Align Unicode numbers in Android lab marker keys (rebase of #2552)

Packages/StrandImport/Sources/StrandImport/LabMarkerCsvImport.swift

@@ -356,6 +356,7 @@ public enum LabMarkerCsvImport {
     public static func customKey(_ name: String) -> String {
         let lowered = Array(name.precomposedStringWithCanonicalMapping
             .trimmingCharacters(in: .whitespaces).lowercased())
+        // Kotlin twin: `LabMarkerCsvImport.isWordChar`.
         func isWordChar(_ ch: Character) -> Bool { ch.isLetter || ch.isNumber }
         var mapped = ""
         for (i, ch) in lowered.enumerated() {

Packages/StrandImport/Tests/StrandImportTests/LabMarkerCsvImportTests.swift

@@ -327,13 +327,13 @@ final class LabMarkerCsvImportTests: XCTestCase {
     }
 
     /// Pinned by oracle: the same literals are asserted by the Android twin
-    /// (LabMarkerCsvImportTest.customKeyIsPinned). Names without "%" keep their pre-"pct" keys.
+    /// (LabMarkerCsvImportTest.customKeyIsPinned). The Unicode numeric cases pin #2351 parity.
     func testCustomKeyIsPinned() {
         let names = [
-            "LYMPH", "LYMPH %", "LYMPH%", "Lymph (%)", "BASO", "BASO %", "% Neutrophils", "NEUT  %", "a%%b", "%", "Apo B", "Magnesium", "  ???  ", "A - B", "Vitamin D (25-OH)", "Caf\u{e9} Marker", "Cafe\u{301} Marker", "MCHC g/dL", "HbA1c (IFCC)",
+            "LYMPH", "LYMPH %", "LYMPH%", "Lymph (%)", "BASO", "BASO %", "% Neutrophils", "NEUT  %", "a%%b", "%", "Apo B", "Magnesium", "  ???  ", "A - B", "Vitamin D (25-OH)", "Caf\u{e9} Marker", "Cafe\u{301} Marker", "MCHC g/dL", "HbA1c (IFCC)", "eGFR mL/min/1.73m²", "Factor Ⅻ", "Vitamin B₁₂", "①%", "½ dose", "³", "2", "𐄇",
         ]
         let expected = [
-            "custom_lymph", "custom_lymph_pct", "custom_lymph_pct", "custom_lymph_pct", "custom_baso", "custom_baso_pct", "custom_pct_neutrophils", "custom_neut_pct", "custom_a_pctpct_b", "custom_pct", "custom_apo_b", "custom_magnesium", "", "custom_a__b", "custom_vitamin_d_25_oh", "custom_café_marker", "custom_café_marker", "custom_mchc_g_dl", "custom_hba1c_ifcc",
+            "custom_lymph", "custom_lymph_pct", "custom_lymph_pct", "custom_lymph_pct", "custom_baso", "custom_baso_pct", "custom_pct_neutrophils", "custom_neut_pct", "custom_a_pctpct_b", "custom_pct", "custom_apo_b", "custom_magnesium", "", "custom_a__b", "custom_vitamin_d_25_oh", "custom_café_marker", "custom_café_marker", "custom_mchc_g_dl", "custom_hba1c_ifcc", "custom_egfr_ml_min_1_73m²", "custom_factor_ⅻ", "custom_vitamin_b₁₂", "custom_①_pct", "custom_½_dose", "custom_³", "custom_2", "custom_𐄇",
         ]
         XCTAssertEqual(names.map(LabMarkerCsvImport.customKey), expected)
     }

Tools/parity_twin_map.json

@@ -19,15 +19,15 @@
   },
   "authority": {
     "files": {"count": 507, "sha256": "efc6950759de76273adccde8b6457f5e8577bdaf6ed9c6538cde79b0eaf0c612"},
-    "functions": {"count": 4517, "sha256": "05dc56778990d1a6313fc8429b783110ef534ebe49693d91ab7da9c736df040c"},
+    "functions": {"count": 4518, "sha256": "f424a4e32aceb96b06b2dc790e858bb55b2548fb326c1be1cfb85c6308d013aa"},
     "properties": {"count": 488, "sha256": "77d60fe0f6218252233b9e60da8bf5135968e49be3c6d2027d0973ea20ddf792"},
     "constants": {"count": 1985, "sha256": "0c67e7c378db2c0518339f073b37967e30cec0d5b586c1408841ca5e7977f9b4"},
-    "file_pairs": {"count": 78, "sha256": "ec0d88b4e0535c3b1051b5c10205f21174944018669fbe3479a3c21801a6dc38"},
-    "function_pairs": {"count": 215, "sha256": "f4ffaafbdf720d27594002cc2337a7b1be5635ec8d02720b7722f01656e8720b"},
+    "file_pairs": {"count": 79, "sha256": "02ee127080cf87c98925a60b30b6e87d25e7a82ef6cf43b6251fe24c20600dd1"},
+    "function_pairs": {"count": 217, "sha256": "c31664b757a428dcadbf56068d92c681397dc2f378e5e6d6ea65be6e70c3a381"},
     "property_pairs": {"count": 163, "sha256": "b7d4a29086beec661993370657c0bcdae86f79b221fdc0043206161df5e1117f"},
     "constant_pairs": {"count": 694, "sha256": "ae11bdde04d49f9dfe83170f0aa362e6bf18d3fdc381c1bdf5bbc256caa400ab"},
-    "unpaired_files": {"count": 373, "sha256": "2b0c9bc717b301037f735ea7cc32f7a7cd9b4a7887d46550b9e727ffd66d3639"},
-    "unpaired_functions": {"count": 4100, "sha256": "cc10b9887488dddae0c270071d3fde36a0c6c764df318e18a27e569e181051a9"},
+    "unpaired_files": {"count": 371, "sha256": "d9e682eadcdd5b1279a69a45799fd68b4589f98c83cfb7e15d3016ad890a98c0"},
+    "unpaired_functions": {"count": 4097, "sha256": "e54745434cdd990e42445672ec7ae137bfe1111def4f701948fa1ae7398e5871"},
     "unpaired_properties": {"count": 162, "sha256": "dabff97ba9bce0aebdc8ef84b9c4f107ce574020e8b0455d8f9214082948246b"},
     "unpaired_constants": {"count": 597, "sha256": "578c607cc05b1cd1f867c206c9fab70237ac40707cd49d4e0c5cc66b400184c7"}
   }

android/app/src/main/java/com/noop/ingest/LabMarkerCsvImport.kt

@@ -453,19 +453,27 @@ object LabMarkerCsvImport {
      * other punctuation gave both `custom_lymph`, so the second row of the day overwrote the
      * first. The token is joined to a letter/digit neighbour by one "_" (a non-alphanumeric
      * neighbour already maps to "_"), so "LYMPH %", "LYMPH%" and "Lymph (%)" all give
-     * `custom_lymph_pct`. Names without "%" keep the keys they always had. Byte-identical to
-     * Swift `LabMarkerCsvImport.customKey`.
+     * `custom_lymph_pct`. Names without "%" keep their keys except those containing Unicode
+     * number letters or other numbers, which Android previously replaced with underscores (#2351).
+     * Swift twin: `LabMarkerCsvImport.customKey`. NFC-composable names produce the same key;
+     * Swift's grapheme clusters and Kotlin's code points can differ for uncomposed marks.
      */
     internal fun customKey(name: String): String {
         val lowered = Normalizer.normalize(name, Normalizer.Form.NFC).trim().lowercase()
+        // Swift Character.isNumber also keeps Unicode letter/other numbers (e.g. Ⅻ, ²).
+        // Swift twin: `LabMarkerCsvImport.isWordChar`.
+        fun isWordChar(cp: Int): Boolean = Character.isLetterOrDigit(cp) ||
+            Character.getType(cp) == Character.LETTER_NUMBER.toInt() ||
+            Character.getType(cp) == Character.OTHER_NUMBER.toInt()
+        val chars = lowered.codePoints().toArray()
         val mapped = buildString {
-            lowered.forEachIndexed { i, ch ->
-                if (ch == '%') {
-                    if (i > 0 && lowered[i - 1].isLetterOrDigit()) append('_')
+            chars.forEachIndexed { i, ch ->
+                if (ch == '%'.code) {
+                    if (i > 0 && isWordChar(chars[i - 1])) append('_')
                     append("pct")
-                    if (i + 1 < lowered.length && lowered[i + 1].isLetterOrDigit()) append('_')
+                    if (i + 1 < chars.size && isWordChar(chars[i + 1])) append('_')
                 } else {
-                    append(if (ch.isLetterOrDigit()) ch else '_')
+                    if (isWordChar(ch)) appendCodePoint(ch) else append('_')
                 }
             }
         }

android/app/src/test/java/com/noop/ingest/LabMarkerCsvImportTest.kt

@@ -346,15 +346,15 @@ class LabMarkerCsvImportTest {
 
     /**
      * Expected literals are the verbatim stdout of Swift `LabMarkerCsvImport.customKey` compiled
-     * standalone over [names] (the Swift suite pins the same list). Names without "%" keep their
-     * pre-"pct" keys.
+     * standalone over [names] (the Swift suite pins the same list). The Unicode numeric cases
+     * guard the Android/Swift category boundary from #2351.
      */
     @Test fun customKeyIsPinned() {
         val names = listOf(
-            "LYMPH", "LYMPH %", "LYMPH%", "Lymph (%)", "BASO", "BASO %", "% Neutrophils", "NEUT  %", "a%%b", "%", "Apo B", "Magnesium", "  ???  ", "A - B", "Vitamin D (25-OH)", "Caf\u00e9 Marker", "Cafe\u0301 Marker", "MCHC g/dL", "HbA1c (IFCC)",
+            "LYMPH", "LYMPH %", "LYMPH%", "Lymph (%)", "BASO", "BASO %", "% Neutrophils", "NEUT  %", "a%%b", "%", "Apo B", "Magnesium", "  ???  ", "A - B", "Vitamin D (25-OH)", "Caf\u00e9 Marker", "Cafe\u0301 Marker", "MCHC g/dL", "HbA1c (IFCC)", "eGFR mL/min/1.73m²", "Factor Ⅻ", "Vitamin B₁₂", "①%", "½ dose", "³", "2", "𐄇",
         )
         val expected = listOf(
-            "custom_lymph", "custom_lymph_pct", "custom_lymph_pct", "custom_lymph_pct", "custom_baso", "custom_baso_pct", "custom_pct_neutrophils", "custom_neut_pct", "custom_a_pctpct_b", "custom_pct", "custom_apo_b", "custom_magnesium", "", "custom_a__b", "custom_vitamin_d_25_oh", "custom_café_marker", "custom_café_marker", "custom_mchc_g_dl", "custom_hba1c_ifcc",
+            "custom_lymph", "custom_lymph_pct", "custom_lymph_pct", "custom_lymph_pct", "custom_baso", "custom_baso_pct", "custom_pct_neutrophils", "custom_neut_pct", "custom_a_pctpct_b", "custom_pct", "custom_apo_b", "custom_magnesium", "", "custom_a__b", "custom_vitamin_d_25_oh", "custom_café_marker", "custom_café_marker", "custom_mchc_g_dl", "custom_hba1c_ifcc", "custom_egfr_ml_min_1_73m²", "custom_factor_ⅻ", "custom_vitamin_b₁₂", "custom_①_pct", "custom_½_dose", "custom_³", "custom_2", "custom_𐄇",
         )
         assertEquals(expected, names.map { LabMarkerCsvImport.customKey(it) })
     }