india2actual 0.1.0

package/dist/extract/pdf.js

@@ -0,0 +1,439 @@
+import { getDocument } from 'pdfjs-dist/legacy/build/pdf.mjs';
+import { roleForHeader } from '../interpret/synonyms.js';
+import { parseStatementDate } from '../interpret/values.js';
+/**
+ * Text items on the same visual line can differ slightly in baseline, so
+ * y values are clustered rather than compared exactly.
+ */
+const LINE_TOLERANCE = 2;
+/**
+ * How far above its own row a wrapped cell's first line may sit.
+ *
+ * Statement PDFs render a multi-line description cell centred on its row, so
+ * its first line can begin slightly *above* the line carrying the date and
+ * amounts. Without this, that line would attach to the previous transaction.
+ */
+const ATTACH_ABOVE = 6;
+/**
+ * How far short of its column's wrap width a fragment must end to count as a
+ * deliberate line break rather than a wrap.
+ */
+const SHORT_LINE_MARGIN = 8;
+/**
+ * How far below its row a wrapped line may sit and still belong to it.
+ *
+ * Page footers live well below the last transaction on a page, and without a
+ * limit they are folded into it — which is how ICICI's `www.icici.bank.in`
+ * and its toll-free number ended up inside a narration and an amount.
+ */
+const MAX_CONTINUATION_BELOW = 40;
+async function readLines(path, options) {
+    const doc = await getDocument({
+        url: path,
+        useSystemFonts: true,
+        ...(options.password ? { password: options.password } : {}),
+    }).promise;
+    const lines = [];
+    for (let pageNumber = 1; pageNumber <= doc.numPages; pageNumber += 1) {
+        const page = await doc.getPage(pageNumber);
+        const content = await page.getTextContent();
+        const items = [];
+        for (const raw of content.items) {
+            if (!('str' in raw) || !raw.str.trim()) {
+                continue;
+            }
+            const x = raw.transform[4];
+            const width = typeof raw.width === 'number' ? raw.width : 0;
+            items.push({
+                x,
+                right: x + width,
+                y: raw.transform[5],
+                text: raw.str,
+            });
+        }
+        // Pages are independent: y coordinates restart, so lines must not be
+        // clustered across a page boundary.
+        lines.push(...clusterLines(items, pageNumber));
+    }
+    return { lines, pages: doc.numPages };
+}
+/** Group items into lines, top of page first. */
+function clusterLines(items, page) {
+    const sorted = [...items].sort((a, b) => b.y - a.y);
+    const lines = [];
+    for (const item of sorted) {
+        const current = lines[lines.length - 1];
+        if (current && Math.abs(current.y - item.y) <= LINE_TOLERANCE) {
+            current.items.push(item);
+        }
+        else {
+            lines.push({ y: item.y, page, items: [item] });
+        }
+    }
+    for (const line of lines) {
+        line.items.sort((a, b) => a.x - b.x);
+    }
+    return lines;
+}
+/**
+ * Does this line start a transaction?
+ *
+ * Judged from the raw items rather than from assembled cells, because column
+ * inference depends on this and must not depend on columns in turn.
+ */
+function isAnchorLine(line) {
+    return line.items.some(item => parseStatementDate(item.text) !== null);
+}
+/**
+ * Is this a summary row rather than a transaction?
+ *
+ * Statements bracket the table with totals and carried-forward balances:
+ * Federal prints `GRAND TOTAL` with per-column sums, and an
+ * `Opening Balance` row above the first transaction. Neither has a date, so
+ * both would otherwise be folded into the nearest transaction — the grand
+ * total's column sums landing in an amount field, which the balance check
+ * then rejects.
+ */
+function isSummaryLine(line) {
+    const text = line.items
+        .map(item => item.text)
+        .join(' ')
+        .trim();
+    return /^(grand\s*total|sub\s*total|total|opening\s*balance|closing\s*balance|balance\s*(carried|brought)\s*forward|[bc]\/f)\b/i.test(text);
+}
+/** Does this line contribute column labels? */
+function isHeaderLine(line) {
+    if (isAnchorLine(line) || isSummaryLine(line)) {
+        return false;
+    }
+    return (line.items.filter(item => roleForHeader(item.text) !== null).length >= 2);
+}
+/**
+ * Narrowest run of horizontal whitespace that separates two columns.
+ *
+ * Measured against real statements: Federal's tightest genuine gutter is 7
+ * points and ICICI's is 5, while gaps *within* a column are smaller still.
+ */
+const MIN_GUTTER = 4;
+/** Index of the first line contributing column labels, or -1. */
+function firstHeaderIndex(lines) {
+    return lines.findIndex(isHeaderLine);
+}
+/**
+ * The lines making up the transaction table: the header, plus the dated rows
+ * below it.
+ *
+ * Anchors are restricted to lines *after* the header because statement
+ * preambles contain dates of their own — Federal prints `Address Last Updated
+ * On : 13/08/2024` and `Account Open Date : 25/03/2013` above the table, and
+ * without this each becomes a phantom transaction.
+ */
+function tableLines(lines) {
+    const header = firstHeaderIndex(lines);
+    return lines.filter((line, index) => {
+        if (isHeaderLine(line)) {
+            return true;
+        }
+        return isAnchorLine(line) && (header < 0 || index > header);
+    });
+}
+/**
+ * Infer columns from the vertical whitespace that separates them.
+ *
+ * Every column of a statement table is divided from its neighbour by a gutter
+ * that stays empty on every row, so accumulating the horizontal extent of all
+ * table text and splitting on runs of emptiness recovers the columns directly.
+ *
+ * This is markedly more robust than comparing where text starts or merging
+ * overlapping spans, both of which were tried first and failed on real files:
+ *
+ * - Start positions alone split a right-aligned amount from its own header,
+ *   which can begin 36 points to its left.
+ * - Merging overlapping spans bridges columns whenever a header label is wider
+ *   than the column spacing. Federal packs columns 45 points apart with labels
+ *   nearly that wide, collapsing `Withdrawals` and `Deposits` into one cell.
+ *
+ * Only header and dated rows take part. Wrapped narration lines and page
+ * furniture would otherwise bridge gutters — ICICI's footer URL spans the gap
+ * between the serial-number and date columns.
+ */
+export function inferBands(lines) {
+    const source = tableLines(lines);
+    const items = (source.length ? source : lines).flatMap(line => line.items);
+    if (!items.length) {
+        return [];
+    }
+    const width = Math.ceil(Math.max(...items.map(item => item.right))) + MIN_GUTTER;
+    const occupancy = new Uint32Array(width + 1);
+    for (const item of items) {
+        const from = Math.max(0, Math.floor(item.x));
+        const to = Math.min(width, Math.ceil(Math.max(item.right, item.x + 1)));
+        for (let x = from; x < to; x += 1) {
+            occupancy[x] = (occupancy[x] ?? 0) + 1;
+        }
+    }
+    const bands = [];
+    let x = 0;
+    while (x < width) {
+        if (!occupancy[x]) {
+            x += 1;
+            continue;
+        }
+        const left = x;
+        let right = x;
+        while (x < width) {
+            if (occupancy[x]) {
+                x += 1;
+                right = x;
+                continue;
+            }
+            // Bridge gaps too narrow to be a column separator.
+            let gapEnd = x;
+            while (gapEnd < width && !occupancy[gapEnd]) {
+                gapEnd += 1;
+            }
+            if (gapEnd - x < MIN_GUTTER && gapEnd < width) {
+                x = gapEnd;
+                continue;
+            }
+            break;
+        }
+        bands.push({ left, right });
+    }
+    return bands;
+}
+/**
+ * Assign an item to a column.
+ *
+ * Normally the band it overlaps most. When it overlaps none, it goes to the
+ * first band starting at or after it, because a cell's text begins at the
+ * cell's left edge, which sits left of the header label naming it — ICICI's
+ * description cell starts at x=192 while its `Transaction Remarks` header
+ * starts at x=247. Choosing the merely *nearest* band would instead drop a
+ * short wrapped line such as `9f03689dd` (x=192..237) into the preceding
+ * column, whose right edge is closer to it.
+ */
+function bandOf(item, bands) {
+    let bestIndex = -1;
+    let bestOverlap = 0;
+    for (const [index, band] of bands.entries()) {
+        const overlap = Math.min(item.right, band.right) - Math.max(item.x, band.left);
+        if (overlap > bestOverlap) {
+            bestOverlap = overlap;
+            bestIndex = index;
+        }
+    }
+    if (bestIndex >= 0) {
+        return bestIndex;
+    }
+    const toTheRight = bands.findIndex(band => band.left >= item.x);
+    return toTheRight >= 0 ? toTheRight : Math.max(0, bands.length - 1);
+}
+/** Split one line into per-column fragments. */
+function toFragments(line, bands) {
+    const fragments = [];
+    for (const item of line.items) {
+        const column = bandOf(item, bands);
+        const existing = fragments[column];
+        fragments[column] = existing
+            ? {
+                text: `${existing.text} ${item.text}`,
+                right: Math.max(existing.right, item.right),
+            }
+            : { text: item.text, right: item.right };
+    }
+    return fragments;
+}
+/**
+ * The right-hand edge each column's text reaches, i.e. its wrap width.
+ * Used to tell a wrapped line from a deliberate line break.
+ */
+function wrapWidths(rows) {
+    const widths = [];
+    for (const row of rows) {
+        for (const [column, fragment] of row.entries()) {
+            if (!fragment) {
+                continue;
+            }
+            widths[column] = Math.max(widths[column] ?? 0, fragment.right);
+        }
+    }
+    return widths;
+}
+/**
+ * Join a cell's fragments back into one string.
+ *
+ * A fragment ending well short of its column's wrap width was a deliberate
+ * line break, so the next fragment starts a new word and needs a space. One
+ * that runs to the wrap width was broken mid-flow, where inserting a space
+ * would corrupt a reference split across lines (`...837e576` + `b9cc0dd2ee`).
+ *
+ * Statement PDFs need this distinction: ICICI prints a short descriptor line
+ * (`NACH trxn`) above each wrapped narration, and joining that without a
+ * separator yields `NACH trxnACH/HDFC...`.
+ *
+ * A space lost at a wrap that happened to fall on a word boundary is not
+ * recoverable from the PDF, so `FEDERAL BA` can come back as `FEDERALBA`.
+ */
+function joinFragments(fragments, wrapWidth) {
+    let text = '';
+    let previous;
+    for (const fragment of fragments) {
+        if (previous && previous.right < wrapWidth - SHORT_LINE_MARGIN) {
+            text += ' ';
+        }
+        text += fragment.text;
+        previous = fragment;
+    }
+    return text;
+}
+function mergeHeader(rows) {
+    const merged = [];
+    for (const row of rows) {
+        for (const [column, fragment] of row.entries()) {
+            if (!fragment) {
+                continue;
+            }
+            merged[column] = merged[column]
+                ? `${merged[column]} ${fragment.text}`
+                : fragment.text;
+        }
+    }
+    for (let index = 0; index < merged.length; index += 1) {
+        merged[index] ??= '';
+    }
+    return merged;
+}
+export function assembleRows(lines, bands) {
+    const fragmentsFor = lines.map(line => toFragments(line, bands));
+    const isAnchor = lines.map(isAnchorLine);
+    const isHeader = lines.map(isHeaderLine);
+    const widths = wrapWidths(fragmentsFor);
+    const rows = [];
+    // Only the first run of header lines is used. Statement PDFs repeat the
+    // header on every page, and merging all of them would produce labels like
+    // "Balance (INR) Balance (INR) Balance (INR)".
+    const firstHeader = isHeader.indexOf(true);
+    if (firstHeader >= 0) {
+        const group = [];
+        for (let index = firstHeader; index < lines.length; index += 1) {
+            if (isAnchor[index]) {
+                break;
+            }
+            if (isHeader[index]) {
+                group.push(fragmentsFor[index] ?? []);
+            }
+        }
+        rows.push(mergeHeader(group));
+    }
+    // Restricted to lines below the header for the same reason as in
+    // `tableLines`: dates in the statement preamble are not transactions.
+    const anchorIndexes = isAnchor
+        .map((anchor, index) => (anchor ? index : -1))
+        .filter(index => index >= 0 && (firstHeader < 0 || index > firstHeader));
+    // Which transaction each continuation line belongs to. Lines left unowned
+    // are discarded, which is what removes the statement preamble.
+    const owner = new Map();
+    for (const [position, anchorIndex] of anchorIndexes.entries()) {
+        const previousAnchor = anchorIndexes[position - 1];
+        const start = previousAnchor === undefined ? 0 : previousAnchor + 1;
+        const anchor = lines[anchorIndex];
+        if (!anchor) {
+            continue;
+        }
+        for (let index = start; index < anchorIndex; index += 1) {
+            const line = lines[index];
+            if (!line || isHeader[index] || isSummaryLine(line)) {
+                continue;
+            }
+            const previous = previousAnchor === undefined ? undefined : lines[previousAnchor];
+            const gap = line.y - anchor.y;
+            // Requiring the same page, and a small *positive* gap, is what keeps a
+            // page footer off the next page's first transaction: across a page
+            // boundary y restarts, making the gap wildly negative and so trivially
+            // "just above".
+            if (line.page === anchor.page && gap >= 0 && gap <= ATTACH_ABOVE) {
+                owner.set(index, anchorIndex);
+            }
+            else if (previousAnchor !== undefined &&
+                previous &&
+                previous.page === line.page &&
+                previous.y - line.y <= MAX_CONTINUATION_BELOW) {
+                owner.set(index, previousAnchor);
+            }
+            // Otherwise it belongs to no transaction on this page — a preamble,
+            // page header or footer — and is dropped.
+        }
+    }
+    // Lines after the last anchor continue it, but only on its own page and
+    // only while they are close enough to be part of its cell.
+    const lastAnchor = anchorIndexes[anchorIndexes.length - 1];
+    const lastAnchorLine = lastAnchor === undefined ? undefined : lines[lastAnchor];
+    if (lastAnchor !== undefined && lastAnchorLine) {
+        for (let index = lastAnchor + 1; index < lines.length; index += 1) {
+            const line = lines[index];
+            if (!isHeader[index] &&
+                line &&
+                !isSummaryLine(line) &&
+                line.page === lastAnchorLine.page &&
+                lastAnchorLine.y - line.y <= MAX_CONTINUATION_BELOW) {
+                owner.set(index, lastAnchor);
+            }
+        }
+    }
+    const continuations = new Map();
+    for (const [index, anchorIndex] of owner) {
+        const list = continuations.get(anchorIndex) ?? [];
+        list.push(index);
+        continuations.set(anchorIndex, list);
+    }
+    for (const anchorIndex of anchorIndexes) {
+        // Document order, so wrapped text reassembles in reading order.
+        const lineIndexes = [
+            anchorIndex,
+            ...(continuations.get(anchorIndex) ?? []).filter(index => index !== anchorIndex),
+        ].sort((a, b) => a - b);
+        // Split by position relative to the anchor. ICICI prints a descriptor
+        // above each row (`Debit trxn`, `NACH trxn`) that is *not* part of the
+        // bank's narration — the source spreadsheet contains no occurrence of
+        // "trxn" at all. Where such a line does carry real text, that text also
+        // appears in the wrapped narration below. So below-anchor content wins,
+        // and the line above is used only when there is nothing below it.
+        const main = new Map();
+        const above = new Map();
+        for (const index of lineIndexes) {
+            const target = index < anchorIndex ? above : main;
+            for (const [column, fragment] of (fragmentsFor[index] ?? []).entries()) {
+                if (!fragment) {
+                    continue;
+                }
+                const list = target.get(column) ?? [];
+                list.push(fragment);
+                target.set(column, list);
+            }
+        }
+        const cells = [];
+        for (const [column, fragments] of main) {
+            cells[column] = joinFragments(fragments, widths[column] ?? 0);
+        }
+        for (const [column, fragments] of above) {
+            if (!cells[column]) {
+                cells[column] = joinFragments(fragments, widths[column] ?? 0);
+            }
+        }
+        for (let index = 0; index < cells.length; index += 1) {
+            cells[index] ??= '';
+        }
+        rows.push(cells);
+    }
+    return rows;
+}
+export async function extractPdf(path, options = {}) {
+    const { lines, pages } = await readLines(path, options);
+    const bands = inferBands(lines);
+    return {
+        rows: assembleRows(lines, bands),
+        source: { path, format: 'pdf', part: `${pages} page(s)` },
+    };
+}
+//# sourceMappingURL=pdf.js.map

package/dist/extract/pdf.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf.js","sourceRoot":"","sources":["../../src/extract/pdf.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,iCAAiC,CAAC;AAE9D,OAAO,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAC;AACzD,OAAO,EAAE,kBAAkB,EAAE,MAAM,wBAAwB,CAAC;AAkB5D;;;GAGG;AACH,MAAM,cAAc,GAAG,CAAC,CAAC;AAEzB;;;;;;GAMG;AACH,MAAM,YAAY,GAAG,CAAC,CAAC;AAEvB;;;GAGG;AACH,MAAM,iBAAiB,GAAG,CAAC,CAAC;AAE5B;;;;;;GAMG;AACH,MAAM,sBAAsB,GAAG,EAAE,CAAC;AAElC,KAAK,UAAU,SAAS,CACtB,IAAY,EACZ,OAA0B;IAE1B,MAAM,GAAG,GAAG,MAAM,WAAW,CAAC;QAC5B,GAAG,EAAE,IAAI;QACT,cAAc,EAAE,IAAI;QACpB,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,OAAO,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;KAC5D,CAAC,CAAC,OAAO,CAAC;IAEX,MAAM,KAAK,GAAW,EAAE,CAAC;IAEzB,KAAK,IAAI,UAAU,GAAG,CAAC,EAAE,UAAU,IAAI,GAAG,CAAC,QAAQ,EAAE,UAAU,IAAI,CAAC,EAAE,CAAC;QACrE,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC;QAC3C,MAAM,OAAO,GAAG,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;QAE5C,MAAM,KAAK,GAAW,EAAE,CAAC;QACzB,KAAK,MAAM,GAAG,IAAI,OAAO,CAAC,KAAK,EAAE,CAAC;YAChC,IAAI,CAAC,CAAC,KAAK,IAAI,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,EAAE,CAAC;gBACvC,SAAS;YACX,CAAC;YACD,MAAM,CAAC,GAAG,GAAG,CAAC,SAAS,CAAC,CAAC,CAAW,CAAC;YACrC,MAAM,KAAK,GAAG,OAAO,GAAG,CAAC,KAAK,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YAC5D,KAAK,CAAC,IAAI,CAAC;gBACT,CAAC;gBACD,KAAK,EAAE,CAAC,GAAG,KAAK;gBAChB,CAAC,EAAE,GAAG,CAAC,SAAS,CAAC,CAAC,CAAW;gBAC7B,IAAI,EAAE,GAAG,CAAC,GAAG;aACd,CAAC,CAAC;QACL,CAAC;QAED,qEAAqE;QACrE,oCAAoC;QACpC,KAAK,CAAC,IAAI,CAAC,GAAG,YAAY,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC,CAAC;IACjD,CAAC;IAED,OAAO,EAAE,KAAK,EAAE,KAAK,EAAE,GAAG,CAAC,QAAQ,EAAE,CAAC;AACxC,CAAC;AAED,iDAAiD;AACjD,SAAS,YAAY,CAAC,KAAa,EAAE,IAAY;IAC/C,MAAM,MAAM,GAAG,CAAC,GAAG,KAAK,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IACpD,MAAM,KAAK,GAAW,EAAE,CAAC;IAEzB,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC;QAC1B,MAAM,OAAO,GAAG,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;QACxC,IAAI,OAAO,IAAI,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,CAAC,IAAI,cAAc,EAAE,CAAC;YAC9D,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAC3B,CAAC;aAAM,CAAC;YACN,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,IAAI,CAAC,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACjD,CAAC;IACH,CAAC;IAED,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IACvC,CAAC;IAED,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;;;;GAKG;AACH,SAAS,YAAY,CAAC,IAAU;IAC9B,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,IAAI,CAAC,CAAC;AACzE,CAAC;AAED;;;;;;;;;GASG;AACH,SAAS,aAAa,CAAC,IAAU;IAC/B,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK;SACpB,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC;SACtB,IAAI,CAAC,GAAG,CAAC;SACT,IAAI,EAAE,CAAC;IAEV,OAAO,yHAAyH,CAAC,IAAI,CACnI,IAAI,CACL,CAAC;AACJ,CAAC;AAED,+CAA+C;AAC/C,SAAS,YAAY,CAAC,IAAU;IAC9B,IAAI,YAAY,CAAC,IAAI,CAAC,IAAI,aAAa,CAAC,IAAI,CAAC,EAAE,CAAC;QAC9C,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,CACL,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,KAAK,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,CACzE,CAAC;AACJ,CAAC;AAKD;;;;;GAKG;AACH,MAAM,UAAU,GAAG,CAAC,CAAC;AAErB,iEAAiE;AACjE,SAAS,gBAAgB,CAAC,KAAa;IACrC,OAAO,KAAK,CAAC,SAAS,CAAC,YAAY,CAAC,CAAC;AACvC,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,UAAU,CAAC,KAAa;IAC/B,MAAM,MAAM,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAC;IAEvC,OAAO,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE;QAClC,IAAI,YAAY,CAAC,IAAI,CAAC,EAAE,CAAC;YACvB,OAAO,IAAI,CAAC;QACd,CAAC;QACD,OAAO,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,GAAG,MAAM,CAAC,CAAC;IAC9D,CAAC,CAAC,CAAC;AACL,CAAC;AAED;;;;;;;;;;;;;;;;;;;GAmBG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACtC,MAAM,MAAM,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC;IACjC,MAAM,KAAK,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IAE3E,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;QAClB,OAAO,EAAE,CAAC;IACZ,CAAC;IAED,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC;IACjF,MAAM,SAAS,GAAG,IAAI,WAAW,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;IAE7C,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;QAC7C,MAAM,EAAE,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;QACxE,KAAK,IAAI,CAAC,GAAG,IAAI,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC;YAClC,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC;QACzC,CAAC;IACH,CAAC;IAED,MAAM,KAAK,GAAW,EAAE,CAAC;IACzB,IAAI,CAAC,GAAG,CAAC,CAAC;IAEV,OAAO,CAAC,GAAG,KAAK,EAAE,CAAC;QACjB,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC;YAClB,CAAC,IAAI,CAAC,CAAC;YACP,SAAS;QACX,CAAC;QAED,MAAM,IAAI,GAAG,CAAC,CAAC;QACf,IAAI,KAAK,GAAG,CAAC,CAAC;QAEd,OAAO,CAAC,GAAG,KAAK,EAAE,CAAC;YACjB,IAAI,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC;gBACjB,CAAC,IAAI,CAAC,CAAC;gBACP,KAAK,GAAG,CAAC,CAAC;gBACV,SAAS;YACX,CAAC;YAED,mDAAmD;YACnD,IAAI,MAAM,GAAG,CAAC,CAAC;YACf,OAAO,MAAM,GAAG,KAAK,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,EAAE,CAAC;gBAC5C,MAAM,IAAI,CAAC,CAAC;YACd,CAAC;YACD,IAAI,MAAM,GAAG,CAAC,GAAG,UAAU,IAAI,MAAM,GAAG,KAAK,EAAE,CAAC;gBAC9C,CAAC,GAAG,MAAM,CAAC;gBACX,SAAS;YACX,CAAC;YACD,MAAM;QACR,CAAC;QAED,KAAK,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;IAC9B,CAAC;IAED,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;;;;;;;;;GAUG;AACH,SAAS,MAAM,CAAC,IAAkC,EAAE,KAAa;IAC/D,IAAI,SAAS,GAAG,CAAC,CAAC,CAAC;IACnB,IAAI,WAAW,GAAG,CAAC,CAAC;IAEpB,KAAK,MAAM,CAAC,KAAK,EAAE,IAAI,CAAC,IAAI,KAAK,CAAC,OAAO,EAAE,EAAE,CAAC;QAC5C,MAAM,OAAO,GACX,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,CAAC,CAAC;QACjE,IAAI,OAAO,GAAG,WAAW,EAAE,CAAC;YAC1B,WAAW,GAAG,OAAO,CAAC;YACtB,SAAS,GAAG,KAAK,CAAC;QACpB,CAAC;IACH,CAAC;IAED,IAAI,SAAS,IAAI,CAAC,EAAE,CAAC;QACnB,OAAO,SAAS,CAAC;IACnB,CAAC;IAED,MAAM,UAAU,GAAG,KAAK,CAAC,SAAS,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC;IAChE,OAAO,UAAU,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;AACtE,CAAC;AAED,gDAAgD;AAChD,SAAS,WAAW,CAClB,IAAU,EACV,KAAa;IAEb,MAAM,SAAS,GAAgC,EAAE,CAAC;IAElD,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,KAAK,EAAE,CAAC;QAC9B,MAAM,MAAM,GAAG,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;QACnC,MAAM,QAAQ,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;QACnC,SAAS,CAAC,MAAM,CAAC,GAAG,QAAQ;YAC1B,CAAC,CAAC;gBACE,IAAI,EAAE,GAAG,QAAQ,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,EAAE;gBACrC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC;aAC5C;YACH,CAAC,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,IAAI,CAAC,KAAK,EAAE,CAAC;IAC7C,CAAC;IAED,OAAO,SAAS,CAAC;AACnB,CAAC;AAED;;;GAGG;AACH,SAAS,UAAU,CAAC,IAAwC;IAC1D,MAAM,MAAM,GAAa,EAAE,CAAC;IAE5B,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,KAAK,MAAM,CAAC,MAAM,EAAE,QAAQ,CAAC,IAAI,GAAG,CAAC,OAAO,EAAE,EAAE,CAAC;YAC/C,IAAI,CAAC,QAAQ,EAAE,CAAC;gBACd,SAAS;YACX,CAAC;YACD,MAAM,CAAC,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,EAAE,QAAQ,CAAC,KAAK,CAAC,CAAC;QACjE,CAAC;IACH,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED;;;;;;;;;;;;;;GAcG;AACH,SAAS,aAAa,CAAC,SAAqB,EAAE,SAAiB;IAC7D,IAAI,IAAI,GAAG,EAAE,CAAC;IACd,IAAI,QAA8B,CAAC;IAEnC,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;QACjC,IAAI,QAAQ,IAAI,QAAQ,CAAC,KAAK,GAAG,SAAS,GAAG,iBAAiB,EAAE,CAAC;YAC/D,IAAI,IAAI,GAAG,CAAC;QACd,CAAC;QACD,IAAI,IAAI,QAAQ,CAAC,IAAI,CAAC;QACtB,QAAQ,GAAG,QAAQ,CAAC;IACtB,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED,SAAS,WAAW,CAAC,IAAwC;IAC3D,MAAM,MAAM,GAAa,EAAE,CAAC;IAE5B,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,KAAK,MAAM,CAAC,MAAM,EAAE,QAAQ,CAAC,IAAI,GAAG,CAAC,OAAO,EAAE,EAAE,CAAC;YAC/C,IAAI,CAAC,QAAQ,EAAE,CAAC;gBACd,SAAS;YACX,CAAC;YACD,MAAM,CAAC,MAAM,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC;gBAC7B,CAAC,CAAC,GAAG,MAAM,CAAC,MAAM,CAAC,IAAI,QAAQ,CAAC,IAAI,EAAE;gBACtC,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC;QACpB,CAAC;IACH,CAAC;IAED,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,MAAM,CAAC,MAAM,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;QACtD,MAAM,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;IACvB,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,KAAa,EAAE,KAAa;IACvD,MAAM,YAAY,GAAG,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,WAAW,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC;IACjE,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC;IACzC,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC;IACzC,MAAM,MAAM,GAAG,UAAU,CAAC,YAAY,CAAC,CAAC;IAExC,MAAM,IAAI,GAAe,EAAE,CAAC;IAE5B,wEAAwE;IACxE,0EAA0E;IAC1E,+CAA+C;IAC/C,MAAM,WAAW,GAAG,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IAC3C,IAAI,WAAW,IAAI,CAAC,EAAE,CAAC;QACrB,MAAM,KAAK,GAAuC,EAAE,CAAC;QACrD,KAAK,IAAI,KAAK,GAAG,WAAW,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;YAC/D,IAAI,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;gBACpB,MAAM;YACR,CAAC;YACD,IAAI,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;gBACpB,KAAK,CAAC,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC;YACxC,CAAC;QACH,CAAC;QACD,IAAI,CAAC,IAAI,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC,CAAC;IAChC,CAAC;IAED,iEAAiE;IACjE,sEAAsE;IACtE,MAAM,aAAa,GAAG,QAAQ;SAC3B,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;SAC7C,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,IAAI,CAAC,IAAI,CAAC,WAAW,GAAG,CAAC,IAAI,KAAK,GAAG,WAAW,CAAC,CAAC,CAAC;IAE3E,0EAA0E;IAC1E,+DAA+D;IAC/D,MAAM,KAAK,GAAG,IAAI,GAAG,EAAkB,CAAC;IAExC,KAAK,MAAM,CAAC,QAAQ,EAAE,WAAW,CAAC,IAAI,aAAa,CAAC,OAAO,EAAE,EAAE,CAAC;QAC9D,MAAM,cAAc,GAAG,aAAa,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC;QACnD,MAAM,KAAK,GAAG,cAAc,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,cAAc,GAAG,CAAC,CAAC;QACpE,MAAM,MAAM,GAAG,KAAK,CAAC,WAAW,CAAC,CAAC;QAClC,IAAI,CAAC,MAAM,EAAE,CAAC;YACZ,SAAS;QACX,CAAC;QAED,KAAK,IAAI,KAAK,GAAG,KAAK,EAAE,KAAK,GAAG,WAAW,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;YACxD,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;YAC1B,IAAI,CAAC,IAAI,IAAI,QAAQ,CAAC,KAAK,CAAC,IAAI,aAAa,CAAC,IAAI,CAAC,EAAE,CAAC;gBACpD,SAAS;YACX,CAAC;YAED,MAAM,QAAQ,GACZ,cAAc,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,cAAc,CAAC,CAAC;YACnE,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,CAAC;YAE9B,uEAAuE;YACvE,mEAAmE;YACnE,uEAAuE;YACvE,gBAAgB;YAChB,IAAI,IAAI,CAAC,IAAI,KAAK,MAAM,CAAC,IAAI,IAAI,GAAG,IAAI,CAAC,IAAI,GAAG,IAAI,YAAY,EAAE,CAAC;gBACjE,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,WAAW,CAAC,CAAC;YAChC,CAAC;iBAAM,IACL,cAAc,KAAK,SAAS;gBAC5B,QAAQ;gBACR,QAAQ,CAAC,IAAI,KAAK,IAAI,CAAC,IAAI;gBAC3B,QAAQ,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,IAAI,sBAAsB,EAC7C,CAAC;gBACD,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,cAAc,CAAC,CAAC;YACnC,CAAC;YACD,oEAAoE;YACpE,0CAA0C;QAC5C,CAAC;IACH,CAAC;IAED,wEAAwE;IACxE,2DAA2D;IAC3D,MAAM,UAAU,GAAG,aAAa,CAAC,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAC3D,MAAM,cAAc,GAClB,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC;IAC3D,IAAI,UAAU,KAAK,SAAS,IAAI,cAAc,EAAE,CAAC;QAC/C,KAAK,IAAI,KAAK,GAAG,UAAU,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;YAClE,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;YAC1B,IACE,CAAC,QAAQ,CAAC,KAAK,CAAC;gBAChB,IAAI;gBACJ,CAAC,aAAa,CAAC,IAAI,CAAC;gBACpB,IAAI,CAAC,IAAI,KAAK,cAAc,CAAC,IAAI;gBACjC,cAAc,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,IAAI,sBAAsB,EACnD,CAAC;gBACD,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC;YAC/B,CAAC;QACH,CAAC;IACH,CAAC;IAED,MAAM,aAAa,GAAG,IAAI,GAAG,EAAoB,CAAC;IAClD,KAAK,MAAM,CAAC,KAAK,EAAE,WAAW,CAAC,IAAI,KAAK,EAAE,CAAC;QACzC,MAAM,IAAI,GAAG,aAAa,CAAC,GAAG,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC;QAClD,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QACjB,aAAa,CAAC,GAAG,CAAC,WAAW,EAAE,IAAI,CAAC,CAAC;IACvC,CAAC;IAED,KAAK,MAAM,WAAW,IAAI,aAAa,EAAE,CAAC;QACxC,gEAAgE;QAChE,MAAM,WAAW,GAAG;YAClB,WAAW;YACX,GAAG,CAAC,aAAa,CAAC,GAAG,CAAC,WAAW,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,CAC9C,KAAK,CAAC,EAAE,CAAC,KAAK,KAAK,WAAW,CAC/B;SACF,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;QAExB,sEAAsE;QACtE,uEAAuE;QACvE,sEAAsE;QACtE,wEAAwE;QACxE,wEAAwE;QACxE,kEAAkE;QAClE,MAAM,IAAI,GAAG,IAAI,GAAG,EAAsB,CAAC;QAC3C,MAAM,KAAK,GAAG,IAAI,GAAG,EAAsB,CAAC;QAE5C,KAAK,MAAM,KAAK,IAAI,WAAW,EAAE,CAAC;YAChC,MAAM,MAAM,GAAG,KAAK,GAAG,WAAW,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC;YAClD,KAAK,MAAM,CAAC,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,YAAY,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC;gBACvE,IAAI,CAAC,QAAQ,EAAE,CAAC;oBACd,SAAS;gBACX,CAAC;gBACD,MAAM,IAAI,GAAG,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;gBACtC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;gBACpB,MAAM,CAAC,GAAG,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC;YAC3B,CAAC;QACH,CAAC;QAED,MAAM,KAAK,GAAa,EAAE,CAAC;QAC3B,KAAK,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,IAAI,IAAI,EAAE,CAAC;YACvC,KAAK,CAAC,MAAM,CAAC,GAAG,aAAa,CAAC,SAAS,EAAE,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;QAChE,CAAC;QACD,KAAK,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,IAAI,KAAK,EAAE,CAAC;YACxC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC;gBACnB,KAAK,CAAC,MAAM,CAAC,GAAG,aAAa,CAAC,SAAS,EAAE,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;YAChE,CAAC;QACH,CAAC;QACD,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,IAAI,CAAC,EAAE,CAAC;YACrD,KAAK,CAAC,KAAK,CAAC,KAAK,EAAE,CAAC;QACtB,CAAC;QAED,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IACnB,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAC9B,IAAY,EACZ,UAA6B,EAAE;IAE/B,MAAM,EAAE,KAAK,EAAE,KAAK,EAAE,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IACxD,MAAM,KAAK,GAAG,UAAU,CAAC,KAAK,CAAC,CAAC;IAEhC,OAAO;QACL,IAAI,EAAE,YAAY,CAAC,KAAK,EAAE,KAAK,CAAC;QAChC,MAAM,EAAE,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,KAAK,UAAU,EAAE;KAC1D,CAAC;AACJ,CAAC"}

package/dist/extract/sniff.d.ts

@@ -0,0 +1,12 @@
+/**
+ * What a file actually is, as opposed to what its extension claims.
+ *
+ * This distinction is not pedantry for Indian bank statements: a file named
+ * `.xls` is very often an HTML table or a SpreadsheetML document rather than a
+ * real Excel workbook, because that is the cheapest thing for a bank's
+ * reporting layer to emit. Trusting the extension gets you a parser error on a
+ * perfectly readable file.
+ */
+export type DetectedFormat = 'xlsx' | 'html' | 'spreadsheetml' | 'biff' | 'pdf' | 'text';
+export declare function detectFormat(buffer: Buffer): DetectedFormat;
+export declare function describeFormat(format: DetectedFormat): string;

package/dist/extract/sniff.js

@@ -0,0 +1,57 @@
+/** OOXML/ZIP: `PK\x03\x04`. */
+const ZIP_MAGIC = [0x50, 0x4b, 0x03, 0x04];
+/** OLE2 compound document, used by legacy .xls and .doc. */
+const OLE2_MAGIC = [0xd0, 0xcf, 0x11, 0xe0, 0xa1, 0xb1, 0x1a, 0xe1];
+/** `%PDF-`. */
+const PDF_MAGIC = [0x25, 0x50, 0x44, 0x46, 0x2d];
+function startsWithBytes(buffer, magic) {
+    if (buffer.length < magic.length) {
+        return false;
+    }
+    return magic.every((byte, index) => buffer[index] === byte);
+}
+export function detectFormat(buffer) {
+    if (startsWithBytes(buffer, ZIP_MAGIC)) {
+        return 'xlsx';
+    }
+    if (startsWithBytes(buffer, OLE2_MAGIC)) {
+        return 'biff';
+    }
+    if (startsWithBytes(buffer, PDF_MAGIC)) {
+        return 'pdf';
+    }
+    // Only the head matters, and only as text. A binary file decoded this way
+    // produces junk that matches none of the patterns below, which is fine.
+    const head = buffer.subarray(0, 4096).toString('utf8').toLowerCase();
+    // Excel 2003 XML declares this namespace. Checked before the HTML test
+    // because such files can also contain table-ish markup.
+    if (head.includes('urn:schemas-microsoft-com:office:spreadsheet') ||
+        /<workbook[\s>]/.test(head)) {
+        return 'spreadsheetml';
+    }
+    if (/<table[\s>]/.test(head) ||
+        /<html[\s>]/.test(head) ||
+        head.includes('<!doctype html')) {
+        return 'html';
+    }
+    return 'text';
+}
+export function describeFormat(format) {
+    switch (format) {
+        case 'xlsx':
+            return 'Excel workbook (xlsx)';
+        case 'html':
+            return 'HTML table (a .xls file that is really HTML)';
+        case 'spreadsheetml':
+            return 'Excel 2003 XML (SpreadsheetML)';
+        case 'biff':
+            return 'legacy binary Excel (.xls, OLE2)';
+        case 'pdf':
+            return 'PDF';
+        case 'text':
+            return 'delimited text';
+        default:
+            return format;
+    }
+}
+//# sourceMappingURL=sniff.js.map

package/dist/extract/sniff.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sniff.js","sourceRoot":"","sources":["../../src/extract/sniff.ts"],"names":[],"mappings":"AAiBA,+BAA+B;AAC/B,MAAM,SAAS,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;AAE3C,4DAA4D;AAC5D,MAAM,UAAU,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;AAEpE,eAAe;AACf,MAAM,SAAS,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;AAEjD,SAAS,eAAe,CAAC,MAAc,EAAE,KAAe;IACtD,IAAI,MAAM,CAAC,MAAM,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC;QACjC,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC,KAAK,IAAI,CAAC,CAAC;AAC9D,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,MAAc;IACzC,IAAI,eAAe,CAAC,MAAM,EAAE,SAAS,CAAC,EAAE,CAAC;QACvC,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,IAAI,eAAe,CAAC,MAAM,EAAE,UAAU,CAAC,EAAE,CAAC;QACxC,OAAO,MAAM,CAAC;IAChB,CAAC;IACD,IAAI,eAAe,CAAC,MAAM,EAAE,SAAS,CAAC,EAAE,CAAC;QACvC,OAAO,KAAK,CAAC;IACf,CAAC;IAED,0EAA0E;IAC1E,wEAAwE;IACxE,MAAM,IAAI,GAAG,MAAM,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,WAAW,EAAE,CAAC;IAErE,uEAAuE;IACvE,wDAAwD;IACxD,IACE,IAAI,CAAC,QAAQ,CAAC,8CAA8C,CAAC;QAC7D,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,EAC3B,CAAC;QACD,OAAO,eAAe,CAAC;IACzB,CAAC;IAED,IACE,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC;QACxB,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC;QACvB,IAAI,CAAC,QAAQ,CAAC,gBAAgB,CAAC,EAC/B,CAAC;QACD,OAAO,MAAM,CAAC;IAChB,CAAC;IAED,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,MAAM,UAAU,cAAc,CAAC,MAAsB;IACnD,QAAQ,MAAM,EAAE,CAAC;QACf,KAAK,MAAM;YACT,OAAO,uBAAuB,CAAC;QACjC,KAAK,MAAM;YACT,OAAO,8CAA8C,CAAC;QACxD,KAAK,eAAe;YAClB,OAAO,gCAAgC,CAAC;QAC1C,KAAK,MAAM;YACT,OAAO,kCAAkC,CAAC;QAC5C,KAAK,KAAK;YACR,OAAO,KAAK,CAAC;QACf,KAAK,MAAM;YACT,OAAO,gBAAgB,CAAC;QAC1B;YACE,OAAO,MAAM,CAAC;IAClB,CAAC;AACH,CAAC"}

package/dist/extract/spreadsheetml.d.ts

@@ -0,0 +1,3 @@
+import type { Table } from './types.js';
+export declare function tableFromSpreadsheetMl(xml: string, path?: string): Table;
+export declare function extractSpreadsheetMl(path: string): Promise<Table>;

package/dist/extract/spreadsheetml.js

@@ -0,0 +1,93 @@
+import { readFile } from 'node:fs/promises';
+import { XMLParser } from 'fast-xml-parser';
+/**
+ * Excel 2003 XML (SpreadsheetML). Shape, once namespace prefixes are stripped:
+ *
+ *   <Workbook>
+ *     <Worksheet Name="Sheet1">
+ *       <Table>
+ *         <Row>
+ *           <Cell><Data Type="String">Date</Data></Cell>
+ *           <Cell Index="3"><Data Type="Number">450.5</Data></Cell>
+ *         </Row>
+ *
+ * `Index` is the load-bearing detail: it is a 1-based absolute column, emitted
+ * when a run of cells is empty. Ignoring it silently shifts every following
+ * value left, which reads amounts out of the wrong column — the kind of bug
+ * that produces plausible numbers rather than an error.
+ */
+const parser = new XMLParser({
+    ignoreAttributes: false,
+    attributeNamePrefix: '@_',
+    textNodeName: '#text',
+    // Strips `ss:` from both tags and attributes, so `ss:Index` -> `@_Index`.
+    removeNSPrefix: true,
+    // Keep everything as text; the interpreter handles coercion.
+    parseTagValue: false,
+    parseAttributeValue: false,
+    trimValues: true,
+});
+function toArray(value) {
+    if (value === undefined || value === null) {
+        return [];
+    }
+    return (Array.isArray(value) ? value : [value]);
+}
+function dataText(cell) {
+    const data = cell['Data'];
+    if (data === undefined || data === null) {
+        // Some writers put the value directly on the Cell.
+        const direct = cell['#text'];
+        return direct === undefined || direct === null ? '' : String(direct);
+    }
+    if (typeof data === 'object') {
+        const text = data['#text'];
+        return text === undefined || text === null ? '' : String(text);
+    }
+    return String(data);
+}
+function rowsOf(table) {
+    const rows = [];
+    for (const row of toArray(table['Row'])) {
+        const cells = [];
+        for (const cell of toArray(row['Cell'])) {
+            const index = Number.parseInt(String(cell['@_Index'] ?? ''), 10);
+            if (Number.isFinite(index) && index >= 1) {
+                // Pad out the columns this cell skipped over.
+                while (cells.length < index - 1) {
+                    cells.push('');
+                }
+            }
+            cells.push(dataText(cell).trim());
+        }
+        rows.push(cells);
+    }
+    return rows;
+}
+export function tableFromSpreadsheetMl(xml, path = 'inline') {
+    const parsed = parser.parse(xml);
+    const workbook = (parsed['Workbook'] ?? parsed);
+    let best = null;
+    for (const worksheet of toArray(workbook['Worksheet'])) {
+        const name = String(worksheet['@_Name'] ?? 'Sheet');
+        for (const table of toArray(worksheet['Table'])) {
+            const rows = rowsOf(table);
+            if (!best || rows.length > best.rows.length) {
+                best = { rows, name };
+            }
+        }
+    }
+    return {
+        rows: best?.rows ?? [],
+        source: {
+            path,
+            format: 'xlsx',
+            part: best ? `${best.name} (SpreadsheetML)` : 'SpreadsheetML',
+        },
+    };
+}
+export async function extractSpreadsheetMl(path) {
+    const contents = await readFile(path, 'utf8');
+    return tableFromSpreadsheetMl(contents, path);
+}
+//# sourceMappingURL=spreadsheetml.js.map

package/dist/extract/spreadsheetml.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"spreadsheetml.js","sourceRoot":"","sources":["../../src/extract/spreadsheetml.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,MAAM,kBAAkB,CAAC;AAE5C,OAAO,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AAI5C;;;;;;;;;;;;;;;GAeG;AACH,MAAM,MAAM,GAAG,IAAI,SAAS,CAAC;IAC3B,gBAAgB,EAAE,KAAK;IACvB,mBAAmB,EAAE,IAAI;IACzB,YAAY,EAAE,OAAO;IACrB,0EAA0E;IAC1E,cAAc,EAAE,IAAI;IACpB,6DAA6D;IAC7D,aAAa,EAAE,KAAK;IACpB,mBAAmB,EAAE,KAAK;IAC1B,UAAU,EAAE,IAAI;CACjB,CAAC,CAAC;AAIH,SAAS,OAAO,CAAC,KAAc;IAC7B,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;QAC1C,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,OAAO,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAW,CAAC;AAC5D,CAAC;AAED,SAAS,QAAQ,CAAC,IAAU;IAC1B,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC;IAE1B,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,IAAI,EAAE,CAAC;QACxC,mDAAmD;QACnD,MAAM,MAAM,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC;QAC7B,OAAO,MAAM,KAAK,SAAS,IAAI,MAAM,KAAK,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;IACvE,CAAC;IAED,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;QAC7B,MAAM,IAAI,GAAI,IAAa,CAAC,OAAO,CAAC,CAAC;QACrC,OAAO,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC;IACjE,CAAC;IAED,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC;AACtB,CAAC;AAED,SAAS,MAAM,CAAC,KAAW;IACzB,MAAM,IAAI,GAAe,EAAE,CAAC;IAE5B,KAAK,MAAM,GAAG,IAAI,OAAO,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;QACxC,MAAM,KAAK,GAAa,EAAE,CAAC;QAE3B,KAAK,MAAM,IAAI,IAAI,OAAO,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC;YACxC,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC;YACjE,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,IAAI,KAAK,IAAI,CAAC,EAAE,CAAC;gBACzC,8CAA8C;gBAC9C,OAAO,KAAK,CAAC,MAAM,GAAG,KAAK,GAAG,CAAC,EAAE,CAAC;oBAChC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;gBACjB,CAAC;YACH,CAAC;YACD,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;QACpC,CAAC;QAED,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IACnB,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED,MAAM,UAAU,sBAAsB,CAAC,GAAW,EAAE,IAAI,GAAG,QAAQ;IACjE,MAAM,MAAM,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAS,CAAC;IACzC,MAAM,QAAQ,GAAG,CAAC,MAAM,CAAC,UAAU,CAAC,IAAI,MAAM,CAAS,CAAC;IAExD,IAAI,IAAI,GAA8C,IAAI,CAAC;IAE3D,KAAK,MAAM,SAAS,IAAI,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC,EAAE,CAAC;QACvD,MAAM,IAAI,GAAG,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,OAAO,CAAC,CAAC;QACpD,KAAK,MAAM,KAAK,IAAI,OAAO,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,EAAE,CAAC;YAChD,MAAM,IAAI,GAAG,MAAM,CAAC,KAAK,CAAC,CAAC;YAC3B,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC;gBAC5C,IAAI,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC;YACxB,CAAC;QACH,CAAC;IACH,CAAC;IAED,OAAO;QACL,IAAI,EAAE,IAAI,EAAE,IAAI,IAAI,EAAE;QACtB,MAAM,EAAE;YACN,IAAI;YACJ,MAAM,EAAE,MAAM;YACd,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,IAAI,kBAAkB,CAAC,CAAC,CAAC,eAAe;SAC9D;KACF,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oBAAoB,CAAC,IAAY;IACrD,MAAM,QAAQ,GAAG,MAAM,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;IAC9C,OAAO,sBAAsB,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;AAChD,CAAC"}

package/dist/extract/types.d.ts

@@ -0,0 +1,20 @@
+export type SourceFormat = 'csv' | 'xlsx' | 'pdf';
+/**
+ * The single boundary between extraction and interpretation.
+ *
+ * Every extractor — CSV, XLS/XLSX, PDF — produces this and nothing else, and
+ * the interpreter never sees anything else. That is what keeps a bank's column
+ * logic written once even when the same bank ships two formats.
+ *
+ * Cells are always strings, exactly as they appeared. Type coercion (dates,
+ * amounts) belongs to interpretation, not extraction.
+ */
+export type Table = {
+    rows: string[][];
+    source: {
+        path: string;
+        format: SourceFormat;
+        /** Sheet name (spreadsheets) or page number (PDF), when meaningful. */
+        part?: string;
+    };
+};

package/dist/extract/types.js

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=types.js.map