notfire/Iceshrimp.NET
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

[parsing/mfm] Improve urlNode(Brackets) performance & reliability, improve performance of long inputs for some node types

Browse source
This commit is contained in:
Laura Hausmann 2024-11-27 20:50:54 +01:00
parent 5a3d4b9bb4
commit 112d6affd0
No known key found for this signature in database
GPG key ID: D044E84C5BE01605
1 changed files with 29 additions and 17 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

46
Iceshrimp.Parsing/Mfm.fs
View file

@ -318,6 +318,17 @@ module private MfmParser =
else
Reply(Error, error)
let anyCharExcept ch: Parser<char, UserState> =
let error = messageError "anyCharExcept"
fun stream ->
let c = stream.ReadCharOrNewline()
if c <> EOS && c <> ch then
Reply(c)
else
Reply(Error, error)
let createParameterizedParserRef () =
let dummyParser _ =
fun _ -> failwith "a parser created with createParameterizedParserRef was not initialized"
@ -438,7 +449,7 @@ module private MfmParser =
let emojiCodeNode =
skipChar ':'
>>. restOfLineContains ":"
>>. restOfLineContainsChar ':'
>>. manyCharsTill (satisfy isAsciiLetter <|> satisfy isDigit <|> anyOf "+-_") (skipChar ':')
|>> fun e -> MfmEmojiCodeNode(e) :> MfmNode
@ -494,40 +505,41 @@ module private MfmParser =
let urlNode =
lookAhead (skipString "https://" <|> skipString "http://")
>>. manyCharsTill
((pchar '(' .>> pushParen) <|> (pchar ')' .>> popParen) <|> anyChar)
>>. previousCharSatisfiesNot (fun c -> c = '<')
>>. many1CharsTill
((pchar '(' .>> pushParen) <|> (pchar ')' .>> popParen) <|> anyCharExcept '>')
(nextCharSatisfies isWhitespace
<|> (assertNoParen >>. followedByChar ')')
<|> eof)
.>> clearParen
>>= fun uri ->
|>> fun uri ->
match Uri.TryCreate(uri, UriKind.Absolute) with
| true, NonNullQuick finalUri ->
match finalUri.Scheme with
| "http" -> preturn (MfmUrlNode(uri, false) :> MfmNode)
| "https" -> preturn (MfmUrlNode(uri, false) :> MfmNode)
| _ -> fail "invalid scheme"
| _ -> fail "invalid url"
| "http" -> MfmUrlNode(uri, false) :> MfmNode
| "https" -> MfmUrlNode(uri, false) :> MfmNode
| _ -> MfmTextNode(uri) :> MfmNode
| _ -> MfmTextNode(uri) :> MfmNode
let urlNodeBrackets =
skipChar '<'
>>. lookAhead (skipString "https://" <|> skipString "http://")
>>. restOfLineContains ">" // This intentionally breaks compatibility with mfm-js, as there's no reason to allow newlines in urls
>>. manyCharsTill anyChar (skipChar '>')
>>= fun uri ->
>>. restOfLineContainsChar '>' // This intentionally breaks compatibility with mfm-js, as there's no reason to allow newlines in urls
>>. many1CharsTill anyChar (skipChar '>')
|>> fun uri ->
match Uri.TryCreate(uri, UriKind.Absolute) with
| true, NonNullQuick finalUri ->
match finalUri.Scheme with
| "http" -> preturn (MfmUrlNode(uri, true) :> MfmNode)
| "https" -> preturn (MfmUrlNode(uri, true) :> MfmNode)
| _ -> fail "invalid scheme"
| _ -> fail "invalid url"
| "http" -> MfmUrlNode(uri, true) :> MfmNode
| "https" -> MfmUrlNode(uri, true) :> MfmNode
| _ -> MfmTextNode(uri) :> MfmNode
| _ -> MfmTextNode(uri) :> MfmNode
let linkNode =
(opt (pchar '?'))
.>>. (pchar '[' >>. restOfLineContains "]" >>. manyCharsTill anyChar (pchar ']'))
.>>. (pchar '[' >>. restOfLineContainsChar ']' >>. manyCharsTill anyChar (pchar ']'))
.>>. (pchar '('
>>. restOfLineContains ")"
>>. restOfLineContainsChar ')'
>>. lookAhead (skipString "https://" <|> skipString "http://")
>>. manyCharsTill
((pchar '(' .>> pushParen) <|> (pchar ')' .>> popParen) <|> anyCharExceptNewline)