Press n or j to go to the next uncovered block, b, p or k for the previous block.
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 | 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 8x 56x 90x 4x 4x 4x 86x 86x 86x 86x 425x 425x 425x 212x 212x 425x 8x 211x 211x 3567x 3567x 3567x 178x 110x 110x 110x 2140x 2140x 2140x 110x 630x 2306x 2306x 2306x 56x 56x 56x 56x 8x 8x 34x 34x 34x 45x 45x 45x 44x 44x 30x 34x 85x 85x 85x 77x 77x 77x 77x 85x 3x 3x 2x 2x 2x 2x 2x 2x 2x 2x 2x 2x 26x 26x 26x 8x 4x 4x 18x 4x 26x 2x 79x 237x 79x 79x 79x 79x 79x 79x 79x 79x 79x 79x 79x 79x 79x 87x 83x 83x 4x 8x 34x 34x 34x 34x 34x 34x 34x 34x 170x 170x 170x 12x 26x 34x 34x 18x 166x 18x 16x 16x 16x 16x 32x 32x 32x 18x 18x 17x 17x 17x 17x 17x 18x 16x 9x 9x 1x 1x 8x 8x 3x 17x 17x 17x 17x 17x 353x 185x 185x 168x 336x 177x 316x 45x 45x 45x 45x 45x 45x 45x 45x 45x 45x 45x 45x 45x 45x 45x 360x 70x 290x 45x 45x 45x 45x 45x 45x 55x 55x 55x 55x 55x 55x 55x 55x 55x 55x 55x 55x 55x 15x 15x 55x 10x 10x 10x 55x 9x 9x 55x 16x 16x 55x 42x 42x 42x 42x 34x 31x 31x 25x 24x 8x 7x 7x 43x 43x 43x 43x 35x 35x 43x 43x 43x 43x 35x 8x 3x 35x 28x 35x 28x 35x 31x 35x 29x 35x 30x 35x 28x 35x 31x 35x 28x 43x 32x 11x 1x 43x 36x 7x 43x 39x 4x 43x 32x 11x 1x 43x 32x 11x 8x 34x 34x 34x 34x 34x 34x 34x 34x 34x 34x 34x 34x 34x 34x 34x 34x | // SPDX-FileCopyrightText: 2024-2026 Hack23 AB
// SPDX-License-Identifier: Apache-2.0
/**
* @module Utils/ArticleQualityScorer
* @description Comprehensive quality assessment engine for generated EU Parliament Monitor articles.
*
* Analyses HTML article content across four dimensions:
* - **Analysis depth** — political context, coalition dynamics, historical evidence, scenarios
* - **Stakeholder coverage** — breadth of perspectives from MEPs, Commission, civil society, etc.
* - **Visualization quality** — SWOT, dashboard metrics, mindmap branches, deep-analysis evidence
* - **Content integrity** — word count, evidence references, section structure
*
* Produces an {@link ArticleQualityReport} with a 0–100 overall score, letter grade (A–F),
* pass/fail quality gate, and actionable recommendations.
*/
import type {
ArticleQualityReport,
AnalysisDepthScore,
StakeholderCoverage,
VisualizationQuality,
ArticleGrade,
} from '../types/quality.js';
import { stripScriptBlocks } from './html-sanitize.js';
// ─── Scoring constants ────────────────────────────────────────────────────────
/** Weight applied to analysis depth score in overall calculation */
const WEIGHT_ANALYSIS_DEPTH = 0.25;
/** Weight applied to stakeholder balance score in overall calculation */
const WEIGHT_STAKEHOLDER = 0.2;
/** Weight applied to visualization quality score in overall calculation */
const WEIGHT_VISUALIZATION = 0.25;
/** Weight applied to word-count score in overall calculation */
const WEIGHT_WORD_COUNT = 0.15;
/** Weight applied to evidence-reference score in overall calculation */
const WEIGHT_EVIDENCE = 0.15;
/** Minimum word count to score 0 on the word-count dimension */
const WORD_COUNT_MIN = 0;
/** Word count that earns the maximum word-count dimension score */
const WORD_COUNT_MAX = 1500;
/** Evidence-reference count that earns the maximum evidence dimension score */
const EVIDENCE_MAX = 10;
/** Overall score threshold for passing the quality gate (Grade C floor) */
const QUALITY_GATE_THRESHOLD = 40;
/** Grade boundary — score >= this earns an A */
const GRADE_A_MIN = 80;
/** Grade boundary — score >= this earns a B */
const GRADE_B_MIN = 65;
/** Grade boundary — score >= this earns a C */
const GRADE_C_MIN = 40;
/** Grade boundary — score >= this earns a D */
const GRADE_D_MIN = 25;
// ─── Analysis-depth keyword sets ─────────────────────────────────────────────
/** Keywords indicating political context discussion */
const POLITICAL_CONTEXT_KEYWORDS: ReadonlyArray<string> = [
'political',
'coalition',
'majority',
'opposition',
'parliament',
];
/** Keywords indicating coalition-dynamics analysis */
const COALITION_DYNAMICS_KEYWORDS: ReadonlyArray<string> = [
'coalition',
'alliance',
'EPP',
'S&D',
'Renew',
'Greens',
];
/** Keywords indicating historical context */
const HISTORICAL_CONTEXT_KEYWORDS: ReadonlyArray<string> = [
'historically',
'since 2019',
'previous term',
'compared to',
];
/** Keywords indicating evidence-based reasoning */
const EVIDENCE_BASED_KEYWORDS: ReadonlyArray<string> = [
'according to',
'data shows',
'evidence suggests',
'figures',
];
/** Keywords indicating scenario planning or projections */
const SCENARIO_PLANNING_KEYWORDS: ReadonlyArray<string> = [
'if ',
'could',
'scenario',
'projection',
'forecast',
];
/** Keywords indicating stated confidence levels */
const CONFIDENCE_LEVEL_KEYWORDS: ReadonlyArray<string> = [
'likely',
'probably',
'uncertain',
'confidence',
];
// ─── Stakeholder detection sets ───────────────────────────────────────────────
/** All known stakeholder categories and their keyword signals */
const STAKEHOLDER_KEYWORDS: ReadonlyArray<{ name: string; keywords: ReadonlyArray<string> }> = [
{ name: 'MEPs/Parliament', keywords: ['MEP', 'parliament', 'parliamentarian', 'deputy'] },
{
name: 'Commission',
keywords: ['Commission', 'commissioner', 'European Commission'],
},
{ name: 'Council', keywords: ['Council', 'presidency', 'member states'] },
{
name: 'member states/governments',
keywords: ['government', 'national', 'member state', 'minister'],
},
{
name: 'civil society/NGOs',
keywords: ['civil society', 'NGO', 'non-governmental', 'advocacy'],
},
{
name: 'industry/business',
keywords: ['industry', 'business', 'corporate', 'sector', 'company'],
},
{ name: 'citizens', keywords: ['citizen', 'public', 'voter', 'constituent'] },
{ name: 'media', keywords: ['media', 'press', 'journalist', 'outlet'] },
];
// ─── Placeholder / generic-phrase patterns ────────────────────────────────────
/** Patterns indicating vague or un-replaced generic phrases */
const GENERIC_PHRASE_PATTERNS: ReadonlyArray<RegExp> = [
/various committees/iu,
/several MEPs/iu,
/multiple documents/iu,
/some countries/iu,
];
// ─── EP document-reference pattern ───────────────────────────────────────────
/**
* Patterns matching known EP document reference formats.
* Uses separate patterns to avoid alternation complexity flagged by security/detect-unsafe-regex.
* Covers: TA-10-2026-0123, PE-123, PE-123.456, A9-0123, B9-0123, C9-0123, P9_TA(2024)0001
* Excludes broad matches like EU-27 or EEA-32.
*/
const EP_DOC_PATTERNS: ReadonlyArray<RegExp> = [
/\bTA-\d+-\d+-\d+\b/gu, // TA-10-2026-0001 (TA prefix + three numeric segments)
/\bPE-\d+\.\d+\b/gu, // PE-123.456 (dotted PE reference)
/\bPE-\d+(?!\.\d)\b/gu, // PE-123 (simple PE reference, excludes dotted)
/\b[A-C]\d-\d+\b/gu, // A9-0123, B9-0002, C9-0003 (variable-length digits)
/\bP\d_TA\(\d{4}\)\d+\b/gu, // P9_TA(2024)0001
];
/** CSS class selector for deep-analysis sections (extracted to avoid duplication) */
const CLASS_DEEP_ANALYSIS = 'class="deep-analysis"';
/** Pattern to extract a leading ISO date (YYYY-MM-DD) from an article identifier */
const ARTICLE_DATE_PATTERN = /^(\d{4}-\d{2}-\d{2})/u;
// ─── HTML entity map ──────────────────────────────────────────────────────────
/** Common HTML entities to decode when extracting plain text */
const HTML_ENTITY_MAP: Readonly<Record<string, string>> = {
'&': '&',
'<': '<',
'>': '>',
'"': '"',
''': "'",
''': "'",
' ': ' ',
};
/** Pattern matching named and numeric HTML entities */
const HTML_ENTITY_PATTERN = /&(?:#(\d+)|#x([0-9a-fA-F]+)|([a-zA-Z]+));/gu;
// ─── Helpers ──────────────────────────────────────────────────────────────────
/**
* Decode HTML entities in a string.
* Handles named entities (&, <, >, ", ', ', )
* and numeric references ({, {).
*
* @param text - Text possibly containing HTML entities
* @returns Text with entities replaced by their character equivalents
*/
function decodeHtmlEntities(text: string): string {
return text.replace(HTML_ENTITY_PATTERN, (match, decimal, hex, named) => {
if (decimal !== undefined) {
const cp = parseInt(decimal, 10);
try {
return String.fromCodePoint(cp);
} catch {
return match;
}
}
Iif (hex !== undefined) {
const cp = parseInt(hex, 16);
try {
return String.fromCodePoint(cp);
} catch {
return match;
}
}
Eif (named !== undefined) {
const key = `&${named};`;
return HTML_ENTITY_MAP[key] ?? match;
}
return match;
});
}
/**
* Count non-overlapping occurrences of a CSS class or id string in HTML.
*
* @param html - HTML string to search
* @param selector - CSS class or id token to count (e.g. `class="metric"`)
* @returns Number of occurrences found
*/
function countOccurrences(html: string, selector: string): number {
let count = 0;
let index = html.indexOf(selector);
while (index !== -1) {
count++;
index = html.indexOf(selector, index + selector.length);
}
return count;
}
/** Pattern matching all class attribute values in HTML */
const CLASS_ATTR_PATTERN = /class="([^"]*)"/gu;
/**
* Check whether any `class="…"` attribute in the HTML contains the given token
* as an exact CSS class (whitespace-delimited). Unlike `\b` word-boundary matching,
* this prevents false positives from hyphenated classes (e.g. `dashboard-grid` does
* not match `dashboard`).
*
* @param html - HTML string to scan
* @param token - Exact CSS class name to detect
* @returns true if an exact class token match is found
*/
function hasExactClassToken(html: string, token: string): boolean {
CLASS_ATTR_PATTERN.lastIndex = 0;
let match: RegExpExecArray | null;
while ((match = CLASS_ATTR_PATTERN.exec(html)) !== null) {
const value = match[1] ?? '';
const classes = value.split(/\s+/);
if (classes.includes(token)) return true;
}
return false;
}
/**
* Count the number of elements whose `class` attribute contains the given token
* as an exact CSS class (whitespace-delimited). Unlike simple substring counting,
* this correctly counts multi-class attributes like `class="metric-card pipeline-on-track"`.
*
* @param html - HTML string to scan
* @param token - Exact CSS class name to count
* @returns Number of elements with the given class token
*/
function countExactClassToken(html: string, token: string): number {
CLASS_ATTR_PATTERN.lastIndex = 0;
let count = 0;
let match: RegExpExecArray | null;
while ((match = CLASS_ATTR_PATTERN.exec(html)) !== null) {
const value = match[1] ?? '';
const classes = value.split(/\s+/);
if (classes.includes(token)) count++;
}
return count;
}
/**
* Check whether at least one keyword from a list is present in a text string.
*
* Uses a leading word-boundary anchor (`\b`) so that keywords like "national"
* do not false-match inside longer words like "international", while still
* matching inflected forms such as "citizens" for the keyword "citizen".
*
* @param text - Text to search (comparison is case-insensitive)
* @param keywords - Keywords to look for
* @returns true if any keyword is found
*/
function containsAnyKeyword(text: string, keywords: ReadonlyArray<string>): boolean {
return keywords.some((kw) => {
const escaped = kw.replace(/[.*+?^${}()|[\]\\]/gu, '\\$&');
const pattern = new RegExp(`\\b${escaped}`, 'iu');
return pattern.test(text);
});
}
// stripScriptBlocks is imported from html-sanitize.ts
/**
* Extract the plain text content from the `<main>` element of an HTML string.
* Falls back to the full document when no `<main>` is found.
* Decodes HTML entities so keyword detection works on real article HTML.
*
* @param html - Raw HTML string
* @returns Plain text stripped of tags, scripts, and HTML entities
*/
function extractPlainText(html: string): string {
const mainMatch = /<main[^>]*>([\s\S]*?)<\/main>/u.exec(html);
const source = mainMatch?.[1] ?? html;
const stripped = stripScriptBlocks(source)
.replace(/<[^>]+>/gu, ' ')
.replace(/\s+/gu, ' ')
.trim();
return decodeHtmlEntities(stripped);
}
/**
* Tokens that identify a `<section>` element as analysis content.
* Non-analysis sections (e.g. `article-sources`, `sitemap-section`) are excluded.
*/
const ANALYSIS_SECTION_TOKENS: ReadonlyArray<string> = [
'analysis',
'analysis-section',
'deep-analysis',
'swot-analysis',
'dashboard',
'mindmap-section',
'sankey-section',
];
/** Pattern to extract the class attribute value from a single HTML tag */
const CLASS_VALUE_PATTERN = /class="([^"]*)"/iu;
/**
* Count structural analysis sections in HTML.
* Only counts `<section>` elements whose class attribute contains a known
* analysis-related token, preventing inflation from non-analysis sections
* like sources or footer wrappers.
*
* @param html - Raw HTML string
* @returns Number of analysis-content sections found
*/
function countAnalysisSections(html: string): number {
const SECTION_TAG = /<section\b[^>]*>/giu;
let count = 0;
let m: RegExpExecArray | null;
while ((m = SECTION_TAG.exec(html)) !== null) {
const tag = m[0];
const cv = CLASS_VALUE_PATTERN.exec(tag);
if (cv?.[1]) {
const tokens = cv[1].split(/\s+/).filter(Boolean);
if (tokens.some((t) => ANALYSIS_SECTION_TOKENS.includes(t))) {
count++;
}
}
}
return count;
}
/**
* Count `<li>` elements inside containers matching the given class attribute.
* Used to count evidence items in `<ul class="perspective-evidence"><li>…</li></ul>`
* and `<ul class="evidence-refs"><li lang="…">…</li></ul>` structures produced by
* the deep-analysis and stakeholder perspective generators.
*
* Locates each container by its class attribute, determines the enclosing element
* tag name, finds the end of the opening tag (`>`), and extracts content up to
* the balanced closing tag for that specific element — ensuring correct scoping
* even when the container is a `<ul>` (which `findBalancedContent` does not track).
*
* Counts both `<li>` (bare) and `<li ` (with attributes like `lang="…"`) to handle
* attributed list items while avoiding false matches on `<link>` or `<listing>` tags.
*
* @param html - HTML string to search
* @param containerClass - Class attribute string to match (e.g. `class="perspective-evidence"`)
* @returns Number of `<li>` children found across all matching containers
*/
function countListItemsInClass(html: string, containerClass: string): number {
let total = 0;
let idx = html.indexOf(containerClass);
while (idx !== -1) {
const content = extractContainerContent(html, idx, containerClass);
Eif (content) {
// Count both bare <li> and attributed <li ...> (e.g. <li lang="en">)
// Using '<li>' and '<li ' avoids false matches on <link> or <listing>
total += countOccurrences(content, '<li>') + countOccurrences(content, '<li ');
}
idx = html.indexOf(containerClass, idx + 1);
}
return total;
}
/**
* Count only direct (top-level) `<li>` children of the first container matching
* the given class attribute prefix. Tracks nesting depth of list elements
* (`<ul>`, `<ol>`) so that `<li>` tags inside nested sublists are excluded.
*
* This avoids inflating the count for deep-but-narrow structures where nested
* subnodes are also wrapped in `<li>` elements.
*
* @param html - HTML string to search
* @param containerClassPrefix - Class attribute prefix to locate (e.g. `class="mindmap-branches`)
* @returns Number of direct `<li>` children in the first matching container
*/
function countDirectListChildren(html: string, containerClassPrefix: string): number {
const idx = html.indexOf(containerClassPrefix);
if (idx < 0) return 0;
// Complete the attribute value to find the closing quote
const quoteEnd = html.indexOf('"', idx + containerClassPrefix.length);
Iif (quoteEnd < 0) return 0;
const fullAttr = html.slice(idx, quoteEnd + 1);
const content = extractContainerContent(html, idx, fullAttr);
Iif (!content) return 0;
// Scan through the container content tracking list nesting depth.
// Only count <li> tags at depth 0 (direct children of the container).
let count = 0;
let listDepth = 0;
const tagPattern = /<(\/?)(?:ul|ol|li)(?=[\s>/])/giu;
let m = tagPattern.exec(content);
while (m) {
const isClosing = m[1] === '/';
const tagLower = m[0].replace(/^<\/?/u, '').toLowerCase();
if (tagLower === 'ul' || tagLower === 'ol') {
if (isClosing) {
listDepth = Math.max(0, listDepth - 1);
} else {
listDepth++;
}
} else if (tagLower === 'li' && !isClosing && listDepth === 0) {
count++;
}
m = tagPattern.exec(content);
}
return count;
}
/**
* attribute match at the given position. Identifies the tag name by searching
* backwards for `<tagname`, then finds the end of the opening tag (`>`), and
* uses balanced tag matching on that specific element to locate the matching
* closing tag.
*
* @param html - Full HTML string
* @param attrIdx - Index where the matched attribute starts within `html`
* @param attr - The attribute string that was matched
* @returns Inner HTML of the container, or empty string if extraction fails
*/
function extractContainerContent(html: string, attrIdx: number, attr: string): string {
// Search backwards from the attribute to find the opening `<`
let openBracket = attrIdx - 1;
while (openBracket >= 0 && html[openBracket] !== '<') openBracket--;
Iif (openBracket < 0) return '';
// Extract the tag name (e.g. "ul", "div", "section")
const tagSlice = html.slice(openBracket + 1, attrIdx).trim();
const tagNameMatch = /^([a-z][a-z0-9]*)/iu.exec(tagSlice);
Iif (!tagNameMatch) return '';
const tagName = tagNameMatch[1];
// Find the end of the opening tag
const closeAngle = html.indexOf('>', attrIdx + attr.length);
Iif (closeAngle < 0) return '';
const contentStart = closeAngle + 1;
// Balanced matching for this specific tag name
// tagName is validated by /^([a-z][a-z0-9]*)/ — alphanumeric only, safe for RegExp
// eslint-disable-next-line security/detect-non-literal-regexp
const balancePattern = new RegExp(`</?${tagName}[\\s>/]`, 'giu');
balancePattern.lastIndex = contentStart;
let depth = 1;
let m = balancePattern.exec(html);
while (m) {
if (m[0].startsWith('</')) {
depth--;
if (depth === 0) return html.slice(contentStart, m.index);
} else {
depth++;
}
m = balancePattern.exec(html);
}
return '';
}
/**
* Count evidence and document references in HTML.
* Detects evidence markers from the actual generator output:
* - `<ul class="perspective-evidence"><li>…</li></ul>` — deep-analysis evidence items
* - `<ul class="evidence-refs"><li lang="…">…</li></ul>` — reasoning-chain evidence items
* - `class="swot-ref-evidence"` — SWOT cross-reference evidence markers
* - `class="evidence"` — generic evidence markers (legacy / tests)
* - `data-reference` attributes
* - EP document reference codes (TA-, PE-, A9-, P9_TA patterns)
*
* Strips `<script>` blocks once up front so that JSON-LD metadata and other
* inline scripts do not inflate any evidence counts.
*
* @param html - Raw HTML string
* @returns Number of evidence references found
*/
function countEvidenceRefs(html: string): number {
// Strip script blocks (e.g. JSON-LD) once for all evidence counting
// to avoid inflated counts from matching substrings inside scripts.
const htmlNoScripts = stripScriptBlocks(html);
// Count <li> items inside perspective-evidence containers (deep-analysis generator)
const perspectiveEvidenceItems = countListItemsInClass(
htmlNoScripts,
'class="perspective-evidence"'
);
// Count <li> items inside evidence-refs containers (reasoning-chain generator)
const evidenceRefsItems = countListItemsInClass(htmlNoScripts, 'class="evidence-refs"');
// Count SWOT cross-reference evidence markers (swot-content generator)
const swotRefEvidence = countOccurrences(htmlNoScripts, 'class="swot-ref-evidence"');
// Legacy / generic evidence markers
const evidenceClasses = countOccurrences(htmlNoScripts, 'class="evidence"');
const dataRefs = countOccurrences(htmlNoScripts, 'data-reference');
// EP document reference codes
const matched = new Set<string>();
for (const pattern of EP_DOC_PATTERNS) {
pattern.lastIndex = 0;
const hits = htmlNoScripts.match(pattern);
if (hits) {
for (const hit of hits) {
matched.add(hit);
}
}
}
const epRefs = matched.size;
return (
perspectiveEvidenceItems +
evidenceRefsItems +
swotRefEvidence +
evidenceClasses +
dataRefs +
epRefs
);
}
/**
* Search backwards from a given index to find the opening `<` of the
* enclosing HTML tag.
*
* @param html - HTML string to search
* @param fromIdx - Index to start searching backwards from
* @param fallback - Value to return if no `<` is found
* @returns Index of the opening `<`, or `fallback` if none is found
*/
function findTagStartBefore(html: string, fromIdx: number, fallback: number): number {
let pos = fromIdx - 1;
while (pos >= 0 && html[pos] !== '<') pos--;
return pos >= 0 ? pos : fallback;
}
/**
* Count evidence markers inside deep-analysis sections only, preventing
* inflation from evidence markers elsewhere in the article.
*
* Iterates ALL matching deep-analysis sections (not just the first),
* so articles with multiple deep-analysis blocks are fully counted.
* Deduplicates matches across class and id patterns using the opening-tag
* boundary (`<` position) so that a tag matching both class and id is counted
* only once.
*
* Detects:
* - `<li>` items inside `<ul class="perspective-evidence">` — real generator output
* - `class="swot-ref-evidence"` — SWOT cross-reference evidence
* - `class="evidence"` — generic/legacy evidence markers
* - `data-reference` attributes
*
* @param html - Raw HTML string
* @returns Evidence count restricted to deep-analysis section(s)
*/
function countDeepAnalysisSectionEvidence(html: string): number {
const openPatterns = [/class="deep-analysis"[^>]*>/giu, /id="[^"]*deep[^"]*"[^>]*>/giu];
let total = 0;
const countedTagStarts = new Set<number>();
for (const pattern of openPatterns) {
pattern.lastIndex = 0;
let openMatch = pattern.exec(html);
while (openMatch) {
const tagStart = findTagStartBefore(html, openMatch.index, openMatch.index);
if (!countedTagStarts.has(tagStart)) {
countedTagStarts.add(tagStart);
const startIdx = openMatch.index + openMatch[0].length;
const sectionContent = findBalancedContent(html, startIdx);
Eif (sectionContent) {
total +=
countListItemsInClass(sectionContent, 'class="perspective-evidence"') +
countOccurrences(sectionContent, 'class="swot-ref-evidence"') +
countOccurrences(sectionContent, 'class="evidence"') +
countOccurrences(sectionContent, 'data-reference');
}
}
openMatch = pattern.exec(html);
}
}
return total;
}
/**
* Compute the mindmap branch count.
*
* Priority order:
* 1. `data-branch-count="N"` attribute on `.mindmap-container` — the generators
* set this to the exact number of top-level branches (`config.branches.length`
* or `domainNodes.length`), so it is the most reliable source.
* 2. `class="mindmap-branch"` element count.
* 3. Direct `<li>` children of the first `.mindmap-branches` container (layer-1
* only), using nesting-depth tracking to exclude nested subnodes.
*
* @param html - Raw HTML string
* @returns Number of mindmap branches detected
*/
function computeMindmapBranches(html: string): number {
// 1. Prefer the explicit data-branch-count attribute set by the generators
const attrMatch = /data-branch-count="(\d+)"/u.exec(html);
if (attrMatch?.[1]) {
const parsed = parseInt(attrMatch[1], 10);
Eif (parsed > 0) return parsed;
}
// 2. Count class="mindmap-branch" elements
const branchCount = countOccurrences(html, 'class="mindmap-branch"');
if (branchCount > 0) return branchCount;
// 3. Count only direct <li> children of the mindmap-branches container
return countDirectListChildren(html, 'class="mindmap-branches');
}
/**
* Walk forward from a starting position to find balanced closing tag content.
*
* @param html - HTML string
* @param startIdx - Position right after the opening tag
* @returns Content between the opening and its balanced closing tag, or empty string
*/
function findBalancedContent(html: string, startIdx: number): string {
let depth = 1;
const closeTagPattern = /<\/?(?:div|section|article)[\s>/]/giu;
closeTagPattern.lastIndex = startIdx;
let tagMatch = closeTagPattern.exec(html);
while (tagMatch) {
if (tagMatch[0].startsWith('</')) {
depth--;
if (depth === 0) return html.slice(startIdx, tagMatch.index);
} else {
depth++;
}
tagMatch = closeTagPattern.exec(html);
}
return '';
}
/**
* Check whether generic/placeholder phrases appear in the article text.
*
* @param html - Raw HTML string
* @returns true if any generic phrase pattern is detected
*/
function hasGenericPhrases(html: string): boolean {
return GENERIC_PHRASE_PATTERNS.some((pattern) => pattern.test(html));
}
/**
* Clamp a numeric value between 0 and 100.
*
* @param value - Value to clamp
* @returns Value clamped to [0, 100]
*/
function clamp100(value: number): number {
return Math.max(0, Math.min(100, value));
}
// ─── Analysis depth assessment ────────────────────────────────────────────────
/**
* Assess the analytical depth of an article by detecting keyword signals.
*
* Accepts either raw HTML or pre-extracted plain text. When called from
* {@link scoreArticleQuality} the text is already extracted, avoiding
* redundant HTML stripping.
*
* @param htmlOrText - Raw HTML string or pre-extracted plain text
* @param preExtracted - If true, treat `htmlOrText` as already-extracted plain text
* @returns Analysis depth score with per-dimension flags and composite score
*/
export function assessAnalysisDepth(htmlOrText: string, preExtracted = false): AnalysisDepthScore {
const text = preExtracted ? htmlOrText : extractPlainText(htmlOrText);
const politicalContextPresent = containsAnyKeyword(text, POLITICAL_CONTEXT_KEYWORDS);
const coalitionDynamicsAnalyzed = containsAnyKeyword(text, COALITION_DYNAMICS_KEYWORDS);
const historicalContextProvided = containsAnyKeyword(text, HISTORICAL_CONTEXT_KEYWORDS);
const evidenceBasedConclusions = containsAnyKeyword(text, EVIDENCE_BASED_KEYWORDS);
const scenarioPlanning = containsAnyKeyword(text, SCENARIO_PLANNING_KEYWORDS);
const confidenceLevelsIndicated = containsAnyKeyword(text, CONFIDENCE_LEVEL_KEYWORDS);
const dimensions = [
politicalContextPresent,
coalitionDynamicsAnalyzed,
historicalContextProvided,
evidenceBasedConclusions,
scenarioPlanning,
confidenceLevelsIndicated,
];
const presentCount = dimensions.filter(Boolean).length;
const score = clamp100(Math.round((presentCount / dimensions.length) * 100));
return {
politicalContextPresent,
coalitionDynamicsAnalyzed,
historicalContextProvided,
evidenceBasedConclusions,
scenarioPlanning,
confidenceLevelsIndicated,
score,
};
}
// ─── Stakeholder coverage assessment ─────────────────────────────────────────
/**
* Assess how many stakeholder perspectives are covered in the article text.
*
* Accepts either raw HTML or pre-extracted plain text. When called from
* {@link scoreArticleQuality} the text is already extracted, avoiding
* redundant HTML stripping.
*
* @param htmlOrText - Raw HTML string or pre-extracted plain text
* @param preExtracted - If true, treat `htmlOrText` as already-extracted plain text
* @returns Stakeholder coverage assessment with present/missing lists and scores
*/
export function assessStakeholderCoverage(
htmlOrText: string,
preExtracted = false
): StakeholderCoverage {
const text = preExtracted ? htmlOrText : extractPlainText(htmlOrText);
const perspectivesPresent: string[] = [];
const perspectivesMissing: string[] = [];
for (const stakeholder of STAKEHOLDER_KEYWORDS) {
if (containsAnyKeyword(text, stakeholder.keywords)) {
perspectivesPresent.push(stakeholder.name);
} else {
perspectivesMissing.push(stakeholder.name);
}
}
const total = STAKEHOLDER_KEYWORDS.length;
const balanceScore = clamp100(Math.round((perspectivesPresent.length / total) * 100));
// Reasoning quality: bonus for not using generic phrases, penalty if they are present
const genericPenalty = hasGenericPhrases(text) ? 20 : 0;
const baseReasoningScore = balanceScore;
const reasoningQuality = clamp100(baseReasoningScore - genericPenalty);
return {
perspectivesPresent,
perspectivesMissing,
balanceScore,
reasoningQuality,
};
}
// ─── Visualization quality assessment ────────────────────────────────────────
/**
* Assess the quality of embedded visual elements (SWOT, dashboard, mindmap, deep analysis).
*
* @param html - Raw HTML string of the article
* @returns Visualization quality assessment with per-element flags and composite score
*/
export function assessVisualizationQuality(html: string): VisualizationQuality {
// SWOT: exact class-token match supports multi-class attributes
// (e.g. class="swot-analysis swot-multidimensional")
const swotPresent =
hasExactClassToken(html, 'swot-analysis') || html.includes('id="swot-analysis"');
// Partial match: quadrant classes include a variant suffix (e.g. "swot-quadrant swot-strengths")
const swotDimensions =
countOccurrences(html, 'swot-quadrant') + countOccurrences(html, 'data-dimension');
// Dashboard: exact class-token match prevents false positives from hyphenated
// classes like "dashboard-grid", "dashboard-panel", "dashboard-chart"
const dashboardPresent = hasExactClassToken(html, 'dashboard') || html.includes('id="dashboard"');
const dashboardMetrics =
countExactClassToken(html, 'metric-card') + countExactClassToken(html, 'dashboard-metric');
// Trend indicators: metric-trend-up/-down/-stable classes, or arrow symbols
const dashboardTrends =
html.includes('class="metric-trend-') || html.includes('↑') || html.includes('↓');
// Mindmap: exact class-token match supports multi-class attributes
const mindmapPresent =
hasExactClassToken(html, 'mindmap-section') ||
hasExactClassToken(html, 'mindmap-container') ||
html.includes('id="mindmap"');
const mindmapBranches = mindmapPresent ? computeMindmapBranches(html) : 0;
const deepAnalysisPresent =
html.includes(CLASS_DEEP_ANALYSIS) || /id="[^"]*deep[^"]*"/iu.test(html);
// Restrict evidence counting to deep-analysis section(s) only to avoid
// inflating the metric with evidence markers elsewhere in the article.
const deepAnalysisEvidence = deepAnalysisPresent ? countDeepAnalysisSectionEvidence(html) : 0;
const score = computeVisualizationScore({
swotPresent,
swotDimensions,
dashboardPresent,
dashboardMetrics,
dashboardTrends,
mindmapPresent,
mindmapBranches,
deepAnalysisPresent,
deepAnalysisEvidence,
});
return {
swotPresent,
swotDimensions,
dashboardPresent,
dashboardMetrics,
dashboardTrends,
mindmapPresent,
mindmapBranches,
deepAnalysisPresent,
deepAnalysisEvidence,
score,
};
}
/**
* Compute a 0–100 composite visualization score from individual element assessments.
*
* @param v - Visualization dimensions (without the score field)
* @returns Composite visualization score clamped to [0, 100]
*/
function computeVisualizationScore(v: Omit<VisualizationQuality, 'score'>): number {
let score = 0;
// SWOT contribution (max 25 points)
if (v.swotPresent) {
score += 10;
score += Math.min(15, v.swotDimensions * 5);
}
// Dashboard contribution (max 25 points)
if (v.dashboardPresent) {
score += 10;
score += Math.min(10, v.dashboardMetrics * 2);
if (v.dashboardTrends) score += 5;
}
// Mindmap contribution (max 25 points)
if (v.mindmapPresent) {
score += 10;
score += Math.min(15, v.mindmapBranches * 5);
}
// Deep analysis contribution (max 25 points)
if (v.deepAnalysisPresent) {
score += 10;
score += Math.min(15, v.deepAnalysisEvidence * 3);
}
return clamp100(score);
}
// ─── Overall score calculation ────────────────────────────────────────────────
/**
* Compute the weighted overall quality score (0–100) from component scores.
*
* Weights:
* - Analysis depth: 25 %
* - Stakeholder balance: 20 %
* - Visualization: 25 %
* - Word count: 15 %
* - Evidence references: 15 %
*
* @param depth - Analysis depth score object
* @param coverage - Stakeholder coverage score object
* @param viz - Visualization quality score object
* @param wordCount - Plain-text word count of the article
* @param evidenceRefs - Number of evidence/document references
* @returns Overall quality score clamped to [0, 100]
*/
export function calculateOverallScore(
depth: AnalysisDepthScore,
coverage: StakeholderCoverage,
viz: VisualizationQuality,
wordCount: number,
evidenceRefs: number
): number {
const wordCountScore = clamp100(
Math.round(((wordCount - WORD_COUNT_MIN) / (WORD_COUNT_MAX - WORD_COUNT_MIN)) * 100)
);
const evidenceScore = clamp100(Math.round((evidenceRefs / EVIDENCE_MAX) * 100));
const overall =
depth.score * WEIGHT_ANALYSIS_DEPTH +
coverage.balanceScore * WEIGHT_STAKEHOLDER +
viz.score * WEIGHT_VISUALIZATION +
wordCountScore * WEIGHT_WORD_COUNT +
evidenceScore * WEIGHT_EVIDENCE;
return clamp100(Math.round(overall));
}
// ─── Grade assignment ─────────────────────────────────────────────────────────
/**
* Convert an overall score to a letter grade.
*
* @param score - Overall quality score (0–100)
* @returns Letter grade A–F
*/
function scoreToGrade(score: number): ArticleGrade {
if (score >= GRADE_A_MIN) return 'A';
Iif (score >= GRADE_B_MIN) return 'B';
if (score >= GRADE_C_MIN) return 'C';
if (score >= GRADE_D_MIN) return 'D';
return 'F';
}
// ─── Non-English language adjustments ─────────────────────────────────────────
/**
* Baseline score assigned to keyword-based analysis dimensions for non-English
* articles, so that translated content is not systematically penalised by the
* English-only keyword lists.
*/
const NON_ENGLISH_BASELINE = 50;
/**
* Adjust analysis depth for non-English articles.
*
* English keyword lists do not apply to translated text, so we raise the
* composite score to at least a baseline when the raw keyword scan scored low.
* This prevents non-English articles from being systematically under-scored.
*
* @param depth - Raw analysis depth from keyword scanning
* @returns Adjusted analysis depth with a baseline floor
*/
function adjustNonEnglishAnalysisDepth(depth: AnalysisDepthScore): AnalysisDepthScore {
return {
...depth,
score: Math.max(depth.score, NON_ENGLISH_BASELINE),
};
}
/**
* Adjust stakeholder coverage for non-English articles.
*
* English stakeholder keyword lists may not match translated terms, so we apply
* a baseline floor to prevent systematically low balance and reasoning scores.
*
* @param coverage - Raw stakeholder coverage from keyword scanning
* @returns Adjusted coverage with baseline floors on balance and reasoning scores
*/
function adjustNonEnglishStakeholderCoverage(coverage: StakeholderCoverage): StakeholderCoverage {
return {
...coverage,
balanceScore: Math.max(coverage.balanceScore, NON_ENGLISH_BASELINE),
reasoningQuality: Math.max(coverage.reasoningQuality, NON_ENGLISH_BASELINE),
};
}
// ─── Recommendation generation ────────────────────────────────────────────────
/**
* Generate actionable improvement recommendations based on a partial quality report.
*
* For non-English articles, keyword-based analysis-depth and stakeholder recommendations
* are omitted because the underlying boolean flags derive from English-only keyword
* lists, making them unreliable for translated content.
*
* @param report - Quality report without the recommendations field
* @returns Array of recommendation strings (may be empty for high-quality articles)
*/
export function generateRecommendations(
report: Omit<ArticleQualityReport, 'recommendations'>
): string[] {
const recs: string[] = [];
const isEnglish = report.lang === 'en';
addWordCountRecommendations(report, recs);
// Only emit keyword-dependent recommendations for English articles; non-English
// articles have baseline-adjusted scores and keyword detection is not reliable.
if (isEnglish) {
addAnalysisDepthRecommendations(report.analysisDepth, recs);
addStakeholderRecommendations(report.stakeholderCoverage, recs);
}
addVisualizationRecommendations(report.visualizationQuality, recs);
addEvidenceRecommendations(report, recs);
return recs;
}
/**
* Add word-count related recommendations.
*
* @param report - Partial quality report
* @param recs - Mutable array to push recommendations into
*/
function addWordCountRecommendations(
report: Omit<ArticleQualityReport, 'recommendations'>,
recs: string[]
): void {
if (report.wordCount < 500) {
recs.push('Expand article length to at least 500 words for Grade C quality');
} else if (report.wordCount < WORD_COUNT_MAX) {
recs.push(
`Increase article depth to ${WORD_COUNT_MAX} words for Grade A quality (currently ${report.wordCount})`
);
}
}
/**
* Add analysis-depth recommendations.
*
* @param depth - Analysis depth score
* @param recs - Mutable array to push recommendations into
*/
function addAnalysisDepthRecommendations(depth: AnalysisDepthScore, recs: string[]): void {
if (!depth.politicalContextPresent) {
recs.push('Add political context: discuss coalitions, majorities, and opposition dynamics');
}
if (!depth.coalitionDynamicsAnalyzed) {
recs.push('Analyse coalition dynamics between EPP, S&D, Renew, and other groups');
}
if (!depth.historicalContextProvided) {
recs.push('Provide historical context by comparing to previous terms or key milestones');
}
if (!depth.evidenceBasedConclusions) {
recs.push('Support conclusions with data, figures, or cited evidence');
}
if (!depth.scenarioPlanning) {
recs.push('Include forward-looking scenarios or projections');
}
if (!depth.confidenceLevelsIndicated) {
recs.push('State confidence levels or acknowledge uncertainty in assessments');
}
}
/**
* Add stakeholder coverage recommendations.
*
* @param coverage - Stakeholder coverage assessment
* @param recs - Mutable array to push recommendations into
*/
function addStakeholderRecommendations(coverage: StakeholderCoverage, recs: string[]): void {
if (coverage.perspectivesMissing.length > 0) {
recs.push(
`Add perspectives from missing stakeholders: ${coverage.perspectivesMissing.join(', ')}`
);
}
if (coverage.reasoningQuality < 60) {
recs.push(
'Replace generic phrases (e.g. "several MEPs", "some countries") with specific named entities'
);
}
}
/**
* Add visualization quality recommendations.
*
* @param viz - Visualization quality assessment
* @param recs - Mutable array to push recommendations into
*/
function addVisualizationRecommendations(viz: VisualizationQuality, recs: string[]): void {
if (!viz.swotPresent) {
recs.push('Add a SWOT analysis section to strengthen political assessment');
} else if (viz.swotDimensions < 3) {
recs.push(`Expand SWOT dimensions to at least 3 (currently ${viz.swotDimensions})`);
}
if (!viz.dashboardPresent) {
recs.push('Add a data dashboard with key metrics for quantitative support');
} else Iif (viz.dashboardMetrics < 5) {
recs.push(`Add more dashboard metrics to reach 5 (currently ${viz.dashboardMetrics})`);
}
if (!viz.mindmapPresent) {
recs.push('Add a mindmap to illustrate relationships and conceptual structure');
} else Iif (viz.mindmapBranches < 3) {
recs.push(`Add more mindmap branches to reach 3 (currently ${viz.mindmapBranches})`);
}
if (!viz.deepAnalysisPresent) {
recs.push('Add deep-analysis sections to provide substantive investigative content');
} else if (viz.deepAnalysisEvidence < 3) {
recs.push(
`Include more evidence items in deep-analysis sections (currently ${viz.deepAnalysisEvidence})`
);
}
}
/**
* Add evidence-reference recommendations.
*
* @param report - Partial quality report
* @param recs - Mutable array to push recommendations into
*/
function addEvidenceRecommendations(
report: Omit<ArticleQualityReport, 'recommendations'>,
recs: string[]
): void {
if (report.evidenceReferences < 3) {
recs.push('Add at least 3 evidence references or EP document citations');
} else if (report.evidenceReferences < 10) {
recs.push(
`Increase evidence references to 10 for Grade A quality (currently ${report.evidenceReferences})`
);
}
}
// ─── Public API ───────────────────────────────────────────────────────────────
/**
* Score the quality of a generated article and produce a comprehensive report.
*
* This is the primary entry point for the quality assessment pipeline.
*
* @param html - Complete HTML string of the generated article
* @param articleId - Unique identifier for the article (typically the filename slug)
* @param lang - Language code of the article (e.g. `"en"`, `"de"`)
* @param articleType - Article category string (e.g. `"week-ahead"`)
* @returns Comprehensive quality report including grade, score and recommendations
*/
export function scoreArticleQuality(
html: string,
articleId: string,
lang: string,
articleType: string
): ArticleQualityReport {
// Extract plain text once to avoid redundant HTML stripping in sub-assessors
const plainText = extractPlainText(html);
const wordCount = plainText ? plainText.split(' ').length : 0;
const analysisSections = countAnalysisSections(html);
const evidenceReferences = countEvidenceRefs(html);
// For non-English articles, keyword-based analysis-depth and stakeholder scoring
// uses English keywords which may not appear in translated text. Weight structural
// signals (visualization, word count, evidence) more heavily by treating keyword
// dimensions as partially present when the language is not English.
const isEnglish = lang === 'en';
const analysisDepth = isEnglish
? assessAnalysisDepth(plainText, true)
: adjustNonEnglishAnalysisDepth(assessAnalysisDepth(plainText, true));
const stakeholderCoverage = isEnglish
? assessStakeholderCoverage(plainText, true)
: adjustNonEnglishStakeholderCoverage(assessStakeholderCoverage(plainText, true));
const visualizationQuality = assessVisualizationQuality(html);
const overallScore = calculateOverallScore(
analysisDepth,
stakeholderCoverage,
visualizationQuality,
wordCount,
evidenceReferences
);
const grade = scoreToGrade(overallScore);
const passesQualityGate = overallScore >= QUALITY_GATE_THRESHOLD;
// Derive the article's date from its ID when possible (slug format: YYYY-MM-DD-…),
// falling back to the current execution date only if the ID does not contain a date prefix.
const dateMatch = ARTICLE_DATE_PATTERN.exec(articleId);
const date = dateMatch?.[1] ?? new Date().toISOString().split('T')[0] ?? '';
const partial: Omit<ArticleQualityReport, 'recommendations'> = {
articleId,
date,
type: articleType,
lang,
wordCount,
analysisSections,
evidenceReferences,
analysisDepth,
stakeholderCoverage,
visualizationQuality,
overallScore,
grade,
passesQualityGate,
};
const recommendations = generateRecommendations(partial);
return { ...partial, recommendations };
}
|