Diagnostics reference

Every recoverable issue the parser or writer encounters is reported as a LrcDiagnostic carrying a stable code (LRC00xx), severity, 1-based line/column location, and a human-readable message. Codes are exposed as constants on LrcDiagnosticIds so consumers can switch on them without coupling to an enum that changes between versions.

In tolerant mode (the default), every code below is collected on LrcParseResult.Diagnostics and the parser keeps going. In strict mode, the first Error-severity diagnostic causes LrcParseException to be thrown.

Codes

Structural

Code	Severity	Trigger	Recovery
`LRC0001` `UnclosedTag`	Error	A `[` opens with no matching `]` before line end.	Skip rest of line.
`LRC0002` `InvalidTimestamp`	Error	Digit-prefixed `[…]` that fails timestamp parsing (e.g. `[00:99.00]` — seconds out of range).	Skip the bracketed group; rest of line continues.
`LRC0003` `MalformedIdTag`	Error	`[key]` with no `:` (e.g. `[bad]junk`).	Skip the bracketed group.
`LRC0004` `UnknownIdTag`	Info	`[xx:value]` where `xx` isn't one of the known keys.	Preserved verbatim in `Metadata.RawTags`.
`LRC0005` `InvalidOffset`	Warning	`[offset:not-a-number]`.	Tag preserved in `RawTags`; typed `Offset` unchanged.
`LRC0006` `InvalidLength`	Warning	`[length:invalid]`.	Tag preserved in `RawTags`; typed `Length` unchanged.
`LRC0007` `InvalidEnhancedTimestamp`	Error	`<bad>` inside an enhanced line (text isn't a valid timestamp).	Word skipped; following text is preserved as literal line text.
`LRC0008` `UnclosedEnhancedTimestamp`	Error	`<` with no matching `>` before line end.	Skip rest of line.
`LRC0009` `EmptyTimestamp`	Warning	Literal `[]` group.	Group dropped; subsequent timestamps still parsed.

Encoding

Code	Severity	Trigger	Recovery
`LRC0010` `EncodingFallback`	Error	BOM detection failed and bytes are not valid UTF-8 — the configured `FallbackEncoding` was used.	Falls back; result still produced. The Error severity is intentional — fallback should be a conscious choice.

Metadata

Code	Severity	Trigger	Recovery
`LRC0020` `ConflictingMetadata`	Warning	Same key encountered twice with different values.	Last-wins on the typed accessor; both entries kept in `RawTags`.
`LRC0021` `DuplicateMetadata`	Info	Same key encountered twice with the same value.	Kept once on the typed accessor; both entries in `RawTags`.

Timestamp variants (tolerant accept)

The parser accepts these non-canonical forms; each emits an Info diagnostic so callers can detect them if they care.

Code	Severity	Trigger	Example
`LRC0030` `NonStandardTimestamp`	Info	Any timestamp that is not strictly `mm:ss.xx`.	`[00:01]`, `[00:01.500]`, `[00:01:50]`, `[1:02:33.45]`, `[01:23,45]`
`LRC0031` `TruncatedPrecision`	Info	Sub-millisecond fractional digits supplied. Fires for any fraction separator: `.`, `,`, or colon-fraction (`mm:ss:ffff` / `h:mm:ss:ffff`).	`[00:01.12345]`, `[01:23,4567]`, `[00:23:4567]` → millisecond truncation

Voice markers

The parser recognises Walaoke M: / F: / D: prefixes and propagates voice state across lines. There are no diagnostics emitted by the voice path today — unrecognised prefixes are simply treated as line text.

Lyric flow

Code	Severity	Trigger	Recovery
`LRC0050` `DroppedUntimedText`	Info	Free text on a line with no timestamp.	Text dropped from `Lines`.
`LRC0051` `TimestampWithoutText`	Warning	Timestamp followed only by line terminator.	Empty `Text` line still emitted.
`LRC0060` `LinesReordered`	Info	Lines were not in monotonic timestamp order; output sorted.	Document is sorted on the way out.
`LRC0070` `ImplausibleTimestamp`	Info	Timestamp exceeds `ImplausibleTimestampThreshold` (default 24 h).	Value still parsed and kept.
`LRC0080` `EmptyEnhancedWord`	Info	Enhanced line had no parseable `<ts>word` pairs.	Demoted to plain line preserving verbatim text.

Write-side

Code	Severity	Trigger	Recovery
`LRC0090` `UnrepresentableVoiceTransition`	Warning	A line transitions back to `LrcVoice.Default` after a non-default voice — Walaoke has no "clear voice" marker.	No marker emitted; the previously emitted voice persists in re-parsed output. Surface via `LrcWriter.ValidateForWrite`.

Tolerance / pragmatic

Code	Severity	Trigger	Recovery
`LRC0091` `BracketedContentTolerance`	Info	A `[…]` group after the first that fails to parse as a timestamp.	Treated as part of the line text.
`LRC0092` `Id3LanguagePrefixStripped`	Info	Input begins with `xx\\|\\|` (2-letter ISO 639-1) or `xxx\\|\\|` (3-letter ISO 639-2) language code, as some streaming-service exports use.	Prefix stripped before scanning.

Diagnostic cap

Code	Severity	Trigger	Recovery
`LRC0099` `MaxDiagnosticsReached`	Warning	More diagnostics fired than `MaxDiagnostics` allows.	Subsequent diagnostics suppressed. Set `MaxDiagnostics = 0` to suppress everything (including this notice).

Recipes

Suppress all diagnostics

var quiet = new LrcParseOptions { MaxDiagnostics = 0 };
var result = LrcParser.Parse(text, quiet);
result.Diagnostics.Length.ShouldBe(0);

Strict mode still throws on Error-severity codes even when the cap suppresses the diagnostic body.

Strict-mode error handling

var strict = new LrcParseOptions { Strictness = LrcStrictness.Strict };
try
{
    var result = LrcParser.Parse(text, strict);
}
catch (LrcParseException ex)
{
    Console.WriteLine($"{ex.FirstError?.Code} at L{ex.FirstError?.Line}");
    // ex.PartialResult holds whatever was parsed before the failure
}

Triage Error / Warning / Info

var byTier = result.Diagnostics
    .GroupBy(d => d.Severity)
    .ToDictionary(g => g.Key, g => g.Count());

Table of Contents