polyglot-piranha 0.3.31__tar.gz → 0.3.33__tar.gz

{polyglot_piranha-0.3.31 → polyglot_piranha-0.3.33}/Cargo.lock

@@ -126,9 +126,9 @@ checksum = "613afe47fcd5fac7ccf1db93babcb082c5994d996f20b8b159f2ad1658eb5724"
 
 [[package]]
 name = "clap"
-version = "4.5.38"
+version = "4.5.39"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ed93b9805f8ba930df42c2590f05453d5ec36cbb85d018868a5b24d31f6ac000"
+checksum = "fd60e63e9be68e5fb56422e397cf9baddded06dae1d2e523401542383bc72a9f"
 dependencies = [
  "clap_builder",
  "clap_derive",
@@ -136,9 +136,9 @@ dependencies = [
 
 [[package]]
 name = "clap_builder"
-version = "4.5.38"
+version = "4.5.39"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "379026ff283facf611b0ea629334361c4211d1b12ee01024eec1591133b04120"
+checksum = "89cc6392a1f72bbeb820d71f32108f61fdaf18bc526e1d23954168a67759ef51"
 dependencies = [
  "anstream",
  "anstyle",
@@ -502,9 +502,9 @@ checksum = "d750af042f7ef4f724306de029d18836c26c1765a54a6a3f094cbd23a7267ffa"
 
 [[package]]
 name = "lock_api"
-version = "0.4.12"
+version = "0.4.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "07af8b9cdd281b7915f413fa73f29ebd5d55d0d3f0155584dade1ff18cea1b17"
+checksum = "96936507f153605bddfcda068dd804796c84324ed2510809e5b2a624c81da765"
 dependencies = [
  "autocfg",
  "scopeguard",
@@ -572,9 +572,9 @@ checksum = "a4895175b425cb1f87721b59f0f286c2092bd4af812243672510e1ac53e2e0ad"
 
 [[package]]
 name = "parking_lot"
-version = "0.12.3"
+version = "0.12.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f1bf18183cf54e8d6059647fc3063646a1801cf30896933ec2311622cc4b9a27"
+checksum = "70d58bf43669b5795d1576d0641cfb6fbb2057bf629506267a92807158584a13"
 dependencies = [
  "lock_api",
  "parking_lot_core",
@@ -582,9 +582,9 @@ dependencies = [
 
 [[package]]
 name = "parking_lot_core"
-version = "0.9.10"
+version = "0.9.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1e401f977ab385c9e4e3ab30627d6f26d00e2c73eef317493c4ec6d468726cf8"
+checksum = "bc838d2a56b5b1a6c25f55575dfc605fabb63bb2365f6c2353ef9159aa69e4a5"
 dependencies = [
  "cfg-if",
  "libc",
@@ -595,7 +595,7 @@ dependencies = [
 
 [[package]]
 name = "piranha"
-version = "0.3.31"
+version = "0.3.33"
 dependencies = [
  "assert_cmd",
  "cc",
@@ -1087,8 +1087,7 @@ dependencies = [
 [[package]]
 name = "tree-sitter-kotlin"
 version = "0.3.5"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8df217a0e1fec649f3e13157de932439f3d37ea4e265038dd0873971ef56e726"
+source = "git+https://github.com/danieltrt/tree-sitter-kotlin?rev=5e0b07cb2ba628842028313b60b97699cc5e0fee#5e0b07cb2ba628842028313b60b97699cc5e0fee"
 dependencies = [
  "cc",
  "tree-sitter",

{polyglot_piranha-0.3.31 → polyglot_piranha-0.3.33}/Cargo.toml

@@ -2,7 +2,7 @@
 authors = ["Uber Technologies Inc."]
 name = "piranha"
 description = "Polyglot Piranha is a library for performing structural find and replace with deep cleanup."
-version = "0.3.31"
+version = "0.3.33"
 edition = "2021"
 include = ["pyproject.toml", "src/"]
 exclude = ["legacy"]
@@ -47,7 +47,9 @@ serde_json = "1.0.82"
 ctrlc = "3.4"
 
 # TODO: Update if we upgrade tree-sitter to >=0.21
-tree-sitter-kotlin = "=0.3.5"
+# Kotlin's grammar needs to fix the way it parses unary expressions (currently -a + b gets parsed as - (a +b)).
+# Once we upstream the fix, we can point to the official version.
+tree-sitter-kotlin = { git = "https://github.com/danieltrt/tree-sitter-kotlin", rev = "5e0b07cb2ba628842028313b60b97699cc5e0fee" }
 tree-sitter-java = "0.20.2"
 # TODO: Update after: https://github.com/alex-pinkus/tree-sitter-swift/issues/278 resolves
 tree-sitter-swift = { git = "https://github.com/satyam1749/tree-sitter-swift.git", rev = "08a28993599f1968bc81631a89690503e1db7704" }

{polyglot_piranha-0.3.31 → polyglot_piranha-0.3.33}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: polyglot_piranha
-Version: 0.3.31
+Version: 0.3.33
 License-File: LICENSE
 License-File: NOTICE
 Summary: Polyglot Piranha is a library for performing structural find and replace with deep cleanup.

{polyglot_piranha-0.3.31 → polyglot_piranha-0.3.33}/src/models/default_configs.rs

@@ -192,8 +192,8 @@ pub(crate) fn default_is_seed_rule() -> bool {
   true
 }
 
-pub(crate) fn default_delete_comments() -> bool {
-  true
+pub(crate) fn default_keep_comment_regexes() -> HashSet<String> {
+  HashSet::new()
 }
 
 pub(crate) fn default_allow_dirty_ast() -> bool {

{polyglot_piranha-0.3.31 → polyglot_piranha-0.3.33}/src/models/matches.rs

@@ -18,6 +18,7 @@ use getset::{Getters, MutGetters};
 use itertools::Itertools;
 use log::trace;
 use pyo3::prelude::{pyclass, pymethods};
+use regex::Regex;
 use serde_derive::{Deserialize, Serialize};
 use tree_sitter::Node;
 
@@ -120,10 +121,10 @@ impl Match {
   // Populates the leading and trailing comma and comment ranges for the match.
   fn populate_associated_elements(
     &mut self, node: &Node, code: &String, piranha_arguments: &PiranhaArguments,
-    delete_comments: bool,
+    compiled_regexes: Vec<Regex>,
   ) {
-    self.get_associated_elements(node, code, piranha_arguments, true, delete_comments);
-    self.get_associated_elements(node, code, piranha_arguments, false, delete_comments);
+    self.get_associated_elements(node, code, piranha_arguments, true, &compiled_regexes);
+    self.get_associated_elements(node, code, piranha_arguments, false, &compiled_regexes);
     self.get_associated_leading_empty_lines(node, code);
   }
 
@@ -183,7 +184,7 @@ impl Match {
   /// We currently capture leading and trailing comments and commas.
   fn get_associated_elements(
     &mut self, node: &Node, code: &String, piranha_arguments: &PiranhaArguments, trailing: bool,
-    delete_comments: bool,
+    keep_comment_regexes: &[Regex],
   ) {
     let mut current_node = *node;
     let mut buf = *piranha_arguments.cleanup_comments_buffer();
@@ -201,11 +202,14 @@ impl Match {
           self.associated_comma = Some(sibling.range().into());
           current_node = sibling;
           continue; // Continue the inner loop (i.e. evaluate next sibling)
-        } else if delete_comments
-          && self._is_comment_safe_to_delete(&sibling, node, piranha_arguments, trailing)
-        {
-          // Add the comment to the associated matches
-          self.associated_comments.push(sibling.range().into());
+        } else if self._is_comment_safe_to_delete(&sibling, node, piranha_arguments, trailing) {
+          let comment_text = sibling.utf8_text(code.as_bytes()).unwrap();
+          if !keep_comment_regexes
+            .iter()
+            .any(|re| re.is_match(comment_text))
+          {
+            self.associated_comments.push(sibling.range().into());
+          }
           current_node = sibling;
           continue; // Continue the inner loop (i.e. evaluate next sibling)
         }
@@ -494,11 +498,17 @@ impl SourceCodeUnit {
         p_match.range().end_byte,
       );
       if self.is_satisfied(matched_node, rule, p_match.matches(), rule_store) {
+        let compiled_regexes: Vec<Regex> = rule
+          .rule()
+          .keep_comment_regexes()
+          .iter()
+          .filter_map(|pattern| Regex::new(pattern).ok())
+          .collect();
         p_match.populate_associated_elements(
           &matched_node,
           self.code(),
           self.piranha_arguments(),
-          *rule.rule().delete_comments(),
+          compiled_regexes,
         );
         trace!("Found match {:#?}", p_match);
         output.push(p_match.clone());

{polyglot_piranha-0.3.31 → polyglot_piranha-0.3.33}/src/models/rule.rs

@@ -25,8 +25,9 @@ use crate::utilities::Instantiate;
 use super::{
   capture_group_patterns::CGPattern,
   default_configs::{
-    default_delete_comments, default_filters, default_groups, default_holes, default_is_seed_rule,
-    default_query, default_replace, default_replace_idx, default_replace_node, default_rule_name,
+    default_filters, default_groups, default_holes, default_is_seed_rule,
+    default_keep_comment_regexes, default_query, default_replace, default_replace_idx,
+    default_replace_node, default_rule_name,
   },
   filter::Filter,
   Validator,
@@ -97,11 +98,11 @@ pub struct Rule {
   is_seed_rule: bool,
 
   /// Marks comments as deletable
-  #[builder(default = "default_delete_comments()")]
-  #[serde(default = "default_delete_comments")]
+  #[builder(default = "default_keep_comment_regexes()")]
+  #[serde(default = "default_keep_comment_regexes")]
   #[get = "pub"]
   #[pyo3(get)]
-  delete_comments: bool,
+  keep_comment_regexes: HashSet<String>,
 }
 
 impl Rule {
@@ -148,7 +149,7 @@ macro_rules! piranha_rule {
                 $(, is_seed_rule = $is_seed_rule:expr)?
                 $(, groups = [$($group_name: expr)*])?
                 $(, filters = [$($filter:tt)*])?
-                $(, delete_comments = $delete_comments:expr)?
+                $(, keep_comment_regexes = [$($rgex: expr)*])?
               ) => {
     $crate::models::rule::RuleBuilder::default()
     .name($name.to_string())
@@ -160,7 +161,7 @@ macro_rules! piranha_rule {
     $(.holes(std::collections::HashSet::from([$($hole.to_string(),)*])))?
     $(.groups(std::collections::HashSet::from([$($group_name.to_string(),)*])))?
     $(.filters(std::collections::HashSet::from([$($filter)*])))?
-    $(.delete_comments($delete_comments))?
+    $(.keep_comment_regexes(std::collections::HashSet::from([$($rgex.to_string(),)*])))?
     .build().unwrap()
   };
 }
@@ -171,7 +172,8 @@ impl Rule {
   fn py_new(
     name: String, query: Option<String>, replace: Option<String>, replace_idx: Option<u8>,
     replace_node: Option<String>, holes: Option<HashSet<String>>, groups: Option<HashSet<String>>,
-    filters: Option<HashSet<Filter>>, is_seed_rule: Option<bool>, delete_comments: Option<bool>,
+    filters: Option<HashSet<Filter>>, is_seed_rule: Option<bool>,
+    keep_comment_regexes: Option<HashSet<String>>,
   ) -> Self {
     let mut rule_builder = RuleBuilder::default();
 
@@ -208,8 +210,8 @@ impl Rule {
       rule_builder.is_seed_rule(is_seed_rule);
     }
 
-    if let Some(delete_comments) = delete_comments {
-      rule_builder.delete_comments(delete_comments);
+    if let Some(keep_comment_regexes) = keep_comment_regexes {
+      rule_builder.keep_comment_regexes(keep_comment_regexes);
     }
 
     rule_builder.build().unwrap()

{polyglot_piranha-0.3.31 → polyglot_piranha-0.3.33}/src/models/unit_tests/concrete_syntax_test.rs

@@ -14,7 +14,10 @@
 use crate::models::capture_group_patterns::ConcreteSyntax;
 use crate::models::concrete_syntax::get_all_matches_for_concrete_syntax;
 use crate::models::default_configs::GO;
-use crate::models::{default_configs::JAVA, language::PiranhaLanguage};
+use crate::models::{
+  default_configs::{JAVA, KOTLIN},
+  language::PiranhaLanguage,
+};
 
 fn run_test(
   code: &str, pattern: &str, expected_matches: usize, expected_vars: Vec<Vec<(&str, &str)>>,
@@ -39,6 +42,19 @@ fn run_test(
   }
 }
 
+#[test]
+fn test_single_match_kotlin() {
+  run_test(
+    "class Something : BaseParameters(namespace = \"data\") {
+              val something_else: Boolean by parameter(\"something_else\")
+          }",
+    "val :[stale_property_name] : Boolean by parameter(\"something_else\")",
+    1,
+    vec![vec![("stale_property_name", "something_else")]],
+    KOTLIN,
+  );
+}
+
 #[test]
 fn test_single_match() {
   run_test(

{polyglot_piranha-0.3.31 → polyglot_piranha-0.3.33}/src/models/unit_tests/rule_test.rs

@@ -393,14 +393,17 @@ fn test_rule_delete_comments() {
     query= "((method_invocation) @m (#eq? @m \"variable.set_value(true)\"))",
     replace_node = "m",
     replace = "",
-    delete_comments = false
+    keep_comment_regexes = ["// Given .*"]
   };
 
   let rule = InstantiatedRule::new(&_rule, &HashMap::new());
   let source_code = "class Test {
           public void foobar(){
-            // Given
+            // Given something
+            // This is an explanatory comment
+            // This is another explanatory comment
             variable.set_value(true);
+            other_variable.set_value(false);
 
             // When
             var result = data.Something();
@@ -433,16 +436,12 @@ fn test_rule_delete_comments() {
   let edit = source_code_unit.get_edit(&rule, &mut rule_store, node, true);
   assert!(edit.is_some());
   let edit = edit.unwrap();
-
   // Apply the edit to the source code unit
   source_code_unit.apply_edit(&edit, &mut parser);
 
-  // Inspect the resulting code
   let edited_code = source_code_unit.code();
-
-  // Now assert on the full edited code
-  assert!(edited_code.contains("// Given"));
-  assert!(edited_code.contains("// When"));
-  assert!(edited_code.contains("// Then"));
+  assert!(edited_code.contains("// Given something"));
+  assert!(!edited_code.contains("// This is another explanatory comment"));
+  assert!(!edited_code.contains("// This is an explanatory comment"));
   assert!(!edited_code.contains("variable.set_value(true);"));
 }