Remove unnecessary wording from codebase (#26)

Restore required words
microsoft · Apr 3, 2024 · 71277e2 · 71277e2
1 parent b2ee5ab
commit 71277e2
Show file tree

Hide file tree

Showing 4 changed files with 12 additions and 36 deletions.
diff --git a/dictionary.txt b/dictionary.txt
@@ -117,7 +117,6 @@ scatterplot
 SITREP
 megapipeline
 columnwise
-janky
 docstore
 datasource
 devcontainers
@@ -131,28 +130,6 @@ MSRC
 Arrary
 
 # Prompt Inputs
-VOLKIMIR
-VLODIMIR
-VOLODYMYR
-VOLADIMIR
-VOLODYMYR
-VOLDEMIR
-VOLDEMIR
-VOLYNMYR
-VOLYNMYR
-VOLYN
-ZELENS
-ZELENSKY
-ZELENSKI
-ZELENSKIY
-ZELENSKYI
-ZELENSKII
-ROSIYA
-ROSSIA
-ROSIYA
-ROSIJA
-ROSIA
-OLEKSANDROVYCH
 Siamak
 Namazi
 Emad
@@ -163,7 +140,6 @@ Tahbaz
 abila
 abilas
 payus
-derp
 POKRALLY
 dulce
 

diff --git a/python/graphrag/graphrag/index/workflows/v1/create_final_communities.py b/python/graphrag/graphrag/index/workflows/v1/create_final_communities.py
@@ -148,7 +148,7 @@ def build_steps(
 
 
 create_community_title_wf = [
-    # janky way of string concat "Community " + id
+    # Hack to string concat "Community " + id
     {
         "verb": "fill",
         "args": {

diff --git a/python/graphrag/tests/unit/indexing/cache/test_file_pipeline_cache.py b/python/graphrag/tests/unit/indexing/cache/test_file_pipeline_cache.py
@@ -45,11 +45,11 @@ async def test_child_cache(self):
         await self.cache.set("test1", "test1")
         assert os.path.exists(f"{TEMP_DIR}/test1")
 
-        child = self.cache.child("derp")
-        assert os.path.exists(f"{TEMP_DIR}/derp")
+        child = self.cache.child("test")
+        assert os.path.exists(f"{TEMP_DIR}/test")
 
         await child.set("test2", "test2")
-        assert os.path.exists(f"{TEMP_DIR}/derp/test2")
+        assert os.path.exists(f"{TEMP_DIR}/test/test2")
 
         await self.cache.set("test1", "test1")
         await self.cache.delete("test1")

diff --git a/python/graphrag/tests/unit/indexing/default_config/test_default_config.py b/python/graphrag/tests/unit/indexing/default_config/test_default_config.py
@@ -90,10 +90,10 @@ def test_text_input_returns_correct_config(self):
             "GRAPHRAG_ENTITY_EXTRACTION_MAX_GLEANINGS": "112",
             "GRAPHRAG_SUMMARIZE_DESCRIPTIONS_MAX_LENGTH": "12345",
             "GRAPHRAG_COMMUNITY_REPORT_MAX_LENGTH": "23456",
-            "GRAPHRAG_CLAIM_EXTRACTION_DESCRIPTION": "derp 123",
+            "GRAPHRAG_CLAIM_EXTRACTION_DESCRIPTION": "test 123",
             "GRAPHRAG_MAX_CLUSTER_SIZE": "123",
             "GRAPHRAG_UMAP_ENABLED": "true",
-            "GRAPHRAG_ENCODING_MODEL": "derp123",
+            "GRAPHRAG_ENCODING_MODEL": "test123",
             "GRAPHRAG_SKIP_WORKFLOWS": "a,b,c",
         },
         clear=True,
@@ -103,7 +103,7 @@ def test_create_parameters_from_env_vars(self) -> None:
         assert parameters.llm["api_key"] == "test"
         assert parameters.llm["model"] == "test-llm"
         assert parameters.parallelization["num_threads"] == 987
-        assert parameters.encoding_model == "derp123"
+        assert parameters.encoding_model == "test123"
         assert parameters.skip_workflows == ["a", "b", "c"]
         assert parameters.storage.type == PipelineStorageType.blob
         assert parameters.storage.connection_string == "test_cs"
@@ -128,7 +128,7 @@ def test_create_parameters_from_env_vars(self) -> None:
         assert parameters.entity_extraction.max_gleanings == 112
         assert parameters.summarize_descriptions.max_length == 12345
         assert parameters.community_reports.max_length == 23456
-        assert parameters.claim_extraction.description == "derp 123"
+        assert parameters.claim_extraction.description == "test 123"
         assert parameters.cluster_graph.max_cluster_size == 123
         assert parameters.umap.enabled
 
@@ -179,19 +179,19 @@ def test_create_parameters(self) -> None:
                 community_reports=CommunityReportsConfigModel(
                     max_length=23456,
                 ),
-                claim_extraction=ClaimExtractionConfigModel(description="derp 123"),
+                claim_extraction=ClaimExtractionConfigModel(description="test 123"),
                 cluster_graph=ClusterGraphConfigModel(
                     max_cluster_size=123,
                 ),
                 umap=UmapConfigModel(enabled=True),
-                encoding_model="derp123",
+                encoding_model="test123",
                 skip_workflows=["a", "b", "c"],
             ),
             ".",
         )
         assert parameters.llm["api_key"] == "test"
         assert parameters.llm["model"] == "test-llm"
-        assert parameters.encoding_model == "derp123"
+        assert parameters.encoding_model == "test123"
         assert parameters.skip_workflows == ["a", "b", "c"]
         assert parameters.storage.type == PipelineStorageType.blob
         assert parameters.storage.connection_string == "test_cs"
@@ -217,7 +217,7 @@ def test_create_parameters(self) -> None:
         assert parameters.entity_extraction.max_gleanings == 112
         assert parameters.summarize_descriptions.max_length == 12345
         assert parameters.community_reports.max_length == 23456
-        assert parameters.claim_extraction.description == "derp 123"
+        assert parameters.claim_extraction.description == "test 123"
         assert parameters.cluster_graph.max_cluster_size == 123
         assert parameters.umap.enabled