From 365a6ee384bcb2e9ec9b28262d65d99d785a544e Mon Sep 17 00:00:00 2001
From: Aaron Gyes <var.log@icloud.com>
Date: Tue, 21 Dec 2021 02:26:41 -0800
Subject: [PATCH] Use source_offset_t (uint32) in tokenizer.

Seems like size_t is unnecessarily large as well, as elsewhere
in the code we are clamping down to uint32_t / source_offset_t.

This makes tok_t more like 16 bytes. More cleanup seems desirable,
this is not very well hamrnoized across our code base.
---
 src/complete.cpp      |  5 +----
 src/parse_constants.h |  9 ++++++---
 src/parse_tree.h      |  4 ----
 src/tokenizer.h       | 18 +++++++++---------
 4 files changed, 16 insertions(+), 20 deletions(-)

diff --git a/src/complete.cpp b/src/complete.cpp
index 9bda3c61e..559791c7f 100644
--- a/src/complete.cpp
+++ b/src/complete.cpp
@@ -1674,10 +1674,7 @@ void completer_t::perform_for_commandline(wcstring cmdline) {
         custom_arg_data_t arg_data{&var_assignments};
         arg_data.had_ddash = had_ddash;
 
-        assert(cmd_tok.offset < std::numeric_limits<uint32_t>::max());
-        assert(cmd_tok.length < std::numeric_limits<uint32_t>::max());
-        source_range_t command_range = {static_cast<uint32_t>(cmd_tok.offset),
-                                        static_cast<uint32_t>(cmd_tok.length)};
+        source_range_t command_range = {cmd_tok.offset, cmd_tok.length};
 
         wcstring exp_command = cmd_tok.get_source(cmdline);
         bool unescaped =
diff --git a/src/parse_constants.h b/src/parse_constants.h
index 904e75ad5..a03e416fc 100644
--- a/src/parse_constants.h
+++ b/src/parse_constants.h
@@ -7,6 +7,9 @@
 #include "common.h"
 #include "enum_map.h"
 
+using source_offset_t = uint32_t;
+constexpr source_offset_t SOURCE_OFFSET_INVALID = static_cast<source_offset_t>(-1);
+
 #define PARSER_DIE()                   \
     do {                               \
         FLOG(error, L"Parser dying!"); \
@@ -15,10 +18,10 @@
 
 // A range of source code.
 struct source_range_t {
-    uint32_t start;
-    uint32_t length;
+    source_offset_t start;
+    source_offset_t length;
 
-    uint32_t end() const {
+    source_offset_t end() const {
         assert(start + length >= start && "Overflow");
         return start + length;
     }
diff --git a/src/parse_tree.h b/src/parse_tree.h
index dae6c8cd1..7f9d9e1e2 100644
--- a/src/parse_tree.h
+++ b/src/parse_tree.h
@@ -16,10 +16,6 @@
 #include "parse_constants.h"
 #include "tokenizer.h"
 
-typedef uint32_t source_offset_t;
-
-constexpr source_offset_t SOURCE_OFFSET_INVALID = static_cast<source_offset_t>(-1);
-
 /// A struct representing the token type that we use internally.
 struct parse_token_t {
     enum parse_token_type_t type;  // The type of the token as represented by the parser
diff --git a/src/tokenizer.h b/src/tokenizer.h
index 8a860cc13..f0504db02 100644
--- a/src/tokenizer.h
+++ b/src/tokenizer.h
@@ -10,8 +10,8 @@
 #include "parse_constants.h"
 #include "redirection.h"
 
-/// Token types.
-enum class token_type_t {
+/// Token types. XXX Why this isn't parse_token_type_t, I'm not really sure.
+enum class token_type_t : uint8_t {
     error,       /// Error reading token
     string,      /// String token
     pipe,        /// Pipe token
@@ -39,7 +39,7 @@ enum class token_type_t {
 
 using tok_flags_t = unsigned int;
 
-enum class tokenizer_error_t {
+enum class tokenizer_error_t : uint8_t {
     none,
     unterminated_quote,
     unterminated_subshell,
@@ -61,20 +61,20 @@ const wchar_t *tokenizer_get_error_message(tokenizer_error_t err);
 
 struct tok_t {
     // Offset of the token.
-    size_t offset{0};
+    source_offset_t offset{0};
     // Length of the token.
-    size_t length{0};
+    source_offset_t length{0};
 
     // If an error, this is the offset of the error within the token. A value of 0 means it occurred
     // at 'offset'.
-    size_t error_offset_within_token{size_t(-1)};
-
-    // The type of the token.
-    token_type_t type;
+    source_offset_t error_offset_within_token{SOURCE_OFFSET_INVALID};
 
     // If an error, this is the error code.
     tokenizer_error_t error{tokenizer_error_t::none};
 
+    // The type of the token.
+    token_type_t type;
+
     // Construct from a token type.
     explicit tok_t(token_type_t type);