refactored everything to use linkname helpers

benchmark/job/cli.py

@@ -6,6 +6,7 @@
     get_tables_dirname,
     get_workload_dirname,
     get_workload_suffix,
+    name_to_linkname,
 )
 
 from benchmark.constants import DEFAULT_SCALE_FACTOR
@@ -255,7 +256,7 @@ def _generate_job_workload(
         get_workload_suffix("job", query_subset=query_subset),
     )
     expected_workload_symlink_path = dbgym_workspace.dbgym_cur_symlinks_path / (
-        workload_name + ".link"
+        name_to_linkname(workload_name)
     )
     if expected_workload_symlink_path.exists():
         logging.getLogger(DBGYM_LOGGER_NAME).info(
@@ -281,7 +282,7 @@ def _generate_job_workload(
 
     with open(workload_path / "order.txt", "w") as f:
         queries_parent_path = dbgym_workspace.dbgym_cur_symlinks_path / (
-            JOB_QUERIES_DNAME + ".link"
+            name_to_linkname(JOB_QUERIES_DNAME)
         )
 
         for qname in query_names:

benchmark/tpch/cli.py

@@ -8,6 +8,7 @@
     get_workload_suffix,
     get_workload_symlink_path,
     linkname_to_name,
+    name_to_linkname,
 )
 
 from benchmark.constants import DEFAULT_SCALE_FACTOR
@@ -98,7 +99,7 @@ def _get_queries_dirname(seed: int, scale_factor: float) -> str:
 
 def _clone_tpch_kit(dbgym_workspace: DBGymWorkspace) -> None:
     expected_symlink_path = dbgym_workspace.dbgym_cur_symlinks_path / (
-        TPCH_KIT_DIRNAME + ".link"
+        name_to_linkname(TPCH_KIT_DIRNAME)
     )
     if expected_symlink_path.exists():
         logging.getLogger(DBGYM_LOGGER_NAME).info(
@@ -122,14 +123,14 @@ def _generate_tpch_queries(
     dbgym_workspace: DBGymWorkspace, seed_start: int, seed_end: int, scale_factor: float
 ) -> None:
     tpch_kit_path = dbgym_workspace.dbgym_cur_symlinks_path / (
-        TPCH_KIT_DIRNAME + ".link"
+        name_to_linkname(TPCH_KIT_DIRNAME)
     )
     logging.getLogger(DBGYM_LOGGER_NAME).info(
         f"Generating queries: [{seed_start}, {seed_end}]"
     )
     for seed in range(seed_start, seed_end + 1):
         expected_queries_symlink_path = dbgym_workspace.dbgym_cur_symlinks_path / (
-            _get_queries_dirname(seed, scale_factor) + ".link"
+            name_to_linkname(_get_queries_dirname(seed, scale_factor))
         )
         if expected_queries_symlink_path.exists():
             continue
@@ -155,7 +156,7 @@ def _generate_tpch_queries(
 
 def _generate_tpch_tables(dbgym_workspace: DBGymWorkspace, scale_factor: float) -> None:
     tpch_kit_path = dbgym_workspace.dbgym_cur_symlinks_path / (
-        TPCH_KIT_DIRNAME + ".link"
+        name_to_linkname(TPCH_KIT_DIRNAME)
     )
     expected_tables_symlink_path = get_tables_symlink_path(
         dbgym_workspace.dbgym_workspace_path, "tpch", scale_factor
@@ -225,7 +226,7 @@ def _generate_tpch_workload(
     with open(workload_path / "order.txt", "w") as f:
         for seed in range(seed_start, seed_end + 1):
             queries_parent_path = dbgym_workspace.dbgym_cur_symlinks_path / (
-                _get_queries_dirname(seed, scale_factor) + ".link"
+                name_to_linkname(_get_queries_dirname(seed, scale_factor))
             )
 
             for qname in query_names:

gymlib_package/gymlib/symlinks_paths.py

@@ -50,7 +50,7 @@ def get_tables_symlink_path(
         workspace_path
         / SYMLINKS_DNAME
         / DBGYM_APP_NAME
-        / (get_tables_dirname(benchmark, scale_factor) + ".link")
+        / name_to_linkname(get_tables_dirname(benchmark, scale_factor))
     )
 
 
@@ -61,7 +61,7 @@ def get_workload_symlink_path(
         workspace_path
         / SYMLINKS_DNAME
         / DBGYM_APP_NAME
-        / (get_workload_dirname(benchmark, scale_factor, suffix) + ".link")
+        / name_to_linkname(get_workload_dirname(benchmark, scale_factor, suffix))
     )
 
 
@@ -80,16 +80,20 @@ def get_dbdata_tgz_symlink_path(
         workspace_path
         / SYMLINKS_DNAME
         / DBGYM_APP_NAME
-        / (get_dbdata_tgz_filename(benchmark_name, scale_factor) + ".link")
+        / name_to_linkname(get_dbdata_tgz_filename(benchmark_name, scale_factor))
     )
 
 
-# TODO: refactor stuff to use this
+def is_linkname(name: str) -> bool:
+    assert not name.endswith(".link.link")
+    return name.endswith(".link")
+
+
 def name_to_linkname(name: str) -> str:
-    assert not name.endswith(".link")
+    assert not is_linkname(name)
     return f"{name}.link"
 
 
 def linkname_to_name(linkname: str) -> str:
-    assert linkname.endswith(".link")
+    assert is_linkname(linkname)
     return linkname[: -len(".link")]

util/workspace.py

@@ -12,6 +12,7 @@
 from typing import IO, Any, Optional
 
 import yaml
+from gymlib.symlinks_paths import is_linkname, name_to_linkname
 
 from util.log import DBGYM_LOGGER_NAME
 
@@ -143,9 +144,9 @@ def link_result(
             link_name = custom_link_name
         else:
             if os.path.isfile(result_path):
-                link_name = basename_of_path(result_path) + ".link"
+                link_name = name_to_linkname(basename_of_path(result_path))
             elif os.path.isdir(result_path):
-                link_name = basename_of_path(result_path) + ".link"
+                link_name = name_to_linkname(basename_of_path(result_path))
             else:
                 raise AssertionError("result_path must be either a file or dir")
 
@@ -156,8 +157,8 @@ def link_result(
         # Note that in a multi-threaded setting, this might remove one created by a process in the same run,
         #   meaning it's not "old" by our definition of "old". However, we'll always end up with a symlink
         #   file of the current run regardless of the order of threads.
-        assert link_name.endswith(".link") and not link_name.endswith(
-            ".link.link"
+        assert is_linkname(
+            link_name
         ), f'link_name ({link_name}) should end with ".link"'
         symlink_path = symlink_parent_path / link_name
         try_remove_file(symlink_path)
@@ -210,7 +211,7 @@ def save_file(self, path: Path) -> None:
             parent_path = parent_path_of_path(path)
             if parent_path.samefile(run_path):
                 fname = basename_of_path(path)
-                symlink_path = self.dbgym_this_run_path / (fname + ".link")
+                symlink_path = self.dbgym_this_run_path / name_to_linkname(fname)
                 try_remove_file(symlink_path)
                 try_create_symlink(path, symlink_path)
             # Otherwise, we know the path file is _not_ directly inside run_path dir.
@@ -225,7 +226,9 @@ def save_file(self, path: Path) -> None:
 
                 # Create symlink
                 open_base_dname = basename_of_path(base_path)
-                symlink_path = self.dbgym_this_run_path / (open_base_dname + ".link")
+                symlink_path = self.dbgym_this_run_path / name_to_linkname(
+                    open_base_dname
+                )
                 try_remove_file(symlink_path)
                 try_create_symlink(base_path, symlink_path)
         # If the file wasn't generated by a run, we can't just symlink it because we don't know that it's immutable.
@@ -461,7 +464,7 @@ def try_create_symlink(src_path: Path, dst_path: Path) -> None:
     Our functions that create symlinks might be called by multiple processes at once
     during HPO. Thus, this is a thread-safe way to create a symlink.
     """
-    assert dst_path.suffix == ".link"
+    assert is_linkname(dst_path.name)
     try:
         os.symlink(src_path, dst_path)
     except FileExistsError: