Fix parse package import
This commit is contained in:
66
parse/posts.go
Normal file
66
parse/posts.go
Normal file
@@ -0,0 +1,66 @@
|
||||
package parse
|
||||
|
||||
import (
|
||||
"github.com/writeas/web-core/stringmanip"
|
||||
"regexp"
|
||||
"strings"
|
||||
)
|
||||
|
||||
var (
|
||||
titleElementReg = regexp.MustCompile("</?p>")
|
||||
urlReg = regexp.MustCompile("https?://")
|
||||
imgReg = regexp.MustCompile(`!\[([^]]+)\]\([^)]+\)`)
|
||||
)
|
||||
|
||||
// PostLede attempts to extract the first thought of the given post, generally
|
||||
// contained within the first line or sentence of text.
|
||||
func PostLede(t string, includePunc bool) string {
|
||||
// Adjust where we truncate if we want to include punctuation
|
||||
iAdj := 0
|
||||
if includePunc {
|
||||
iAdj = 1
|
||||
}
|
||||
|
||||
// Find lede within first line of text
|
||||
nl := strings.IndexRune(t, '\n')
|
||||
if nl > -1 {
|
||||
t = t[:nl]
|
||||
}
|
||||
|
||||
// Strip certain HTML tags
|
||||
t = titleElementReg.ReplaceAllString(t, "")
|
||||
|
||||
// Strip URL protocols
|
||||
t = urlReg.ReplaceAllString(t, "")
|
||||
|
||||
// Strip image URL, leaving only alt text
|
||||
t = imgReg.ReplaceAllString(t, " $1 ")
|
||||
|
||||
// Find lede within first sentence
|
||||
punc := strings.Index(t, ". ")
|
||||
if punc > -1 {
|
||||
t = t[:punc+iAdj]
|
||||
}
|
||||
punc = stringmanip.IndexRune(t, '。')
|
||||
if punc > -1 {
|
||||
c := []rune(t)
|
||||
t = string(c[:punc+iAdj])
|
||||
}
|
||||
|
||||
return t
|
||||
}
|
||||
|
||||
// TruncToWord truncates the given text to the provided limit.
|
||||
func TruncToWord(s string, l int) (string, bool) {
|
||||
truncated := false
|
||||
c := []rune(s)
|
||||
if len(c) > l {
|
||||
truncated = true
|
||||
s = string(c[:l])
|
||||
spaceIdx := strings.LastIndexByte(s, ' ')
|
||||
if spaceIdx > -1 {
|
||||
s = s[:spaceIdx]
|
||||
}
|
||||
}
|
||||
return s, truncated
|
||||
}
|
Reference in New Issue
Block a user