romdo/go-conver
1
0
Fork 0
mirror of https://github.com/romdo/go-conver.git synced 2026-02-19 08:16:40 +00:00
Code Issues Packages Projects Releases Wiki Activity

wip: improve whitespace handling in paragraph parsing

Browse Source
This commit is contained in:
Jim Myhrberg
2020-10-31 19:55:56 +00:00
parent 67c89f0df1
commit 5d205a639c
2 changed files with 59 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
pkg/commit/parser.go
View File

@@ -34,10 +34,14 @@ func parseHeader(header []byte) (*Commit, error) {
}, nil
}
func paragraphs(input []byte) [][]byte {
paras := bytes.Split(normlizeLinefeeds(input), []byte{lf, lf})
func paragraphs(commitMsg []byte) [][]byte {
paras := bytes.Split(
bytes.TrimSpace(normlizeLinefeeds(commitMsg)),
[]byte{lf, lf},
)
for i, p := range paras {
paras[i] = bytes.Trim(p, crlf)
paras[i] = bytes.TrimSpace(p)
}
return paras

70
pkg/commit/parser_test.go
View File

@@ -194,12 +194,17 @@ func Test_paragraph(t *testing.T) {
args: args{input: []byte("hello world\nthe brown fox\n")},
want: [][]byte{[]byte("hello world\nthe brown fox")},
},
{
name: "excess whitespace",
args: args{input: []byte(" \n hello world\nthe brown fox \n ")},
want: [][]byte{[]byte("hello world\nthe brown fox")},
},
{
name: "multiple paragraphs",
args: args{input: []byte(
"Lorem ipsum dolor sit amet, consectetur adipiscing elit.\n" +
"Praesent eleifend lorem non purus finibus, interdum\n" +
"hendrerit sem bibendum.\n" +
"Lorem ipsum dolor sit amet, consectetur adipiscing\n" +
"elit.Praesent eleifend lorem non purus finibus,\n" +
"interdum hendrerit sem bibendum.\n" +
"\n" +
"Etiam porttitor mollis nulla, egestas facilisis nisi\n" +
"molestie ut. Quisque mi mi, commodo ut mattis a,\n" +
@@ -211,9 +216,9 @@ func Test_paragraph(t *testing.T) {
)},
want: [][]byte{
[]byte(
"Lorem ipsum dolor sit amet, consectetur adipiscing elit.\n" +
"Praesent eleifend lorem non purus finibus, interdum\n" +
"hendrerit sem bibendum.",
"Lorem ipsum dolor sit amet, consectetur adipiscing\n" +
"elit.Praesent eleifend lorem non purus finibus,\n" +
"interdum hendrerit sem bibendum.",
),
[]byte(
"Etiam porttitor mollis nulla, egestas facilisis nisi\n" +
@@ -227,12 +232,41 @@ func Test_paragraph(t *testing.T) {
),
},
},
{
name: "paragraphs with surrounding whitespace",
args: args{input: []byte(
"\n" +
" \n" +
" Lorem ipsum dolor sit amet, consectetur adipiscing\n" +
"elit.Praesent eleifend lorem non purus finibus,\n" +
"interdum hendrerit sem bibendum. \n" +
"\n" +
"\n" +
" Etiam porttitor mollis nulla, egestas facilisis nisi\n" +
"molestie ut. Quisque mi mi, commodo ut mattis a,\n" +
"scelerisque eu elit.\n" +
" \n" +
" ",
)},
want: [][]byte{
[]byte(
"Lorem ipsum dolor sit amet, consectetur adipiscing\n" +
"elit.Praesent eleifend lorem non purus finibus,\n" +
"interdum hendrerit sem bibendum.",
),
[]byte(
"Etiam porttitor mollis nulla, egestas facilisis nisi\n" +
"molestie ut. Quisque mi mi, commodo ut mattis a,\n" +
"scelerisque eu elit.",
),
},
},
{
name: "CRLF line separator",
args: args{input: []byte(
"Lorem ipsum dolor sit amet, consectetur adipiscing elit.\r\n" +
"Praesent eleifend lorem non purus finibus, interdum\r\n" +
"hendrerit sem bibendum.\r\n" +
"Lorem ipsum dolor sit amet, consectetur adipiscing\r\n" +
"elit.Praesent eleifend lorem non purus finibus,\r\n" +
"interdum hendrerit sem bibendum.\r\n" +
"\r\n" +
"Etiam porttitor mollis nulla, egestas facilisis nisi\r\n" +
"molestie ut. Quisque mi mi, commodo ut mattis a,\r\n" +
@@ -240,9 +274,9 @@ func Test_paragraph(t *testing.T) {
)},
want: [][]byte{
[]byte(
"Lorem ipsum dolor sit amet, consectetur adipiscing elit.\n" +
"Praesent eleifend lorem non purus finibus, interdum\n" +
"hendrerit sem bibendum.",
"Lorem ipsum dolor sit amet, consectetur adipiscing\n" +
"elit.Praesent eleifend lorem non purus finibus,\n" +
"interdum hendrerit sem bibendum.",
),
[]byte(
"Etiam porttitor mollis nulla, egestas facilisis nisi\n" +
@@ -254,9 +288,9 @@ func Test_paragraph(t *testing.T) {
{
name: "CR line separator",
args: args{input: []byte(
"Lorem ipsum dolor sit amet, consectetur adipiscing elit.\r" +
"Praesent eleifend lorem non purus finibus, interdum\r" +
"hendrerit sem bibendum.\r" +
"Lorem ipsum dolor sit amet, consectetur adipiscing\r" +
"elit.Praesent eleifend lorem non purus finibus,\r" +
"interdum hendrerit sem bibendum.\r" +
"\r" +
"Etiam porttitor mollis nulla, egestas facilisis nisi\r" +
"molestie ut. Quisque mi mi, commodo ut mattis a,\r" +
@@ -264,9 +298,9 @@ func Test_paragraph(t *testing.T) {
)},
want: [][]byte{
[]byte(
"Lorem ipsum dolor sit amet, consectetur adipiscing elit.\n" +
"Praesent eleifend lorem non purus finibus, interdum\n" +
"hendrerit sem bibendum.",
"Lorem ipsum dolor sit amet, consectetur adipiscing\n" +
"elit.Praesent eleifend lorem non purus finibus,\n" +
"interdum hendrerit sem bibendum.",
),
[]byte(
"Etiam porttitor mollis nulla, egestas facilisis nisi\n" +