Skip to content

Latest commit

 

History

History
130 lines (111 loc) · 6.06 KB

CHANGELOG.md

File metadata and controls

130 lines (111 loc) · 6.06 KB

Cloudhopper by Twitter

cloudhopper-commons-charset

3.0.1 - 2012-11-06

  • Added T-Mobile NL charset to CharsetUtils and updated tests, also to Readme
  • Changed byte constant in T-Mobile NL charset to uppercase for readability

3.0.0 - 2012-10-30

  • No major code changes, mostly project layout changes in prep of release to Maven Central Repository
  • Maven artifact group changed from cloudhopper to com.cloudhopper
  • Readme and changelog switched to markdown
  • Log4J dependency switched to SLF4J
  • Updated underlying maven license plugin - every java file had its license positioned after package declaration.

2.3.2 - 2012-10-29

  • Added custom charset for T-Mobile NL, TMobileNlGSMCharset

2.3.1 - 2012-02-17

  • Removed Turkish "dotless i" safe unicode replacements in MobileTextUtil. At recommendation of locals in Turkey, these replacements changed the meaning of of a message too much.

2.3.0 - 2011-10-26

  • Upgrade parent POM dependency from 1.5 to 1.6
  • Added support for OSGI in final jar
  • UTF8Charset.calculateByteLength() is deprecated. The algorithm only calculates the byte length for a "modified UTF-8" string and is wrong on higher range characters. Will be removed in a future release since it now calls ModifiedUTF8Charset.calculateByteLength() for backwards compatability.
  • GSMCharset.canRepresent is now ~10X faster by switching from a lookup table to a switch statement. See test "PrintGSMMain.class" for automated class for re-generating the switch statement if needed.
  • Added "Modified UTF-8" charset for improved performance on encoding/decoding vs. Java's native UTF-8 charset. Decoding is generally ~35% faster and encoding is generally between 0 and 25% faster than Java's UTF-8 impl. Encoding performance will be higher assuming you already had a byte buffer you planned on copying the bytes into anyway (very useful for serialization).
  • Much better performance on CharsetUtil.decode() into a String. Each charset can now optimize this usage vs. the previous method of creating an underlying StringBuilder. Performance is ~25% better on most charsets.
  • Some behavior of encode() and decode() with NULL parameters was changed to always either do a NOOP or return a NULL. Unit tests now check every registered charset for similar behavior.

2.2.1 - 2011-08-18

  • Fixed a performance bug with GSMCharset.canRepresent() where an inner loop was not broken as soon as a char was found and the search continued. Thanks to Yury Korolev (https://github.com/yury) for finding.
  • Fixed a correctness bug with GSMCharset.canRepresent() where the '`' (0x60) character was included as part of the GSM charset. This character was not included on encodes or decodes, but was included in this specific method as part of an optimization. Thanks to Yury Korolev (https://github.com/yury) for finding.
  • Removed any external dependency on log4j.
  • Cleaned up documentation

2.2.0 - 2011-07-26

  • Upgrade parent POM dependency from 1.1 to 1.5
  • Switched to new version format.
  • Added Version class to compile build info into jar

2.1 - 2011-05-27

  • Dotless i (used mainly in Turkic language) is now replaced with the char "1" (digit one) in MobileTextUtil. Please see wiki page for common SMS usage: http://en.wikipedia.org/wiki/Dotted_and_dotless_I
  • U+201B and U+201F unicode chars are now replaced with ascii equivalents of ' and "" respectively.

2.0 - 2011-03-14

  • Migrated build system from Ant to Maven
  • Deleted all legacy Ant build files (build.xml, Dependency.txt)
  • Added Apache license to top of all source files
  • Moved demo source code from src/demo/java into src/test/java directory and placed in separate "demo" package.
  • Added "Makefile" with targets that match previous demo ant tasks.

1.8 - 2011-02-14

  • Added MobileTextUtil.replaceAccentedChars to downgrade accented chars with their ascii equivalent. For example, an accented e to a non-accented e.
  • Added additional ascii equivalents for unicode chars that MS-Word typically uses for text.

1.7 - 2011-01-08

  • Added UTF8Charset.calculateByteLength method to calculate the byte length required to convert a Java String into an array of UTF-8 bytes. Highly efficient method since it doesn't require any byte array allocation.

1.6 - 2010-12-16

  • Added normalize() feature: filters a source Java string into a new Java string with any unsupported characters converted into ?'s.
  • Added MobileTextUtil utility class: helps replace unicode characters with their ascii equivalents (e.g. smart quotes into dumb quotes)
  • Fixed extra E on name of charset for AIRWIDE.
  • Modified log4j.properties to ouput UTF-8 to console during demos.
  • Fixed unit tests with encoding issue on linux vs. mac os x due to encoding of unicode chars in the source files.

1.5 - 2010-08-23

  • Added support for ISO-8859-15 charset
  • Added support for VFTR-GSM charset which is a custom charset specifically for the Vodafone Turkey SMSC.

1.4 - 2010-06-08

  • Added aliases for GSM -> GSM8, PACKED-GSM -> GSM7, AIRWIDE-IA5 -> AIRWIDE-GSM
  • Added support for "VFD2-GSM" which is a custom charset Vodafone D2 uses to interpret the @ and some German letters from the standard GSM charset.
  • Fixed bug with some charsets modifying the byte array parameter values in the decoding process. Wrote a unit test to validate common byte values commonly replaced does not actually change the byte array.
  • Added extra unit tests to check for null parameters against every charset

1.3 - 2010-05-16

  • Added support for verfiying a Java String can be represented by the GSMCharset. Provides a way of automatically figuring out the correct data coding scheme to use when forwarding an SMS to an operator.
  • Fixed deprecated methods used in ch-commons-util

1.2 - 2010-02-28

  • Added support for Airwide-IA5 charset

1.1 - 2010-02-21

  • Added support for charset conversions including GSM, PackedGSM, etc.

1.0 - 2010-02-11

  • Initial release