From 9e22c31a2f95b7f91ad9fec0873f3fa21366c272 Mon Sep 17 00:00:00 2001
From: Ian Moffett <ian@mirocom.org>
Date: Sat, 14 Feb 2026 00:11:26 -0500
Subject: [PATCH] core: Add initial lexer and parser sources

Signed-off-by: Ian Moffett <ian@mirocom.org>
---
 core/lexer.c       | 93 ++++++++++++++++++++++++++++++++++++++++++++++
 core/parser.c      | 51 +++++++++++++++++++++++++
 core/rifle.c       |  7 ++++
 inc/rifle/lexer.h  | 18 +++++++++
 inc/rifle/parser.h | 13 +++++++
 5 files changed, 182 insertions(+)
 create mode 100644 core/lexer.c
 create mode 100644 core/parser.c
 create mode 100644 inc/rifle/lexer.h
 create mode 100644 inc/rifle/parser.h

diff --git a/core/lexer.c b/core/lexer.c
new file mode 100644
index 0000000..bacde50
--- /dev/null
+++ b/core/lexer.c
@@ -0,0 +1,93 @@
+#include <stdint.h>
+#include <stddef.h>
+#include <unistd.h>
+#include <stdbool.h>
+#include "rifle/lexer.h"
+#include "rifle/trace.h"
+
+/*
+ * Returns true if the given input character is a
+ * whitespace character
+ */
+static inline bool
+lexer_is_ws(char c)
+{
+    switch (c) {
+    case '\n':
+    case '\t':
+    case '\r':
+    case '\f':
+    case ' ':
+        return true;
+    }
+
+    return false;
+}
+
+/*
+ * Consume a single byte from the input source file
+ * while skipping any whitespace characters.
+ *
+ * @state: Compiler state
+ * @keep_ws: If true, preserve whitespace
+ *
+ * Returns the consumed character on success, otherwise
+ * '\0' upon failure.
+ */
+static char
+lexer_nom(struct rifle_state *state, bool keep_ws)
+{
+    char c;
+
+    if (state == NULL) {
+        return '\0';
+    }
+
+    while (read(state->in_fd, &c, 1) > 0) {
+        if (lexer_is_ws(c) && !keep_ws) {
+            continue;
+        }
+
+        return c;
+    }
+
+    return '\0';
+}
+
+int
+lexer_scan(struct rifle_state *state, struct token *res)
+{
+    char c;
+
+    if (state == NULL || res == NULL) {
+        return -1;
+    }
+
+    if ((c = lexer_nom(state, false)) == '\0') {
+        return -1;
+    }
+
+    switch (c) {
+    case '+':
+        res->type = TT_PLUS;
+        res->c = c;
+        return 0;
+    case '-':
+        res->type = TT_MINUS;
+        res->c = c;
+        return 0;
+    case '*':
+        res->type = TT_STAR;
+        res->c = c;
+        return 0;
+    case '/':
+        res->type = TT_SLASH;
+        res->c = c;
+        return 0;
+    default:
+        trace_error(state, "unexpected token '%c'\n", c);
+        break;
+    }
+
+    return -1;
+}
diff --git a/core/parser.c b/core/parser.c
new file mode 100644
index 0000000..b2bed0b
--- /dev/null
+++ b/core/parser.c
@@ -0,0 +1,51 @@
+#include <stdint.h>
+#include <stddef.h>
+#include "rifle/parser.h"
+#include "rifle/token.h"
+#include "rifle/trace.h"
+#include "rifle/lexer.h"
+#include "rifle/state.h"
+
+/* Symbolic token */
+#define symtok(tok) \
+    "[" tok "]"
+
+/* Quoted token */
+#define qtok(tok)   \
+    "'" tok "'"
+
+/* Convert token to string */
+#define tokstr1(tt)  \
+    toktab[(tt)]
+
+/* Convert token to string */
+#define tokstr(tok)  \
+    tokstr1((tok)->type)
+
+/*
+ * Table used to convert token constants into human
+ * readable strings
+ */
+static const char *toktab[] = {
+    [TT_NONE]   = symtok("none"),
+    [TT_PLUS]   = qtok("+"),
+    [TT_MINUS]  = qtok("-"),
+    [TT_STAR]   = qtok("*"),
+    [TT_SLASH]  = qtok("/")
+};
+
+int
+parser_parse(struct rifle_state *state)
+{
+    struct token tok;
+
+    if (state == NULL) {
+        return 0;
+    }
+
+    while (lexer_scan(state, &tok) == 0) {
+        trace_debug("got %s\n", tokstr(&tok));
+    }
+
+    return 0;
+}
diff --git a/core/rifle.c b/core/rifle.c
index 76ac466..c7460a8 100644
--- a/core/rifle.c
+++ b/core/rifle.c
@@ -2,6 +2,7 @@
 #include <unistd.h>
 #include <fcntl.h>
 #include "rifle/state.h"
+#include "rifle/parser.h"
 
 /* Compiler version */
 #define RIFLE_VERSION "0.0.1"
@@ -38,6 +39,11 @@ compile(const char *in_path)
         return -1;
     }
 
+    if (parser_parse(&state) < 0) {
+        rifle_state_destroy(&state);
+        return -1;
+    }
+
     rifle_state_destroy(&state);
     return 0;
 }
@@ -72,5 +78,6 @@ main(int argc, char **argv)
         printf("fatal: no input files\n");
         return -1;
     }
+
     return 0;
 }
diff --git a/inc/rifle/lexer.h b/inc/rifle/lexer.h
new file mode 100644
index 0000000..474cbb4
--- /dev/null
+++ b/inc/rifle/lexer.h
@@ -0,0 +1,18 @@
+#ifndef RIFLE_LEXER_H
+#define RIFLE_LEXER_H 1
+
+#include <stdint.h>
+#include "rifle/token.h"
+#include "rifle/state.h"
+
+/*
+ * Scan a single token from the source input
+ *
+ * @state: Compiler state
+ * @res:   Result is written here
+ *
+ * Returns zero on success
+ */
+int lexer_scan(struct rifle_state *state, struct token *res);
+
+#endif  /* !RIFLE_LEXER_H */
diff --git a/inc/rifle/parser.h b/inc/rifle/parser.h
new file mode 100644
index 0000000..0353c42
--- /dev/null
+++ b/inc/rifle/parser.h
@@ -0,0 +1,13 @@
+#ifndef RIFLE_PARSER_H
+#define RIFLE_PARSER_H 1
+
+#include "rifle/state.h"
+
+/*
+ * Begin parsing the input source file
+ *
+ * @state: Compiler state
+ */
+int parser_parse(struct rifle_state *state);
+
+#endif  /* !RIFLE_PARSER_H */