Relational Databases Object Relational Mappers SQLObject BCHB 524

Relational Databases: Object Relational Mappers - SQLObject BCHB 524 Lecture 22 BCHB 524 - Edwards

Outline l Object relational mappers l l l Tables as classes, rows as instances Advantages & disadvantages Minimal SQLObject example Legacy databases Exercises BCHB 524 - Edwards 2

Relational Databases l Store information in a table l l Rows represent items Columns represent items' properties or attributes Name Continent Region Surface Area Population GNP Brazil South America 8547403 170115000 776739 Indonesia Asia Southeast Asia 1904569 212107000 84982 India Asia Southern and Central Asia 3287263 1013662000 447114 China Asia Eastern Asia 9572900 1277558000 982268 Pakistan Asia Southern and Central Asia 796095 156483000 61289 United States North America 9363520 278357000 8510700 BCHB 524 - Edwards 3

. . . as Objects l l l Objects have data members or attributes. Store objects in a list or iterable. Abstract away details of underlying RDBMS c 1 = Country() c 1. name = 'Brazil' c 1. continent = 'South America' c 1. region = 'South America' c 1. surface. Area = 8547403 c 1. population = 170115000 c 1. gnp = 776739 # initialize c 2, . . . , c 6 country. Table = [ c 1, c 2, c 3, c 4, c 5, c 6 ] for cnty in country. Table: if cnty. population > 10000: print cnty. name, cnty. population BCHB 524 - Edwards 4

Pros and Cons Pros: l Learn one language l Ease of development l Simplified joins l One set of data-types l Easy storage of higher-level objects l Can apply the power of python as necessary l Abstract away RDBMS l Distribute CPU load Cons: l Execution speed l Sometimes forced into poor strategies l Optimal SQL construct may be impossible l Tend not to take advantage of RDBMS quirks. l Can be difficult to apply to legacy databases BCHB 524 - Edwards 5

SQLObject l l Just one of many object-relational mappers Each tool makes different tradeoffs in l l l Table/row/attribute abstraction How much SQL you need to know Overhead Ease of adapting to legacy databases SQLObject is almost completely devoid of SQL and is almost entirely "objecty". See http: //sqlobject. org. BCHB 524 - Edwards 6

Minimal SQLObject Example: Define the database model (model. py) from sqlobject import * import os. path dbfile = 'myworld. db 3' # Magic formatting for database URI conn_str = os. path. abspath(dbfile) conn_str = 'sqlite: '+ conn_str sqlhub. process. Connection = connection. For. URI(conn_str) class Country(SQLObject): name = String. Col() continent = String. Col() region = String. Col() surface. Area = Float. Col() population = Int. Col() gnp = Float. Col() BCHB 524 - Edwards 7

Minimal SQLObject Example: Populate the database from model import Country # Initialize the table Country. create. Table() # Add some rows c = Country(name="Brazil", continent="South America", region="South America", surface. Area=8547403, population=170115000, gnp=776739) c = Country(name="China", continent="Asia", region="Eastern Asia", surface. Area=9572900, population=1277558000, gnp=982268) #. . . c = Country(name="United States", continent="North America", region="North America", surface. Area=9363520, population=278357000, gnp=8510700) # Retrieve and print all countries for c in Country. select(): print c. id, c. name, c. continent, c. gnp BCHB 524 - Edwards 8

Minimal SQLObject Example: Access/Change the database from model import Country # Change country #6 c = Country. get(6) c. name = 'United States of America' # Retrieve and print all countries for c in Country. select(): print c. id, c. name, c. continent, c. gnp BCHB 524 - Edwards 9

Minimal SQLObject Example: Access the rows as objects from model import Country # Select countries with more than 500, 000 in population for c in Country. select(Country. q. population >= 50000): print "A: ", c. id, c. name, c. population # Select countries that start with 'U' for c in Country. select(Country. q. name. startswith("U")): print "B: ", c. id, c. name, c. population # Lookup by id, exactly 1 country with each id c = Country. get(5) print "C: ", c. id, c. name, c. population # Get exception for bad id # c = Country. get(100) # Shortcut for select, countries with continent == 'Asia' for c in Country. select. By(continent = 'Asia'): print "D: ", c. id, c. name, c. population BCHB 524 - Edwards 10

Legacy databases l l If the legacy database is well-structured, SQLObject can figure out (most of) the definitions If there is no id column. . . l l l Need to tell SQLObject what to use for the ID. May need to specify the id at instantiation time. Have to fill in Multiple. Joins and Foreign. Keys yourself l l l Need to declare which columns in two different tables should correspond. Enables SQLObject to make relationships explicit Enables SQLObject to turn joins into lists BCHB 524 - Edwards 11

Legacy databases from sqlobject import * import os. path dbfile = 'taxa. db 3' conn_str = os. path. abspath(dbfile) conn_str = 'sqlite: '+ conn_str sqlhub. process. Connection = connection. For. URI(conn_str) class Taxonomy(SQLObject): class sqlmeta: id. Name = "tax_id" from. Database = True names = Multiple. Join('Name', join. Column="tax_id") class Name(SQLObject): class sqlmeta: from. Database = True taxa = Foreign. Key('Taxonomy', db. Name="tax_id") BCHB 524 - Edwards 12

Legacy databases # Set up data-model from model import * # get homo sapiens hs 1 = Taxonomy. get(9606) # select the Taxonomy object # with scientific name Homo sapiens hs 2 = Taxonomy. select. By(scientific. Name= 'Homo sapiens')[0] # get the name human try: hsname = Name. select. By(name= 'human')[0] except Index. Error: print "Can't find name 'human'" sys. exit(1) # get the Taxonomy object from the Name object # Uses the magic Foreign Key attribute hs 3 = hsname. taxa # hs 1, hs 2, hs 3 the same! print hs 1 print hs 2 print hs 3 BCHB 524 - Edwards 13

Legacy databases # Set up data-model from model import * # get homo sapiens hs = Taxonomy. get(9606) # find rows in the Name table with taxa the same as hs # Use Foreign. Key to create condition, equality test # between objects condition = (Name. q. taxa == hs) for n in Name. select(condition): print n # find rows in the Name table corresonding to hs # Easy shortcut, using Multiple. Join iterable for n in hs. names: print n. name, "|", n. name. Class # More general conditions condition = Name. q. name. startswith('Da') for n in Name. select(condition): print n. name, "|", n. name. Class BCHB 524 - Edwards 14

Exercises l Read through the SQLObject documentation l Write a python program using SQLObject to lookup the scientific name for a user-supplied organism name. BCHB 524 - Edwards 15