forgejo/modules/markup/markdown/math/inline_parser.go
João Tiago 2adc3a45fb
Fixes #27605: inline math blocks can't be preceeded/followed by alphanumerical characters (#30175)
- Inline math blocks couldn't be preceeded or succeeded by
alphanumerical characters due to changes introduced in PR #21171.
Removed the condition that caused this (precedingCharacter condition)
and added a new exit condition of the for-loop that checks if a specific
'$' was escaped using '\' so that the math expression can be rendered as
intended.
- Additionally this PR fixes another bug where math blocks of the type
'$xyz$abc$' where the dollar sign was not escaped by the user, generated
an error (shown in the screenshots below)
- Altered the tests to accomodate for the changes

Former behaviour (from try.gitea.io):

![image](https://github.com/go-gitea/gitea/assets/114936010/8f0cbb21-321d-451c-b871-c67a8e1e9235)

Fixed behaviour (from my local build):

![image](https://github.com/go-gitea/gitea/assets/114936010/5c22687c-6f11-4407-b5e7-c14b838bc20d)

(Edit) Source code for the README.md file:
```
$x$ -$x$ $x$-

a$xa$ $xa$a 1$xb$ $xb$1

$a a$b b$

a$b $a a$b b$

$a a\$b b$
```

---------

Signed-off-by: João Tiago <joao.leal.tintas@tecnico.ulisboa.pt>
Co-authored-by: Giteabot <teabot@gitea.io>
(cherry picked from commit e006451ab1509f8d6d43c5974387c05b26517392)
2024-04-07 15:40:31 +02:00

127 lines
3.1 KiB
Go

// Copyright 2022 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package math
import (
"bytes"
"github.com/yuin/goldmark/ast"
"github.com/yuin/goldmark/parser"
"github.com/yuin/goldmark/text"
)
type inlineParser struct {
start []byte
end []byte
}
var defaultInlineDollarParser = &inlineParser{
start: []byte{'$'},
end: []byte{'$'},
}
// NewInlineDollarParser returns a new inline parser
func NewInlineDollarParser() parser.InlineParser {
return defaultInlineDollarParser
}
var defaultInlineBracketParser = &inlineParser{
start: []byte{'\\', '('},
end: []byte{'\\', ')'},
}
// NewInlineDollarParser returns a new inline parser
func NewInlineBracketParser() parser.InlineParser {
return defaultInlineBracketParser
}
// Trigger triggers this parser on $ or \
func (parser *inlineParser) Trigger() []byte {
return parser.start[0:1]
}
func isPunctuation(b byte) bool {
return b == '.' || b == '!' || b == '?' || b == ',' || b == ';' || b == ':'
}
func isAlphanumeric(b byte) bool {
return (b >= 'a' && b <= 'z') || (b >= 'A' && b <= 'Z') || (b >= '0' && b <= '9')
}
// Parse parses the current line and returns a result of parsing.
func (parser *inlineParser) Parse(parent ast.Node, block text.Reader, pc parser.Context) ast.Node {
line, _ := block.PeekLine()
if !bytes.HasPrefix(line, parser.start) {
// We'll catch this one on the next time round
return nil
}
precedingCharacter := block.PrecendingCharacter()
if precedingCharacter < 256 && (isAlphanumeric(byte(precedingCharacter)) || isPunctuation(byte(precedingCharacter))) {
// need to exclude things like `a$` from being considered a start
return nil
}
// move the opener marker point at the start of the text
opener := len(parser.start)
// Now look for an ending line
ender := opener
for {
pos := bytes.Index(line[ender:], parser.end)
if pos < 0 {
return nil
}
ender += pos
// Now we want to check the character at the end of our parser section
// that is ender + len(parser.end) and check if char before ender is '\'
pos = ender + len(parser.end)
if len(line) <= pos {
break
}
suceedingCharacter := line[pos]
if !isPunctuation(suceedingCharacter) && !(suceedingCharacter == ' ') {
return nil
}
if line[ender-1] != '\\' {
break
}
// move the pointer onwards
ender += len(parser.end)
}
block.Advance(opener)
_, pos := block.Position()
node := NewInline()
segment := pos.WithStop(pos.Start + ender - opener)
node.AppendChild(node, ast.NewRawTextSegment(segment))
block.Advance(ender - opener + len(parser.end))
trimBlock(node, block)
return node
}
func trimBlock(node *Inline, block text.Reader) {
if node.IsBlank(block.Source()) {
return
}
// trim first space and last space
first := node.FirstChild().(*ast.Text)
if !(!first.Segment.IsEmpty() && block.Source()[first.Segment.Start] == ' ') {
return
}
last := node.LastChild().(*ast.Text)
if !(!last.Segment.IsEmpty() && block.Source()[last.Segment.Stop-1] == ' ') {
return
}
first.Segment = first.Segment.WithStart(first.Segment.Start + 1)
last.Segment = last.Segment.WithStop(last.Segment.Stop - 1)
}