@@ -29,10 +29,13 @@ main = do else do let options = mconcat optionss - -- Read the page files and the statement they contain - pages <- mapM readFile page_paths + -- Read the page files and the statement they contain. + -- For some reason, the Halifax files contain character 0xA0 (160). + -- This terminates the file reading process in text mode for some reason, + -- so let's just read in binary mode for now. + pages <- mapM readBinaryFile page_paths let (account, transactions) = parseStatement pages - hPutStrLn stderr (show transactions) + --hPutStrLn stderr (show transactions) -- Read the rules (if any) and apply them to the transactions from the pages rules <- maybe (return []) (fmap parseRules . readFile) $ opt_rules_file options @@ -43,4 +46,9 @@ main = do QIF -> outputQIF Ledger -> outputLedger CSV -> outputCSV - output_method options account transactions' \ No newline at end of file + output_method options account transactions' + +readBinaryFile :: FilePath -> IO String +readBinaryFile fp = withBinaryFile fp ReadMode $ \h -> do + res <- hGetContents h + length res `seq` return res Finance/Halifax/Main.hs @@ -12,12 +12,12 @@ import Text.HTML.TagSoup import Finance.Halifax.Core import Finance.Halifax.Utilities - parseStatement :: [String] -> (Account, [Transaction]) parseStatement pages = (parseAccountInfo account_info, transactions) where - tagss = map parseTags pages - account_info = fst $ fragment (~== "<table class=bankingSummaryContainer>") (~== "<br>") (head tagss) + -- Halifax seems to include stray non-ASCII characters in the files, for some reason (these are the ones that mean we have to read in binary mode). + tagss = map (parseTags . map (\c -> if isAscii c then c else ' ')) pages + account_info = fst $ fragment (~== "<table class=bankingSummaryContainer>") (~== "<table class=bankingSummaryInfo>") (head tagss) transaction_infoss = map (fst . fragment (~== "<table class=DataTable>") (~== "</table>")) tagss transaction_rowss = map (drop 1 . partitions (~== "<tr>")) transaction_infoss @@ -30,7 +30,7 @@ parseAccountInfo tags = Account { acc_roll_number = get_summary_for "Roll number" } where - get_summary_for = filter (not . isSpace) . fromJust . flip lookup summaries + get_summary_for key = filter (not . isSpace) . maybe (error $ "No value for " ++ key) id . flip lookup summaries $ key summaries = unfoldr unfold_summary tags unfold_summary tags' = case dropWhile (~/= "<td class=summaryBoxesText>") tags' of [] -> Nothing Finance/Halifax/StatementParser.hs cb426d74a971f5a6e5b3c5b0b5eecac26b694618 Max Bolingbroke batterseapower@hotmail.com http://github.com/batterseapower/halifax-import/commit/4cb74d2b89657ec26d65710ef6861c1f419c2790 4cb74d2b89657ec26d65710ef6861c1f419c2790 2009-11-01T14:40:17-08:00 2009-11-01T14:40:17-08:00 Fix parser on 6.12 - stray non-ASCII in files 8e0ed7d78f4a3f7f55b7e5f56a0ce3e55e44c73c Max Bolingbroke batterseapower@hotmail.com