|
| 1 | +use crate::parser::JSONParser; |
| 2 | +use crate::serializer::JSONSerializer; |
| 3 | +use crate::token_fixer::JSONTokenFixer; |
| 4 | +use crate::tokenizer::JSONTokenizer; |
| 5 | +use serde_json::{Map, Value}; |
| 6 | + |
| 7 | +/// Main JSON auto-correction engine |
| 8 | +/// Combines tokenization, fixing, parsing, and serialization |
| 9 | +pub struct JSONAutoCorrector { |
| 10 | + tokenizer: JSONTokenizer, |
| 11 | + parser: JSONParser, |
| 12 | + serializer: JSONSerializer, |
| 13 | +} |
| 14 | + |
| 15 | +impl JSONAutoCorrector { |
| 16 | + pub fn new() -> Self { |
| 17 | + Self { |
| 18 | + tokenizer: JSONTokenizer::new(), |
| 19 | + parser: JSONParser::new(), |
| 20 | + serializer: JSONSerializer::new(), |
| 21 | + } |
| 22 | + } |
| 23 | + |
| 24 | + /// Auto-correct broken JSON string |
| 25 | + /// This is the main entry point for JSON correction |
| 26 | + pub fn autocorrect(&mut self, input: &str) -> String { |
| 27 | + // Step 1: Tokenize the input |
| 28 | + let tokens = self.tokenizer.tokenize(input); |
| 29 | + |
| 30 | + // Step 2: Fix token stream (add missing brackets, etc.) |
| 31 | + let fixed_tokens = JSONTokenFixer::fix_tokens(tokens); |
| 32 | + |
| 33 | + // Step 3: Parse tokens into JSON value |
| 34 | + let parse_result = self.parser.parse(&fixed_tokens, 0); |
| 35 | + |
| 36 | + // Step 4: Get the result value or default to empty object |
| 37 | + let result = match parse_result.value { |
| 38 | + Some(value) => value, |
| 39 | + None => Value::Object(Map::new()), |
| 40 | + }; |
| 41 | + |
| 42 | + // Step 5: Serialize back to JSON string |
| 43 | + self.serializer.serialize(&result) |
| 44 | + } |
| 45 | + |
| 46 | + /// Auto-correct and return pretty-printed JSON |
| 47 | + pub fn autocorrect_pretty(&mut self, input: &str) -> String { |
| 48 | + let tokens = self.tokenizer.tokenize(input); |
| 49 | + let fixed_tokens = JSONTokenFixer::fix_tokens(tokens); |
| 50 | + let parse_result = self.parser.parse(&fixed_tokens, 0); |
| 51 | + |
| 52 | + let result = match parse_result.value { |
| 53 | + Some(value) => value, |
| 54 | + None => Value::Object(Map::new()), |
| 55 | + }; |
| 56 | + |
| 57 | + self.serializer.serialize_pretty(&result) |
| 58 | + } |
| 59 | +} |
| 60 | + |
| 61 | +impl Default for JSONAutoCorrector { |
| 62 | + fn default() -> Self { |
| 63 | + Self::new() |
| 64 | + } |
| 65 | +} |
| 66 | + |
| 67 | +#[cfg(test)] |
| 68 | +mod tests { |
| 69 | + use super::*; |
| 70 | + |
| 71 | + #[test] |
| 72 | + fn test_autocorrect_missing_brace() { |
| 73 | + let mut corrector = JSONAutoCorrector::new(); |
| 74 | + let result = corrector.autocorrect(r#"{"key":42"#); |
| 75 | + assert_eq!(result, r#"{"key":42}"#); |
| 76 | + } |
| 77 | + |
| 78 | + #[test] |
| 79 | + fn test_autocorrect_missing_bracket() { |
| 80 | + let mut corrector = JSONAutoCorrector::new(); |
| 81 | + let result = corrector.autocorrect(r#"{"key":[1,2,3}"#); |
| 82 | + assert_eq!(result, r#"{"key":[1,2,3]}"#); |
| 83 | + } |
| 84 | + |
| 85 | + #[test] |
| 86 | + fn test_autocorrect_missing_comma() { |
| 87 | + let mut corrector = JSONAutoCorrector::new(); |
| 88 | + let result = corrector.autocorrect(r#"{"key1":42 "key2":true}"#); |
| 89 | + assert_eq!(result, r#"{"key1":42,"key2":true}"#); |
| 90 | + } |
| 91 | + |
| 92 | + #[test] |
| 93 | + fn test_autocorrect_incomplete_string() { |
| 94 | + let mut corrector = JSONAutoCorrector::new(); |
| 95 | + let result = corrector.autocorrect(r#"{"title":"Hello"#); |
| 96 | + assert_eq!(result, r#"{"title":"Hello"}"#); |
| 97 | + } |
| 98 | + |
| 99 | + #[test] |
| 100 | + fn test_autocorrect_partial_literals() { |
| 101 | + let mut corrector = JSONAutoCorrector::new(); |
| 102 | + |
| 103 | + let result = corrector.autocorrect(r#"{"flag":tr}"#); |
| 104 | + assert_eq!(result, r#"{"flag":true}"#); |
| 105 | + |
| 106 | + let result = corrector.autocorrect(r#"{"flag":fals}"#); |
| 107 | + assert_eq!(result, r#"{"flag":false}"#); |
| 108 | + |
| 109 | + let result = corrector.autocorrect(r#"{"value":nul}"#); |
| 110 | + assert_eq!(result, r#"{"value":null}"#); |
| 111 | + } |
| 112 | + |
| 113 | + #[test] |
| 114 | + fn test_autocorrect_trailing_comma() { |
| 115 | + let mut corrector = JSONAutoCorrector::new(); |
| 116 | + let result = corrector.autocorrect(r#"{"key1":1,"key2":2,}"#); |
| 117 | + assert_eq!(result, r#"{"key1":1,"key2":2}"#); |
| 118 | + } |
| 119 | + |
| 120 | + #[test] |
| 121 | + fn test_autocorrect_nested_incomplete() { |
| 122 | + let mut corrector = JSONAutoCorrector::new(); |
| 123 | + let result = corrector.autocorrect(r#"{"outer":{"inner":[1,2,3"#); |
| 124 | + assert_eq!(result, r#"{"outer":{"inner":[1,2,3]}}"#); |
| 125 | + } |
| 126 | +} |
| 127 | + |
0 commit comments