@gmod/tabix 3.5.3 → 3.5.4

package/esm/tbi.d.ts

@@ -1,12 +1,25 @@
-import Chunk from './chunk.ts';
 import IndexFile from './indexFile.ts';
 import type { Options, RefIndex } from './indexFile.ts';
 import type VirtualOffset from './virtualOffset.ts';
 export default class TabixIndex extends IndexFile {
+    /**
+     * The bins that may overlap [beg, end) (zero-based half-open). TBI's scheme
+     * is CSI's with minShift/depth fixed, so this is CSI's reg2bins with the
+     * levels unrolled and its float shifts replaced by int32 ones.
+     */
+    protected reg2bins(beg: number, end: number): (readonly [number, number])[];
+    /**
+     * The linear index is monotonically non-decreasing, so the minimum virtual
+     * offset for chunks that could overlap [beg, ...) is its entry for beg.
+     * SYNC: ~/src/gmod/bam-js/src/bai.ts getLowestChunk
+     */
+    protected lowestOffset(ref: RefIndex, beg: number): VirtualOffset | undefined;
     /** @internal */
     _parse(opts?: Options): Promise<{
         indices: (refId: number) => RefIndex | undefined;
         metaChar: string | undefined;
+        minShift: number;
+        depth: number;
         maxBinNumber: number;
         maxRefLength: number;
         skipLines: number;
@@ -22,5 +35,4 @@ export default class TabixIndex extends IndexFile {
         refNameToId: Record<string, number>;
         maxBlockSize: number;
     }>;
-    blocksForRange(refName: string, min: number, max: number, opts?: Options): Promise<Chunk[]>;
 }

package/esm/tbi.js

@@ -1,36 +1,48 @@
-import { unzip } from '@gmod/bgzf-filehandle';
 import Chunk from "./chunk.js";
 import IndexFile from "./indexFile.js";
-import { clampChunkEnds, memoizeByRefId, minVirtualOffset, optimizeChunks, parseAuxData, parsePseudoBin, } from "./util.js";
+import { clampChunkEnds, memoizeByRefId, minVirtualOffset, parseAuxData, parsePseudoBin, } from "./util.js";
 import { fromBytes } from "./virtualOffset.js";
 const TBI_MAGIC = 21_578_324; // TBI\1
 // TBI is always the fixed depth-5, 14-bit-leaf binning scheme
 const TBI_DEPTH = 5;
-const TBI_MAX_REF_LENGTH = 2 ** (14 + TBI_DEPTH * 3);
-/**
- * calculate the list of bins that may overlap with region [beg,end)
- * (zero-based half-open)
- */
-function reg2bins(beg, end) {
-    // Clamp into the binning scheme's range. Past 2**31 the shifts below
-    // truncate to int32 and go negative, which would silently yield no bins.
-    const b = Math.min(Math.max(beg, 0), TBI_MAX_REF_LENGTH);
-    const e = Math.min(end, TBI_MAX_REF_LENGTH) - 1;
-    const bins = [];
-    if (b <= e) {
-        bins.push([0, 0], [1 + (b >> 26), 1 + (e >> 26)], [9 + (b >> 23), 9 + (e >> 23)], [73 + (b >> 20), 73 + (e >> 20)], [585 + (b >> 17), 585 + (e >> 17)], [4681 + (b >> 14), 4681 + (e >> 14)]);
-    }
-    return bins;
+const TBI_MIN_SHIFT = 14;
+const TBI_MAX_REF_LENGTH = 2 ** (TBI_MIN_SHIFT + TBI_DEPTH * 3);
+// Clamp a query coordinate into the binning scheme's range. Past 2**31 the
+// int32 shifts below truncate and go negative, which would silently yield no
+// bins.
+function clampToRefLength(pos) {
+    return Math.min(Math.max(pos, 0), TBI_MAX_REF_LENGTH);
 }
 export default class TabixIndex extends IndexFile {
+    /**
+     * The bins that may overlap [beg, end) (zero-based half-open). TBI's scheme
+     * is CSI's with minShift/depth fixed, so this is CSI's reg2bins with the
+     * levels unrolled and its float shifts replaced by int32 ones.
+     */
+    reg2bins(beg, end) {
+        if (beg > TBI_MAX_REF_LENGTH) {
+            console.warn('querying outside of possible tabix range');
+        }
+        const b = clampToRefLength(beg);
+        const e = Math.min(end, TBI_MAX_REF_LENGTH) - 1;
+        const bins = [];
+        if (b <= e) {
+            bins.push([0, 0], [1 + (b >> 26), 1 + (e >> 26)], [9 + (b >> 23), 9 + (e >> 23)], [73 + (b >> 20), 73 + (e >> 20)], [585 + (b >> 17), 585 + (e >> 17)], [4681 + (b >> 14), 4681 + (e >> 14)]);
+        }
+        return bins;
+    }
+    /**
+     * The linear index is monotonically non-decreasing, so the minimum virtual
+     * offset for chunks that could overlap [beg, ...) is its entry for beg.
+     * SYNC: ~/src/gmod/bam-js/src/bai.ts getLowestChunk
+     */
+    lowestOffset(ref, beg) {
+        const linearIndex = ref.linearIndex;
+        return linearIndex?.[Math.min(clampToRefLength(beg) >> TBI_MIN_SHIFT, linearIndex.length - 1)];
+    }
     /** @internal */
     async _parse(opts = {}) {
-        const buf = await this.filehandle.readFile({
-            signal: opts.signal,
-            onProgress: opts.onProgress,
-        });
-        const bytes = (await unzip(buf));
-        const dataView = new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength);
+        const { bytes, dataView } = await this.readIndexBytes(opts);
         if (dataView.getUint32(0, true) !== TBI_MAGIC) {
             throw new Error('Not a TBI file');
         }
@@ -119,6 +131,8 @@ export default class TabixIndex extends IndexFile {
         return {
             indices: memoizeByRefId(getIndices),
             metaChar,
+            minShift: TBI_MIN_SHIFT,
+            depth: TBI_DEPTH,
             maxBinNumber,
             maxRefLength: TBI_MAX_REF_LENGTH,
             skipLines,
@@ -131,41 +145,5 @@ export default class TabixIndex extends IndexFile {
             maxBlockSize: 1 << 16,
         };
     }
-    async blocksForRange(refName, min, max, opts = {}) {
-        const indexData = await this.parse(opts);
-        const refId = indexData.refNameToId[refName];
-        if (refId === undefined) {
-            return [];
-        }
-        const ba = indexData.indices(refId);
-        if (!ba) {
-            return [];
-        }
-        if (min > TBI_MAX_REF_LENGTH) {
-            console.warn('querying outside of possible tabix range');
-        }
-        // clamp before the >> 14 below, which would truncate to int32 and go
-        // negative past 2**31
-        const beg = Math.min(Math.max(min, 0), TBI_MAX_REF_LENGTH);
-        const overlappingBins = reg2bins(min, max); // List of bin #s that overlap min, max
-        const chunks = [];
-        // Find chunks in overlapping bins.  Leaf bins (< 4681) are not pruned
-        for (const [start, end] of overlappingBins) {
-            for (let bin = start; bin <= end; bin++) {
-                const binChunks = ba.binIndex[bin];
-                if (binChunks) {
-                    for (const c of binChunks) {
-                        chunks.push(c);
-                    }
-                }
-            }
-        }
-        // The linear index is monotonically non-decreasing, so the minimum virtual
-        // offset for chunks that could overlap [min, ...) is at index minLin.
-        // SYNC: ~/src/gmod/bam-js/src/bai.ts getLowestChunk
-        const linearIndex = ba.linearIndex;
-        const lowest = linearIndex?.[Math.min(beg >> 14, linearIndex.length - 1)];
-        return optimizeChunks(chunks, lowest);
-    }
 }
 //# sourceMappingURL=tbi.js.map

package/esm/tbi.js.map

@@ -1 +1 @@
-{"version":3,"file":"tbi.js","sourceRoot":"","sources":["../src/tbi.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,EAAE,MAAM,uBAAuB,CAAA;AAE7C,OAAO,KAAK,MAAM,YAAY,CAAA;AAC9B,OAAO,SAAS,MAAM,gBAAgB,CAAA;AACtC,OAAO,EACL,cAAc,EACd,cAAc,EACd,gBAAgB,EAChB,cAAc,EACd,YAAY,EACZ,cAAc,GACf,MAAM,WAAW,CAAA;AAClB,OAAO,EAAE,SAAS,EAAE,MAAM,oBAAoB,CAAA;AAK9C,MAAM,SAAS,GAAG,UAAU,CAAA,CAAC,QAAQ;AAErC,8DAA8D;AAC9D,MAAM,SAAS,GAAG,CAAC,CAAA;AACnB,MAAM,kBAAkB,GAAG,CAAC,IAAI,CAAC,EAAE,GAAG,SAAS,GAAG,CAAC,CAAC,CAAA;AAEpD;;;GAGG;AACH,SAAS,QAAQ,CAAC,GAAW,EAAE,GAAW;IACxC,qEAAqE;IACrE,yEAAyE;IACzE,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAA;IACxD,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,kBAAkB,CAAC,GAAG,CAAC,CAAA;IAC/C,MAAM,IAAI,GAAkC,EAAE,CAAA;IAC9C,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;QACX,IAAI,CAAC,IAAI,CACP,CAAC,CAAC,EAAE,CAAC,CAAC,EACN,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAC9B,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAC9B,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAChC,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAClC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CACrC,CAAA;IACH,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED,MAAM,CAAC,OAAO,OAAO,UAAW,SAAQ,SAAS;IAC/C,gBAAgB;IAChB,KAAK,CAAC,MAAM,CAAC,OAAgB,EAAE;QAC7B,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,UAAU,CAAC,QAAQ,CAAC;YACzC,MAAM,EAAE,IAAI,CAAC,MAAM;YACnB,UAAU,EAAE,IAAI,CAAC,UAAU;SAC5B,CAAC,CAAA;QACF,MAAM,KAAK,GAAG,CAAC,MAAM,KAAK,CAAC,GAAG,CAAC,CAAe,CAAA;QAC9C,MAAM,QAAQ,GAAG,IAAI,QAAQ,CAC3B,KAAK,CAAC,MAAM,EACZ,KAAK,CAAC,UAAU,EAChB,KAAK,CAAC,UAAU,CACjB,CAAA;QAED,IAAI,QAAQ,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,SAAS,EAAE,CAAC;YAC9C,MAAM,IAAI,KAAK,CAAC,gBAAgB,CAAC,CAAA;QACnC,CAAC;QAED,MAAM,QAAQ,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,CAAA;QAC5C,MAAM,YAAY,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,SAAS,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAA;QAE3D,oEAAoE;QACpE,MAAM,EACJ,WAAW,EACX,WAAW,EACX,cAAc,EACd,MAAM,EACN,aAAa,EACb,QAAQ,EACR,SAAS,GACV,GAAG,YAAY,CAAC,KAAK,EAAE,CAAC,CAAC,CAAA;QAC1B,+EAA+E;QAC/E,MAAM,iBAAiB,GAAG,QAAQ,CAAC,QAAQ,CAAC,EAAE,EAAE,IAAI,CAAC,CAAA;QAErD,iEAAiE;QACjE,oEAAoE;QACpE,EAAE;QACF,oEAAoE;QACpE,wEAAwE;QACxE,2EAA2E;QAC3E,0EAA0E;QAC1E,qEAAqE;QACrE,gBAAgB;QAChB,IAAI,IAAI,GAAG,EAAE,GAAG,iBAAiB,CAAA;QACjC,IAAI,aAAwC,CAAA;QAC5C,MAAM,OAAO,GAAa,EAAE,CAAA;QAE5B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;YAClC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAClB,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;YAC9C,IAAI,IAAI,CAAC,CAAA;YACT,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;gBAClC,MAAM,GAAG,GAAG,QAAQ,CAAC,SAAS,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;gBAC1C,IAAI,IAAI,CAAC,CAAA;gBACT,MAAM,UAAU,GAAG,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;gBAChD,IAAI,IAAI,CAAC,CAAA;gBACT,IAAI,GAAG,GAAG,YAAY,GAAG,CAAC,EAAE,CAAC;oBAC3B,MAAM,IAAI,KAAK,CACb,4DAA4D,CAC7D,CAAA;gBACH,CAAC;gBACD,IAAI,IAAI,EAAE,GAAG,UAAU,CAAA;YACzB,CAAC;YACD,MAAM,WAAW,GAAG,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;YACjD,IAAI,IAAI,CAAC,CAAA;YACT,aAAa,GAAG,gBAAgB,CAAC,KAAK,EAAE,IAAI,EAAE,WAAW,EAAE,aAAa,CAAC,CAAA;YACzE,IAAI,IAAI,CAAC,GAAG,WAAW,CAAA;QACzB,CAAC;QAED,SAAS,UAAU,CAAC,KAAa;YAC/B,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAA;YAC5B,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;gBACxB,OAAO,SAAS,CAAA;YAClB,CAAC;YACD,IAAI,GAAG,GAAG,KAAK,CAAA;YACf,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YAC7C,GAAG,IAAI,CAAC,CAAA;YACR,MAAM,QAAQ,GAA4B,EAAE,CAAA;YAC5C,IAAI,KAAK,CAAA;YACT,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;gBAClC,MAAM,GAAG,GAAG,QAAQ,CAAC,SAAS,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;gBACzC,GAAG,IAAI,CAAC,CAAA;gBACR,IAAI,GAAG,GAAG,YAAY,GAAG,CAAC,EAAE,CAAC;oBAC3B,MAAM,IAAI,KAAK,CACb,4DAA4D,CAC7D,CAAA;gBACH,CAAC;qBAAM,IAAI,GAAG,KAAK,YAAY,GAAG,CAAC,EAAE,CAAC;oBACpC,MAAM,UAAU,GAAG,QAAQ,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;oBAC/C,GAAG,IAAI,CAAC,CAAA;oBACR,IAAI,UAAU,KAAK,CAAC,EAAE,CAAC;wBACrB,KAAK,GAAG,cAAc,CAAC,KAAK,EAAE,GAAG,GAAG,EAAE,CAAC,CAAA;oBACzC,CAAC;oBACD,GAAG,IAAI,EAAE,GAAG,UAAU,CAAA;gBACxB,CAAC;qBAAM,CAAC;oBACN,MAAM,UAAU,GAAG,QAAQ,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;oBAC/C,GAAG,IAAI,CAAC,CAAA;oBACR,MAAM,MAAM,GAAG,KAAK,CAAC,IAAI,CAAQ,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC,CAAA;oBACxD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;wBACpC,MAAM,CAAC,CAAC,CAAC,GAAG,IAAI,KAAK,CACnB,SAAS,CAAC,KAAK,EAAE,GAAG,CAAC,EACrB,SAAS,CAAC,KAAK,EAAE,GAAG,GAAG,CAAC,CAAC,EACzB,GAAG,CACJ,CAAA;wBACD,GAAG,IAAI,EAAE,CAAA;oBACX,CAAC;oBACD,QAAQ,CAAC,GAAG,CAAC,GAAG,MAAM,CAAA;gBACxB,CAAC;YACH,CAAC;YACD,MAAM,WAAW,GAAG,QAAQ,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YAChD,GAAG,IAAI,CAAC,CAAA;YACR,MAAM,WAAW,GAAG,KAAK,CAAC,IAAI,CAAgB,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC,CAAA;YACtE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,WAAW,EAAE,CAAC,EAAE,EAAE,CAAC;gBACrC,WAAW,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,KAAK,EAAE,GAAG,CAAC,CAAA;gBACtC,GAAG,IAAI,CAAC,CAAA;YACV,CAAC;YACD,cAAc,CACZ,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,IAAI,EAAE,EAC9B,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,aAAa,CAAC,CACtC,CAAA;YACD,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,KAAK,EAAE,CAAA;QACzC,CAAC;QAED,OAAO;YACL,OAAO,EAAE,cAAc,CAAC,UAAU,CAAC;YACnC,QAAQ;YACR,YAAY;YACZ,YAAY,EAAE,kBAAkB;YAChC,SAAS;YACT,aAAa;YACb,aAAa;YACb,cAAc;YACd,MAAM;YACN,WAAW;YACX,WAAW;YACX,YAAY,EAAE,CAAC,IAAI,EAAE;SACtB,CAAA;IACH,CAAC;IAED,KAAK,CAAC,cAAc,CAClB,OAAe,EACf,GAAW,EACX,GAAW,EACX,OAAgB,EAAE;QAElB,MAAM,SAAS,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,KAAK,GAAG,SAAS,CAAC,WAAW,CAAC,OAAO,CAAC,CAAA;QAC5C,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;YACxB,OAAO,EAAE,CAAA;QACX,CAAC;QACD,MAAM,EAAE,GAAG,SAAS,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;QACnC,IAAI,CAAC,EAAE,EAAE,CAAC;YACR,OAAO,EAAE,CAAA;QACX,CAAC;QAED,IAAI,GAAG,GAAG,kBAAkB,EAAE,CAAC;YAC7B,OAAO,CAAC,IAAI,CAAC,0CAA0C,CAAC,CAAA;QAC1D,CAAC;QACD,qEAAqE;QACrE,sBAAsB;QACtB,MAAM,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAA;QAC1D,MAAM,eAAe,GAAG,QAAQ,CAAC,GAAG,EAAE,GAAG,CAAC,CAAA,CAAC,uCAAuC;QAClF,MAAM,MAAM,GAAY,EAAE,CAAA;QAE1B,sEAAsE;QACtE,KAAK,MAAM,CAAC,KAAK,EAAE,GAAG,CAAC,IAAI,eAAe,EAAE,CAAC;YAC3C,KAAK,IAAI,GAAG,GAAG,KAAK,EAAE,GAAG,IAAI,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC;gBACxC,MAAM,SAAS,GAAG,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAA;gBAClC,IAAI,SAAS,EAAE,CAAC;oBACd,KAAK,MAAM,CAAC,IAAI,SAAS,EAAE,CAAC;wBAC1B,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;oBAChB,CAAC;gBACH,CAAC;YACH,CAAC;QACH,CAAC;QAED,2EAA2E;QAC3E,sEAAsE;QACtE,oDAAoD;QACpD,MAAM,WAAW,GAAG,EAAE,CAAC,WAAW,CAAA;QAClC,MAAM,MAAM,GAAG,WAAW,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,IAAI,EAAE,EAAE,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAA;QAEzE,OAAO,cAAc,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IACvC,CAAC;CACF"}
+{"version":3,"file":"tbi.js","sourceRoot":"","sources":["../src/tbi.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,YAAY,CAAA;AAC9B,OAAO,SAAS,MAAM,gBAAgB,CAAA;AACtC,OAAO,EACL,cAAc,EACd,cAAc,EACd,gBAAgB,EAChB,YAAY,EACZ,cAAc,GACf,MAAM,WAAW,CAAA;AAClB,OAAO,EAAE,SAAS,EAAE,MAAM,oBAAoB,CAAA;AAK9C,MAAM,SAAS,GAAG,UAAU,CAAA,CAAC,QAAQ;AAErC,8DAA8D;AAC9D,MAAM,SAAS,GAAG,CAAC,CAAA;AACnB,MAAM,aAAa,GAAG,EAAE,CAAA;AACxB,MAAM,kBAAkB,GAAG,CAAC,IAAI,CAAC,aAAa,GAAG,SAAS,GAAG,CAAC,CAAC,CAAA;AAE/D,2EAA2E;AAC3E,6EAA6E;AAC7E,QAAQ;AACR,SAAS,gBAAgB,CAAC,GAAW;IACnC,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAA;AACvD,CAAC;AAED,MAAM,CAAC,OAAO,OAAO,UAAW,SAAQ,SAAS;IAC/C;;;;OAIG;IACO,QAAQ,CAAC,GAAW,EAAE,GAAW;QACzC,IAAI,GAAG,GAAG,kBAAkB,EAAE,CAAC;YAC7B,OAAO,CAAC,IAAI,CAAC,0CAA0C,CAAC,CAAA;QAC1D,CAAC;QACD,MAAM,CAAC,GAAG,gBAAgB,CAAC,GAAG,CAAC,CAAA;QAC/B,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,kBAAkB,CAAC,GAAG,CAAC,CAAA;QAC/C,MAAM,IAAI,GAAkC,EAAE,CAAA;QAC9C,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACX,IAAI,CAAC,IAAI,CACP,CAAC,CAAC,EAAE,CAAC,CAAC,EACN,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAC9B,CAAC,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAC9B,CAAC,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAChC,CAAC,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,GAAG,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,EAClC,CAAC,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CACrC,CAAA;QACH,CAAC;QACD,OAAO,IAAI,CAAA;IACb,CAAC;IAED;;;;OAIG;IACO,YAAY,CAAC,GAAa,EAAE,GAAW;QAC/C,MAAM,WAAW,GAAG,GAAG,CAAC,WAAW,CAAA;QACnC,OAAO,WAAW,EAAE,CAClB,IAAI,CAAC,GAAG,CAAC,gBAAgB,CAAC,GAAG,CAAC,IAAI,aAAa,EAAE,WAAW,CAAC,MAAM,GAAG,CAAC,CAAC,CACzE,CAAA;IACH,CAAC;IAED,gBAAgB;IAChB,KAAK,CAAC,MAAM,CAAC,OAAgB,EAAE;QAC7B,MAAM,EAAE,KAAK,EAAE,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,CAAA;QAE3D,IAAI,QAAQ,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,SAAS,EAAE,CAAC;YAC9C,MAAM,IAAI,KAAK,CAAC,gBAAgB,CAAC,CAAA;QACnC,CAAC;QAED,MAAM,QAAQ,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,CAAA;QAC5C,MAAM,YAAY,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,SAAS,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAA;QAE3D,oEAAoE;QACpE,MAAM,EACJ,WAAW,EACX,WAAW,EACX,cAAc,EACd,MAAM,EACN,aAAa,EACb,QAAQ,EACR,SAAS,GACV,GAAG,YAAY,CAAC,KAAK,EAAE,CAAC,CAAC,CAAA;QAC1B,+EAA+E;QAC/E,MAAM,iBAAiB,GAAG,QAAQ,CAAC,QAAQ,CAAC,EAAE,EAAE,IAAI,CAAC,CAAA;QAErD,iEAAiE;QACjE,oEAAoE;QACpE,EAAE;QACF,oEAAoE;QACpE,wEAAwE;QACxE,2EAA2E;QAC3E,0EAA0E;QAC1E,qEAAqE;QACrE,gBAAgB;QAChB,IAAI,IAAI,GAAG,EAAE,GAAG,iBAAiB,CAAA;QACjC,IAAI,aAAwC,CAAA;QAC5C,MAAM,OAAO,GAAa,EAAE,CAAA;QAE5B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;YAClC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAClB,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;YAC9C,IAAI,IAAI,CAAC,CAAA;YACT,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;gBAClC,MAAM,GAAG,GAAG,QAAQ,CAAC,SAAS,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;gBAC1C,IAAI,IAAI,CAAC,CAAA;gBACT,MAAM,UAAU,GAAG,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;gBAChD,IAAI,IAAI,CAAC,CAAA;gBACT,IAAI,GAAG,GAAG,YAAY,GAAG,CAAC,EAAE,CAAC;oBAC3B,MAAM,IAAI,KAAK,CACb,4DAA4D,CAC7D,CAAA;gBACH,CAAC;gBACD,IAAI,IAAI,EAAE,GAAG,UAAU,CAAA;YACzB,CAAC;YACD,MAAM,WAAW,GAAG,QAAQ,CAAC,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;YACjD,IAAI,IAAI,CAAC,CAAA;YACT,aAAa,GAAG,gBAAgB,CAAC,KAAK,EAAE,IAAI,EAAE,WAAW,EAAE,aAAa,CAAC,CAAA;YACzE,IAAI,IAAI,CAAC,GAAG,WAAW,CAAA;QACzB,CAAC;QAED,SAAS,UAAU,CAAC,KAAa;YAC/B,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,CAAA;YAC5B,IAAI,KAAK,KAAK,SAAS,EAAE,CAAC;gBACxB,OAAO,SAAS,CAAA;YAClB,CAAC;YACD,IAAI,GAAG,GAAG,KAAK,CAAA;YACf,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YAC7C,GAAG,IAAI,CAAC,CAAA;YACR,MAAM,QAAQ,GAA4B,EAAE,CAAA;YAC5C,IAAI,KAAK,CAAA;YACT,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;gBAClC,MAAM,GAAG,GAAG,QAAQ,CAAC,SAAS,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;gBACzC,GAAG,IAAI,CAAC,CAAA;gBACR,IAAI,GAAG,GAAG,YAAY,GAAG,CAAC,EAAE,CAAC;oBAC3B,MAAM,IAAI,KAAK,CACb,4DAA4D,CAC7D,CAAA;gBACH,CAAC;qBAAM,IAAI,GAAG,KAAK,YAAY,GAAG,CAAC,EAAE,CAAC;oBACpC,MAAM,UAAU,GAAG,QAAQ,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;oBAC/C,GAAG,IAAI,CAAC,CAAA;oBACR,IAAI,UAAU,KAAK,CAAC,EAAE,CAAC;wBACrB,KAAK,GAAG,cAAc,CAAC,KAAK,EAAE,GAAG,GAAG,EAAE,CAAC,CAAA;oBACzC,CAAC;oBACD,GAAG,IAAI,EAAE,GAAG,UAAU,CAAA;gBACxB,CAAC;qBAAM,CAAC;oBACN,MAAM,UAAU,GAAG,QAAQ,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;oBAC/C,GAAG,IAAI,CAAC,CAAA;oBACR,MAAM,MAAM,GAAG,KAAK,CAAC,IAAI,CAAQ,EAAE,MAAM,EAAE,UAAU,EAAE,CAAC,CAAA;oBACxD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;wBACpC,MAAM,CAAC,CAAC,CAAC,GAAG,IAAI,KAAK,CACnB,SAAS,CAAC,KAAK,EAAE,GAAG,CAAC,EACrB,SAAS,CAAC,KAAK,EAAE,GAAG,GAAG,CAAC,CAAC,EACzB,GAAG,CACJ,CAAA;wBACD,GAAG,IAAI,EAAE,CAAA;oBACX,CAAC;oBACD,QAAQ,CAAC,GAAG,CAAC,GAAG,MAAM,CAAA;gBACxB,CAAC;YACH,CAAC;YACD,MAAM,WAAW,GAAG,QAAQ,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,CAAC,CAAA;YAChD,GAAG,IAAI,CAAC,CAAA;YACR,MAAM,WAAW,GAAG,KAAK,CAAC,IAAI,CAAgB,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC,CAAA;YACtE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,WAAW,EAAE,CAAC,EAAE,EAAE,CAAC;gBACrC,WAAW,CAAC,CAAC,CAAC,GAAG,SAAS,CAAC,KAAK,EAAE,GAAG,CAAC,CAAA;gBACtC,GAAG,IAAI,CAAC,CAAA;YACV,CAAC;YACD,cAAc,CACZ,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,IAAI,EAAE,EAC9B,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,aAAa,CAAC,CACtC,CAAA;YACD,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,KAAK,EAAE,CAAA;QACzC,CAAC;QAED,OAAO;YACL,OAAO,EAAE,cAAc,CAAC,UAAU,CAAC;YACnC,QAAQ;YACR,QAAQ,EAAE,aAAa;YACvB,KAAK,EAAE,SAAS;YAChB,YAAY;YACZ,YAAY,EAAE,kBAAkB;YAChC,SAAS;YACT,aAAa;YACb,aAAa;YACb,cAAc;YACd,MAAM;YACN,WAAW;YACX,WAAW;YACX,YAAY,EAAE,CAAC,IAAI,EAAE;SACtB,CAAA;IACH,CAAC;CACF"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@gmod/tabix",
-  "version": "3.5.3",
+  "version": "3.5.4",
   "packageManager": "pnpm@11.15.1",
   "description": "Read Tabix-indexed files, supports both .tbi and .csi indexes",
   "type": "module",
@@ -50,7 +50,7 @@
   ],
   "dependencies": {
     "@gmod/abortable-promise-cache": "^3.0.4",
-    "@gmod/bgzf-filehandle": "^6.2.0",
+    "@gmod/bgzf-filehandle": "^6.3.2",
     "@jbrowse/quick-lru": "^7.3.5",
     "generic-filehandle2": "^2.2.0"
   },

package/src/csi.ts

@@ -1,18 +1,15 @@
-import { unzip } from '@gmod/bgzf-filehandle'
-
 import Chunk from './chunk.ts'
 import IndexFile from './indexFile.ts'
 import {
   clampChunkEnds,
   memoizeByRefId,
   minVirtualOffset,
-  optimizeChunks,
   parseAuxData,
   parsePseudoBin,
 } from './util.ts'
 import { fromBytes } from './virtualOffset.ts'
 
-import type { Options, RefIndex } from './indexFile.ts'
+import type { IndexData, Options, RefIndex } from './indexFile.ts'
 import type VirtualOffset from './virtualOffset.ts'
 
 const CSI1_MAGIC = 21_582_659 // CSI\1
@@ -28,22 +25,9 @@ function rshift(num: number, bits: number) {
 }
 
 export default class CSI extends IndexFile {
-  private maxBinNumber = 0
-  private depth = 0
-  private minShift = 0
-
   /** @internal */
   async _parse(opts: Options = {}) {
-    const buf = await this.filehandle.readFile({
-      signal: opts.signal,
-      onProgress: opts.onProgress,
-    })
-    const bytes = (await unzip(buf)) as Uint8Array
-    const dataView = new DataView(
-      bytes.buffer,
-      bytes.byteOffset,
-      bytes.byteLength,
-    )
+    const { bytes, dataView } = await this.readIndexBytes(opts)
 
     const magic = dataView.getUint32(0, true)
     let csiVersion
@@ -55,11 +39,10 @@ export default class CSI extends IndexFile {
       throw new Error(`Not a CSI file (magic=${magic})`)
     }
 
-    this.minShift = dataView.getInt32(4, true)
-    this.depth = dataView.getInt32(8, true)
-    this.maxBinNumber = ((1 << ((this.depth + 1) * 3)) - 1) / 7
-    const maxBinNumber = this.maxBinNumber
-    const maxRefLength = 2 ** (this.minShift + this.depth * 3)
+    const minShift = dataView.getInt32(4, true)
+    const depth = dataView.getInt32(8, true)
+    const maxBinNumber = ((1 << ((depth + 1) * 3)) - 1) / 7
+    const maxRefLength = 2 ** (minShift + depth * 3)
     const auxLength = dataView.getInt32(12, true)
     const aux =
       auxLength >= 30
@@ -144,51 +127,13 @@ export default class CSI extends IndexFile {
       firstDataLine,
       csiVersion,
       indices: memoizeByRefId(getIndices),
-      depth: this.depth,
+      minShift,
+      depth,
       maxBinNumber,
       maxRefLength,
     }
   }
 
-  async blocksForRange(
-    refName: string,
-    min: number,
-    max: number,
-    opts: Options = {},
-  ) {
-    if (min < 0) {
-      min = 0
-    }
-
-    const indexData = await this.parse(opts)
-    const refId = indexData.refNameToId[refName]
-    if (refId === undefined) {
-      return []
-    }
-    const ba = indexData.indices(refId)
-    if (!ba) {
-      return []
-    }
-
-    // List of bin #s that overlap min, max
-    const overlappingBins = this.reg2bins(min, max)
-    const chunks: Chunk[] = []
-
-    // Find chunks in overlapping bins.  Leaf bins (< 4681) are not pruned
-    for (const [start, end] of overlappingBins) {
-      for (let bin = start; bin <= end; bin++) {
-        const binChunks = ba.binIndex[bin]
-        if (binChunks) {
-          for (const c of binChunks) {
-            chunks.push(c)
-          }
-        }
-      }
-    }
-
-    return optimizeChunks(chunks, this.minOffset(ba.loffsets, min))
-  }
-
   /**
    * CSI's equivalent of the BAI/TBI linear index: the loffset of the deepest
    * indexed bin covering `beg` is the earliest virtual offset any record
@@ -196,14 +141,17 @@ export default class CSI extends IndexFile {
    * Walks leaf -> previous sibling -> parent until an indexed bin is found.
    * SYNC: htslib hts_itr_query min_off computation
    */
-  private minOffset(
-    loffsets: Record<number, VirtualOffset> | undefined,
+  protected lowestOffset(
+    ref: RefIndex,
     beg: number,
+    { minShift, depth }: IndexData,
   ) {
+    const { loffsets } = ref
     let found: VirtualOffset | undefined
     if (loffsets) {
       // first bin of the deepest level, i.e. (8**depth - 1) / 7
-      let bin = (lshift(1, 3 * this.depth) - 1) / 7 + rshift(beg, this.minShift)
+      let bin =
+        (lshift(1, 3 * depth) - 1) / 7 + rshift(Math.max(beg, 0), minShift)
       while (found === undefined && bin > 0) {
         found = loffsets[bin]
         if (found === undefined) {
@@ -217,23 +165,27 @@ export default class CSI extends IndexFile {
     return found
   }
 
-  /** @internal */
-  reg2bins(beg: number, end: number) {
-    const maxPos = 2 ** (this.minShift + this.depth * 3)
+  protected reg2bins(
+    beg: number,
+    end: number,
+    { minShift, depth, maxBinNumber }: IndexData,
+  ) {
+    beg = Math.max(beg, 0)
+    const maxPos = 2 ** (minShift + depth * 3)
     if (end > maxPos) {
       end = maxPos
     }
     end -= 1
     let l = 0
     let t = 0
-    let s = this.minShift + this.depth * 3
-    const bins = []
-    for (; l <= this.depth; s -= 3, t += lshift(1, l * 3), l += 1) {
+    let s = minShift + depth * 3
+    const bins: (readonly [number, number])[] = []
+    for (; l <= depth; s -= 3, t += lshift(1, l * 3), l += 1) {
       const b = t + rshift(beg, s)
       const e = t + rshift(end, s)
-      if (e - b + bins.length > this.maxBinNumber) {
+      if (e - b + bins.length > maxBinNumber) {
         throw new Error(
-          `query ${beg}-${end} is too large for current binning scheme (shift ${this.minShift}, depth ${this.depth}), try a smaller query or a coarser index binning scheme`,
+          `query ${beg}-${end} is too large for current binning scheme (shift ${minShift}, depth ${depth}), try a smaller query or a coarser index binning scheme`,
         )
       }
       bins.push([b, e] as const)

package/src/indexFile.ts

@@ -1,3 +1,7 @@
+import { unzip } from '@gmod/bgzf-filehandle'
+
+import { optimizeChunks } from './util.ts'
+
 import type Chunk from './chunk.ts'
 import type VirtualOffset from './virtualOffset.ts'
 import type { GenericFilehandle } from 'generic-filehandle2'
@@ -30,13 +34,19 @@ export interface IndexData {
   indices: (refId: number) => RefIndex | undefined
   maxRefLength: number
   skipLines?: number
-  maxBinNumber?: number
   maxBlockSize: number
   firstDataLine?: VirtualOffset
   refCount?: number
   csi?: boolean
   csiVersion?: number
-  depth?: number
+  /**
+   * The binning scheme. TBI reports these too, though its own are fixed by the
+   * format (minShift 14, depth 5) rather than read from the file: they are the
+   * same scheme, and CSI exists to let a file choose other values for them.
+   */
+  minShift: number
+  depth: number
+  maxBinNumber: number
 }
 
 export default abstract class IndexFile {
@@ -49,6 +59,43 @@ export default abstract class IndexFile {
 
   protected abstract _parse(opts: Options): Promise<IndexData>
 
+  /**
+   * The bins that may overlap [beg, end), as one inclusive [first, last] range
+   * per level of the binning scheme.
+   */
+  protected abstract reg2bins(
+    beg: number,
+    end: number,
+    indexData: IndexData,
+  ): (readonly [number, number])[]
+
+  /**
+   * The earliest virtual offset any record overlapping `beg` can have, so that
+   * chunks ending at or before it can be dropped. TBI reads it off the linear
+   * index; CSI, which has none, off the bins' loffsets.
+   */
+  protected abstract lowestOffset(
+    ref: RefIndex,
+    beg: number,
+    indexData: IndexData,
+  ): VirtualOffset | undefined
+
+  /**
+   * The whole index file, decompressed, with a DataView over it. Both .tbi and
+   * .csi are bgzf-compressed and read whole, so both parsers start here.
+   */
+  protected async readIndexBytes(opts: Options) {
+    const buf = await this.filehandle.readFile({
+      signal: opts.signal,
+      onProgress: opts.onProgress,
+    })
+    const bytes = await unzip(buf)
+    return {
+      bytes,
+      dataView: new DataView(bytes.buffer, bytes.byteOffset, bytes.byteLength),
+    }
+  }
+
   /** @internal */
   public async lineCount(refName: string, opts: Options = {}) {
     const indexData = await this.parse(opts)
@@ -65,13 +112,44 @@ export default abstract class IndexFile {
     return rest
   }
 
-  /** @internal */
-  public abstract blocksForRange(
+  /**
+   * The chunks of the data file that may hold records overlapping the region.
+   * The two index formats differ only in their binning scheme and in where
+   * they keep the pruning floor, which is what the two hooks above supply.
+   *
+   * @internal
+   */
+  public async blocksForRange(
     refName: string,
-    start: number,
-    end: number,
-    opts: Options,
-  ): Promise<Chunk[]>
+    min: number,
+    max: number,
+    opts: Options = {},
+  ) {
+    const indexData = await this.parse(opts)
+    const refId = indexData.refNameToId[refName]
+    if (refId === undefined) {
+      return []
+    }
+    const ba = indexData.indices(refId)
+    if (!ba) {
+      return []
+    }
+
+    // Find chunks in overlapping bins. Leaf bins are not pruned.
+    const chunks: Chunk[] = []
+    for (const [start, end] of this.reg2bins(min, max, indexData)) {
+      for (let bin = start; bin <= end; bin++) {
+        const binChunks = ba.binIndex[bin]
+        if (binChunks) {
+          for (const c of binChunks) {
+            chunks.push(c)
+          }
+        }
+      }
+    }
+
+    return optimizeChunks(chunks, this.lowestOffset(ba, min, indexData))
+  }
 
   /** @internal */
   async parse(opts: Options = {}) {