stacks, rebased on observablehq/plot#176

Author: tigrevol8888

src/transforms/stack.js

@@ -1,36 +1,114 @@
-import {InternMap} from "d3-array";
-import {valueof} from "../mark.js";
+import {InternMap, ascending, cumsum, descending, group, groupSort, maxIndex, range, rollup, sum} from "d3-array";
+import {field, maybeSort, take, valueof} from "../mark.js";
 
 export function stackX({x, y, ...options}) {
-  const [transform, Y, x1, x2] = stack(y, x);
+  const [transform, Y, x1, x2] = stack(y, x, options);
   return {...options, transform, y: Y, x1, x2};
 }
 
 export function stackY({x, y, ...options}) {
-  const [transform, X, y1, y2] = stack(x, y);
+  const [transform, X, y1, y2] = stack(x, y, options);
   return {...options, transform, x: X, y1, y2};
 }
 
-// TODO configurable series order
-function stack(x, y) {
+function stack(x, y = () => 1, {
+  z, fill, stroke, title,
+  rank,
+  reverse = ["descending", "reverse"].includes(rank),
+  offset,
+  sort
+}) {
   const [X, setX] = lazyChannel(x);
   const [Y1, setY1] = lazyChannel(y);
   const [Y2, setY2] = lazyChannel(y);
   return [
-    data => {
+    (data, facets) => {
       const X = setX(valueof(data, x));
       const Y = valueof(data, y);
-      const n = X.length;
-      const Y0 = new InternMap();
+      const Z = valueof(data, z || fill || stroke || title);
+      const R = maybeRank(rank, data, X, Y, Z);
+      const n = data.length;
+      const I = range(n);
       const Y1 = setY1(new Float64Array(n));
       const Y2 = setY2(new Float64Array(n));
-      for (let i = 0; i < n; ++i) {
-        const k = X[i];
-        const y1 = Y1[i] = Y0.has(k) ? Y0.get(k) : 0;
-        const y2 = Y2[i] = y1 + +Y[i];
-        Y0.set(k, isNaN(y2) ? y1 : y2);
+      sort = maybeSort(sort);
+    
+      for (const index of (facets === undefined ? [I] : facets)) {
+        
+        if (sort) {
+          const facet = take(data, index);
+          const index0 = index.slice();
+          const sorted = sort(facet);
+          for (let k = 0; k < index.length; k++) index[k] = index0[facet.indexOf(sorted[k])];
+        }
+        
+        const Yp = new InternMap();
+        const Yn = new InternMap();
+        
+        const stacks = group(index, i => X[i]);
+        
+        // rank sort
+        if (R) {
+          const a = reverse ? descending : ascending;
+          for (const [, stack] of stacks) stack.sort((i, j) => a(R[i], R[j]));
+        }
+        
+        // stack
+        for (const [x, stack] of stacks) {
+          for (const i of stack) {
+            const v = +Y[i];
+            const [Y0, ceil, floor] = v < 0 ? [Yn, Y1, Y2] : [Yp, Y2, Y1];
+            const y1 = floor[i] = Y0.has(x) ? Y0.get(x) : 0;
+            const y2 = ceil[i] = y1 + +Y[i];
+            Y0.set(x, isNaN(y2) ? y1 : y2);
+          }
+        }
+  
+        // offset
+        if (offset === "expand") {
+          for (const i of index) {
+            const x = X[i];
+            const floor = Yn.has(x) ? Yn.get(x) : 0;
+            const ceil = Yp.has(x) ? Yp.get(x) : 0;
+            const m = 1 / (ceil - floor || 1);
+            Y1[i] = m * (-floor + Y1[i]);
+            Y2[i] = m * (-floor + Y2[i]);
+          }
+        }
+        if (offset === "silhouette") {
+          for (const i of index) {
+            const x = X[i];
+            const floor = Yn.has(x) ? Yn.get(x) : 0;
+            const ceil = Yp.has(x) ? Yp.get(x) : 0;
+            const m = (ceil + floor) / 2;
+            Y1[i] -= m;
+            Y2[i] -= m;
+          }
+        }
+        if (offset === "wiggle") {
+          const prev = new InternMap();
+          let y = 0;
+          for (const [, stack] of stacks) {
+            let j = -1;
+            const Fi = stack.map(i => Math.abs(Y2[i] - Y1[i]));
+            const Df = stack.map(i => {
+              j = z ? Z[i] : ++j;
+              const value = Y2[i] - Y1[i];
+              const diff = prev.has(j) ? value - prev.get(j) : 0;
+              prev.set(j, value);
+              return diff;
+            });
+            const Cf1 = [0, ...cumsum(Df)];
+            for (const i of stack) {
+              Y1[i] += y;
+              Y2[i] += y;
+            }
+            const s1 = sum(Fi);
+            if (s1) y -= sum(Fi, (d, i) => (Df[i] / 2 + Cf1[i]) * d) / s1;
+          }
+        }
       }
-      return data;
+      return {index: facets === undefined ? I : facets, data};
     },
     X,
     Y1,
@@ -45,11 +123,74 @@ function lazyChannel(source) {
   let value;
   return [
     {
-      transform() { return value; },
+      transform: () => value,
       label: typeof source === "string" ? source
         : source ? source.label
         : undefined
     },
     v => value = v
   ];
 }
+
+// well-known ranking strategies by series
+function maybeRank(rank, data, X, Y, Z) {
+  if (rank == null) return [null];
+  // d3.stackOrderNone, sorts series by key, ascending
+  // d3.stackOrderReverse, sorts series by key, descending
+  if (rank === "key" || rank === "none" || rank === "reverse") {
+    return Z;
+  }
+  // d3.stackOrderAscending, sorts series by sum of value, ascending
+  if (rank === "sum" || rank === "ascending" || rank === "descending") {
+    const S = groupSort(range(data.length), g => sum(g, i => Y[i]), i => Z[i]);
+    return Z.map(z => S.indexOf(z));
+  }
+  // ranks items by value
+  if (rank === "value") {
+    return Y;
+  }
+  // d3.stackOrderAppearance, sorts series by x = argmax of value
+  if (rank === "appearance") {
+    const K = groupSort(
+      range(data.length),
+      v => X[v[maxIndex(v, i => Y[i])]],
+      i => Z[i]
+    );
+    return Z.map(z => K.indexOf(z));
+  }
+  // d3.stackOrderInsideOut, sorts series by x = argmax of value, then rearranges them
+  // inside out by alternating series according to the sign of a running divergence
+  // of their sums
+  if (rank === "insideOut") {
+    const K = groupSort(
+      range(data.length),
+      v => X[v[maxIndex(v, i => Y[i])]],
+      i => Z[i]
+    );
+    const sums = rollup(range(data.length), v => sum(v, i => Y[i]), i => Z[i]);
+    const order = [];
+    let diff = 0;
+    for (const k of K) {
+      if (diff < 0) {
+        diff += sums.get(k);
+        order.push(k);
+      } else {
+        diff -= sums.get(k);
+        order.unshift(k);
+      }
+    }
+    return Z.map(z => order.indexOf(z));
+  }
+  // any other string is a datum accessor
+  if (typeof rank === "string") {
+    return valueof(data, field(rank));
+  }
+  // rank can be an array of z (particularly useful with groupSort)
+  if (rank.indexOf) {
+    return Z.map(z => rank.indexOf(z));
+  }
+  // final case, a generic function
+  if (typeof rank === "function") {
+    return valueof(data, rank);
+  }
+}

test/data/caltrain.csv

@@ -0,0 +1,79 @@
+station,time,line,type,orientation,hours,minutes
+Palo Alto,5:01am,101N,N,N,5,01
+Palo Alto,8:01pm,191N,N,N,20,01
+Palo Alto,9:01pm,193N,N,N,21,01
+Palo Alto,10:01pm,195N,N,N,22,01
+Palo Alto,11:01pm,197N,N,N,23,01
+Palo Alto,8:01am,216L,L,S,8,01
+Palo Alto,9:01am,226L,L,S,9,01
+Palo Alto,5:01pm,264L,L,S,17,01
+Palo Alto,6:02pm,274L,L,S,18,02
+Palo Alto,10:03am,134N,N,S,10,03
+Palo Alto,11:03am,138N,N,S,11,03
+Palo Alto,12:03pm,142N,N,S,12,03
+Palo Alto,1:03pm,146N,N,S,13,03
+Palo Alto,2:03pm,150N,N,S,14,03
+Palo Alto,3:03pm,154N,N,S,15,03
+Palo Alto,4:03pm,158N,N,S,16,03
+Palo Alto,6:05am,305B,B,N,6,05
+Palo Alto,7:05am,313B,B,N,7,05
+Palo Alto,8:05am,323B,B,N,8,05
+Palo Alto,5:06pm,369B,B,N,17,06
+Palo Alto,6:06pm,379B,B,N,18,06
+Palo Alto,7:10pm,287L,L,N,19,10
+Palo Alto,9:11am,233L,L,N,9,11
+Palo Alto,10:11am,237L,L,N,10,11
+Palo Alto,3:11pm,257L,L,N,15,11
+Palo Alto,5:12pm,368B,B,S,17,12
+Palo Alto,6:12pm,378B,B,S,18,12
+Palo Alto,7:12pm,386B,B,S,19,12
+Palo Alto,7:16am,215L,L,N,7,16
+Palo Alto,8:16am,225L,L,N,8,16
+Palo Alto,4:16pm,261L,L,N,16,16
+Palo Alto,5:16pm,267L,L,N,17,16
+Palo Alto,6:16pm,277L,L,N,18,16
+Palo Alto,7:18am,208L,L,S,7,18
+Palo Alto,8:18am,218L,L,S,8,18
+Palo Alto,9:18am,228L,L,S,9,18
+Palo Alto,7:21pm,189N,N,N,19,21
+Palo Alto,6:21am,104N,N,S,6,21
+Palo Alto,6:23am,309B,B,N,6,23
+Palo Alto,7:23am,319B,B,N,7,23
+Palo Alto,8:23am,329B,B,N,8,23
+Palo Alto,4:24pm,263L,L,N,16,24
+Palo Alto,5:24pm,271L,L,N,17,24
+Palo Alto,6:24pm,281L,L,N,18,24
+Palo Alto,10:25am,236L,L,S,10,25
+Palo Alto,3:25pm,256L,L,S,15,25
+Palo Alto,4:25pm,260L,L,S,16,25
+Palo Alto,7:26am,210L,L,S,7,26
+Palo Alto,8:26am,220L,L,S,8,26
+Palo Alto,9:26am,230L,L,S,9,26
+Palo Alto,8:26pm,190N,N,S,20,26
+Palo Alto,5:36am,103N,N,N,5,36
+Palo Alto,6:36am,207L,L,N,6,36
+Palo Alto,7:36am,217L,L,N,7,36
+Palo Alto,8:36am,227L,L,N,8,36
+Palo Alto,9:36pm,192N,N,S,21,36
+Palo Alto,10:36pm,194N,N,S,22,36
+Palo Alto,11:36pm,196N,N,S,23,36
+Palo Alto,3:38pm,159N,N,N,15,38
+Palo Alto,5:38pm,270L,L,S,17,38
+Palo Alto,6:38pm,280L,L,S,18,38
+Palo Alto,7:38pm,288L,L,S,19,38
+Palo Alto,9:41am,135N,N,N,9,41
+Palo Alto,10:41am,139N,N,N,10,41
+Palo Alto,11:41am,143N,N,N,11,41
+Palo Alto,12:41pm,147N,N,N,12,41
+Palo Alto,1:41pm,151N,N,N,13,41
+Palo Alto,2:41pm,155N,N,N,14,41
+Palo Alto,4:44pm,362B,B,S,16,44
+Palo Alto,5:49pm,372B,B,S,17,49
+Palo Alto,6:49pm,382B,B,S,18,49
+Palo Alto,5:51am,102N,N,S,5,51
+Palo Alto,7:51am,314B,B,S,7,51
+Palo Alto,8:51am,324B,B,S,8,51
+Palo Alto,5:54pm,275L,L,N,17,54
+Palo Alto,6:54pm,285L,L,N,18,54
+Palo Alto,6:57am,206L,L,S,6,57
+Palo Alto,12:57am,198N,N,S,24,57

test/data/learning-poverty.csv

@@ -0,0 +1,101 @@
+Country Name,Out-of-School (OoS),Below Minimum Proficiency (in School),Learning Poverty,Assessment Year,Assessment,population,density
+Afghanistan,49.6,87,93.4,2013,NLA,38928346,60
+Argentina,0.6,53.6,53.9,2013,LLECE,45195774,17
+Armenia,7.2,30,35,2015,TIMSS,2963243,104
+Australia,3.2,5.5,8.6,2016,PIRLS,25499884,3
+Austria,0,2.4,2.4,2016,PIRLS,9006398,109
+Azerbaijan,5,19.2,23.3,2016,PIRLS,10139177,123
+Bahrain,2.1,30.6,32.1,2016,PIRLS,1701575,2239
+Bangladesh,4.9,56,58.1,2017,NLA,164689383,1265
+Belgium,1.3,5.1,6.4,2016,PIRLS,11589623,383
+Benin,3.6,77.3,78.2,2014,PASEC,12123200,108
+Botswana,7.2,44.3,48.3,2011,PIRLS,2351627,4
+Brazil,2.7,46.9,48.4,2013,LLECE,212559417,25
+Bulgaria,6.8,5.2,11.7,2016,PIRLS,6948445,64
+Burkina Faso,31.7,78.6,85.4,2014,PASEC,20903273,76
+Burundi,2.7,92.7,92.9,2014,PASEC,11890784,463
+Cambodia,2.6,49.8,51.1,2013,NLA,16718965,95
+Cameroon,5.2,75.9,77.2,2014,PASEC,26545863,56
+Canada,0,4.3,4.3,2016,PIRLS,37742154,4
+Chad,21.1,97,97.7,2014,PASEC,16425864,13
+Chile,9.3,30.3,36.8,2013,LLECE,19116201,26
+China,0,18.2,18.2,2016,NLA,1439323776,153
+Colombia,6.9,44.7,48.6,2013,LLECE,50882891,46
+"Congo, Dem Rep",63.2,62,86,2011,NLA,89561403,40
+"Congo, Rep",12.8,82.9,85.1,2014,PASEC,5518087,16
+Costa Rica,1.1,31.7,32.5,2013,LLECE,5094118,100
+Cote d’Ivoire,21.1,77.6,82.3,2014,PASEC,26378274,83
+Croatia,3,1,4,2011,PIRLS,4105267,73
+Cyprus,2.2,14.3,16.2,2015,TIMSS,1207359,131
+Czech Republic,0,3,3,2016,PIRLS,10708981,139
+Denmark,1,2.6,3.6,2016,PIRLS,5792202,137
+Dominican Republic,6.6,79.4,80.7,2013,LLECE,10847910,225
+Ecuador,1.9,62.1,62.8,2013,LLECE,17643054,71
+"Egypt, Arab Rep",1.4,69.2,69.6,2016,PIRLS,102334404,103
+Ethiopia,14,88.7,90.3,2015,NLA,114963588,115
+Finland,0.9,1.7,2.6,2016,PIRLS,5540720,18
+France,0.9,6.3,7.1,2016,PIRLS,65273511,119
+Georgia,0.4,13.5,13.8,2016,PIRLS,3989167,57
+Germany,0.2,5.5,5.7,2016,PIRLS,83783942,240
+Guatemala,10.1,63.6,67.3,2013,LLECE,17915568,167
+Honduras,17.1,69.4,74.7,2013,LLECE,9904607,89
+"Hong Kong SAR, China",1.9,1.4,3.2,2016,PIRLS,7496981,7140
+Hungary,3.1,2.9,5.9,2016,PIRLS,9660351,107
+India,2.3,53.7,54.8,2017,NLA,1380004385,464
+Indonesia,2.4,33.8,35.4,2011,PIRLS,273523615,151
+"Iran, Islamic Rep",0.9,35.1,35.7,2016,PIRLS,83992949,52
+Ireland,0,2.3,2.3,2016,PIRLS,4937786,72
+Israel,2.9,9,11.7,2016,PIRLS,8655535,400
+Italy,1.4,2.1,3.5,2016,PIRLS,60461826,206
+Japan,1.2,1,2.2,2015,TIMSS,126476461,347
+Jordan,4,50,52,2015,TIMSS,10203134,115
+Kazakhstan,0.3,1.9,2.2,2016,PIRLS,18776707,7
+"Korea, Rep",2.7,0.3,3,2015,TIMSS,51269185,527
+Kuwait,3.3,49.4,51,2016,PIRLS,4270571,240
+Kyrgyz Republic,1.9,63.8,64.5,2014,NLA,6524195,34
+Latvia,3.2,0.8,4,2016,PIRLS,1886198,30
+Lithuania,0.3,2.7,3,2016,PIRLS,2722289,43
+"Macao SAR, China",1.3,2.4,3.7,2016,PIRLS,649335,21645
+Madagascar,21.9,95.8,96.7,2015,NLA,27691018,48
+Malaysia,1.4,11.7,12.9,2017,NLA,32365999,99
+Mali,33,86.6,91,2012,NLA,20250833,17
+Malta,2.4,26.8,28.6,2016,PIRLS,441543,1380
+Mexico,1.2,42.5,43.2,2013,LLECE,128932753,66
+Morocco,5.4,63.8,65.8,2016,PIRLS,36910560,83
+Netherlands,0.3,1.3,1.6,2016,PIRLS,17134872,508
+New Zealand,1.5,10,11.4,2016,PIRLS,4822233,18
+Nicaragua,1.6,69.3,69.8,2013,LLECE,6624554,55
+Niger,38.9,97.9,98.7,2014,PASEC,24206644,19
+Norway,0.2,5.8,6,2016,PIRLS,5421241,15
+Oman,1.5,40.9,41.8,2016,PIRLS,5106626,16
+Pakistan,27.3,65,74.5,2014,NLA,220892340,287
+Panama,7.1,64.1,66.6,2013,LLECE,4314767,58
+Paraguay,10.8,71.3,74.4,2013,LLECE,7132538,18
+Peru,4.2,53.7,55.7,2013,LLECE,32971854,26
+Poland,4.4,2,6.3,2016,PIRLS,37846611,124
+Portugal,3.6,3,6.5,2016,PIRLS,10196709,111
+Qatar,2.2,33.8,35.3,2016,PIRLS,2881053,248
+Romania,6.9,14.1,20,2011,PIRLS,19237691,84
+Russian Federation,2.4,0.9,3.3,2016,PIRLS,145934462,9
+Saudi Arabia,2.5,36.7,38.3,2016,PIRLS,34813871,16
+Senegal,25.7,65.2,74.1,2014,PASEC,16743927,87
+Serbia,0.8,7.4,8.1,2015,TIMSS,8737371,100
+Singapore,0.1,2.7,2.8,2016,PIRLS,5850342,8358
+Slovak Republic,2.1,6.6,8.5,2016,PIRLS,5459642,114
+Slovenia,2.2,3.7,5.8,2016,PIRLS,2078938,103
+South Africa,8.4,77.9,79.8,2016,PIRLS,59308690,49
+Spain,1.5,3.4,4.9,2016,PIRLS,46754778,94
+Sri Lanka,0.9,14,14.8,2015,NLA,21413249,341
+Sweden,0.4,1.9,2.3,2016,PIRLS,10099265,25
+Thailand,2,21.9,23.5,2011,TIMSS,69799978,137
+Togo,8.5,84.2,85.6,2014,PASEC,8278724,152
+Trinidad and Tobago,1.3,19.7,20.7,2016,PIRLS,1399488,273
+Tunisia,0.4,65.1,65.3,2011,TIMSS,11818619,76
+Turkey,5,17.6,21.7,2015,TIMSS,84339067,110
+Uganda,9,81.1,82.8,2014,NLA,45741007,229
+United Arab Emirates,2.8,32.4,34.3,2016,PIRLS,9890402,118
+United Kingdom,0.2,3.2,3.4,2016,PIRLS,67886011,281
+United States,4.1,3.9,7.9,2016,PIRLS,331002651,36
+Uruguay,0.5,41.4,41.7,2013,LLECE,3473730,20
+Vietnam,0.6,1.1,1.7,2011,NLA,97338579,314
+"Yemen, Rep",18.9,93.5,94.7,2011,TIMSS,29825964,56