tour/puzzle-generator
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

introduce bitloops

Browse Source
This commit is contained in:
mike
2026-01-20 19:00:27 +01:00
parent 58b8b57688
commit dadde53f76
28 changed files with 42000 additions and 41654 deletions
Download Patch File Download Diff File Expand all files Collapse all files

62
src/test/java/puzzle/DictJavaGeneratorMulti.java
View File

@@ -5,7 +5,6 @@ import lombok.Getter;
import lombok.NoArgsConstructor;
import lombok.experimental.Accessors;
import lombok.val;
import puzzle.DictJavaGeneratorMulti.DictEntryDTO.IntListDTO;
import puzzle.SwedishGenerator.Dict;
import puzzle.SwedishGenerator.DictEntry;
import puzzle.SwedishGenerator.Lemma;
@@ -16,14 +15,14 @@ public final class DictJavaGeneratorMulti {
// Smaller = more files, but safer for javac/class limits.
private static final int WORDS_CHUNK = 8_192;
private static final int POS_CHUNK = 8_192;
public static final int THRESS = 800;
public static void main(String[] args) throws Exception {
Path wordsFile = Path.of(args.length > 0 ? args[0] : "nl_score_hints_v4.csv");
Path outDir = Path.of(args.length > 1 ? args[1] : "src/main/generated-sources/puzzle/dict" + THRESS);
String pkg = "puzzle.dict" + THRESS;
Path wordsFile = Path.of(args.length > 0 ? args[0] : "nl_score_hints_v4.csv");
Path outDir = Path.of(args.length > 1 ? args[1] : "src/main/generated-sources/puzzle/dict" + THRESS);
String pkg = "puzzle.dict" + THRESS;
HashMap<String, ShardBuilder> builders = new HashMap<String, ShardBuilder>(16);
SwedishGenerator.Dict dict = buildDict(wordsFile, builders);
SwedishGenerator.Dict dict = buildDict(wordsFile, builders, THRESS);
Files.createDirectories(outDir);
@@ -41,34 +40,30 @@ public final class DictJavaGeneratorMulti {
System.out.println("Generated sources into: " + outDir.toAbsolutePath());
}
public static final int THRESS = 800;
static String shardKey(long word) {
return ""+Lemma.unpackSize(word) + 1;
return "" + Lemma.unpackSize(word) + 1;
}
private static SwedishGenerator.Dict buildDict(Path wordsPath, HashMap<String, ShardBuilder> builders) throws IOException {
private static SwedishGenerator.Dict buildDict(Path wordsPath, HashMap<String, ShardBuilder> builders, int thress) throws IOException {
var map = new LongArrayList(100_000);
try (var lines = Files.lines(wordsPath, StandardCharsets.UTF_8)) {
lines.forEach(line -> {
CsvIndexService.lineToLemma(line, w -> {
String word = Lemma.asWord(w, Export.BYTES.get());
String[] clues = CsvIndexService.lineToClue(line);
int simpel = CsvIndexService.lineToSimpel(line);
// serialize to: WORD \t JSON \n
// (als je al JSON string wilt bewaren: gebruik Gson.toJson(clues))
String json = Meta.GSON.toJson(clues);
String recStr = word + "\t" + simpel + "\t" + json + "\n";
byte[] rec = recStr.getBytes(StandardCharsets.UTF_8);
var key = shardKey(w);
ShardBuilder sb = builders.computeIfAbsent(key, k -> new ShardBuilder());
try {
map.add(Lemma.pack(w, sb.addRecord(rec)));
} catch (IOException e) {
throw new UncheckedIOException(e);
}
});
var parts = line.split(",", 4);
var word = parts[0].trim();
long w = SwedishGenerator.Lemma.from(word.getBytes(US_ASCII));
if (!word.equals(SwedishGenerator.Lemma.asWord(w, Export.BYTES.get()))) {
throw new RuntimeException();
}
int score = Integer.parseInt(parts[1].trim());
var simpel = Integer.parseInt(parts[CsvIndexService.SIMPEL_IDX].trim());
if (score < 1 || simpel > thress) {
if (Main.VERBOSE) System.err.println("Word too complex: " + line);
return;
}
var key = shardKey(w);
ShardBuilder sb = builders.computeIfAbsent(key, k -> new ShardBuilder());
map.add(Lemma.pack(w, sb.addRecord()));
});
}
return Dicts.makeDict(map.toArray());
@@ -114,13 +109,10 @@ public final class DictJavaGeneratorMulti {
static final class ShardBuilder {
final IntListDTO offsets = new IntListDTO(4096);
final ByteArrayOutputStream data = new ByteArrayOutputStream(1 << 20); // grows
int addRecord(byte[] rec) throws IOException {
var size = data.size();
val currSize = offsets.size();
offsets.add(size);
data.write(rec);
int c;
int addRecord() {
val currSize = c;
c++;
return currSize;
}
}