[DCP - SigFigs] Preserve SigFigs through Dataflow. Stop converting to double/int

pipeline/util/src/main/java/org/datacommons/ingestion/util/GraphReader.java

@@ -496,10 +496,12 @@ static TimeSeriesKey toTimeSeriesKey(McfStatVarObsSeries.Key key, String importN
 
   static Observation toObservation(TimeSeriesKey seriesKey, StatVarObs obs) {
     String value = "";
-    if (obs.hasNumber()) {
-      value = Double.toString(obs.getNumber());
-    } else if (obs.hasText()) {
+    // Prefer text representation to preserve significant figures.
+    // Fall back to legacy number field for backward compatibility with older serialized protos.
+    if (obs.hasText()) {
       value = obs.getText();
+    } else if (obs.hasNumber()) {
+      value = Double.toString(obs.getNumber());
     }
 
     return Observation.builder().seriesKey(seriesKey).date(obs.getDate()).value(value).build();

pipeline/util/src/test/java/org/datacommons/ingestion/util/GraphReaderTest.java

@@ -389,18 +389,38 @@ public void testToObservation() {
         StatVarObs.newBuilder().setDcid("obs1").setDate("2020").setNumber(10.0).build();
     StatVarObs obs2 =
         StatVarObs.newBuilder().setDcid("obs2").setDate("2021").setText("someText").build();
+    StatVarObs obs3 =
+        StatVarObs.newBuilder().setDcid("obs3").setDate("2022").setText("12.5000").build();
+    StatVarObs obs4 =
+        StatVarObs.newBuilder()
+            .setDcid("obs4")
+            .setDate("2023")
+            .setText("100000000000000001")
+            .build();
 
     Observation expected1 =
         Observation.builder().seriesKey(seriesKey).date("2020").value("10.0").build();
 
     Observation expected2 =
         Observation.builder().seriesKey(seriesKey).date("2021").value("someText").build();
 
+    Observation expected3 =
+        Observation.builder().seriesKey(seriesKey).date("2022").value("12.5000").build();
+
+    Observation expected4 =
+        Observation.builder().seriesKey(seriesKey).date("2023").value("100000000000000001").build();
+
     Observation actual1 = GraphReader.toObservation(seriesKey, obs1);
     assertEquals(expected1, actual1);
 
     Observation actual2 = GraphReader.toObservation(seriesKey, obs2);
     assertEquals(expected2, actual2);
+
+    Observation actual3 = GraphReader.toObservation(seriesKey, obs3);
+    assertEquals(expected3, actual3);
+
+    Observation actual4 = GraphReader.toObservation(seriesKey, obs4);
+    assertEquals(expected4, actual4);
   }
 
   @Test

pipeline/util/src/test/java/org/datacommons/ingestion/util/PipelineUtilsTest.java

@@ -62,10 +62,10 @@ private McfGraph createStatVarObservationGraph(
     return graph.build();
   }
 
-  private McfStatVarObsSeries.StatVarObs createStatVarObs(String date, double value, String dcid) {
+  private McfStatVarObsSeries.StatVarObs createStatVarObs(String date, String value, String dcid) {
     McfStatVarObsSeries.StatVarObs.Builder svObs = McfStatVarObsSeries.StatVarObs.newBuilder();
     svObs.setDate(date);
-    svObs.setNumber(value);
+    svObs.setText(value);
     svObs.setDcid(dcid);
     svObs.setPvs(PropertyValues.newBuilder().build());
     return svObs.build();
@@ -114,23 +114,53 @@ public void testBuildOptimizedMcfGraph() {
                     "count_person",
                     "country/USA",
                     Arrays.asList(
-                        createStatVarObs("2020", 32.0, "obs1"),
-                        createStatVarObs("2021", 33.0, "obs2"))))
+                        createStatVarObs("2020", "32.0", "obs1"),
+                        createStatVarObs("2021", "33.0", "obs2"))))
             .build();
     McfOptimizedGraph expected2 =
         McfOptimizedGraph.newBuilder()
             .setSvObsSeries(
                 createMcfStatVarObsSeries(
                     "count_person",
                     "country/India",
-                    List.of(createStatVarObs("2022", 36.0, "obs4"))))
+                    List.of(createStatVarObs("2022", "36.0", "obs4"))))
             .build();
 
     PAssert.that(result).containsInAnyOrder(expected1, expected2);
     PipelineResult.State state = p.run().waitUntilFinish();
     Assert.assertEquals(PipelineResult.State.DONE, state);
   }
 
+  @Test
+  public void testBuildOptimizedMcfGraph_preservesSigFigs() {
+    options.setStableUniqueNames(PipelineOptions.CheckEnabled.OFF);
+    p.getCoderRegistry()
+        .registerCoderForClass(
+            McfStatVarObsSeries.Key.class, ProtoCoder.of(McfStatVarObsSeries.Key.class));
+
+    PCollection<McfGraph> input =
+        p.apply(
+            Create.of(
+                createStatVarObservationGraph(
+                    "obsSigFig", "measurement_rate", "country/USA", "2020", "12.5000")));
+
+    PCollection<McfOptimizedGraph> result =
+        PipelineUtils.buildOptimizedMcfGraph("testSigFig", input);
+
+    McfOptimizedGraph expected =
+        McfOptimizedGraph.newBuilder()
+            .setSvObsSeries(
+                createMcfStatVarObsSeries(
+                    "measurement_rate",
+                    "country/USA",
+                    List.of(createStatVarObs("2020", "12.5000", "obsSigFig"))))
+            .build();
+
+    PAssert.that(result).containsInAnyOrder(expected);
+    PipelineResult.State state = p.run().waitUntilFinish();
+    Assert.assertEquals(PipelineResult.State.DONE, state);
+  }
+
   @Test
   public void testCombineGraphNodes() {
     // Input Graph 1

util/src/main/java/org/datacommons/util/GraphUtils.java

@@ -191,25 +191,7 @@ public static List<String> getPropertyValues(Map<String, McfGraph.Values> pvs, S
   }
 
   /**
-   * Gets the double value of a specific property from a graph node.
-   *
-   * @param node The graph node (PropertyValues) to read from.
-   * @param prop The property name whose value should be retrieved as a double.
-   * @return The double value of the property, or Double.NaN if the value is not a valid number.
-   */
-  public static Double nodeDoubleValue(McfGraph.PropertyValues node, String prop) {
-    String str_val = getPropVal(node, prop);
-    if (str_val.isEmpty()) throw new IllegalArgumentException("Failed to get double value.");
-    try {
-      double v = Double.parseDouble(str_val);
-      return v;
-    } catch (NumberFormatException nfe) {
-      return Double.NaN;
-    }
-  }
-
-  /**
-   * Flattens an optimized MCF graph into a list of graph node
+   * Flattens an optimized MCF graph into a list of graph nodes.
    *
    * @param optimized_graph input optimized graph
    * @return list of McfGraph instances, each representing a single StatVarObservation.
@@ -256,10 +238,14 @@ public static List<McfGraph> convertMcfStatVarObsSeriesToMcfGraph(
       // Set required PVs.
       setPropVal(Property.dcid.name(), ValueType.TEXT, o.getDcid(), node);
       setPropVal(Property.observationDate.name(), ValueType.TEXT, o.getDate(), node);
-      if (o.hasNumber()) {
+      // Prefer text representation to preserve significant figures (SigFigs).
+      // Fall back to legacy number field for backward compatibility with older serialized protos.
+      if (o.hasText()) {
+        String valText = o.getText();
+        ValueType valType = StringUtil.isNumber(valText) ? ValueType.NUMBER : ValueType.TEXT;
+        setPropVal(Property.value.name(), valType, valText, node);
+      } else if (o.hasNumber()) {
         setPropVal(Property.value.name(), ValueType.NUMBER, Double.toString(o.getNumber()), node);
-      } else if (o.hasText()) {
-        setPropVal(Property.value.name(), ValueType.TEXT, o.getText(), node);
       }
 
       // Set optional PVs.
@@ -329,12 +315,8 @@ public static McfStatVarObsSeries convertMcfGraphToMcfStatVarObsSeries(
     if (!useDcidForLocalNodeIdInOptimizedMcf(dcid, nodeId)) {
       svo.setLocalNodeId(nodeId);
     }
-    Double value;
-    if (!(value = nodeDoubleValue(node, "value")).isNaN()) {
-      svo.setNumber(value);
-    } else { // Non-number value.
-      svo.setText(getPropVal(node, "value"));
-    }
+    // Preserve raw string representation to maintain significant figures (SigFigs).
+    svo.setText(getPropVal(node, "value"));
     McfGraph.PropertyValues.Builder pvs = svo.getPvsBuilder();
     for (Map.Entry<String, McfGraph.Values> entry : node.getPvsMap().entrySet()) {
       String prop = entry.getKey();

util/src/main/java/org/datacommons/util/StatChecker.java

@@ -81,10 +81,9 @@ public class StatChecker {
   private final ConcurrentMap<String, Set<String>> samplePlaces;
   private final boolean shouldGenerateSamplePlaces;
   // Tracks global state on StatVarObservations to detect whether there are multiple of the
-  // same StatVarObservation with inconsistent values. The key is a hash made up of a set of
-  // properties that distinguish a StatVarObservation and the value is the first value seen of that
-  // StatVarObservation.
-  private final ConcurrentMap<Long, Float> svObValues;
+  // same StatVarObservation with inconsistent values. Stored as raw String to preserve
+  // significant figures (SigFigs) and avoid floating-point precision collisions.
+  private final ConcurrentMap<Long, String> svObValues;
   private final String EMPTY_PROP_STRING = "EMPTY_PROP";
   private StatVarState statVarState;
   private ExistenceChecker existenceChecker;
@@ -604,12 +603,7 @@ private boolean checkSvObsValueInconsistency(McfGraph.PropertyValues node) {
       }
     }
     Long fp = hasher.hash().asLong();
-    Float val = null;
-    try {
-      val = Float.parseFloat(McfUtil.getPropVal(node, Vocabulary.VALUE));
-    } catch (NumberFormatException e) {
-      // If value is not a float, val will stay as null and this will be handled later.
-    }
+    String val = McfUtil.getPropVal(node, Vocabulary.VALUE);
     if (this.svObValues.containsKey(fp) && !this.svObValues.get(fp).equals(val)) {
       logCtx.addEntry(
           Level.LEVEL_ERROR,

util/src/test/java/org/datacommons/util/GraphUtilsTest.java

@@ -107,7 +107,7 @@ public void testBuildOptimizedMcfGraph_singleObservation() {
     McfStatVarObsSeries.StatVarObs obs = series.getSvObsList(0);
     assertEquals("2023-01-15", obs.getDate());
     assertEquals("obs1", obs.getDcid());
-    assertEquals(100.5, obs.getNumber(), 0.001);
+    assertEquals("100.5", obs.getText());
     assertEquals("A test observation", GraphUtils.getPropVal(obs.getPvs(), "description"));
   }
 
@@ -169,12 +169,12 @@ public void testBuildOptimizedMcfGraph_multipleObservations_sameKey() {
     McfStatVarObsSeries.StatVarObs obs1 = series.getSvObsList(0);
     assertEquals("2023-01-15", obs1.getDate());
     assertEquals("obs1", obs1.getDcid());
-    assertEquals(100.5, obs1.getNumber(), 0.001);
+    assertEquals("100.5", obs1.getText());
 
     McfStatVarObsSeries.StatVarObs obs2 = series.getSvObsList(1);
     assertEquals("2023-02-15", obs2.getDate());
     assertEquals("obs2", obs2.getDcid());
-    assertEquals(102.0, obs2.getNumber(), 0.001);
+    assertEquals("102.0", obs2.getText());
   }
 
   @Test
@@ -216,4 +216,35 @@ public void testBuildOptimizedMcfGraph_multipleObservations_differentKeys() {
     assertNotNull(result);
     assertEquals(2, result.size());
   }
+
+  @Test
+  public void testBuildOptimizedMcfGraph_preservesSigFigsAndLargeNumbers() {
+    McfGraph.PropertyValues svoNode =
+        createSVObsNode(
+            "obsSigFig",
+            "dcid:placeA",
+            "dcid:svPrecision",
+            "2023-01-15",
+            "12.5000",
+            ValueType.NUMBER,
+            "P1M",
+            "dcid:methodA",
+            "dcid:unitX",
+            "1",
+            null);
+
+    McfGraph mcfGraph =
+        McfGraph.newBuilder().setType(McfType.INSTANCE_MCF).putNodes("l:node1", svoNode).build();
+
+    List<McfOptimizedGraph> result = GraphUtils.buildOptimizedMcfGraph(List.of(mcfGraph));
+    assertEquals(1, result.size());
+    McfStatVarObsSeries.StatVarObs obs = result.get(0).getSvObsSeries().getSvObsList(0);
+    assertEquals("12.5000", obs.getText());
+
+    // Also verify roundtrip back to McfGraph
+    List<McfGraph> roundTripGraphs = GraphUtils.convertMcfStatVarObsSeriesToMcfGraph(result.get(0));
+    assertEquals(1, roundTripGraphs.size());
+    McfGraph.PropertyValues node = roundTripGraphs.get(0).getNodesOrThrow("l:node1");
+    assertEquals("12.5000", GraphUtils.getPropVal(node, "value"));
+  }
 }

util/src/test/java/org/datacommons/util/StatCheckerTest.java

@@ -125,7 +125,7 @@ public void testCheckSvObsInGraph() throws IOException {
         TestUtil.checkLog(
             lw.getLog(),
             "Sanity_InconsistentSvObsValues",
-            "Found nodes with different values for the same StatVarObservation :: observationAbout: 'geoId/SF', variableMeasured: 'WomenIncome', observationDate: '2020', value1: 1.0E7, value2: 1.0000001E7"));
+            "Found nodes with different values for the same StatVarObservation :: observationAbout: 'geoId/SF', variableMeasured: 'WomenIncome', observationDate: '2020', value1: 10000000.0, value2: 10000001.0"));
 
     // check node that differs only in one property from an existing StatVarObservation node.
     mcf =