TSV parsing

.gitignore

@@ -0,0 +1,38 @@
+notes.txt
+notes.lhs
+dist
+.cabal-sandbox
+cabal.sanbox.config
+cabal.config
+
+# emacs stuff
+*~
+\#*\#
+/.emacs.desktop
+/.emacs.desktop.lock
+*.elc
+auto-save-list
+tramp
+.\#*
+
+# Org-mode
+.org-id-locations
+*_archive
+
+# flymake-mode
+*_flymake.*
+
+# eshell files
+/eshell/history
+/eshell/lastdir
+
+# elpa packages
+/elpa/
+
+# vim stuff
+*.swp
+*.swo
+
+*.key
+_darcs
+darcs*

MissingH.cabal

@@ -52,6 +52,7 @@ Library
   Control.Concurrent.Thread.Utils,
   Network.Email.Sendmail,
     Data.CSV,
+    Data.TSV,
   System.Cmd.Utils,
   Data.BinPacking,
   Data.Progress.Tracker,

README

@@ -175,6 +175,8 @@ MissingH.Str              * Leading/trailing whitespace removal
 
 MissingH.Str.CSV          * Parsing of comma-separated value (CSV) files
 
+MissingH.Str.TSV          * Parsing of tab-separated value (TSV) files
+
 MissingH.Threads          * Threaded callbacks
 
 MissingH.Time             * Utilities for working with times and dates

src/Data/CSV.hs

@@ -1,4 +1,4 @@
-{- arch-tag: CSV and TSV utilities
+{- arch-tag: CSV utilities
 Copyright (c) 2005-2011 John Goerzen <[email protected]>
 
 All rights reserved.
@@ -22,28 +22,13 @@ Written by John Goerzen, jgoerzen\@complete.org
 
 module Data.CSV (csvFile, genCsvFile) where
 import Text.ParserCombinators.Parsec
-import Data.List (intersperse)
-
-eol :: forall st. GenParser Char st String
-eol = (try $ string "\n\r") <|> (try $ string "\r\n") <|> string "\n" <|>
-      string "\r" <?> "End of line"
+import Data.SeperatingValues.SeperatingValues
 
 cell :: GenParser Char st String
-cell = quotedcell <|> many (noneOf ",\n\r")
-
-quotedchar :: GenParser Char st Char
-quotedchar = noneOf "\""
-             <|> (try $ do string "\"\""
-                           return '"'
-                 )
-quotedcell :: CharParser st String
-quotedcell = do char '"'
-                content <- many quotedchar
-                char '"'
-                return content
+cell = cellOfX ','
 
 line :: GenParser Char st [String]
-line = sepBy cell (char ',')
+line = lineOfX ','
 
 {- | Parse a Comma-Separated Value (CSV) file.  The return value is a list of
 lines; each line is a list of cells; and each cell is a String.
@@ -89,14 +74,4 @@ csvFile = endBy line eol
 {- | Generate CSV data for a file.  The resulting string can be
 written out to disk directly. -}
 genCsvFile :: [[String]] -> String
-genCsvFile inp =
-    unlines . map csvline $ inp
-    where csvline :: [String] -> String
-          csvline l = concat . intersperse "," . map csvcells $ l
-          csvcells :: String -> String
-          csvcells "" = ""
-          csvcells c = '"' : convcell c ++ "\""
-          convcell :: String -> String
-          convcell c = concatMap convchar c
-          convchar '"' = "\"\""
-          convchar x = [x]
+genCsvFile inp = genXsvFile "," inp

src/Data/SeperatingValues/SeperatingValues.hs

@@ -0,0 +1,52 @@
+{- |
+   Module     : Data.SeperatingValues.SeperatingValues
+   Copyright  : Copyright (C) 2005-2011 John Goerzen
+   License    : BSD3
+
+   Maintainer : John Goerzen <[email protected]>
+   Stability  : provisional
+   Portability: portable
+
+Misc/Helper Haskell Parsec parsers for any(X)-separated values (XSV) files.
+
+Written by: Aistis Raulinaitis, [email protected]
+-}
+
+module Data.SeperatingValues.SeperatingValues(eol, cellOfX, quotedchar, lineOfX, genXsvFile) where
+
+import Text.ParserCombinators.Parsec
+import Data.List (intersperse)
+
+eol :: forall st. GenParser Char st String
+eol = (try $ string "\n\r") <|> (try $ string "\r\n") <|> string "\n" <|>
+      string "\r" <?> "End of line"
+
+cellOfX :: Char -> GenParser Char st String
+cellOfX x = quotedcell <|> many (noneOf (x : "\n\r"))
+
+quotedchar :: GenParser Char st Char
+quotedchar = noneOf "\""
+             <|> (try $ do string "\"\""
+                           return '"'
+                 )
+quotedcell :: CharParser st String
+quotedcell = do char '"'
+                content <- many quotedchar
+                char '"'
+                return content
+
+lineOfX :: Char -> GenParser Char st [String]
+lineOfX x = sepBy (cellOfX x) (char x)
+
+genXsvFile :: String -> [[String]] -> String
+genXsvFile x inp =
+    unlines . map xsvline $ inp
+    where xsvline :: [String] -> String
+          xsvline l = concat . intersperse x . map xsvcells $ l
+          xsvcells :: String -> String
+          xsvcells "" = ""
+          xsvcells c = '"' : convcell c ++ "\""
+          convcell :: String -> String
+          convcell c = concatMap convchar c
+          convchar '"' = "\"\""
+          convchar x = [x]

src/Data/TSV.hs

@@ -0,0 +1,70 @@
+{- |
+   Module     : Data.TSV
+   Copyright  : Copyright (C) 2005-2011 John Goerzen
+   License    : BSD3
+
+   Maintainer : John Goerzen <[email protected]>
+   Stability  : provisional
+   Portability: portable
+
+Haskell Parsec parsers for tab-separated value (TSV) files.
+
+Written by: Aistis Raulinaitis, [email protected]
+-}
+module Data.TSV where
+
+import Text.ParserCombinators.Parsec
+import Data.SeperatingValues.SeperatingValues
+
+cell :: GenParser Char st String
+cell = cellOfX '\t'
+
+line :: GenParser Char st [String]
+line = lineOfX '\t'
+
+{- | Parse a Tab-Separated Value (TSV) file.  The return value is a list of
+lines; each line is a list of cells; and each cell is a String.
+
+Please note that TSV files may have a different number of cells on each line.
+Also, it is impossible to distinguish a TSV line that has a call with no data
+from a TSV line that has no cells.
+
+Here are some examples:
+
+>Input (literal strings)          Parses As (Haskell String syntax)
+>-------------------------------- ---------------------------------
+
+>1	2	3                 [["1", "2", "3"]]
+>
+>l1                               [["l1"], ["l2"]]
+>l2
+>
+> (empty line)                    [[""]]
+>
+>NQ	"Quoted"                  [["NQ", "Quoted"]]
+>
+>NQ	"Embedded""Quote"         [["NQ", "Embedded\"Quote"]]
+
+To parse a String, you might use:
+
+>import Text.ParserCombinators.Parsec
+>import Data.String.TSV
+>....
+>parse tsvFile "" mystring
+
+To parse a file, you might instead use:
+
+>do result <- parseFromFile tsvFile "/path/to/file"
+
+Please note that the result of parsing will be of type
+(Either ParseError [[String]]).  A Left result indicates an error.
+For more details, see the Parsec information.
+-}
+
+tsvFile :: CharParser st [[String]]
+tsvFile = endBy line eol
+
+{- | Generate TSV data for a file.  The resulting string can be
+written out to disk directly. -}
+genTsvFile :: [[String]] -> String
+genTsvFile inp = genXsvFile "\t" inp

testsrc/Str/TSVtest.hs

@@ -0,0 +1,33 @@
+{- arch-tag: CSV tests main file
+Copyright (C) 2005-2011 John Goerzen <[email protected]>
+
+All rights reserved.
+
+For license and copyright information, see the file LICENSE
+
+-}
+
+module Str.CSVtest(tests) where
+import Test.HUnit
+import Data.CSV
+import Text.ParserCombinators.Parsec
+
+test_tsv =
+    let f inp exp = TestLabel inp $ TestCase $ 
+                    exp @=? case parse tsvFile "" inp of
+                                  Right x -> Right x
+                                  Left y -> Left (show y)
+        in [
+        f "" (Right []),
+        f "\n" (Right [[""]]),
+        f "1	2	3\n" (Right [["1", "2", "3"]]),
+        f "This is a	Test	Really\n" (Right [["This is a", "Test", "Really"]]),
+        f "l1\nl2\n" (Right [["l1"], ["l2"]]),
+        f "NQ	\"Quoted\"\n" (Right [["NQ", "Quoted"]]),
+        f "1Q	\"\"\"\"\n" (Right [["1Q", "\""]]),
+        f "	\"\"\n" (Right [["", ""]]),
+        f "\"Embedded\"\"Quote\"\n" (Right [["Embedded\"Quote"]])
+        ]
+
+tests = TestList [TestLabel "tsv" (TestList test_csv)]
+

testsrc/Tests.hs

@@ -23,6 +23,7 @@ import qualified HVIOtest
 import qualified HVFStest
 import qualified Timetest
 import qualified Str.CSVtest
+import qualified Str.TSVtest
 import qualified WildMatchtest
 import qualified Globtest
 import qualified ProgressTrackertest
@@ -33,6 +34,7 @@ tests = TestList [TestLabel "test1" test1,
                  TestLabel "List" Listtest.tests,
                  TestLabel "Str" Strtest.tests,
                  TestLabel "CSV" Str.CSVtest.tests,
+                 TestLabel "TSV" Str.TSVtest.tests,
                  TestLabel "Time" Timetest.tests,
                  TestLabel "Map" Maptest.tests,
                  TestLabel "ProgressTracker" ProgressTrackertest.tests,