Skip to content

Preserve absent source counters as unknown measurements #704

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
rudycelekli wants to merge 3 commits into from
+210 −73
Closed

Preserve absent source counters as unknown measurements #704

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
575860e
Preserve absent source counters as unknown measurements
rudycelekli Sep 29, 2026
16e7afe
fix(sources): version optional-counter parsing semantics
rudycelekli Oct 6, 2026
32e5e2d
fix(snapshots): keep title backfill provenance independent
rudycelekli Oct 6, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
132 changes: 72 additions & 60 deletions src/benchmark_radar/sources.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,7 +29,7 @@ class ConnectorPayloadError(ValueError):


FUTURE_TIMESTAMP_TOLERANCE = timedelta(minutes=5)
GITHUB_RELEASE_PARSER_VERSION = "github-releases/3"
GITHUB_RELEASE_PARSER_VERSION = "github-releases/4"
Comment thread
rudycelekli marked this conversation as resolved.


def _xml_local_name(tag: str) -> str:
Expand Down Expand Up @@ -225,6 +225,17 @@ def _request_options(config: dict[str, Any]) -> dict[str, Any]:
}


def _reported_metrics(payload: dict[str, Any], fields: dict[str, str]) -> dict[str, float]:
# An omitted or null upstream counter is unknown, not a measured zero.
# Keep an explicit 0 so downstream adoption and trend calculations can
# distinguish actual zero activity from an unreported measurement.
return {
Comment thread
rudycelekli marked this conversation as resolved.
metric: float(payload[source])
for metric, source in fields.items()
if payload.get(source) is not None and payload[source] != ""
}


def _openreview_value(content: dict[str, Any], key: str, default: Any = None) -> Any:
value = content.get(key, default)
if isinstance(value, dict) and "value" in value:
Expand Down Expand Up @@ -490,12 +501,9 @@ def fetch_huggingface(config: dict[str, Any], since: datetime, limit: int) -> li
event_kind=(
"released" if created is not None and created >= since else "updated"
),
metrics={
"downloads": float(row.get("downloads") or 0),
"likes": float(row.get("likes") or 0),
},
metrics=_reported_metrics(row, {"downloads": "downloads", "likes": "likes"}),
raw=row,
parser_version="huggingface-hub/1",
parser_version="huggingface-hub/2",
)
_clear_inherited_short_descriptions(found.values())
# `limit` is applied per request, and this fetcher issues one per kind per
Expand Down Expand Up @@ -619,12 +627,11 @@ def fetch_github(config: dict[str, Any], since: datetime, limit: int) -> list[Ra
event_kind=(
"released" if created is not None and created >= since else "updated"
),
metrics={
"stars": float(row.get("stargazers_count") or 0),
"forks": float(row.get("forks_count") or 0),
},
metrics=_reported_metrics(
row, {"stars": "stargazers_count", "forks": "forks_count"}
),
raw=row,
parser_version="github-search/1",
parser_version="github-search/2",
)
if len(rows) < min(limit, page_size):
exhausted.add(index)
Expand All @@ -638,9 +645,9 @@ def fetch_github(config: dict[str, Any], since: datetime, limit: int) -> list[Ra
)[:limit]


GITHUB_ORGANIZATIONS_PARSER_VERSION = "github-organizations/1"
KAGGLE_DATASETS_PARSER_VERSION = "kaggle-datasets/1"
HUGGINGFACE_PAPERS_PARSER_VERSION = "huggingface-papers/1"
GITHUB_ORGANIZATIONS_PARSER_VERSION = "github-organizations/2"
KAGGLE_DATASETS_PARSER_VERSION = "kaggle-datasets/2"
HUGGINGFACE_PAPERS_PARSER_VERSION = "huggingface-papers/2"


def _github_headers() -> dict[str, str]:
Expand Down Expand Up @@ -739,10 +746,9 @@ def fetch_github_organizations(
summary=github_summary(row),
event_kind="released",
organizations=[organization["display_name"]],
metrics={
"stars": float(row.get("stargazers_count") or 0),
"forks": float(row.get("forks_count") or 0),
},
metrics=_reported_metrics(
row, {"stars": "stargazers_count", "forks": "forks_count"}
),
raw={"repository": row, "organization_tier": organization["tier"]},
parser_version=GITHUB_ORGANIZATIONS_PARSER_VERSION,
)
Expand Down Expand Up @@ -822,11 +828,14 @@ def fetch_kaggle_datasets(
authors=(
[str(row.get("creatorName") or "").strip()] if row.get("creatorName") else []
),
metrics={
"downloads": float(row.get("downloadCount") or 0),
"votes": float(row.get("voteCount") or 0),
"views": float(row.get("viewCount") or 0),
},
metrics=_reported_metrics(
row,
{
"downloads": "downloadCount",
"votes": "voteCount",
"views": "viewCount",
},
),
raw=row,
parser_version=KAGGLE_DATASETS_PARSER_VERSION,
)
Expand Down Expand Up @@ -884,7 +893,7 @@ def fetch_huggingface_papers(
event_kind="discovered",
authors=authors,
artifact_urls=sorted(set(artifact_urls)),
metrics={"upvotes": float(paper.get("upvotes") or 0)},
metrics=_reported_metrics(paper, {"upvotes": "upvotes"}),
raw=row,
parser_version=HUGGINGFACE_PAPERS_PARSER_VERSION,
)
Expand Down Expand Up @@ -953,12 +962,9 @@ def fetch_zenodo_records(
if isinstance(creator, dict) and str(creator.get("name") or "").strip()
],
artifact_urls=artifact_urls,
metrics={
"downloads": float(stats.get("downloads") or 0),
"views": float(stats.get("views") or 0),
},
metrics=_reported_metrics(stats, {"downloads": "downloads", "views": "views"}),
raw=row,
parser_version="zenodo-records/1",
parser_version="zenodo-records/2",
)
return sorted(
collapse_batch_deposits(found.values()),
Expand Down Expand Up @@ -1118,9 +1124,9 @@ def fetch_crossref(
authors=author_names,
organizations=list(dict.fromkeys(organizations)),
artifact_urls=[doi_url],
metrics={"citations": float(row.get("is-referenced-by-count") or 0)},
metrics=_reported_metrics(row, {"citations": "is-referenced-by-count"}),
raw=row,
parser_version="crossref-works/1",
parser_version="crossref-works/2",
)
return sorted(found.values(), key=lambda item: item.published_at, reverse=True)[:limit]

Expand Down Expand Up @@ -1380,12 +1386,11 @@ def fetch_openaire(
),
artifact_urls=artifact_urls,
metrics={
"citations": float(citations.get("citationCount") or 0),
"downloads": float(usage.get("downloads") or 0),
"views": float(usage.get("views") or 0),
**_reported_metrics(citations, {"citations": "citationCount"}),
**_reported_metrics(usage, {"downloads": "downloads", "views": "views"}),
},
raw=row,
parser_version="openaire-graph-v3/1",
parser_version="openaire-graph-v3/2",
)
return sorted(found.values(), key=lambda item: item.published_at, reverse=True)[:limit]

Expand Down Expand Up @@ -1646,13 +1651,16 @@ def fetch_datacite(
authors=authors,
organizations=list(dict.fromkeys(organizations)),
artifact_urls=artifact_urls,
metrics={
"citations": float(attributes.get("citationCount") or 0),
"downloads": float(attributes.get("downloadCount") or 0),
"views": float(attributes.get("viewCount") or 0),
},
metrics=_reported_metrics(
attributes,
{
"citations": "citationCount",
"downloads": "downloadCount",
"views": "viewCount",
},
),
raw=row,
parser_version="datacite-dois/1",
parser_version="datacite-dois/2",
)
return sorted(found.values(), key=lambda item: item.published_at, reverse=True)[:limit]

Expand Down Expand Up @@ -1869,12 +1877,15 @@ def fetch_semantic_scholar(
if isinstance(author, dict) and author.get("name")
],
artifact_urls=sorted(set(artifact_urls)),
metrics={
"citations": float(row.get("citationCount") or 0),
"influential_citations": float(row.get("influentialCitationCount") or 0),
},
metrics=_reported_metrics(
row,
{
"citations": "citationCount",
"influential_citations": "influentialCitationCount",
},
),
raw=row,
parser_version="semantic-scholar-graph/1",
parser_version="semantic-scholar-graph/2",
)
next_offset = _payload_dict(payload, "Semantic Scholar").get("next")
if next_offset is None or len(rows) < min(page_size, limit - len(found)):
Expand Down Expand Up @@ -2000,6 +2011,14 @@ def fetch_github_releases(
isinstance(asset, dict) for asset in assets
):
raise ConnectorPayloadError("GitHub release assets must be an array")
# A release with no assets has zero downloads; an omitted assets
# list or an asset without its counter has unknown downloads.
download_metrics = (
{"downloads": float(sum(int(asset["download_count"]) for asset in assets))}
if row.get("assets") is not None
and all(asset.get("download_count") not in (None, "") for asset in assets)
else {}
)
found[f"{repository}@{tag}"] = RadarItem(
source="GitHub Release",
source_id=f"{repository}@{tag}",
Expand All @@ -2015,15 +2034,7 @@ def fetch_github_releases(
else []
),
artifact_urls=[f"https://github.com/{repository}"],
metrics={
"downloads": float(
sum(
int(asset.get("download_count") or 0)
for asset in assets
if isinstance(asset, dict)
)
)
},
metrics=download_metrics,
raw=row,
parser_version=GITHUB_RELEASE_PARSER_VERSION,
)
Expand Down Expand Up @@ -2061,8 +2072,9 @@ def fetch_github_releases(
)
if not isinstance(repository_payload, dict):
raise ConnectorPayloadError("GitHub repository metadata was not an object")
stars = float(repository_payload.get("stargazers_count") or 0)
forks = float(repository_payload.get("forks_count") or 0)
popularity = _reported_metrics(
repository_payload, {"stars": "stargazers_count", "forks": "forks_count"}
)
except Exception as error:
config.setdefault("_source_warnings", []).append(
f"{repository} metadata: {type(error).__name__}: {error}"
Expand All @@ -2071,7 +2083,7 @@ def fetch_github_releases(
for item in found.values():
if not item.source_id.startswith(f"{repository}@"):
continue
item.metrics.update({"stars": stars, "forks": forks})
item.metrics.update(popularity)
item.raw = {"release": item.raw, "repository": repository_payload}
return sorted(found.values(), key=lambda item: item.published_at, reverse=True)[:limit]

Expand Down Expand Up @@ -2170,9 +2182,9 @@ def fetch_openalex(
event_kind="released",
authors=[author for author in authors if author],
organizations=organizations,
metrics={"citations": float(row.get("cited_by_count") or 0)},
metrics=_reported_metrics(row, {"citations": "cited_by_count"}),
raw=row,
parser_version="openalex-works/1",
parser_version="openalex-works/2",
)
return list(found.values())

Expand Down
Loading
Loading