Specs, refactor and cleanup

.github/workflows/ci.yaml

@@ -29,6 +29,12 @@ jobs:
           ruby-version: ${{ matrix.ruby }}
           bundler-cache: true
 
+      - name: Configure Git
+        run: |
+          git config --global user.email "shayonj@gmail.com"
+          git config --global user.name "Shayon Mukherjee"
+          git config --global init.defaultBranch main
+
       - name: Bundle install
         env:
           RAILS_ENV: test
@@ -38,10 +44,10 @@ jobs:
 
       - name: Run Lint
         run: bundle exec rubocop
-      # - name: Setup upterm session
-      #   uses: lhotari/action-upterm@v1
+
       - name: Run RSpec
         run: bundle exec rspec
+
   build-push-image:
     if: startsWith(github.ref, 'refs/tags/v')
     runs-on: ubuntu-latest

Gemfile.lock

@@ -9,9 +9,14 @@ PATH
 GEM
   remote: https://rubygems.org/
   specs:
+    addressable (2.8.5)
+      public_suffix (>= 2.0.2, < 6.0)
     ast (2.4.2)
     coderay (1.1.3)
     diff-lcs (1.5.0)
+    git (1.18.0)
+      addressable (~> 2.8)
+      rchardet (~> 1.8)
     haml (6.0.7)
       temple (>= 0.8.2)
       thor
@@ -27,10 +32,12 @@ GEM
     pry (0.14.2)
       coderay (~> 1.1)
       method_source (~> 1.0)
+    public_suffix (5.0.4)
     racc (1.7.3)
     rainbow (3.1.1)
     rake (13.1.0)
     rbs (1.0.0)
+    rchardet (1.8.0)
     regexp_parser (2.8.2)
     rexml (3.2.6)
     rspec (3.12.0)
@@ -99,6 +106,7 @@ PLATFORMS
 
 DEPENDENCIES
   branch_base!
+  git
   prettier_print
   pry
   rake

branch_base.gemspec

@@ -26,6 +26,7 @@ Gem::Specification.new do |spec|
 
   spec.metadata = { "rubygems_mfa_required" => "true" }
 
+  spec.add_development_dependency("git")
   spec.add_development_dependency("prettier_print")
   spec.add_development_dependency("pry")
   spec.add_development_dependency("rake")

lib/branch_base.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-# lib/branch_base.rb
-
 require "branch_base/database"
 require "branch_base/repository"
 require "branch_base/sync"

lib/branch_base/database.rb

@@ -25,38 +25,6 @@ def last_insert_row_id
       @db.last_insert_row_id
     end
 
-    def commit_exists?(commit_hash)
-      query = "SELECT COUNT(*) FROM commits WHERE commit_hash = ?"
-      execute(query, commit_hash).first[0].positive?
-    end
-
-    def insert_commit(repo_id, commit)
-      return if commit_exists?(commit.oid)
-
-      query =
-        "INSERT INTO commits (commit_hash, repo_id, author, committer, message, timestamp) VALUES (?, ?, ?, ?, ?, ?)"
-      execute(
-        query,
-        commit.oid,
-        repo_id,
-        commit.author[:name],
-        commit.committer[:name],
-        commit.message,
-        commit.time.to_s
-      )
-    end
-
-    def get_or_insert_repo_id(name, path)
-      query = "SELECT repo_id FROM repositories WHERE url = ?"
-      result = execute(query, path).first
-      return result[0] if result
-
-      execute("INSERT INTO repositories (name, url) VALUES (?, ?)", name, path)
-      last_insert_row_id
-    end
-
-    private
-
     def setup_schema
       @db.execute_batch(<<-SQL)
         CREATE TABLE IF NOT EXISTS repositories (

lib/branch_base/sync.rb

@@ -2,43 +2,37 @@
 
 module BranchBase
   class Sync
-    BATCH_SIZE = 10_000 # Adjust this based on performance testing
+    BATCH_SIZE = 1000
 
     def initialize(database, repository)
       @db = database
       @repo = repository
     end
 
     def run
-      # Disable foreign key checks for performance
       @db.execute("PRAGMA foreign_keys = OFF")
 
-      # @db.transaction do
       repo_id = sync_repository
       sync_branches(repo_id)
       sync_commits(repo_id)
-      # end
 
-      # Re-enable foreign key checks
       @db.execute("PRAGMA foreign_keys = ON")
     end
 
-    private
-
     def sync_repository
       repo_path = @repo.path.chomp(".git/")
       repo_name = File.basename(repo_path)
 
       existing_repo_id =
         @db.execute(
           "SELECT repo_id FROM repositories WHERE url = ?",
-          [repo_path]
+          [repo_path],
         ).first
       return existing_repo_id[0] if existing_repo_id
 
       @db.execute(
         "INSERT INTO repositories (name, url) VALUES (?, ?)",
-        [repo_name, repo_path]
+        [repo_name, repo_path],
       )
       @db.last_insert_row_id
     end
@@ -81,7 +75,7 @@ def sync_commits(repo_id)
           commit.author[:name],
           commit.committer[:name],
           commit.message,
-          commit.time.to_s
+          commit.time.to_s,
         ]
 
         if batched_commits.size >= BATCH_SIZE
@@ -106,10 +100,12 @@ def sync_commits(repo_id)
       insert_files_and_commit_files(batched_files) unless batched_files.empty?
     end
 
+    private
+
     def commit_exists?(commit_hash)
       @db.execute(
         "SELECT COUNT(*) FROM commits WHERE commit_hash = ?",
-        [commit_hash]
+        [commit_hash],
       ).first[
         0
       ].positive?
@@ -120,12 +116,12 @@ def insert_commit_files(commit, repo_id)
         file_path = patch.delta.new_file[:path]
         @db.execute(
           "INSERT OR IGNORE INTO files (repo_id, file_path, latest_commit) VALUES (?, ?, ?)",
-          [repo_id, file_path, commit.oid]
+          [repo_id, file_path, commit.oid],
         )
         file_id = @db.last_insert_row_id
         @db.execute(
           "INSERT INTO commit_files (commit_hash, file_id, changes) VALUES (?, ?, ?)",
-          [commit.oid, file_id, patch.to_s]
+          [commit.oid, file_id, patch.to_s],
         )
       end
     end
@@ -134,7 +130,7 @@ def insert_commit_parents(commit)
       commit.parent_ids.each do |parent_id|
         @db.execute(
           "INSERT INTO commit_parents (commit_hash, parent_hash) VALUES (?, ?)",
-          [commit.oid, parent_id]
+          [commit.oid, parent_id],
         )
       end
     end
@@ -144,7 +140,7 @@ def insert_branches(batched_branches)
         batched_branches.each do |data|
           @db.execute(
             "INSERT OR IGNORE INTO branches (repo_id, name, head_commit) VALUES (?, ?, ?)",
-            data
+            data,
           )
         end
       end
@@ -155,7 +151,7 @@ def insert_commits(batched_commits)
         batched_commits.each do |data|
           @db.execute(
             "INSERT INTO commits (commit_hash, repo_id, author, committer, message, timestamp) VALUES (?, ?, ?, ?, ?, ?)",
-            data
+            data,
           )
         end
       end
@@ -167,12 +163,12 @@ def insert_files_and_commit_files(batched_data)
           repo_id, file_path, commit_hash, changes = data
           @db.execute(
             "INSERT OR IGNORE INTO files (repo_id, file_path, latest_commit) VALUES (?, ?, ?)",
-            [repo_id, file_path, commit_hash]
+            [repo_id, file_path, commit_hash],
           )
           file_id = @db.last_insert_row_id
           @db.execute(
             "INSERT INTO commit_files (commit_hash, file_id, changes) VALUES (?, ?, ?)",
-            [commit_hash, file_id, changes]
+            [commit_hash, file_id, changes],
           )
         end
       end

spec/branch_base/cli_spec.rb

@@ -0,0 +1,52 @@
+# frozen_string_literal: true
+require "branch_base/cli"
+require "branch_base/database"
+require "branch_base/repository"
+require "branch_base/sync"
+require "thor"
+require "rspec"
+
+RSpec.describe(BranchBase::CLI) do
+  let(:cli) { BranchBase::CLI.new }
+  let(:repo_path) { "path/to/repo" }
+  let(:expanded_repo_path) { File.expand_path(repo_path) }
+  let(:repo_name) { File.basename(expanded_repo_path) }
+  let(:db_filename) { "#{repo_name}_git_data.db" }
+
+  describe "#sync" do
+    context "when the repository path is valid" do
+      before do
+        allow(File).to receive(:directory?).and_call_original
+        allow(File).to receive(:directory?).with(
+          "#{expanded_repo_path}/.git"
+        ).and_return(true)
+        allow(BranchBase::Database).to receive(:new).with(
+          db_filename
+        ).and_return(double("Database"))
+        allow(BranchBase::Repository).to receive(:new).with(
+          expanded_repo_path
+        ).and_return(double("Repository"))
+        allow(BranchBase::Sync).to receive(:new).and_return(
+          double("Sync", run: nil)
+        )
+      end
+
+      it "syncs the repository data" do
+        expect { cli.sync(repo_path) }.to output(
+          "Repository data synced successfully for\n"
+        ).to_stdout
+      end
+    end
+
+    context "when the repository path is not valid" do
+      it "exits with an error message" do
+        allow(File).to receive(:directory?).with(
+          "#{expanded_repo_path}/.git"
+        ).and_return(false)
+        expect { cli.sync(repo_path) }.to output(
+          "The specified path is not a valid Git repository: #{expanded_repo_path}\n"
+        ).to_stdout.and raise_error(SystemExit)
+      end
+    end
+  end
+end

spec/branch_base/database_spec.rb

@@ -0,0 +1,70 @@
+# frozen_string_literal: true
+require "branch_base/database"
+require "rspec"
+
+RSpec.describe(BranchBase::Database) do
+  let(:database) { BranchBase::Database.new(":memory:") }
+
+  describe "#initialize" do
+    it "initializes with the correct schema" do
+      tables =
+        database.execute("SELECT name FROM sqlite_master WHERE type='table'")
+      expected_tables = %w[
+        repositories
+        commits
+        branches
+        files
+        commit_files
+        commit_parents
+        sqlite_sequence
+      ]
+      expect(tables.flatten).to match_array(expected_tables)
+    end
+  end
+
+  describe "#execute" do
+    it "executes a given SQL statement" do
+      result =
+        database.execute(
+          "INSERT INTO repositories (name, url) VALUES (?, ?)",
+          %w[mock_repo mock_repo/],
+        )
+      expect(result).to be_empty
+    end
+  end
+
+  describe "#prepare" do
+    it "prepares a SQL statement" do
+      statement =
+        database.prepare("INSERT INTO repositories (name, url) VALUES (?, ?)")
+      expect(statement).to be_a(SQLite3::Statement)
+    end
+  end
+
+  describe "#transaction" do
+    it "executes a block within a transaction" do
+      expect {
+        database.transaction do
+          database.execute(
+            "INSERT INTO repositories (name, url) VALUES (?, ?)",
+            %w[mock_repo mock_repo/],
+          )
+          raise "Rollback transaction"
+        end
+      }.to raise_error(RuntimeError, "Rollback transaction")
+
+      count = database.execute("SELECT COUNT(*) FROM repositories").first.first
+      expect(count).to eq(0)
+    end
+  end
+
+  describe "#last_insert_row_id" do
+    it "returns the last insert row ID" do
+      database.execute(
+        "INSERT INTO repositories (name, url) VALUES (?, ?)",
+        %w[mock_repo mock_repo/],
+      )
+      expect(database.last_insert_row_id).to be > 0
+    end
+  end
+end