Fix searching for files by path

3 files changed, 48 insertions, 14 deletions

diff --git a/changelogs/unreleased/36939-fix-find-blobs-by-path.yml b/changelogs/unreleased/36939-fix-find-blobs-by-path.yml
new file mode 100644
index 00000000000..b48b10049ed
--- /dev/null
+++ b/changelogs/unreleased/36939-fix-find-blobs-by-path.yml
@@ -0,0 +1,5 @@
+---
+title: Fix searching for files by path
+merge_request: 13798
+author:
+type: fixed
diff --git a/lib/gitlab/file_finder.rb b/lib/gitlab/file_finder.rb
index 093d9ed8092..10ffc345bd5 100644
--- a/lib/gitlab/file_finder.rb
+++ b/lib/gitlab/file_finder.rb
@@ -6,27 +6,48 @@ module Gitlab
 
     attr_reader :project, :ref
 
+    delegate :repository, to: :project
+
     def initialize(project, ref)
       @project = project
       @ref = ref
     end
 
     def find(query)
-      blobs = project.repository.search_files_by_content(query, ref).first(BATCH_SIZE)
-      found_file_names = Set.new
+      by_content = find_by_content(query)
 
-      results = blobs.map do |blob|
-        blob = Gitlab::ProjectSearchResults.parse_search_result(blob)
-        found_file_names << blob.filename
+      already_found = Set.new(by_content.map(&:filename))
+      by_filename = find_by_filename(query, except: already_found)
 
-        [blob.filename, blob]
-      end
+      (by_content + by_filename)
+        .sort_by(&:filename)
+        .map { |blob| [blob.filename, blob] }
+    end
 
-      project.repository.search_files_by_name(query, ref).first(BATCH_SIZE).each do |filename|
-        results << [filename, OpenStruct.new(ref: ref)] unless found_file_names.include?(filename)
-      end
+    private
 
-      results.sort_by(&:first)
+    def find_by_content(query)
+      results = repository.search_files_by_content(query, ref).first(BATCH_SIZE)
+      results.map { |result| Gitlab::ProjectSearchResults.parse_search_result(result) }
+    end
+
+    def find_by_filename(query, except: [])
+      filenames = repository.search_files_by_name(query, ref).first(BATCH_SIZE)
+      filenames.delete_if { |filename| except.include?(filename) } unless except.empty?
+
+      blob_refs = filenames.map { |filename| [ref, filename] }
+      blobs = Gitlab::Git::Blob.batch(repository, blob_refs, blob_size_limit: 1024)
+
+      blobs.map do |blob|
+        Gitlab::SearchResults::FoundBlob.new(
+          id: blob.id,
+          filename: blob.path,
+          basename: File.basename(blob.path),
+          ref: ref,
+          startline: 1,
+          data: blob.data
+        )
+      end
     end
   end
 end
diff --git a/spec/lib/gitlab/file_finder_spec.rb b/spec/lib/gitlab/file_finder_spec.rb
index 3fb6315a39a..07cb10e563e 100644
--- a/spec/lib/gitlab/file_finder_spec.rb
+++ b/spec/lib/gitlab/file_finder_spec.rb
@@ -7,15 +7,23 @@ describe Gitlab::FileFinder do
 
     it 'finds by name' do
       results = finder.find('files')
-      expect(results.map(&:first)).to include('files/images/wm.svg')
+
+      filename,  blob = results.find { |_, blob| blob.filename == 'files/images/wm.svg' }
+      expect(filename).to eq('files/images/wm.svg')
+      expect(blob).to be_a(Gitlab::SearchResults::FoundBlob)
+      expect(blob.ref).to eq(finder.ref)
+      expect(blob.data).not_to be_empty
     end
 
     it 'finds by content' do
       results = finder.find('files')
 
-      blob = results.select { |result| result.first == "CHANGELOG" }.flatten.last
+      filename, blob = results.find { |_, blob| blob.filename == 'CHANGELOG' }
 
-      expect(blob.filename).to eq("CHANGELOG")
+      expect(filename).to eq('CHANGELOG')
+      expect(blob).to be_a(Gitlab::SearchResults::FoundBlob)
+      expect(blob.ref).to eq(finder.ref)
+      expect(blob.data).not_to be_empty
     end
   end
 end