|
| 1 | +using System.Text; |
| 2 | +using System.Text.Json; |
| 3 | + |
| 4 | +namespace Juggle.Domain.Engine.NodeExecutors; |
| 5 | + |
| 6 | +/// <summary> |
| 7 | +/// 数据提取节点：从输入文本中提取内容。 |
| 8 | +/// 提取类型：json（首个 JSON 对象/数组）、code（代码块）、keyword（关键字后内容）、 |
| 9 | +/// between（开始/结束标志之间）、length（按偏移与长度截取）。 |
| 10 | +/// </summary> |
| 11 | +public class DataExtractNodeExecutor : INodeExecutor |
| 12 | +{ |
| 13 | + public Task<string?> ExecuteAsync(FlowNode node, FlowContext context) |
| 14 | + { |
| 15 | + var cfg = node.DataExtractConfig |
| 16 | + ?? throw new InvalidOperationException($"数据提取节点 [{node.Key}] 未配置 dataExtractConfig。"); |
| 17 | + |
| 18 | + var raw = string.IsNullOrWhiteSpace(cfg.Input) ? "" : context.GetVariable(cfg.Input)?.ToString() ?? ""; |
| 19 | + var type = (cfg.ExtractType ?? "json").ToLower(); |
| 20 | + |
| 21 | + object? result = type switch |
| 22 | + { |
| 23 | + "json" => ExtractJson(raw), |
| 24 | + "code" => ExtractCode(raw), |
| 25 | + "keyword" => ExtractAfterKeyword(raw, cfg.Keyword ?? ""), |
| 26 | + "between" => ExtractBetween(raw, cfg.StartFlag ?? "", cfg.EndFlag ?? ""), |
| 27 | + "length" => ExtractLength(raw, cfg.Offset, cfg.Length), |
| 28 | + _ => raw |
| 29 | + }; |
| 30 | + |
| 31 | + if (!string.IsNullOrWhiteSpace(cfg.Output)) |
| 32 | + { |
| 33 | + switch ((cfg.OutputTargetType ?? "VARIABLE").ToUpper()) |
| 34 | + { |
| 35 | + case "OUTPUT": |
| 36 | + context.SetOutputParameter(cfg.Output, result); |
| 37 | + break; |
| 38 | + default: |
| 39 | + context.SetVariable(cfg.Output, result); |
| 40 | + break; |
| 41 | + } |
| 42 | + } |
| 43 | + |
| 44 | + return Task.FromResult(node.Outgoings.FirstOrDefault()); |
| 45 | + } |
| 46 | + |
| 47 | + /// <summary>提取首个 JSON 对象/数组（返回解析对象，失败返回原文）。</summary> |
| 48 | + public static object? ExtractJson(string raw) |
| 49 | + { |
| 50 | + var start = raw.IndexOfAny(new[] { '{', '[' }); |
| 51 | + if (start < 0) return raw; |
| 52 | + var end = FindBalancedEnd(raw, start); |
| 53 | + if (end < 0) return raw; |
| 54 | + var json = raw.Substring(start, end - start + 1); |
| 55 | + try |
| 56 | + { |
| 57 | + using var doc = JsonDocument.Parse(json); |
| 58 | + return FileParseNodeExecutor.CloneToNative(doc.RootElement); |
| 59 | + } |
| 60 | + catch { return json; } |
| 61 | + } |
| 62 | + |
| 63 | + /// <summary>从 { 或 [ 开始找平衡闭合位置。</summary> |
| 64 | + private static int FindBalancedEnd(string s, int start) |
| 65 | + { |
| 66 | + var openers = new Stack<char>(); |
| 67 | + for (int i = start; i < s.Length; i++) |
| 68 | + { |
| 69 | + var ch = s[i]; |
| 70 | + if (ch == '{' || ch == '[') |
| 71 | + { |
| 72 | + openers.Push(ch); |
| 73 | + } |
| 74 | + else if (ch == '}' || ch == ']') |
| 75 | + { |
| 76 | + if (openers.Count == 0) return -1; |
| 77 | + var opener = openers.Pop(); |
| 78 | + if ((ch == '}' && opener != '{') || (ch == ']' && opener != '[')) return -1; |
| 79 | + if (openers.Count == 0) return i; |
| 80 | + } |
| 81 | + } |
| 82 | + return -1; |
| 83 | + } |
| 84 | + |
| 85 | + /// <summary>提取代码块内容（```lang ... ```）。</summary> |
| 86 | + public static string ExtractCode(string raw) |
| 87 | + { |
| 88 | + var fenceIdx = raw.IndexOf("```", StringComparison.Ordinal); |
| 89 | + if (fenceIdx < 0) return raw; |
| 90 | + var contentStart = raw.IndexOf('\n', fenceIdx); |
| 91 | + if (contentStart < 0) return raw; |
| 92 | + var endIdx = raw.IndexOf("```", contentStart + 1, StringComparison.Ordinal); |
| 93 | + if (endIdx < 0) return raw.Substring(contentStart + 1).Trim(); |
| 94 | + return raw.Substring(contentStart + 1, endIdx - contentStart - 1).Trim(); |
| 95 | + } |
| 96 | + |
| 97 | + /// <summary>提取关键字之后的内容（跳过 : ：= 空格等分隔符，取到行尾）。</summary> |
| 98 | + public static string ExtractAfterKeyword(string raw, string keyword) |
| 99 | + { |
| 100 | + if (string.IsNullOrEmpty(keyword)) return raw; |
| 101 | + var idx = raw.IndexOf(keyword, StringComparison.Ordinal); |
| 102 | + if (idx < 0) return ""; |
| 103 | + var pos = idx + keyword.Length; |
| 104 | + while (pos < raw.Length && (raw[pos] == ':' || raw[pos] == '：' || raw[pos] == '=' || raw[pos] == ' ' || raw[pos] == '\t')) |
| 105 | + pos++; |
| 106 | + var lineEnd = raw.IndexOf('\n', pos); |
| 107 | + return lineEnd < 0 ? raw.Substring(pos).Trim() : raw.Substring(pos, lineEnd - pos).Trim(); |
| 108 | + } |
| 109 | + |
| 110 | + /// <summary>提取开始标志与结束标志之间的内容（首次出现）。</summary> |
| 111 | + public static string ExtractBetween(string raw, string startFlag, string endFlag) |
| 112 | + { |
| 113 | + if (string.IsNullOrEmpty(startFlag)) return raw; |
| 114 | + var sIdx = raw.IndexOf(startFlag, StringComparison.Ordinal); |
| 115 | + if (sIdx < 0) return ""; |
| 116 | + var start = sIdx + startFlag.Length; |
| 117 | + if (string.IsNullOrEmpty(endFlag)) return raw.Substring(start); |
| 118 | + var eIdx = raw.IndexOf(endFlag, start, StringComparison.Ordinal); |
| 119 | + return eIdx < 0 ? raw.Substring(start) : raw.Substring(start, eIdx - start); |
| 120 | + } |
| 121 | + |
| 122 | + /// <summary>按偏移与长度截取。</summary> |
| 123 | + public static string ExtractLength(string raw, int offset, int length) |
| 124 | + { |
| 125 | + var off = Math.Max(0, offset); |
| 126 | + if (off >= raw.Length) return ""; |
| 127 | + var len = length > 0 ? length : raw.Length - off; |
| 128 | + return raw.Substring(off, Math.Min(len, raw.Length - off)); |
| 129 | + } |
| 130 | +} |
0 commit comments