In MySQL, the character set defines what kind of characters you can store, but not how they are compared or sorted. This is determined by the character set’s collation setting.

For example, take the following table:

CREATE TABLE `CHEM_FRAGMENT` (
`CFR_ID` bigint(20) NOT NULL auto_increment,
`CFR_SMILES` varchar(255) NOT NULL,
...
) ENGINE=InnoDB DEFAULT CHARSET=latin1


The character set is ‘latin1′. The collation settings are not shown because they are set to the default value. What’s that? Well:


mysql> show collation where charset = 'latin1';

+-------------------+---------+----+---------+----------+---------+
| Collation         | Charset | Id | Default | Compiled | Sortlen |
+-------------------+---------+----+---------+----------+---------+
| latin1_german1_ci | latin1  |  5 |         | Yes      |       1 |
| latin1_swedish_ci | latin1  |  8 | Yes     | Yes      |       1 |
| latin1_danish_ci  | latin1  | 15 |         | Yes      |       1 |
| latin1_german2_ci | latin1  | 31 |         | Yes      |       2 |
| latin1_bin        | latin1  | 47 |         | Yes      |       1 |
| latin1_general_ci | latin1  | 48 |         | Yes      |       1 |
| latin1_general_cs | latin1  | 49 |         | Yes      |       1 |
| latin1_spanish_ci | latin1  | 94 |         | Yes      |       1 |
+-------------------+---------+----+---------+----------+---------+

8 rows in set (0.00 sec)


The default collation is ‘latin1_swedish_ci’. Ignoring everything else, the important thing is the ‘_ci’ at the end, which stands for ‘case insensitive’. So change it! ‘latin1_general_cs’ is case sensitive, and so is ‘latin1_bin’.

To alter the collation for an entire table:

ALTER TABLE `THE_TABLE` COLLATE latin1_general_cs;


To alter the collation for a single column:

ALTER TABLE `THE_TABLE` MODIFY `THE_COLUMN` varchar(760) default NULL COLLATE latin1_general_cs;


There’s an easy way to fix it using xargs, that I found via google on Toolbox for IT, a new site for me.

All you do is ask tar to list the contents of the tarball and use xargs to delete those files. Look how easy it seems:

$ tar ztf project_net_installer_linux90b1.tar.gz | xargs rm -rf

HOWEVER. The first entry in this particular tarball is ‘./‘. So the first thing running that script did was to delete everything in my working directory (and tell me that it couldn’t delete the working directory itself). Awesome.

Nothing of value was lost, though I did have to redownload the tarball. And this is nowhere near as elegant, but to save youself from this one could try

$ tar ztf project_net_installer_linux90b1.tar.gz | egrep -v '^./$' | xargs rm -rf

It worked for me.

I wanted to replicate the functionality in groovy, another scripting language that runs on the JVM — one I’ve recently been using quite extensively (in conjunction with its web development framework called, inevitably, grails).

Here’s what I did to get this working

1. Download mx

   $ wget wget http://mx-java.googlecode.com/files/mx-0.107.0.jar

2. Copy to ~/.groovy/lib

   $ cp mx-0.107.0.jar ~/.groovy/lib

3. Test reading an SD file in groovy (this is transliterated from the example provided by Mr Apodaca)

   $ groovysh
   Groovy Shell (1.5.6, JVM: 1.6.0_0-b12)
   Type 'help' or '\h' for help.
   -------------------------------------------------------------------------------
   groovy:000> import com.metamolecular.mx.io.mdl.SDFileReader
   groovy:000> r = new SDFileReader("example.sdf")
   ===> com.metamolecular.mx.io.mdl.SDFileReader@4e2892b
   groovy:000> r.nextRecord()
   ===> null
   groovy:000> m = r.molecule
   ===> com.metamolecular.mx.model.DefaultMolecule@64c272bc
   groovy:000> m.countAtoms()
   ===> 20
   groovy:000> r.keys.each{ key -> println "$key : ${r.getData(key)}" }
   Mol Weight : 275.2203
   Formula : C11H9N5O4
   ===> [Mol Weight, Formula]

There you have it! I am impressed with the relatively straightforward API, though it is rather tempting to wrap it inside a groovy builder to make a mini-DSL. SDFileBuilder has a certain ring to it…