Legacy Database Replacement with LAMP
The most recent feature that was added was the alias search. An alias replaces a more complex set of terms that might be used a lot. An example of this is a search for New Zealand content—music that contains NZ artists or was composed by a New Zealander. This is useful as we have self-imposed NZ music quotas for both networks.
Over many years of data entry and staff changes, different fields and identifiers were used to indicate NZ status in the main Works database. The NZ Music alias automatically adds the required terms and fields to the query as an OR search. This was achieved by building a new class on top of the term parser and using it to extract any aliases from the query. The parser then adds the required parameters to the query stack maintained by the Query Complier. Here are some BRAD Alias examples. The query:
gives us Mozart and any NZ content field true. The actual query looks like this:
SELECT * FROM brs.works WHERE (cf REGEXP '[[:<:]]local[[:>:]]' OR cf REGEXP '[[:<:]]nz[[:>:]]' OR lq REGEXP '[[:<:]]nz[[:>:]]') AND MATCH ti,ra,cf,cd,cp,so,at,notes,lq AGAINST ('+Mozart' IN BOOLEAN MODE) ORDER BY ti asc LIMIT 1000
A duration search also was added so that producers can find material within certain ranges—it is quite common to need music by a certain composer of approximately a known duration. In BRAD, numbers in square brackets are treated as a duration query. BRAD can do approximate searches or searches within a range of durations. See the sidebar for some examples.
BRAD Duration Examples
Between (you also can specify a range of times). The following looks for anything with mozart in it that is between 20 minutes, 30 seconds and 30 minutes, 15 seconds:
Approximate matches—this looks for the time you specify plus or minus 10%; c is short for circa:
mozart [ c 24 ]
You also can add a time range. The following input retrieves items 24 minutes in length, plus or minus 1 minute:
mozart [ c 24 r 1 ]
Complex duration searches— the following searches for pieces with Beethoven as the composer that last between 20 and 22 minutes:
The query compiled for the last search:
SELECT * FROM cfm.cfms WHERE (du <= 1320) AND (du >= 1200) AND MATCH ti,ca,ma, ra,cd,cp,so,at,notes AGAINST ('+beethoven' IN BOOLEAN MODE) AND MATCH cp AGAINST ('+beethoven' IN BOOLEAN MODE) ORDER BY ti asc LIMIT 1000
The Concert FM Selector data mentioned earlier has NZ artist and duration fields set correctly for all data, so these aliases can be used reliably on the whole data set. Because there is a mix of item types in the Works data, only those with a valid duration are searched. In the past, it was not possible to do any duration search at all within Works, so this is an improvement.
At the time of writing, I was asked about putting the company phone directory into BRAD, and a proof-of-concept pronunciation guide was added for our News department.
This project has enabled us to replace a key Radio NZ cataloging system and provide enhanced functionality to staff at a low TCO. It also has provided a storage platform for new and legacy data.
In the future, it may be possible for programme producers to do a single search on a composer or artist and get back a whole set of results that includes music tracks, interviews and archival material. It even could indicate the correct pronunciation of the person's name and provide his or her phone number.
BRAD probably will continue to be a work in progress as we find more uses for it. This is one of the benefits of DIY-IT—the system is ours to extend or modify as we see fit, whenever we need to do so.
Resources for this article: www.linuxjournal.com/article/7968.
Richard Hulse is a Senior Recording Engineer for Radio New Zealand and currently is working on a number of IT projects, including improving the Radio NZ Web site (www.radionz.co.nz).
Free DevOps eBooks, Videos, and more!
Regardless of where you are in your DevOps process, Linux Journal can help!
We offer here the DEFINITIVE DevOps for Dummies, a mobile Application Development Primer, and advice & help from the expert sources like:
- Linux Journal