Add owner integrity check

Author: vinistock

.github/workflows/indexing_top_gems.yml

@@ -31,6 +31,8 @@ jobs:
           key: ${{ runner.os }}-rust
 
       - name: Run Ruby tests
+        env:
+          RUST_BACKTRACE: full
         run: bundle exec rake index_top_gems
 
       - name: Save Rust compile cache

ext/rubydex/graph.c

@@ -9,6 +9,7 @@
 #include "utils.h"
 
 static VALUE cGraph;
+static VALUE mRubydex;
 
 // Free function for the custom Graph allocator. We always have to call into Rust to free data allocated by it
 static void graph_free(void *ptr) {
@@ -449,6 +450,32 @@ static VALUE rdxr_graph_require_paths(VALUE self, VALUE load_path) {
     return array;
 }
 
+// Graph#check_integrity: () -> Array[Rubydex::IntegrityFailure]
+// Returns an array of IntegrityFailure objects, empty if no issues found
+static VALUE rdxr_graph_check_integrity(VALUE self) {
+    void *graph;
+    TypedData_Get_Struct(self, void *, &graph_type, graph);
+
+    size_t error_count = 0;
+    const char *const *errors = rdx_check_integrity(graph, &error_count);
+
+    if (errors == NULL) {
+        return rb_ary_new();
+    }
+
+    VALUE cIntegrityError = rb_const_get(mRubydex, rb_intern("IntegrityFailure"));
+    VALUE array = rb_ary_new_capa((long)error_count);
+
+    for (size_t i = 0; i < error_count; i++) {
+        VALUE argv[] = {rb_utf8_str_new_cstr(errors[i])};
+        VALUE error = rb_class_new_instance(1, argv, cIntegrityError);
+        rb_ary_push(array, error);
+    }
+
+    free_c_string_array(errors, error_count);
+    return array;
+}
+
 // Graph#diagnostics -> Array[Rubydex::Diagnostic]
 static VALUE rdxr_graph_diagnostics(VALUE self) {
     void *graph;
@@ -482,7 +509,8 @@ static VALUE rdxr_graph_diagnostics(VALUE self) {
     return diagnostics;
 }
 
-void rdxi_initialize_graph(VALUE mRubydex) {
+void rdxi_initialize_graph(VALUE moduleRubydex) {
+    mRubydex = moduleRubydex;
     cGraph = rb_define_class_under(mRubydex, "Graph", rb_cObject);
     rb_define_alloc_func(cGraph, rdxr_graph_alloc);
     rb_define_method(cGraph, "index_all", rdxr_graph_index_all, 1);
@@ -495,6 +523,7 @@ void rdxi_initialize_graph(VALUE mRubydex) {
     rb_define_method(cGraph, "constant_references", rdxr_graph_constant_references, 0);
     rb_define_method(cGraph, "method_references", rdxr_graph_method_references, 0);
     rb_define_method(cGraph, "diagnostics", rdxr_graph_diagnostics, 0);
+    rb_define_method(cGraph, "check_integrity", rdxr_graph_check_integrity, 0);
     rb_define_method(cGraph, "[]", rdxr_graph_aref, 1);
     rb_define_method(cGraph, "search", rdxr_graph_search, 1);
     rb_define_method(cGraph, "encoding=", rdxr_graph_set_encoding, 1);

lib/rubydex.rb

@@ -2,11 +2,6 @@
 
 require "bundler"
 require "uri"
-
-module Rubydex
-  class Error < StandardError; end
-end
-
 require "rubydex/version"
 
 begin
@@ -18,6 +13,7 @@ class Error < StandardError; end
   require "rubydex/rubydex"
 end
 
+require "rubydex/failures"
 require "rubydex/location"
 require "rubydex/comment"
 require "rubydex/diagnostic"

lib/rubydex/failures.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Rubydex
+  class Failure
+    #: String
+    attr_reader :message
+
+    #: (String) -> void
+    def initialize(message)
+      @message = message
+    end
+  end
+
+  class IntegrityFailure < Failure; end
+end

lib/rubydex/location.rb

@@ -4,6 +4,8 @@ module Rubydex
   # A zero based internal location. Intended to be used for tool-to-tool communication, such as a language server
   # communicating with an editor.
   class Location
+    class NotFileUriError < StandardError; end
+
     include Comparable
 
     #: String
@@ -22,9 +24,9 @@ def initialize(uri:, start_line:, end_line:, start_column:, end_column:)
     end
 
     #: () -> String
-    def path
+    def to_file_path
       uri = URI(@uri)
-      raise Rubydex::Error, "URI is not a file:// URI: #{@uri}" unless uri.scheme == "file"
+      raise NotFileUriError, "URI is not a file:// URI: #{@uri}" unless uri.scheme == "file"
 
       path = uri.path
       # TODO: This has to go away once we have a proper URI abstraction
@@ -59,7 +61,7 @@ def to_display
 
     #: -> String
     def to_s
-      "#{path}:#{@start_line + 1}:#{@start_column + 1}-#{@end_line + 1}:#{@end_column + 1}"
+      "#{to_file_path}:#{@start_line + 1}:#{@start_column + 1}-#{@end_line + 1}:#{@end_column + 1}"
     end
   end
 
@@ -82,7 +84,7 @@ def comparable_values
 
     #: -> String
     def to_s
-      "#{path}:#{@start_line}:#{@start_column}-#{@end_line}:#{@end_column}"
+      "#{to_file_path}:#{@start_line}:#{@start_column}-#{@end_line}:#{@end_column}"
     end
   end
 end

rakelib/index_top_gems.rake

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 desc "Index the top 100 gems from rubygems.org"
-task index_top_gems: :compile_release do
+task index_top_gems: :compile do
   $LOAD_PATH.unshift(File.expand_path("../lib", __dir__))
   require "net/http"
   require "rubygems/package"
@@ -51,7 +51,10 @@ task index_top_gems: :compile_release do
 
             # Index the gem's files and yield errors back to the main Ractor
             graph = Rubydex::Graph.new
-            errors = graph.index_all(Dir.glob("#{gem_dir}/**/*.rb"))
+            indexing_errors = graph.index_all(Dir.glob("#{gem_dir}/**/*.rb"))
+            graph.resolve
+
+            errors = indexing_errors + graph.check_integrity.map(&:message)
             next if errors.empty?
 
             @mutex.synchronize { @errors << "#{gem} => #{errors.join(", ")}" }

rbi/rubydex.rbi

@@ -204,6 +204,9 @@ module Rubydex
     sig { returns(T::Array[String]) }
     def workspace_paths; end
 
+    sig { void }
+    def check_integrity; end
+
     private
 
     # Gathers the paths we have to index for all workspace dependencies
@@ -235,7 +238,7 @@ module Rubydex
     def end_line; end
 
     sig { returns(String) }
-    def path; end
+    def to_file_path; end
 
     sig { returns(Integer) }
     def start_column; end

rust/rubydex-sys/src/graph_api.rs

@@ -10,7 +10,7 @@ use rubydex::model::encoding::Encoding;
 use rubydex::model::graph::Graph;
 use rubydex::model::ids::DeclarationId;
 use rubydex::resolution::Resolver;
-use rubydex::{indexing, listing, query};
+use rubydex::{indexing, integrity, listing, query};
 use std::ffi::CString;
 use std::path::PathBuf;
 use std::{mem, ptr};
@@ -196,6 +196,42 @@ pub extern "C" fn rdx_graph_resolve(pointer: GraphPointer) {
     });
 }
 
+/// Checks the integrity of the graph and returns an array of error message strings. Returns NULL if there are no
+/// errors. Caller must free with `free_c_string_array`.
+///
+/// # Safety
+///
+/// - `pointer` must be a valid `GraphPointer` previously returned by this crate.
+/// - `out_error_count` must be a valid, writable pointer.
+#[unsafe(no_mangle)]
+pub unsafe extern "C" fn rdx_check_integrity(
+    pointer: GraphPointer,
+    out_error_count: *mut usize,
+) -> *const *const c_char {
+    with_graph(pointer, |graph| {
+        let errors = integrity::check_integrity(graph);
+
+        if errors.is_empty() {
+            unsafe { *out_error_count = 0 };
+            return ptr::null();
+        }
+
+        let c_strings: Vec<*const c_char> = errors
+            .into_iter()
+            .filter_map(|error| {
+                CString::new(error.to_string())
+                    .ok()
+                    .map(|c_string| c_string.into_raw().cast_const())
+            })
+            .collect();
+
+        unsafe { *out_error_count = c_strings.len() };
+
+        let boxed = c_strings.into_boxed_slice();
+        Box::into_raw(boxed).cast::<*const c_char>()
+    })
+}
+
 /// # Safety
 ///
 /// Expects both the graph pointer and encoding string pointer to be valid