convert tsv quicken reports to json lines with invisible XML

[dckc]

As noted in the 2006 item, I have years of data in this tab-separated format; for example:

3/28/08	D0962	            S	STAPLES 00108084 ANKENY IA	Canon Powershot SD	Stuff:Major:AV	R	-249.99

I played with invisible XML and its playground and turned that into...

<tx num="S">
   <date month="3" day="28" year="08"/>
   <account>D0962</account>
   <description>STAPLES 00108084 ANKENY IA</description>
   <split status="R">
      <memo>Canon Powershot SD</memo>
      <category>Stuff:Major:AV</category>
      <amount>-249.99</amount>
   </split>
</tx>

using...

tx: date, account, @num, description, split.

split: memo, category, @status, amount.

date: @month, -"/", @day, -"/", @year, tab.
month: ["0"-"9"]+.
day: ["0"-"9"]+.
year: ["0"-"9"]+.

account: field, tab.
num: -space?, field, tab.
description: field, tab.

memo: field, tab.
category: field, tab { TODO: split on : ? }.
status: field, tab.
amount: field { TODO: constrain to number syntax? }.

-field: ~[" "], (~[#9])*. { don't start with space to avoid ambiguity in num }
-tab: -#9.
-space: -[" "]+.

TODO:

• handle >1 tx
• handle >1 split
• write JSON lines instead of XML
• find a simple implementation, preferably by aggressively pre-compiling
  • perhaps just use this as a spec for a hand-rolled parser

[dckc]

p.s. I had a hard time finding my earlier implementation: trxtsv.py d5b0312 2008