+    "args": ["ptr", "word-count"]
   },
   "print-tree-bitmask": {
-    "args": [
-      "mask",
-      "count"
-    ]
+    "args": ["mask", "count"]
   },
   "valid?": {
     "args": [
@@ -319,29 +168,15 @@
     ]
   },
   "matrix-transpose!": {
-    "args": [
-      "dst",
-      "src"
-    ]
+    "args": ["dst", "src"]
   },
   "matrix-axis-sin-cos!": {
-    "args": [
-      "dst",
-      "axis",
-      "s",
-      "c"
-    ]
+    "args": ["dst", "axis", "s", "c"]
   },
   "vector-rad<-vector-deg/2!": {
-    "args": [
-      "out",
-      "in"
-    ]
+    "args": ["out", "in"]
   },
   "vector-rad<-vector-deg!": {
-    "args": [
-      "out",
-      "in"
-    ]
+    "args": ["out", "in"]
   }
-}
\ No newline at end of file
+}
diff --git a/lsp/state/workspace.cpp b/lsp/state/workspace.cpp
index 6447df0e48..e8ac977f57 100644
--- a/lsp/state/workspace.cpp
+++ b/lsp/state/workspace.cpp
@@ -326,13 +326,19 @@ WorkspaceIRFile::WorkspaceIRFile(const std::string& content) {
   const auto line_ending = file_util::get_majority_file_line_endings(content);
   m_lines = str_util::split_string(content, line_ending);
 
+  bool in_opengoal_block = false;
   for (int i = 0; i < m_lines.size(); i++) {
     const auto& line = m_lines.at(i);
     if (m_all_types_uri == "") {
       find_all_types_path(line);
     }
+    if (str_util::contains(line, ";;-*-OpenGOAL-Start-*-")) {
+      in_opengoal_block = true;
+    } else if (str_util::contains(line, ";;-*-OpenGOAL-End-*-")) {
+      in_opengoal_block = false;
+    }
     find_function_symbol(i, line);
-    identify_diagnostics(i, line);
+    identify_diagnostics(i, line, in_opengoal_block);
   }
 
   lg::info("Added new IR file. {} lines with {} symbols and {} diagnostics", m_lines.size(),
@@ -400,7 +406,10 @@ void WorkspaceIRFile::find_function_symbol(const uint32_t line_num_zero_based,
 }
 
 void WorkspaceIRFile::identify_diagnostics(const uint32_t line_num_zero_based,
-                                           const std::string& line) {
+                                           const std::string& line,
+                                           const bool in_opengoal_block) {
+  std::regex unnamed_variables_regex(
+      "(?:(?:arg\\d+)|(?:f\\d+|at|v[0-1]|a[0-3]|t[0-9]|s[0-7]|k[0-1]|gp|sp|sv|fp|ra)\\-\\d+)");
   std::regex info_regex(";; INFO: (.*)");
   std::regex warn_regex(";; WARN: (.*)");
   std::regex error_regex(";; ERROR: (.*)");
@@ -408,6 +417,24 @@ void WorkspaceIRFile::identify_diagnostics(const uint32_t line_num_zero_based,
   std::smatch warn_matches;
   std::smatch error_matches;
 
+  if (in_opengoal_block) {
+    for (auto regex_iterator =
+             std::sregex_iterator(line.begin(), line.end(), unnamed_variables_regex);
+         regex_iterator != std::sregex_iterator(); ++regex_iterator) {
+      std::smatch match = *regex_iterator;
+      LSPSpec::Diagnostic new_diag;
+      new_diag.m_severity = LSPSpec::DiagnosticSeverity::Warning;
+      new_diag.m_message = fmt::format("{} - Unnamed variable", match.str());
+      LSPSpec::Range diag_range;
+      diag_range.m_start = {line_num_zero_based, (u32)match.position(0)};
+      diag_range.m_end = {line_num_zero_based,
+                          diag_range.m_start.m_character + (u32)match.length()};
+      new_diag.m_range = diag_range;
+      new_diag.m_source = "OpenGOAL LSP";
+      m_diagnostics.push_back(new_diag);
+    }
+  }
+
   LSPSpec::Range diag_range;
   diag_range.m_start = {line_num_zero_based, 0};
   diag_range.m_end = {line_num_zero_based, (uint32_t)line.length() - 1};
@@ -440,6 +467,7 @@ void WorkspaceIRFile::identify_diagnostics(const uint32_t line_num_zero_based,
       return;
     }
   }
+
   // Check for a error level warnings
   if (std::regex_search(line, error_matches, error_regex)) {
     // NOTE - assumes we can only find 1 function per line
diff --git a/lsp/state/workspace.h b/lsp/state/workspace.h
index 54825c9d97..997a9bb6c5 100644
--- a/lsp/state/workspace.h
+++ b/lsp/state/workspace.h
@@ -53,7 +53,9 @@ class WorkspaceIRFile {
   /// Make any relevant diagnostics on the IR line.
   /// It's assumed each line in an IR can have atmost one diagnostic, and they are contained to just
   /// that line!
-  void identify_diagnostics(const uint32_t line_num_zero_based, const std::string& line);
+  void identify_diagnostics(const uint32_t line_num_zero_based,
+                            const std::string& line,
+                            const bool in_opengoal_block);
 };
 
 class WorkspaceAllTypesFile {