From 8e0073a1e60325507c9776ba41b469ed29918f46 Mon Sep 17 00:00:00 2001
From: Yehowshua Immanuel <yehowshua@yehowshuaimmanuel.com>
Date: Mon, 9 Dec 2024 11:34:26 -0500
Subject: [PATCH] RTLIL parser presumably finished except for comment parser

---
 README.md                 |  7 +++++-
 TODO.md                   |  1 +
 src/RTLILParser/AST.hs    | 30 +++++++++++++++--------
 src/RTLILParser/Parser.hs | 50 +++++++++++++++++++++++++++++++++------
 4 files changed, 70 insertions(+), 18 deletions(-)

diff --git a/README.md b/README.md
index fb802d3..1ec2597 100644
--- a/README.md
+++ b/README.md
@@ -70,4 +70,9 @@ $ rtlil-parse
    I have yet to evaluate the implications of how this affects
    multi-clock domain circuits and their associated primitives such
    as asynchronous FIFOs, but I plan to make sure simulation of such
-   circuits is possible and correct.
\ No newline at end of file
+   circuits is possible and correct.
+
+# Lessons Learned
+ - Should have written parser to be token based where after consuming
+   and capturing token, we consume and discard all following whitespaces
+   as well as comments...
\ No newline at end of file
diff --git a/TODO.md b/TODO.md
index 656df2a..16cf4e7 100644
--- a/TODO.md
+++ b/TODO.md
@@ -42,6 +42,7 @@
  - [x] Remove all instances of `_ <-`
  - [ ] Module
  - [ ] Remove weird GHC imports
+ - [x] Are the `try` statements in `pWireOption` correctly constructed?
  - [ ] Consider the very weird case where the process body has nothing,
        thus, `pEolAndAdvanceToNextNonWs` may never get invoked in any of
        the sub-parsers encapsulated in `pProcessBody`. Do we need to 
diff --git a/src/RTLILParser/AST.hs b/src/RTLILParser/AST.hs
index 8a6d3a7..56d0461 100644
--- a/src/RTLILParser/AST.hs
+++ b/src/RTLILParser/AST.hs
@@ -5,11 +5,15 @@ module RTLILParser.AST (
     -- Values
     Value(..),
 
+    -- File
+    File(..),
+
     -- Autoindex statements
     AutoIdxStmt(..),
 
     -- Module
-    ParamStmt(..), Constant(..),
+    Module(..), ModuleStmt(..), ModuleBody(..),
+    ModuleBodyVariant(..), ParamStmt(..), Constant(..),
 
     -- Attribute statements
     AttrStmt(..),
@@ -68,20 +72,26 @@ data Value = Value
 -- strings
 -- comments
 -- file
+data File           = File AutoIdxStmt [Module] deriving (Show)
 
 -- Autoindex statements
 data AutoIdxStmt    = AutoIdxStmt   Int     deriving (Show)
 
 -- Module
-data ParamStmt = ParamStmt
-    { paramId       :: Id
-    , paramConstant :: Maybe Constant
-    }
-    deriving (Show)
-data Constant = ConstantValue   Value
-              | ConstantInteger Int
-              | ConstantString  String
-              deriving (Show)
+data Module         = Module ModuleStmt [AttrStmt] ModuleBody deriving (Show)
+data ModuleStmt     = ModuleStmt Id deriving (Show)
+data ModuleBody     = ModuleBody [ModuleBodyVariant] deriving (Show)
+data ModuleBodyVariant  = ModuleBodyParamStmt ParamStmt
+                        | ModuleBodyWire Wire
+                        | ModuleBodyMemory Memory
+                        | ModuleBodyCell Cell
+                        | ModuleBodyProcess Process
+                        deriving (Show)
+data ParamStmt      = ParamStmt Id (Maybe Constant) deriving (Show)
+data Constant       = ConstantValue   Value
+                    | ConstantInteger Int
+                    | ConstantString  String
+                    deriving (Show)
 
 -- Attribute statements
 data AttrStmt       = AttrStmt      Id Constant deriving (Show)
diff --git a/src/RTLILParser/Parser.hs b/src/RTLILParser/Parser.hs
index 9a1e318..94f6dcd 100644
--- a/src/RTLILParser/Parser.hs
+++ b/src/RTLILParser/Parser.hs
@@ -11,11 +11,15 @@ import RTLILParser.AST (
     -- Values
     Value(..),
 
+    -- File
+    File(..),
+
     -- Autoindex statements
     AutoIdxStmt(..),
 
     -- Module
-    ParamStmt(..), Constant(..),
+    Module(..), ModuleStmt(..), ModuleBody(..),
+    ModuleBodyVariant(..), ParamStmt(..), Constant(..),
 
     -- Attribute statements
     AttrStmt(..),
@@ -106,6 +110,10 @@ pString =
 
 -- comments
 -- file
+pFile :: Parser File
+pFile = File
+    <$> pAutoIdxStmt
+    <*> many pModule
 
 -- Autoindex statements
 pAutoIdxStmt :: Parser AutoIdxStmt
@@ -114,9 +122,37 @@ pAutoIdxStmt = AutoIdxStmt
          pInteger <* pEolAndAdvanceToNextNonWs)
 
 -- Module
-pModuleStmt :: Parser Id
-pModuleStmt = string "module" *> pWs *> pId <*
-              pEolAndAdvanceToNextNonWs
+pModule :: Parser Module
+pModule = do
+    attrs       <- many pAttrStmt
+    moduleStmt  <- pModuleStmt
+    moduleBody  <- pModuleBody
+    pModuleEndStmt
+    return $ Module moduleStmt attrs moduleBody
+
+pModuleStmt :: Parser ModuleStmt
+pModuleStmt = ModuleStmt 
+    <$> (string "module" *> pWs *> 
+         pId <* pEolAndAdvanceToNextNonWs)
+
+pModuleBody :: Parser ModuleBody
+pModuleBody = ModuleBody
+    <$> many pModuleBodyVariant
+
+pModuleBodyVariant :: Parser ModuleBodyVariant
+pModuleBodyVariant = 
+    -- `pWire`, `pMemory`, `pCell`, `pProcess` all
+    -- start by parsing attribute statements, so we
+    -- need backtracking since we can't determin which
+    -- parser will succeed based on the first character
+    -- we encounter alone. `pParamStmt` technically doesn't
+    -- need to be prefixed by `try`, so that is a stylistic
+    -- choice.
+    try (ModuleBodyParamStmt    <$> pParamStmt) <|>
+    try (ModuleBodyWire         <$> pWire     ) <|>
+    try (ModuleBodyMemory       <$> pMemory   ) <|>
+    try (ModuleBodyCell         <$> pCell     ) <|>
+        (ModuleBodyProcess      <$> pProcess  )
 
 pParamStmt :: Parser ParamStmt
 pParamStmt = ParamStmt
@@ -126,9 +162,9 @@ pParamStmt = ParamStmt
 
 pConstant :: Parser Constant
 pConstant =
-    try (ConstantValue <$> pValue)
-    <|> (ConstantInteger <$> pInteger)
-    <|> (ConstantString <$> pString)
+    try (ConstantValue      <$> pValue  )
+    <|> (ConstantInteger    <$> pInteger)
+    <|> (ConstantString     <$> pString )
 
 pModuleEndStmt :: Parser ()
 pModuleEndStmt = void (string "end")