ICU-23214 generate brkitr/root.txt lstm from data

- update cldr-icu.md
- see TODOs in ICU-23215

Author: srl295

docs/processes/cldr-icu.md

@@ -297,17 +297,8 @@ ant copy-cldr-testdata
 5d. NOP
 (This step has been subsumed into 5c above)
 
-5e. For now, manually re-add the `lstm` entries in `data/brkitr/root.txt`
-```sh
-open $ICU4C_DIR/source/data/brkitr/root.txt
-```
-Paste the following block after the dictionaries block and before the final closing '}':
-```
-    lstm{
-        Thai{"Thai_graphclust_model4_heavy.res"}
-        Mymr{"Burmese_graphclust_model5_heavy.res"}
-    }
-```
+5e. NOP
+(This step is no longer necessary, see [ICU-23215](https://unicode-org.atlassian.net/browse/ICU-23215) for details.)
 
 5f. Update hard-coded lists in ICU
 
@@ -499,7 +490,7 @@ rebuilding of other kinds of data and/or code. For example:
 
 If you see a failure such as
 ```
-MeasureUnitTest	testCLDRUnitAvailability	Failure	(MeasureUnitTest.java:3410) : Unit present in CLDR but not available via constant in MeasureUnit: speed-beaufort 
+MeasureUnitTest	testCLDRUnitAvailability	Failure	(MeasureUnitTest.java:3410) : Unit present in CLDR but not available via constant in MeasureUnit: speed-beaufort
 ```
 then you will need to update the C and J library and test code for new measurement
 units, see the procedure at

icu4c/source/data/brkitr/root.txt

@@ -23,7 +23,7 @@ root{
         Thai:process(dependency){"thaidict.dict"}
     }
     lstm{
-        Thai{"Thai_graphclust_model4_heavy.res"}
         Mymr{"Burmese_graphclust_model5_heavy.res"}
+        Thai{"Thai_graphclust_model4_heavy.res"}
     }
 }

icu4c/source/data/xml/brkitr/root.xml

@@ -13,7 +13,7 @@
 <ldml>
     <identity>
         <version number="$Revision$"/>
-        <language type="root"/> 
+        <language type="root"/>
     </identity>
     <special xmlns:icu="http://www.icu-project.org/">
         <icu:breakIteratorData>
@@ -36,13 +36,10 @@
                 <icu:dictionary type="Mymr" icu:dependency="burmesedict.dict"/>
                 <icu:dictionary type="Thai" icu:dependency="thaidict.dict"/>
             </icu:dictionaries>
-            <!--
             <icu:lstm>
                 <icu:lstmdata type="Thai" icu:dependency="Thai_graphclust_model4_heavy.res"/>
                 <icu:lstmdata type="Mymr" icu:dependency="Burmese_graphclust_model5_heavy.res"/>
             </icu:lstm>
-            -->
         </icu:breakIteratorData>
     </special>
 </ldml>
-

icu4c/source/test/testdata/BUILDRULES.py

@@ -44,7 +44,7 @@ def generate_rb(config, io, common_vars):
         "testaliases",
         "testempty",
         "testtypes",
-        # LSTM models
+        # LSTM models - TODO ICU-23215 this should be dynamic
         "Thai_graphclust_model4_heavy",
         "Thai_codepoints_exclusive_model5_heavy",
         "Burmese_graphclust_model5_heavy"

tools/cldr/cldr-to-icu/src/main/java/org/unicode/icu/tool/cldrtoicu/mapper/BreakIteratorMapper.java

@@ -110,7 +110,7 @@ private void addExtension(CldrValue v) {
 
     private void addLstmdata(CldrValue v) {
         //System.out.println("addLstmdata: " + v.toString()); // debug
-        addDependency(
+        addLstmDependency(
             getDependencyName(v),
             LSTMDATA_TYPE.valueFrom(v),
             LSTMDATA_DEP.optionalValueFrom(v));
@@ -123,6 +123,15 @@ private void addDependency(String name, String type, Optional<String> dependency
             dependency.orElseThrow(() -> new IllegalArgumentException("missing dependency")));
     }
 
+    private void addLstmDependency(String name, String type, Optional<String> dependency) {
+        // this should be :process(dependency) but that's not what ICU expects
+        // BUILDRULES.py hard codes the inclusion of the .res files, see the TODO ticket.
+        System.out.println("addLstmDependency: name " + name + ", type " + type + ", dependency " + dependency + " - see ICU-23215");
+        icuData.add(
+            RbPath.of(name, type /* + ":process(dependency)") */), // TODO ICU-23215
+            dependency.orElseThrow(() -> new IllegalArgumentException("missing lstm dependency")));
+    }
+
     // Must match the BOUNDARIES or DICTIONARY path.
     private static String getDependencyName(CldrValue value) {
         return stripXmlNamespace(value.getPath().getParent().getName());