First draft of construct tree (infinite loop bug for now)

ChillerDragon · Aug 3, 2024 · 4d13f37 · 4d13f37
1 parent 379a586
commit 4d13f37
Show file tree

Hide file tree

Showing 5 changed files with 203 additions and 1 deletion.
diff --git a/src/bss/huffman.asm b/src/bss/huffman.asm
@@ -1,6 +1,12 @@
 ; m_aNodes array of CNODE structs
 huff_nodes resb HUFF_CNODE_SIZE * HUFFMAN_MAX_NODES
 
+; aNodesLeftStorage[HUFFMAN_MAX_SYMBOLS];
+huff_nodes_left_storage resb HUFF_CNODE_SIZE * HUFFMAN_MAX_SYMBOLS
+
+; *apNodesLeft[HUFFMAN_MAX_SYMBOLS];
+huff_nodes_left resb 4 * HUFFMAN_MAX_SYMBOLS
+
 ; m_apDecodeLut array of CNODE structs
 huff_decode_lut resb HUFFMAN_LUTSIZE * 8
 

diff --git a/src/data/huffman.asm b/src/data/huffman.asm
@@ -33,3 +33,19 @@ HUFF_CCONSTRUCTION_NODE_NODE_ID_OFFSET equ 0
 HUFF_CCONSTRUCTION_NODE_FREQUENCY_OFFSET equ 2
 HUFF_CCONSTRUCTION_NODE_SIZE equ 6
 
+HUFF_FREQ_TABLE dd 1073741824,4545,2657,431,1950,919,444,482,2244,617,838,542,715,1814,304,240,754,212,647,186, \
+	283,131,146,166,543,164,167,136,179,859,363,113,157,154,204,108,137,180,202,176, \
+	872,404,168,134,151,111,113,109,120,126,129,100,41,20,16,22,18,18,17,19, \
+	16,37,13,21,362,166,99,78,95,88,81,70,83,284,91,187,77,68,52,68, \
+	59,66,61,638,71,157,50,46,69,43,11,24,13,19,10,12,12,20,14,9, \
+	20,20,10,10,15,15,12,12,7,19,15,14,13,18,35,19,17,14,8,5, \
+	15,17,9,15,14,18,8,10,2173,134,157,68,188,60,170,60,194,62,175,71, \
+	148,67,167,78,211,67,156,69,1674,90,174,53,147,89,181,51,174,63,163,80, \
+	167,94,128,122,223,153,218,77,200,110,190,73,174,69,145,66,277,143,141,60, \
+	136,53,180,57,142,57,158,61,166,112,152,92,26,22,21,28,20,26,30,21, \
+	32,27,20,17,23,21,30,22,22,21,27,25,17,27,23,18,39,26,15,21, \
+	12,18,18,27,20,18,15,19,11,17,33,12,18,15,19,18,16,26,17,18, \
+	9,10,25,22,22,17,20,16,6,16,15,20,14,18,24,335,1517
+HUFF_FREQ_TABLE_SIZE equ ($ - HUFF_FREQ_TABLE) / 4
+; print_int32_array HUFF_FREQ_TABLE, HUFF_FREQ_TABLE_SIZE
+
diff --git a/src/huffman/construct_tree.asm b/src/huffman/construct_tree.asm
@@ -11,5 +11,179 @@
 ; huff_num_nodes resb 4
 
 _huff_construct_tree:
+    push_registers
+    mov rbp, rsp
+
+    sub rsp, 4
+    ; int NumNodesLeft = HUFFMAN_MAX_SYMBOLS
+    mov dword [rbp-4], HUFFMAN_MAX_SYMBOLS
+
+    mov rcx, 0
+    ; for(int i = 0; i < HUFFMAN_MAX_SYMBOLS; i++)
+._huff_construct_tree_for_loop:
+    cmp rcx, HUFFMAN_MAX_SYMBOLS
+    jge ._huff_construct_tree_for_loop_end
+
+    mov rbx, rcx
+    imul rbx, HUFF_CNODE_SIZE
+    ; m_aNodes[i].m_NumBits = 0xFFFFFFFF;
+    mov dword [huff_nodes + rbx + HUFF_CNODE_NUM_BITS_OFFSET], 0xFFFF_FFFF
+    ; m_aNodes[i].m_Symbol = i;
+    mov byte [huff_nodes + rbx + HUFF_CNODE_SYMBOL_OFFSET], cl
+    mov word [huff_nodes + rbx + HUFF_CNODE_LEAF_0_OFFSET], 0xffff
+    mov word [huff_nodes + rbx + HUFF_CNODE_LEAF_1_OFFSET], 0xffff
+
+    ; get i offset into aNodesLeftStorage[i]
+    mov rbx, rcx
+    imul rbx, HUFF_CCONSTRUCTION_NODE_SIZE
+
+    ; if(i == HUFFMAN_EOF_SYMBOL)
+    cmp rcx, HUFFMAN_EOF_SYMBOL
+    jne ._huff_construct_tree_i_not_eof
+    mov dword [huff_nodes_left_storage + rbx + HUFF_CCONSTRUCTION_NODE_FREQUENCY_OFFSET], 1
+    jmp ._huff_construct_tree_for_loop_if_end
+._huff_construct_tree_i_not_eof:
+    ; eax = pFrequencies[i]
+    ; r9 = i
+    mov r9, rcx
+    ; 4 is the size of int / unsigned int
+    ; which is the element size in the freq table array
+    imul r9, 4
+    mov rax, 0
+    mov eax, [HUFF_FREQ_TABLE + r9]
+    mov dword [huff_nodes_left_storage + rbx + HUFF_CCONSTRUCTION_NODE_FREQUENCY_OFFSET], eax
+._huff_construct_tree_for_loop_if_end:
+
+    ; aNodesLeftStorage[i].m_NodeId = i;
+    ; rbx is still the correct [i] offset
+    mov word [huff_nodes_left_storage + rbx + HUFF_CCONSTRUCTION_NODE_NODE_ID_OFFSET], cx
+
+    ; rax = &aNodesLeftStorage[i];
+    lea rax, [huff_nodes_left_storage + rbx]
+
+    mov rbx, rcx
+    ; 8 is the size of a pointer on 64 bit machines
+    ; huff_nodes_left is an array of pointers
+    imul rbx, 8
+    mov qword [huff_nodes_left + rbx], rax
+
+    inc rcx
+    jmp ._huff_construct_tree_for_loop
+._huff_construct_tree_for_loop_end:
+
+    ; m_NumNodes = HUFFMAN_MAX_SYMBOLS;
+    mov dword [huff_num_nodes], HUFFMAN_MAX_SYMBOLS
+
+    ; while(NumNodesLeft > 1)
+._huff_construct_tree_while_loop:
+    cmp dword [rbp-4], 1
+    jle ._huff_construct_tree_while_loop_end
+
+    ;  rax = **ppList
+    mov rax, huff_nodes_left
+    ;  rdi = Size
+    ; rbp-4 is NumNodesLeft
+    mov rdi, 0
+    mov edi, [rbp-4]
+    call _huff_bubble_sort
+
+    ; m_aNodes[m_NumNodes].m_NumBits = 0;
+    mov rbx, [rbp-4]
+    imul rbx, HUFF_CNODE_SIZE
+    mov dword [huff_nodes + rbx + HUFF_CNODE_NUM_BITS_OFFSET], 0
+
+    ; r9 is the offset NumNodesLeft into apNodesLeft
+    mov r9, 0
+    mov r9d, [rbp-4]
+    ; apNodesLeft is an array of pointers
+    ; so the offset is 8 (pointer size)
+    imul r9, 8
+
+    ; m_aNodes[m_NumNodes].m_aLeafs[0] = apNodesLeft[NumNodesLeft-1]->m_NodeId;
+    ; rax = apNodesLeft[NumNodesLeft-1]->m_NodeId;
+    ;                 NumNodesLeft - 1
+    mov rsi, [huff_nodes_left + r9 - (1 * 8)] ; get pointer from array
+    ; get 2 byte field from struct at dereferenced value
+    ; ax = unsigned int NodeId
+    mov rax, 0
+    mov ax, [rsi + HUFF_CCONSTRUCTION_NODE_NODE_ID_OFFSET]
+    mov word [huff_nodes + rbx + HUFF_CNODE_LEAF_0_OFFSET], ax
+
+    ; m_aNodes[m_NumNodes].m_aLeafs[1] = apNodesLeft[NumNodesLeft-2]->m_NodeId;
+    ; rax = apNodesLeft[NumNodesLeft-2]->m_NodeId;
+    ;                 NumNodesLeft - 2
+    mov rsi, [huff_nodes_left + r9 - (2 * 8)] ; get pointer from array
+    ; get 2 byte field from struct at dereferenced value
+    ; ax = unsigned int NodeId
+    mov rax, 0
+    mov ax, [rsi + HUFF_CCONSTRUCTION_NODE_NODE_ID_OFFSET]
+    mov word [huff_nodes + rbx + HUFF_CNODE_LEAF_1_OFFSET], ax
+
+    ; apNodesLeft[NumNodesLeft-2]->m_NodeId = m_NumNodes;
+    mov rax, [huff_nodes_left + r9 - (2 * 8)] ; get pointer from array
+    add rax, HUFF_CCONSTRUCTION_NODE_NODE_ID_OFFSET ; ->m_NodeId
+    mov si, [huff_num_nodes]
+    mov word [rax], si
+
+
+    ; apNodesLeft[NumNodesLeft-2]->m_Frequency = apNodesLeft[NumNodesLeft-1]->m_Frequency + apNodesLeft[NumNodesLeft-2]->m_Frequency;
+
+    ; r10 = apNodesLeft[NumNodesLeft-1]->m_Frequency
+    ;                 NumNodesLeft - 1
+    mov rax, [huff_nodes_left + r9 - (1 * 8)] ; get pointer from array
+    ; get 4 byte field from struct at dereferenced value
+    ; r10d = int m_Frequency
+    mov r10, 0
+    mov r10d, [rax + HUFF_CCONSTRUCTION_NODE_NODE_ID_OFFSET]
+
+    ; r11 = apNodesLeft[NumNodesLeft-2]->m_Frequency
+    ;                 NumNodesLeft - 2
+    mov rax, [huff_nodes_left + r9 - (2 * 8)] ; get pointer from array
+    ; get 4 byte field from struct at dereferenced value
+    ; r10d = int m_Frequency
+    mov r11, 0
+    mov r11d, [rax + HUFF_CCONSTRUCTION_NODE_NODE_ID_OFFSET]
+
+    ; apNodesLeft[NumNodesLeft-1]->m_Frequency + apNodesLeft[NumNodesLeft-2]->m_Frequency;
+    add r10, r11
+
+    ; apNodesLeft[NumNodesLeft-2]->m_Frequency = r10
+    mov rax, [huff_nodes_left + r9 - (2 * 8)] ; get pointer from array
+    add rax, HUFF_CCONSTRUCTION_NODE_FREQUENCY_OFFSET ; ->m_Frequency
+    mov dword [rax], r10d
+
+    ; m_NumNodes++
+    mov rax, 0
+    mov eax, [huff_num_nodes]
+    inc eax
+    mov [huff_num_nodes], eax
+
+    ; NumNodesLeft--
+    dec dword [rbp-4]
+
+    jmp ._huff_construct_tree_while_loop
+._huff_construct_tree_while_loop_end:
+
+    ; ebx = m_NumNodes - 1
+    mov rbx, 0
+    mov ebx, [huff_num_nodes]
+    dec ebx
+
+    ; rax = &m_aNodes[m_NumNodes-1]
+    imul ebx, HUFF_CNODE_SIZE
+    lea rax, [huff_nodes + ebx]
+    ; m_pStartNode = &m_aNodes[m_NumNodes-1];
+    mov qword [huff_start_node], rax
+
+    ;  rax = *pNode (still set)
+    ;  rdi = int Bits
+    mov rdi, 0
+    ;  rsi = unsigned Depth
+    mov rsi, 0
+    call _huff_setbits_r
+
+
+    mov rsp, rbp
+    pop_registers
     ret
 
diff --git a/src/logger.asm b/src/logger.asm
@@ -75,7 +75,7 @@ print_any_int:
     sub rsp, 16
 
     ; buffer
-    mov [rbp-16], %1
+    mov qword [rbp-16], %1
 
     ; this matches so the pointer move worked fine
     ; mov rax, [rbp-14]
@@ -121,6 +121,11 @@ print_any_int:
     pop_registers
 %endmacro
 
+%macro print_int32_array 2
+    ; print_int32_array [array buffer] [array size]
+    print_int_array %1, 4, %2
+%endmacro
+
 print_newline:
     push_registers
 

diff --git a/tests/huffman_decompress_test.asm b/tests/huffman_decompress_test.asm
@@ -1,5 +1,6 @@
 %include "tests/assert.asm"
 
 _start:
+    ; call _huff_construct_tree
     exit 0