a fork of iceshrimp.net but a tweaked frontend to my personal liking. waow
fediverse
social-media
social
iceshrimp
fedi
[parsing] Add parenthesis tracking to MfmUrlNode and MfmLinkNode parsers (ISH-608)
+98
-12
+45
-9
Iceshrimp.Parsing/Mfm.fs
+45
-9
Iceshrimp.Parsing/Mfm.fs
···
101
101
inherit MfmInlineNode([])
102
102
member val Char = v
103
103
104
+
type internal UserState =
105
+
{ ParenthesisStack: char list
106
+
LastLine: int64 }
107
+
108
+
static member Default = { ParenthesisStack = []; LastLine = 0 }
109
+
104
110
open MfmNodeTypes
105
111
106
112
module private MfmParser =
···
114
120
let isLetterOrNumber c = Char.IsLetterOrDigit c
115
121
let isNewline c = '\n'.Equals(c)
116
122
let isNotNewline c = not (isNewline c)
123
+
124
+
let followedByChar c = nextCharSatisfies <| fun ch -> c = ch
117
125
118
126
let (|CharNode|MfmNode|) (x: MfmNode) =
119
127
if x :? MfmCharNode then
···
174
182
| None -> None
175
183
| Some items -> items |> dict |> Some
176
184
177
-
let pushLine: Parser<unit, int64> =
185
+
let pushLine: Parser<unit, UserState> =
178
186
fun stream ->
179
-
stream.UserState <- stream.Line
187
+
stream.UserState <-
188
+
{ stream.UserState with
189
+
LastLine = stream.Line }
190
+
180
191
Reply(())
181
192
182
-
let assertLine: Parser<unit, int64> =
193
+
let assertLine: Parser<unit, UserState> =
183
194
fun stream ->
184
-
match stream.UserState = stream.Line with
195
+
match stream.UserState.LastLine = stream.Line with
185
196
| true -> Reply(())
186
197
| false -> Reply(Error, messageError "Line changed")
187
198
199
+
let assertParen = userStateSatisfies <| fun u -> u.ParenthesisStack.Length > 0
200
+
let assertNoParen = userStateSatisfies <| fun u -> u.ParenthesisStack.Length = 0
201
+
202
+
let pushParen =
203
+
updateUserState
204
+
<| fun u ->
205
+
{ u with
206
+
ParenthesisStack = '(' :: u.ParenthesisStack }
207
+
208
+
let popParen =
209
+
assertParen
210
+
>>. updateUserState (fun u ->
211
+
{ u with
212
+
ParenthesisStack = List.tail u.ParenthesisStack })
213
+
214
+
let clearParen = updateUserState <| fun u -> { u with ParenthesisStack = [] }
215
+
188
216
// References
189
217
let node, nodeRef = createParserForwardedToRef ()
190
218
let inlineNode, inlineNodeRef = createParserForwardedToRef ()
···
328
356
329
357
let urlNode =
330
358
lookAhead (skipString "https://" <|> skipString "http://")
331
-
>>. manyCharsTill anyChar (nextCharSatisfies isWhitespace <|> nextCharSatisfies (isAnyOf "()") <|> eof) //FIXME: this needs significant improvements
359
+
>>. manyCharsTill
360
+
((pchar '(' .>> pushParen) <|> (pchar ')' .>> popParen) <|> anyChar)
361
+
(nextCharSatisfies isWhitespace
362
+
<|> (assertNoParen >>. followedByChar ')')
363
+
<|> eof)
364
+
.>> clearParen
332
365
>>= fun uri ->
333
366
match Uri.TryCreate(uri, UriKind.Absolute) with
334
367
| true, finalUri ->
···
356
389
.>>. (pchar '[' >>. manyCharsTill anyChar (pchar ']'))
357
390
.>>. (pchar '('
358
391
>>. lookAhead (skipString "https://" <|> skipString "http://")
359
-
>>. manyCharsTill anyChar (pchar ')'))
392
+
>>. manyCharsTill
393
+
((pchar '(' .>> pushParen) <|> (pchar ')' .>> popParen) <|> anyChar)
394
+
(assertNoParen >>. skipChar ')'))
395
+
.>> clearParen
360
396
>>= fun ((silent, text), uri) ->
361
397
match Uri.TryCreate(uri, UriKind.Absolute) with
362
398
| true, finalUri ->
···
394
430
| Simple
395
431
396
432
let parseNode (m: ParseMode) =
397
-
let prefixedNode (m: ParseMode) : Parser<MfmNode, int64> =
433
+
let prefixedNode (m: ParseMode) : Parser<MfmNode, UserState> =
398
434
fun (stream: CharStream<_>) ->
399
435
match (stream.Peek(), m) with
400
436
// Block nodes, ordered by expected frequency
···
439
475
440
476
module Mfm =
441
477
let parse str =
442
-
match runParserOnString parse 0 "" str with
478
+
match runParserOnString parse UserState.Default "" str with
443
479
| Success(result, _, _) -> aggregateText result
444
480
| Failure(s, _, _) -> failwith $"Failed to parse MFM: {s}"
445
481
446
482
let parseSimple str =
447
-
match runParserOnString parseSimple 0 "" str with
483
+
match runParserOnString parseSimple UserState.Default "" str with
448
484
| Success(result, _, _) -> aggregateText result
449
485
| Failure(s, _, _) -> failwith $"Failed to parse MFM: {s}"
+53
-3
Iceshrimp.Tests/Parsing/MfmTests.cs
+53
-3
Iceshrimp.Tests/Parsing/MfmTests.cs
···
262
262
[TestMethod]
263
263
public void TestUrl()
264
264
{
265
-
const string input = "https://example.org/path/Name_(test)_asdf";
266
-
//TODO: List<MfmNode> expected = [new MfmUrlNode(input, false),];
267
-
List<MfmNode> expected = [new MfmUrlNode(input[..30], false), new MfmTextNode(input[30..])];
265
+
const string input = "https://example.org/path/Name_(test)_asdf";
266
+
List<MfmNode> expected = [new MfmUrlNode(input, false)];
268
267
var res = Mfm.parse(input);
268
+
269
+
AssertionOptions.FormattingOptions.MaxDepth = 100;
270
+
res.ToList().Should().Equal(expected, MfmNodeEqual);
271
+
MfmSerializer.Serialize(res).Should().BeEquivalentTo(input);
272
+
}
273
+
274
+
[TestMethod]
275
+
public void TestUrlAlt()
276
+
{
277
+
const string input = "https://example.org/path/Name_(test";
278
+
List<MfmNode> expected = [new MfmUrlNode(input, false)];
279
+
var res = Mfm.parse(input);
280
+
281
+
AssertionOptions.FormattingOptions.MaxDepth = 100;
282
+
res.ToList().Should().Equal(expected, MfmNodeEqual);
283
+
MfmSerializer.Serialize(res).Should().BeEquivalentTo(input);
284
+
}
285
+
286
+
[TestMethod]
287
+
public void TestUrlNeg()
288
+
{
289
+
const string input = "https://example.org/path/Name_test)_asdf";
290
+
List<MfmNode> expected = [new MfmUrlNode(input[..34], false), new MfmTextNode(input[34..])];
291
+
var res = Mfm.parse(input);
292
+
293
+
AssertionOptions.FormattingOptions.MaxDepth = 100;
294
+
res.ToList().Should().Equal(expected, MfmNodeEqual);
295
+
MfmSerializer.Serialize(res).Should().BeEquivalentTo(input);
296
+
}
297
+
298
+
[TestMethod]
299
+
public void TestLink()
300
+
{
301
+
const string input = "[test](https://example.org/path/Name_(test)_asdf)";
302
+
List<MfmNode> expected = [new MfmLinkNode("https://example.org/path/Name_(test)_asdf", "test", false)];
303
+
var res = Mfm.parse(input);
304
+
305
+
AssertionOptions.FormattingOptions.MaxDepth = 100;
306
+
res.ToList().Should().Equal(expected, MfmNodeEqual);
307
+
MfmSerializer.Serialize(res).Should().BeEquivalentTo(input);
308
+
}
309
+
310
+
[TestMethod]
311
+
public void TestLinkNeg()
312
+
{
313
+
const string input = "[test](https://example.org/path/Name_(test_asdf)";
314
+
List<MfmNode> expected =
315
+
[
316
+
new MfmTextNode("[test]("), new MfmUrlNode("https://example.org/path/Name_(test_asdf)", false)
317
+
];
318
+
var res = Mfm.parse(input);
269
319
270
320
AssertionOptions.FormattingOptions.MaxDepth = 100;
271
321
res.ToList().Should().Equal(expected, MfmNodeEqual);