support tsv: and ssv: file path prefixes, as well as csv:

[simonmichael]

As reported in chat, -f tsv:- and -f ssv:- don't work like -f csv:-. That seems natural to expect, so we'd like to support it.

Currently hledger's file path prefixes must be a canonical format (internally: reader) name, and the name of the reader that handles CSV, SSV and TSV is csv.

[hledger internally conflates "reader" and "data format", and thinks of csv/ssv/tsv as one format with just a different separator setting, but humans think of csv/ssv/tsv as three different formats.]

Here is a patch that partially adds support for ssv: and tsv: prefixes:

--- a/hledger-lib/Hledger/Read/JournalReader.hs
+++ b/hledger-lib/Hledger/Read/JournalReader.hs
@@ -151,2 +151,4 @@ readerNames = map rFormat (readers'::[Reader IO])
 -- Find the reader named by @mformat@, if provided.
+-- ("ssv" and "tsv" are recognised as alternate names for the csv reader,
+-- which also handles those formats.)
 -- Or, if a file path is provided, find the first reader that handles
@@ -155,3 +157,3 @@ findReader :: MonadIO m => Maybe StorageFormat -> Maybe FilePath -> Maybe (Reade
 findReader Nothing Nothing     = Nothing
-findReader (Just fmt) _        = headMay [r | r <- readers', rFormat r == fmt]
+findReader (Just fmt) _        = headMay [r | r <- readers', let rname = rFormat r, rname == fmt || (rname=="csv" && fmt `elem` ["ssv","tsv"])]
 findReader Nothing (Just path) =
@@ -170,2 +172,7 @@ type PrefixedFilePath = FilePath
 -- split that off. Eg "csv:-" -> (Just "csv", "-").
+-- These reader prefixes can be used to force a specific reader,
+-- overriding the file extension. They are the readers' canonical names,
+-- not all the file extensions they recognise. But as a special case,
+-- "csv", "ssv" and "tsv" are all recognised as prefixes selecting the csv reader,
+-- since it handles all three of those formats.
 splitReaderPrefix :: PrefixedFilePath -> (Maybe String, FilePath)
@@ -173,3 +180,3 @@ splitReaderPrefix f =
   headDef (Nothing, f) $
-  [(Just r, drop (length r + 1) f) | r <- readerNames, (r++":") `isPrefixOf` f]
+  [(Just r, drop (length r + 1) f) | r <- readerNames ++ ["ssv","tsv"], (r++":") `isPrefixOf` f]

But it needs to not just select the csv reader, but also somehow instruct it to expect SSV or TSV, ie adjust the separator. See https://hledger.org/1.32/hledger.html#separator , which also seems to imply that ssv: and tsv: already work, wrongly I think.

Help welcome!

[reesmichael1]

I mentioned this over in Matrix too, but just so it's here, I'm going to try diving into this and aim to have something put together within the next few days!