Implement time ago parser and improve localization handling

- Handle special cases for languages where the number is not shown
- Rework the Downloader base implementation, allowing for more
advanced things to be done
- Separate the localization from the content country (just like
YouTube let's the user choose both).
This commit is contained in:
Mauricio Colli 2019-04-28 17:03:16 -03:00
parent 180836c180
commit 3638f0e0ea
No known key found for this signature in database
GPG key ID: F200BFD6F29DDD85
274 changed files with 4770 additions and 3468 deletions

View file

@ -0,0 +1,116 @@
import com.grack.nanojson.JsonArray;
import com.grack.nanojson.JsonObject;
import com.grack.nanojson.JsonParser;
import com.grack.nanojson.JsonParserException;
import org.schabi.newpipe.extractor.timeago.PatternsHolder;
import org.schabi.newpipe.extractor.timeago.TimeAgoUnit;
import java.io.*;
import java.util.*;
public class GeneratePatternClasses {
public static void main(String[] args) throws FileNotFoundException, JsonParserException {
final InputStream resourceAsStream =
new FileInputStream("timeago-parser/raw/unique_patterns.json");
final JsonObject from = JsonParser.object().from(resourceAsStream);
final TreeMap<String, Object> map = new TreeMap<>(from);
for (Map.Entry<String, Object> entry : map.entrySet()) {
final String languageCode = entry.getKey().replace('-', '_');
final Map<String, Object> unitsList = (Map<String, Object>) entry.getValue();
final String wordSeparator = (String) unitsList.get("word_separator");
final JsonArray seconds = (JsonArray) unitsList.get("seconds");
final JsonArray minutes = (JsonArray) unitsList.get("minutes");
final JsonArray hours = (JsonArray) unitsList.get("hours");
final JsonArray days = (JsonArray) unitsList.get("days");
final JsonArray weeks = (JsonArray) unitsList.get("weeks");
final JsonArray months = (JsonArray) unitsList.get("months");
final JsonArray years = (JsonArray) unitsList.get("years");
final StringBuilder specialCasesString = new StringBuilder();
specialCasesConstruct(TimeAgoUnit.SECONDS, seconds, specialCasesString);
specialCasesConstruct(TimeAgoUnit.MINUTES, minutes, specialCasesString);
specialCasesConstruct(TimeAgoUnit.HOURS, hours, specialCasesString);
specialCasesConstruct(TimeAgoUnit.DAYS, days, specialCasesString);
specialCasesConstruct(TimeAgoUnit.WEEKS, weeks, specialCasesString);
specialCasesConstruct(TimeAgoUnit.MONTHS, months, specialCasesString);
specialCasesConstruct(TimeAgoUnit.YEARS, years, specialCasesString);
System.out.println("Generating \"" + languageCode + "\" pattern class...");
try (final FileWriter fileOut = new FileWriter(
"timeago-parser/src/main/java/org/schabi/newpipe/extractor/timeago/patterns/" +
languageCode + ".java")) {
final String test = INFO_CLASS_GENERATED + "\n" +
"\n" +
"package org.schabi.newpipe.extractor.timeago.patterns;\n\n" +
"import org.schabi.newpipe.extractor.timeago.PatternsHolder;\n" +
(specialCasesString.length() > 0 ? "import org.schabi.newpipe.extractor.timeago.TimeAgoUnit;\n" : "") +
"\n" +
"public class " + languageCode + " extends PatternsHolder {\n" +
" private static final String WORD_SEPARATOR = \"" + wordSeparator + "\";\n" +
" private static final String[]\n" +
" SECONDS /**/ = {" + join(seconds) + "},\n" +
" MINUTES /**/ = {" + join(minutes) + "},\n" +
" HOURS /**/ = {" + join(hours) + "},\n" +
" DAYS /**/ = {" + join(days) + "},\n" +
" WEEKS /**/ = {" + join(weeks) + "},\n" +
" MONTHS /**/ = {" + join(months) + "},\n" +
" YEARS /**/ = {" + join(years) + "};\n" +
"\n" +
" private static final " + languageCode + " INSTANCE = new " + languageCode + "();\n" +
"\n" +
" public static " + languageCode + " getInstance() {\n" +
" return INSTANCE;\n" +
" }\n" +
"\n" +
" private " + languageCode + "() {\n" +
" super(WORD_SEPARATOR, SECONDS, MINUTES, HOURS, DAYS, WEEKS, MONTHS, YEARS);\n" +
specialCasesString.toString() +
" }\n" +
"}";
fileOut.write(test);
} catch (IOException e) {
e.printStackTrace();
}
}
}
private static void specialCasesConstruct(TimeAgoUnit unit, JsonArray array, StringBuilder stringBuilder) {
final Iterator<Object> iterator = array.iterator();
while (iterator.hasNext()) {
final Object o = iterator.next();
if (o instanceof JsonObject) {
final JsonObject caseObject = (JsonObject) o;
for (Map.Entry<String, Object> caseEntry : caseObject.entrySet()) {
final int caseAmount = Integer.parseInt(caseEntry.getKey());
final String caseText = (String) caseEntry.getValue();
iterator.remove();
stringBuilder.append(" ")
.append("putSpecialCase(TimeAgoUnit.").append(unit.name())
.append(", \"").append(caseText).append("\"")
.append(", ").append(caseAmount).append(");").append("\n");
}
}
}
}
private static final String INFO_CLASS_GENERATED = "/**/// DO NOT MODIFY THIS FILE MANUALLY\n" +
"/**/// This class was automatically generated by \"GeneratePatternClasses.java\",\n" +
"/**/// modify the \"unique_patterns.json\" and re-generate instead.";
private static String join(List<Object> list) {
final StringBuilder toReturn = new StringBuilder();
for (Object o : list) {
toReturn.append('"').append(o).append('"').append(", ");
}
toReturn.setLength(Math.max(toReturn.length() - 2, 0));
return toReturn.toString();
}
}

View file

@ -1,59 +0,0 @@
import com.grack.nanojson.JsonArray;
import com.grack.nanojson.JsonObject;
import com.grack.nanojson.JsonParser;
import java.io.File;
import java.io.FileInputStream;
import java.io.FileOutputStream;
import java.io.OutputStream;
import java.util.Iterator;
import java.util.List;
import java.util.Map;
import java.util.TreeMap;
import static org.schabi.newpipe.extractor.timeago.TimeAgoPatternsManager.RESOURCE_BUNDLE_ARRAY_SEPARATOR;
public class GenerateResourceBundles {
public static void main(String[] args) throws Exception {
File outDir = new File("timeago-parser/outBundle");
if (!outDir.isDirectory()) outDir.mkdir();
JsonObject object = JsonParser.object().from(new FileInputStream(new File("timeago-parser/raw/unique_patterns.json")));
for (Map.Entry<String, Object> langTimeEntry : new TreeMap<>(object).entrySet()) {
final String langName = langTimeEntry.getKey();
StringBuilder outString = new StringBuilder();
final TreeMap<String, Object> sortedMap = new TreeMap<>(Utils.compareByUnitName());
sortedMap.putAll((JsonObject) langTimeEntry.getValue());
final Iterator<Map.Entry<String, Object>> unitEntriesIterator = sortedMap.entrySet().iterator();
while (unitEntriesIterator.hasNext()) {
final Map.Entry<String, Object> unitEntry = unitEntriesIterator.next();
final String unitName = unitEntry.getKey();
final List<Object> unitList = (JsonArray) unitEntry.getValue();
outString.append(unitName).append("=\\\n");
for (int i = 0; i < unitList.size(); i++) {
final String s = unitList.get(i).toString();
outString.append(" ").append(s);
if (i < unitList.size() - 1) {
outString.append(RESOURCE_BUNDLE_ARRAY_SEPARATOR).append("\\").append("\n");
}
}
if (unitEntriesIterator.hasNext()) outString.append("\n\n");
}
String fileName = "time_units_" + langName.replaceAll("-", "_") + ".properties";
System.out.println("Writing " + fileName + "...");
try (OutputStream out = new FileOutputStream(new File(outDir, fileName))) {
out.write(outString.toString().getBytes("UTF-8"));
}
}
}
}