Count number of words across an array of strings - mongodb

Question

I have the following aggregation:
db.subtitles.aggregate()
.match({})
.group({_id: {chunkId: "$chunk_id"}, text: { $push:"$text"}})
What this will render is a result so:
{
"_id" : {
"chunkId" : "ffdd704b-c441-4b49-a32e-fc2277d99250"
},
"text" : [
"Mula doon, sumasama ako sa grocery, sa palengke, sinusundan ko saan napupunta ang pera.",
"Nagkakaroon sila ng resibo na makikita sa kanilang device.",
"Parang ganun na nga, pero…",
"Kaya parang akong naging buhay na QuickBooks. Gusto ko malaman kung ano ang ginagawa ng mga tao sa pera, magkano kinita nila. ",
"Sa kanilang email o text ay may impormasyon na masasabi mo na \"Itong numero na ito, itong text ay galing halimbawa sa Bank of America, at kumpirmado ito\"",
"Mga 4,500 na interbyu o mahigit pa. Sa buong Silangang Africa, sub Saharan Africa at sa Timog Asia.",
"Sa mga umuusbong na merkado, kapag nagbabayad sila ng kuryente, o kapag sumweldo sila.",
"Hindi ko na gustong makita ang nangyari 3 taon nakalipas. Nais ko lang malaman kung kaya mo itong bayaran sa katapusan ng buwan.",
"Saan ako magpunta?"
]
},
…
What I'd like to do is add another field to this group that gives me a total word count for the text array. In this case roughly 136 words.
How could I adjust my aggregation to accomplish this?

Answer 1

You can calculate words before grouping, so you don't need to deal with array of strings but with a single "text" field.
Starting from v4.2 you can benefit from $regexFindAll operator:
db.subtitles.aggregate([
{ $match: {} },
{ $addFields: { words: { $size: { $regexFindAll: { input: "$text", regex: /\w+/ }}}}},
{ $group: {_id: {chunkId: "$chunk_id"}, text: { $push:"$text"}, words: {$sum: "$words"}}}
])
Please read the docs regarding collation to ensure proper behaviour of \w+ regexp. You may want to add some other characters there, e.g. apostrophe etc depending on language. Precise counting may require quite sophisticated regexs especially for non-english strings. See Regex word count - matching words with apostrophe for inspiration.

Answer 2

You can use $stLenCP and $addFields stage
db.subtitle.aggregate([
{ $match: { "_id": ObjectId("5d5b889c33acba0b89b97cda") } },
{ $addFields: {
"length": {
$strLenCP: {
$reduce: {
input: "$text",
initialValue: "",
in: { $concat: ["$$value", "$$this"] }
}
}
}
}}
])