jiazhizhong/wechatmp2markdown
1
0
Fork 0
mirror of https://github.com/fengxxc/wechatmp2markdown.git synced 2026-06-09 03:27:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: 修复文本节点丢失的bug

Browse Source
This commit is contained in:
fengxxc
2021-12-02 15:56:23 +08:00
parent 32558cc8e8
commit 800c28772c
1 changed files with 25 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

64
parse/parse.go
View File

@@ -15,33 +15,38 @@ import (
) )
func parseSection(s *goquery.Selection) []Piece { func parseSection(s *goquery.Selection) []Piece {
var piece []Piece var pieces []Piece
s.Children().Each(func(i int, sc *goquery.Selection) { s.Contents().Each(func(i int, sc *goquery.Selection) {
attr := make(map[string]string) attr := make(map[string]string)
if sc.Is("span") { if sc.Is("span") {
// p = Piece{NORMAL_TEXT, s.Text(), nil} pieces = append(pieces, Piece{NORMAL_TEXT, sc.Text(), nil})
piece = append(piece, Piece{NORMAL_TEXT, sc.Text(), nil})
} else if sc.Is("a") { } else if sc.Is("a") {
attr["href"], _ = sc.Attr("href") attr["href"], _ = sc.Attr("href")
// p = Piece{LINK, removeBrAndBlank(s.Text()), attr} pieces = append(pieces, Piece{LINK, removeBrAndBlank(sc.Text()), attr})
piece = append(piece, Piece{LINK, removeBrAndBlank(sc.Text()), attr})
} else if sc.Is("img") { } else if sc.Is("img") {
attr["src"], _ = sc.Attr("data-src") attr["src"], _ = sc.Attr("data-src")
attr["alt"], _ = sc.Attr("alt") attr["alt"], _ = sc.Attr("alt")
attr["title"], _ = sc.Attr("title") attr["title"], _ = sc.Attr("title")
// p = Piece{IMAGE, "", attr} pieces = append(pieces, Piece{IMAGE, "", attr}, Piece{BR, nil, nil})
piece = append(piece, Piece{IMAGE, "", attr})
} else if sc.Is("ol") { } else if sc.Is("ol") {
piece = append(piece, parseList(sc, O_LIST)...) pieces = append(pieces, parseList(sc, O_LIST)...)
} else if sc.Is("ul") { } else if sc.Is("ul") {
piece = append(piece, parseList(sc, U_LIST)...) pieces = append(pieces, parseList(sc, U_LIST)...)
} else if sc.Is("section") { } else if sc.Is("pre") || sc.Is("section.code-snippet__fix") {
piece = append(piece, parseSection(sc)...) // 代码块
pieces = append(pieces, parsePre(sc)...)
} else if sc.Is("p") || sc.Is("section") {
pieces = append(pieces, parseSection(sc)...)
} else if sc.Is("h1") || sc.Is("h2") || sc.Is("h3") || sc.Is("h4") || sc.Is("h5") || sc.Is("h6") {
pieces = append(pieces, parseHeader(sc)...)
} else if sc.Is("blockquote") {
pieces = append(pieces, parseBlockQuote(sc)...)
} else { } else {
piece = append(piece, Piece{NORMAL_TEXT, sc.Text(), nil}) pieces = append(pieces, Piece{NORMAL_TEXT, sc.Text(), nil})
} }
}) })
return piece pieces = append(pieces, Piece{BR, nil, nil})
return pieces
} }
func parseHeader(s *goquery.Selection) []Piece { func parseHeader(s *goquery.Selection) []Piece {
@@ -62,7 +67,7 @@ func parseHeader(s *goquery.Selection) []Piece {
} }
attr := map[string]string{"level": strconv.Itoa(level)} attr := map[string]string{"level": strconv.Itoa(level)}
p := Piece{HEADER, removeBrAndBlank(s.Text()), attr} p := Piece{HEADER, removeBrAndBlank(s.Text()), attr}
return []Piece{p} return []Piece{p, {BR, nil, nil}}
} }
func parsePre(s *goquery.Selection) []Piece { func parsePre(s *goquery.Selection) []Piece {
@@ -71,7 +76,7 @@ func parsePre(s *goquery.Selection) []Piece {
codeRows = append(codeRows, sc.Text()) codeRows = append(codeRows, sc.Text())
}) })
p := Piece{CODE_BLOCK, codeRows, nil} p := Piece{CODE_BLOCK, codeRows, nil}
return []Piece{p} return []Piece{p, {BR, nil, nil}}
} }
func parseList(s *goquery.Selection, ptype PieceType) []Piece { func parseList(s *goquery.Selection, ptype PieceType) []Piece {
@@ -79,14 +84,16 @@ func parseList(s *goquery.Selection, ptype PieceType) []Piece {
s.Find("li").Each(func(i int, sc *goquery.Selection) { s.Find("li").Each(func(i int, sc *goquery.Selection) {
list = append(list, Piece{ptype, parseSection(sc), nil}) list = append(list, Piece{ptype, parseSection(sc), nil})
}) })
list = append(list, Piece{BR, nil, nil})
return list return list
} }
func parseBlockQuote(s *goquery.Selection) []Piece { func parseBlockQuote(s *goquery.Selection) []Piece {
var bq []Piece var bq []Piece
s.Children().Each(func(i int, sc *goquery.Selection) { s.Contents().Each(func(i int, sc *goquery.Selection) {
bq = append(bq, Piece{BLOCK_QUOTES, parseSection(sc), nil}) bq = append(bq, Piece{BLOCK_QUOTES, parseSection(sc), nil})
}) })
bq = append(bq, Piece{BR, nil, nil})
return bq return bq
} }
@@ -118,28 +125,7 @@ func ParseFromReader(r io.Reader) Article {
// p[style="line-height: 1.5em;"] => 项目列表（有序/无序） // p[style="line-height: 1.5em;"] => 项目列表（有序/无序）
// section[style=".*text-align:center"]>img => 居中段落（图片） // section[style=".*text-align:center"]>img => 居中段落（图片）
content := mainContent.Find("#js_content") content := mainContent.Find("#js_content")
// var sections []Paragraph pieces := parseSection(content)
var pieces []Piece
content.Children().Each(func(i int, s *goquery.Selection) {
// var paragraph Paragraph
if s.Is("pre") || s.Is("section.code-snippet__fix") {
// 代码块
pieces = append(pieces, parsePre(s)...)
} else if s.Is("p") || s.Is("section") {
pieces = append(pieces, parseSection(s)...)
} else if s.Is("h1") || s.Is("h2") || s.Is("h3") || s.Is("h4") || s.Is("h5") || s.Is("h6") {
pieces = append(pieces, parseHeader(s)...)
} else if s.Is("ol") {
pieces = append(pieces, parseList(s, O_LIST)...)
} else if s.Is("ul") {
pieces = append(pieces, parseList(s, U_LIST)...)
} else if s.Is("blockquote") {
pieces = append(pieces, parseBlockQuote(s)...)
}
// sections = append(sections, paragraph)
pieces = append(pieces, Piece{BR, nil, nil})
})
// article.Content = sections
article.Content = pieces article.Content = pieces
return article return article