Merge pull request 'Non-blank line num display' (#7) from number-nonblank into master

Reviewed-on: #7

README.md

@@ -16,4 +16,8 @@ Run the `make` command in the root of the repository.
 
 Options:
 
-`-h, --help` - Show a help page.
+`-b, --number-nonblank` - number nonempty output lines, overrides -n
+`-E, --show-ends` - display $ at end of each line\
+`-n, --number` - number all output lines
+`--help` - Show the help page.\
+`--version` - Show the version page.

src/args.S

@@ -2,7 +2,7 @@
 sp_start:       .space 8
 
 .section .text
-.globl save_sp, fetch_argc, fetch_argv, has_switch
+.globl save_sp, fetch_argc, fetch_argv, has_switch, str_cmp
 
 save_sp:
     sd sp, sp_start, t0 # save the stack pointer
@@ -125,13 +125,13 @@ has_switch:
     
     mv a0, s3 # load argv buff addr into a0 (string 0)
     mv a1, s0 # move switch ptr 1 addr into a1 (string 1)
-    call _str_cmp
+    call str_cmp
     bnez a1, .LSwitchFound # if the strings match, exit loop
     beqz s1, .LSkipSecondSwitch # if we have no second switch, skip the check
 
     mv a0, s3 # move argv buff addr into a0 (string 0)
     mv a1, s1 # move switch ptr 2 addr into a1 (string 1)
-    call _str_cmp
+    call str_cmp
     bnez a1, .LSwitchFound # if the strings match, exit loop
 
     .LSkipSecondSwitch:
@@ -170,7 +170,7 @@ has_switch:
 # @param[in] a1: str buffer 2
 # @param[out] a0: result code.
 # @param[out] a1: is equal (0-no, 1-yes).
-_str_cmp:
+str_cmp:
     li t2, 0
 
     .LByteCmp:

src/cat.S

@@ -0,0 +1,152 @@
+.data
+mod_buf_0:      .space 36864
+mod_buf_1:      .space 36864
+
+.section .text
+.globl cat_file, open_fd
+
+# Cat out the given file to stdout
+# @param[in] a0: ptr to null-terminated file str. 0 = stdin
+# @param[out] a0: result code.
+cat_file:
+    addi sp, sp, -32
+    sd s0, 0(sp) 
+    sd ra, 8(sp)
+    sd s1, 16(sp) # next modbuf to WRITE TO (inverse is read) 0=modbuf_0, 1=modbuf_1
+    sd s2, 24(sp) # current modbuf byte count
+
+    mv a1, a0 # since the function takes in a0, move it to a1 for syscall
+    li s1, 1
+    li s2, 0
+    
+    beqz a0, 1f
+    call open_fd
+    j 2f
+    1:
+    li s0, 0 # stdin fd
+    2:
+
+    mv s0, a1 # put fd in s0
+
+    beqz a0, .LReadFileBlock # if we don't have an error, jump to reading
+    li a0, 1
+    j .LCatFileRet
+
+    .LReadFileBlock:
+    mv a0, s0
+    la a1, mod_buf_0
+    li a2, 4096
+    li a7, 63
+    ecall
+    blez a0, .LCloseFD # if we haven't read any bytes, EOF, close FD
+    mv s2, a0
+    li s1, 1
+
+    la t1, number_nonblank
+    lb t1, 0(t1)
+    beqz t1, .LNoNonblankNumbers
+
+    mv a1, s2 # load mod buf size to a1
+
+    beqz s1, 1f
+    la a0, mod_buf_0 # load mod_buf_0 as read buf
+    la a2, mod_buf_1 # load mod_buf_1 as write buf
+    j 2f
+    1:
+    la a0, mod_buf_1 # load mod_buf_1 as read buf
+    la a2, mod_buf_0 # load mod_buf_0 as write buf
+    2:
+
+    call add_line_numbers
+    xori s1, s1, 1 # flip the mod buf to write to
+    mv s2, a1 # update the current byte count
+    j .LNoNumbers # -n and -b are exclusive
+    .LNoNonblankNumbers:
+    
+    la t1, number
+    lb t1, 0(t1)
+    beqz t1, .LNoNumbers
+
+    mv a1, s2 # load mod buf size to a1
+
+    beqz s1, 1f
+    la a0, mod_buf_0 # load mod_buf_0 as read buf
+    la a2, mod_buf_1 # load mod_buf_1 as write buf
+    j 2f
+    1:
+    la a0, mod_buf_1 # load mod_buf_1 as read buf
+    la a2, mod_buf_0 # load mod_buf_0 as write buf
+    2:
+
+    call add_line_numbers
+    xori s1, s1, 1 # flip the mod buf to write to
+    mv s2, a1 # update the current byte count
+    .LNoNumbers:
+
+    la t1, show_ends
+    lb t1, 0(t1)
+    beqz t1, .LNoEndings
+
+    mv a1, s2 # load mod buf size to a1
+
+    beqz s1, 1f
+    la a0, mod_buf_0 # load mod_buf_0 as read buf
+    la a2, mod_buf_1 # load mod_buf_1 as write buf
+    j 2f
+    1:
+    la a0, mod_buf_1 # load mod_buf_1 as read buf
+    la a2, mod_buf_0 # load mod_buf_0 as write buf
+    2:
+
+    call add_endings
+    xori s1, s1, 1 # flip the mod buf to write to
+    mv s2, a1 # update the current byte count
+    .LNoEndings:
+
+    beqz s1, 1f 
+    la a1, mod_buf_0
+    j 2f
+    1:
+    la a1, mod_buf_1
+    2:
+
+    mv a2, s2 # load the amount of bytes read in previous step into bytes needed to write
+    li a0, 1
+    li a7, 64
+    ecall
+    j .LReadFileBlock
+    
+    .LCloseFD:
+    mv a0, s0
+    li a7, 57 # close fd
+    ecall
+
+    .LCatFileRet:
+    ld s0, 0(sp)
+    ld ra, 8(sp)
+    ld s1, 16(sp)
+    ld s2, 24(sp)
+    addi sp, sp, 32
+    ret
+
+# Open fd
+# @param[in] a0: ptr to null-terminated file str.
+# @param[out] a0: result code. (0-exists 1-doesn't exist)
+# @param[out] a1: fd.
+open_fd:
+    mv a1, a0 # since the function takes in a0, move it to a1 for syscall
+    
+    li a0, -100 # relative/absolute path
+    li a2, 0 # readonly
+    li a7, 56 # open
+    ecall
+    bltz a0, 1f # file doesn't exist
+    mv a1, a0 # move fd to a1
+    li a0, 0
+    j 2f
+
+    1:
+    li a0, 1
+
+    2:
+    ret

src/file.S

@@ -1,107 +0,0 @@
-.data
-cat_buf:    .space 4096
-
-.section .text
-.globl cat_file, open_fd, cat_stdin
-
-# Cat out the given file to stdout
-# @param[in] a0: ptr to null-terminated file str.
-# @param[out] a0: result code.
-cat_file:
-    mv a1, a0 # since the function takes in a0, move it to a1 for syscall
-    
-    addi sp, sp, -8
-    sd ra, 0(sp)
-    call open_fd
-    ld ra, 0(sp)
-    addi sp, sp, 8
-
-    mv t0, a1 # put fd in t0
-    beqz a0, .LReadFileBlock # if we don't have an error, jump to reading
-    li a0, 1
-    j .LCatFileRet
-
-
-    .LReadFileBlock:
-    mv a0, t0
-    la a1, cat_buf
-    li a2, 4096
-    li a7, 63
-    ecall
-    blez a0, .LCloseFD # if we haven't read any bytes, EOF, close FD
-    mv a2, a0 # move bytes read into bytes needed to write
-
-    li a0, 1
-    la a1, cat_buf
-    li a7, 64
-    ecall
-    j .LReadFileBlock
-    
-    .LCloseFD:
-    mv a0, t0
-    li a7, 57 # close fd
-    ecall
-
-    .LCatFileRet:
-    ret
-
-# Cat out stdin to stdout
-# @param[in] a0: Not used.
-# @param[out] a0: result code.
-cat_stdin:
-    .LReadStdinBlock:
-    li a0, 0
-    la a1, cat_buf
-    li a2, 4096
-    li a7, 63
-    ecall
-    blez a0, .LCatStdinRet # if we haven't read any bytes, EOF, close FD
-    mv a2, a0 # move bytes read into bytes needed to write
-
-    li a0, 1
-    la a1, cat_buf
-    li a7, 64
-    ecall
-    j .LReadStdinBlock
-
-    .LCatStdinRet:
-    ret
-
-# Open fd
-# @param[in] a0: ptr to null-terminated file str.
-# @param[out] a0: result code. (0-exists 1-doesn't exist)
-# @param[out] a1: fd.
-open_fd:
-    mv a1, a0 # since the function takes in a0, move it to a1 for syscall
-    
-    li a0, -100 # relative/absolute path
-    li a2, 0 # readonly
-    li a7, 56 # open
-    ecall
-    bltz a0, 1f # file doesn't exist
-    mv a1, a0 # move fd to a1
-    li a0, 0
-    j 2f
-
-    1:
-    li a0, 1
-
-    2:
-    ret
-
-# # Check if the given file exists without opening fd
-# # @param[in] a0: ptr to null-terminated file str.
-# # @param[out] a0: result code (0-exists 1-doesn't exist).
-# file_exists:
-#     mv a1, a0
-#     li a0, -100 # relative/absolute path
-#     li a2, 0
-#     li a7, 48
-#     ecall
-#     beqz a0, 1f
-#     li a0, 1
-#     ret
-
-#     1:
-#     li a0, 0
-#     ret

src/main.S

@@ -1,5 +1,8 @@
 .data
 argc:               .space 8
+show_ends:          .space 1
+number:             .space 1
+number_nonblank:    .space 1
 
 cat_prefix:     .ascii "cat: "
 .equ CAT_PREFIX_LEN, . - cat_prefix
@@ -11,13 +14,16 @@ help:
     .ascii "Usage: cat [OPTION]... [FILE]...\n"
     .ascii "Concatenate FILE(S) to standard output.\n\n"
     .ascii "With no FILE, or when FILE is -, read standard input.\n\n"
+    .ascii "    -b, --number-nonblank   number nonempty output lines, overrides -n\n"
+    .ascii "    -E, --show-ends         display $ at end of each line\n"
+    .ascii "    -n, --number            number all output lines\n"
     .ascii "    --help                  display this help and exit\n"
     .ascii "    --version               output version information and exit\n\n"
     .asciz "Git repository: <https://git.lexza.ch/Lexzach/riscv-cat>\n"
 .equ HELP_LEN, . - help
 
 version:
-    .ascii "cat 1.0.0\n"
+    .ascii "cat 1.4.0\n"
     .ascii "Copyright (C) 2026 lexzach\n"
     .ascii "License MIT\n"
     .ascii "This is free software: you are free to change and redistribute it.\n"
@@ -27,10 +33,17 @@ version:
 
 flag_help:     .asciz "--help"
 flag_version:  .asciz "--version"
+flag_show_ends:  .asciz "--show-ends"
+flag_E:  .asciz "-E"
+flag_number: .asciz "--number"
+flag_n: .asciz "-n"
+flag_number_nonblank: .asciz "--number-nonblank"
+flag_b: .asciz "-b"
 
 .section .text
-.globl _start
+.globl _start, show_ends, number, number_nonblank
 
+# main.
 _start:
     .option push
     .option norelax
@@ -41,25 +54,38 @@ _start:
     call fetch_argc
     mv s0, a1 # store argc
     li s1, 1
+    li s2, 0 # processed files
     
+    call zero_line_count
+    bnez a0, .LMainReturnErr
     call _check_flags
 
     li t0, 1
-    ble s0, t0, .LProcessStdin # use stdin if no commands
+    ble s0, t0, .LProcessStdin # use stdin if no commands, although this is checked later, if no flags are used, we can go ahead and jump
     
     .LProcessFile:
     mv a0, s1
     call fetch_argv
     bnez a0, .LMainReturnErr # error if failed
     mv a0, a1
+    addi sp, sp, -8
+    sd a0, 0(sp)
+    call _is_known_flag
+    ld a0, 0(sp)
+    addi sp, sp, 8
+    bnez a1, .LSkipArg # check if arg is flag
     call cat_file
+    addi s2, s2, 1
     bnez a0, .LReturnFileErr # error if failed
+    .LSkipArg:
     addi s1, s1, 1
     blt s1, s0, .LProcessFile
+    beqz s2, .LProcessStdin # if we reach the end of the file queue, and no files have actually been processed, that means they were all flags, switch to stdin
     j .LMainReturnOk
     
     .LProcessStdin:
-    call cat_stdin
+    li a0, 0
+    call cat_file
     j .LMainReturnOk
     
     .LReturnFileErr:
@@ -94,7 +120,9 @@ _start:
     ecall   # return 0
 
 
-
+# Check for flags, branch or set buffers as needed.
+# @param[in] a0: None.
+# @param[out] a0: None.
 _check_flags:
     addi sp, sp, -8
     sd ra, 0(sp)
@@ -113,7 +141,7 @@ _check_flags:
 
     .LSkipDispHelp:
     
-    la a0, flag_version # help flag
+    la a0, flag_version # version flag
     li a1, 0 # no second flag
     call has_switch
     bnez a0, .LMainReturnErr
@@ -126,6 +154,82 @@ _check_flags:
     j .LMainReturnOk
     .LSkipDispVersion:
 
+    la a0, flag_show_ends # show ends flag
+    la a1, flag_E # show ends -E flag
+    call has_switch
+    bnez a0, .LMainReturnErr
+    sb a1, show_ends, t1 # since the return for has switch is either 0 for not found or 1 for found, we can just write it directly into the buf
+    
+    la a0, flag_number
+    la a1, flag_n
+    call has_switch
+    bnez a0, .LMainReturnErr
+    sb a1, number, t1
+
+    la a0, flag_number_nonblank
+    la a1, flag_b
+    call has_switch
+    bnez a0, .LMainReturnErr
+    sb a1, number_nonblank, t1
+
     ld ra, 0(sp)
     addi sp, sp, 8
     ret
+
+# Check if given argv is a flag to determine if cat_file should be used
+# @param[in] a0: ptr to null-terminated argv string.
+# @param[out] a0: result code.
+# @param[out] a1: found flag (0- not a flag, 1- flag).
+_is_known_flag:
+    addi sp, sp, -16
+    sd ra, 8(sp)
+    sd s0, 0(sp)
+    
+    mv s0, a0
+
+    la a0, flag_help
+    mv a1, s0
+    call str_cmp
+    bnez a1, .LFlagFound
+
+    la a0, flag_version
+    mv a1, s0
+    call str_cmp
+    bnez a1, .LFlagFound
+
+    la a0, flag_show_ends
+    mv a1, s0
+    call str_cmp
+    bnez a1, .LFlagFound
+
+    la a0, flag_E
+    mv a1, s0
+    call str_cmp
+    bnez a1, .LFlagFound
+
+    la a0, flag_number
+    mv a1, s0
+    call str_cmp
+    bnez a1, .LFlagFound
+
+    la a0, flag_n
+    mv a1, s0
+    call str_cmp
+    bnez a1, .LFlagFound
+
+    la a0, flag_number_nonblank
+    mv a1, s0
+    call str_cmp
+    bnez a1, .LFlagFound
+
+    la a0, flag_b
+    mv a1, s0
+    call str_cmp
+    bnez a1, .LFlagFound
+
+    li a1, 0
+    .LFlagFound:
+    ld s0, 0(sp)
+    ld ra, 8(sp)
+    addi sp, sp, 16
+    ret

src/mod.S

@@ -0,0 +1,235 @@
+.data
+line_count:     .space 6
+
+.section .text
+.globl add_endings, add_line_numbers, zero_line_count
+
+# Print to stdout while adding "$" characters.
+# @param[in] a0: ptr to null-terminated str.
+# @param[in] a1: length of str.
+# @param[in] a2: mod buf addr.
+# @param[out] a0: result code.
+# @param[out] a1: bytes written to mod buf.
+add_endings:
+    addi sp, sp, -56
+    sd s5, 48(sp)
+    sd s4, 40(sp)
+    sd ra, 32(sp)
+    sd s3, 24(sp)
+    sd s2, 16(sp)
+    sd s1, 8(sp)
+    sd s0, 0(sp)
+
+
+    mv s0, a0 # put the str buf ptr in s0
+    mv s1, a1
+    li s2, 0 # set the str buf char counter to 0
+    li s3, 0 # set the mod buf char counter to 0
+    
+    mv s5, a2
+    .LEndingsProcessChunk:
+    add s4, s0, s2 # Calculate the offset address
+    lb t6, 0(s4)
+    li t4, 0xA # newline
+    bne t6, t4, 1f # if the current byte is not a newline, skip to writing
+
+    add s4, s5, s3 # s4 = mod buf[s2]
+    li t5, '$'
+    sb t5, 0(s4) # mod buf[s2] = $
+    li t5, '\n'
+    addi s4, s4, 1
+    sb t5, 0(s4) # mod buf[s2] = \n
+    addi s3, s3, 2
+    j 2f
+
+    1:
+    add s4, s5, s3 # s4 = mod buf[s2]
+    sb t6, 0(s4) # mod buf[s2] = t6
+
+    addi s3, s3, 1 # inc mod buf pointer
+    2:
+    addi s2, s2, 1 # inc str buf pointer
+    blt s2, s1, .LEndingsProcessChunk
+
+    li a0, 0
+    mv a1, s3
+
+    ld s0, 0(sp)
+    ld s1, 8(sp)
+    ld s2, 16(sp)
+    ld s3, 24(sp)
+    ld ra, 32(sp)
+    ld s4, 40(sp)
+    ld s5, 48(sp)
+    addi sp, sp, 56
+    ret
+
+# Prefix newlines with ######+TAB line count
+# @param[in] a0: ptr to null-terminated str.
+# @param[in] a1: length of str.
+# @param[in] a2: mod buf addr.
+# @param[out] a0: result code.
+# @param[out] a1: bytes written to mod buf.
+add_line_numbers:
+    addi sp, sp, -72
+    sd s7, 64(sp) # skip empty
+    sd s6, 56(sp) # mod buf ptr
+    sd s5, 48(sp) # mod buf bp
+    sd s4, 40(sp) # str buf ptr
+    sd ra, 32(sp)
+    sd s3, 24(sp) # mod buf offset
+    sd s2, 16(sp) # str buf offset
+    sd s1, 8(sp) # len of str
+    sd s0, 0(sp) # str buf addr
+
+    la t0, number_nonblank
+    lb s7, 0(t0)
+
+    mv s0, a0 # put the str ptr in s0
+    mv s1, a1 # move length of str into s1
+    li s2, 0 # set the str buf char offset to 0
+    li s3, 0 # set the mod buf char offset to 0
+    mv s5, a2 # move mod buf addr into s5
+
+    # there is always one line number at the very start
+    call _inc_line_count
+    li t2, 0 # lc char buf offset
+    li t3, 6 # lc char buf max offset
+    la t1, line_count # load the line count buffer addr
+    add s6, s5, s3 # update mod buf + offset addr
+    4:
+    add t4, t1, t2
+    lb t5, 0(t4)
+    sb t5, 0(s6) # mod buf[s4] = t4
+    addi t2, t2, 1
+    addi s3, s3, 1
+    add s6, s5, s3 # update mod buf + offset addr
+    blt t2, t3, 4b
+    li t5, 0x9 # tab character
+    sb t5, 0(s6)
+    addi s3, s3, 1
+    
+    .LLineNumbersProcessChunk:
+    add s4, s0, s2 # Calculate the offset address
+    lb t6, 0(s4)
+    li t4, 0xA # newline
+    bne t6, t4, 1f # if the current byte IS NOT a newline, skip to writing
+    mv t0, s2
+    addi t0, t0, 1 # move ahead one char and see if EOF
+    bge t0, s1, 1f # if the newline is at the very end, we don't indicate a new line number
+    beqz s7, .LLineNumbersSkipEmptyCheck
+    
+    mv t0, s4 # move the current string buf pointer into t0
+    addi t0, t0, 1 # move the str pointer forward one
+    lb t0, 0(t0)
+    # t4 still contains a newline char
+    beq t0, t4, 1f # if the next byte IS a newline, skip to writing
+    
+    .LLineNumbersSkipEmptyCheck:
+    
+    call _inc_line_count
+    
+    addi s3, s3, 1 # we want the ptr at the spot right after the newline
+    add s6, s5, s3 # update mod buf + offset addr
+    
+    li t2, 0 # lc char buf offset
+    li t3, 6 # lc char buf max offset
+    la t1, line_count # load the line count buffer addr
+    li t5, 0xA # newline character
+    sb t5, 0(s6)
+    addi s3, s3, 1
+    add s6, s5, s3 # update mod buf + offset addr
+    3:
+    add t4, t1, t2
+    lb t5, 0(t4)
+    sb t5, 0(s6) # mod buf[s4] = t4
+    addi t2, t2, 1
+    addi s3, s3, 1
+    add s6, s5, s3 # update mod buf + offset addr
+    blt t2, t3, 3b
+
+    li t5, 0x9 # tab character
+    sb t5, 0(s6)
+
+    addi s3, s3, 1
+    
+    j 2f
+
+    1:
+    add s6, s5, s3 # s4 = mod buf[s2]
+    sb t6, 0(s6) # mod buf[s2] = t6
+
+    addi s3, s3, 1 # inc mod buf pointer
+    2:
+    addi s2, s2, 1 # inc str buf pointer
+    blt s2, s1, .LLineNumbersProcessChunk
+
+    li a0, 0
+    mv a1, s3
+
+    ld s0, 0(sp)
+    ld s1, 8(sp)
+    ld s2, 16(sp)
+    ld s3, 24(sp)
+    ld ra, 32(sp)
+    ld s4, 40(sp)
+    ld s5, 48(sp)
+    ld s6, 56(sp)
+    ld s6, 64(sp)
+    addi sp, sp, 72
+    ret
+
+
+# Increment the line count string
+# @param[in] a0: not used
+# @param[out] a0: result code
+_inc_line_count:
+    la t0, line_count # load the line count buffer address
+    li t1, 5 # current offset, we move from right to left first
+
+    .LIncLine:
+    bltz t1, .LLineCountDoneInc
+    add t2, t0, t1 # load the correct offset address into t2
+    lb t3, 0(t2) # load the current ascii character into t3
+    
+    li t4, 32 # load space character
+    bne t3, t4, .LLineCountNotSpaceChar # if the character is not a space, skip setting to zero
+    li t4, 48 # ascii '0'
+    sb t4, 0(t2) # store this at the current pointer
+
+    .LLineCountNotSpaceChar:
+    add t2, t0, t1 # load the correct offset address into t2
+    lb t3, 0(t2) # load the current ascii character into t3
+    li t4, 58 # ascii ':', right after ascii '9', overflow
+    
+    addi t3, t3, 1
+    beq t3, t4, .LLineCountOverflow
+    sb t3, 0(t2)
+    j .LLineCountDoneInc
+    .LLineCountOverflow:
+    li t4, 48 # ascii '0'
+    sb t4, 0(t2) # store that ascii '0'
+    li t4, 1
+    sub t1, t1, t4 # move to the next digit to the left
+    j .LIncLine
+
+    .LLineCountDoneInc:
+    ret
+
+# Set the line count to all space characters
+# @param[in] a0: not used
+# @param[out] a0: result code
+zero_line_count:
+    la t0, line_count # load the line count buffer address
+    li t1, 0 # char ptr
+    li t2, 6 # max char
+    li t3, 32 # space character
+
+    1:
+    add t4, t0, t1 # load offset addr
+    sb t3, 0(t4) # store the space char in offset addr
+    addi t1, t1, 1
+    blt t1, t2, 1b
+
+    li a0, 0
+    ret