PlotDirector/PlotLine/Services/StoryIntelligenceJsonRepair.cs

497 lines
14 KiB
C#

using System.Text;
using System.Text.Json;
using System.Text.RegularExpressions;
namespace PlotLine.Services;
public static partial class StoryIntelligenceJsonRepair
{
public static StoryIntelligenceJsonRepairResult Repair(string value)
{
var warnings = new List<string>();
var extraction = ExtractJsonObject(value);
if (!extraction.Success)
{
return StoryIntelligenceJsonRepairResult.Failed(value, extraction.ErrorMessage);
}
warnings.AddRange(extraction.Warnings);
var numericWhitespaceRepaired = RepairWhitespaceInDecimalNumbers(extraction.Json, warnings);
var numericRepaired = RepairTrailingDecimalNumbers(numericWhitespaceRepaired, warnings);
var slashRepaired = RepairInvalidBackslashesInStrings(numericRepaired, warnings);
if (!IsValidJson(slashRepaired, out var error))
{
return StoryIntelligenceJsonRepairResult.Failed(slashRepaired, error);
}
return new StoryIntelligenceJsonRepairResult(
RepairedJson: slashRepaired,
RepairApplied: warnings.Count > 0,
Warnings: warnings,
ErrorMessage: null);
}
public static StoryIntelligenceJsonExtractionResult ExtractJsonObject(string value)
{
var warnings = new List<string>();
if (string.IsNullOrWhiteSpace(value))
{
return StoryIntelligenceJsonExtractionResult.Failed(value, "Output was empty.");
}
var candidate = StripMarkdownFence(value.Trim(), warnings);
candidate = TrimToJsonObject(candidate, warnings);
if (string.IsNullOrWhiteSpace(candidate))
{
return StoryIntelligenceJsonExtractionResult.Failed(value, "No complete JSON object was found.");
}
return new StoryIntelligenceJsonExtractionResult(
Json: candidate,
Warnings: warnings,
ErrorMessage: null);
}
private static string StripMarkdownFence(string value, List<string> warnings)
{
var match = MarkdownFenceRegex().Match(value);
if (!match.Success)
{
return value;
}
warnings.Add("Removed markdown code fence.");
return match.Groups["json"].Value.Trim();
}
private static string TrimToJsonObject(string value, List<string> warnings)
{
var start = value.IndexOf('{');
if (start < 0)
{
return string.Empty;
}
var depth = 0;
var inString = false;
var escaped = false;
for (var i = start; i < value.Length; i++)
{
var ch = value[i];
if (inString)
{
if (escaped)
{
escaped = false;
}
else if (ch == '\\')
{
escaped = true;
}
else if (ch == '"')
{
inString = false;
}
continue;
}
if (ch == '"')
{
inString = true;
}
else if (ch == '{')
{
depth++;
}
else if (ch == '}')
{
depth--;
if (depth == 0)
{
if (start > 0 || i < value.Length - 1)
{
warnings.Add("Trimmed text outside the JSON object.");
}
return value[start..(i + 1)].Trim();
}
}
}
return string.Empty;
}
private static string RepairTrailingDecimalNumbers(string value, List<string> warnings)
{
var repaired = TrailingDecimalRegex().Replace(value, match =>
{
warnings.Add("Normalised trailing decimal number to valid JSON number.");
return $"{match.Groups["prefix"].Value}{match.Groups["number"].Value}.0";
});
return repaired;
}
private static string RepairWhitespaceInDecimalNumbers(string value, List<string> warnings)
{
var builder = new StringBuilder(value.Length);
var inString = false;
var escaped = false;
var changed = false;
for (var i = 0; i < value.Length; i++)
{
var ch = value[i];
if (inString)
{
builder.Append(ch);
if (escaped)
{
escaped = false;
}
else if (ch == '\\')
{
escaped = true;
}
else if (ch == '"')
{
inString = false;
}
continue;
}
if (ch == '"')
{
inString = true;
builder.Append(ch);
continue;
}
if ((ch == '-' || char.IsDigit(ch))
&& IsJsonNumberValueStart(value, i)
&& TryReadWhitespaceDecimalNumber(value, i, out var repairedNumber, out var endIndex))
{
builder.Append(repairedNumber);
i = endIndex - 1;
changed = true;
continue;
}
builder.Append(ch);
}
if (changed)
{
warnings.Add("Normalised whitespace inside decimal JSON number.");
}
return builder.ToString();
}
private static bool TryReadWhitespaceDecimalNumber(string value, int startIndex, out string repairedNumber, out int endIndex)
{
repairedNumber = string.Empty;
endIndex = startIndex;
var index = startIndex;
var builder = new StringBuilder();
var removedWhitespace = false;
if (value[index] == '-')
{
builder.Append(value[index]);
index++;
if (index >= value.Length || !char.IsDigit(value[index]))
{
return false;
}
}
var integerStart = index;
while (index < value.Length && char.IsDigit(value[index]))
{
builder.Append(value[index]);
index++;
}
if (index == integerStart)
{
return false;
}
var beforeDecimalWhitespaceStart = index;
while (index < value.Length && char.IsWhiteSpace(value[index]))
{
index++;
}
if (index >= value.Length || value[index] != '.')
{
return false;
}
removedWhitespace = index > beforeDecimalWhitespaceStart;
builder.Append('.');
index++;
var afterDecimalWhitespaceStart = index;
while (index < value.Length && char.IsWhiteSpace(value[index]))
{
index++;
}
removedWhitespace = removedWhitespace || index > afterDecimalWhitespaceStart;
var fractionStart = index;
while (index < value.Length && char.IsDigit(value[index]))
{
builder.Append(value[index]);
index++;
}
if (!removedWhitespace || index == fractionStart || !IsJsonValueEnd(value, index))
{
return false;
}
repairedNumber = builder.ToString();
endIndex = index;
return true;
}
private static bool IsJsonNumberValueStart(string value, int startIndex)
{
for (var i = startIndex - 1; i >= 0; i--)
{
if (char.IsWhiteSpace(value[i]))
{
continue;
}
return value[i] is ':' or '[' or ',';
}
return false;
}
private static bool IsJsonValueEnd(string value, int index)
{
for (var i = index; i < value.Length; i++)
{
if (char.IsWhiteSpace(value[i]))
{
continue;
}
return value[i] is ',' or '}' or ']';
}
return true;
}
private static string RepairInvalidBackslashesInStrings(string value, List<string> warnings)
{
var builder = new StringBuilder(value.Length);
var inString = false;
var changed = false;
for (var i = 0; i < value.Length; i++)
{
var ch = value[i];
if (!inString)
{
builder.Append(ch);
if (ch == '"')
{
inString = true;
}
continue;
}
if (ch == '"')
{
inString = false;
builder.Append(ch);
continue;
}
if (ch == '\\')
{
var next = i + 1 < value.Length ? value[i + 1] : '\0';
if (IsValidJsonEscape(next))
{
builder.Append(ch);
if (next != '\0')
{
builder.Append(next);
i++;
}
continue;
}
builder.Append(@"\\");
changed = true;
continue;
}
builder.Append(ch);
}
if (changed)
{
warnings.Add("Escaped invalid backslash inside string value.");
}
return builder.ToString();
}
private static bool IsValidJsonEscape(char value)
=> value is '"' or '\\' or '/' or 'b' or 'f' or 'n' or 'r' or 't' or 'u';
private static bool IsValidJson(string value, out string? error)
{
try
{
using var _ = JsonDocument.Parse(value);
error = null;
return true;
}
catch (JsonException ex)
{
error = ex.Message;
return false;
}
}
[GeneratedRegex("^```(?:json)?\\s*(?<json>.*?)\\s*```$", RegexOptions.Singleline | RegexOptions.IgnoreCase)]
private static partial Regex MarkdownFenceRegex();
[GeneratedRegex("(?<prefix>:\\s*)(?<number>-?\\d+)\\.\\s*(?=[,}\\]])")]
private static partial Regex TrailingDecimalRegex();
}
public static class StoryIntelligenceAiJsonParser
{
public static StoryIntelligenceAiJsonParseResult<T> Parse<T>(string rawAssistantOutput, JsonSerializerOptions jsonOptions)
{
var extraction = StoryIntelligenceJsonRepair.ExtractJsonObject(rawAssistantOutput);
if (!extraction.Success)
{
return StoryIntelligenceAiJsonParseResult<T>.Failed(rawAssistantOutput, extraction.ErrorMessage ?? "No complete JSON object was found.");
}
try
{
var parsed = JsonSerializer.Deserialize<T>(extraction.Json, jsonOptions);
return StoryIntelligenceAiJsonParseResult<T>.Ok(
extraction.Json,
repairedJson: null,
parsed,
extraction.Warnings);
}
catch (JsonException strictEx)
{
var repair = StoryIntelligenceJsonRepair.Repair(rawAssistantOutput);
if (!repair.Success)
{
return StoryIntelligenceAiJsonParseResult<T>.Failed(
extraction.Json,
repair.ErrorMessage ?? strictEx.Message,
strictEx.Message,
strictEx.Path);
}
try
{
var parsed = JsonSerializer.Deserialize<T>(repair.RepairedJson, jsonOptions);
var warnings = BuildRepairWarnings(repair.Warnings, strictEx.Path);
return StoryIntelligenceAiJsonParseResult<T>.Ok(
extraction.Json,
repair.RepairedJson,
parsed,
warnings,
strictEx.Message,
strictEx.Path);
}
catch (JsonException repairedEx)
{
return StoryIntelligenceAiJsonParseResult<T>.Failed(extraction.Json, repairedEx.Message, strictEx.Message, strictEx.Path);
}
}
}
private static IReadOnlyList<string> BuildRepairWarnings(IReadOnlyList<string> repairWarnings, string? strictErrorPath)
{
var warnings = new List<string>();
foreach (var warning in repairWarnings)
{
if (warning.Contains("trailing decimal", StringComparison.OrdinalIgnoreCase)
|| warning.Contains("whitespace inside decimal", StringComparison.OrdinalIgnoreCase))
{
warnings.Add(string.IsNullOrWhiteSpace(strictErrorPath)
? "Repaired malformed AI JSON numeric value."
: $"Repaired malformed AI JSON numeric value at {strictErrorPath}.");
}
else
{
warnings.Add(warning);
}
}
return warnings;
}
}
public sealed record StoryIntelligenceJsonRepairResult(
string RepairedJson,
bool RepairApplied,
IReadOnlyList<string> Warnings,
string? ErrorMessage)
{
public bool Success => string.IsNullOrWhiteSpace(ErrorMessage);
public static StoryIntelligenceJsonRepairResult Failed(string value, string? errorMessage)
=> new(value, false, [], errorMessage ?? "JSON repair failed.");
}
public sealed record StoryIntelligenceJsonExtractionResult(
string Json,
IReadOnlyList<string> Warnings,
string? ErrorMessage)
{
public bool Success => string.IsNullOrWhiteSpace(ErrorMessage);
public static StoryIntelligenceJsonExtractionResult Failed(string value, string? errorMessage)
=> new(value, [], errorMessage ?? "JSON extraction failed.");
}
public sealed record StoryIntelligenceAiJsonParseResult<T>(
bool Success,
string RawJson,
string? RepairedJson,
T? Parsed,
IReadOnlyList<string> Warnings,
string? ErrorMessage,
string? InitialErrorMessage = null,
string? InitialErrorPath = null)
{
public static StoryIntelligenceAiJsonParseResult<T> Ok(
string rawJson,
string? repairedJson,
T? parsed,
IReadOnlyList<string> warnings,
string? initialErrorMessage = null,
string? initialErrorPath = null)
=> new(true, rawJson, repairedJson, parsed, warnings, null, initialErrorMessage, initialErrorPath);
public static StoryIntelligenceAiJsonParseResult<T> Failed(
string rawJson,
string errorMessage,
string? initialErrorMessage = null,
string? initialErrorPath = null)
=> new(false, rawJson, null, default, [], errorMessage, initialErrorMessage, initialErrorPath);
}