96 lines
3.5 KiB
JavaScript
96 lines
3.5 KiB
JavaScript
/* source-validator.v1.js
|
||
Centralizes logic for deciding if a "Source" string should link to WOL.
|
||
Exposes:
|
||
- SourceValidator.isWOLSource(text) -> boolean
|
||
- SourceValidator.buildWOLSearchURL(text) -> string
|
||
*/
|
||
window.SourceValidator = (function () {
|
||
// Publications / codes that produce valid WOL links.
|
||
// Added: uw, su, re, lvs, rs (rs was already present).
|
||
const PUB_CODES = [
|
||
"wp","ws","yb","mwb","w","g","ap","apf","be","bh","br","bt","btg","cf","cl","ct","dp",
|
||
"fg","fy","gt","hb","im","ip","it","jv","ka","kj","kl","lf","lff","ll","ly","my","od",
|
||
"pe","po","pt","rr","rs","sg","sh","si","td","tp","tr","ts","un","jy",
|
||
"uw","su","re","lvs","lp","yy","yp2","yp"// new
|
||
];
|
||
|
||
// Year validation rules (applies only if a year can be parsed from the source).
|
||
// Watchtower (w/wp/ws) back to 1950; Awake (g) back to 1970.
|
||
const YEAR_RULES = [
|
||
{ codes: ["w","wp","ws"], minYear: 1950 },
|
||
{ codes: ["g"], minYear: 1970 }
|
||
];
|
||
|
||
// Normalize helper
|
||
function normalize(s) { return (s || "").trim().toLowerCase(); }
|
||
|
||
// Choose the longest matching code at the start (so "ws" beats "w").
|
||
const PUB_CODES_SORTED = [...PUB_CODES].sort((a,b)=>b.length-a.length);
|
||
|
||
function leadingCode(textLower) {
|
||
for (const code of PUB_CODES_SORTED) {
|
||
if (textLower.startsWith(code)) return code;
|
||
}
|
||
return null;
|
||
}
|
||
|
||
// Try to extract a year that appears right after the leading code (allow spaces),
|
||
// accepting either 4-digit (e.g., 1955, 2001) or 2-digit (e.g., 55, 95, 12) forms.
|
||
function extractYearAfterCode(textLower, code) {
|
||
let s = textLower.slice(code.length).trim();
|
||
|
||
// 1) Look for a 4-digit year first
|
||
let m = s.match(/\b(1[89]\d{2}|20\d{2})\b/); // 1800-2099 (broad, but OK)
|
||
if (m) {
|
||
return parseInt(m[1], 10);
|
||
}
|
||
|
||
// 2) If not found, accept a 2-digit year at the *start* of the remainder,
|
||
// or right after an optional space: e.g., "w55 1/1", "w 95", "g70 1/22"
|
||
m = s.match(/^\s*(\d{2})\b/);
|
||
if (m) {
|
||
const yy = parseInt(m[1], 10);
|
||
// Infer century based on publication + threshold logic
|
||
// - For Watchtower: 50–99 -> 1950–1999; 00–49 -> 2000–2049
|
||
// - For Awake: 70–99 -> 1970–1999; 00–69 -> 2000–2069
|
||
if (code === "g") {
|
||
return yy >= 70 ? (1900 + yy) : (2000 + yy);
|
||
}
|
||
if (code === "w" || code === "wp" || code === "ws") {
|
||
return yy >= 50 ? (1900 + yy) : (2000 + yy);
|
||
}
|
||
// For other pubs, if they ever include 2-digit years, assume 1900+yy≥70 else 2000+yy
|
||
return yy >= 70 ? (1900 + yy) : (2000 + yy);
|
||
}
|
||
|
||
// No recognizable year → don't enforce year limits
|
||
return null;
|
||
}
|
||
|
||
function passesYearRuleIfPresent(textLower, code) {
|
||
const rule = YEAR_RULES.find(r => r.codes.includes(code));
|
||
if (!rule) return true; // no year rule for this pub
|
||
|
||
const y = extractYearAfterCode(textLower, code);
|
||
if (y == null) return true; // no year present → allow
|
||
return y >= rule.minYear;
|
||
}
|
||
|
||
function isWOLSource(text) {
|
||
const t = normalize(text);
|
||
if (!t) return false;
|
||
|
||
const code = leadingCode(t);
|
||
if (!code) return false;
|
||
|
||
// If starts with a known pub code, it’s WOL-capable — but enforce year rules where applicable.
|
||
return passesYearRuleIfPresent(t, code);
|
||
}
|
||
|
||
function buildWOLSearchURL(text) {
|
||
const q = encodeURIComponent(text || "");
|
||
return `https://wol.jw.org/en/wol/l/r1/lp-e?q=${q}`;
|
||
}
|
||
|
||
return { isWOLSource, buildWOLSearchURL };
|
||
})(); |