@alpaca-software/40kdc-data 0.5.11 → 0.5.14

package/dist/extract-faction-pack.js

@@ -0,0 +1,318 @@
+/**
+ * Standardized extractor for GW 11e Faction Pack PDFs → IP-safe staging JSON.
+ *
+ * The faction packs are the structural source of truth for detachments,
+ * stratagems, and enhancements. This tool pulls the *facts* out of a pack —
+ * detachment names, stratagem names + CP cost + type + phase, enhancement names,
+ * detachment-rule names — and writes them to `data/_audit/faction-pack-input/
+ * <faction>.json` for human review before they are merged into `data/core/`.
+ *
+ * IP firewall: this tool MUST NOT emit GW rules/effect prose. It reads the
+ * `WHEN:` line of a stratagem card only to *derive* the structured phase /
+ * player-turn fields; the prose itself is never written to the staging file.
+ * Stored output is names + numeric/enum metadata only.
+ *
+ * The packs are *supplements*, not complete codexes: a pack contains only the
+ * extra/updated detachments. Absence from a pack does not imply removal. This
+ * tool reports only what a pack actually contains; completeness is reconciled
+ * against GW's full Detachment-Points list separately.
+ *
+ * Parsing uses `pdftotext -bbox-layout`, which emits per-word coordinates. GW
+ * lays each card out as discrete text blocks; we locate the "<DET> – <TYPE>
+ * STRATAGEM" label block, then read the name block directly above it, the CP
+ * block in the column gutter, and the `WHEN:` block below — all by coordinate.
+ * This is layout-independent (column positions differ across packs and pages).
+ * Anything the parser cannot determine is left null and recorded in a per-entry
+ * `flags` array for the reviewer.
+ *
+ * Requires `pdftotext` (poppler) on PATH — authoring-time only, not CI.
+ *
+ * Usage:
+ *   npx tsx tools/src/extract-faction-pack.ts <pdf-path> --faction <faction-id>
+ *   npx tsx tools/src/extract-faction-pack.ts --all --dir <pack-pdf-dir>
+ */
+import { readdirSync, writeFileSync, mkdirSync } from "node:fs";
+import { resolve, basename } from "node:path";
+import { fileURLToPath } from "node:url";
+import { runPdftotext, allBlocks, slug, titleCase, normCaps, isCapsHeader, sameColumn, stripCp, } from "./pack-blocks.js";
+export { slug, titleCase } from "./pack-blocks.js";
+const __dirname = fileURLToPath(new URL(".", import.meta.url));
+const DATA_ROOT = resolve(__dirname, "../../data");
+const OUT_DIR = resolve(DATA_ROOT, "_audit", "faction-pack-input");
+/** Pack-filename fragment → 40kdc faction id. `--all` uses this to route packs. */
+const PACK_FACTION = [
+    ["space-marines", "adeptus-astartes"],
+    ["black_templars", "black-templars"],
+    ["blood_angels", "blood-angels"],
+    ["dark_angels", "dark-angels"],
+    ["space_wolves", "space-wolves"],
+    ["deathwatch", "deathwatch"],
+    ["grey_knights", "grey-knights"],
+];
+const STRAT_TYPE = {
+    "BATTLE TACTIC": "battle-tactic",
+    "STRATEGIC PLOY": "strategic-ploy",
+    "EPIC DEED": "epic-deed",
+    WARGEAR: "wargear",
+};
+const TYPE_RE = /(BATTLE TACTIC|STRATEGIC PLOY|EPIC DEED|WARGEAR)\s+STRATAGEM$/;
+const PHASES = ["command", "movement", "shooting", "charge", "fight"];
+// Note: "Legends Datasheets" (not bare "Legends") — else it eats the detachment
+// "Legends of Saga and Song".
+const SECTION_TERMINATORS = /^(Datasheets|Rules Updates|Legends Datasheets|Imperial Armour|Index)\b/i;
+/** Datasheet/table headers that can masquerade as caps "enhancement" names. */
+const DATASHEET_NOISE = new Set([
+    "RANGED WEAPONS", "MELEE WEAPONS", "RANGE", "WARGEAR OPTIONS", "UNIT COMPOSITION",
+    "KEYWORDS", "LEADER", "TRANSPORT", "DAMAGED", "SUPREME COMMANDER",
+]);
+/** Non-detachment section headers that bound the last detachment's content. */
+const SECTION_MARKER = /^(DATASHEETS|RULES UPDATES|LEGENDS DATASHEETS|IMPERIAL ARMOUR|INDEX)\b/;
+/** Derive phases / player-turn from a `WHEN:` line (the prose itself is discarded). */
+function parseWhen(when) {
+    const low = when.toLowerCase();
+    const phases = PHASES.filter((p) => low.includes(`${p} phase`));
+    let player_turn = null;
+    if (/opponent['’]?s?\b/.test(low))
+        player_turn = "opponent-turn";
+    else if (/\byour\b/.test(low) && phases.length)
+        player_turn = "your-turn";
+    else if (phases.length)
+        player_turn = "either";
+    const flag = phases.length ? undefined : "phase-review"; // e.g. "any phase"
+    return { phases, player_turn, flag };
+}
+/** Extract stratagems, enhancements, and rule name from a detachment's blocks. */
+export function parseDetachmentBlocks(blocks, detachmentId) {
+    const flags = [];
+    const labels = blocks.filter((b) => /\bSTRATAGEM$/.test(b.text) && !/^STRATAGEMS?$/.test(b.text));
+    const cpBlocks = blocks.filter((b) => /^\d+\s?CP$/.test(b.text));
+    const whenBlocks = blocks.filter((b) => /^WHEN:/i.test(b.text));
+    // --- detachment rule name: first caps header under a DETACHMENT RULE(S) marker
+    let detachment_rule_name = null;
+    const ruleHeader = blocks.filter((b) => /^DETACHMENT RULES?$/.test(b.text)).sort((a, b) => a.gy - b.gy)[0];
+    if (ruleHeader) {
+        const rn = blocks
+            .filter((b) => sameColumn(b.x, ruleHeader.x) && b.gy > ruleHeader.gy && isCapsHeader(b.text))
+            .sort((a, b) => a.gy - b.gy)[0];
+        if (rn)
+            detachment_rule_name = titleCase(rn.text);
+    }
+    // --- stratagems: keyed off each "<…> <TYPE> STRATAGEM" label block.
+    const stratagems = [];
+    const seenS = new Set();
+    for (const label of labels) {
+        const typeM = label.text.match(TYPE_RE);
+        const type = typeM ? STRAT_TYPE[typeM[1]] : null;
+        // name: nearest caps-header block directly above the label, same column
+        const nameBlock = blocks
+            .filter((b) => sameColumn(b.x, label.x) && b.gy < label.gy && isCapsHeader(stripCp(b.text).name))
+            .sort((a, b) => b.gy - a.gy)[0];
+        if (!nameBlock)
+            continue;
+        const { name, cp: inlineCp } = stripCp(nameBlock.text);
+        const id = slug(name);
+        if (!id || seenS.has(id))
+            continue;
+        seenS.add(id);
+        // cp: inline on the name block, else the CP block within the card window.
+        // Card layouts differ: new detachments put CP at the name's right edge on
+        // the name row; reprints put it in the left gutter below the label. So scan
+        // a window [name row → just below label] × [card width] and pick the CP
+        // nearest the name horizontally (avoids grabbing the adjacent column's CP).
+        let cp = inlineCp;
+        if (cp === null) {
+            const inCard = cpBlocks
+                .filter((b) => b.gy >= nameBlock.gy - 15 && b.gy <= label.gy + 70 && b.x >= nameBlock.x - 70 && b.x <= nameBlock.x + 260)
+                .sort((a, b) => Math.abs(a.x - nameBlock.x) - Math.abs(b.x - nameBlock.x))[0];
+            if (inCard)
+                cp = Number(inCard.text.match(/(\d+)/)[1]);
+        }
+        // phase: the first WHEN: block below the label in the same column
+        const when = whenBlocks
+            .filter((b) => sameColumn(b.x, label.x) && b.gy >= label.gy)
+            .sort((a, b) => a.gy - b.gy)[0];
+        const sFlags = [];
+        let phases = [];
+        let player_turn = null;
+        if (when) {
+            const w = parseWhen(when.text);
+            phases = w.phases;
+            player_turn = w.player_turn;
+            if (w.flag)
+                sFlags.push(w.flag);
+        }
+        else {
+            sFlags.push("phase-missing");
+        }
+        if (!type)
+            sFlags.push("type-missing");
+        if (cp === null)
+            sFlags.push("cp-missing");
+        sFlags.push("timing-missing"); // pack rarely states once-per-phase/turn/battle
+        stratagems.push({ id, name: titleCase(name), type, cp_cost: cp, phases, player_turn, timing: null, flags: sFlags });
+    }
+    if (!detachment_rule_name)
+        flags.push("rule-name-missing");
+    // --- enhancements (subtractive): a detachment page's caps headers are the
+    //     title, tagline, rule name, section words, stratagem names, type labels,
+    //     and enhancement names. Remove every known category; the rest are
+    //     enhancements. Each must be a name+description pair, i.e. followed by a
+    //     mixed-case prose block in its column. Cost is never printed in the pack.
+    const ruleSlug = detachment_rule_name ? slug(detachment_rule_name) : "";
+    const hasProseBelow = (b) => blocks
+        .filter((o) => sameColumn(o.x, b.x) && o.gy > b.gy)
+        .sort((p, q) => p.gy - q.gy)
+        .some((o, i) => i === 0 && /[a-z]/.test(o.text)); // nearest block below is prose
+    const enhancements = [];
+    const seenE = new Set();
+    for (const b of blocks) {
+        if (!isCapsHeader(b.text) || DATASHEET_NOISE.has(normCaps(b.text)))
+            continue;
+        const isUpgrade = / UPGRADE$/.test(b.text);
+        const name = b.text.replace(/ UPGRADE$/, "").trim();
+        const id = slug(name);
+        if (!id || seenE.has(id) || seenS.has(id))
+            continue; // skip stratagem names
+        if (id === detachmentId || id === ruleSlug)
+            continue; // skip title / rule name
+        if (!hasProseBelow(b))
+            continue; // enhancements always have a description
+        seenE.add(id);
+        enhancements.push({
+            id,
+            name: titleCase(name),
+            cost: null,
+            flags: ["cost-missing", "name-needs-review", ...(isUpgrade ? ["upgrade-tag?"] : [])],
+        });
+    }
+    if (!enhancements.length)
+        flags.push("no-enhancements-parsed");
+    return { detachment_rule_name, stratagems, enhancements, flags };
+}
+/**
+ * Parse the pack's table of contents (first two pages) into the ordered list of
+ * detachments with their start pages. Uses non-layout extraction so the two
+ * columns don't interleave; dot leaders are private-font glyphs (non-word
+ * chars), so a TOC row is `<Name><2+ non-word><page#>`. Entries are kept only
+ * between the "Detachments" header and the first non-detachment section.
+ */
+export function parseToc(tocText) {
+    const rows = [];
+    let inDetachments = false;
+    let sawHeader = false;
+    for (const line of tocText.split("\n")) {
+        const m = line.match(/^\s*([A-Z][A-Za-z0-9'’ \-]+?)[^\w\s]{2,}\s*(\d+)\s*$/);
+        if (!m)
+            continue;
+        const name = m[1].trim();
+        const page = Number(m[2]);
+        if (/^Detachments$/i.test(name)) {
+            inDetachments = true;
+            sawHeader = true;
+            continue;
+        }
+        if (sawHeader && SECTION_TERMINATORS.test(name)) {
+            if (inDetachments)
+                rows.push({ name, page, section: true }); // bounds the last detachment
+            inDetachments = false;
+            continue;
+        }
+        if (inDetachments)
+            rows.push({ name, page });
+    }
+    return rows;
+}
+/**
+ * Segment a pack into per-detachment block slices. Detachment names come (ordered)
+ * from the TOC; content boundaries come from block positions: each detachment runs
+ * from its title block to the next detachment, the next non-detachment section
+ * marker, or a ~2-page backstop — whichever comes first. Shared by `extractPack`
+ * (names/metadata) and `author-input-pack` (rule bodies) so both segment identically.
+ */
+export function detachmentSegments(pdf) {
+    const names = parseToc(runPdftotext(["-f", "1", "-l", "2", pdf, "-"]))
+        .filter((t) => !t.section)
+        .map((t) => t.name);
+    const blocks = allBlocks(pdf);
+    // start of each detachment = its title block (exact uppercase name match, first
+    // occurrence). Stratagem labels like "NAME – TYPE STRATAGEM" are not exact, so
+    // they don't collide with the title.
+    const starts = names
+        .map((name) => {
+        const title = blocks.filter((b) => normCaps(b.text) === normCaps(name)).sort((a, b) => a.gy - b.gy)[0];
+        return title ? { name, gy: title.gy } : null;
+    })
+        .filter((s) => s !== null)
+        .sort((a, b) => a.gy - b.gy);
+    const sectionGys = blocks.filter((b) => SECTION_MARKER.test(normCaps(b.text))).map((b) => b.gy).sort((a, b) => a - b);
+    const segments = [];
+    for (let i = 0; i < starts.length; i++) {
+        const start = starts[i].gy;
+        const nextDet = starts[i + 1]?.gy ?? Infinity;
+        const nextSection = sectionGys.find((gy) => gy > start) ?? Infinity;
+        // Backstop: a detachment is at most ~2 pages (gy page stride is 100000).
+        // Bounds the last detachment when no datasheets/section divider block is
+        // detected, so it can't bleed into the datasheet pages that follow.
+        const end = Math.min(nextDet, nextSection, start + 200000);
+        segments.push({
+            id: slug(starts[i].name),
+            name: starts[i].name,
+            blocks: blocks.filter((b) => b.gy >= start && b.gy < end),
+        });
+    }
+    return segments;
+}
+export function extractPack(pdf, faction) {
+    const detachments = detachmentSegments(pdf).map((seg) => ({
+        id: seg.id,
+        name: seg.name,
+        ...parseDetachmentBlocks(seg.blocks, seg.id),
+    }));
+    return { faction_id: faction, source_pack: basename(pdf), detachments };
+}
+function factionForPack(file) {
+    const lower = file.toLowerCase();
+    return PACK_FACTION.find(([frag]) => lower.includes(frag))?.[1];
+}
+function main() {
+    const argv = process.argv.slice(2);
+    mkdirSync(OUT_DIR, { recursive: true });
+    let jobs = [];
+    if (argv[0] === "--all") {
+        const dirIdx = argv.indexOf("--dir");
+        const dir = dirIdx === -1 ? undefined : argv[dirIdx + 1];
+        if (!dir) {
+            console.error("Usage: extract-faction-pack.ts --all --dir <pack-pdf-dir>");
+            process.exit(1);
+        }
+        for (const f of readdirSync(dir)) {
+            if (!/faction_pack.*\.pdf$/i.test(f))
+                continue;
+            const faction = factionForPack(f);
+            if (faction)
+                jobs.push({ pdf: resolve(dir, f), faction });
+        }
+    }
+    else {
+        const facIdx = argv.indexOf("--faction");
+        const pdf = argv[0];
+        if (!pdf || facIdx === -1) {
+            console.error("Usage: extract-faction-pack.ts <pdf-path> --faction <faction-id>");
+            process.exit(1);
+        }
+        jobs = [{ pdf: resolve(pdf), faction: argv[facIdx + 1] }];
+    }
+    for (const { pdf, faction } of jobs) {
+        const result = extractPack(pdf, faction);
+        const out = resolve(OUT_DIR, `${faction}.json`);
+        writeFileSync(out, JSON.stringify(result, null, 2) + "\n");
+        const nStrat = result.detachments.reduce((a, d) => a + d.stratagems.length, 0);
+        const nEnh = result.detachments.reduce((a, d) => a + d.enhancements.length, 0);
+        console.log(`  ${faction}: ${result.detachments.length} detachments, ${nStrat} stratagems, ${nEnh} enhancements → ${out}`);
+    }
+}
+const isMain = process.argv[1] &&
+    resolve(process.argv[1]).replace(/\.\w+$/, "") === fileURLToPath(import.meta.url).replace(/\.\w+$/, "");
+if (isMain)
+    main();
+//# sourceMappingURL=extract-faction-pack.js.map

package/dist/extract-faction-pack.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract-faction-pack.js","sourceRoot":"","sources":["../src/extract-faction-pack.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAgCG;AACH,OAAO,EAAE,WAAW,EAAE,aAAa,EAAE,SAAS,EAAE,MAAM,SAAS,CAAC;AAChE,OAAO,EAAE,OAAO,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAC;AAC9C,OAAO,EAAE,aAAa,EAAE,MAAM,UAAU,CAAC;AACzC,OAAO,EAEL,YAAY,EACZ,SAAS,EACT,IAAI,EACJ,SAAS,EACT,QAAQ,EACR,YAAY,EACZ,UAAU,EACV,OAAO,GACR,MAAM,kBAAkB,CAAC;AAE1B,OAAO,EAAE,IAAI,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAEnD,MAAM,SAAS,GAAG,aAAa,CAAC,IAAI,GAAG,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;AAC/D,MAAM,SAAS,GAAG,OAAO,CAAC,SAAS,EAAE,YAAY,CAAC,CAAC;AACnD,MAAM,OAAO,GAAG,OAAO,CAAC,SAAS,EAAE,QAAQ,EAAE,oBAAoB,CAAC,CAAC;AAEnE,mFAAmF;AACnF,MAAM,YAAY,GAA+C;IAC/D,CAAC,eAAe,EAAE,kBAAkB,CAAC;IACrC,CAAC,gBAAgB,EAAE,gBAAgB,CAAC;IACpC,CAAC,cAAc,EAAE,cAAc,CAAC;IAChC,CAAC,aAAa,EAAE,aAAa,CAAC;IAC9B,CAAC,cAAc,EAAE,cAAc,CAAC;IAChC,CAAC,YAAY,EAAE,YAAY,CAAC;IAC5B,CAAC,cAAc,EAAE,cAAc,CAAC;CACjC,CAAC;AAEF,MAAM,UAAU,GAA2B;IACzC,eAAe,EAAE,eAAe;IAChC,gBAAgB,EAAE,gBAAgB;IAClC,WAAW,EAAE,WAAW;IACxB,OAAO,EAAE,SAAS;CACnB,CAAC;AACF,MAAM,OAAO,GAAG,+DAA+D,CAAC;AAChF,MAAM,MAAM,GAAG,CAAC,SAAS,EAAE,UAAU,EAAE,UAAU,EAAE,QAAQ,EAAE,OAAO,CAAU,CAAC;AAC/E,gFAAgF;AAChF,8BAA8B;AAC9B,MAAM,mBAAmB,GAAG,yEAAyE,CAAC;AAgCtG,+EAA+E;AAC/E,MAAM,eAAe,GAAG,IAAI,GAAG,CAAC;IAC9B,gBAAgB,EAAE,eAAe,EAAE,OAAO,EAAE,iBAAiB,EAAE,kBAAkB;IACjF,UAAU,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,mBAAmB;CAClE,CAAC,CAAC;AACH,+EAA+E;AAC/E,MAAM,cAAc,GAAG,wEAAwE,CAAC;AAEhG,uFAAuF;AACvF,SAAS,SAAS,CAAC,IAAY;IAC7B,MAAM,GAAG,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;IAC/B,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC;IAChE,IAAI,WAAW,GAAkB,IAAI,CAAC;IACtC,IAAI,mBAAmB,CAAC,IAAI,CAAC,GAAG,CAAC;QAAE,WAAW,GAAG,eAAe,CAAC;SAC5D,IAAI,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,MAAM,CAAC,MAAM;QAAE,WAAW,GAAG,WAAW,CAAC;SACrE,IAAI,MAAM,CAAC,MAAM;QAAE,WAAW,GAAG,QAAQ,CAAC;IAC/C,MAAM,IAAI,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,mBAAmB;IAC5E,OAAO,EAAE,MAAM,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;AACvC,CAAC;AAED,kFAAkF;AAClF,MAAM,UAAU,qBAAqB,CAAC,MAAe,EAAE,YAAoB;IACzE,MAAM,KAAK,GAAa,EAAE,CAAC;IAC3B,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;IAClG,MAAM,QAAQ,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;IACjE,MAAM,UAAU,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;IAEhE,gFAAgF;IAChF,IAAI,oBAAoB,GAAkB,IAAI,CAAC;IAC/C,MAAM,UAAU,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,qBAAqB,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;IAC3G,IAAI,UAAU,EAAE,CAAC;QACf,MAAM,EAAE,GAAG,MAAM;aACd,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,GAAG,UAAU,CAAC,EAAE,IAAI,YAAY,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;aAC5F,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAClC,IAAI,EAAE;YAAE,oBAAoB,GAAG,SAAS,CAAC,EAAE,CAAC,IAAI,CAAC,CAAC;IACpD,CAAC;IAED,qEAAqE;IACrE,MAAM,UAAU,GAAyB,EAAE,CAAC;IAC5C,MAAM,KAAK,GAAG,IAAI,GAAG,EAAU,CAAC;IAChC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,MAAM,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;QACxC,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;QAEjD,wEAAwE;QACxE,MAAM,SAAS,GAAG,MAAM;aACrB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,GAAG,KAAK,CAAC,EAAE,IAAI,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC;aAChG,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAClC,IAAI,CAAC,SAAS;YAAE,SAAS;QACzB,MAAM,EAAE,IAAI,EAAE,EAAE,EAAE,QAAQ,EAAE,GAAG,OAAO,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC;QACvD,MAAM,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;QACtB,IAAI,CAAC,EAAE,IAAI,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YAAE,SAAS;QACnC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAEd,0EAA0E;QAC1E,0EAA0E;QAC1E,4EAA4E;QAC5E,wEAAwE;QACxE,4EAA4E;QAC5E,IAAI,EAAE,GAAG,QAAQ,CAAC;QAClB,IAAI,EAAE,KAAK,IAAI,EAAE,CAAC;YAChB,MAAM,MAAM,GAAG,QAAQ;iBACpB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,SAAS,CAAC,EAAE,GAAG,EAAE,IAAI,CAAC,CAAC,EAAE,IAAI,KAAK,CAAC,EAAE,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,IAAI,SAAS,CAAC,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,IAAI,SAAS,CAAC,CAAC,GAAG,GAAG,CAAC;iBACxH,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;YAChF,IAAI,MAAM;gBAAE,EAAE,GAAG,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAC1D,CAAC;QAED,kEAAkE;QAClE,MAAM,IAAI,GAAG,UAAU;aACpB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,KAAK,CAAC,EAAE,CAAC;aAC3D,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QAClC,MAAM,MAAM,GAAa,EAAE,CAAC;QAC5B,IAAI,MAAM,GAAa,EAAE,CAAC;QAC1B,IAAI,WAAW,GAAkB,IAAI,CAAC;QACtC,IAAI,IAAI,EAAE,CAAC;YACT,MAAM,CAAC,GAAG,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAC/B,MAAM,GAAG,CAAC,CAAC,MAAM,CAAC;YAClB,WAAW,GAAG,CAAC,CAAC,WAAW,CAAC;YAC5B,IAAI,CAAC,CAAC,IAAI;gBAAE,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QAClC,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,eAAe,CAAC,CAAC;QAC/B,CAAC;QACD,IAAI,CAAC,IAAI;YAAE,MAAM,CAAC,IAAI,CAAC,cAAc,CAAC,CAAC;QACvC,IAAI,EAAE,KAAK,IAAI;YAAE,MAAM,CAAC,IAAI,CAAC,YAAY,CAAC,CAAC;QAC3C,MAAM,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAC,CAAC,gDAAgD;QAE/E,UAAU,CAAC,IAAI,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,SAAS,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,EAAE,EAAE,MAAM,EAAE,WAAW,EAAE,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC,CAAC;IACtH,CAAC;IAED,IAAI,CAAC,oBAAoB;QAAE,KAAK,CAAC,IAAI,CAAC,mBAAmB,CAAC,CAAC;IAE3D,2EAA2E;IAC3E,8EAA8E;IAC9E,uEAAuE;IACvE,6EAA6E;IAC7E,+EAA+E;IAC/E,MAAM,QAAQ,GAAG,oBAAoB,CAAC,CAAC,CAAC,IAAI,CAAC,oBAAoB,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;IACxE,MAAM,aAAa,GAAG,CAAC,CAAQ,EAAW,EAAE,CAC1C,MAAM;SACH,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC;SAClD,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC;SAC3B,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,+BAA+B;IACrF,MAAM,YAAY,GAA2B,EAAE,CAAC;IAChD,MAAM,KAAK,GAAG,IAAI,GAAG,EAAU,CAAC;IAChC,KAAK,MAAM,CAAC,IAAI,MAAM,EAAE,CAAC;QACvB,IAAI,CAAC,YAAY,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,eAAe,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;YAAE,SAAS;QAC7E,MAAM,SAAS,GAAG,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QAC3C,MAAM,IAAI,GAAG,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;QACpD,MAAM,EAAE,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC;QACtB,IAAI,CAAC,EAAE,IAAI,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC,IAAI,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;YAAE,SAAS,CAAC,uBAAuB;QAC5E,IAAI,EAAE,KAAK,YAAY,IAAI,EAAE,KAAK,QAAQ;YAAE,SAAS,CAAC,yBAAyB;QAC/E,IAAI,CAAC,aAAa,CAAC,CAAC,CAAC;YAAE,SAAS,CAAC,yCAAyC;QAC1E,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QACd,YAAY,CAAC,IAAI,CAAC;YAChB,EAAE;YACF,IAAI,EAAE,SAAS,CAAC,IAAI,CAAC;YACrB,IAAI,EAAE,IAAI;YACV,KAAK,EAAE,CAAC,cAAc,EAAE,mBAAmB,EAAE,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;SACrF,CAAC,CAAC;IACL,CAAC;IACD,IAAI,CAAC,YAAY,CAAC,MAAM;QAAE,KAAK,CAAC,IAAI,CAAC,wBAAwB,CAAC,CAAC;IAE/D,OAAO,EAAE,oBAAoB,EAAE,UAAU,EAAE,YAAY,EAAE,KAAK,EAAE,CAAC;AACnE,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,QAAQ,CAAC,OAAe;IACtC,MAAM,IAAI,GAA6D,EAAE,CAAC;IAC1E,IAAI,aAAa,GAAG,KAAK,CAAC;IAC1B,IAAI,SAAS,GAAG,KAAK,CAAC;IACtB,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QACvC,MAAM,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,sDAAsD,CAAC,CAAC;QAC7E,IAAI,CAAC,CAAC;YAAE,SAAS;QACjB,MAAM,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAC1B,IAAI,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAChC,aAAa,GAAG,IAAI,CAAC;YACrB,SAAS,GAAG,IAAI,CAAC;YACjB,SAAS;QACX,CAAC;QACD,IAAI,SAAS,IAAI,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAChD,IAAI,aAAa;gBAAE,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,6BAA6B;YAC1F,aAAa,GAAG,KAAK,CAAC;YACtB,SAAS;QACX,CAAC;QACD,IAAI,aAAa;YAAE,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC;IAC/C,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AASD;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,GAAW;IAC5C,MAAM,KAAK,GAAG,QAAQ,CAAC,YAAY,CAAC,CAAC,IAAI,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;SACnE,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC;SACzB,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;IACtB,MAAM,MAAM,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC;IAE9B,gFAAgF;IAChF,+EAA+E;IAC/E,qCAAqC;IACrC,MAAM,MAAM,GAAG,KAAK;SACjB,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QACZ,MAAM,KAAK,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QACvG,OAAO,KAAK,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,KAAK,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;IAC/C,CAAC,CAAC;SACD,MAAM,CAAC,CAAC,CAAC,EAAqC,EAAE,CAAC,CAAC,KAAK,IAAI,CAAC;SAC5D,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC;IAC/B,MAAM,UAAU,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;IAEtH,MAAM,QAAQ,GAAwB,EAAE,CAAC;IACzC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACvC,MAAM,KAAK,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QAC3B,MAAM,OAAO,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,IAAI,QAAQ,CAAC;QAC9C,MAAM,WAAW,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,EAAE,GAAG,KAAK,CAAC,IAAI,QAAQ,CAAC;QACpE,yEAAyE;QACzE,yEAAyE;QACzE,oEAAoE;QACpE,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,EAAE,WAAW,EAAE,KAAK,GAAG,MAAM,CAAC,CAAC;QAC3D,QAAQ,CAAC,IAAI,CAAC;YACZ,EAAE,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;YACxB,IAAI,EAAE,MAAM,CAAC,CAAC,CAAC,CAAC,IAAI;YACpB,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,KAAK,IAAI,CAAC,CAAC,EAAE,GAAG,GAAG,CAAC;SAC1D,CAAC,CAAC;IACL,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED,MAAM,UAAU,WAAW,CAAC,GAAW,EAAE,OAAe;IACtD,MAAM,WAAW,GAA0B,kBAAkB,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;QAC/E,EAAE,EAAE,GAAG,CAAC,EAAE;QACV,IAAI,EAAE,GAAG,CAAC,IAAI;QACd,GAAG,qBAAqB,CAAC,GAAG,CAAC,MAAM,EAAE,GAAG,CAAC,EAAE,CAAC;KAC7C,CAAC,CAAC,CAAC;IACJ,OAAO,EAAE,UAAU,EAAE,OAAO,EAAE,WAAW,EAAE,QAAQ,CAAC,GAAG,CAAC,EAAE,WAAW,EAAE,CAAC;AAC1E,CAAC;AAED,SAAS,cAAc,CAAC,IAAY;IAClC,MAAM,KAAK,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;IACjC,OAAO,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;AAClE,CAAC;AAED,SAAS,IAAI;IACX,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IACnC,SAAS,CAAC,OAAO,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;IAExC,IAAI,IAAI,GAA4C,EAAE,CAAC;IACvD,IAAI,IAAI,CAAC,CAAC,CAAC,KAAK,OAAO,EAAE,CAAC;QACxB,MAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;QACrC,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACzD,IAAI,CAAC,GAAG,EAAE,CAAC;YACT,OAAO,CAAC,KAAK,CAAC,2DAA2D,CAAC,CAAC;YAC3E,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAClB,CAAC;QACD,KAAK,MAAM,CAAC,IAAI,WAAW,CAAC,GAAG,CAAC,EAAE,CAAC;YACjC,IAAI,CAAC,uBAAuB,CAAC,IAAI,CAAC,CAAC,CAAC;gBAAE,SAAS;YAC/C,MAAM,OAAO,GAAG,cAAc,CAAC,CAAC,CAAC,CAAC;YAClC,IAAI,OAAO;gBAAE,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,OAAO,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,OAAO,EAAE,CAAC,CAAC;QAC5D,CAAC;IACH,CAAC;SAAM,CAAC;QACN,MAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC;QACzC,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;QACpB,IAAI,CAAC,GAAG,IAAI,MAAM,KAAK,CAAC,CAAC,EAAE,CAAC;YAC1B,OAAO,CAAC,KAAK,CAAC,kEAAkE,CAAC,CAAC;YAClF,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;QAClB,CAAC;QACD,IAAI,GAAG,CAAC,EAAE,GAAG,EAAE,OAAO,CAAC,GAAG,CAAC,EAAE,OAAO,EAAE,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC;IAC5D,CAAC;IAED,KAAK,MAAM,EAAE,GAAG,EAAE,OAAO,EAAE,IAAI,IAAI,EAAE,CAAC;QACpC,MAAM,MAAM,GAAG,WAAW,CAAC,GAAG,EAAE,OAAO,CAAC,CAAC;QACzC,MAAM,GAAG,GAAG,OAAO,CAAC,OAAO,EAAE,GAAG,OAAO,OAAO,CAAC,CAAC;QAChD,aAAa,CAAC,GAAG,EAAE,IAAI,CAAC,SAAS,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC;QAC3D,MAAM,MAAM,GAAG,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,UAAU,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QAC/E,MAAM,IAAI,GAAG,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,YAAY,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QAC/E,OAAO,CAAC,GAAG,CACT,KAAK,OAAO,KAAK,MAAM,CAAC,WAAW,CAAC,MAAM,iBAAiB,MAAM,gBAAgB,IAAI,mBAAmB,GAAG,EAAE,CAC9G,CAAC;IACJ,CAAC;AACH,CAAC;AAED,MAAM,MAAM,GACV,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC;IACf,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,KAAK,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;AAC1G,IAAI,MAAM;IAAE,IAAI,EAAE,CAAC","sourcesContent":["/**\n * Standardized extractor for GW 11e Faction Pack PDFs → IP-safe staging JSON.\n *\n * The faction packs are the structural source of truth for detachments,\n * stratagems, and enhancements. This tool pulls the *facts* out of a pack —\n * detachment names, stratagem names + CP cost + type + phase, enhancement names,\n * detachment-rule names — and writes them to `data/_audit/faction-pack-input/\n * <faction>.json` for human review before they are merged into `data/core/`.\n *\n * IP firewall: this tool MUST NOT emit GW rules/effect prose. It reads the\n * `WHEN:` line of a stratagem card only to *derive* the structured phase /\n * player-turn fields; the prose itself is never written to the staging file.\n * Stored output is names + numeric/enum metadata only.\n *\n * The packs are *supplements*, not complete codexes: a pack contains only the\n * extra/updated detachments. Absence from a pack does not imply removal. This\n * tool reports only what a pack actually contains; completeness is reconciled\n * against GW's full Detachment-Points list separately.\n *\n * Parsing uses `pdftotext -bbox-layout`, which emits per-word coordinates. GW\n * lays each card out as discrete text blocks; we locate the \"<DET> – <TYPE>\n * STRATAGEM\" label block, then read the name block directly above it, the CP\n * block in the column gutter, and the `WHEN:` block below — all by coordinate.\n * This is layout-independent (column positions differ across packs and pages).\n * Anything the parser cannot determine is left null and recorded in a per-entry\n * `flags` array for the reviewer.\n *\n * Requires `pdftotext` (poppler) on PATH — authoring-time only, not CI.\n *\n * Usage:\n *   npx tsx tools/src/extract-faction-pack.ts <pdf-path> --faction <faction-id>\n *   npx tsx tools/src/extract-faction-pack.ts --all --dir <pack-pdf-dir>\n */\nimport { readdirSync, writeFileSync, mkdirSync } from \"node:fs\";\nimport { resolve, basename } from \"node:path\";\nimport { fileURLToPath } from \"node:url\";\nimport {\n  type Block,\n  runPdftotext,\n  allBlocks,\n  slug,\n  titleCase,\n  normCaps,\n  isCapsHeader,\n  sameColumn,\n  stripCp,\n} from \"./pack-blocks.js\";\n\nexport { slug, titleCase } from \"./pack-blocks.js\";\n\nconst __dirname = fileURLToPath(new URL(\".\", import.meta.url));\nconst DATA_ROOT = resolve(__dirname, \"../../data\");\nconst OUT_DIR = resolve(DATA_ROOT, \"_audit\", \"faction-pack-input\");\n\n/** Pack-filename fragment → 40kdc faction id. `--all` uses this to route packs. */\nconst PACK_FACTION: Array<[fragment: string, faction: string]> = [\n  [\"space-marines\", \"adeptus-astartes\"],\n  [\"black_templars\", \"black-templars\"],\n  [\"blood_angels\", \"blood-angels\"],\n  [\"dark_angels\", \"dark-angels\"],\n  [\"space_wolves\", \"space-wolves\"],\n  [\"deathwatch\", \"deathwatch\"],\n  [\"grey_knights\", \"grey-knights\"],\n];\n\nconst STRAT_TYPE: Record<string, string> = {\n  \"BATTLE TACTIC\": \"battle-tactic\",\n  \"STRATEGIC PLOY\": \"strategic-ploy\",\n  \"EPIC DEED\": \"epic-deed\",\n  WARGEAR: \"wargear\",\n};\nconst TYPE_RE = /(BATTLE TACTIC|STRATEGIC PLOY|EPIC DEED|WARGEAR)\\s+STRATAGEM$/;\nconst PHASES = [\"command\", \"movement\", \"shooting\", \"charge\", \"fight\"] as const;\n// Note: \"Legends Datasheets\" (not bare \"Legends\") — else it eats the detachment\n// \"Legends of Saga and Song\".\nconst SECTION_TERMINATORS = /^(Datasheets|Rules Updates|Legends Datasheets|Imperial Armour|Index)\\b/i;\n\nexport interface ExtractedStratagem {\n  id: string;\n  name: string;\n  type: string | null;\n  cp_cost: number | null;\n  phases: string[];\n  player_turn: string | null;\n  timing: string | null;\n  flags: string[];\n}\nexport interface ExtractedEnhancement {\n  id: string;\n  name: string;\n  cost: null; // never in the pack — comes from the Munitorum Field Manual\n  flags: string[];\n}\nexport interface ExtractedDetachment {\n  id: string;\n  name: string;\n  detachment_rule_name: string | null;\n  stratagems: ExtractedStratagem[];\n  enhancements: ExtractedEnhancement[];\n  flags: string[];\n}\nexport interface PackExtract {\n  faction_id: string;\n  source_pack: string;\n  detachments: ExtractedDetachment[];\n}\n\n/** Datasheet/table headers that can masquerade as caps \"enhancement\" names. */\nconst DATASHEET_NOISE = new Set([\n  \"RANGED WEAPONS\", \"MELEE WEAPONS\", \"RANGE\", \"WARGEAR OPTIONS\", \"UNIT COMPOSITION\",\n  \"KEYWORDS\", \"LEADER\", \"TRANSPORT\", \"DAMAGED\", \"SUPREME COMMANDER\",\n]);\n/** Non-detachment section headers that bound the last detachment's content. */\nconst SECTION_MARKER = /^(DATASHEETS|RULES UPDATES|LEGENDS DATASHEETS|IMPERIAL ARMOUR|INDEX)\\b/;\n\n/** Derive phases / player-turn from a `WHEN:` line (the prose itself is discarded). */\nfunction parseWhen(when: string): { phases: string[]; player_turn: string | null; flag?: string } {\n  const low = when.toLowerCase();\n  const phases = PHASES.filter((p) => low.includes(`${p} phase`));\n  let player_turn: string | null = null;\n  if (/opponent['’]?s?\\b/.test(low)) player_turn = \"opponent-turn\";\n  else if (/\\byour\\b/.test(low) && phases.length) player_turn = \"your-turn\";\n  else if (phases.length) player_turn = \"either\";\n  const flag = phases.length ? undefined : \"phase-review\"; // e.g. \"any phase\"\n  return { phases, player_turn, flag };\n}\n\n/** Extract stratagems, enhancements, and rule name from a detachment's blocks. */\nexport function parseDetachmentBlocks(blocks: Block[], detachmentId: string): Omit<ExtractedDetachment, \"id\" | \"name\"> {\n  const flags: string[] = [];\n  const labels = blocks.filter((b) => /\\bSTRATAGEM$/.test(b.text) && !/^STRATAGEMS?$/.test(b.text));\n  const cpBlocks = blocks.filter((b) => /^\\d+\\s?CP$/.test(b.text));\n  const whenBlocks = blocks.filter((b) => /^WHEN:/i.test(b.text));\n\n  // --- detachment rule name: first caps header under a DETACHMENT RULE(S) marker\n  let detachment_rule_name: string | null = null;\n  const ruleHeader = blocks.filter((b) => /^DETACHMENT RULES?$/.test(b.text)).sort((a, b) => a.gy - b.gy)[0];\n  if (ruleHeader) {\n    const rn = blocks\n      .filter((b) => sameColumn(b.x, ruleHeader.x) && b.gy > ruleHeader.gy && isCapsHeader(b.text))\n      .sort((a, b) => a.gy - b.gy)[0];\n    if (rn) detachment_rule_name = titleCase(rn.text);\n  }\n\n  // --- stratagems: keyed off each \"<…> <TYPE> STRATAGEM\" label block.\n  const stratagems: ExtractedStratagem[] = [];\n  const seenS = new Set<string>();\n  for (const label of labels) {\n    const typeM = label.text.match(TYPE_RE);\n    const type = typeM ? STRAT_TYPE[typeM[1]] : null;\n\n    // name: nearest caps-header block directly above the label, same column\n    const nameBlock = blocks\n      .filter((b) => sameColumn(b.x, label.x) && b.gy < label.gy && isCapsHeader(stripCp(b.text).name))\n      .sort((a, b) => b.gy - a.gy)[0];\n    if (!nameBlock) continue;\n    const { name, cp: inlineCp } = stripCp(nameBlock.text);\n    const id = slug(name);\n    if (!id || seenS.has(id)) continue;\n    seenS.add(id);\n\n    // cp: inline on the name block, else the CP block within the card window.\n    // Card layouts differ: new detachments put CP at the name's right edge on\n    // the name row; reprints put it in the left gutter below the label. So scan\n    // a window [name row → just below label] × [card width] and pick the CP\n    // nearest the name horizontally (avoids grabbing the adjacent column's CP).\n    let cp = inlineCp;\n    if (cp === null) {\n      const inCard = cpBlocks\n        .filter((b) => b.gy >= nameBlock.gy - 15 && b.gy <= label.gy + 70 && b.x >= nameBlock.x - 70 && b.x <= nameBlock.x + 260)\n        .sort((a, b) => Math.abs(a.x - nameBlock.x) - Math.abs(b.x - nameBlock.x))[0];\n      if (inCard) cp = Number(inCard.text.match(/(\\d+)/)![1]);\n    }\n\n    // phase: the first WHEN: block below the label in the same column\n    const when = whenBlocks\n      .filter((b) => sameColumn(b.x, label.x) && b.gy >= label.gy)\n      .sort((a, b) => a.gy - b.gy)[0];\n    const sFlags: string[] = [];\n    let phases: string[] = [];\n    let player_turn: string | null = null;\n    if (when) {\n      const w = parseWhen(when.text);\n      phases = w.phases;\n      player_turn = w.player_turn;\n      if (w.flag) sFlags.push(w.flag);\n    } else {\n      sFlags.push(\"phase-missing\");\n    }\n    if (!type) sFlags.push(\"type-missing\");\n    if (cp === null) sFlags.push(\"cp-missing\");\n    sFlags.push(\"timing-missing\"); // pack rarely states once-per-phase/turn/battle\n\n    stratagems.push({ id, name: titleCase(name), type, cp_cost: cp, phases, player_turn, timing: null, flags: sFlags });\n  }\n\n  if (!detachment_rule_name) flags.push(\"rule-name-missing\");\n\n  // --- enhancements (subtractive): a detachment page's caps headers are the\n  //     title, tagline, rule name, section words, stratagem names, type labels,\n  //     and enhancement names. Remove every known category; the rest are\n  //     enhancements. Each must be a name+description pair, i.e. followed by a\n  //     mixed-case prose block in its column. Cost is never printed in the pack.\n  const ruleSlug = detachment_rule_name ? slug(detachment_rule_name) : \"\";\n  const hasProseBelow = (b: Block): boolean =>\n    blocks\n      .filter((o) => sameColumn(o.x, b.x) && o.gy > b.gy)\n      .sort((p, q) => p.gy - q.gy)\n      .some((o, i) => i === 0 && /[a-z]/.test(o.text)); // nearest block below is prose\n  const enhancements: ExtractedEnhancement[] = [];\n  const seenE = new Set<string>();\n  for (const b of blocks) {\n    if (!isCapsHeader(b.text) || DATASHEET_NOISE.has(normCaps(b.text))) continue;\n    const isUpgrade = / UPGRADE$/.test(b.text);\n    const name = b.text.replace(/ UPGRADE$/, \"\").trim();\n    const id = slug(name);\n    if (!id || seenE.has(id) || seenS.has(id)) continue; // skip stratagem names\n    if (id === detachmentId || id === ruleSlug) continue; // skip title / rule name\n    if (!hasProseBelow(b)) continue; // enhancements always have a description\n    seenE.add(id);\n    enhancements.push({\n      id,\n      name: titleCase(name),\n      cost: null,\n      flags: [\"cost-missing\", \"name-needs-review\", ...(isUpgrade ? [\"upgrade-tag?\"] : [])],\n    });\n  }\n  if (!enhancements.length) flags.push(\"no-enhancements-parsed\");\n\n  return { detachment_rule_name, stratagems, enhancements, flags };\n}\n\n/**\n * Parse the pack's table of contents (first two pages) into the ordered list of\n * detachments with their start pages. Uses non-layout extraction so the two\n * columns don't interleave; dot leaders are private-font glyphs (non-word\n * chars), so a TOC row is `<Name><2+ non-word><page#>`. Entries are kept only\n * between the \"Detachments\" header and the first non-detachment section.\n */\nexport function parseToc(tocText: string): Array<{ name: string; page: number; section?: boolean }> {\n  const rows: Array<{ name: string; page: number; section?: boolean }> = [];\n  let inDetachments = false;\n  let sawHeader = false;\n  for (const line of tocText.split(\"\\n\")) {\n    const m = line.match(/^\\s*([A-Z][A-Za-z0-9'’ \\-]+?)[^\\w\\s]{2,}\\s*(\\d+)\\s*$/);\n    if (!m) continue;\n    const name = m[1].trim();\n    const page = Number(m[2]);\n    if (/^Detachments$/i.test(name)) {\n      inDetachments = true;\n      sawHeader = true;\n      continue;\n    }\n    if (sawHeader && SECTION_TERMINATORS.test(name)) {\n      if (inDetachments) rows.push({ name, page, section: true }); // bounds the last detachment\n      inDetachments = false;\n      continue;\n    }\n    if (inDetachments) rows.push({ name, page });\n  }\n  return rows;\n}\n\n/** One detachment's slice of the pack: its id, display name, and bounded blocks. */\nexport interface DetachmentSegment {\n  id: string;\n  name: string;\n  blocks: Block[];\n}\n\n/**\n * Segment a pack into per-detachment block slices. Detachment names come (ordered)\n * from the TOC; content boundaries come from block positions: each detachment runs\n * from its title block to the next detachment, the next non-detachment section\n * marker, or a ~2-page backstop — whichever comes first. Shared by `extractPack`\n * (names/metadata) and `author-input-pack` (rule bodies) so both segment identically.\n */\nexport function detachmentSegments(pdf: string): DetachmentSegment[] {\n  const names = parseToc(runPdftotext([\"-f\", \"1\", \"-l\", \"2\", pdf, \"-\"]))\n    .filter((t) => !t.section)\n    .map((t) => t.name);\n  const blocks = allBlocks(pdf);\n\n  // start of each detachment = its title block (exact uppercase name match, first\n  // occurrence). Stratagem labels like \"NAME – TYPE STRATAGEM\" are not exact, so\n  // they don't collide with the title.\n  const starts = names\n    .map((name) => {\n      const title = blocks.filter((b) => normCaps(b.text) === normCaps(name)).sort((a, b) => a.gy - b.gy)[0];\n      return title ? { name, gy: title.gy } : null;\n    })\n    .filter((s): s is { name: string; gy: number } => s !== null)\n    .sort((a, b) => a.gy - b.gy);\n  const sectionGys = blocks.filter((b) => SECTION_MARKER.test(normCaps(b.text))).map((b) => b.gy).sort((a, b) => a - b);\n\n  const segments: DetachmentSegment[] = [];\n  for (let i = 0; i < starts.length; i++) {\n    const start = starts[i].gy;\n    const nextDet = starts[i + 1]?.gy ?? Infinity;\n    const nextSection = sectionGys.find((gy) => gy > start) ?? Infinity;\n    // Backstop: a detachment is at most ~2 pages (gy page stride is 100000).\n    // Bounds the last detachment when no datasheets/section divider block is\n    // detected, so it can't bleed into the datasheet pages that follow.\n    const end = Math.min(nextDet, nextSection, start + 200000);\n    segments.push({\n      id: slug(starts[i].name),\n      name: starts[i].name,\n      blocks: blocks.filter((b) => b.gy >= start && b.gy < end),\n    });\n  }\n  return segments;\n}\n\nexport function extractPack(pdf: string, faction: string): PackExtract {\n  const detachments: ExtractedDetachment[] = detachmentSegments(pdf).map((seg) => ({\n    id: seg.id,\n    name: seg.name,\n    ...parseDetachmentBlocks(seg.blocks, seg.id),\n  }));\n  return { faction_id: faction, source_pack: basename(pdf), detachments };\n}\n\nfunction factionForPack(file: string): string | undefined {\n  const lower = file.toLowerCase();\n  return PACK_FACTION.find(([frag]) => lower.includes(frag))?.[1];\n}\n\nfunction main(): void {\n  const argv = process.argv.slice(2);\n  mkdirSync(OUT_DIR, { recursive: true });\n\n  let jobs: Array<{ pdf: string; faction: string }> = [];\n  if (argv[0] === \"--all\") {\n    const dirIdx = argv.indexOf(\"--dir\");\n    const dir = dirIdx === -1 ? undefined : argv[dirIdx + 1];\n    if (!dir) {\n      console.error(\"Usage: extract-faction-pack.ts --all --dir <pack-pdf-dir>\");\n      process.exit(1);\n    }\n    for (const f of readdirSync(dir)) {\n      if (!/faction_pack.*\\.pdf$/i.test(f)) continue;\n      const faction = factionForPack(f);\n      if (faction) jobs.push({ pdf: resolve(dir, f), faction });\n    }\n  } else {\n    const facIdx = argv.indexOf(\"--faction\");\n    const pdf = argv[0];\n    if (!pdf || facIdx === -1) {\n      console.error(\"Usage: extract-faction-pack.ts <pdf-path> --faction <faction-id>\");\n      process.exit(1);\n    }\n    jobs = [{ pdf: resolve(pdf), faction: argv[facIdx + 1] }];\n  }\n\n  for (const { pdf, faction } of jobs) {\n    const result = extractPack(pdf, faction);\n    const out = resolve(OUT_DIR, `${faction}.json`);\n    writeFileSync(out, JSON.stringify(result, null, 2) + \"\\n\");\n    const nStrat = result.detachments.reduce((a, d) => a + d.stratagems.length, 0);\n    const nEnh = result.detachments.reduce((a, d) => a + d.enhancements.length, 0);\n    console.log(\n      `  ${faction}: ${result.detachments.length} detachments, ${nStrat} stratagems, ${nEnh} enhancements → ${out}`,\n    );\n  }\n}\n\nconst isMain =\n  process.argv[1] &&\n  resolve(process.argv[1]).replace(/\\.\\w+$/, \"\") === fileURLToPath(import.meta.url).replace(/\\.\\w+$/, \"\");\nif (isMain) main();\n"]}

package/dist/gen-conformance.js

@@ -114,6 +114,15 @@ function decodeCanonicalSeed(caseDir) {
     if (existsSync(lfTextSeed)) {
         return readFileSync(lfTextSeed, "utf8");
     }
+    // A hand-authored canonical Roster (roster-json) — the lossless carrier used
+    // for cases that no upstream format can yet express, e.g. multi-detachment
+    // 11e lists. Like the legacy seeds it is import-only (no derived round-trip
+    // inputs are generated), since the text/header exporters print every
+    // detachment but their importers read only one.
+    const rosterJsonSeed = join(caseDir, "input.roster-json.json");
+    if (existsSync(rosterJsonSeed)) {
+        return JSON.parse(readFileSync(rosterJsonSeed, "utf8"));
+    }
     throw new Error(`no canonical input found in ${caseDir}`);
 }
 /** Run a decoded payload through the adapter pipeline up to (but not past)
@@ -169,10 +178,15 @@ function genRosters() {
         // exporters can't fully preserve, so the round-trip would fail
         // structurally rather than uncover a parser bug.
         const isNewRecruitSeed = existsSync(join(caseDir, "input.newrecruit-json.json"));
+        // The header text formats (wtc-compact/full, simple) print every detachment
+        // but their parsers read only one, so a multi-detachment roster can't
+        // round-trip through them — emit the export golden but not the derived
+        // round-trip input (the same reasoning as the legacy-ListForge skip).
+        const textRoundTrippable = isNewRecruitSeed && seed.detachments.length <= 1;
         for (const { format, inputName, goldenName } of TEXT_FORMATS) {
             const out = exportRoster(seed, format);
             writeText(join(caseDir, goldenName), out);
-            if (isNewRecruitSeed) {
+            if (textRoundTrippable) {
                 writeText(join(caseDir, inputName), out);
             }
         }
@@ -915,6 +929,11 @@ function genEffectTranslation() {
         .sort((a, b) => (a.id < b.id ? -1 : a.id > b.id ? 1 : 0));
     const seen = new Map();
     const CAP = 5;
+    // Also cover the `applies_to` render path: force-include a few abilities that
+    // carry a curated filter even when their effect-node types are already
+    // exhausted, so the trailing `Applies to:` line stays pinned cross-impl.
+    const APPLIES_CAP = 3;
+    let appliesToSeen = 0;
     const cases = [];
     for (const a of abilities) {
         const types = new Set();
@@ -924,16 +943,30 @@ function genEffectTranslation() {
             if ((seen.get(t) ?? 0) < CAP)
                 fresh = true;
         }
+        const coversAppliesTo = a.raw.applies_to != null && appliesToSeen < APPLIES_CAP;
+        if (coversAppliesTo)
+            fresh = true;
         if (!fresh)
             continue;
         for (const t of types)
             seen.set(t, (seen.get(t) ?? 0) + 1);
-        cases.push({
+        if (a.raw.applies_to != null)
+            appliesToSeen += 1;
+        const entry = {
             caseId: `${a.id}#${cases.length}`,
             effect: a.raw.effect,
             scope: a.raw.scope ?? null,
-            expected: { text: describeAbility({ effect: a.raw.effect, scope: a.raw.scope }) },
-        });
+        };
+        if (a.raw.applies_to != null)
+            entry.applies_to = a.raw.applies_to;
+        entry.expected = {
+            text: describeAbility({
+                effect: a.raw.effect,
+                scope: a.raw.scope,
+                applies_to: a.raw.applies_to,
+            }),
+        };
+        cases.push(entry);
     }
     writeJson(join(CONFORMANCE, "effect-translation", "cases.json"), cases);
     console.log(`effect-translation/cases.json: ${cases.length} cases (${seen.size} node types)`);