tree_sitter_language_pack 1.2.1 → 1.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e39e2e8f16d680a03503e0fc7a5340cc8900a9c11250d423dcafc70dbb38a308
-  data.tar.gz: 22863be51d8d97eef161733d4864edac0a07ad96cef21672b3e4da9950c039c4
+  metadata.gz: 6db995bcd67cec0642ed17f3ce6da23f5a8f4dc71118f2cc06cdbc3ce6e5b8b5
+  data.tar.gz: c2fee6f228184924cf2b7a908aec4d9e7dba3a6e607d0dfbb40de6be0efa296a
 SHA512:
-  metadata.gz: 6e9afb8bbf4644808bac37295187463bbc9d8f702fc6e3e239a530044e31d95e57f966b1d97e6c8077832102edc439c19bbfc0261cb7aaedd16fba931b6e014b
-  data.tar.gz: 9248459136e6233d742c6832618663cc5bb209d0b56e583abcc45f0ba1aae9bc3ebd56e75cba149dbf92b8e061fa3c8db9637a56770c4f113a3b3a530edd42f6
+  metadata.gz: 2c924f922d81454f9b02727e9f4883d31d80de10b4823cdff445f1b590fb60cc4851b86d57612a64ee4a52f65ff7ec962b16b85fe4ac55453d178afe34ae4169
+  data.tar.gz: e3323b3ed5f2f61fd4942ab6157d34ef998228d0e668b5b01873e2c57f8ab5e9ff5566df2b8ab76023f2a9e529a44ee255521ca9bb799a1ebff81460435f7427

data/src/lib.rs

@@ -1,4 +1,4 @@
-use magnus::{Error, Ruby, function, method, prelude::*};
+use magnus::{Error, IntoValue, Ruby, Value as RbValue, function, method, prelude::*};
 use std::sync::Mutex;
 
 /// Wraps a tree-sitter Tree for safe sharing across the Ruby boundary.
@@ -84,20 +84,93 @@ fn parse_string(ruby: &Ruby, language: String, source: String) -> Result<TreeWra
     Ok(TreeWrapper(Mutex::new(tree)))
 }
 
-/// Unified process method that accepts a JSON config string and returns a JSON result string.
+/// Convert a serde_json::Value to a native Ruby object (Hash, Array, String, Integer, Float, true/false, nil).
+fn json_value_to_ruby(ruby: &Ruby, value: &serde_json::Value) -> Result<RbValue, Error> {
+    Ok(match value {
+        serde_json::Value::Null => ruby.qnil().as_value(),
+        serde_json::Value::Bool(b) => (*b).into_value_with(ruby),
+        serde_json::Value::Number(n) => {
+            if let Some(i) = n.as_i64() {
+                i.into_value_with(ruby)
+            } else if let Some(u) = n.as_u64() {
+                u.into_value_with(ruby)
+            } else if let Some(f) = n.as_f64() {
+                f.into_value_with(ruby)
+            } else {
+                ruby.qnil().as_value()
+            }
+        }
+        serde_json::Value::String(s) => s.as_str().into_value_with(ruby),
+        serde_json::Value::Array(arr) => {
+            let ary = ruby.ary_new_capa(arr.len());
+            for item in arr {
+                ary.push(json_value_to_ruby(ruby, item)?)?;
+            }
+            ary.as_value()
+        }
+        serde_json::Value::Object(map) => {
+            let hash = ruby.hash_new();
+            for (k, v) in map {
+                hash.aset(k.as_str(), json_value_to_ruby(ruby, v)?)?;
+            }
+            hash.as_value()
+        }
+    })
+}
+
+/// Unified process method that accepts a JSON config string and returns a native Ruby Hash.
 ///
 /// The config JSON must contain at least `"language"`. Optional fields:
 /// - `structure`, `imports`, `exports`, `comments`, `docstrings`, `symbols`, `diagnostics` (booleans, default true)
 /// - `chunk_max_size` (integer or null, default null meaning no chunking)
-fn process(ruby: &Ruby, source: String, config_json: String) -> Result<String, Error> {
+fn process(ruby: &Ruby, source: String, config_json: String) -> Result<RbValue, Error> {
     let core_config: tree_sitter_language_pack::ProcessConfig = serde_json::from_str(&config_json)
         .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("invalid config JSON: {e}")))?;
 
     let result = tree_sitter_language_pack::process(&source, &core_config)
         .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("{e}")))?;
 
-    serde_json::to_string(&result)
-        .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("serialization failed: {e}")))
+    let json_value = serde_json::to_value(&result)
+        .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("serialization failed: {e}")))?;
+    json_value_to_ruby(ruby, &json_value)
+}
+
+/// Extract patterns from source code using a JSON configuration.
+///
+/// The config JSON must contain:
+/// - `language` (string): the language name
+/// - `patterns` (object): named patterns to run, each with a `query` field
+///
+/// Returns a native Ruby Hash with extraction results.
+fn extract(ruby: &Ruby, source: String, config_json: String) -> Result<RbValue, Error> {
+    let config: tree_sitter_language_pack::ExtractionConfig = serde_json::from_str(&config_json)
+        .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("invalid config JSON: {e}")))?;
+
+    let result = tree_sitter_language_pack::extract_patterns(&source, &config)
+        .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("{e}")))?;
+
+    let json_value = serde_json::to_value(&result)
+        .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("serialization failed: {e}")))?;
+    json_value_to_ruby(ruby, &json_value)
+}
+
+/// Validate extraction patterns without running them.
+///
+/// The config JSON must contain:
+/// - `language` (string): the language name
+/// - `patterns` (object): named patterns to validate
+///
+/// Returns a native Ruby Hash with validation results.
+fn validate_extraction(ruby: &Ruby, config_json: String) -> Result<RbValue, Error> {
+    let config: tree_sitter_language_pack::ExtractionConfig = serde_json::from_str(&config_json)
+        .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("invalid config JSON: {e}")))?;
+
+    let result = tree_sitter_language_pack::validate_extraction(&config)
+        .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("{e}")))?;
+
+    let json_value = serde_json::to_value(&result)
+        .map_err(|e| Error::new(ruby.exception_runtime_error(), format!("serialization failed: {e}")))?;
+    json_value_to_ruby(ruby, &json_value)
 }
 
 /// Initialize the language pack with configuration.
@@ -187,6 +260,8 @@ fn init(ruby: &Ruby) -> Result<(), Error> {
     module.define_module_function("get_language_ptr", function!(get_language_ptr, 1))?;
     module.define_module_function("parse_string", function!(parse_string, 2))?;
     module.define_module_function("process", function!(process, 2))?;
+    module.define_module_function("extract", function!(extract, 2))?;
+    module.define_module_function("validate_extraction", function!(validate_extraction, 1))?;
 
     // Download API functions
     module.define_module_function("init", function!(rb_init, 1))?;

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tree_sitter_language_pack
 version: !ruby/object:Gem::Version
-  version: 1.2.1
+  version: 1.3.1
 platform: ruby
 authors:
 - kreuzberg.dev