Upgrade Prism to v1.8.0 (#1067)

marcoroth · web-flow · commit ee64cce102b4 · 2026-01-17T10:54:38.000+01:00
Prism 1.8.0 (ruby/prism#3827) changed how the `closing_loc` is populated for unclosed blocks. Previously it was `NULL` for unclosed blocks, but now it's always set (pointing to body content). This broke our detection logic that relied on checking if `closing_loc` was empty. https://github.com/ruby/prism/releases/tag/v1.8.0
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -12,7 +12,7 @@ If you encounter any issues when following along with this file please don't hes
 - [**Clang 21**](https://clang.llvm.org): The compiler used to build this project.
 - [**Clang Format 21**](https://clang.llvm.org/docs/ClangFormat.html): For formatting the project.
 - [**Clang Tidy 21**](https://clang.llvm.org/extra/clang-tidy/): For linting the project.
-- [**Prism Ruby Parser v1.7.0**](https://github.com/ruby/prism/releases/tag/v1.7.0): We use Prism for Parsing the Ruby Source Code in the HTML+ERB files.
+- [**Prism Ruby Parser v1.8.0**](https://github.com/ruby/prism/releases/tag/v1.8.0): We use Prism for Parsing the Ruby Source Code in the HTML+ERB files.
 - [**Ruby**](https://www.ruby-lang.org/en/): We need Ruby as a dependency for `bundler`.
 - [**Bundler**](https://bundler.io): We are using `bundler` to build [`prism`](https://github.com/ruby/prism) from source so we can build `herb` against it.
 - [**Emscripten**](https://emscripten.org): For the WebAssembly build of `libherb` so it can be used in the browser using the [`@herb-tools/browser`](https://github.com/marcoroth/herb/blob/main/javascript/packages/browser) package.
diff --git a/Gemfile b/Gemfile
@@ -4,7 +4,7 @@ source "https://rubygems.org"
 
 gemspec
 
-gem "prism", github: "ruby/prism", tag: "v1.7.0"
+gem "prism", github: "ruby/prism", tag: "v1.8.0"
 
 gem "actionview", "~> 8.0"
 gem "digest", "~> 3.2"
diff --git a/Gemfile.lock b/Gemfile.lock
@@ -1,9 +1,9 @@
 GIT
   remote: https://github.com/ruby/prism.git
-  revision: df554035afb63c754b9edbe531ff48e6569eaf07
-  tag: v1.7.0
+  revision: 90c05789e4f90dec600d7748a002058046852370
+  tag: v1.8.0
   specs:
-    prism (1.7.0)
+    prism (1.8.0)
 
 PATH
   remote: .
@@ -110,7 +110,7 @@ GEM
     psych (5.3.1)
       date
       stringio
-    racc (1.8.1)
+    racc (1.8.0)
     rails-dom-testing (2.3.0)
       activesupport (>= 5.0.0)
       minitest
diff --git a/docs/docs/bindings/java/reference.md b/docs/docs/bindings/java/reference.md
@@ -185,7 +185,7 @@ Returns the full version information including Herb, Prism, and JNI details:
 import org.herb.Herb;
 
 System.out.println(Herb.version());
-// Output: "herb java v0.8.7, libprism v1.7.0, libherb v0.8.7 (Java JNI)"
+// Output: "herb java v0.8.7, libprism v1.8.0, libherb v0.8.7 (Java JNI)"
 ```
 :::
 
@@ -211,7 +211,7 @@ Returns the Prism parser version:
 import org.herb.Herb;
 
 System.out.println(Herb.prismVersion());
-// Output: "1.7.0"
+// Output: "1.8.0"
 ```
 :::
 
diff --git a/docs/docs/bindings/rust/reference.md b/docs/docs/bindings/rust/reference.md
@@ -229,7 +229,7 @@ Returns the full version information including Herb, Prism, and FFI details:
 use herb::version;
 
 println!("{}", version());
-// Output: "herb rust v0.8.7, libprism v1.7.0, libherb v0.8.7 (Rust FFI)"
+// Output: "herb rust v0.8.7, libprism v1.8.0, libherb v0.8.7 (Rust FFI)"
 ```
 :::
 
@@ -255,7 +255,7 @@ Returns the Prism parser version:
 use herb::prism_version;
 
 println!("{}", prism_version());
-// Output: "1.7.0"
+// Output: "1.8.0"
 ```
 :::
 
diff --git a/javascript/packages/browser/test/browser.test.ts b/javascript/packages/browser/test/browser.test.ts
@@ -17,7 +17,7 @@ describe("@herb-tools/browser", () => {
   test("version() returns a string", async () => {
     const version = Herb.version
     expect(typeof version).toBe("string")
-    expect(version).toBe("@herb-tools/browser@0.8.7, @herb-tools/core@0.8.7, libprism@1.7.0, libherb@0.8.7 (WebAssembly)")
+    expect(version).toBe("@herb-tools/browser@0.8.7, @herb-tools/core@0.8.7, libprism@1.8.0, libherb@0.8.7 (WebAssembly)")
   })
 
   test("parse() can process a simple template", async () => {
diff --git a/javascript/packages/node-wasm/test/node-wasm.test.ts b/javascript/packages/node-wasm/test/node-wasm.test.ts
@@ -17,7 +17,7 @@ describe("@herb-tools/node-wasm", () => {
   test("version() returns a string", async () => {
     const version = Herb.version
     expect(typeof version).toBe("string")
-    expect(version).toBe("@herb-tools/node-wasm@0.8.7, @herb-tools/core@0.8.7, libprism@1.7.0, libherb@0.8.7 (WebAssembly)")
+    expect(version).toBe("@herb-tools/node-wasm@0.8.7, @herb-tools/core@0.8.7, libprism@1.8.0, libherb@0.8.7 (WebAssembly)")
   })
 
   test("parse() can process a simple template", async () => {
diff --git a/javascript/packages/node/test/node.test.ts b/javascript/packages/node/test/node.test.ts
@@ -17,7 +17,7 @@ describe("@herb-tools/node", () => {
   test("version() returns a string", async () => {
     const version = Herb.version
     expect(typeof version).toBe("string")
-    expect(version).toBe("@herb-tools/node@0.8.7, @herb-tools/core@0.8.7, libprism@1.7.0, libherb@0.8.7 (Node.js C++ native extension)")
+    expect(version).toBe("@herb-tools/node@0.8.7, @herb-tools/core@0.8.7, libprism@1.8.0, libherb@0.8.7 (Node.js C++ native extension)")
   })
 
   test("parse() can process a simple template", async () => {
diff --git a/rust/tests/cli_commands_test.rs b/rust/tests/cli_commands_test.rs
@@ -4,7 +4,7 @@ use herb::{extract_html, extract_ruby, version};
 fn test_version_functions() {
   assert_eq!(
     version(),
-    "herb rust v0.8.7, libprism v1.7.0, libherb v0.8.7 (Rust FFI)"
+    "herb rust v0.8.7, libprism v1.8.0, libherb v0.8.7 (Rust FFI)"
   );
 }
 
diff --git a/rust/tests/snapshots/snapshot_test__prism_version_output.snap b/rust/tests/snapshots/snapshot_test__prism_version_output.snap
diff --git a/rust/tests/snapshots/snapshot_test__version_output.snap b/rust/tests/snapshots/snapshot_test__version_output.snap
diff --git a/src/analyze.c b/src/analyze.c
@@ -194,14 +194,12 @@ static bool find_earliest_control_keyword_walker(const pm_node_t* node, void* da
 
       if (call->block != NULL && call->block->type == PM_BLOCK_NODE) {
         pm_block_node_t* block_node = (pm_block_node_t*) call->block;
-        size_t opening_length = block_node->opening_loc.end - block_node->opening_loc.start;
-        bool has_do_opening =
-          opening_length == 2 && block_node->opening_loc.start[0] == 'd' && block_node->opening_loc.start[1] == 'o';
-        bool has_brace_opening = opening_length == 1 && block_node->opening_loc.start[0] == '{';
-        bool has_closing_location = block_node->closing_loc.start != NULL && block_node->closing_loc.end != NULL
-                                 && (block_node->closing_loc.end - block_node->closing_loc.start) > 0;
-
-        if (has_do_opening || (has_brace_opening && !has_closing_location)) {
+
+        bool has_do_opening = is_do_block(block_node->opening_loc);
+        bool has_brace_opening = is_brace_block(block_node->opening_loc);
+        bool has_valid_brace_closing = is_closing_brace(block_node->closing_loc);
+
+        if (has_do_opening || (has_brace_opening && !has_valid_brace_closing)) {
           current_type = CONTROL_TYPE_BLOCK;
           keyword_offset = (uint32_t) (node->location.start - context->source_start);
         }
diff --git a/src/analyze_helpers.c b/src/analyze_helpers.c
@@ -102,13 +102,13 @@ bool search_if_nodes(const pm_node_t* node, void* data) {
   return false;
 }
 
-static bool is_do_block(pm_location_t opening_location) {
+bool is_do_block(pm_location_t opening_location) {
   size_t length = opening_location.end - opening_location.start;
 
   return length == 2 && opening_location.start[0] == 'd' && opening_location.start[1] == 'o';
 }
 
-static bool is_brace_block(pm_location_t opening_location) {
+bool is_brace_block(pm_location_t opening_location) {
   size_t length = opening_location.end - opening_location.start;
 
   return length == 1 && opening_location.start[0] == '{';
@@ -118,14 +118,40 @@ static bool has_location(pm_location_t location) {
   return location.start != NULL && location.end != NULL && (location.end - location.start) > 0;
 }
 
+static bool is_end_keyword(pm_location_t location) {
+  if (location.start == NULL || location.end == NULL) { return false; }
+
+  size_t length = location.end - location.start;
+
+  return length == 3 && location.start[0] == 'e' && location.start[1] == 'n' && location.start[2] == 'd';
+}
+
+bool is_closing_brace(pm_location_t location) {
+  if (location.start == NULL || location.end == NULL) { return false; }
+
+  size_t length = location.end - location.start;
+
+  return length == 1 && location.start[0] == '}';
+}
+
+bool has_valid_block_closing(pm_location_t opening_loc, pm_location_t closing_loc) {
+  if (is_do_block(opening_loc)) {
+    return is_end_keyword(closing_loc);
+  } else if (is_brace_block(opening_loc)) {
+    return is_closing_brace(closing_loc);
+  }
+
+  return false;
+}
+
 bool search_block_nodes(const pm_node_t* node, void* data) {
   analyzed_ruby_T* analyzed = (analyzed_ruby_T*) data;
 
   if (node->type == PM_BLOCK_NODE) {
     pm_block_node_t* block_node = (pm_block_node_t*) node;
 
     bool has_opening = is_do_block(block_node->opening_loc) || is_brace_block(block_node->opening_loc);
-    bool is_unclosed = !has_location(block_node->closing_loc);
+    bool is_unclosed = !has_valid_block_closing(block_node->opening_loc, block_node->closing_loc);
 
     if (has_opening && is_unclosed) { analyzed->block_node_count++; }
   }
@@ -314,7 +340,7 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
     case PM_IF_NODE: {
       const pm_if_node_t* if_node = (const pm_if_node_t*) node;
 
-      if (has_location(if_node->if_keyword_loc) && !has_location(if_node->end_keyword_loc)) {
+      if (has_location(if_node->if_keyword_loc) && !is_end_keyword(if_node->end_keyword_loc)) {
         if (!is_postfix_conditional(if_node->statements, if_node->if_keyword_loc)) {
           analyzed->unclosed_control_flow_count++;
         }
@@ -326,7 +352,7 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
     case PM_UNLESS_NODE: {
       const pm_unless_node_t* unless_node = (const pm_unless_node_t*) node;
 
-      if (has_location(unless_node->keyword_loc) && !has_location(unless_node->end_keyword_loc)) {
+      if (has_location(unless_node->keyword_loc) && !is_end_keyword(unless_node->end_keyword_loc)) {
         if (!is_postfix_conditional(unless_node->statements, unless_node->keyword_loc)) {
           analyzed->unclosed_control_flow_count++;
         }
@@ -338,7 +364,7 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
     case PM_CASE_NODE: {
       const pm_case_node_t* case_node = (const pm_case_node_t*) node;
 
-      if (has_location(case_node->case_keyword_loc) && !has_location(case_node->end_keyword_loc)) {
+      if (has_location(case_node->case_keyword_loc) && !is_end_keyword(case_node->end_keyword_loc)) {
         analyzed->unclosed_control_flow_count++;
       }
 
@@ -348,7 +374,7 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
     case PM_CASE_MATCH_NODE: {
       const pm_case_match_node_t* case_match_node = (const pm_case_match_node_t*) node;
 
-      if (has_location(case_match_node->case_keyword_loc) && !has_location(case_match_node->end_keyword_loc)) {
+      if (has_location(case_match_node->case_keyword_loc) && !is_end_keyword(case_match_node->end_keyword_loc)) {
         analyzed->unclosed_control_flow_count++;
       }
 
@@ -358,7 +384,7 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
     case PM_WHILE_NODE: {
       const pm_while_node_t* while_node = (const pm_while_node_t*) node;
 
-      if (has_location(while_node->keyword_loc) && !has_location(while_node->closing_loc)) {
+      if (has_location(while_node->keyword_loc) && !is_end_keyword(while_node->closing_loc)) {
         analyzed->unclosed_control_flow_count++;
       }
 
@@ -368,7 +394,7 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
     case PM_UNTIL_NODE: {
       const pm_until_node_t* until_node = (const pm_until_node_t*) node;
 
-      if (has_location(until_node->keyword_loc) && !has_location(until_node->closing_loc)) {
+      if (has_location(until_node->keyword_loc) && !is_end_keyword(until_node->closing_loc)) {
         analyzed->unclosed_control_flow_count++;
       }
 
@@ -378,7 +404,7 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
     case PM_FOR_NODE: {
       const pm_for_node_t* for_node = (const pm_for_node_t*) node;
 
-      if (has_location(for_node->for_keyword_loc) && !has_location(for_node->end_keyword_loc)) {
+      if (has_location(for_node->for_keyword_loc) && !is_end_keyword(for_node->end_keyword_loc)) {
         analyzed->unclosed_control_flow_count++;
       }
 
@@ -388,7 +414,7 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
     case PM_BEGIN_NODE: {
       const pm_begin_node_t* begin_node = (const pm_begin_node_t*) node;
 
-      if (has_location(begin_node->begin_keyword_loc) && !has_location(begin_node->end_keyword_loc)) {
+      if (has_location(begin_node->begin_keyword_loc) && !is_end_keyword(begin_node->end_keyword_loc)) {
         analyzed->unclosed_control_flow_count++;
       }
 
@@ -399,7 +425,9 @@ bool search_unclosed_control_flows(const pm_node_t* node, void* data) {
       const pm_block_node_t* block_node = (const pm_block_node_t*) node;
       bool has_opening = is_do_block(block_node->opening_loc) || is_brace_block(block_node->opening_loc);
 
-      if (has_opening && !has_location(block_node->closing_loc)) { analyzed->unclosed_control_flow_count++; }
+      if (has_opening && !has_valid_block_closing(block_node->opening_loc, block_node->closing_loc)) {
+        analyzed->unclosed_control_flow_count++;
+      }
       break;
     }
 
diff --git a/src/include/analyze_helpers.h b/src/include/analyze_helpers.h
@@ -28,6 +28,11 @@ bool has_yield_node(analyzed_ruby_T* analyzed);
 
 bool has_error_message(analyzed_ruby_T* anlayzed, const char* message);
 
+bool is_do_block(pm_location_t opening_location);
+bool is_brace_block(pm_location_t opening_location);
+bool is_closing_brace(pm_location_t location);
+bool has_valid_block_closing(pm_location_t opening_loc, pm_location_t closing_loc);
+
 bool search_if_nodes(const pm_node_t* node, void* data);
 bool search_block_nodes(const pm_node_t* node, void* data);
 bool search_case_nodes(const pm_node_t* node, void* data);
diff --git a/test/herb_test.rb b/test/herb_test.rb
@@ -4,6 +4,6 @@
 
 class HerbTest < Minitest::Spec
   test "version" do
-    assert_equal "herb gem v0.8.7, libprism v1.7.0, libherb v0.8.7 (Ruby C native extension)", Herb.version
+    assert_equal "herb gem v0.8.7, libprism v1.8.0, libherb v0.8.7 (Ruby C native extension)", Herb.version
   end
 end
diff --git a/test/snapshots/analyze/block_test/test_0020_unclosed_brace_block_with_end_should_error_ce5658e89dd99cb6740835d0d0666353.txt b/test/snapshots/analyze/block_test/test_0020_unclosed_brace_block_with_end_should_error_ce5658e89dd99cb6740835d0d0666353.txt
diff --git a/test/snapshots/parser/multiple_control_flow_test/test_0011_case_and_block_in_same_ERB_tag_5a8200a187bd81188eb49932478f8972.txt b/test/snapshots/parser/multiple_control_flow_test/test_0011_case_and_block_in_same_ERB_tag_5a8200a187bd81188eb49932478f8972.txt

Original file line number	Diff line number	Diff line change
`@@ -4,7 +4,7 @@ use herb::{extract_html, extract_ruby, version};`
`4`	`4`	`fn test_version_functions() {`
`5`	`5`	`assert_eq!(`
`6`	`6`	`version(),`
`7`		`- "herb rust v0.8.7, libprism v1.7.0, libherb v0.8.7 (Rust FFI)"`
	`7`	`+ "herb rust v0.8.7, libprism v1.8.0, libherb v0.8.7 (Rust FFI)"`
`8`	`8`	`);`
`9`	`9`	`}`
`10`	`10`