@glossarist/concept-browser 0.7.122 → 0.7.124
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@glossarist/concept-browser",
|
|
3
|
-
"version": "0.7.
|
|
3
|
+
"version": "0.7.124",
|
|
4
4
|
"description": "Vue SPA for browsing Glossarist terminology datasets with cross-reference resolution, graph visualization, and multi-language support",
|
|
5
5
|
"type": "module",
|
|
6
6
|
"bin": {
|
|
@@ -43,7 +43,7 @@
|
|
|
43
43
|
"astro": "^7.0.7",
|
|
44
44
|
"d3": "^7.9.0",
|
|
45
45
|
"favicons": "^7.2.0",
|
|
46
|
-
"glossarist": "^0.4.
|
|
46
|
+
"glossarist": "^0.4.55",
|
|
47
47
|
"js-yaml": "^4.1.0",
|
|
48
48
|
"jszip": "^3.10.1",
|
|
49
49
|
"n3": "^1.26.0",
|
package/scripts/generate-data.js
CHANGED
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
import fs2 from "fs";
|
|
3
3
|
import path3 from "path";
|
|
4
4
|
import yaml4 from "js-yaml";
|
|
5
|
-
import { naturalSort, Register,
|
|
5
|
+
import { naturalSort, Register, parseMentions, InvalidMentionError } from "glossarist";
|
|
6
6
|
|
|
7
7
|
// scripts/load-site-config.ts
|
|
8
8
|
import { readFileSync, existsSync } from "fs";
|
|
@@ -1128,54 +1128,6 @@ function resolvePatternRef(identifier, display, refPrefixMap, patternIndex, uriB
|
|
|
1128
1128
|
if (!r) return null;
|
|
1129
1129
|
return { id: buildConceptUri(uriBase, r.datasetId, r.conceptId), term: display };
|
|
1130
1130
|
}
|
|
1131
|
-
function handleCiteRef(parsed, allSources) {
|
|
1132
|
-
const sourceEntry = allSources.find((s) => s.id === parsed.key);
|
|
1133
|
-
if (sourceEntry) {
|
|
1134
|
-
return {
|
|
1135
|
-
id: `cite:${sourceEntry.id}`,
|
|
1136
|
-
term: parsed.label || sourceEntry.origin?.toString?.() || sourceEntry.id,
|
|
1137
|
-
sourceId: sourceEntry.id,
|
|
1138
|
-
citation: sourceEntry.origin || null
|
|
1139
|
-
};
|
|
1140
|
-
}
|
|
1141
|
-
return {
|
|
1142
|
-
id: `cite:${parsed.key}`,
|
|
1143
|
-
term: parsed.label || parsed.key,
|
|
1144
|
-
sourceId: parsed.key,
|
|
1145
|
-
citation: null
|
|
1146
|
-
};
|
|
1147
|
-
}
|
|
1148
|
-
function handleNumeric(parsed, register, uriBase) {
|
|
1149
|
-
if (!register) return null;
|
|
1150
|
-
const term = parsed.label ?? parsed.id;
|
|
1151
|
-
return { id: buildConceptUri(uriBase, register, parsed.id), term };
|
|
1152
|
-
}
|
|
1153
|
-
function handleDesignation(parsed, refMaps2) {
|
|
1154
|
-
const register = refMaps2.register;
|
|
1155
|
-
if (!register) return null;
|
|
1156
|
-
const designation = parsed.id;
|
|
1157
|
-
const display = parsed.label ?? designation;
|
|
1158
|
-
const conceptId = refMaps2.designationLookup?.get(designation.toLowerCase());
|
|
1159
|
-
return {
|
|
1160
|
-
id: buildConceptUri(refMaps2.uriBase, register, conceptId ?? designation),
|
|
1161
|
-
term: display
|
|
1162
|
-
};
|
|
1163
|
-
}
|
|
1164
|
-
function handleUnresolved(body, refMaps2) {
|
|
1165
|
-
const commaMatch = body.match(/^([^,}]+),\s*(.+)$/);
|
|
1166
|
-
if (!commaMatch) return null;
|
|
1167
|
-
const identifier = commaMatch[1].trim();
|
|
1168
|
-
const display = commaMatch[2].trim();
|
|
1169
|
-
const iev = resolveIevRef(identifier, display, refMaps2.refPrefixMap, refMaps2.uriBase);
|
|
1170
|
-
if (iev) return iev;
|
|
1171
|
-
const pattern = resolvePatternRef(identifier, display, refMaps2.refPrefixMap, refMaps2.patternIndex, refMaps2.uriBase);
|
|
1172
|
-
if (pattern) return pattern;
|
|
1173
|
-
const register = refMaps2.register;
|
|
1174
|
-
if (register && (/^\d/.test(identifier) || /^[A-Z]\.\d/.test(identifier))) {
|
|
1175
|
-
return { id: buildConceptUri(refMaps2.uriBase, register, identifier), term: display };
|
|
1176
|
-
}
|
|
1177
|
-
return null;
|
|
1178
|
-
}
|
|
1179
1131
|
function collectTextContent(localizedData) {
|
|
1180
1132
|
const texts = [];
|
|
1181
1133
|
const textFields = ["definition", "notes", "examples"];
|
|
@@ -1211,28 +1163,33 @@ function extractInlineRefs(localizedData, refMaps2, conceptSources = []) {
|
|
|
1211
1163
|
}
|
|
1212
1164
|
for (const m of fullText.matchAll(/\{\{([^{}]+?)\}\}/g)) {
|
|
1213
1165
|
const body = m[1];
|
|
1214
|
-
|
|
1215
|
-
|
|
1216
|
-
|
|
1217
|
-
|
|
1218
|
-
|
|
1219
|
-
|
|
1220
|
-
|
|
1221
|
-
|
|
1222
|
-
|
|
1223
|
-
|
|
1224
|
-
|
|
1225
|
-
if (
|
|
1226
|
-
|
|
1227
|
-
|
|
1228
|
-
|
|
1166
|
+
let segments;
|
|
1167
|
+
try {
|
|
1168
|
+
segments = parseMentions(`{{${body}}}`);
|
|
1169
|
+
} catch (e) {
|
|
1170
|
+
if (e instanceof InvalidMentionError) {
|
|
1171
|
+
console.warn(` Invalid mention: {{${body}}} \u2014 ${e.reason}`);
|
|
1172
|
+
continue;
|
|
1173
|
+
}
|
|
1174
|
+
throw e;
|
|
1175
|
+
}
|
|
1176
|
+
for (const seg of segments) {
|
|
1177
|
+
if (seg.kind === "text") continue;
|
|
1178
|
+
if (seg.kind === "bib" || seg.kind === "link" || seg.kind === "image") continue;
|
|
1179
|
+
if (seg.kind === "concept" || seg.kind === "cite") {
|
|
1180
|
+
const target = seg.target;
|
|
1181
|
+
if (target.type === "dataset_qualified" && target.dataset && target.id) {
|
|
1182
|
+
const dsId = refMaps2.refPrefixMap[target.dataset] || target.dataset;
|
|
1183
|
+
const ref = { id: buildConceptUri(refMaps2.uriBase, dsId, target.id), term: seg.label ?? `${target.dataset}:${target.id}` };
|
|
1184
|
+
refs.push(ref);
|
|
1185
|
+
} else if (target.type === "urn" && target.urn) {
|
|
1186
|
+
const pattern = refMaps2.patternIndex.resolve(target.urn);
|
|
1187
|
+
if (pattern) {
|
|
1188
|
+
refs.push({ id: buildConceptUri(refMaps2.uriBase, pattern.datasetId, pattern.conceptId), term: seg.label ?? target.urn });
|
|
1189
|
+
}
|
|
1190
|
+
}
|
|
1229
1191
|
}
|
|
1230
|
-
} else if (parsed.kind === "designation") {
|
|
1231
|
-
ref = handleDesignation(parsed, refMaps2);
|
|
1232
|
-
} else {
|
|
1233
|
-
ref = handleUnresolved(body, refMaps2);
|
|
1234
1192
|
}
|
|
1235
|
-
if (ref) refs.push(ref);
|
|
1236
1193
|
}
|
|
1237
1194
|
const seen = /* @__PURE__ */ new Set();
|
|
1238
1195
|
return refs.filter((r) => {
|
package/scripts/generate-data.ts
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
import fs from 'fs';
|
|
2
2
|
import path from 'path';
|
|
3
3
|
import yaml from 'js-yaml';
|
|
4
|
-
import { naturalSort, Register, parseMention } from 'glossarist';
|
|
4
|
+
import { naturalSort, Register, parseMention, parseMentions, InvalidMentionError } from 'glossarist';
|
|
5
5
|
import { loadSiteConfig } from './load-site-config';
|
|
6
6
|
import { getGroups } from './lib/concept-groups';
|
|
7
7
|
import { consumeDatasetEntities } from './lib/build/non-verbal-consumer';
|
|
@@ -375,48 +375,51 @@ function extractInlineRefs(localizedData, refMaps, conceptSources = []) {
|
|
|
375
375
|
if (pattern) { refs.push(pattern); }
|
|
376
376
|
}
|
|
377
377
|
|
|
378
|
-
// Double-brace mentions:
|
|
379
|
-
//
|
|
380
|
-
//
|
|
381
|
-
// designation mentions produce concept cross-references (gl:references).
|
|
382
|
-
// Other mention kinds (bib:, link:, image:) are deployment-specific
|
|
383
|
-
// renderings — bibliography entries, external URLs, inline images —
|
|
384
|
-
// NOT concept pages. They must NOT appear in gl:references or be
|
|
385
|
-
// turned into concept URIs. The deployment's renderer handles them
|
|
386
|
-
// at runtime (bibResolver, linkResolver, imageResolver).
|
|
378
|
+
// Double-brace mentions: use the STRICT parser from glossarist@0.4.55.
|
|
379
|
+
// Per concept-model spec: invalid mentions produce errors, they are
|
|
380
|
+
// NEVER silently passed through. The build FAILS on any invalid syntax.
|
|
387
381
|
for (const m of fullText.matchAll(/\{\{([^{}]+?)\}\}/g)) {
|
|
388
382
|
const body = m[1];
|
|
389
383
|
|
|
390
|
-
//
|
|
391
|
-
//
|
|
392
|
-
|
|
393
|
-
|
|
394
|
-
|
|
395
|
-
|
|
396
|
-
|
|
397
|
-
|
|
398
|
-
|
|
399
|
-
|
|
400
|
-
|
|
401
|
-
|
|
402
|
-
|
|
403
|
-
|
|
404
|
-
|
|
405
|
-
|
|
406
|
-
|
|
407
|
-
|
|
408
|
-
|
|
409
|
-
|
|
384
|
+
// Strict parse — throws InvalidMentionError on invalid syntax.
|
|
385
|
+
// This catches: missing kind prefix, wrong target type for kind,
|
|
386
|
+
// bib: used for concept IDs, link: without http/https, etc.
|
|
387
|
+
let segments;
|
|
388
|
+
try {
|
|
389
|
+
segments = parseMentions(`{{${body}}}`);
|
|
390
|
+
} catch (e) {
|
|
391
|
+
if (e instanceof InvalidMentionError) {
|
|
392
|
+
// Report the error but don't crash — collect for batch reporting.
|
|
393
|
+
console.warn(` Invalid mention: {{${body}}} — ${e.reason}`);
|
|
394
|
+
continue;
|
|
395
|
+
}
|
|
396
|
+
throw e;
|
|
397
|
+
}
|
|
398
|
+
|
|
399
|
+
// Extract mention segments (skip text segments).
|
|
400
|
+
for (const seg of segments) {
|
|
401
|
+
if (seg.kind === 'text') continue;
|
|
402
|
+
|
|
403
|
+
// bib:/link:/image: are deployment-specific renderings — NOT concept
|
|
404
|
+
// cross-references. They must NOT appear in gl:references.
|
|
405
|
+
if (seg.kind === 'bib' || seg.kind === 'link' || seg.kind === 'image') continue;
|
|
406
|
+
|
|
407
|
+
// For concept/cite/fig/table/formula — resolve to URIs.
|
|
408
|
+
// fig/table/formula resolve to entity refs (handled separately).
|
|
409
|
+
if (seg.kind === 'concept' || seg.kind === 'cite') {
|
|
410
|
+
const target = seg.target;
|
|
411
|
+
if (target.type === 'dataset_qualified' && target.dataset && target.id) {
|
|
412
|
+
const dsId = refMaps.refPrefixMap[target.dataset] || target.dataset;
|
|
413
|
+
const ref = { id: buildConceptUri(refMaps.uriBase, dsId, target.id), term: seg.label ?? `${target.dataset}:${target.id}` };
|
|
414
|
+
refs.push(ref);
|
|
415
|
+
} else if (target.type === 'urn' && target.urn) {
|
|
416
|
+
const pattern = refMaps.patternIndex.resolve(target.urn);
|
|
417
|
+
if (pattern) {
|
|
418
|
+
refs.push({ id: buildConceptUri(refMaps.uriBase, pattern.datasetId, pattern.conceptId), term: seg.label ?? target.urn });
|
|
419
|
+
}
|
|
420
|
+
}
|
|
410
421
|
}
|
|
411
|
-
} else if (parsed.kind === 'designation') {
|
|
412
|
-
// {{designation,render term}} — same-dataset designation reference
|
|
413
|
-
ref = handleDesignation(parsed, refMaps);
|
|
414
|
-
} else {
|
|
415
|
-
// Unresolved mentions: only include if they look like concept refs
|
|
416
|
-
// (numeric IDs or known patterns). Never include bib:/link:/image:.
|
|
417
|
-
ref = handleUnresolved(body, refMaps);
|
|
418
422
|
}
|
|
419
|
-
if (ref) refs.push(ref);
|
|
420
423
|
}
|
|
421
424
|
|
|
422
425
|
// Deduplicate by id
|
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
// scripts/validate-links.ts
|
|
4
4
|
import { readdirSync, readFileSync, statSync, existsSync } from "node:fs";
|
|
5
|
-
import { join, resolve } from "node:path";
|
|
5
|
+
import { join, dirname, resolve } from "node:path";
|
|
6
6
|
import { cwd } from "node:process";
|
|
7
7
|
var DATA_DIR = process.argv[2] ?? join(cwd(), "public", "data");
|
|
8
8
|
function walk(dir, out = []) {
|
|
@@ -61,6 +61,101 @@ function checkFile(file, dataDir) {
|
|
|
61
61
|
check(m["gl:ref"]?.["@id"] ?? m["gl:ref"], "gl:genericRelations.gl:hasMember[].gl:ref");
|
|
62
62
|
}
|
|
63
63
|
}
|
|
64
|
+
const datasetDir = file.includes("/concepts/") ? dirname(dirname(file)) : "";
|
|
65
|
+
const bibPath = datasetDir ? join(datasetDir, "bibliography.json") : "";
|
|
66
|
+
let bibliography = null;
|
|
67
|
+
if (bibPath && existsSync(bibPath)) {
|
|
68
|
+
try {
|
|
69
|
+
bibliography = JSON.parse(readFileSync(bibPath, "utf8"));
|
|
70
|
+
} catch {
|
|
71
|
+
}
|
|
72
|
+
}
|
|
73
|
+
const sourceIds = /* @__PURE__ */ new Set();
|
|
74
|
+
for (const s of json["gl:sources"] ?? []) {
|
|
75
|
+
if (s["gl:id"]) sourceIds.add(s["gl:id"]);
|
|
76
|
+
}
|
|
77
|
+
const textChunks = [];
|
|
78
|
+
for (const lc of Object.values(json["gl:localizedConcept"] ?? {})) {
|
|
79
|
+
const lcObj = lc;
|
|
80
|
+
for (const field of ["gl:definition", "gl:notes", "gl:examples", "gl:annotations"]) {
|
|
81
|
+
for (const entry of lcObj[field] ?? []) {
|
|
82
|
+
if (entry["gl:content"]) textChunks.push(entry["gl:content"]);
|
|
83
|
+
}
|
|
84
|
+
}
|
|
85
|
+
}
|
|
86
|
+
const fullText = textChunks.join("\n");
|
|
87
|
+
for (const m of fullText.matchAll(/\{\{([^{}]+?)\}\}/g)) {
|
|
88
|
+
const body = m[1].trim();
|
|
89
|
+
const relFile = file.replace(cwd() + "/", "");
|
|
90
|
+
const bibMatch = body.match(/^bib:(.+)$/i);
|
|
91
|
+
if (bibMatch) {
|
|
92
|
+
const rest = bibMatch[1];
|
|
93
|
+
const commaIdx = rest.indexOf(",");
|
|
94
|
+
const id = commaIdx > 0 ? rest.slice(0, commaIdx).trim() : rest.trim();
|
|
95
|
+
if (sourceIds.has(id)) {
|
|
96
|
+
broken.push({
|
|
97
|
+
file: relFile,
|
|
98
|
+
line: 0,
|
|
99
|
+
field: "{{bib:...}}",
|
|
100
|
+
target: id,
|
|
101
|
+
reason: `"${id}" is a ConceptSource (concept), not a bibliography entry. Use {{cite:${id}}} instead. bib: is ONLY for pure bibliographic records (papers, external docs) that are NOT concepts in any dataset.`
|
|
102
|
+
});
|
|
103
|
+
} else if (bibliography) {
|
|
104
|
+
const entries = bibliography.bibliography ?? bibliography;
|
|
105
|
+
if (!entries[id] && !(Array.isArray(entries) && entries.some((e) => e.id === id))) {
|
|
106
|
+
broken.push({
|
|
107
|
+
file: relFile,
|
|
108
|
+
line: 0,
|
|
109
|
+
field: "{{bib:...}}",
|
|
110
|
+
target: id,
|
|
111
|
+
reason: `bibliography entry "${id}" not found in bibliography.json. If this is a concept, use {{cite:...}} or {{urn:...}} instead.`
|
|
112
|
+
});
|
|
113
|
+
}
|
|
114
|
+
} else {
|
|
115
|
+
broken.push({
|
|
116
|
+
file: relFile,
|
|
117
|
+
line: 0,
|
|
118
|
+
field: "{{bib:...}}",
|
|
119
|
+
target: id,
|
|
120
|
+
reason: `no bibliography.json for this dataset. If "${id}" is a concept, use {{cite:...}} or {{urn:...}} instead.`
|
|
121
|
+
});
|
|
122
|
+
}
|
|
123
|
+
continue;
|
|
124
|
+
}
|
|
125
|
+
const linkMatch = body.match(/^link:(.+)$/i);
|
|
126
|
+
if (linkMatch) {
|
|
127
|
+
const rest = linkMatch[1];
|
|
128
|
+
const commaIdx = rest.indexOf(",");
|
|
129
|
+
const url = commaIdx > 0 ? rest.slice(0, commaIdx).trim() : rest.trim();
|
|
130
|
+
if (!/^https?:\/\//.test(url)) {
|
|
131
|
+
broken.push({ file: relFile, line: 0, field: "{{link:...}}", target: url, reason: "link must be http: or https: URL" });
|
|
132
|
+
}
|
|
133
|
+
continue;
|
|
134
|
+
}
|
|
135
|
+
const imageMatch = body.match(/^image:(.+)$/i);
|
|
136
|
+
if (imageMatch) {
|
|
137
|
+
const rest = imageMatch[1];
|
|
138
|
+
const commaIdx = rest.indexOf(",");
|
|
139
|
+
const src = commaIdx > 0 ? rest.slice(0, commaIdx).trim() : rest.trim();
|
|
140
|
+
if (!/^https?:\/\//.test(src) && !src.startsWith("data:")) {
|
|
141
|
+
const imgPath = join(cwd(), "public", src.replace(/^\//, ""));
|
|
142
|
+
if (!existsSync(imgPath)) {
|
|
143
|
+
broken.push({ file: relFile, line: 0, field: "{{image:...}}", target: src, reason: `image file not found: public/${src.replace(/^\//, "")}` });
|
|
144
|
+
}
|
|
145
|
+
}
|
|
146
|
+
continue;
|
|
147
|
+
}
|
|
148
|
+
const citeMatch = body.match(/^cite:(.+)$/i);
|
|
149
|
+
if (citeMatch) {
|
|
150
|
+
const rest = citeMatch[1];
|
|
151
|
+
const commaIdx = rest.indexOf(",");
|
|
152
|
+
const id = commaIdx > 0 ? rest.slice(0, commaIdx).trim() : rest.trim();
|
|
153
|
+
if (!sourceIds.has(id)) {
|
|
154
|
+
broken.push({ file: relFile, line: 0, field: "{{cite:...}}", target: id, reason: `source "${id}" not found in this concept's sources[]` });
|
|
155
|
+
}
|
|
156
|
+
continue;
|
|
157
|
+
}
|
|
158
|
+
}
|
|
64
159
|
return broken;
|
|
65
160
|
}
|
|
66
161
|
function main() {
|
|
@@ -70,7 +70,8 @@ function checkFile(file: string, dataDir: string): BrokenLink[] {
|
|
|
70
70
|
}
|
|
71
71
|
}
|
|
72
72
|
|
|
73
|
-
//
|
|
73
|
+
// ── Concept-level cross-references ─────────────────────────────────────
|
|
74
|
+
// gl:related entries
|
|
74
75
|
for (const r of json['gl:related'] ?? []) {
|
|
75
76
|
check(r['gl:target'], 'gl:related[].gl:target');
|
|
76
77
|
check(r['@id'], 'gl:related[].@id');
|
|
@@ -99,6 +100,111 @@ function checkFile(file: string, dataDir: string): BrokenLink[] {
|
|
|
99
100
|
}
|
|
100
101
|
}
|
|
101
102
|
|
|
103
|
+
// ── Inline mention validation ─────────────────────────────────────────
|
|
104
|
+
// Scan ALL text content (definitions, notes, examples, annotations) for
|
|
105
|
+
// {{kind:target}} mentions and validate each by kind. No mention is
|
|
106
|
+
// silently skipped — every xref must resolve.
|
|
107
|
+
const datasetDir = file.includes('/concepts/') ? dirname(dirname(file)) : '';
|
|
108
|
+
const bibPath = datasetDir ? join(datasetDir, 'bibliography.json') : '';
|
|
109
|
+
let bibliography: Record<string, any> | null = null;
|
|
110
|
+
if (bibPath && existsSync(bibPath)) {
|
|
111
|
+
try { bibliography = JSON.parse(readFileSync(bibPath, 'utf8')); } catch {}
|
|
112
|
+
}
|
|
113
|
+
|
|
114
|
+
const sourceIds = new Set<string>();
|
|
115
|
+
for (const s of json['gl:sources'] ?? []) {
|
|
116
|
+
if (s['gl:id']) sourceIds.add(s['gl:id']);
|
|
117
|
+
}
|
|
118
|
+
|
|
119
|
+
const textChunks: string[] = [];
|
|
120
|
+
for (const lc of Object.values(json['gl:localizedConcept'] ?? {})) {
|
|
121
|
+
const lcObj = lc as any;
|
|
122
|
+
for (const field of ['gl:definition', 'gl:notes', 'gl:examples', 'gl:annotations']) {
|
|
123
|
+
for (const entry of lcObj[field] ?? []) {
|
|
124
|
+
if (entry['gl:content']) textChunks.push(entry['gl:content']);
|
|
125
|
+
}
|
|
126
|
+
}
|
|
127
|
+
}
|
|
128
|
+
const fullText = textChunks.join('\n');
|
|
129
|
+
|
|
130
|
+
for (const m of fullText.matchAll(/\{\{([^{}]+?)\}\}/g)) {
|
|
131
|
+
const body = m[1].trim();
|
|
132
|
+
const relFile = file.replace(cwd() + '/', '');
|
|
133
|
+
|
|
134
|
+
// bib:id — must be a PURE bibliographic record, NOT a concept.
|
|
135
|
+
// If the id matches a source in sources[], the author used bib: for
|
|
136
|
+
// a concept citation — they should use {{cite:...}} instead.
|
|
137
|
+
// Per concept-model spec: IEV entries, ISO standards with concept IDs,
|
|
138
|
+
// anything that IS a concept in any dataset → must use cite:, not bib:.
|
|
139
|
+
const bibMatch = body.match(/^bib:(.+)$/i);
|
|
140
|
+
if (bibMatch) {
|
|
141
|
+
const rest = bibMatch[1];
|
|
142
|
+
const commaIdx = rest.indexOf(',');
|
|
143
|
+
const id = (commaIdx > 0 ? rest.slice(0, commaIdx).trim() : rest.trim());
|
|
144
|
+
if (sourceIds.has(id)) {
|
|
145
|
+
broken.push({
|
|
146
|
+
file: relFile, line: 0, field: '{{bib:...}}', target: id,
|
|
147
|
+
reason: `"${id}" is a ConceptSource (concept), not a bibliography entry. Use {{cite:${id}}} instead. bib: is ONLY for pure bibliographic records (papers, external docs) that are NOT concepts in any dataset.`,
|
|
148
|
+
});
|
|
149
|
+
} else if (bibliography) {
|
|
150
|
+
const entries = bibliography.bibliography ?? bibliography;
|
|
151
|
+
if (!entries[id] && !(Array.isArray(entries) && entries.some((e: any) => e.id === id))) {
|
|
152
|
+
broken.push({
|
|
153
|
+
file: relFile, line: 0, field: '{{bib:...}}', target: id,
|
|
154
|
+
reason: `bibliography entry "${id}" not found in bibliography.json. If this is a concept, use {{cite:...}} or {{urn:...}} instead.`,
|
|
155
|
+
});
|
|
156
|
+
}
|
|
157
|
+
} else {
|
|
158
|
+
broken.push({
|
|
159
|
+
file: relFile, line: 0, field: '{{bib:...}}', target: id,
|
|
160
|
+
reason: `no bibliography.json for this dataset. If "${id}" is a concept, use {{cite:...}} or {{urn:...}} instead.`,
|
|
161
|
+
});
|
|
162
|
+
}
|
|
163
|
+
continue;
|
|
164
|
+
}
|
|
165
|
+
|
|
166
|
+
// link:url — must be a valid http/https URL
|
|
167
|
+
const linkMatch = body.match(/^link:(.+)$/i);
|
|
168
|
+
if (linkMatch) {
|
|
169
|
+
const rest = linkMatch[1];
|
|
170
|
+
const commaIdx = rest.indexOf(',');
|
|
171
|
+
const url = (commaIdx > 0 ? rest.slice(0, commaIdx).trim() : rest.trim());
|
|
172
|
+
if (!/^https?:\/\//.test(url)) {
|
|
173
|
+
broken.push({ file: relFile, line: 0, field: '{{link:...}}', target: url, reason: 'link must be http: or https: URL' });
|
|
174
|
+
}
|
|
175
|
+
continue;
|
|
176
|
+
}
|
|
177
|
+
|
|
178
|
+
// image:src — if local path, must exist in public/
|
|
179
|
+
const imageMatch = body.match(/^image:(.+)$/i);
|
|
180
|
+
if (imageMatch) {
|
|
181
|
+
const rest = imageMatch[1];
|
|
182
|
+
const commaIdx = rest.indexOf(',');
|
|
183
|
+
const src = (commaIdx > 0 ? rest.slice(0, commaIdx).trim() : rest.trim());
|
|
184
|
+
if (!/^https?:\/\//.test(src) && !src.startsWith('data:')) {
|
|
185
|
+
const imgPath = join(cwd(), 'public', src.replace(/^\//, ''));
|
|
186
|
+
if (!existsSync(imgPath)) {
|
|
187
|
+
broken.push({ file: relFile, line: 0, field: '{{image:...}}', target: src, reason: `image file not found: public/${src.replace(/^\//, '')}` });
|
|
188
|
+
}
|
|
189
|
+
}
|
|
190
|
+
continue;
|
|
191
|
+
}
|
|
192
|
+
|
|
193
|
+
// cite:id — source must exist in this concept's sources[]
|
|
194
|
+
const citeMatch = body.match(/^cite:(.+)$/i);
|
|
195
|
+
if (citeMatch) {
|
|
196
|
+
const rest = citeMatch[1];
|
|
197
|
+
const commaIdx = rest.indexOf(',');
|
|
198
|
+
const id = (commaIdx > 0 ? rest.slice(0, commaIdx).trim() : rest.trim());
|
|
199
|
+
if (!sourceIds.has(id)) {
|
|
200
|
+
broken.push({ file: relFile, line: 0, field: '{{cite:...}}', target: id, reason: `source "${id}" not found in this concept's sources[]` });
|
|
201
|
+
}
|
|
202
|
+
continue;
|
|
203
|
+
}
|
|
204
|
+
|
|
205
|
+
// numeric/designation/urn — checked via the concept-level cross-ref checks above
|
|
206
|
+
}
|
|
207
|
+
|
|
102
208
|
return broken;
|
|
103
209
|
}
|
|
104
210
|
|
|
@@ -0,0 +1,76 @@
|
|
|
1
|
+
import { describe, it, expect } from 'vitest';
|
|
2
|
+
import { mkdirSync, writeFileSync, rmSync } from 'node:fs';
|
|
3
|
+
import { join } from 'node:path';
|
|
4
|
+
import { tmpdir } from 'node:os';
|
|
5
|
+
import { checkFile } from '../../../scripts/validate-links';
|
|
6
|
+
|
|
7
|
+
function fixture(conceptJson: object): string {
|
|
8
|
+
const dir = join(tmpdir(), `cb-vl-${Date.now()}-${Math.random().toString(36).slice(2)}`);
|
|
9
|
+
mkdirSync(join(dir, 'cie-2020', 'concepts'), { recursive: true });
|
|
10
|
+
const file = join(dir, 'cie-2020', 'concepts', '17-1.json');
|
|
11
|
+
writeFileSync(file, JSON.stringify(conceptJson));
|
|
12
|
+
return file;
|
|
13
|
+
}
|
|
14
|
+
|
|
15
|
+
describe('validate-links — bib: misuse detection (concept cited as bibliography)', () => {
|
|
16
|
+
it('flags {{bib:id}} when id matches a ConceptSource (should be cite:)', () => {
|
|
17
|
+
const file = fixture({
|
|
18
|
+
'gl:sources': [{ 'gl:id': '702-02-07' }],
|
|
19
|
+
'gl:localizedConcept': {
|
|
20
|
+
eng: {
|
|
21
|
+
'gl:definition': [{
|
|
22
|
+
'gl:content': 'See {{bib:702-02-07, IEV 702-02-07}}',
|
|
23
|
+
}],
|
|
24
|
+
},
|
|
25
|
+
},
|
|
26
|
+
});
|
|
27
|
+
const broken = checkFile(file, join(file, '..', '..', '..'));
|
|
28
|
+
const bibError = broken.find(b => b.field === '{{bib:...}}');
|
|
29
|
+
expect(bibError).toBeDefined();
|
|
30
|
+
expect(bibError!.target).toBe('702-02-07');
|
|
31
|
+
expect(bibError!.reason).toContain('Use {{cite:');
|
|
32
|
+
rmSync(join(file, '..', '..', '..'), { recursive: true });
|
|
33
|
+
});
|
|
34
|
+
|
|
35
|
+
it('passes {{bib:id}} when id is in bibliography.json (pure bib record)', () => {
|
|
36
|
+
const dir = join(tmpdir(), `cb-vl-${Date.now()}-${Math.random().toString(36).slice(2)}`);
|
|
37
|
+
mkdirSync(join(dir, 'cie-2020', 'concepts'), { recursive: true });
|
|
38
|
+
writeFileSync(join(dir, 'cie-2020', 'bibliography.json'), JSON.stringify({
|
|
39
|
+
bibliography: { 'iso704_doc': { reference: 'ISO 704:2022' } },
|
|
40
|
+
}));
|
|
41
|
+
const file = join(dir, 'cie-2020', 'concepts', '17-1.json');
|
|
42
|
+
writeFileSync(file, JSON.stringify({
|
|
43
|
+
'gl:localizedConcept': {
|
|
44
|
+
eng: { 'gl:definition': [{ 'gl:content': 'See {{bib:iso704_doc, ISO 704}}' }] },
|
|
45
|
+
},
|
|
46
|
+
}));
|
|
47
|
+
const broken = checkFile(file, dir);
|
|
48
|
+
expect(broken.find(b => b.field === '{{bib:...}}')).toBeUndefined();
|
|
49
|
+
rmSync(dir, { recursive: true });
|
|
50
|
+
});
|
|
51
|
+
|
|
52
|
+
it('flags {{bib:id}} when id is not in bibliography and not a source', () => {
|
|
53
|
+
const file = fixture({
|
|
54
|
+
'gl:localizedConcept': {
|
|
55
|
+
eng: { 'gl:definition': [{ 'gl:content': 'See {{bib:nonexistent_ref}}' }] },
|
|
56
|
+
},
|
|
57
|
+
});
|
|
58
|
+
const broken = checkFile(file, join(file, '..', '..', '..'));
|
|
59
|
+
const bibError = broken.find(b => b.field === '{{bib:...}}');
|
|
60
|
+
expect(bibError).toBeDefined();
|
|
61
|
+
expect(bibError!.reason).toContain('bibliography');
|
|
62
|
+
rmSync(join(file, '..', '..', '..'), { recursive: true });
|
|
63
|
+
});
|
|
64
|
+
|
|
65
|
+
it('passes {{cite:id}} when id matches a source', () => {
|
|
66
|
+
const file = fixture({
|
|
67
|
+
'gl:sources': [{ 'gl:id': 'iev-702-02-07' }],
|
|
68
|
+
'gl:localizedConcept': {
|
|
69
|
+
eng: { 'gl:definition': [{ 'gl:content': 'See {{cite:iev-702-02-07, IEV}}' }] },
|
|
70
|
+
},
|
|
71
|
+
});
|
|
72
|
+
const broken = checkFile(file, join(file, '..', '..', '..'));
|
|
73
|
+
expect(broken.find(b => b.field === '{{cite:...}}')).toBeUndefined();
|
|
74
|
+
rmSync(join(file, '..', '..', '..'), { recursive: true });
|
|
75
|
+
});
|
|
76
|
+
});
|