|
| 1 | +use super::LanguageExtractor; |
| 2 | +use crate::extractor::models::{ChunkType, Language}; |
| 3 | +use crate::{GitTypeError, Result}; |
| 4 | +use tree_sitter::{Node, Parser}; |
| 5 | + |
| 6 | +pub struct HaskellExtractor; |
| 7 | + |
| 8 | +impl LanguageExtractor for HaskellExtractor { |
| 9 | + fn language(&self) -> Language { |
| 10 | + Language::Haskell |
| 11 | + } |
| 12 | + |
| 13 | + fn file_extensions(&self) -> &[&str] { |
| 14 | + &["hs", "lhs"] |
| 15 | + } |
| 16 | + |
| 17 | + fn tree_sitter_language(&self) -> tree_sitter::Language { |
| 18 | + tree_sitter_haskell::LANGUAGE.into() |
| 19 | + } |
| 20 | + |
| 21 | + fn query_patterns(&self) -> &str { |
| 22 | + r#" |
| 23 | + ; Function definitions with name extraction |
| 24 | + (function name: (variable) @function.name) @function.definition |
| 25 | + (function (variable) @function.name) @function.named |
| 26 | +
|
| 27 | + ; Type signatures with name extraction |
| 28 | + (signature name: (variable) @signature.name) @signature.definition |
| 29 | + (signature (variable) @signature.name) @signature.named |
| 30 | +
|
| 31 | + ; Module declarations |
| 32 | + (module) @module.definition |
| 33 | +
|
| 34 | + ; Import declarations (if present) |
| 35 | + (import) @import.definition |
| 36 | +
|
| 37 | + ; General declarations with names |
| 38 | + (declaration name: (variable) @declaration.name) @declaration.definition |
| 39 | + (decl name: (variable) @decl.name) @decl.definition |
| 40 | +
|
| 41 | + ; Type elements |
| 42 | + (type name: (variable) @type.name) @type.definition |
| 43 | +
|
| 44 | + ; Constructors |
| 45 | + (constructor) @constructor.definition |
| 46 | +
|
| 47 | + ; Fallback patterns for completeness |
| 48 | + (function) @function.basic |
| 49 | + (signature) @signature.basic |
| 50 | + (declaration) @declaration.basic |
| 51 | + "# |
| 52 | + } |
| 53 | + |
| 54 | + fn comment_query(&self) -> &str { |
| 55 | + "[(comment)] @comment" |
| 56 | + } |
| 57 | + |
| 58 | + fn capture_name_to_chunk_type(&self, capture_name: &str) -> Option<ChunkType> { |
| 59 | + match capture_name { |
| 60 | + "function.definition" | "function.named" | "function.basic" => { |
| 61 | + Some(ChunkType::Function) |
| 62 | + } |
| 63 | + "signature.definition" | "signature.named" | "signature.basic" => { |
| 64 | + Some(ChunkType::Function) |
| 65 | + } |
| 66 | + "module.definition" => Some(ChunkType::Module), |
| 67 | + "import.definition" => Some(ChunkType::Module), |
| 68 | + "declaration.definition" | "declaration.basic" => Some(ChunkType::Function), |
| 69 | + "decl.definition" => Some(ChunkType::Function), |
| 70 | + "type.definition" => Some(ChunkType::TypeAlias), |
| 71 | + "constructor.definition" => Some(ChunkType::Struct), |
| 72 | + _ => None, |
| 73 | + } |
| 74 | + } |
| 75 | + |
| 76 | + fn extract_name(&self, node: Node, source_code: &str, capture_name: &str) -> Option<String> { |
| 77 | + match capture_name { |
| 78 | + name if name.contains("function") => self.extract_function_name(node, source_code), |
| 79 | + name if name.contains("signature") => self.extract_signature_name(node, source_code), |
| 80 | + name if name.contains("declaration") || name.contains("decl") => { |
| 81 | + self.extract_declaration_name(node, source_code) |
| 82 | + } |
| 83 | + name if name.contains("type") => self.extract_type_name(node, source_code), |
| 84 | + name if name.contains("constructor") => { |
| 85 | + self.extract_constructor_name(node, source_code) |
| 86 | + } |
| 87 | + name if name.contains("module") => self.extract_module_name(node, source_code), |
| 88 | + name if name.contains("import") => self.extract_import_name(node, source_code), |
| 89 | + _ => self.extract_name_from_node(node, source_code), |
| 90 | + } |
| 91 | + } |
| 92 | +} |
| 93 | + |
| 94 | +impl HaskellExtractor { |
| 95 | + fn extract_function_name(&self, node: Node, source_code: &str) -> Option<String> { |
| 96 | + // Look for variable node in function definition |
| 97 | + self.find_child_by_kind(node, source_code, "variable") |
| 98 | + } |
| 99 | + |
| 100 | + fn extract_signature_name(&self, node: Node, source_code: &str) -> Option<String> { |
| 101 | + // Look for variable node in type signature |
| 102 | + self.find_child_by_kind(node, source_code, "variable") |
| 103 | + } |
| 104 | + |
| 105 | + fn extract_type_name(&self, node: Node, source_code: &str) -> Option<String> { |
| 106 | + // Look for type node in data/newtype declaration |
| 107 | + self.find_child_by_kind(node, source_code, "type") |
| 108 | + } |
| 109 | + |
| 110 | + fn extract_declaration_name(&self, node: Node, source_code: &str) -> Option<String> { |
| 111 | + // Look for variable node in general declarations |
| 112 | + self.find_child_by_kind(node, source_code, "variable") |
| 113 | + } |
| 114 | + |
| 115 | + fn extract_constructor_name(&self, node: Node, source_code: &str) -> Option<String> { |
| 116 | + // Look for constructor node |
| 117 | + self.find_child_by_kind(node, source_code, "constructor") |
| 118 | + } |
| 119 | + |
| 120 | + fn extract_module_name(&self, node: Node, source_code: &str) -> Option<String> { |
| 121 | + // Look for module_name node |
| 122 | + self.find_child_by_kind(node, source_code, "module_name") |
| 123 | + } |
| 124 | + |
| 125 | + fn extract_import_name(&self, node: Node, source_code: &str) -> Option<String> { |
| 126 | + // Look for module_name node in import |
| 127 | + self.find_child_by_kind(node, source_code, "module_name") |
| 128 | + } |
| 129 | + |
| 130 | + fn find_child_by_kind(&self, node: Node, source_code: &str, kind: &str) -> Option<String> { |
| 131 | + let mut cursor = node.walk(); |
| 132 | + if cursor.goto_first_child() { |
| 133 | + loop { |
| 134 | + let child = cursor.node(); |
| 135 | + if child.kind() == kind { |
| 136 | + return child |
| 137 | + .utf8_text(source_code.as_bytes()) |
| 138 | + .ok() |
| 139 | + .map(|s| s.to_string()); |
| 140 | + } |
| 141 | + // Recursively search in child nodes |
| 142 | + if let Some(name) = self.find_child_by_kind(child, source_code, kind) { |
| 143 | + return Some(name); |
| 144 | + } |
| 145 | + if !cursor.goto_next_sibling() { |
| 146 | + break; |
| 147 | + } |
| 148 | + } |
| 149 | + } |
| 150 | + None |
| 151 | + } |
| 152 | + |
| 153 | + fn extract_name_from_node(&self, node: Node, source_code: &str) -> Option<String> { |
| 154 | + let mut cursor = node.walk(); |
| 155 | + if cursor.goto_first_child() { |
| 156 | + loop { |
| 157 | + let child = cursor.node(); |
| 158 | + match child.kind() { |
| 159 | + "variable" | "type" | "module_name" | "constructor" => { |
| 160 | + return child |
| 161 | + .utf8_text(source_code.as_bytes()) |
| 162 | + .ok() |
| 163 | + .map(|s| s.to_string()); |
| 164 | + } |
| 165 | + _ => { |
| 166 | + // Recursively search in child nodes |
| 167 | + if let Some(name) = self.extract_name_from_node(child, source_code) { |
| 168 | + return Some(name); |
| 169 | + } |
| 170 | + } |
| 171 | + } |
| 172 | + if !cursor.goto_next_sibling() { |
| 173 | + break; |
| 174 | + } |
| 175 | + } |
| 176 | + } |
| 177 | + None |
| 178 | + } |
| 179 | + |
| 180 | + pub fn create_parser() -> Result<Parser> { |
| 181 | + let mut parser = Parser::new(); |
| 182 | + parser |
| 183 | + .set_language(&tree_sitter_haskell::LANGUAGE.into()) |
| 184 | + .map_err(|e| { |
| 185 | + GitTypeError::ExtractionFailed(format!("Failed to set Haskell language: {}", e)) |
| 186 | + })?; |
| 187 | + Ok(parser) |
| 188 | + } |
| 189 | +} |
0 commit comments