79: convert: add a skip-lines directive (Magnus Henoch)

MANUAL.md

@@ -517,11 +517,13 @@ file with `--rules-file` (useful when reading from standard input).
 An example - sample.csv:
 
     sample.csv:
+    "Date","Note","Amount"
     "2012/3/22","TRANSFER TO SAVINGS","-10.00"
     "2012/3/23","SOMETHING ELSE","5.50"
 
 sample.rules:
 
+    skip-lines 1
     date-field 0
     description-field 1
     amount-field 2
@@ -632,6 +634,11 @@ results.)
 
 > Which field contains the transaction cleared status (`*`).
 
+`skip-lines`
+
+> How many lines to skip in the beginning of the file, e.g. to skip a
+> line of column headings.
+
 Account-assigning rules select an account to transfer to based on the
 description field (unless `account2-field` is used.) Each
 account-assigning rule is a paragraph consisting of one or more

hledger-lib/Hledger/Data/Types.hs

@@ -203,7 +203,8 @@ data CsvRules = CsvRules {
       account2Field :: Maybe FieldPosition,
       effectiveDateField :: Maybe FieldPosition,
       baseAccount :: AccountName,
-      accountRules :: [AccountRule]
+      accountRules :: [AccountRule],
+      skipLines :: Int
 } deriving (Show, Eq)
 
 type FieldPosition = Int

hledger-lib/Hledger/Read/CsvReader.hs

@@ -98,7 +98,8 @@ nullrules = CsvRules {
       account2Field=Nothing,
       effectiveDateField=Nothing,
       baseAccount="unknown",
-      accountRules=[]
+      accountRules=[],
+      skipLines=0
 }
 
 type CsvRecord = [String]
@@ -137,9 +138,10 @@ readJournalFromCsv mrulesfile csvfile csvdata =
   rules <- liftM (either (throw.userError.show) id) $ parseCsvRulesFile rulesfile
 
   let requiredfields = (maxFieldIndex rules + 1)
-      badrecords = take 1 $ filter ((< requiredfields).length) records
+      realrecords = drop (skipLines rules) records
+      badrecords = take 1 $ filter ((< requiredfields).length) realrecords
   return $ case badrecords of
-            []    -> Right nulljournal{jtxns=sortBy (comparing tdate) $ map (transactionFromCsvRecord rules) records}
+            []    -> Right nulljournal{jtxns=sortBy (comparing tdate) $ map (transactionFromCsvRecord rules) realrecords}
             (_:_) -> Left $ "Parse error: at least one CSV record does not contain a field referenced by the conversion rules file:\n"++(show $ head badrecords)
 
 -- | Ensure there is a conversion rules file at the given path, creating a
@@ -189,6 +191,7 @@ newRulesFileContent = let prognameandversion = "hledger" in
     "# Add rules to this file for more accurate conversion, see\n"++
     "# http://hledger.org/MANUAL.html#convert\n" ++
     "\n" ++
+    "skip-lines 0\n" ++
     "base-account assets:bank:checking\n" ++
     "date-field 0\n" ++
     "description-field 4\n" ++
@@ -251,6 +254,7 @@ definitions = do
    ,effectivedatefield
    ,basecurrency
    ,baseaccount
+   ,skiplines
    ,commentline
    ] <?> "definition"
   return ()
@@ -350,6 +354,12 @@ baseaccount = do
   optional newline
   updateState (\r -> r{baseAccount=v})
 
+skiplines = do
+  string "skip-lines"
+  many1 spacenonewline
+  v <- restofline
+  updateState (\r -> r{skipLines=read v})
+
 accountrule :: GenParser Char CsvRules AccountRule
 accountrule = do
   many blankorcommentline