Where one of the range endpoints is mapped with `addr:flats` and the range itself has no `addr:flats` then the range is removed (for example at _116 Anderson Street, South Yarra_)

3 files changed, 29 insertions, 5 deletions

diff --git a/bin/reduceDuplicates.js b/bin/reduceDuplicates.js
index 0880abf..e4373c1 100755
--- a/bin/reduceDuplicates.js
+++ b/bin/reduceDuplicates.js
@@ -45,7 +45,6 @@ const index = new Transform({
     }
 
     const key = [
-      // feature.properties['addr:unit:prefix'],
       feature.properties['addr:unit'],
       feature.properties['addr:housenumber'],
       feature.properties['addr:street'],
diff --git a/bin/reduceOverlap.js b/bin/reduceOverlap.js
index 90ca036..7414241 100755
--- a/bin/reduceOverlap.js
+++ b/bin/reduceOverlap.js
@@ -1,5 +1,9 @@
 #!/usr/bin/env node
 
+/**
+ * Reduce overlapping features which can be combined together into a single feature.
+ */
+
 const fs = require('fs')
 const { Readable, Transform, pipeline } = require('stream')
 const ndjson = require('ndjson')
@@ -85,7 +89,7 @@ const reduce = new Transform({
       // multiple features with the same geometry
 
       // group by housenumber, street, suburb, state, postcode to reduce units into addr:flats
-      // groupedFeatures all all the features at the same point
+      // groupedFeatures is all the features at the same point
       const featuresGroupByNonUnit = {}
       groupedFeatures.forEach(feature => {
         const key = [
@@ -175,7 +179,9 @@ const reduce = new Transform({
 })
 
 /**
- * limit values
+ * Per https://wiki.openstreetmap.org/wiki/API_v0.6#Maximum_string_lengths
+ * tag values are limited to 255 characters. Because some addr:flats values can
+ * exceed this, they are split into addr:flatsN tags.
  */
 let limitValuesIndex = 0
 const limitValues = new Transform({
diff --git a/bin/reduceRangeDuplicates.js b/bin/reduceRangeDuplicates.js
index 89f8a11..6fac0b5 100755
--- a/bin/reduceRangeDuplicates.js
+++ b/bin/reduceRangeDuplicates.js
@@ -134,6 +134,9 @@ const reduceRange = new Transform({
         let foundStart = false
         let foundEnd = false
 
+        let matchedStart
+        let matchedEnd
+
         let startNum
         let endNum
         let pre = ''
@@ -142,6 +145,7 @@ const reduceRange = new Transform({
         for (const matchCandidate of matchCandidates) {
           if (!foundStart && start === matchCandidate.properties['addr:housenumber']) {
             foundStart = true
+            matchedStart = matchCandidate
 
             const match = start.match(regexp)
             if (match && match.groups) {
@@ -152,6 +156,7 @@ const reduceRange = new Transform({
           }
           if (!foundEnd && end === matchCandidate.properties['addr:housenumber']) {
             foundEnd = true
+            matchedEnd = matchCandidate
 
             const match = end.match(regexp)
             if (match && match.groups) {
@@ -203,8 +208,22 @@ const reduceRange = new Transform({
           // keep track of removed features for filter B, so we don't double remove both range and midpoints
           rangesRemovedInFilterA[hash(feature)] = true
         } else {
-          // since not both start and end found, then still include the range
-          this.push(feature)
+          // not both start and end found,
+          // if one of start or end found and that start/end has addr:flats...
+          if (foundStart || foundEnd) {
+            // ...if the range has no flats AND the non-range has addr:flats
+            if (!feature.properties['addr:flats'] && (
+              (matchedStart && matchedStart.properties['addr:flats']) || (matchedEnd && matchedEnd.properties['addr:flats'])
+            )) {
+              // drop the range, eg "112-116 Anderson Street, South Yarra"
+            } else {
+              // then still include the range
+              this.push(feature)
+            }
+          } else {
+            // then still include the range
+            this.push(feature)
+          }
         }
       } else {
         // there are no non-ranges on this street so still include the range