@genome-spy/core 0.64.0 → 0.65.0

package/dist/bundle/{index-C08YCM2T.js → index-D-w7Mmt9.js}

@@ -1,4 +1,78 @@
-const h = {
+function R(d, t, e) {
+  const r = /* @__PURE__ */ Object.create(null), i = e.length, a = t.length, l = 9, s = 58;
+  let n = 0;
+  if (d === "GT") {
+    for (let o = 0; o < i; o++) {
+      const p = n;
+      for (; n < a && t.charCodeAt(n) !== l; )
+        n++;
+      r[e[o]] = t.slice(p, n), n++;
+    }
+    return r;
+  }
+  const c = d.indexOf("GT");
+  if (c === -1)
+    return r;
+  if (c === 0) {
+    for (let o = 0; o < i; o++) {
+      const p = n;
+      for (; n < a && t.charCodeAt(n) !== s && t.charCodeAt(n) !== l; )
+        n++;
+      for (r[e[o]] = t.slice(p, n); n < a && t.charCodeAt(n) !== l; )
+        n++;
+      n++;
+    }
+    return r;
+  }
+  let b = 0;
+  for (let o = 0; o < c; o++)
+    d.charCodeAt(o) === s && b++;
+  for (let o = 0; o < i; o++) {
+    const p = n;
+    let u = n;
+    for (; u < a && t.charCodeAt(u) !== l; )
+      u++;
+    let g = 0, N = p;
+    for (let h = p; h <= u; h++)
+      if (h === u || t.charCodeAt(h) === s) {
+        if (g === b) {
+          r[e[o]] = t.slice(N, h);
+          break;
+        }
+        g++, N = h + 1;
+      }
+    n = u + 1;
+  }
+  return r;
+}
+function E(d) {
+  const t = [], e = [];
+  let r = !1, i = !1;
+  const a = d.length;
+  for (let l = 0; l < a; l++) {
+    const s = d[l];
+    s === '"' ? (r = !r, e.push(s)) : s === "[" ? (i = !0, e.push(s)) : s === "]" ? (i = !1, e.push(s)) : s === "," && !r && !i ? (t.push(e.join("").trim()), e.length = 0) : e.push(s);
+  }
+  return e.length > 0 && t.push(e.join("").trim()), t;
+}
+function F(d, t) {
+  const e = d.indexOf(t);
+  return [d.slice(0, e), d.slice(e + 1)];
+}
+function C(d) {
+  const t = d.slice(1, -1), e = E(t), r = [];
+  for (let i = 0; i < e.length; i++) {
+    const a = e[i], [l, s] = F(a, "=");
+    if (s && s.startsWith("[") && s.endsWith("]")) {
+      const n = s.slice(1, -1).split(",");
+      for (let c = 0; c < n.length; c++)
+        n[c] = n[c].trim();
+      r.push([l, n]);
+    } else s && s.startsWith('"') && s.endsWith('"') ? r.push([l, s.slice(1, -1)]) : r.push([l, s]);
+  }
+  return Object.fromEntries(r);
+}
+const S = {
   // INFO fields
   InfoFields: {
     // from the VCF4.3 spec, https://samtools.github.io/hts-specs/VCFv4.3.pdf
@@ -343,35 +417,37 @@ const h = {
     }
   }
 };
-function v(b) {
+function L(d) {
   try {
-    return decodeURIComponent(b);
+    return decodeURIComponent(d);
   } catch {
-    return b;
+    return d;
   }
 }
 class M {
-  constructor({ header: a = "", strict: r = !0 }) {
-    if (!a.length)
+  constructor({ header: t = "", strict: e = !0 }) {
+    if (!t.length)
       throw new Error("empty header received");
-    const n = a.split(/[\r\n]+/).filter((o) => o);
-    if (!n.length)
+    const r = t.split(/[\r\n]+/).filter(Boolean);
+    if (!r.length)
       throw new Error("no non-empty header lines specified");
-    this.strict = r, this.metadata = JSON.parse(JSON.stringify({
-      INFO: h.InfoFields,
-      FORMAT: h.GenotypeFields,
-      ALT: h.AltTypes,
-      FILTER: h.FilterTypes
-    }));
-    let e;
-    if (n.forEach((o) => {
-      if (o.startsWith("#"))
-        o.startsWith("##") ? this.parseMetadata(o) : e = o;
+    this.strict = e, this.metadata = {
+      INFO: { ...S.InfoFields },
+      FORMAT: { ...S.GenotypeFields },
+      ALT: { ...S.AltTypes },
+      FILTER: { ...S.FilterTypes }
+    };
+    let i;
+    for (let n = 0; n < r.length; n++) {
+      const c = r[n];
+      if (c.startsWith("#"))
+        c.startsWith("##") ? this.parseMetadata(c) : i = c;
       else throw new Error(`Bad line in header:
-${o}`);
-    }), !e)
+${c}`);
+    }
+    if (!i)
       throw new Error("No format line found in header");
-    const t = e.trim().split("	"), s = t.slice(0, 8), i = [
+    const a = i.trim().split("	"), l = a.slice(0, 8), s = [
       "#CHROM",
       "POS",
       "ID",
@@ -381,54 +457,88 @@ ${o}`);
       "FILTER",
       "INFO"
     ];
-    if (t.length < 8)
+    if (a.length < 8)
       throw new Error(`VCF header missing columns:
-${e}`);
-    if (s.length !== i.length || !s.every((o, p) => o === i[p]))
+${i}`);
+    if (l.length !== s.length || !l.every((n, c) => n === s[c]))
       throw new Error(`VCF column headers not correct:
-${e}`);
-    this.samples = t.slice(9);
+${i}`);
+    this.samples = a.slice(9);
   }
-  parseSamples(a, r) {
-    const n = {};
-    if (a) {
-      const e = r.split("	"), t = a.split(":"), s = t.map((i) => {
-        const o = this.getMetadata("FORMAT", i, "Type");
-        return o === "Integer" || o === "Float";
-      });
-      for (let i = 0; i < this.samples.length; i++) {
-        const o = this.samples[i];
-        n[o] = {};
-        const p = e[i].split(":");
-        for (let c = 0; c < p.length; c++) {
-          const m = p[c];
-          n[o][t[c]] = m === "" || m === "." ? void 0 : m.split(",").map((u) => u === "." ? void 0 : s[c] ? +u : u);
+  parseInfo(t) {
+    var s;
+    const e = {}, r = t.includes("%"), i = t.split(";"), a = this.metadata.INFO, l = i.length;
+    for (let n = 0; n < l; n++) {
+      const c = i[n], b = c.indexOf("="), o = b === -1 ? c : c.slice(0, b), p = b === -1 ? void 0 : c.slice(b + 1), u = (s = a[o]) == null ? void 0 : s.Type;
+      if (u === "Flag")
+        e[o] = !0;
+      else if (!p)
+        e[o] = !0;
+      else {
+        const g = u === "Integer" || u === "Float", N = p.split(","), h = N.length;
+        if (r) {
+          const f = [];
+          for (let m = 0; m < h; m++) {
+            const y = N[m];
+            if (y === ".")
+              f.push(void 0);
+            else {
+              const D = L(y);
+              f.push(g ? Number(D) : D);
+            }
+          }
+          e[o] = f;
+        } else {
+          const f = [];
+          for (let m = 0; m < h; m++) {
+            const y = N[m];
+            y === "." ? f.push(void 0) : f.push(g ? Number(y) : y);
+          }
+          e[o] = f;
         }
       }
     }
-    return n;
+    return e;
   }
-  parseGenotypesOnly(a, r) {
-    const n = r.split("	"), e = {};
-    let t = 0;
-    const s = a.split(":");
-    if (s.length === 1)
-      for (const i of this.samples)
-        e[i] = n[t++];
-    else {
-      const i = s.findIndex((o) => o === "GT");
-      if (i === 0)
-        for (const o of this.samples) {
-          const p = n[t++], c = p.indexOf(":");
-          c !== -1 ? e[o] = p.slice(0, c) : console.warn("unknown");
-        }
-      else
-        for (const o of this.samples) {
-          const p = n[t++].split(":");
-          e[o] = p[i];
-        }
+  parseSamples(t, e) {
+    var i;
+    const r = {};
+    if (t) {
+      const a = e.split("	"), l = t.split(":"), s = this.metadata.FORMAT, n = [];
+      for (let o = 0; o < l.length; o++) {
+        const p = (i = s[l[o]]) == null ? void 0 : i.Type;
+        n.push(p === "Integer" || p === "Float");
+      }
+      const c = l.length, b = this.samples.length;
+      for (let o = 0; o < b; o++) {
+        const p = this.samples[o], u = {}, g = a[o], N = g.length;
+        let h = 0, f = 0;
+        for (let m = 0; m <= N; m++)
+          if (m === N || g[m] === ":") {
+            const y = g.slice(h, m);
+            if (y === "" || y === ".")
+              u[l[f]] = void 0;
+            else {
+              const D = y.split(","), I = [];
+              if (n[f])
+                for (let T = 0; T < D.length; T++) {
+                  const A = D[T];
+                  I.push(A === "." ? void 0 : +A);
+                }
+              else
+                for (let T = 0; T < D.length; T++) {
+                  const A = D[T];
+                  I.push(A === "." ? void 0 : A);
+                }
+              u[l[f]] = I;
+            }
+            if (h = m + 1, f += 1, f >= c)
+              break;
+          }
+        r[p] = u;
+      }
     }
-    return e;
+    return r;
   }
   /**
    * Parse a VCF metadata line (i.e. a line that starts with "##") and add its
@@ -437,17 +547,17 @@ ${e}`);
    * @param {string} line - A line from the VCF. Supports both LF and CRLF
    * newlines.
    */
-  parseMetadata(a) {
-    const r = /^##(.+?)=(.*)/.exec(a.trim());
-    if (!r)
-      throw new Error(`Line is not a valid metadata line: ${a}`);
-    const [n, e] = r.slice(1, 3), t = n;
-    if (e != null && e.startsWith("<")) {
-      t in this.metadata || (this.metadata[t] = {});
-      const [s, i] = this.parseStructuredMetaVal(e);
-      this.metadata[t][s] = i;
+  parseMetadata(t) {
+    const e = /^##(.+?)=(.*)/.exec(t.trim());
+    if (!e)
+      throw new Error(`Line is not a valid metadata line: ${t}`);
+    const [r, i] = e.slice(1, 3), a = r;
+    if (i != null && i.startsWith("<")) {
+      a in this.metadata || (this.metadata[a] = {});
+      const [l, s] = this.parseStructuredMetaVal(i);
+      l ? this.metadata[a][l] = s : this.metadata[a] = s;
     } else
-      this.metadata[t] = e;
+      this.metadata[a] = i;
   }
   /**
    * Parse a VCF header structured meta string (i.e. a meta value that starts
@@ -458,9 +568,9 @@ ${e}`);
    * @returns {Array} - Array with two entries, 1) a string of the metadata ID
    * and 2) an object with the other key-value pairs in the metadata
    */
-  parseStructuredMetaVal(a) {
-    const r = this.parseKeyValue(a.replace(/^<|>$/g, ""), ","), n = r.ID;
-    return delete r.ID, "Number" in r && (Number.isNaN(Number(r.Number)) || (r.Number = Number(r.Number))), [n, r];
+  parseStructuredMetaVal(t) {
+    const e = C(t), r = e.ID;
+    return delete e.ID, "Number" in e && (Number.isNaN(Number(e.Number)) || (e.Number = Number(e.Number))), [r, e];
   }
   /**
    * Get metadata filtered by the elements in args. For example, can pass
@@ -471,66 +581,76 @@ ${e}`);
    *
    * @returns {any} An object, string, or number, depending on the filtering
    */
-  getMetadata(...a) {
-    let r = this.metadata;
-    for (const n of a)
-      if (r = r[n], !r)
-        return r;
-    return r;
+  getMetadata(...t) {
+    let e = this.metadata;
+    const r = t.length;
+    for (let i = 0; i < r; i++)
+      if (e = e[t[i]], !e)
+        return e;
+    return e;
   }
   /**
-   * Sometimes VCFs have key-value strings that allow the separator within the
-   * value if it's in quotes, like:
-   * 'ID=DB,Number=0,Type=Flag,Description="dbSNP membership, build 129"'
+   * Parse a VCF line into an object like
    *
-   * Parse this at a low level since we can't just split at "," (or whatever
-   * separator). Above line would be parsed to: {ID: 'DB', Number: '0', Type:
-   * 'Flag', Description: 'dbSNP membership, build 129'}
-   */
-  parseKeyValue(a, r = ";") {
-    const n = {};
-    let e = "", t = "", s = 1;
-    for (const i of a)
-      s === 1 ? i === "=" ? s = 2 : i !== r ? e += i : t === "" && (n[e] = void 0, e = "") : s === 2 ? i === r ? (n[e] = t, e = "", t = "", s = 1) : i === '"' ? s = 3 : t += i : s === 3 && (i !== '"' ? t += i : s = 2);
-    return s === 2 || s === 3 ? n[e] = t : s === 1 && (n[e] = void 0), n;
-  }
-  /**
-   * Parse a VCF line into an object like { CHROM POS ID REF ALT QUAL FILTER
-   * INFO } with SAMPLES optionally included if present in the VCF
+   * ```typescript
+   * {
+   *   CHROM: 'contigA',
+   *   POS: 3000,
+   *   ID: ['rs17883296'],
+   *   REF: 'G',
+   *   ALT: ['T', 'A'],
+   *   QUAL: 100,
+   *   FILTER: 'PASS',
+   *   INFO: {
+   *     NS: [3],
+   *     DP: [14],
+   *     AF: [0.5],
+   *     DB: true,
+   *     XYZ: ['5'],
+   *   },
+   *   SAMPLES: () => ({
+   *     HG00096: {
+   *       GT: ['0|0'],
+   *       AP: ['0.000', '0.000'],
+   *     }
+   *   }),
+   *   GENOTYPES: () => ({
+   *     HG00096: '0|0'
+   *   })
+   * }
+   * ```
+   *
+   * SAMPLES and GENOTYPES methods are functions instead of static data fields
+   * because it avoids parsing the potentially long list of samples from e.g.
+   * 1000 genotypes data unless requested.
    *
-   * @param {string} line - A string of a line from a VCF. Supports both LF and
-   * CRLF newlines.
+   * The SAMPLES function gives all info about the samples
+   *
+   * The GENOTYPES function only extracts the raw GT string if it exists, for
+   * potentially optimized parsing by programs that need it
+   *
+   * @param {string} line - A string of a line from a VCF
    */
-  parseLine(a) {
-    var r, n;
-    let e = 0;
-    for (let l = 0; e < a.length && (a[e] === "	" && (l += 1), l !== 9); e += 1)
-      ;
-    const t = a.slice(0, e).split("	"), s = a.slice(e + 1), [i, o, p, c, m, u, T] = t, g = i, I = +o, A = p === "." ? void 0 : p.split(";"), S = c, R = m === "." ? void 0 : m.split(","), E = u === "." ? void 0 : +u, y = T === "." ? void 0 : T.split(";");
-    if (this.strict && !t[7])
+  parseLine(t) {
+    let e = 0, r = 0;
+    for (; e < t.length && r < 9; )
+      t[e] === "	" && (r += 1), e += 1;
+    const i = r === 9 ? e - 1 : e, a = t.slice(0, i).split("	"), l = t.slice(i + 1), [s, n, c, b, o, p, u] = a, g = s, N = +n, h = c === "." ? void 0 : c.split(";"), f = b, m = o === "." ? void 0 : o.split(","), y = p === "." ? void 0 : +p, D = u === "." ? void 0 : u.split(";"), I = a[8];
+    if (this.strict && !a[7])
       throw new Error("no INFO field specified, must contain at least a '.' (turn off strict mode to allow)");
-    const F = (r = t[7]) === null || r === void 0 ? void 0 : r.includes("%"), f = t[7] === void 0 || t[7] === "." ? {} : this.parseKeyValue(t[7]);
-    for (const l of Object.keys(f)) {
-      const D = (n = f[l]) === null || n === void 0 ? void 0 : n.split(",").map((d) => d === "." ? void 0 : d).map((d) => d && F ? v(d) : d), N = this.getMetadata("INFO", l, "Type");
-      N === "Integer" || N === "Float" ? f[l] = D == null ? void 0 : D.map((d) => d === void 0 ? void 0 : Number(d)) : N === "Flag" ? f[l] = !0 : f[l] = D;
-    }
+    const T = a[7] === void 0 || a[7] === "." ? {} : this.parseInfo(a[7]);
     return {
       CHROM: g,
-      POS: I,
-      ALT: R,
-      INFO: f,
-      REF: S,
-      FILTER: y && y.length === 1 && y[0] === "PASS" ? "PASS" : y,
-      ID: A,
-      QUAL: E,
-      SAMPLES: () => {
-        var l;
-        return this.parseSamples((l = t[8]) !== null && l !== void 0 ? l : "", s);
-      },
-      GENOTYPES: () => {
-        var l;
-        return this.parseGenotypesOnly((l = t[8]) !== null && l !== void 0 ? l : "", s);
-      }
+      POS: N,
+      ALT: m,
+      INFO: T,
+      REF: f,
+      FILTER: (D == null ? void 0 : D.length) === 1 && D[0] === "PASS" ? "PASS" : D,
+      ID: h,
+      QUAL: y,
+      FORMAT: I,
+      SAMPLES: () => this.parseSamples(a[8] ?? "", l),
+      GENOTYPES: () => R(a[8] ?? "", l, this.samples)
     };
   }
 }