I'm trying to automate writing CSV files to an RSQLite DB.
I am doing so by indexing csvFiles, which is a list of data.frame variables stored in the environment.
I can't seem to figure out why my dbWriteTable() code works perfectly fine when I enter it manually but not when I try to index the name and value fields.
### CREATE DB ###
mydb <- dbConnect(RSQLite::SQLite(),"")
# FOR LOOP TO BATCH IMPORT DATA INTO DATABASE
for (i in 1:length(csvFiles)) {
dbWriteTable(mydb,name = csvFiles[i], value = csvFiles[i], overwrite=T)
i=i+1
}
# EXAMPLE CODE THAT SUCCESSFULLY MANUAL IMPORTS INTO mydb
dbWriteTable(mydb,"DEPARTMENT",DEPARTMENT)
When I run the for loop above, I'm given this error:
"Error in file(file, "rt") : cannot open the connection
In addition: Warning message:
In file(file, "rt") :
cannot open file 'DEPARTMENT': No such file or directory
# note that 'DEPARTMENT' is the value of csvFiles[1]
Here's the dput output of csvFiles:
c("DEPARTMENT", "EMPLOYEE_PHONE", "PRODUCT", "EMPLOYEE", "SALES_ORDER_LINE",
"SALES_ORDER", "CUSTOMER", "INVOICES", "STOCK_TOTAL")
I've researched this error and it seems to be related to my working directory; however, I don't really understand what to change, as I'm not even trying to manipulate files from my computer, simply data.frames already in my environment.
Please help!
Simply use get() for the value argument as you are passing a string value when a dataframe object is expected. Notice your manual version does not have DEPARTMENT quoted for value.
# FOR LOOP TO BATCH IMPORT DATA INTO DATABASE
for (i in seq_along(csvFiles)) {
dbWriteTable(mydb,name = csvFiles[i], value = get(csvFiles[i]), overwrite=T)
}
Alternatively, consider building a list of named dataframes with mget and loop element-wise between list's names and df elements with Map:
dfs <- mget(csvfiles)
output <- Map(function(n, d) dbWriteTable(mydb, name = n, value = d, overwrite=T), names(dfs), dfs)
Related
I have a snakemake rule aggregating several result files to a single file, per study. So to make it a bit more understandable; I have two roles ['big','small'] that each produce data for 5 studies ['a','b','c','d','e'], and each study produces 3 output files, one per phenotype ['xxx','yyy','zzz']. Now what I want is a rule to aggregate the phenotype results from each study to a single summary file per study (so merging the phenotypes into a single table). In the merge_results rule I give the rule a list of files (per study and role), and aggregate these using a pandas frame, and then spit out the result as a single file.
In the process of merging the results I need the 'pheno' variable from the input file being iterated over. Since pheno is not needed in the aggregated output file, it is not provided in output and as a consequence it is also not available in the wildcards object. Now to get a hold of the pheno I parse the filename to grab it, however this all feels very hacky and I suspect there is something here I have not understood properly. Is there a better way to grab wildcards from input files not used in output files in a better way?
runstudy = ['a','b','c','d','e']
runpheno = ['xxx','yyy','zzz']
runrole = ['big','small']
rule all:
input:
expand(os.path.join(output, '{role}-additive', '{study}', '{study}-summary-merge.txt'), role=runrole, study=runstudy)
rule merge_results:
input:
expand(os.path.join(output, '{{role}}', '{{study}}', '{pheno}', '{pheno}.summary'), pheno=runpheno)
output:
os.path.join(output, '{role}', '{study}', '{study}-summary-merge.txt')
run:
import pandas as pd
import os
# Iterate over input files, read into pandas df
tmplist = []
for f in input:
data = pd.read_csv(f, sep='\t')
# getting the pheno from the input file and adding it to the data frame
pheno = os.path.split(f)[1].split('.')[0]
data['pheno'] = pheno
tmplist.append(data)
resmerged = pd.concat(tmplist)
resmerged.to_csv(output, sep='\t')
You are doing it the right way !
In your line:
expand(os.path.join(output, '{{role}}', '{{study}}', '{pheno}', '{pheno}.summary'), pheno=runpheno)
you have to understand that role and study are wildcards. pheno is not a wildcard and is set by the second argument of the expand function.
In order to get the phenotype if your for loop, you can either parse the file name like you are doing or directly reconstruct the file name since you know the different values that pheno takes and you can access the wildcards:
run:
import pandas as pd
import os
# Iterate over phenotypes, read into pandas df
tmplist = []
for pheno in runpheno:
# conflicting variable name 'output' between a global variable and the rule variable here. Renamed global var outputDir for example
file = os.path.join(outputDir, wildcards.role, wildcards.study, pheno, pheno+'.summary')
data = pd.read_csv(file, sep='\t')
data['pheno'] = pheno
tmplist.append(data)
resmerged = pd.concat(tmplist)
resmerged.to_csv(output, sep='\t')
I don't know if this is better than parsing the file name like you were doing though. I wanted to show that you can access wildcards in the code. Either way, you are defining the input and output correctly.
There are different lists available in pelicanconf.py such as
SOCIAL = (('Facebook','www.facebook.com'),)
LINKS =
etc.
I want to manage these content and create my own lists by loading these values from an external file which can be edited independently. I tried importing data as a text file using python but it doesn't work. Is there any other way?
What exactly did not work? Can you provide code?
You can execute arbitrary python code in your pelicanconf.py.
Example for a very simple CSV reader:
# in pelicanconf.py
def fn_to_list(fn):
with open(fn, 'r') as res:
return tuple(map(lambda line: tuple(line[:-1].split(';')), res.readlines()))
print(fn_to_list("data"))
CSV file data:
A;1
B;2
C;3
D;4
E;5
F;6
Together, this yields the following when running pelican:
# ...
((u'A', u'1'), (u'B', u'2'), (u'C', u'3'), (u'D', u'4'), (u'E', u'5'), (u'F', u'6'))
# ...
Instead of printing you can also assign this list to a variable, say LINKS.
I am using Jython and wish to import a text file that contains many configuration values such as:
QManager = MYQM
ProdDBName = MYDATABASE
etc.
.. and then I am reading the file line by line.
What I am unable to figure out is now that as I read each line and have assigned whatever is before the = sign to a local loop variable named MYVAR and assigned whatever is after the = sign to a local loop variable MYVAL - how do I ensure that once the loop finishes I have a bunch of global variables such as QManager & ProdDBName etc.
I've been working on this for days - I really hope someone can help.
Many thanks,
Bret.
See other question: Properties file in python (similar to Java Properties)
Automatically setting global variables is not a good idea for me. I would prefer global ConfigParser object or dictionary. If your config file is similar to Windows .ini files then you can read it and set some global variables with something like:
def read_conf():
global QManager
import ConfigParser
conf = ConfigParser.ConfigParser()
conf.read('my.conf')
QManager = conf.get('QM', 'QManager')
print('Conf option QManager: [%s]' % (QManager))
(this assumes you have [QM] section in your my.conf config file)
If you want to parse config file without help of ConfigParser or similar module then try:
my_options = {}
f = open('my.conf')
for line in f:
if '=' in line:
k, v = line.split('=', 1)
k = k.strip()
v = v.strip()
print('debug [%s]:[%s]' % (k, v))
my_options[k] = v
f.close()
print('-' * 20)
# this will show just read value
print('Option QManager: [%s]' % (my_options['QManager']))
# this will fail with KeyError exception
# you must be aware of non-existing values or values
# where case differs
print('Option qmanager: [%s]' % (my_options['qmanager']))
I am processing data from a set of files which contain a date stamp as part of the filename. The data within the file does not contain the date stamp. I would like to process the filename and add it to one of the data structures within the script. Is there a way to do that within Pig Latin (an extension to PigStorage maybe?) or do I need to preprocess all of the files using Perl or the like beforehand?
I envision something like the following:
-- Load two fields from file, then generate a third from the filename
rawdata = LOAD '/directory/of/files/' USING PigStorage AS (field1:chararray, field2:int, field3:filename);
-- Reformat the filename into a datestamp
annotated = FOREACH rawdata GENERATE
REGEX_EXTRACT(field3,'*-(20\d{6})-*',1) AS datestamp,
field1, field2;
Note the special "filename" datatype in the LOAD statement. Seems like it would have to happen there as once the data has been loaded it's too late to get back to the source filename.
You can use PigStorage by specify -tagsource as following
A = LOAD 'input' using PigStorage(',','-tagsource');
B = foreach A generate INPUT_FILE_NAME;
The first field in each Tuple will contain input path (INPUT_FILE_NAME)
According to API doc http://pig.apache.org/docs/r0.10.0/api/org/apache/pig/builtin/PigStorage.html
Dan
The Pig wiki as an example of PigStorageWithInputPath which had the filename in an additional chararray field:
Example
A = load '/directory/of/files/*' using PigStorageWithInputPath()
as (field1:chararray, field2:int, field3:chararray);
UDF
// Note that there are several versions of Path and FileSplit. These are intended:
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.mapreduce.RecordReader;
import org.apache.hadoop.mapreduce.lib.input.FileSplit;
import org.apache.pig.backend.hadoop.executionengine.mapReduceLayer.PigSplit;
import org.apache.pig.builtin.PigStorage;
import org.apache.pig.data.Tuple;
public class PigStorageWithInputPath extends PigStorage {
Path path = null;
#Override
public void prepareToRead(RecordReader reader, PigSplit split) {
super.prepareToRead(reader, split);
path = ((FileSplit)split.getWrappedSplit()).getPath();
}
#Override
public Tuple getNext() throws IOException {
Tuple myTuple = super.getNext();
if (myTuple != null)
myTuple.append(path.toString());
return myTuple;
}
}
-tagSource is deprecated in Pig 0.12.0 .
Instead use
-tagFile - Appends input source file name to beginning of each tuple.
-tagPath - Appends input source file path to beginning of each tuple.
A = LOAD '/user/myFile.TXT' using PigStorage(',','-tagPath');
DUMP A ;
will give you the full file path as first column
( hdfs://myserver/user/blo/input/2015.TXT,439,43,05,4,NAVI,PO,P&C,P&CR,UC,40)
Refrence: http://pig.apache.org/docs/r0.12.0/api/org/apache/pig/builtin/PigStorage.html
A way to do this in Bash and PigLatin can be found at: How Can I Load Every File In a Folder Using PIG?.
What I've been doing lately though, and find to be much cleaner is embedding Pig in Python. That let's you throw all sorts of variables and such between the two. A simple example is:
#!/path/to/jython.jar
# explicitly import Pig class
from org.apache.pig.scripting import Pig
# COMPILE: compile method returns a Pig object that represents the pipeline
P = Pig.compile(
"a = load '$in'; store a into '$out';")
input = '/path/to/some/file.txt'
output = '/path/to/some/output/on/hdfs'
# BIND and RUN
results = P.bind({'in':input, 'out':output}).runSingle()
if results.isSuccessful() :
print 'Pig job succeeded'
else :
raise 'Pig job failed'
Have a look at Julien Le Dem's great slides as an introduction to this, if you're interested. There's also a ton of documentation at http://pig.apache.org/docs/r0.9.2/cont.pdf.
How can I execute whole sql file into database using SQLAlchemy? There can be many different sql queries in the file including begin and commit/rollback.
sqlalchemy.text or sqlalchemy.sql.text
The text construct provides a straightforward method to directly execute .sql files.
from sqlalchemy import create_engine
from sqlalchemy import text
# or from sqlalchemy.sql import text
engine = create_engine('mysql://{USR}:{PWD}#localhost:3306/db', echo=True)
with engine.connect() as con:
with open("src/models/query.sql") as file:
query = text(file.read())
con.execute(query)
SQLAlchemy: Using Textual SQL
text()
I was able to run .sql schema files using pure SQLAlchemy and some string manipulations. It surely isn't an elegant approach, but it works.
# Open the .sql file
sql_file = open('file.sql','r')
# Create an empty command string
sql_command = ''
# Iterate over all lines in the sql file
for line in sql_file:
# Ignore commented lines
if not line.startswith('--') and line.strip('\n'):
# Append line to the command string
sql_command += line.strip('\n')
# If the command string ends with ';', it is a full statement
if sql_command.endswith(';'):
# Try to execute statement and commit it
try:
session.execute(text(sql_command))
session.commit()
# Assert in case of error
except:
print('Ops')
# Finally, clear command string
finally:
sql_command = ''
It iterates over all lines in a .sql file ignoring commented lines.
Then it concatenates lines that form a full statement and tries to execute the statement. You just need a file handler and a session object.
You can do it with SQLalchemy and psycopg2.
file = open(path)
engine = sqlalchemy.create_engine(db_url)
escaped_sql = sqlalchemy.text(file.read())
engine.execute(escaped_sql)
Unfortunately I'm not aware of a good general answer for this. Some dbapi's (psycopg2 for instance) support executing many statements at a time. If the files aren't huge you can just load them into a string and execute them on a connection. For others, I would try to use a command-line client for that db and pipe the data into that using the subprocess module.
If those approaches aren't acceptable, then you'll have to go ahead and implement a small SQL parser that can split the file apart into separate statements. This is really tricky to get 100% correct, as you'll have to factor in database dialect specific literal escaping rules, the charset used, any database configuration options that affect literal parsing (e.g. PostgreSQL standard_conforming_strings).
If you only need to get this 99.9% correct, then some regexp magic should get you most of the way there.
If you are using sqlite3 it has a useful extension to dbapi called conn.executescript(str), I've hooked this up via something like this and it seemed to work: (Not all context is shown but it should be enough to get the drift)
def init_from_script(script):
Base.metadata.drop_all(db_engine)
Base.metadata.create_all(db_engine)
# HACK ALERT: we can do this using sqlite3 low level api, then reopen session.
f = open(script)
script_str = f.read().strip()
global db_session
db_session.close()
import sqlite3
conn = sqlite3.connect(db_file_name)
conn.executescript(script_str)
conn.commit()
db_session = Session()
Is this pure evil I wonder? I looked in vain for a 'pure' sqlalchemy equivalent, perhaps that could be added to the library, something like db_session.execute_script(file_name) ? I'm hoping that db_session will work just fine after all that (ie no need to restart engine) but not sure yet... further research needed (ie do we need to get a new engine or just a session after going behind sqlalchemy's back?)
FYI sqlite3 includes a related routine: sqlite3.complete_statement(sql) if you roll your own parser...
You can access the raw DBAPI connection through this
raw_connection = mySqlAlchemyEngine.raw_connection()
raw_cursor = raw_connection() #get a hold of the proxied DBAPI connection instance
but then it will depend on which dialect/driver you are using which can be referred to through this list.
For pyscog2, you can just do
raw_cursor.execute(open("my_script.sql").read())
but pysqlite you would need to do
raw_cursor.executescript(open("my_script").read())
and in line with that you would need to check the documentation of whichever DBAPI driver you are using to see if multiple statements are allowed in one execute or if you would need to use a helper like executescript which is unique to pysqlite.
Here's how to run the script splitting the statements, and running each statement directly with a "connectionless" execution with the SQLAlchemy Engine. This assumes that each statement ends with a ; and that there's no more than one statement per line.
engine = create_engine(url)
with open('script.sql') as file:
statements = re.split(r';\s*$', file.read(), flags=re.MULTILINE)
for statement in statements:
if statement:
engine.execute(text(statement))
In the current answers, I did not found a solution which works when a combination of these features in the .SQL file is present:
Comments with "--"
Multi-line statements with additional comments after "--"
Function definitions which have multiple SQL-queries ending with ";" butmust be executed as a whole statement
A found a rather simple solution:
# check for /* */
with open(file, 'r') as f:
assert '/*' not in f.read(), 'comments with /* */ not supported in SQL file python interface'
# we check out the SQL file line-by-line into a list of strings (without \n, ...)
with open(file, 'r') as f:
queries = [line.strip() for line in f.readlines()]
# from each line, remove all text which is behind a '--'
def cut_comment(query: str) -> str:
idx = query.find('--')
if idx >= 0:
query = query[:idx]
return query
# join all in a single line code with blank spaces
queries = [cut_comment(q) for q in queries]
sql_command = ' '.join(queries)
# execute in connection (e.g. sqlalchemy)
conn.execute(sql_command)
Code bellow works for me in alembic migrations
from alembic import op
import sqlalchemy as sa
from ekrec.common import get_project_root
def upgrade():
path = f'{get_project_root()}/migrations/versions/fdb8492f75b2_.sql'
op.execute(open(path).read())
I had success with David's answer here, with two slight modifications:
Use get_bind() as I was working with a Session rather than an Engine
Call cursor() on the raw connection
raw_connection = myDbSession.get_bind().raw_connection()
raw_cursor = raw_connection.cursor()
raw_cursor.execute(open("my_script.sql").read())