refactor(main): worked on string handling in C and other stuff

25 files changed, 196 insertions, 176 deletions

diff --git a/.gitignore b/.gitignore
index 00fe5ef..05a7b85 100644
--- a/.gitignore
+++ b/.gitignore
@@ -9,3 +9,4 @@ idea/*
 /helper.sh
 /tests
 /tags
+/.cache/clangd/index
diff --git a/Makefile b/Makefile
index f9c5906..0224a42 100644
--- a/Makefile
+++ b/Makefile
@@ -1,5 +1,5 @@
 BIN = build/engine
-SRC = source/engine/engine.c
+SRC = source/engine.c
 CC = clang
 CFLAGS = -Wall -Wextra -Wfloat-equal -Wswitch-default -Wswitch-enum \
          -Wno-unused-parameter -Wno-implicit-fallthrough -Wno-unused-function -g -Werror
diff --git a/source/storage/b_tree.c b/source/b_tree.c
index 6a0e76d..e42f709 100644
--- a/source/storage/b_tree.c
+++ b/source/b_tree.c
@@ -10,10 +10,10 @@ node_alloc(mem_arena *arena)
 }
 
 // NOTE(nasr): @return the index of of the found element
-internal i32
+internal s32
 node_find_pos(mem_arena *arena, string8 value)
 {
-    i32 i = 0;
+    s32 i = 0;
     while (i < n->key_count && str8_cmp(n->keys[i], k) < 0)
     {
         ++i;
@@ -22,10 +22,31 @@ node_find_pos(mem_arena *arena, string8 value)
     return i;
 }
 
+interal void
+b_tree_create(mem_arena *arena, b_tree *tree)
+{
+    tree->root = node_alloc(arena);
+    tree->root->leaf = 1;
+    tree->root->key_count = 0;
+}
+
 // NOTE(nasr): nodes that get  passed as parameters should've already been loaded into memory
 internal void
-b_tree_search(node *node)
+b_tree_search(b_tree_node *node, string8 key)
 {
+    s32 found_index = node_find_pos(node, key);
+
+     if (found_index < n->key_count && string_compare(n->keys[i], key) == 0)
+     {
+         return n->rows[i];
+     }
+     if (n->leaf)
+     {
+         return NULL;
+
+     }
+
+     return b_tree_search(n->children[i], key);
 
 
 }
diff --git a/source/storage/b_tree.h b/source/b_tree.h
index 10ad00d..0fc4d3c 100644
--- a/source/storage/b_tree.h
+++ b/source/b_tree.h
@@ -21,9 +21,9 @@ struct  b_tree_node
 
     // NOTE(nasr): reference count ::: check how many leaves are using this node
     // also not needed for now because we don't free individual node because of arena allocator
-    // i32 *refc;
+    // s32 *refc;
 
-    i32 key_count;
+    s32 key_count;
 
     b32 leaf;
 
diff --git a/source/base/base.h b/source/base/base.h
index ef23391..cf2a15f 100755
--- a/source/base/base.h
+++ b/source/base/base.h
@@ -52,22 +52,22 @@ typedef uint32_t u32;
 typedef uint16_t u16;
 typedef uint8_t  u8;
 
-typedef int8_t  i8;
-typedef int16_t i16;
-typedef int32_t i32;
-typedef int64_t i64;
+typedef int8_t  s8;
+typedef int16_t s16;
+typedef int32_t s32;
+typedef int64_t s64;
 
 typedef float  f32;
 typedef double f64;
 
-typedef i32 b32;
-typedef i16 b16;
+typedef s32 b32;
+typedef s16 b16;
 typedef u8  b8;
 
 typedef uintptr_t umm;
 typedef intptr_t  smm;
 
-#define TRUE (0 == 0)
-#define FALSE (0 != 0)
+#define TRUE (1 == 1)
+#define FALSE (1 != 1)
 
 #endif
diff --git a/source/base/base_include.h b/source/base/base_include.h
index 40ae5ea..07856d0 100755
--- a/source/base/base_include.h
+++ b/source/base/base_include.h
@@ -4,6 +4,7 @@
 #include <dirent.h>
 #include <sys/mman.h>
 #include <sys/stat.h>
+#include <sys/syscall.h>
 #include <fcntl.h>
 #include <stdint.h>
 #include <stddef.h>
@@ -16,6 +17,7 @@
 #include "base_stack.h"
 #include "base_test.h"
 #include "base_string.h"
+#include "base_io.h"
 #include "base_os.h"
 
 #ifdef BASE_UNITY
diff --git a/source/base/base_io.h b/source/base/base_io.h
index ece4d7c..85fedc7 100644
--- a/source/base/base_io.h
+++ b/source/base/base_io.h
@@ -1,11 +1,38 @@
 #ifndef BASE_IO_H
 #define BASE_IO_H
 
+#define STDIN_FD  0
+#define STDOUT_FD 1
+#define STDERR_FD 2
+
+internal s64
+os_write(s32 fd, void const *buf, u64 count) 
+{
+    return syscall(SYS_write, fd, buf, count);
+}
+
+internal s64
+os_read(s32 fd, void *buf, u64 count)
+{
+    return syscall(SYS_read, fd, buf, count);
+}
+
 internal void
-input_read()
+print_s8(string8 s)
 {
+    os_write(STDOUT_FILENO, s.data, s.size);
+}
 
+internal void
+print(const char *str)
+{
+    s32 len = 0;
+    while (str[len]) len++;
+    os_write(STDOUT_FILENO, str, len);
 
 }
 
+#define Os_read(buffer, buffer_count) 
+
+
 #endif /* BASE_IO_H */
diff --git a/source/base/base_os.h b/source/base/base_os.h
index 23587c6..abe8628 100644
--- a/source/base/base_os.h
+++ b/source/base/base_os.h
@@ -1,21 +1,13 @@
 #ifndef BASE_OS_H
 #define BASE_OS_H
 
-internal void
-print(const char *str)
-{
-    i32 len = 0;
-    while (str[len]) len++;
-    write(STDOUT_FILENO, str, len);
-}
-
 internal string8
 load_file(const char *path)
 {
     string8 result = {0};
     struct stat sbuf = {0};
 
-    i32 file = open(path, O_RDONLY);
+    s32 file = open(path, O_RDONLY);
     if(file == -1) return result;
 
     if(fstat(file, &sbuf) == -1)
diff --git a/source/base/base_string.h b/source/base/base_string.h
index 189b38a..4e0d923 100644
--- a/source/base/base_string.h
+++ b/source/base/base_string.h
@@ -1,13 +1,18 @@
 #ifndef BASE_STRING_H
 #define BASE_STRING_H
 
-#include <string.h>
-
 #define StringLit(string) \
     (string8){ .data = (u8 *)(string), .size = (sizeof(string) - 1) }
 
- #define PushString(arena, size) \
-    (string8){ (u8 *)PushArray((arena), u8, (size)), (u64)(size) }
+#define PushString(arena, size) \
+    ({ \
+        string8 *result = PushStruct((arena), string8); \
+        result->data = PushArray((arena), u8, (size)); \
+        result->size = (u64)(size); \
+        result; \
+    })
+
+#define String8Cast(literal, literal_count) ( string8 ){( u8 * )( literal ),( u64 )( literal_count ) } 
 
 #define StringFmt "%.*s"
 #define ULongFmt  "%lu"
@@ -50,10 +55,8 @@ string8_append_char(string8 *buf, u8 c)
 read_only global_variable
 string8 nil_string =
 {
-
     .data = NULL,
     .size = 0,
-
 };
 
 #endif /* BASE_STRING_H */
diff --git a/source/base/base_test.h b/source/base/base_test.h
index 412797b..7619cfb 100644
--- a/source/base/base_test.h
+++ b/source/base/base_test.h
@@ -9,7 +9,7 @@
 #define LEN(s) (sizeof(s) - 1)
 
 internal void
-write_int(i32 num)
+write_int(s32 num)
 {
 
     if (num < 0)
diff --git a/source/csv_parser.c b/source/csv_parser.c
new file mode 100644
index 0000000..048e089
--- /dev/null
+++ b/source/csv_parser.c
@@ -0,0 +1,32 @@
+internal void
+strip_new_line(string8 buffer)
+{
+
+    for (u64 index = 0; index < buffer.size; index++)
+        {
+
+        }
+
+    return;
+
+}
+
+internal void
+read_csv(string8 buffer)
+{
+    printf("\nsize:%lu\ndata %s\n", buffer.size, buffer.data);
+
+}
+
+internal csv_table *
+parse_csv(token *tokens, csv_table *table)
+{
+
+
+
+
+
+
+
+    return NULL;
+}
diff --git a/source/storage/csv_reader.h b/source/csv_parser.h
index 36e07a4..1aa96b9 100644
--- a/source/storage/csv_reader.h
+++ b/source/csv_parser.h
@@ -1,12 +1,12 @@
-#ifndef CSV_READER_H
-#define CSV_READER_H
+#ifndef CSV_PARSER_H
+#define CSV_PARSER_H
 
 typedef struct csv_row csv_row;
 struct csv_row
 {
     // array of size col_count, points into mmap buffer
     string8 *fields;
-    i32      count;
+    s32      count;
 };
 
 typedef struct csv_table csv_table;
@@ -16,8 +16,8 @@ struct csv_table
     // all data rows
     string8  *headers;
     csv_row  *rows;
-    i32       col_count;
-    i32       row_count;
+    s32       col_count;
+    s32       row_count;
 };
 
 read_only global_variable
@@ -27,8 +27,6 @@ csv_row  nil_csv_row =
     .count      = 0,
 };
 
-
-
 read_only global_variable
 csv_table nil_csv_table =
 {
@@ -39,5 +37,4 @@ csv_table nil_csv_table =
 };
 
 
-
-#endif /* CSV_READER_H */
+#endif /* CSV_PARSER_H */
diff --git a/source/engine.c b/source/engine.c
new file mode 100644
index 0000000..8f701e3
--- /dev/null
+++ b/source/engine.c
@@ -0,0 +1,46 @@
+#define BASE_UNITY
+#include "base/base_include.h"
+
+#include <stdio.h>
+
+#include "csv_parser.h"
+#include "lexer.h"
+
+#include "lexer.c"
+#include "csv_parser.c"
+
+int main(int c, char **v)
+{
+    if(c < 2) return -999;
+
+    b32 running = 0;
+
+    mem_arena *global_arena = arena_create(MiB(30));
+    csv_table *global_table = PushStruct(global_arena, csv_table);
+
+    string8 buffer = load_file(v[1]);
+
+    print("database engine in nasr");
+
+    for(;;)
+    {
+        if (running)
+        {
+            // TODO(nasr): check for return key
+            {
+                u8 line_buffer[256] = {};
+
+                s64 codepoint = os_read(STDIN_FD, line_buffer, 256);
+                unused(codepoint);
+
+                read_csv(buffer);
+                token *tokens = tokenize_csv(buffer, global_arena);
+                global_table = parse_csv(tokens, global_table);
+            }
+
+            sleep(1);
+        }
+    }
+
+    return 0;
+}
diff --git a/source/engine/engine.c b/source/engine/engine.c
deleted file mode 100644
index 18b52be..0000000
--- a/source/engine/engine.c
+++ /dev/null
@@ -1,37 +0,0 @@
-#define BASE_UNITY
-#include "../base/base_include.h"
-
-#include <stdio.h>
-
-
-
-#include "../parser/parser.h"
-#include "../parser/parser.c"
-
-#include "../repl/repl.h"
-#include "../repl/repl.c"
-
-#include "../storage/csv_reader.h"
-#include "../storage/csv_reader.c"
-
-#include "../lexer/lexer.h"
-#include "../lexer/lexer.c"
-
-
-
-int main(int c, char **v)
-{
-    if(c < 2) return -999;
-
-    mem_arena *global_arena = arena_create(MiB(30));
-    csv_table *global_table = PushStruct(global_arena, csv_table);
-
-    string8 buffer = load_file(v[1]);
-    // read_csv(buffer);
-    tokenize_csv(buffer, global_table, global_arena);
-
-
-    return 0;
-}
-
-
diff --git a/source/lexer/lexer.c b/source/lexer.c
index 948afd0..c84a831 100644
--- a/source/lexer/lexer.c
+++ b/source/lexer.c
@@ -32,19 +32,20 @@ is_delimiter(u8 point)
 }
 
 internal token *
-tokenize_csv(string8 buffer, csv_table *global_table, mem_arena *arena)
+tokenize_csv(string8 buffer, mem_arena *arena)
 {
-    i32 count = 0;
-    string8 **tokens = PushArray(arena, string8 *, buffer.size / 10);
-    b32 first_line = 1;
+    s32 count = 0;
+    string8 **tokens = PushString(arena, buffer.size);
+
+    b32 FL = TRUE;
 
     if(buffer.size < 0) return NULL;
-    for(i32 index = 0;
+    for(s32 index = 0;
         buffer.data[index] != '\0';
         ++index)
     {
         csv_row *row = PushStruct(arena, csv_row);
-        string8 token = {0};
+        token *tok = PushStruct(arena, token);
 
         u8 point = buffer.data[index];
 
@@ -53,28 +54,25 @@ tokenize_csv(string8 buffer, csv_table *global_table, mem_arena *arena)
 
         unused(row);
 
-        switch (point)
+        switch(point)
         {
-            case '\n':
-                {
-                    first_line = -1;
-                    break;
-                }
-            case ',':
+            case('\n'):
                 {
-                    end = start - 1;
 
-                    if (first_line)
+                    if(FL)
                     {
-                        global_table->headers = &token;
-                        ++global_table->headers;
-                        break;
+                        FL = FALSE;
+                        tok->flags |= END_FL;
                     }
-                    else
-                    {
 
-                        break;
-                    }
+                    break;
+
+                }
+
+            case(','):
+                {
+                    end = start - 1;
+                    break;
                 }
 
             default:
@@ -85,16 +83,12 @@ tokenize_csv(string8 buffer, csv_table *global_table, mem_arena *arena)
                 }
         }
 
-        token = (string8){
-            .data = start,
-            .size = end - start,
-        };
+        token->lexeme = String8Cast(start, end - start);
 
-        **tokens = token;
-        ++*tokens;
-    }
+        *tokens = token;
+        ++tokens;
 
-    printf("%d", count);
 
-    return NULL;
+        return NULL;
+    }
 }
diff --git a/source/lexer/lexer.h b/source/lexer.h
index 86f8427..7bafc0d 100644
--- a/source/lexer/lexer.h
+++ b/source/lexer.h
@@ -1,6 +1,14 @@
 #ifndef ENGINE_LEXER_H
 #define ENGINE_LEXER_H
 
+typedef enum token_flags token_flags;
+enum token_flags 
+{
+    START_FL    = 1 << 1,
+    END_FL      = 1 << 2,
+};
+
+
 typedef enum token_type token_type;
 enum token_type
 {
@@ -16,8 +24,9 @@ struct token
 {
     string8 lexeme;
     token_type type;
-
+    token_flags flags;
 };
 
 
+
 #endif /* ENGINE_LEXER_H */
diff --git a/source/parser/parser.c b/source/parser/parser.c
deleted file mode 100644
index 4c57345..0000000
--- a/source/parser/parser.c
+++ /dev/null
@@ -1,16 +0,0 @@
-#ifndef ENGINE_REPL_H
-#define ENGINE_REPL_H
-
-typedef struct node node;
-struct node
-{
-
-};
-
-typedef struct btree btree;
-struct btree
-{
-
-};
-
-#endif /* ENGINE_H */
diff --git a/source/parser/parser.h b/source/parser/parser.h
deleted file mode 100644
index 4c57345..0000000
--- a/source/parser/parser.h
+++ /dev/null
@@ -1,16 +0,0 @@
-#ifndef ENGINE_REPL_H
-#define ENGINE_REPL_H
-
-typedef struct node node;
-struct node
-{
-
-};
-
-typedef struct btree btree;
-struct btree
-{
-
-};
-
-#endif /* ENGINE_H */
diff --git a/source/query/query.c b/source/query.c
index e69de29..e69de29 100644
--- a/source/query/query.c
+++ b/source/query.c
diff --git a/source/query/query.h b/source/query.h
index e69de29..e69de29 100644
--- a/source/query/query.h
+++ b/source/query.h
diff --git a/source/repl/repl.c b/source/repl/repl.c
deleted file mode 100644
index dd289d8..0000000
--- a/source/repl/repl.c
+++ /dev/null
@@ -1,12 +0,0 @@
-internal void
-init_repl()
-{
-    for(;;)
-    {
-        print("reading user input...");
-        // TODO(nasr): design a repl system
-
-        sleep(1);
-    }
-
-}
diff --git a/source/repl/repl.h b/source/repl/repl.h
deleted file mode 100644
index 4c57345..0000000
--- a/source/repl/repl.h
+++ /dev/null
@@ -1,16 +0,0 @@
-#ifndef ENGINE_REPL_H
-#define ENGINE_REPL_H
-
-typedef struct node node;
-struct node
-{
-
-};
-
-typedef struct btree btree;
-struct btree
-{
-
-};
-
-#endif /* ENGINE_H */
diff --git a/source/storage/csv_reader.c b/source/storage/csv_reader.c
deleted file mode 100644
index 2fcbe04..0000000
--- a/source/storage/csv_reader.c
+++ /dev/null
@@ -1,7 +0,0 @@
-internal void
-read_csv(string8 buffer)
-{
-    printf("\nsize:%lu\ndata %s\n", buffer.size, buffer.data);
-
-}
-
diff --git a/source/storage/table.c b/source/storage/table.c
deleted file mode 100644
index e69de29..0000000
--- a/source/storage/table.c
+++ /dev/null
diff --git a/source/storage/table.h b/source/storage/table.h
deleted file mode 100644
index e69de29..0000000
--- a/source/storage/table.h
+++ /dev/null