scholarlib 0.3.0__tar.gz

scholarlib-0.3.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Heikki Wilenius
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

scholarlib-0.3.0/PKG-INFO

@@ -0,0 +1,232 @@
+Metadata-Version: 2.4
+Name: scholarlib
+Version: 0.3.0
+Summary: Shared bibliographic-API library behind the scholarfocus and litreview agent skills
+Author: Heikki Wilenius
+License-Expression: MIT
+Project-URL: Homepage, https://github.com/wilenius/scholarfocus-skill
+Project-URL: Repository, https://github.com/wilenius/scholarfocus-skill
+Project-URL: Issues, https://github.com/wilenius/scholarfocus-skill/issues
+Keywords: bibliography,openalex,orcid,literature-review,agent-skills
+Classifier: Development Status :: 4 - Beta
+Classifier: Environment :: Console
+Classifier: Intended Audience :: Science/Research
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Scientific/Engineering :: Information Analysis
+Classifier: Topic :: Text Processing :: Indexing
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: requests>=2.31.0
+Requires-Dist: pyyaml>=6.0
+Provides-Extra: fast
+Requires-Dist: rapidfuzz>=3.6; extra == "fast"
+Provides-Extra: zotero
+Requires-Dist: pyzotero>=1.5; extra == "zotero"
+Provides-Extra: dev
+Requires-Dist: pytest>=8.0; extra == "dev"
+Dynamic: license-file
+
+# scholarfocus-skill
+
+Two agent skills over one shared bibliographic library, for Claude Code, OpenClaw
+and any other assistant that can run a shell command.
+
+| Skill | Object | Question it answers |
+|---|---|---|
+| [`skills/scholarfocus`](skills/scholarfocus/SKILL.md) | **people** | What does this researcher work on? Who with? What do they cite? |
+| [`skills/litreview`](skills/litreview/SKILL.md) | **topics** | What is the state of the field on X? What are the key works? |
+
+Both sit on [`scholarlib`](https://pypi.org/project/scholarlib/), which wraps OpenAlex,
+ORCID, CrossRef, CORE, Unpaywall, OpenAIRE, Semantic Scholar and an optional local
+JSTOR metadata index behind one cache, one credit budget and one deduplication layer.
+
+## Install
+
+Two steps: the command-line tool, then the skills.
+
+### 1. The tool
+
+Install `scholarlib` as a standalone tool, not into your system Python — on Arch,
+Manjaro, Debian, Ubuntu and Fedora the system interpreter is marked
+externally-managed and a bare `pip install` will refuse with *"This environment is
+externally managed"*. That refusal is correct; do not override it.
+
+**macOS**
+
+```bash
+brew install uv          # or: brew install pipx && pipx ensurepath
+uv tool install scholarlib
+```
+
+**Linux**
+
+```bash
+# Arch / Manjaro
+sudo pacman -S uv        # or: sudo pacman -S python-pipx
+
+# Debian / Ubuntu
+sudo apt install pipx    # then use `pipx install` below
+
+# Fedora
+sudo dnf install uv
+
+uv tool install scholarlib      # or: pipx install scholarlib
+```
+
+No `uv` or `pipx` available? `python3 -m venv ~/.venvs/scholarlib && ~/.venvs/scholarlib/bin/pip install scholarlib`
+works anywhere, and the commands land in `~/.venvs/scholarlib/bin/`.
+
+Verify, and create a config:
+
+```bash
+scholarfocus --version
+scholarfocus --init-config     # writes ~/.config/scholarlib/config.yaml
+```
+
+Everything in that file is optional, but **get the free OpenAlex API key** at
+<https://openalex.org/settings/api> — it raises the daily budget roughly tenfold and
+takes about thirty seconds. Fill in an email address too; it puts you in the polite
+pool at several APIs.
+
+Faster fuzzy matching is worth having on large corpora: `uv tool install 'scholarlib[fast]'`.
+
+### 2. The skills
+
+**Claude Code**, as a plugin:
+
+```
+/plugin marketplace add wilenius/scholarfocus-skill
+/plugin install scholarlib@wilenius-skills
+```
+
+**Anything else** — copy the skill directories wherever your assistant looks for
+them:
+
+```bash
+git clone https://github.com/wilenius/scholarfocus-skill
+cp -r scholarfocus-skill/skills/* ~/.claude/skills/
+```
+
+The skills call `scholarfocus` and `litreview` as ordinary commands, so they work
+from any directory and need nothing else on the path.
+
+## Usage
+
+```bash
+scholarfocus --researchers "Jane Doe"
+scholarfocus --researchers "0000-0002-1234-5678"          # ORCID is always better
+litreview --query "multispecies ethnography" --no-jstor --dry-run
+```
+
+`--dry-run` prints the credit plan and spends nothing. Use it first on an
+unfamiliar topic.
+
+### What is deterministic
+
+The library makes no LLM calls. Retrieval, ranking, clustering and rendering are
+plain code, and the output depends only on the API responses. The review prose is
+written by the agent running the skill, from that output.
+
+litreview's **thematic strands** (`--cluster`) are counts, not semantics:
+
+- `topics` (narrative default): the most frequent OpenAlex topics across the corpus,
+  up to 7 with at least 3 works each, become strands. Each work joins the strand of
+  its highest-ranked topic. Labels are OpenAlex's topic names, used verbatim.
+- `cocitation`: the references the corpus cites most become anchors. Each strand is
+  the works citing one anchor, labelled with the anchor's title.
+
+OpenAlex assigns topics with its own classifier, so strands inherit its taxonomy
+and its errors. Treat them as a skeleton to rename, merge or split, not as findings.
+
+## Zotero — recommended
+
+**Off by default, and worth turning on.** litreview can read your Zotero library
+over its local API, which changes the output in ways nothing else can:
+
+- Results you already own are marked, so a reading list separates what you have
+  from what you need to find.
+- Your Better BibTeX citekeys are reused verbatim in exported bibliographies, so
+  they match the keys already in your documents.
+- A collection can seed a review through free OpenAlex lookups — a high-precision
+  starting set at zero credit cost.
+
+Install [Zotero](https://www.zotero.org/download/), then enable *Settings →
+Advanced → "Allow other applications on this computer to communicate with
+Zotero"*. Set `zotero.enabled: true` in your config and pass `--zotero mark` (or
+`both`).
+
+For conceptual search over your library — "find things I own that are *about*
+this" — add [`zotero-mcp`](https://github.com/54yyyu/zotero-mcp), an MCP server
+with a vector index over your library. It complements this integration rather
+than replacing it: `zotero-mcp` answers semantic questions, while `--zotero mark`
+does exact ownership matching. Feed DOIs from the former back in via `--seed-doi`.
+
+The integration degrades quietly: if Zotero is closed, litreview logs one warning
+and carries on.
+
+## JSTOR — optional
+
+litreview can check a local index of JSTOR metadata for books and chapters that
+citation databases miss. It is the difference between a review that sees
+monographs and one that does not, which matters most in the humanities and
+qualitative social sciences.
+
+It requires a metadata dump that **JSTOR distributes only to subscribing
+institutions**, at <https://www.jstor.org/ta-support/metadata>. With an
+institutional account you get a gzipped JSONL file of about 1.3 GB, which builds
+into a ~6.7 GB SQLite FTS5 index in roughly eight minutes:
+
+```bash
+# point jstor.source_path at the downloaded file, then
+jstor-index build --all-disciplines     # or --disciplines Anthropology,History
+```
+
+Without it, pass `--no-jstor` and litreview will say what the gap costs. See
+[references/JSTOR.md](skills/litreview/references/JSTOR.md) for the join rate and
+what the index does and does not contain.
+
+## Data that does not live in this repo
+
+The JSTOR dump belongs outside the working tree — `~/data/jstor/` by default,
+configurable at `jstor.source_path`. The built index lands in
+`~/.local/share/scholarlib/`, the HTTP cache in `~/.cache/scholarlib/`. All are
+gitignored.
+
+## Development
+
+```bash
+git clone https://github.com/wilenius/scholarfocus-skill
+cd scholarfocus-skill
+python -m venv .venv && .venv/bin/pip install -e '.[dev]'
+.venv/bin/python -m pytest
+```
+
+A `config.yaml` at the repo root takes precedence over the user-level one, and is
+gitignored. See [AGENTS.md](AGENTS.md) for versioning, release and packaging
+conventions.
+
+## API status
+
+See [docs/api-status.md](docs/api-status.md) for live-verified credential status and
+the measured OpenAlex credit costs that the budget logic is built around.
+
+## Layout
+
+```
+scholarlib/           shared library, published to PyPI
+  apis/               one client per data source
+  http/               BaseClient, cache, credit ledger
+  pipeline/           retrieval and analysis stages
+  render/             markdown / json / bibtex output
+  jstor/              FTS5 index build and query
+  cli/                entry points
+skills/               the two SKILL.md entry points
+tests/                fixtures and unit tests
+```
+
+## License
+
+MIT — see [LICENSE](LICENSE).

scholarlib-0.3.0/README.md

@@ -0,0 +1,201 @@
+# scholarfocus-skill
+
+Two agent skills over one shared bibliographic library, for Claude Code, OpenClaw
+and any other assistant that can run a shell command.
+
+| Skill | Object | Question it answers |
+|---|---|---|
+| [`skills/scholarfocus`](skills/scholarfocus/SKILL.md) | **people** | What does this researcher work on? Who with? What do they cite? |
+| [`skills/litreview`](skills/litreview/SKILL.md) | **topics** | What is the state of the field on X? What are the key works? |
+
+Both sit on [`scholarlib`](https://pypi.org/project/scholarlib/), which wraps OpenAlex,
+ORCID, CrossRef, CORE, Unpaywall, OpenAIRE, Semantic Scholar and an optional local
+JSTOR metadata index behind one cache, one credit budget and one deduplication layer.
+
+## Install
+
+Two steps: the command-line tool, then the skills.
+
+### 1. The tool
+
+Install `scholarlib` as a standalone tool, not into your system Python — on Arch,
+Manjaro, Debian, Ubuntu and Fedora the system interpreter is marked
+externally-managed and a bare `pip install` will refuse with *"This environment is
+externally managed"*. That refusal is correct; do not override it.
+
+**macOS**
+
+```bash
+brew install uv          # or: brew install pipx && pipx ensurepath
+uv tool install scholarlib
+```
+
+**Linux**
+
+```bash
+# Arch / Manjaro
+sudo pacman -S uv        # or: sudo pacman -S python-pipx
+
+# Debian / Ubuntu
+sudo apt install pipx    # then use `pipx install` below
+
+# Fedora
+sudo dnf install uv
+
+uv tool install scholarlib      # or: pipx install scholarlib
+```
+
+No `uv` or `pipx` available? `python3 -m venv ~/.venvs/scholarlib && ~/.venvs/scholarlib/bin/pip install scholarlib`
+works anywhere, and the commands land in `~/.venvs/scholarlib/bin/`.
+
+Verify, and create a config:
+
+```bash
+scholarfocus --version
+scholarfocus --init-config     # writes ~/.config/scholarlib/config.yaml
+```
+
+Everything in that file is optional, but **get the free OpenAlex API key** at
+<https://openalex.org/settings/api> — it raises the daily budget roughly tenfold and
+takes about thirty seconds. Fill in an email address too; it puts you in the polite
+pool at several APIs.
+
+Faster fuzzy matching is worth having on large corpora: `uv tool install 'scholarlib[fast]'`.
+
+### 2. The skills
+
+**Claude Code**, as a plugin:
+
+```
+/plugin marketplace add wilenius/scholarfocus-skill
+/plugin install scholarlib@wilenius-skills
+```
+
+**Anything else** — copy the skill directories wherever your assistant looks for
+them:
+
+```bash
+git clone https://github.com/wilenius/scholarfocus-skill
+cp -r scholarfocus-skill/skills/* ~/.claude/skills/
+```
+
+The skills call `scholarfocus` and `litreview` as ordinary commands, so they work
+from any directory and need nothing else on the path.
+
+## Usage
+
+```bash
+scholarfocus --researchers "Jane Doe"
+scholarfocus --researchers "0000-0002-1234-5678"          # ORCID is always better
+litreview --query "multispecies ethnography" --no-jstor --dry-run
+```
+
+`--dry-run` prints the credit plan and spends nothing. Use it first on an
+unfamiliar topic.
+
+### What is deterministic
+
+The library makes no LLM calls. Retrieval, ranking, clustering and rendering are
+plain code, and the output depends only on the API responses. The review prose is
+written by the agent running the skill, from that output.
+
+litreview's **thematic strands** (`--cluster`) are counts, not semantics:
+
+- `topics` (narrative default): the most frequent OpenAlex topics across the corpus,
+  up to 7 with at least 3 works each, become strands. Each work joins the strand of
+  its highest-ranked topic. Labels are OpenAlex's topic names, used verbatim.
+- `cocitation`: the references the corpus cites most become anchors. Each strand is
+  the works citing one anchor, labelled with the anchor's title.
+
+OpenAlex assigns topics with its own classifier, so strands inherit its taxonomy
+and its errors. Treat them as a skeleton to rename, merge or split, not as findings.
+
+## Zotero — recommended
+
+**Off by default, and worth turning on.** litreview can read your Zotero library
+over its local API, which changes the output in ways nothing else can:
+
+- Results you already own are marked, so a reading list separates what you have
+  from what you need to find.
+- Your Better BibTeX citekeys are reused verbatim in exported bibliographies, so
+  they match the keys already in your documents.
+- A collection can seed a review through free OpenAlex lookups — a high-precision
+  starting set at zero credit cost.
+
+Install [Zotero](https://www.zotero.org/download/), then enable *Settings →
+Advanced → "Allow other applications on this computer to communicate with
+Zotero"*. Set `zotero.enabled: true` in your config and pass `--zotero mark` (or
+`both`).
+
+For conceptual search over your library — "find things I own that are *about*
+this" — add [`zotero-mcp`](https://github.com/54yyyu/zotero-mcp), an MCP server
+with a vector index over your library. It complements this integration rather
+than replacing it: `zotero-mcp` answers semantic questions, while `--zotero mark`
+does exact ownership matching. Feed DOIs from the former back in via `--seed-doi`.
+
+The integration degrades quietly: if Zotero is closed, litreview logs one warning
+and carries on.
+
+## JSTOR — optional
+
+litreview can check a local index of JSTOR metadata for books and chapters that
+citation databases miss. It is the difference between a review that sees
+monographs and one that does not, which matters most in the humanities and
+qualitative social sciences.
+
+It requires a metadata dump that **JSTOR distributes only to subscribing
+institutions**, at <https://www.jstor.org/ta-support/metadata>. With an
+institutional account you get a gzipped JSONL file of about 1.3 GB, which builds
+into a ~6.7 GB SQLite FTS5 index in roughly eight minutes:
+
+```bash
+# point jstor.source_path at the downloaded file, then
+jstor-index build --all-disciplines     # or --disciplines Anthropology,History
+```
+
+Without it, pass `--no-jstor` and litreview will say what the gap costs. See
+[references/JSTOR.md](skills/litreview/references/JSTOR.md) for the join rate and
+what the index does and does not contain.
+
+## Data that does not live in this repo
+
+The JSTOR dump belongs outside the working tree — `~/data/jstor/` by default,
+configurable at `jstor.source_path`. The built index lands in
+`~/.local/share/scholarlib/`, the HTTP cache in `~/.cache/scholarlib/`. All are
+gitignored.
+
+## Development
+
+```bash
+git clone https://github.com/wilenius/scholarfocus-skill
+cd scholarfocus-skill
+python -m venv .venv && .venv/bin/pip install -e '.[dev]'
+.venv/bin/python -m pytest
+```
+
+A `config.yaml` at the repo root takes precedence over the user-level one, and is
+gitignored. See [AGENTS.md](AGENTS.md) for versioning, release and packaging
+conventions.
+
+## API status
+
+See [docs/api-status.md](docs/api-status.md) for live-verified credential status and
+the measured OpenAlex credit costs that the budget logic is built around.
+
+## Layout
+
+```
+scholarlib/           shared library, published to PyPI
+  apis/               one client per data source
+  http/               BaseClient, cache, credit ledger
+  pipeline/           retrieval and analysis stages
+  render/             markdown / json / bibtex output
+  jstor/              FTS5 index build and query
+  cli/                entry points
+skills/               the two SKILL.md entry points
+tests/                fixtures and unit tests
+```
+
+## License
+
+MIT — see [LICENSE](LICENSE).

scholarlib-0.3.0/pyproject.toml

@@ -0,0 +1,57 @@
+[build-system]
+requires = ["setuptools>=77"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "scholarlib"
+dynamic = ["version"]
+description = "Shared bibliographic-API library behind the scholarfocus and litreview agent skills"
+readme = "README.md"
+license = "MIT"
+license-files = ["LICENSE"]
+requires-python = ">=3.11"
+authors = [{ name = "Heikki Wilenius" }]
+keywords = ["bibliography", "openalex", "orcid", "literature-review", "agent-skills"]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Environment :: Console",
+    "Intended Audience :: Science/Research",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Topic :: Scientific/Engineering :: Information Analysis",
+    "Topic :: Text Processing :: Indexing",
+]
+dependencies = [
+    "requests>=2.31.0",
+    "pyyaml>=6.0",
+]
+
+[project.optional-dependencies]
+fast = ["rapidfuzz>=3.6"]
+zotero = ["pyzotero>=1.5"]
+dev = ["pytest>=8.0"]
+
+[project.urls]
+Homepage = "https://github.com/wilenius/scholarfocus-skill"
+Repository = "https://github.com/wilenius/scholarfocus-skill"
+Issues = "https://github.com/wilenius/scholarfocus-skill/issues"
+
+[project.scripts]
+scholarfocus = "scholarlib.cli.scholarfocus:main"
+litreview = "scholarlib.cli.litreview:main"
+jstor-index = "scholarlib.cli.jstor_index:main"
+
+[tool.setuptools.dynamic]
+version = { attr = "scholarlib.__version__" }
+
+[tool.setuptools.packages.find]
+include = ["scholarlib*"]
+
+[tool.setuptools.package-data]
+scholarlib = ["config.example.yaml", "jstor/*.sql"]
+
+[tool.pyright]
+pythonVersion = "3.11"
+include = ["scholarlib", "tests"]
+reportMissingImports = "warning"

scholarlib-0.3.0/scholarlib/__init__.py

@@ -0,0 +1,3 @@
+"""Shared bibliographic-API library behind the scholarfocus and litreview skills."""
+
+__version__ = "0.3.0"

scholarlib-0.3.0/scholarlib/adapters.py

@@ -0,0 +1,94 @@
+"""Convert each source's native shape into the canonical Record."""
+
+from __future__ import annotations
+
+from typing import Optional
+
+from scholarlib.apis.openalex import reconstruct_abstract
+from scholarlib.records import Record
+
+
+def _authors_from_openalex(work: dict) -> list[str]:
+    out = []
+    for a in work.get("authorships") or []:
+        name = ((a.get("author") or {}).get("display_name"))
+        if name:
+            out.append(name)
+    return out
+
+
+def _topics_from_openalex(work: dict) -> list[str]:
+    names = []
+    for t in (work.get("topics") or [])[:5]:
+        if t.get("display_name"):
+            names.append(t["display_name"])
+    for c in (work.get("concepts") or [])[:5]:
+        if c.get("display_name") and c["display_name"] not in names:
+            names.append(c["display_name"])
+    return names
+
+
+def from_openalex_work(work: dict, *, provenance: Optional[str] = None) -> Record:
+    abstract = reconstruct_abstract(work.get("abstract_inverted_index"))
+    loc = work.get("primary_location") or {}
+    src = loc.get("source") or {}
+    best_oa = work.get("best_oa_location") or {}
+    oa = work.get("open_access") or {}
+    return Record(
+        doi=work.get("doi"),
+        openalex_id=work.get("id"),
+        title=work.get("title") or work.get("display_name"),
+        authors=_authors_from_openalex(work),
+        year=work.get("publication_year"),
+        venue=src.get("display_name"),
+        type=work.get("type"),
+        language=work.get("language"),
+        abstract=abstract,
+        abstract_source="openalex" if abstract else None,
+        cited_by_count=work.get("cited_by_count"),
+        referenced_works=list(work.get("referenced_works") or []),
+        related_works=list(work.get("related_works") or []),
+        topics=_topics_from_openalex(work),
+        oa_status=oa.get("oa_status"),
+        oa_url=best_oa.get("pdf_url") or best_oa.get("landing_page_url"),
+        provenance=[provenance] if provenance else [],
+    )
+
+
+def from_crossref_item(item: dict, *, provenance: str = "crossref") -> Record:
+    titles = item.get("title") or []
+    containers = item.get("container-title") or []
+    authors = []
+    for a in item.get("author") or []:
+        nm = " ".join(x for x in (a.get("given"), a.get("family")) if x) or a.get("name")
+        if nm:
+            authors.append(nm)
+    issued = ((item.get("issued") or {}).get("date-parts") or [[None]])[0]
+    return Record(
+        doi=item.get("DOI"),
+        title=titles[0] if titles else None,
+        authors=authors,
+        year=issued[0] if issued else None,
+        venue=containers[0] if containers else None,
+        type=item.get("type"),
+        cited_by_count=item.get("is-referenced-by-count"),
+        provenance=[provenance],
+    )
+
+
+def from_s2_paper(paper: dict, *, provenance: str = "semantic_scholar") -> Record:
+    ext = paper.get("externalIds") or {}
+    tldr = (paper.get("tldr") or {}).get("text")
+    abstract = paper.get("abstract") or tldr
+    return Record(
+        doi=ext.get("DOI"),
+        title=paper.get("title"),
+        year=paper.get("year"),
+        venue=paper.get("venue"),
+        abstract=abstract,
+        abstract_source=("semantic_scholar" if paper.get("abstract")
+                         else ("s2_tldr" if tldr else None)),
+        cited_by_count=paper.get("citationCount"),
+        topics=list(paper.get("fieldsOfStudy") or []),
+        provenance=[provenance],
+    )

scholarlib-0.3.0/scholarlib/apis/core_api.py

@@ -0,0 +1,41 @@
+"""CORE client. Key is optional but raises the rate limit from 10/min to 150/min."""
+
+from __future__ import annotations
+
+import logging
+from typing import Optional
+
+from scholarlib.http.base import BaseClient
+
+logger = logging.getLogger(__name__)
+
+BASE_URL = "https://api.core.ac.uk/v3"
+
+
+class COREClient(BaseClient):
+    name = "core"
+    base_url = BASE_URL
+
+    def __init__(self, api_key: Optional[str] = None, **kw):
+        super().__init__(**kw)
+        self.api_key = api_key
+        if not api_key:
+            logger.debug("CORE: no API key — rate limit is 10/min instead of 150/min")
+
+    def _auth_headers(self) -> dict:
+        return {"Authorization": f"Bearer {self.api_key}"} if self.api_key else {}
+
+    def search_works(self, query: str, limit: int = 10) -> list[dict]:
+        data = self.get("/search/works", {"q": query, "limit": limit})
+        return (data or {}).get("results", [])
+
+    def get_work_by_doi(self, doi: str) -> Optional[dict]:
+        d = str(doi).replace("https://doi.org/", "").strip()
+        results = self.search_works(f'doi:"{d}"', limit=1)
+        return results[0] if results else None
+
+    def get_abstract(self, doi: str) -> Optional[str]:
+        w = self.get_work_by_doi(doi)
+        if not w:
+            return None
+        return w.get("abstract") or w.get("description") or None