From 30fd15dec535007340ffb0d8b7617a7b0d76ef3a Mon Sep 17 00:00:00 2001
From: Ákos Kőrösi <korakos99@gmail.com>
Date: Tue, 12 May 2026 00:51:29 +0200
Subject: start outlining lexing infrastructure

---
 main.c | 147 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 147 insertions(+)
 create mode 100644 main.c

diff --git a/main.c b/main.c
new file mode 100644
index 0000000..0052ad9
--- /dev/null
+++ b/main.c
@@ -0,0 +1,147 @@
+#include <stddef.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdbool.h>
+#include <stdio.h>
+
+
+#define DECLARE_VEC(T, Name)                                        \
+    typedef struct {                                                \
+        T* items;                                                   \
+        size_t count;                                               \
+        size_t cap;                                                 \
+        size_t incr_step;                                           \
+    } Name;                                                         \
+    static inline Name make_##Name(size_t step) {                   \
+        return (Name) {                                             \
+            .items = malloc(step* sizeof(T)),                       \
+            .count = 0,                                             \
+            .cap = step,                                            \
+            .incr_step = step,                                      \
+        };                                                          \
+    }                                                               \
+    static inline void push_to_##Name(Name* vec, T item) {          \
+        if (vec->count == vec->cap) {                               \
+            vec->cap += vec->incr_step;                             \
+            vec->items = realloc(                                   \
+                vec->items,                                         \
+                vec->cap * sizeof(T)                                \
+            );                                                      \
+        }                                                           \
+        vec->items[vec->count++] = item;                            \
+    }                                                               \
+
+
+DECLARE_VEC(char, String);
+
+void append_str_to_String(String* target, char* src_str) {
+    size_t required_size = 
+        ((target->count + strlen(src_str)) / target->incr_step + 1) * target->incr_step; 
+    if (required_size > target->cap) {
+        target->cap = required_size;
+        target->items = realloc(target->items, required_size * sizeof(char));
+    }
+    memcpy(target->items + target->count, src_str, strlen(src_str));
+    target->count += strlen(src_str);
+}
+
+
+#define DECLARE_MAYBE(T)                                            \
+    typedef struct {                                                \
+        bool some;                                                  \
+        T value;                                                    \
+    } Maybe##T;                                                     \
+
+
+
+typedef enum {
+    UN_Q,
+    SINGLE_Q,
+    DOUBLE_Q,
+} QuotationMode;
+
+typedef struct Region Region;
+
+typedef enum {
+    LITERAL,
+    CMD_SUB,
+    ARITHM,
+    PARAM,
+} ChunkKind;
+
+
+typedef struct {
+    ChunkKind kind;
+    QuotationMode mode;
+
+    union {
+        char* lit_str;
+        Region* sub_cmd_region;
+        Region* arithm_expr_region;
+        Region* param_region;
+    } as;
+} Chunk;
+
+
+DECLARE_VEC(Chunk, ChunkVec)
+
+typedef enum {
+    TOK_WORD,
+    TOK_SPECIAL,
+} TokenKind;
+
+typedef enum {
+    PIPE,
+    WRITE_REDIR_TRUNC,
+    WRITE_REDIR_APPEND,
+    READ_REDIR,
+    EOL,
+} SpecialToken;
+
+typedef struct {
+    TokenKind kind;
+
+    union {
+        ChunkVec word;
+        SpecialToken spec; 
+    } as;
+} Token;
+    
+
+DECLARE_VEC(Token, TokenVec)
+
+DECLARE_MAYBE(SpecialToken)  
+
+struct Region {
+    QuotationMode mode;
+    TokenVec tokens; 
+};
+
+DECLARE_MAYBE(Chunk);
+
+
+MaybeChunk unquoted_next_chunk(char* input, size_t* pos) {
+    String chunk;
+    char curr_c;
+    for (;(curr_c = input[*pos], curr_c != '$' && curr_c != '\'' && curr_c != '\"' && curr_c != '\0'); pos++) {
+        exit(1);
+    }
+    exit(1);
+}
+
+
+TokenVec lex_region(char* input, size_t* pos, char until) {
+    TokenVec tokens = make_TokenVec(256);    
+
+    while (input[*pos] != until) {
+        while (input[*pos] == ' ') (*pos)++;
+        
+
+    }
+    return tokens;
+}
+
+TokenVec lex(char* input) {
+    size_t pos = 0;
+    return lex_region(input, &pos, '\0');
+}
-- 
cgit v1.2.3