feat: implement sum (#121)

* feat: implement sum feat: implement buffered read for BSD algorithm feat: implement buffered read for SysV algorithm ref: extract block count fns chore: clean uo todo feat: add tests ref: split printing and data collect feat: add sysv quirk tests feat: add rjust fn test: add bsd main behanior tests feat: add zero padding to BSD sum, add tests feat: wide block count col * cleanup: remove unused import * chore: apply v fmt * chore: apply v fmt to wc/ * fix: whitespace output * fix: apply vfmt * fix: prod build * fix: vfmt, again * Update sum_test.v Can't use echo in Windows as it produces different outputs; example: `echo hhh | od -t x1` Linux 0000000 68 68 68 0a 0000004 Windows 0000000 68 68 68 20 0d 0a 0000006 * v fmt * Update sum_test.v Avoid `echo` because it behaves differently in Windows * Account for carriage returns Windows emits new lines as \r\n but the reference result was calculated on *nix where it was only \n * Account for carriage returns with proper escaping --------- Co-authored-by: syrmel <[email protected]> Co-authored-by: JalonSolov <[email protected]>
vlang · Nov 30, 2024 · 13a874e · 13a874e
1 parent e975b67
commit 13a874e
Show file tree

Hide file tree

Showing 4 changed files with 279 additions and 29 deletions.
diff --git a/src/sum/parse_args.v b/src/sum/parse_args.v
@@ -2,19 +2,32 @@ import common
 import os
 import time
 
+const app_name = 'sum'
+const app_description = '
+Print checksum and block counts for each FILE.
+
+With no FILE, or when FILE is -, read standard input.'
+
+struct Args {
+	sys_v bool
+	files []string
+}
+
 fn parse_args(args []string) Args {
 	mut fp := common.flag_parser(args)
 	fp.application(app_name)
-	fp.description('
-	Print or check BSD (16-bit) checksums.
+	fp.description(app_description)
 
-	With no FILE, or when FILE is -, read standard input.'.trim_indent())
-
-	fp.bool('', `r`, true, 'use BSD sum algorithm (the default), use 1K blocks')
-	sys_v := fp.bool('sysv', `s`, false, 'use System V sum algorithm, use 512 bytes blocks')
+	fp.bool('', `r`, true, 'use BSD sum algorithm, use 1K blocks')
+	mut sys_v := fp.bool('sysv', `s`, false, 'use System V sum algorithm, use 512 bytes blocks')
 	files_arg := fp.finalize() or { exit_error(err.msg()) }
 	files := scan_files_arg(files_arg)
 
+	// emulate original algorithm switches behavior
+	if '-rs' in args {
+		sys_v = true
+	}
+
 	return Args{
 		sys_v: sys_v
 		files: files

diff --git a/src/sum/sum.v b/src/sum/sum.v
@@ -1,33 +1,85 @@
 import os
 
-const app_name = 'sum'
-
-struct Args {
-	sys_v bool
-	files []string
+struct Sum {
+	checksum    u16
+	block_count u64
+mut:
+	file_name string
 }
 
+const bsd_block_size = 1024
+const sysv_block_size = 512
+
 fn main() {
 	args := parse_args(os.args)
+	mut sums := []Sum{}
+	mut block_size := match args.sys_v {
+		true { sysv_block_size }
+		false { bsd_block_size }
+	}
 
 	for file in args.files {
-		println(sum(file, args.sys_v))
+		checksum, mut blocks, file_name := sum(file, args.sys_v)
+		blocks = get_file_block_count(file, block_size)
+		sums << Sum{checksum, blocks, file_name}
+	}
+
+	if args.sys_v {
+		print_sysv(sums)
+	} else {
+		print_bsd(mut sums)
+	}
+}
+
+fn get_file_block_count(file string, block_size int) u64 {
+	file_size := os.file_size(file)
+	mut blocks := file_size / u64(block_size)
+	if file_size % u64(block_size) != 0 {
+		blocks += 1
+	}
+	return blocks
+}
+
+fn print_sysv(sums []Sum) {
+	for sum in sums {
+		println('${sum.checksum} ${sum.block_count}${sum.file_name}'.trim_space())
+	}
+}
+
+fn print_bsd(mut sums []Sum) {
+	if sums.len == 1 {
+		sums[0].file_name = ''
+	}
+	for sum in sums {
+		mut block_str := sum.block_count.str()
+		if block_str.len <= 5 {
+			block_str = rjust(block_str, 5)
+		}
+		checksum_str := '${sum.checksum:05}'
+		println('${checksum_str} ${block_str}${sum.file_name}'.trim_space())
+	}
+}
+
+fn rjust(s string, width int) string {
+	if width == 0 {
+		return s
 	}
+	return ' '.repeat(width - s.len) + s
 }
 
-fn sum(file string, sys_v bool) string {
+fn sum(file string, sys_v bool) (u16, u64, string) {
 	digest, blocks := match sys_v {
 		true { sum_sys_v(file) }
 		else { sum_bsd(file) }
 	}
 
-	name := if file.contains('/sum-') { '' } else { file }
-	return '${digest:5} ${blocks:5} ${name}'
+	name := if file.contains('/sum-') { '' } else { ' ${file}' }
+	return digest, blocks, name
 }
 
-fn sum_bsd(file string) (u16, int) {
-	mut count := 0
+fn sum_bsd(file string) (u16, u64) {
 	mut checksum := u16(0)
+	mut blocks := u64(0)
 	mut f := os.open(file) or { exit_error(err.msg()) }
 	defer { f.close() }
 
@@ -36,27 +88,23 @@ fn sum_bsd(file string) (u16, int) {
 		checksum = (checksum >> 1) + ((checksum & 1) << 15)
 		checksum += c
 		checksum &= 0xffff
-		count += 1
 	}
 
-	blocks := count / 1024 + 1
 	return checksum, blocks
 }
 
-fn sum_sys_v(file string) (u16, int) {
+fn sum_sys_v(file string) (u16, u64) {
 	mut sum := u32(0)
-	mut count := u32(0)
+	mut blocks := u64(0)
 	mut f := os.open(file) or { exit_error(err.msg()) }
 	defer { f.close() }
 
 	for {
 		c := f.read_raw[u8]() or { break }
 		sum += c
-		count += 1
 	}
 
 	r := (sum & 0xffff) + ((sum & 0xffffffff) >> 16)
 	checksum := u16((r & 0xffff) + (r >> 16))
-	blocks := count / 512 + 1
 	return checksum, blocks
 }
diff --git a/src/sum/sum_test.v b/src/sum/sum_test.v
@@ -1,15 +1,204 @@
-module main
-
 import os
+import io.util
+import common.testing
+
+const eol = testing.output_eol()
+const file_sep = os.path_separator
+
+const util = 'sum'
+
+const platform_util = $if !windows {
+	util
+} $else {
+	'coreutils ${util}'
+}
+
+const executable_under_test = testing.prepare_executable(util)
+
+const cmd = testing.new_paired_command(platform_util, executable_under_test)
+
+const test1_txt = os.join_path(testing.temp_folder, 'test1.txt')
+const test2_txt = os.join_path(testing.temp_folder, 'test2.txt')
+const test3_txt = os.join_path(testing.temp_folder, 'test3.txt')
+const long_line = os.join_path(testing.temp_folder, 'long_line')
+const large_file = os.join_path(testing.temp_folder, 'large_file')
+const main_txt = os.join_path(testing.temp_folder, 'test.txt')
+
+fn test_help_and_version() {
+	cmd.ensure_help_and_version_options_work()!
+}
 
 fn testsuite_begin() {
-	os.chdir(os.dir(@FILE))!
+	os.write_file(test1_txt, 'Hello World!\nHow are you?')!
+	os.write_file(test2_txt, '0123456789abcdefghijklmnopqrstuvwxyz')!
+	os.write_file(test3_txt, 'dummy')!
+	os.write_file(long_line, 'z'.repeat(1024 * 151))!
+	os.write_file(large_file, 'z'.repeat(110 * 1024 * 1024))!
+
+	sample_file_name := @FILE.trim_right('sum_test.v') + 'test.txt'
+	os.cp(sample_file_name, main_txt)!
+}
+
+fn testsuite_end() {
+	os.rm(test1_txt)!
+	os.rm(test2_txt)!
+	os.rm(test3_txt)!
+	os.rm(long_line)!
+	os.rm(large_file)!
+	os.rm(main_txt)!
 }
 
+/*
+	tests from main branch for completeness
+*/
 fn test_bsd() {
-	assert sum('test.txt', false) == '38039     1 test.txt'
+	res := os.execute('cat ${main_txt} | ${executable_under_test} -r')
+
+	assert res.exit_code == 0
+	assert res.output == '38039     1${eol}'
 }
 
 fn test_sysv() {
-	assert sum('test.txt', true) == '25426     1 test.txt'
+	res := os.execute('cat ${main_txt} | ${executable_under_test} -s')
+
+	assert res.exit_code == 0
+	assert res.output == '25426 1${eol}'
+}
+
+/*
+	test main SysV switch behavior
+*/
+fn test_sysv_stream_succeeds() {
+	res := os.execute('cat ${test1_txt} | ${executable_under_test} -s')
+
+	assert res.exit_code == 0
+	assert res.output == '2185 1${eol}'
+}
+
+fn test_sysv_one_file_succeeds() {
+	res := os.execute('${executable_under_test} -s ${test1_txt}')
+
+	assert res.exit_code == 0
+	assert res.output == '2185 1 ${test1_txt}${eol}'
+}
+
+fn test_sysv_repeated_files_not_get_filtered() {
+	res := os.execute('${executable_under_test} -s ${test1_txt} ${test1_txt} ${test1_txt}')
+
+	assert res.exit_code == 0
+	assert res.output == '2185 1 ${test1_txt}${eol}2185 1 ${test1_txt}${eol}2185 1 ${test1_txt}${eol}'
+}
+
+fn test_sysv_several_files_succeeds() {
+	res := os.execute('${executable_under_test} -s ${test1_txt} ${test2_txt} ${test3_txt}')
+
+	assert res.exit_code == 0
+	assert res.output == '2185 1 ${test1_txt}${eol}3372 1 ${test2_txt}${eol}556 1 ${test3_txt}${eol}'
+}
+
+fn sum_arbitrary_value(value string, arg string) !os.Result {
+	mut f, path := util.temp_file()!
+	f.write_string('${value}\n')!
+	f.close()
+	res := $if windows {
+		os.execute("cat ${path} | tr -d '\\r' | ${executable_under_test} ${arg}")
+	} $else {
+		os.execute('cat ${path} | ${executable_under_test} ${arg}')
+	}
+	os.rm(path)!
+	return res
+}
+
+/*
+	test SysV output quirks
+*/
+fn test_sysv_width_2_col_no_padding() {
+	res := sum_arbitrary_value('', '-s')!
+
+	assert res.exit_code == 0
+	assert res.output == '10 1${eol}'
+}
+
+fn test_sysv_width_3_col_no_padding() {
+	res := sum_arbitrary_value('\x61', '-s')!
+
+	assert res.exit_code == 0
+	assert res.output == '107 1${eol}'
+}
+
+fn test_sysv_width_4_col_no_padding() {
+	res := sum_arbitrary_value('zzzzzzzzz', '-s')!
+
+	assert res.exit_code == 0
+	assert res.output == '1108 1${eol}'
+}
+
+fn test_sysv_different_col_widths_no_alignment() {
+	res := os.execute('${executable_under_test} -s ${long_line} ${test1_txt} ${test2_txt} ${test3_txt}')
+
+	assert res.exit_code == 0
+	assert res.output == '55583 302 ${long_line}${eol}2185 1 ${test1_txt}${eol}3372 1 ${test2_txt}${eol}556 1 ${test3_txt}${eol}'
+}
+
+/*
+	test main BSD switch behavior
+*/
+fn test_bsd_sum_stream_succeeds() {
+	res := os.execute('cat ${test1_txt} | ${executable_under_test} -r')
+
+	assert res.exit_code == 0
+	assert res.output == '59852     1${eol}'
+}
+
+fn test_bsd_sum_one_file_succeeds() {
+	res := os.execute('${executable_under_test} -r ${test1_txt}')
+
+	assert res.exit_code == 0
+	assert res.output == '59852     1${eol}'
+}
+
+fn test_bsd_sum_repeated_files_not_get_filtered() {
+	res := os.execute('${executable_under_test} -r ${test1_txt} ${test1_txt} ${test1_txt}')
+
+	assert res.exit_code == 0
+	assert res.output == '59852     1 ${test1_txt}${eol}59852     1 ${test1_txt}${eol}59852     1 ${test1_txt}${eol}'
+}
+
+fn test_bsd_sum_several_files_succeeds() {
+	res := os.execute('${executable_under_test} -r ${test1_txt} ${test2_txt} ${test3_txt}')
+
+	assert res.exit_code == 0
+	assert res.output == '59852     1 ${test1_txt}${eol}11628     1 ${test2_txt}${eol}41183     1 ${test3_txt}${eol}'
+}
+
+/*
+	test BSD output quirks
+*/
+fn test_bsd_sum_col_width_2_padded_with_zero() {
+	res := sum_arbitrary_value('\x02', '-r')!
+
+	assert res.exit_code == 0
+	assert res.output == '00011     1${eol}'
+}
+
+fn test_bsd_sum_col_width_3_padded_with_zero() {
+	res := sum_arbitrary_value('hhh', '-r')!
+
+	assert res.exit_code == 0
+	assert res.output == '00101     1${eol}'
+}
+
+fn test_bsd_sum_col_width_4_padded_with_zero() {
+	res := sum_arbitrary_value('hhh', '-r')!
+
+	assert res.exit_code == 0
+	assert res.output == '00101     1${eol}'
+}
+
+fn test_bsd_block_col_width_more_than_5_not_aligned() {
+	// this test needs 100+MB input string and since there's no easy way to mock block count fn,
+	// we need to create an actual file
+	res := os.execute('${executable_under_test} -r ${test1_txt} ${large_file} ${test2_txt}')
+	assert res.exit_code == 0
+	assert res.output == '59852     1 ${test1_txt}${eol}62707 112640 ${large_file}${eol}11628     1 ${test2_txt}${eol}'
 }
diff --git a/src/wc/wc_test.v b/src/wc/wc_test.v
@@ -14,7 +14,7 @@ const long_over_16k = os.join_path(rig.temp_dir, 'long_over_16k')
 const long_under_16k = os.join_path(rig.temp_dir, 'long_under_16k')
 
 // todo add tests
-// - long line (>16k) count max line
+// - test windows \r\n vs \n
 
 fn testsuite_begin() {
 	rig.assert_platform_util()