correct typing for load_graph

handling of empty/nonexistent graph file

conda_forge_tick/auto_tick.py

@@ -67,7 +67,6 @@
 )
 from conda_forge_tick.utils import (
     pluck,
-    load_graph,
     dump_graph,
     CB_CONFIG,
     parse_meta_yaml,
@@ -77,6 +76,7 @@
     parse_munged_run_export,
     fold_log_lines,
     get_keys_default,
+    load_existing_graph,
 )
 from conda_forge_tick.migrators.arch import OSXArm
 from conda_forge_tick.migrators.migration_yaml import (
@@ -988,7 +988,7 @@ def initialize_migrators(
     dry_run: bool = False,
 ) -> Tuple[MigratorSessionContext, list, MutableSequence[Migrator]]:
     temp = glob.glob("/tmp/*")
-    gx = load_graph()
+    gx = load_existing_graph()
     smithy_version = eval_cmd("conda smithy --version").strip()
     pinning_version = json.loads(eval_cmd("conda list conda-forge-pinning --json"))[0][
         "version"
@@ -1415,7 +1415,7 @@ def _setup_limits():
         resource.setrlimit(resource.RLIMIT_AS, (limit_int, limit_int))
 
 
-def _update_nodes_with_bot_rerun(gx):
+def _update_nodes_with_bot_rerun(gx: nx.DiGraph):
     """Go through all the open PRs and check if they are rerun"""
 
     print("processing bot-rerun labels", flush=True)
@@ -1558,7 +1558,7 @@ def _remove_closed_pr_json():
 
 def _update_graph_with_pr_info():
     _remove_closed_pr_json()
-    gx = load_graph()
+    gx = load_existing_graph()
     _update_nodes_with_bot_rerun(gx)
     _update_nodes_with_new_versions(gx)
     dump_graph(gx)

conda_forge_tick/deploy.py

@@ -3,10 +3,10 @@
 
 from doctr.travis import run_command_hiding_token as doctr_run
 
-from .cli_context import CliContext
 from . import sensitive_env
-from .utils import load_graph
+from .cli_context import CliContext
 from .lazy_json_backends import get_lazy_json_backends, CF_TICK_GRAPH_DATA_HASHMAPS
+from .utils import load_existing_graph
 
 BUILD_URL_KEY = "CIRCLE_BUILD_URL"
 
@@ -93,7 +93,7 @@ def deploy(ctx: CliContext):
 
         # make sure the graph can load, if not we will error
         try:
-            gx = load_graph()
+            gx = load_existing_graph()
             # TODO: be more selective about which json to check
             for node, attrs in gx.nodes.items():
                 attrs["payload"]._load()

conda_forge_tick/pypi_name_mapping.py

@@ -5,24 +5,24 @@
 2: Packages MUST have a test: imports section importing it
 """
 
+import functools
 import json
 import math
-import requests
-import yaml
 import pathlib
 import traceback
-import functools
-
 from collections import Counter, defaultdict
-from typing import Dict, List, Literal, Optional, Any, Tuple, Set, TypedDict, Union
 from os.path import commonprefix
+from typing import Dict, List, Literal, Optional, Any, Tuple, Set, TypedDict, Union
+
+import requests
+import yaml
 from packaging.utils import (
     canonicalize_name as canonicalize_pypi_name,
     NormalizedName as PypiName,
 )
 
-from .utils import as_iterable, load_graph
 from .lazy_json_backends import dump, loads, get_all_keys_for_hashmap, LazyJson
+from .utils import as_iterable, load_existing_graph
 
 
 class Mapping(TypedDict):
@@ -300,7 +300,7 @@ def determine_best_matches_for_pypi_import(
         map_by_conda_name[conda_name] = m
 
     graph_file = str(pathlib.Path(".") / "graph.json")
-    gx = load_graph(graph_file)
+    gx = load_existing_graph(graph_file)
     # TODO: filter out archived feedstocks?
 
     try:

conda_forge_tick/update_prs.py

@@ -18,8 +18,8 @@
 from .executors import executor
 from .make_graph import ghctx
 from .utils import (
-    load_graph,
     github_client,
+    load_existing_graph,
 )
 
 # from conda_forge_tick.profiler import profiling
@@ -166,7 +166,7 @@ def close_dirty_prs(
 
 # @profiling
 def main(ctx: CliContext, job: int = 1, n_jobs: int = 1) -> None:
-    gx = load_graph()
+    gx = load_existing_graph()
 
     gx = close_labels(gx, ctx.dry_run, job=job, n_jobs=n_jobs)
     gx = update_graph_pr_status(gx, ctx.dry_run, job=job, n_jobs=n_jobs)

conda_forge_tick/update_upstream_versions.py

@@ -34,8 +34,7 @@
     IncrementAlphaRawURL,
     NVIDIA,
 )
-from .utils import get_keys_default
-from .utils import load_graph
+from .utils import get_keys_default, load_existing_graph
 
 T = TypeVar("T")
 
@@ -393,7 +392,7 @@ def main(
     """
     logger.info("Reading graph")
     # Graph enabled for inspection
-    gx = load_graph()
+    gx = load_existing_graph()
 
     # Check if 'versions' folder exists or create a new one;
     os.makedirs("versions", exist_ok=True)

conda_forge_tick/utils.py

@@ -74,6 +74,8 @@ def _munge_dict_repr(d):
     datetime=datetime,
 )
 
+DEFAULT_GRAPH_FILENAME = "graph.json"
+
 
 @contextlib.contextmanager
 def fold_log_lines(title):
@@ -431,7 +433,6 @@ def dump_graph_json(gx: nx.DiGraph, filename: str = "graph.json") -> None:
     links = nld["links"]
     links2 = sorted(links, key=lambda x: f'{x["source"]}{x["target"]}')
     nld["links"] = links2
-    from conda_forge_tick.lazy_json_backends import LazyJson
 
     lzj = LazyJson(filename)
     with lzj as attrs:
@@ -447,14 +448,32 @@ def dump_graph(
     dump_graph_json(gx, filename)
 
 
-def load_graph(filename: str = "graph.json") -> nx.DiGraph:
-    from conda_forge_tick.lazy_json_backends import LazyJson
+def load_existing_graph(filename: str = DEFAULT_GRAPH_FILENAME) -> nx.DiGraph:
+    """
+    Load the graph from a file using the lazy json backend.
+    If a non-existing or empty file is encountered, a FileNotFoundError is raised.
+    If you expect the graph to possibly not exist, use load_graph.
+
+    :return: the graph
+    """
+    gx = load_graph(filename)
+    if gx is None:
+        raise FileNotFoundError(f"Graph file {filename} does not exist or is empty")
+    return gx
+
 
+def load_graph(filename: str = DEFAULT_GRAPH_FILENAME) -> Optional[nx.DiGraph]:
+    """
+    Load the graph from a file using the lazy json backend.
+    If you expect the graph to exist, use load_existing_graph.
+
+    :return: the graph, or None if the file does not exist or is empty JSON
+    """
     dta = copy.deepcopy(LazyJson(filename).data)
     if dta:
         return nx.node_link_graph(dta)
     else:
-        raise FileNotFoundError("Graph file not found.")
+        return None
 
 
 # TODO: This type does not support generics yet sadly