From c3215249c2fa7e556ce8110d109a02ea2bacb0c3 Mon Sep 17 00:00:00 2001
From: Ryan Jones-Ward <sciguyryan@gmail.com>
Date: Tue, 9 Jul 2024 15:09:23 +0100
Subject: [PATCH] Initial work on the data section parsing.

---
 Cargo.toml                           |  2 +
 redox-core/Cargo.toml                |  2 +
 redox-core/src/parsing/asm_parser.rs | 80 +++++++++++++++++++++++++++-
 redox-terminal/src/main.rs           | 10 +++-
 4 files changed, 92 insertions(+), 2 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index f91862e..69da7d2 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -18,6 +18,8 @@ rand_xoshiro = "0.6.0"
 strum = "0.26.3"
 strum_macros = "0.26.4"
 thiserror = "1.0.61"
+unescape = "0.1.0"
+unicode-segmentation = "1.11.0"
 winres = "0.1.12"
 
 [profile.dev]
diff --git a/redox-core/Cargo.toml b/redox-core/Cargo.toml
index 8713b67..d0b0207 100644
--- a/redox-core/Cargo.toml
+++ b/redox-core/Cargo.toml
@@ -23,6 +23,8 @@ rand.workspace = true
 rand_xoshiro.workspace = true
 strum.workspace = true
 strum_macros.workspace = true
+unescape.workspace = true
+unicode-segmentation.workspace = true
 thiserror.workspace = true
 
 [dev-dependencies]
diff --git a/redox-core/src/parsing/asm_parser.rs b/redox-core/src/parsing/asm_parser.rs
index 78d89a6..643302c 100644
--- a/redox-core/src/parsing/asm_parser.rs
+++ b/redox-core/src/parsing/asm_parser.rs
@@ -157,7 +157,7 @@ impl<'a> AsmParser<'a> {
                 FileSection::Text => {
                     instructions.push(self.parse_code_line(line));
                 }
-                FileSection::Data => todo!(),
+                FileSection::Data => self.parse_data_line(line),
                 FileSection::ReadOnlyData => todo!(),
             }
         }
@@ -369,6 +369,68 @@ impl<'a> AsmParser<'a> {
         AsmParser::try_build_instruction(final_option.opcode, &arguments, label)
     }
 
+    /// Parse a data line of an assembly file.
+    ///
+    /// # Arguments
+    ///
+    /// * `line` - A code line to be parsed.
+    fn parse_data_line(&mut self, line: &str) {
+        use unicode_segmentation::UnicodeSegmentation;
+
+        let graphemes: Vec<&str> = line.graphemes(true).collect();
+
+        let mut in_quoted_string = false;
+
+        let mut escaped_args = vec![];
+        let mut buffer = String::new();
+        for (i, g) in graphemes.iter().enumerate() {
+            let is_escaped = i > 0 && graphemes[i - 1] == "\\";
+
+            if !is_escaped && (*g == "\"" || *g == "'") {
+                // We have reached the end of a quoted segment.
+                if in_quoted_string {
+                    AsmParser::push_buffer_if_not_empty(&mut buffer, &mut escaped_args);
+                }
+
+                in_quoted_string = !in_quoted_string;
+                continue;
+            }
+
+            if !in_quoted_string && (*g == " " || *g == "\t" || *g == ",") {
+                // We have reached the end of a segment.
+                if !buffer.is_empty() {
+                    AsmParser::push_buffer_if_not_empty(&mut buffer, &mut escaped_args);
+                }
+
+                continue;
+            }
+
+            buffer.push_str(g);
+        }
+
+        assert!(
+            !in_quoted_string,
+            "invalid syntax - closing quotation mark not present for line: {line}."
+        );
+
+        // Push the final argument to the list.
+        AsmParser::push_buffer_if_not_empty(&mut buffer, &mut escaped_args);
+
+        assert!(
+            escaped_args.len() >= 3,
+            "invalid syntax - insufficient arguments for line: {line}."
+        );
+
+        // Now we want to ensure we correctly handle any escape sequences.
+        let mut arguments = vec![];
+        for argument in escaped_args {
+            let un = unescape::unescape(&argument).unwrap_or(argument);
+            arguments.push(un);
+        }
+
+        println!("{arguments:?}");
+    }
+
     /// Parse a section line of an ASM file.
     ///
     /// # Arguments
@@ -389,6 +451,22 @@ impl<'a> AsmParser<'a> {
         }
     }
 
+    /// Push a string buffer onto an string argument vector.
+    ///
+    /// # Arguments
+    ///
+    /// * `arg` - The argument string.
+    /// * `vec` - The vector that the argument should be pushed too, if it isn't empty.
+    #[inline]
+    fn push_buffer_if_not_empty(arg: &mut String, vec: &mut Vec<String>) {
+        if arg.is_empty() {
+            return;
+        }
+
+        vec.push(arg.clone());
+        arg.clear();
+    }
+
     /// Try to build an [`Instruction`] from an [`OpCode`] and a set of arguments.
     ///
     /// # Arguments
diff --git a/redox-terminal/src/main.rs b/redox-terminal/src/main.rs
index 4dcb8dc..9ac0edd 100644
--- a/redox-terminal/src/main.rs
+++ b/redox-terminal/src/main.rs
@@ -66,7 +66,15 @@ fn main() {
         panic!("currently unsupported");
     }
 
-    let code = "section .text\r\npush 0\r\ncall :LABEL_1\r\nhlt\r\n:LABEL_1\r\nmov 0xdeadbeef, EAX\r\niret";
+    let code = "section .data
+    banana  db  \"apples\"
+    section .text
+    push 0
+    call :LABEL_1
+    hlt
+    :LABEL_1
+    mov 0xdeadbeef, EAX
+    iret";
     let mut compiler = Compiler::new();
     let data = compiler.compile_assembly(code, true);