.gitignore

*.pyc

csv2dict.py

import csv


def dictify(filename):
	"""
	turn data from CSV file into a list of dictionaries
	"""
	f = open(filename, "rU")

	items = []
	for row in unicode_csv_reader(f):
		try:
			item = dict([(k, row[i]) for i, k in enumerate(keys)])
			item["_source"] = filename # XXX: not a generic requirement
			items.append(item)
		except NameError: # first row
			keys = row

	f.close()
	return items


# Unicode fixes as per documentation
def unicode_csv_reader(data, dialect=csv.excel, **kwargs):
	"""
	expects Unicode input, temporarily encoding it as UTF-8
	"""
	reader = csv.reader(utf8_encoder(data), dialect=dialect, **kwargs)
	for row in reader:
		yield [unicode(cell, "utf-8", 'replace') for cell in row]


def utf8_encoder(data):
	for line in data:
		yield line

csv2tiddlers.py

#!/usr/bin/env python

"""
Usage:
  $ ./csv2tiddlers.py <filename> <titlefields> <textfields> <fieldmap> <tagsmap>

Example:
  $ ./csv2tiddlers.py sample.csv "colA;colC" "colD;colE" "colB=foo;colF=bar"
"""

import sys

from csv2dict import dictify
from tiddler import put


HOST = "http://localhost:8080"
BAG = "common"


def main(args):
	"""
	arguments are filename, titlefields, textfields and fieldmap

	fields are a list of semicolon-separated column names whose values will be
	concatenated for the respective tiddler attribute

	fieldmap provides a way to translate column names to tiddler fields
	tagsmap provides a way to translate column values into tags
	"""
	args = [unicode(arg, "utf-8", 'replace') for arg in args]
	try:
		filename = args[1]
		titlefields = args[2].split(";")
		textfields = args[3].split(";")
	except IndexError, exc:
		print "ERROR: missing argument", exc
		return False
	try:
		fieldmap = dict([i.split("=") for i in args[4].split(";")])
	except IndexError, exc:
		fieldmap = {}
	try:
		tagsmap = args[5].split(";")
	except IndexError, exc:
		tagsmap = []

	items = dictify(filename)
	put_tiddlers(items, titlefields, textfields, fieldmap, tagsmap)

	return True


def put_tiddlers(items, titlefields, textfields, fieldmap, tagsmap):
	for item in items:
		values = []
		for field in titlefields:
			values.append(item[field])
		title = "_".join(values)

		try:
			values = []
			for field in textfields: # TODO: DRY
				values.append(item[field])
			text = "\n\n".join(values)
		except KeyError:
			text = None # XXX: invalid?

		for column, field in fieldmap.items(): # XXX: inefficient!?
			try: # XXX: temporary workaround
				item[field] = item[column]
				del item[column]
			except KeyError:
				pass

		try:
			tagsCol, tagsDelim = tagsmap
			tags = item[tagsCol].split(tagsDelim)
		except (ValueError, KeyError), exc:
			tags = []

		tiddler = {
			"text": text,
			"tags": tags,
			"fields": item
		}

                message = "sending tiddler: %s ..." % title.encode('utf-8')
		sys.stdout.write(message)
		status = put(HOST, BAG, title, tiddler)
		print "\t[%s]" % status


if __name__ == "__main__":
	status = not main(sys.argv)
	sys.exit(status)

tiddler.py

"""
PUT a tiddler.
"""

import json
import urllib2
import httplib2


def put(host, bag, title, tiddler):
	"""
	send a dictionary representing a tiddler to a TiddlyWeb server
	"""
	http = httplib2.Http()
	bag = urllib2.quote(bag.encode('utf-8'))
	title = urllib2.quote(title.encode('utf-8'))
	uri = "%s/bags/%s/tiddlers/%s" % (host, bag, title)
	response, content = http.request(
		uri,
		method="PUT",
		headers={ "Content-Type": "application/json" },
		body=json.dumps(tiddler)
	)
	return response["status"]