Show HN: Convert CSV files into MySQL insert scripts

http://database-import-scriptor.appspot.com/

by d4nt • 12 years ago

41 46 12 years ago

database-import-scriptor.appspot.com

phpnode3• 12 years ago

MySQL already supports loading CSVs directly, why convert them first? http://dev.mysql.com/doc/refman/5.1/en/load-data.html

jeremysmyth3• 12 years ago

It also has the CSV storage engine which actually uses a CSV file as its data back-end, no conversion required.

http://dev.mysql.com/doc/mysql/en/csv-storage-engine.html

corobo• 12 years ago

Yeah.. I'd be interested in something that does the opposite of this tool actually. LOAD DATA INFILE is a lot faster than SQL query files for heavier data

blowski2• 12 years ago

There are occasions where you can't just do a straight insert. Perhaps you need to conditionally create a foreign record, or you want to format a value on insert instead of doing it directly in the CSV.

That said, I can't see how the OP actually helps me with that.

jeremysmyth3• 12 years ago

LOAD DATA INFILE can't conditionally create a foreign record, but you can certainly format a value on insert by using the SET clause, for example:

    LOAD DATA INFILE 'people.txt'
    INTO TABLE Person (@first, @last, date_of_birth)
    SET full_name = CONCAT(LEFT(@first, 1), '. ', @last);

You can even do funky stuff like lookups:

    LOAD DATA INFILE 'projects.txt'
    INTO TABLE Projects(@first, @last, project)
    SET person_id = 
     (SELECT id FROM people 
      WHERE name = CONCAT(@first, ' ', @last) LIMIT 1);

eli2• 12 years ago

Much more frequently I run into the issue that my csv isn't perfectly formed. Some people have funny ideas about how to handle newlines or escape quotes/commas.

abluecloud• 12 years ago

I've found phpMyAdmin's CSV insert feature has been very useful.

Allows for quite a few options.

onion2k• 12 years ago

Tried it with a file containing this:

    Field1,Field2,Field3,Int,Float,Quoted
    This,is,a,12,1.2,"""file"""

...and it returned an empty textarea.

agildehaus2• 12 years ago

This submits to your server. If I have sensitive data, I sure as hell don't want that going somewhere where it might be stored.

Why isn't this just done client-side in Javascript?

jbrooksuk2• 12 years ago

Probably because it'd be awfully slow.

phpnode3• 12 years ago

that's unlikely to be the case, JS is pretty fast

modoc• 12 years ago

Because the author prefers to code in something else?

easytiger2• 12 years ago

why isn't this a trivial bash script more like it.

agildehaus2• 12 years ago

Indeed, but that's already been done.

Gonzih• 12 years ago

I think it can be done in few lines of ruby/python/whatever code. Without submitting sensitive data to remote unknown server. IMHO it would be much better idea to release code as opensource.

pi-rat• 12 years ago

Or just do something like:

  LOAD DATA INFILE 'data.csv' INTO TABLE mytable
    FIELDS TERMINATED BY ',' ENCLOSED BY '"'
    LINES TERMINATED BY '\r\n'
    IGNORE 1 LINES;

akx2• 12 years ago

If you're fortunate enough to be able to put `data.csv` on the MySQL server, that is.

jeremysmyth3• 12 years ago

...or if you can use the LOAD DATA LOCAL INFILE syntax on that server/client.

http://dev.mysql.com/doc/refman/5.6/en/load-data-local.html

masklinn2• 12 years ago

Add LOCAL (as in, "LOAD DATA LOCAL INFILE") and the path is looked up in the client's filesystem, with the file uploaded through the connection before the command is run.

pessimizer• 12 years ago

http://neilb.bitbucket.org/csvfix/manual/csvfix16/sql_insert...

eli2• 12 years ago

The first python script I ever wrote does csv to SQL: https://github.com/elidickinson/csv-tools/blob/master/csv2sq... It's very rough, but I actually still use it on occasion. The Python csv library is quite good. There's a lot that can go wrong that LOAD DATA can't handle.

itazula• 12 years ago

I submitted the following file (note the blanks): $ cat testblanks.csv 1997, Ford, E350

The following was returned: CREATE TABLE testblanks (Column_1 NUMERIC(32, 16),Column_2 VARCHAR(5) CHARACTER SET utf8,Column_3 VARCHAR(5) CHARACTER SET utf8); INSERT INTO testblanks VALUES (1997,' Ford',' E350');

That was in accordance with http://tools.ietf.org/html/rfc4180

Cool!

But the CSV format has a lot of "gotchas." See, for example, http://discuss.fogcreek.com/joelonsoftware3/default.asp?cmd=.... Especially the rant about the Microsoft version of CSV.

masklinn2• 12 years ago

But the CSV format has a lot of "gotchas."

In the wild, CSV is less a format and more a meta-format (and a subset of the *SV family) with plenty of variability between dialects.

sarciszewski2• 12 years ago

In my experience, this is actually not that difficult or novel, but kudos on trying new things. Keep it up :)

lifeisstillgood• 12 years ago

Absolutely. I did come here to do the "so what" dance but this is the simplest and most respectful comment here.

Keep on making new (to you) things, keep a good log of what you have done (GitHub) - eventually you will make new (to anyone) things.

clxl• 12 years ago

I'm surprised by how many things that aren't that difficult or novel end up on the front page. CSV related projects come to the fore every week.

sarciszewski2• 12 years ago

Heh, maybe I should start submitting some of my more trivial projects and rack up the HN karma ;)

i.e. "A testing utility for database migrations"

quasiben• 12 years ago

This is great! People should have better and faster ways of handling CSV -> DB interactions. We recently added native CSV loading into blaze for MySQL, PostgreSQL, and SQLITE, https://github.com/ContinuumIO/blaze/blob/master/blaze/data/...

MongoDB is coming soon.

mathnode• 12 years ago

I am not putting 1TB+ per table dumps of data into a strangers website what can be done with a bit of sed or a script in ${LANGUAGE}.

Loader from Several Nines, is a tool for doing parallel loads of CSV into MySQL and MariaDB https://github.com/severalnines/loader

Mydumper is a threaded dump and load tool for MySQL/MariaDB. https://launchpad.net/mydumper

Tungsten Replicate can do batch load replication into any external datasource http://docs.continuent.com/tungsten-replicator-3.0/deploymen... So long as you have a script for it: https://code.google.com/p/tungsten-replicator/source/browse/...

EDIT: A fun extra, https://mariadb.com/kb/en/mariadb/documentation/storage-engi...

digital-rubber3• 12 years ago

think it would be actually funny if you did, bogus data that is of course.

But uploading your files to some site that reformats plain text, this does not belong on hacker news imho.

squidmccactus• 12 years ago

So does this work with CSV?

NicoJuicy2• 12 years ago

I actually have some code (.Net) that loads xml, csv, ... into objects.

There is also a visual drawing to help you connect the dots between the properties of the uploaded file vs. the properties of the objects.

This way, all information gets loaded into the db from a single import, in the correct table.

I could seperate it in a opensource project i suppose, if anyone is interested.

kstenson• 12 years ago

I would be interested in looking at that if you opensource it.

NicoJuicy2• 12 years ago

You're a .Net developer then?

digital-rubber3• 12 years ago

What is this for ridiculous product/ attempt?

HACKER news, not noobs-that-cant-do-shit.

digital-rubber3• 12 years ago

Quite amusing, people down voting my 'minimal worded critique', apparently really like this tool and see a benefit that is non-existing. Apparently can't think beyond a solution to a problem.

Reformatting of ascii/text data, a csv, is not rocket science. Nor 'hacker (news) quality'.

And i hope that nobody afterwards realises that they should not have uploaded their csv file to your website hosted by nsa friendly google. Also adding a warning that people are submitting their data to (google, 3rd party servers you don't control) servers, would be nicer, and makes people perhaps aware that their data goes further then only their own browser.

easytiger2• 12 years ago

you were not elegant or polite, but are in fact correct

vijayr• 12 years ago

Nice idea.

Can I suggest another idea? Take JSON files, and convert them into MySQL, postgres etc insert/update scripts. There is a need for this. For example, I run this hobby site http://kivatools.com that imports tons of data from kiva.org - their data dumps are in json/xml format. so I've a script that refreshes the db every night, with a couple of gigs of data. It would be very useful if there was a service that can take a url (or file), and convert json/xml into mysql scripts.

sheetjs• 12 years ago

You should be able to do this entire thing in-browser. Going further, leveraging WebSQL or SQL.js, it should be possible to pump the queries into a client-side database and run queries.

Shameless plug: I built an in-browser tool to generate SQL queries from Excel XLS/XLSX/XLSB files: http://sheetjs.com/sexql/

Discussion: https://news.ycombinator.com/item?id=7392665

gregcsv• 12 years ago

Here's one I made a few years ago:

http://utilitymill.com/utility/csv_to_create_table/

You just paste the content from excel or a csv.

It actually tries to figure out the data types and creates a table for you as well.

stef25• 12 years ago

Tested it out with GeoIPCountryWhois.csv from Maxmind, seems to work ok. However selecting the output is almost impossible, locks up Chrome. This isn't your apps fault of course. Usually I use Sequel Pro for this, which allows me to map csv fields to db columns

Show HN: Convert CSV files into MySQL insert scripts

Related Stories