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 07:08:31 +01:00
parent 47ead135d3
commit a764f45041
83 changed files with 205106 additions and 175 deletions
Download Patch File Download Diff File Expand all files Collapse all files

165
src/test/java/puzzle/DictJavaGeneratorMulti.java
View File

@@ -1,20 +1,27 @@
package puzzle;
import module java.base;
import lombok.Getter;
import lombok.NoArgsConstructor;
import lombok.experimental.Accessors;
import lombok.val;
import puzzle.Export.Dicts;
import puzzle.Export.IntListDTO;
import puzzle.DictJavaGeneratorMulti.DictEntryDTO.IntListDTO;
import puzzle.SwedishGenerator.Dict;
import puzzle.SwedishGenerator.DictEntry;
import puzzle.SwedishGenerator.Lemma;
import static java.nio.charset.StandardCharsets.US_ASCII;
import static puzzle.SwedishGenerator.THRESS;
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 void main(String[] args) throws Exception {
Path wordsFile = Path.of(args.length > 0 ? args[0] : "nl_score_hints_v3.csv");
Path outDir = Path.of(args.length > 1 ? args[1] : "src/main/generated-sources/puzzle");
String pkg = "puzzle";
Path outDir = Path.of(args.length > 1 ? args[1] : "src/main/generated-sources/puzzle/dict" + THRESS);
String pkg = "puzzle.dict" + THRESS;
HashMap<Path, ShardBuilder> builders = new HashMap<Path, ShardBuilder>(16);
SwedishGenerator.Dict dict = buildDict(wordsFile, builders);
@@ -33,10 +40,16 @@ public final class DictJavaGeneratorMulti {
// Aggregator
writeAggregator(outDir, pkg, "DictData", dict.length());
System.out.println("Generated sources into: " + outDir.toAbsolutePath());
builders.forEach(DictJavaGeneratorMulti::writeIndexedShard);
}
static final Path[] SHARDS = IntStream.range(0, 10).mapToObj(sId -> Path.of("src/main/generated-sources/puzzle/dict"+THRESS).resolve(sId + ".idx")).toArray(
Path[]::new);
static Path shardKey(long word) {
return SHARDS[Lemma.unpackSize(word) + 1];
}
private static SwedishGenerator.Dict buildDict(Path wordsPath, HashMap<Path, ShardBuilder> builders) throws IOException {
var map = new Export.LongArrayList(100_000);
var map = new LongArrayList(100_000);
try (var lines = Files.lines(wordsPath, StandardCharsets.UTF_8)) {
lines.forEach(line -> {
CsvIndexService.lineToLemma(line, w -> {
@@ -50,7 +63,7 @@ public final class DictJavaGeneratorMulti {
String recStr = word + "\t" + simpel + "\t" + json + "\n";
byte[] rec = recStr.getBytes(StandardCharsets.UTF_8);
var key = Meta.shardKey(w);
var key = shardKey(w) ;
ShardBuilder sb = builders.computeIfAbsent(key, k -> new ShardBuilder());
try {
map.add(Lemma.pack(w, sb.addRecord(rec)));
@@ -64,7 +77,43 @@ public final class DictJavaGeneratorMulti {
return Dicts.makeDict(map.toArray());
}
static final int VERSION = 1;
interface Dicts {
static Dict makeDict(long[] wordz) {
var index = new DictEntryDTO[SwedishGenerator.MAX_WORD_LENGTH_PLUS_ONE];
Arrays.setAll(index, DictEntryDTO::new);
for (var lemma : wordz) {
var L = Lemma.unpackSize(lemma) + 1;//Lemma.unpackSize(lemma) + 2;
val entry = index[L];
val idx = entry.words().size();
val pos = entry.pos();
entry.words().add(lemma);
int i = 0;
for (long w = lemma & Lemma.LETTER_MASK; w != 0; w >>>= 5, i++) {
pos[i][(int) ((w & 31) - 1)].add(idx);
}
}
for (int i = 2; i < index.length; i++) if (index[i].words().size() <= 0) throw new RuntimeException("No words for length " + i);
return new Dict(Arrays.stream(index).map(i -> {
var words = i.words().toArray();
int numWords = words.length;
int numLongs = (numWords + 63) >>> 6;
var bitsets = new long[i.pos().length * 26][numLongs];
for (int p = 0; p < i.pos().length; p++) {
for (int l = 0; l < 26; l++) {
var list = i.pos()[p][l];
var bs = bitsets[p * 26 + l];
for (int k = 0; k < list.size(); k++) {
int wordIdx = list.data()[k];
bs[wordIdx >>> 6] |= (1L << (wordIdx & 63));
}
}
}
return new DictEntry(words, bitsets, words.length, (words.length + 63) >>> 6);
}).toArray(DictEntry[]::new),
Arrays.stream(index).mapToInt(i -> i.words().size()).sum());
}
}
static final class ShardBuilder {
@@ -79,7 +128,8 @@ public final class DictJavaGeneratorMulti {
}
}
static void writeIndexedShard(Path out, ShardBuilder sb) throws IOException {
static final int VERSION = 1;
static void writeIndexedShard(Path out, ShardBuilder sb) {
int n = sb.offsets.size();
int[] offs = sb.offsets.toArray();
byte[] data = sb.data.toByteArray();
@@ -101,6 +151,8 @@ public final class DictJavaGeneratorMulti {
// data
ch.write(ByteBuffer.wrap(data));
}catch (IOException e){
throw new RuntimeException("Failed to write shard to " + out, e);
}
}
@@ -110,11 +162,11 @@ public final class DictJavaGeneratorMulti {
w.write("package " + pkg + ";\n\n");
w.write("public final class " + cls + " {\n");
w.write(" private " + cls + "() {}\n\n");
w.write(" public static final SwedishGenerator.Dict DICT = build();\n\n");
w.write(" private static SwedishGenerator.Dict build() {\n");
w.write(" SwedishGenerator.DictEntry[] idx = new SwedishGenerator.DictEntry[SwedishGenerator.MAX_WORD_LENGTH_PLUS_ONE];\n");
w.write(" public static final puzzle.SwedishGenerator.Dict DICT" + THRESS + " = build();\n\n");
w.write(" private static puzzle.SwedishGenerator.Dict build() {\n");
w.write(" puzzle.SwedishGenerator.DictEntry[] idx = new puzzle.SwedishGenerator.DictEntry[puzzle.SwedishGenerator.MAX_WORD_LENGTH_PLUS_ONE];\n");
for (int L = 2; L <= 8; L++) w.write(" idx[" + L + "] = DictDataL" + L + ".entry();\n");
w.write(" return new SwedishGenerator.Dict(idx, " + totalLen + ");\n");
w.write(" return new puzzle.SwedishGenerator.Dict(idx, " + totalLen + ");\n");
w.write(" }\n");
w.write("}\n");
}
@@ -217,11 +269,11 @@ public final class DictJavaGeneratorMulti {
w.write(" }\n\n");
// entry
w.write(" public static SwedishGenerator.DictEntry entry() {\n");
w.write(" public static puzzle.SwedishGenerator.DictEntry entry() {\n");
w.write(" long[] wds = words();\n");
w.write(" long[] flat = posFlat();\n");
w.write(" long[][] pos = reshape(flat, ROWS, COLS);\n");
w.write(" return new SwedishGenerator.DictEntry(wds, pos, wds.length, (wds.length + 63) >>> 6);\n");
w.write(" return new puzzle.SwedishGenerator.DictEntry(wds, pos, wds.length, (wds.length + 63) >>> 6);\n");
w.write(" }\n\n");
// helpers
@@ -252,4 +304,87 @@ public final class DictJavaGeneratorMulti {
private static String toLongLiteral(long v) {
return "0x" + Long.toUnsignedString(v, 16) + "L";
}
public static final class CsvIndexService {
static int SIMPEL_IDX = 3;
public static int lineToSimpel(String line) {
var parts = line.split(",", 5);
return Integer.parseInt(parts[SIMPEL_IDX].trim());
}
public static String[] lineToClue(String line) {
if (line.isBlank()) throw new RuntimeException("Empty line");
var parts = line.split(",", 5);
var rawClue = parts[4].trim();
if (rawClue.startsWith("\"") && rawClue.endsWith("\"")) {
rawClue = rawClue.substring(1, rawClue.length() - 1).replace("\"\"", "\"");
}
return Meta.GSON.fromJson(rawClue, String[].class);
}
public static void lineToLemma(String line, LongConsumer ok) {
if (line.isBlank()) {
throw new RuntimeException("Empty line");
}
var parts = line.split(",", 5);
var id = Integer.parseInt(parts[0].trim());
var word = parts[1].trim();
int score = Integer.parseInt(parts[2].trim());
int simpel = Integer.parseInt(parts[3].trim());
if (score < 1 || simpel>THRESS) {
if (Main.VERBOSE) System.err.println("Word too complex: " + line);
return;
}
ok.accept(Lemma.from(word.getBytes(US_ASCII)));
}
}
record DictEntryDTO(LongArrayList words, IntListDTO[][] pos) {
public DictEntryDTO(int L) {
this(new LongArrayList(1024), new IntListDTO[L][26]);
for (var i = 0; i < L; i++) for (var j = 0; j < 26; j++) pos[i][j] = new IntListDTO();
}
@Getter
@Accessors(fluent = true)
@NoArgsConstructor
static final class IntListDTO {
int[] data = new int[8];
int size = 0;
public IntListDTO(int size) {
data = new int[size];
}
void add(int v) {
if (size >= data.length) data = Arrays.copyOf(data, data.length * 2);
data[size++] = v;
}
int[] toArray() { return Arrays.copyOf(data, size); }
}
}
static final class LongArrayList {
long[] a;
int size;
LongArrayList(int initialCapacity) {
if (initialCapacity < 0) throw new IllegalArgumentException();
a = new long[initialCapacity];
}
int size() { return size; }
void add(long v) {
if (size == a.length) grow();
a[size++] = v;
}
void grow() {
int newCap = a.length == 0 ? 1 : a.length * 2;
long[] n = new long[newCap];
System.arraycopy(a, 0, n, 0, size);
a = n;
}
long[] toArray() { return Arrays.copyOf(a, this.size); }
}
}