We set out to improve on a cookbook index. github 
CheapandGoodIndex
We start by looking for categories. These are lower case and not indented. Hover for full line.
Line is divided by a string of dots, except when it isn't.
Indented lines are within category. Hover here too.
Preview flattened index before and after kwic index.
http://ward.dojo.fed.wiki/assets/CheapandGoodIndex/kwic.html HEIGHT 600
Some observations as we read what is present. From these we will construct rules and mark when done.
"Index" is a heading, not an item. ✔︎
"Agua Fresca" left-justified and capitalized is an item. ✔︎
"almonds" left-justified might be a category. ✔︎
"Apple Cinnamon Oatmeal" indented is an item. ✔︎
"Thai" indented is ambiguous: "Thai Basil"?
"green" indented is a kind of bean: "green bean"
"sprouts" indented is a kind of sprout: "bean sprouts"?
"tomatoes" splits page number on multiple lines.