Sustainability Fulltext Fetch

Core Goal

• Read relevant DOI entries from RSS metadata DB.
• Write fetched content into a separate fulltext DB.
• Process only relevant entries (is_relevant=1).
• Prefer API metadata retrieval by DOI (OpenAlex first, Semantic Scholar fallback).
• Fallback to webpage fulltext extraction when API metadata is unavailable.
• Persist one content row per DOI in entry_content.

Triggering Conditions

• Receive a request to enrich relevant DOI records with abstract/fulltext content.
• Receive a request to replace webpage-first crawling with API-first enrichment.
• Need retry-safe incremental updates without duplicate rows.

Workflow

1. Ensure upstream DOI/relevance data exists.

export SUSTAIN_RSS_DB_PATH="/absolute/path/to/workspace-rss-bot/sustainability_rss.db"
export SUSTAIN_FULLTEXT_DB_PATH="/absolute/path/to/workspace-rss-bot/sustainability_fulltext.db"
python3 scripts/fulltext_fetch.py init-db --content-db "$SUSTAIN_FULLTEXT_DB_PATH"

1. Run incremental sync (API first, webpage fallback).

python3 scripts/fulltext_fetch.py sync \
  --rss-db "$SUSTAIN_RSS_DB_PATH" \
  --content-db "$SUSTAIN_FULLTEXT_DB_PATH" \
  --limit 50 \
  --openalex-email "[email protected]" \
  --api-min-chars 80 \
  --min-chars 300

1. Fetch one DOI on demand.

python3 scripts/fulltext_fetch.py fetch-entry \
  --rss-db "$SUSTAIN_RSS_DB_PATH" \
  --content-db "$SUSTAIN_FULLTEXT_DB_PATH" \
  --doi "10.1038/nature12373"

1. Inspect stored content state.

python3 scripts/fulltext_fetch.py list-content \
  --rss-db "$SUSTAIN_RSS_DB_PATH" \
  --content-db "$SUSTAIN_FULLTEXT_DB_PATH" \
  --status ready \
  --limit 100

Data Contract

• Reads from RSS DB entries:

- doi, doiissurrogate, isrelevant, canonicalurl, url, title.

• Writes to fulltext DB entry_content (primary key doi):

- source URL/status/extractor - contentkind (abstract or fulltext) - contenttext, contenthash, contentlength - retry fields and timestamps.

Extraction Priority

1. API metadata path:

• OpenAlex by DOI.
• Semantic Scholar fallback by DOI.
• If accepted (--api-min-chars), persist as content_kind=abstract.

1. Webpage fallback path:

• Use canonical_url then url.
• Extract with trafilatura when available, else built-in HTML parser.
• Persist as content_kind=fulltext.

Update Semantics

• Upsert key: doi.
• Success: status ready, reset retry counters.
• Failure with existing ready row: keep old content, record latest error.
• Failure without ready row: set status=failed, increment retry state.

Configurable Parameters

• --rss-db
• --content-db
• SUSTAINRSSDB_PATH
• SUSTAINFULLTEXTDB_PATH
• --limit
• --force
• --only-failed
• --refetch-days
• --timeout
• --max-bytes
• --min-chars
• --openalex-email / OPENALEX_EMAIL
• --s2-api-key / S2APIKEY
• --api-timeout
• --api-min-chars
• --disable-api-metadata
• --max-retries
• --retry-backoff-minutes
• --user-agent
• --disable-trafilatura
• --fail-on-errors

Error Handling

• Missing DOI-keyed entries table: stop with actionable message.
• RSS DB and fulltext DB path collision: fail fast and require separate files.
• API/network/HTTP failures: record failures and continue queue.
• Webpage non-text content: mark failed for that DOI.
• Short extraction: fail by threshold to avoid low-quality content.

References

• references/schema.md
• references/fetch-rules.md

Assets

• assets/config.example.json

Scripts

• scripts/fulltext_fetch.py