Skip to content

Latest commit

 

History

History
641 lines (551 loc) · 22.4 KB

doc.md

File metadata and controls

641 lines (551 loc) · 22.4 KB

Features

CsvJdbc accepts all types of CSV files defined by RFC 4180.

CsvJdbc accepts only SQL SELECT queries from a single table and does not support INSERT, UPDATE, DELETE or CREATE statements.

SQL sub-queries are permitted but joins between tables in SQL SELECT queries are not yet supported.

SQL SELECT queries must be of the following format.

SELECT [DISTINCT] [table-alias.]column [[AS] alias], ...
  FROM table [[AS] table-alias]
  WHERE [NOT] condition [AND | OR condition] ...
  GROUP BY column ... [HAVING condition ...]
  ORDER BY column [ASC | DESC] ...
  LIMIT n [OFFSET n]

Each column is either a named column, *, a constant value, NULL, CURRENT_DATE, CURRENT_TIME, a sub-query, or an expression including functions, aggregate functions, operations +, -, /, *, % (modulo), || (string concatenation), conditional CASE expressions and parentheses.

Arithmetic involving date values has units of days. Arithmetic involving timestamp or time values has units of milliseconds.

Supported comparisons in the optional WHERE clause are <, >, <=, >=, =, !=, <>, NOT, BETWEEN, LIKE, IS NULL, IN, EXISTS.

Use double quotes around table names or column names containing spaces or other special characters.

Both C-style comments /* a comment */ and single line comments following two dashes and a space until the end of the line -- a comment are skipped in SQL statements.

Function Description
ABS(N) Returns absolute value of N
COALESCE(N1, N2, ...) Returns first expression that is not NULL
TO_ARRAY([DISTINCT] N1, N2, ...) Returns java.sql.Array containing (optionally distinct) values
DAYOFMONTH(D) Extracts day of month from date or timestamp D (first day of month is 1)
HOUROFDAY(T) Extracts hour of day from time or timestamp T
LENGTH(S) Returns length of string
LINE_NUMBER() Returns line number of row in CSV file (NULL if row not corresponding to a line)
LOWER(S) Converts string to lower case
LTRIM(S [, T]) Removes leading characters from S that occur in T
MINUTE(T) Extracts minute of hour from time or timestamp T
MONTH(D) Extracts month from date or timestamp D (first month is 1)
NULLIF(X, Y) Returns NULL if X and Y are equal, otherwise X
RANDOM() Returns random number in the range 0 to 1
REPLACE(S, FROM, TO) Replaces all occurrences of string FROM in S with TO
ROUND(N [, D]) Rounds N to the specified number of decimal places D (0 by default)
RTRIM(S, [, T]) Removes trailing characters from S that occur in T
SECOND(T) Extracts seconds value from time or timestamp T
SUBSTRING(S, N [, L]) Extracts substring from S starting at index N (counting from 1) with length L
TO_NUMBER(S, P) Converts string S to a number in current locale, using locale-independent pattern P and Java class DecimalFormat
TRIM(S, [, T]) Removes leading and trailing characters from S that occur in T
UPPER(S) Converts string to upper case
YEAR(D) Extracts year from date or timestamp D
VERSION() Returns a string containing the CsvJdbc version number

Additional functions are defined from java methods using the function.NAME driver property.

Aggregate Function Description
AVG(N) Average of all values
COUNT(N) Count of all values
MAX(N) Maximum value
MIN(N) Minimum value
STRING_AGG(S, D) All values of S concatenated with delimiter D
ARRAY_AGG(S) All values of S into a java.sql.Array
SUM(N) Sum of all values

For queries containing ORDER BY, all records are read into memory and sorted. For queries containing GROUP BY plus an aggregate function, all records are read into memory and grouped. For queries that produce a scrollable result set, all records up to the furthest accessed record are held into memory. For other queries, CsvJdbc holds only one record at a time in memory.

Notes on functions returning an array:

  • ResultSet.getArray(...) returns an object of type org.relique.jdbc.csv.SqlArray, that implements interface java.sql.Array
  • Both java.sql.Array.getArray(...) and getResultSet(...) methods are implemented
  • SqlArray tries to infer the type of the values for its getBaseTypeName() and getBaseType() methods; an IllegalStateException is thrown if values of different types are detected

Dependencies

CsvJdbc requires Java version 11, or later. For reading DBF files, DANS DBF Library must be downloaded and included in the CLASSPATH.

Advanced Usage

Like other databases, creating a scrollable statement enables scrolling forwards and backwards through result sets. This is demonstrated in the following example.

import java.sql.*;

public class DemoDriver2
{
  public static void main(String[] args) throws Exception
  {
    try (Connection conn = DriverManager.getConnection("jdbc:relique:csv:" + args[0]);

      // create a scrollable Statement so we can move forwards and backwards
      // through ResultSets
      Statement stmt = conn.createStatement(ResultSet.TYPE_SCROLL_SENSITIVE,
        ResultSet.CONCUR_READ_ONLY);
      ResultSet results = stmt.executeQuery("SELECT ID,NAME FROM sample"))
    {

      // dump out the last record in the result set, then the first record
      if (results.last())
      {
        System.out.println("ID= " + results.getString("ID") +
              "   NAME= " + results.getString("NAME"));
        if (results.first())
        {
          System.out.println("ID= " + results.getString("ID") +
              "   NAME= " + results.getString("NAME"));
        }
      }
    }
  }
}

To read several files (for example, daily log files) as a single table, set the database connection property indexedFiles. The following example demonstrates how to do this.

import java.sql.*;
import java.util.Properties;

public class DemoDriver3
{
  public static void main(String[] args) throws Exception
  {
    Properties props = new Properties();
    props.put("fileExtension", ".txt");
    props.put("indexedFiles", "true");
    // We want to read test-001-20081112.txt, test-002-20081113.txt and many
    // other files matching this pattern.
    props.put("fileTailPattern", "-(\\d+)-(\\d+)");
    // Make the two groups in the regular expression available as
    // additional table columns.
    props.put("fileTailParts", "Seqnr,Logdatum");
    try (Connection conn = DriverManager.getConnection("jdbc:relique:csv:" +
        args[0], props);
      Statement stmt = conn.createStatement();
      ResultSet results = stmt.executeQuery("SELECT Datum, Station, " +
        "Seqnr, Logdatum FROM test"))
    {
      ResultSetMetaData meta = results.getMetaData();
      while (results.next())
      {
        for (int i = 0; i < meta.getColumnCount(); i++)
        {
          System.out.println(meta.getColumnName(i + 1) + " " +
            results.getString(i + 1));
        }
      }
    }
  }
}

Set the database connection property columnTypes to enable expressions containing numeric, time and date data types to be used in SELECT statements and to enable column values to be fetched using ResultSet.getInt, getDouble, getTime and other ResultSet.get methods.

import java.sql.*;
import java.util.Properties;

public class DemoDriver4
{
  public static void main(String[] args) throws Exception
  {
    Properties props = new Properties();
    // Define column names and column data types here.
    props.put("suppressHeaders", "true");
    props.put("headerline", "ID,ANGLE,MEASUREDATE");
    props.put("columnTypes", "Int,Double,Date");
    try (Connection conn = DriverManager.getConnection("jdbc:relique:csv:" +
        args[0], props);
      Statement stmt = conn.createStatement();
      ResultSet results = stmt.executeQuery("SELECT Id, Angle * 180 / 3.1415 as A, " +
        "MeasureDate FROM t1 where Id > 1001"))
    {
      while (results.next())
      {
        // Fetch column values with methods that match the column data types.
        System.out.println(results.getInt(1));
        System.out.println(results.getDouble(2));
        System.out.println(results.getDate(3));
      }
    }
  }
}

To read the compressed files inside a ZIP file as database tables, make a database connection to the ZIP file using the JDBC connection string format jdbc:relique:csv:zip:filename.zip. This is demonstrated in the following example.

import java.sql.*;

public class DemoDriver5
{
  public static void main(String[] args) throws Exception
  {
    String zipFilename = args[0];
    try (Connection conn = DriverManager.getConnection("jdbc:relique:csv:zip:" +
        zipFilename);
      Statement stmt = conn.createStatement();
      // Read from file mytable.csv inside the ZIP file
      ResultSet results = stmt.executeQuery("SELECT * FROM mytable"))
    {
      while (results.next())
      {
          System.out.println(results.getString("COUNTRY"));
      }
    }
  }
}

To read the resources from classpath as database tables, make a database connection to the classpath path using the JDBC connection string format jdbc:relique:csv:classpath:path/to/resources. The additional dependency is required to scan the paths in the classpath.

...
<dependency>
  <groupId>io.github.classgraph</groupId>
  <artifactId>classgraph</artifactId>
  <version>X.Y.Z</version>
</dependency>
...

The following example demonstrates reading from the classpath.

import java.sql.*;

public class DemoDriver6
{
  public static void main(String[] args) throws Exception
  {
    String path = args[0];
    try (Connection conn = DriverManager.getConnection("jdbc:relique:csv:classpath:" +
            path);
      Statement stmt = conn.createStatement();
      // Read from the classpath resource mytable.csv
      ResultSet results = stmt.executeQuery("SELECT * FROM mytable"))
      {
        while (results.next())
        {
            System.out.println(results.getString("COUNTRY"));
        }
      }
  }
}

To read data that is either held inside some file storage or accessed remotely (for example, using HTTP requests), create a Java class that implements the interface org.relique.io.TableReader and give this class name in the connection URL. CsvJdbc then creates an instance of this class and calls the getReader method to obtain a java.io.Reader for each database table being read. This is demonstrated in the following two Java classes.

import java.io.*;
import java.net.*;
import java.sql.*;
import java.util.*;

import org.relique.io.TableReader;

public class MyHTTPReader implements TableReader
{
  public Reader getReader(Statement statement, String tableName) throws SQLException
  {
    try
    {
      URL url = new URL("http://csvjdbc.sourceforge.net/" + tableName + ".csv");
      HttpURLConnection connection = (HttpURLConnection) url.openConnection();
      InputStreamReader reader = new InputStreamReader(connection.getInputStream());
      return reader;
    }
    catch (Exception e)
    {
      throw new SQLException(e.getMessage());
    }
  }
  public List getTableNames(Connection connection)
  {
    // Return list of available table names
    return List.of("sample");
  }
}


import java.sql.*;
import org.relique.jdbc.csv.CsvDriver;

public class DemoDriver7
{
  public static void main(String []args) throws Exception
  {
    String sql = "SELECT * FROM sample";
    // Give name of Java class that provides database tables.
    try (Connection conn = DriverManager.getConnection("jdbc:relique:csv:class:" +
        MyHTTPReader.class.getName());
      Statement stmt = conn.createStatement();
      ResultSet results = stmt.executeQuery(sql))
    {
      CsvDriver.writeToCsv(results, System.out, true);
    }
  }
}

Driver Properties

The driver also supports a number of parameters that change the default behaviour of the driver.

These properties are:

charset

  • type: String
  • default: Java default
  • Defines the character set name of the files being read, such as UTF-16. See the Java Charset documentation for a list of available character set names.

columnTypes

  • type: String
  • default: all Strings
  • A comma-separated list defining SQL data types for columns in tables. When column values are fetched using getObject (as opposed to getString), the driver will parse the value and return a correctly typed object. If fewer data types are provided than the number of columns in the table, the last data type is repeated for all remaining columns. If columnTypes is set to an empty string then column types are inferred from the data. When working with multiple tables with different column types, define properties named columnTypes.CATS and columnTypes.DOGS to define different column types for tables CATS and DOGS.

commentChar

  • type: String
  • default: null
  • Lines before the header that start with the comment are skipped. After the header has been read, all lines are interpreted as data.

cryptoFilterClassName

  • type: Class
  • default: null
  • The full class name of a Java class that decrypts the file being read. The class must implement interface org.relique.io.CryptoFilter. The class org.relique.io.XORFilter included in CsvJdbc implements an XOR encryption filter.

cryptoFilterParameterTypes

  • type: String
  • default: String
  • Comma-separated list of data types to pass to the constructor of the decryption class set in property cryptoFilterClassName.

cryptoFilterParameters

  • type: String
  • default:
  • Comma-separated list of values to pass to the constructor of the decryption class set in property cryptoFilterClassName.

defectiveHeaders

  • type: Boolean
  • default: False
  • When true, column names that are an emtpy string or a case insensitive duplicate of a previous column name are renamed to COLUMNx, where x is the ordinal identifying the column.

fileExtension

  • type: String
  • default: .csv
  • Specifies file extension of the CSV files. If the extension .dbf is used then files are read as dBase format database files.

fileTailParts

  • type: String
  • default: null
  • Comma-separated list of column names for the additional columns generated by regular expression groups in the property fileTailPattern.

fileTailPattern

  • type: String
  • default: null
  • Regular expression for matching filenames when property indexedFiles is True. If the regular expression contains groups (surrounded by parentheses) then the value of each group in matching filenames is added as an extra column to each line read from that file. For example, when querying table test, the regular expression -(\d+)-(\d+) will match files test-001-20081112.csv and test-002-20081113.csv. The column values 001 and 20081112 are added to each line read from the first file and 002 and 20081113 are added to each line read from the second file.

fileTailPrepend

  • type: Boolean
  • default: False
  • when True, columns generated by regular expression groups in the fileTailPattern property are prepended to the start of each line. When False, the generated columns are appended after the columns read for each line.

fixedWidths

  • type: String
  • default: null
  • Defines character position ranges for each column in a fixed width file. When set, column values are extracted from these ranges in each line instead of separating the line by delimiters. Each column is a pair of character positions separated by a minus sign, or a single character for columns with only a single character. The position of the first character on each line is 1. Character position ranges are separated by commas. For example, 1,2-9,16-19.

function.NAME

  • type: String
  • default: None
  • Defines a java method to use as the SQL function named NAME in SQL statements. The property value is a public static java given as a java package, class and method name followed by parameter list in parentheses. For example, property function.POW with value java.lang.Math.pow(double, double) makes POW available as an SQL function. Methods with variable length argument lists are defined by appending ... after the last parameter. Each method parameter must be a numeric type, String, or Object.

headerline

  • type: string
  • default: None
  • Used in combination with the suppressHeaders property to specify a custom header line for tables. headerline contains a list of column names for tables separated by the separator. When working with multiple tables with different headers, define properties named headerline.CATS and headerline.DOGS to define different header lines for tables CATS and DOGS.

ignoreNonParseableLines

  • type: Boolean
  • default: False
  • when True, lines that have too few or too many column values will not cause an exception but will be ignored. Each ignored line is logged. Call method java.sql.DriverManager.setLogWriter before executing a query to capture a list of ignored lines.

indexedFiles

  • type: Boolean
  • default: False
  • when True, all files with a filename matching the table name plus the regular expression given in property fileTailPattern are read as if they were a single file.

isHeaderFixedWidth

  • type: Boolean
  • default: True
  • Used in combination with the fixedWidths property when reading fixed width files to specify whether the header line containing the column names is also fixed width. If False, column names are separated by the separator.

missingValue

  • type: String
  • default: null
  • When not null, lines with too few column values will use this value for each missing column, instead of throwing an exception.

quotechar

  • type: Character
  • default: "
  • Defines quote character. Column values surrounded with the quote character are parsed with the quote characters removed. This is useful when values contain the separator or line breaks. No more than one character is allowed. An empty value disables quoting.

quoteStyle

  • type: String
  • default: SQL
  • Defines how a quote character is interpreted inside a quoted value. When SQL, a pair of quote characters together is interpreted as a single quote character. When C, a backslash followed by a quote character is interpreted as a single quote character.

locale

  • type: String
  • default: Java default
  • Defines locale to use when parsing timestamps and when parsing numbers with the TO_NUMBER function. This is important when parsing words in a timestamp such as December, or parsing numbers such as 1.234,56 with TO_NUMBER, which vary depending on the locale. Call method java.util.Locale.toString() to convert a locale into a string for this property, such as de or de_AT.

maxDataLines

  • type: Integer
  • default: 0
  • when non-zero, defines the maximum number of lines of data to read from the file. Using this property together with skipLeadingDataLines enables a limited number of lines to be read from the middle of a very large file.

randomSeed

  • type: Long
  • default: None
  • Defines seed value for random number generation. Using the same seed value will repeat the same sequence of random numbers.

separator

  • type: String
  • default: ,
  • Defines column separator. A separator longer than one character is permitted. The separator \t is interpreted as a tab character.

skipLeadingLines

  • type: Integer
  • default: 0
  • after opening a file, skip this many lines before starting to interpret the contents.

skipLeadingDataLines

  • type: Integer
  • default: 0
  • after reading the header from a file, skip this many lines before starting to interpret lines as records.

suppressHeaders

  • type: boolean
  • default: False
  • Used to specify that the file does not contain a column header with column names. If True and headerline is not set, then columns are named sequentially COLUMN1, COLUMN2, ... If False, the column header is read from the first line of the file.

timestampFormat, timeFormat, dateFormat

  • type: String
  • default: yyyy-MM-dd HH:mm:ss, HH:mm:ss, yyyy-MM-dd
  • Defines the format from which columns of type Timestamp, Time and Date are parsed. See the Java SimpleDateFormat documentation for date and timestamp patterns, or DateTimeFormatter when property useDateTimeFormatter is set to true.

timeZoneName

  • type: String
  • default: UTC
  • The time zone of Timestamp columns. To use the time zone of the computer, set this to the value returned by the method java.util.TimeZone.getDefault().getID().

trimHeaders

  • type: Boolean
  • default: True
  • If True, leading and trailing whitespace is trimmed from each column name in the header line. Column names inside quotes are not trimmed.

trimValues

  • type: Boolean
  • default: False
  • If True, leading and trailing whitespace is trimmed from each column value in the file. Column values inside quotes are not trimmed.

useDateTimeFormatter

  • type: Boolean
  • default: False
  • If True, Java class DateTimeFormatter (that was new in Java 8) is used to parse and format timestamps, times and dates instead of the older class SimpleDateFormat.

The following example code shows how some of these properties are used.

...

  Properties props = new java.util.Properties();

  props.put("separator", "|");              // separator is a bar
  props.put("suppressHeaders", "true");     // first line contains data
  props.put("fileExtension", ".txt");       // file extension is .txt
  props.put("timeZoneName", "America/Los_Angeles"); // timestamps are Los Angeles time

  Connection conn1 = Drivermanager.getConnection("jdbc:relique:csv:" + args[0], props);

  ...

  // Connections using a URL string containing both directory and
  // properties are also accepted (class java.net.URLEncoder encodes
  // property values containing special characters).
  Connection conn2 = DriverManager.getConnection("jdbc:relique:csv:" + args[0] + "?" +
    "separator=" + URLEncoder.encode("|", "UTF-8") + "&" +
    "quotechar=" + URLEncoder.encode("'", "UTF-8") + "&" +
    "fileExtension=.txt" + "&" +
    "suppressHeaders=true");