tools: add sample journals with wide characters

generatejournal now has --chinese and --mixed options, which are used to
generate some additional small sample journals.

Makefile

@@ -794,7 +794,17 @@ ghci-web: \
 		$(call def-help,ghci-web, start a GHCI REPL and load the hledger-lib, hledger and hledger-web packages)
 	stack exec $(GHCI) -- $(BUILDFLAGS) hledger-web/app/main.hs
 
-samplejournals: data/sample.journal data/100x100x10.journal data/1000x1000x10.journal data/1000x10000x10.journal data/10000x1000x10.journal data/10000x10000x10.journal data/100000x1000x10.journal \
+samplejournals: \
+	data/sample.journal \
+	data/100x100x10.journal \
+	data/1000x1000x10.journal \
+	data/1000x10000x10.journal \
+	data/10000x1000x10.journal \
+	data/10000x10000x10.journal \
+	data/100000x1000x10.journal \
+	data/ascii.journal \
+	data/chinese.journal \
+	data/mixed.journal \
 	$(call def-help,samplejournals, regenerate standard sample journals in data/ )
 
 data/sample.journal:
@@ -818,6 +828,15 @@ data/10000x10000x10.journal: tools/generatejournal
 data/100000x1000x10.journal: tools/generatejournal
 	tools/generatejournal 100000 1000 10 >$@
 
+data/ascii.journal: tools/generatejournal
+	tools/generatejournal 3 5 5 >$@
+
+data/chinese.journal: tools/generatejournal
+	tools/generatejournal 3 5 5 --chinese >$@
+
+data/mixed.journal: tools/generatejournal
+	tools/generatejournal 3 5 5 --mixed >$@
+
 ###############################################################################
 $(call def-help-subsection,DOCUMENTATION:)
 

doc/developer-guide.md

@@ -477,7 +477,7 @@ Finally, for quick, fine-grained performance measurements when troubleshooting o
 
 ### Generate sample journal files
 
-Synthetic data files like `data/100x100x10.journal` useful for benchmarks and testing.
+Synthetic data files like `data/100x100x10.journal` are useful for benchmarks and testing.
 The numbers describe the number of transactions, number of accounts, and maximum account depth respectively.
 They are generated by [`tools/generatejournal.hs`](https://github.com/simonmichael/hledger/blob/master/tools/generatejournal.hs).
 They should be built as needed, if not you can use `make samplejournals` rule:
@@ -493,6 +493,9 @@ tools/generatejournal 1000 10000 10 >data/1000x10000x10.journal
 tools/generatejournal 10000 1000 10 >data/10000x1000x10.journal
 tools/generatejournal 10000 10000 10 >data/10000x10000x10.journal
 tools/generatejournal 100000 1000 10 >data/100000x1000x10.journal
+tools/generatejournal 3 5 5 >data/ascii.journal
+tools/generatejournal 3 5 5 --chinese >data/chinese.journal
+tools/generatejournal 3 5 5 --mixed >data/mixed.journal
 ```
 
 ### Run developer tests

tools/generatejournal.hs

@@ -1,30 +1,34 @@
 #!/usr/bin/env runhaskell
 {-
-generateledger.hs NUMTXNS NUMACCTS ACCTDEPTH
-
-Outputs a dummy ledger file with the specified number of transactions,
-number of accounts, and account tree depth. Useful for
-testing/profiling/benchmarking.
+generatejournal.hs NUMTXNS NUMACCTS ACCTDEPTH [--chinese|--mixed]
 
+Outputs a dummy journal file with the specified number of
+transactions, number of accounts, and account tree depth. By default
+it uses only ascii characters, with --chinese it uses wide chinese
+characters, or with --mixed it uses both.  These files are used for
+testing, benchmarking, profiling, etc.
 -}
 
 module Main
 where
-import System.Environment
-import Control.Monad
-import Data.Time.LocalTime
+import Data.Char
+import Data.List
 import Data.Time.Calendar
-import Text.Printf
+import Data.Time.LocalTime
 import Numeric
+import System.Environment
+import Text.Printf
+-- import Hledger.Utils.Debug
 
 main = do
-  args <- getArgs
+  rawargs <- getArgs
+  let (opts,args) = partition (isPrefixOf "-") rawargs
   let [numtxns, numaccts, acctdepth] = map read args :: [Int]
   today <- getCurrentDay
   let (year,_,_) = toGregorian today
   let d = fromGregorian (year-1) 1 1
   let dates = iterate (addDays 1) d
-  let accts = pair $ cycle $ take numaccts $ uniqueacctnames acctdepth
+  let accts = pair $ cycle $ take numaccts $ uniqueAccountNames opts acctdepth
   mapM_ (\(n,d,(a,b)) -> putStr $ showtxn n d a b) $ take numtxns $ zip3 [1..] dates accts
   return ()
 
@@ -35,17 +39,41 @@ showtxn txnno date acct1 acct2 =
       d = show date
       amt = 1::Int
 
-uniqueacctnames :: Int -> [String]
-uniqueacctnames depth = uniqueacctnames' depth uniquenames
-    where uniquenames = map hex [1..] where hex = flip showHex ""
+uniqueAccountNames :: [String] -> Int -> [String]
+uniqueAccountNames opts depth =
+  mkacctnames uniquenames
+  where
+    mkacctnames names = mkacctnamestodepth some ++ mkacctnames rest
+      where
+        (some, rest) = splitAt depth names
+        -- mkacctnamestodepth ["a", "b", "c"] = ["a","a:b","a:b:c"]
+        mkacctnamestodepth :: [String] -> [String]
+        mkacctnamestodepth [] = []
+        mkacctnamestodepth (a:as) = a : map ((a++":")++) (mkacctnamestodepth as)
+    uniquenames
+      | "--mixed" `elem` opts   = concat $ zipWith (\a b -> [a,b]) uniqueNamesHex uniqueNamesWide
+      | "--chinese" `elem` opts = uniqueNamesWide
+      | otherwise               = uniqueNamesHex
 
-uniqueacctnames' depth uniquenames = group some ++ uniqueacctnames' depth rest
-    where (some, rest) = splitAt depth uniquenames
+uniqueNamesHex = map hex [1..] where hex = flip showHex ""
+
+uniqueNamesWide = concat [sequences n wideChars | n <- [1..]]
+
+-- Get the sequences of specified size starting at each element of a list,
+-- cycling it if needed to fill the last sequence. If the list's elements
+-- are unique, then the sequences will be too.
+sequences :: Show a => Int -> [a] -> [[a]]
+sequences n l = go l
+  where
+    go [] = []
+    go l' = s : go (tail l')
+      where
+        s' = take n l'
+        s | length s' == n = s'
+          | otherwise      = take n (l' ++ cycle l)
+
+wideChars = map chr [0x3400..0x4db0]
 
--- group ["a", "b", "c"] = ["a","a:b","a:b:c"]
-group :: [String] -> [String]
-group [] = []
-group (a:as) = a : map ((a++":")++) (group as)
 
 pair :: [a] -> [(a,a)]
 pair [] = []