如何在自定义 arangosearch 分析器中删除空字符串

Question

I've got a custom analyzer that is like text_en but doesn't include the hyphen as a delimiter:

{pipeline:[
 {type:"norm",properties:{
  locale: "en.utf-8", accent: false, case: "lower", stemming: false}},
 {type:"delimiter",properties:{delimiter:" "}},
 {type:"delimiter",properties:{delimiter:"!"}},
 {type:"delimiter",properties:{delimiter:"."}},
 {type:"delimiter",properties:{delimiter:","}},
 {type:"delimiter",properties:{delimiter:";"}},
 {type:"delimiter",properties:{delimiter:"?"}},
 {type:"delimiter",properties:{delimiter:"["}},
 {type:"delimiter",properties:{delimiter:"]"}},
 {type:"delimiter",properties:{delimiter:"{"}},
 {type:"delimiter",properties:{delimiter:"}"}},
 {type:"delimiter",properties:{delimiter:"("}},
 {type:"delimiter",properties:{delimiter:")"}},
 {type:"delimiter",properties:{delimiter:"<"}},
 {type:"delimiter",properties:{delimiter:">"}},
 {type:"delimiter",properties:{delimiter:"~"}},
 {type:"delimiter",properties:{delimiter:"@"}},
 {type:"delimiter",properties:{delimiter:"="}},
 {type:"delimiter",properties:{delimiter:"&"}},
 {type:"delimiter",properties:{delimiter:"|"}},
 {type:"delimiter",properties:{delimiter:"\n"}},
 {type:"stem",properties:{locale:"en.utf-8"}}]}'

The issue is that chaining like this will return empty strings. The tokens for the string "HYPNOS2, Aphrodite and other Microcontrollers." is this:

[
 "hypnos2",
 "",
 "aphrodit",
 "and",
 "other",
 "microcontrol",
 ""
]

What do I do to remove the empty string tokens in this analyzer?

Answer 1

I added another stage in the pipeline after the delimiters. It's an aql with keepNull=false and queryString of return @param==""?null:@param. If someone has a simpler suggestion I'm interested.