Art-gallery/Documentation/influence-import.md

Influence links import & curator tool

Curator tool to list / add / delete influence edges, visualize an artist neighborhood, and import CSV / JSON / XLSX files through a mapping wizard.

Entry: header Influences (curator session). Data lives in painting_influence_sources (legacy painting_influences mirrored for painting→painting).

---

Expansion rule (artist-level files)

Input workbooks such as Inputs/artist_influences_web_sources.xlsx are artist-centric. On import:

• Influenced by tokens → attach as sources on all paintings of the subject artist
• Influenced tokens that resolve to an artist → reverse link: subject artist becomes a source on all paintings of the influenced artist (same as PainterPalette Influencedon)

Unresolved names (artists / movements / paintings not in the DB) are skipped with warnings — no auto-create stubs.

---

Wizard column roles

Role	Meaning
subject_artist	Artist the row is about (required)
subject_painting	Optional work hint (contextual; expansion still uses all works)
influenced_by	Who/what influenced the subject (; / , separated)
influenced	Who the subject influenced
notes / reference / source_url	Citation metadata (URLs scraped from reference text)
ignore	Skip column

Presets

Preset	Typical headers	Example file
Web sources	Artist, Painting, Influenced by, Influenced, Reference (source + link)	Inputs/artist_influences_web_sources.xlsx
Story of Art	Same Title Case (+ chapter reference column)	Inputs/story_of_art_influences.xlsx
Gariff influential painters	artist, painting, influenced by, influenced, reference	Inputs/gariff_influential_painters_influences.xlsx
Art influences	artist, painting, influenced_by, influenced, reference	Inputs/art_influences.xlsx
Custom	Map any columns manually	—

Token classification order: artist → movement → painting title (under subject artist, then global).

Committed edges use confidence=curated, discovered_via=import-wizard.

Book extract: Gariff (2008)

Inputs/gariff_influential_painters_influences.xlsx is distilled from David Gariff et al., The World's Most Influential Painters and the Artists They Inspired (Herbert Press / Quarto, 2008; local PDF Inputs/1.pdf, not committed).

• Columns match the import wizard (artist / painting / influenced by / influenced / reference).
• artist and painting are limited to catalog slices in Inputs/artistslistdbdata-*.csv and Inputs/paintinglistdbdata-*.csv (regenerate those from prod/dev as needed).
• influenced by / influenced may name artists, movements, or works from the book (including entities outside those CSVs).
• Multiple rows per subject when the book states several influences.
• Import via curator Influences → Import (Art influences / custom mapping) or keep as a curated source workbook.

See also Inputs/story_of_art_influences.xlsx (Gombrich) and optional local Inputs/janson_short_history_influences.xlsx (same column shape).

Duplicate file / data guard

Each successful commit stores SHA-256 fingerprints in curator_audit_log (influence.import details):

• contentHash — raw file bytes
• payloadHash — normalized mapped rows (same data under another filename still matches)

On parse/preview, if either hash matches a prior import, the UI warns and blocks commit unless the curator checks Import anyway (force). Individual edges remain unique via DB ON CONFLICT either way.

---

API (curator)

Method	Path	Purpose
GET	/api/influences	List (artistId, paintingId, q, limit, offset)
GET	/api/influences/graph?artistId=	Nodes + edges for SVG neighborhood
POST	/api/influences	Create one edge
PATCH	/api/influences/:id	Update metadata / remap source
DELETE	/api/influences/:id	Delete (+ legacy mirror)
POST	/api/influences/import/parse	{ filename, contentBase64, sheet? } → columns, hashes, alreadyImported
POST	/api/influences/import/preview	{ rows, mapping, contentHash?, payloadHash? } → proposals + duplicate check
POST	/api/influences/import/commit	{ proposals, fileName?, contentHash?, payloadHash?, force? } — 409 ALREADY_IMPORTED unless force

Audit: influence.create / update / delete / import in curator_audit_log.

---

CLI still available

• npm run dev:update-influences — curated scripts/art-influences-data.js
• npm run dev:import-painter-palette — PainterPalette CSV

The wizard is the interactive path for ad-hoc spreadsheets under Inputs/.

See also API.md, DB_structure.md, Plans.md.