Examples with ConfigException org.embulk.config.ConfigException used on opensource projects

Example 6 with ConfigException

use of org.embulk.config.ConfigException in project embulk by embulk.

the class RemoveColumnsFilterPlugin method getExistentColumns.

private List<String> getExistentColumns(Schema schema, List<String> specifiedColumns, boolean acceptUnmatch) {
    ImmutableList.Builder<String> existentColumns = ImmutableList.builder();
    for (String column : specifiedColumns) {
        try {
            schema.lookupColumn(column);
            existentColumns.add(column);
        } catch (SchemaConfigException e) {
            if (!acceptUnmatch) {
                throw new ConfigException(String.format(ENGLISH, "Column '%s' doesn't exist in the schema", column));
            }
        }
    }
    return existentColumns.build();
}

Example 7 with ConfigException

use of org.embulk.config.ConfigException in project embulk by embulk.

the class RemoveColumnsFilterPlugin method transaction.

@Override
public void transaction(ConfigSource config, Schema inputSchema, FilterPlugin.Control control) {
    PluginTask task = config.loadConfig(PluginTask.class);
    // validate remove: and keep:
    if (task.getRemove().isPresent() && task.getKeep().isPresent()) {
        throw new ConfigException("remove: and keep: must not be multi-select");
    }
    if (!task.getRemove().isPresent() && !task.getKeep().isPresent()) {
        throw new ConfigException("Must require remove: or keep:");
    }
    boolean acceptUnmatchedColumns = task.getAcceptUnmatchedColumns();
    ImmutableList.Builder<Column> outputColumns = ImmutableList.builder();
    int index = 0;
    int[] indexMapping = new int[inputSchema.size()];
    for (int i = 0; i < indexMapping.length; i++) {
        indexMapping[i] = -1;
    }
    if (task.getRemove().isPresent()) {
        // specify remove:
        List<String> removeColumns = getExistentColumns(inputSchema, task.getRemove().get(), acceptUnmatchedColumns);
        for (Column column : inputSchema.getColumns()) {
            if (!removeColumns.contains(column.getName())) {
                outputColumns.add(new Column(index, column.getName(), column.getType()));
                indexMapping[column.getIndex()] = index;
                index++;
            }
        }
    } else {
        // specify keep:
        List<String> keepColumns = getExistentColumns(inputSchema, task.getKeep().get(), acceptUnmatchedColumns);
        for (Column column : inputSchema.getColumns()) {
            if (keepColumns.contains(column.getName())) {
                outputColumns.add(new Column(index, column.getName(), column.getType()));
                indexMapping[column.getIndex()] = index;
                index++;
            }
        }
    }
    task.setIndexMapping(indexMapping);
    control.run(task.dump(), new Schema(outputColumns.build()));
}

Example 8 with ConfigException

use of org.embulk.config.ConfigException in project embulk by embulk.

the class RenameFilterPlugin method applyFirstCharacterTypesRule.

private Schema applyFirstCharacterTypesRule(Schema inputSchema, FirstCharacterTypesRule rule) {
    final Optional<String> replace = rule.getReplace();
    final List<String> passTypes = rule.getPassTypes();
    final String passCharacters = rule.getPassCharacters();
    final Optional<String> prefix = rule.getPrefix();
    if (replace.isPresent() && replace.get().length() != 1) {
        throw new ConfigException("\"replace\" in \"first_character_types\" must contain just 1 character if specified");
    }
    if (prefix.isPresent() && prefix.get().length() != 1) {
        throw new ConfigException("\"prefix\" in \"first_character_types\" must contain just 1 character if specified");
    }
    if (prefix.isPresent() && replace.isPresent()) {
        throw new ConfigException("\"replace\" and \"prefix\" in \"first_character_types\" must not be specified together");
    }
    if ((!prefix.isPresent()) && (!replace.isPresent())) {
        throw new ConfigException("Either of \"replace\" or \"prefix\" must be specified in \"first_character_types\"");
    }
    // TODO(dmikurube): Revisit this for better escaping.
    if (passCharacters.contains("\\E")) {
        throw new ConfigException("\"pass_characters\" in \"first_character_types\" must not contain \"\\E\"");
    }
    StringBuilder regexBuilder = new StringBuilder();
    regexBuilder.append("^[^");
    for (String target : passTypes) {
        if (CHARACTER_TYPE_KEYWORDS.containsKey(target)) {
            regexBuilder.append(CHARACTER_TYPE_KEYWORDS.get(target));
        } else {
            throw new ConfigException("\"" + target + "\" is an unknown character type keyword");
        }
    }
    if (!passCharacters.isEmpty()) {
        regexBuilder.append("\\Q");
        regexBuilder.append(passCharacters);
        regexBuilder.append("\\E");
    }
    regexBuilder.append("].*");
    Schema.Builder schemaBuidler = Schema.builder();
    for (Column column : inputSchema.getColumns()) {
        String name = column.getName();
        if (name.matches(regexBuilder.toString())) {
            if (replace.isPresent()) {
                name = replace.get() + name.substring(1);
            } else if (prefix.isPresent()) {
                name = prefix.get() + name;
            }
        }
        schemaBuidler.add(name, column.getType());
    }
    return schemaBuidler.build();
}

Example 9 with ConfigException

use of org.embulk.config.ConfigException in project embulk by embulk.

the class RenameFilterPlugin method applyCharacterTypesRule.

private Schema applyCharacterTypesRule(Schema inputSchema, CharacterTypesRule rule) {
    final List<String> passTypes = rule.getPassTypes();
    final String passCharacters = rule.getPassCharacters();
    final String replace = rule.getReplace();
    if (replace.isEmpty()) {
        throw new ConfigException("\"replace\" in \"character_types\" must not be explicitly empty");
    }
    if (replace.length() != 1) {
        throw new ConfigException("\"replace\" in \"character_types\" must contain just 1 character");
    }
    // TODO(dmikurube): Revisit this for better escaping.
    if (passCharacters.contains("\\E")) {
        throw new ConfigException("\"pass_characters\" in \"character_types\" must not contain \"\\E\"");
    }
    StringBuilder regexBuilder = new StringBuilder();
    regexBuilder.append("[^");
    for (String target : passTypes) {
        if (CHARACTER_TYPE_KEYWORDS.containsKey(target)) {
            regexBuilder.append(CHARACTER_TYPE_KEYWORDS.get(target));
        } else {
            throw new ConfigException("\"" + target + "\" is an unknown character type keyword");
        }
    }
    if (!passCharacters.isEmpty()) {
        regexBuilder.append("\\Q");
        regexBuilder.append(passCharacters);
        regexBuilder.append("\\E");
    }
    regexBuilder.append("]");
    Schema.Builder schemaBuilder = Schema.builder();
    for (Column column : inputSchema.getColumns()) {
        schemaBuilder.add(column.getName().replaceAll(regexBuilder.toString(), replace), column.getType());
    }
    return schemaBuilder.build();
}

Example 10 with ConfigException

use of org.embulk.config.ConfigException in project embulk by embulk.

the class DynamicColumnSetterFactory method newColumnSetter.

public DynamicColumnSetter newColumnSetter(PageBuilder pageBuilder, Column column) {
    Type type = column.getType();
    if (type instanceof BooleanType) {
        return new BooleanColumnSetter(pageBuilder, column, defaultValue);
    } else if (type instanceof LongType) {
        return new LongColumnSetter(pageBuilder, column, defaultValue);
    } else if (type instanceof DoubleType) {
        return new DoubleColumnSetter(pageBuilder, column, defaultValue);
    } else if (type instanceof StringType) {
        TimestampFormatter formatter = TimestampFormatter.of(getTimestampFormatForFormatter(column), getTimeZoneId(column));
        return new StringColumnSetter(pageBuilder, column, defaultValue, formatter);
    } else if (type instanceof TimestampType) {
        // TODO use flexible time format like Ruby's Time.parse
        final TimestampParser parser;
        if (this.useColumnForTimestampMetadata) {
            final TimestampType timestampType = (TimestampType) type;
            // https://github.com/embulk/embulk/issues/935
            parser = TimestampParser.of(getFormatFromTimestampTypeWithDepracationSuppressed(timestampType), getTimeZoneId(column));
        } else {
            parser = TimestampParser.of(getTimestampFormatForParser(column), getTimeZoneId(column));
        }
        return new TimestampColumnSetter(pageBuilder, column, defaultValue, parser);
    } else if (type instanceof JsonType) {
        TimestampFormatter formatter = TimestampFormatter.of(getTimestampFormatForFormatter(column), getTimeZoneId(column));
        return new JsonColumnSetter(pageBuilder, column, defaultValue, formatter);
    }
    throw new ConfigException("Unknown column type: " + type);
}