diff --git a/web/wall.html b/web/wall.html index 5a2b6a4..4eea2dd 100644 --- a/web/wall.html +++ b/web/wall.html @@ -1277,11 +1277,7 @@ function showCard(data) { // this piece's dc.subject, one chip each (accent/case duplicates folded) const tagsEl = document.getElementById('card-tags'); tagsEl.replaceChildren(); - const seen = new Set(); - [].concat(meta.subject || []).forEach(t => { - const k = deburr(String(t)).trim(); - if (!k || seen.has(k)) return; - seen.add(k); + subjectsOf(meta).forEach(t => { const chip = document.createElement('span'); chip.textContent = t; tagsEl.appendChild(chip); @@ -1408,6 +1404,20 @@ const authorKey = name => deburr(name).replace(/[^a-z0-9 ]/g, ' ').split(/\s+/). const subjectKey = t => deburr(String(t)).replace(/\s+/g, ' ').trim(); const SUBJECT_MIN = 20; +// the catalogue repeats most subjects in English ("Railroads - Buildings and +// structures - Porto Velho (Rondônia, Brazil)" beside "Ferrovias - …"); keep the +// Portuguese side. An English-only record keeps what it has rather than none. +const EN_WORD = new RegExp('\\b(?:brazil|state|buildings?|structures|and|of|the|railroads?|rivers?|bridges|streets|church(?:es)?|photographs?|public|dwellings|portraits?|stations?|construction|region|waterfalls|plazas|mountains|women|history|life|hospitals?|views?|colon(?:y|ies)|slaves|panoramic|aerial|harbors?|ports?|forts?|islands?|beaches|houses|schools?|palaces?|squares?|paraguay|amazon|falls|lakes?|people|children|boats|ships|trains?|roads?|cities|city|towns?|landscapes?|factories|industry)\\b', 'i'); +function subjectsOf(meta) { + const seen = new Set(), all = []; + [].concat((meta || {}).subject || []).forEach(t => { + const k = subjectKey(t); + if (k && !seen.has(k)) { seen.add(k); all.push(String(t).trim()); } + }); + const pt = all.filter(t => !EN_WORD.test(t)); + return pt.length ? pt : all; +} + function setupFilters(mapData) { const entries = Object.values(mapData.map); const instCounts = new Map(), typeCounts = new Map(), authorGroups = new Map(), subjectGroups = new Map(); @@ -1418,8 +1428,7 @@ function setupFilters(mapData) { if (meta.community) instCounts.set(meta.community, (instCounts.get(meta.community) || 0) + 1); const t = cleanType(meta.type); if (t) typeCounts.set(t, (typeCounts.get(t) || 0) + 1); - new Set([].concat(meta.subject || []).map(subjectKey)).forEach(k => { - if (!k) return; + subjectsOf(meta).map(subjectKey).forEach(k => { subjectGroups.set(k, (subjectGroups.get(k) || 0) + 1); }); const a = cleanAuthor(meta.contributor); @@ -1460,7 +1469,7 @@ function setupFilters(mapData) { }); const subjectNames = new Map(); // key → a readable spelling - entries.forEach(e => [].concat((e.meta || {}).subject || []).forEach(t => { + entries.forEach(e => subjectsOf(e.meta).forEach(t => { const k = subjectKey(t); if (k && subjectGroups.get(k) >= SUBJECT_MIN && !subjectNames.has(k)) subjectNames.set(k, String(t).trim()); })); @@ -1502,7 +1511,7 @@ function setupFilters(mapData) { const a = cleanAuthor(meta.contributor); m.userData._author = a ? authorKey(a) : ''; m.userData._type = cleanType(meta.type); - m.userData._subjects = new Set([].concat(meta.subject || []).map(subjectKey)); + m.userData._subjects = new Set(subjectsOf(meta).map(subjectKey)); // one haystack per piece: every metadata field, not a chosen few // the handle is the map's key, not a field of meta — it still has to be // searchable (people paste a handle or its number straight into the box)