ENH Add read_int and read_double functions

These are very simple. The choice of API was to allow for a default on
empty, but to error on anything else that cannot be parsed

This patch also adds the ``__assert`` function for internal use and
generalizes some of the transformation mechanics

closes #78

Author: luispedro

ChangeLog

@@ -3,6 +3,7 @@ Version 1.2.0+
 	* Use ZStd compression for temporary files from preprocess()
 	* Correctly handle subpaths in samples for collect (fixes #141)
 	* Add to_string() to int and double types (partially fixes #78 & fixes #81)
+	* Add read_int() and read_double() functions (fixes #78)
 
 Version 1.2.0 2020-07-12 by luispedro
 	* Add load_fastq_directory to builtin functions

NGLess/BuiltinFunctions.hs

@@ -43,8 +43,14 @@ builtinFunctions =
     ,Function (FuncName "__check_count") (Just NGLMappedReadSet) [] NGLCounts countCheckArgs False []
     ,Function (FuncName "countfile") (Just NGLString) [ArgCheckFileReadable] NGLCounts [] False [FunctionCheckReturnAssigned]
     ,Function (FuncName "write") (Just NGLAny) [] NGLVoid writeArgs False []
+
     ,Function (FuncName "print") (Just NGLAny) [] NGLVoid [] False []
 
+    ,Function (FuncName "read_int") (Just NGLString) [] NGLInteger [ArgInformation "on_empty_return" False NGLInteger []] False []
+    ,Function (FuncName "read_double") (Just NGLString) [] NGLDouble [ArgInformation "on_empty_return" False NGLDouble []] False []
+
+    ,Function (FuncName "__assert") (Just NGLBool) [] NGLVoid [] False []
+
     ,Function (FuncName "__merge_samfiles") (Just (NGList NGLString)) [] NGLMappedReadSet [] False []
     ]
 

NGLess/Interpret.hs

@@ -80,6 +80,7 @@ import           Control.DeepSeq (NFData(..))
 import qualified Data.Vector as V
 import           Safe (atMay)
 import           Control.Error (note)
+import           Text.Read (readEither)
 
 import System.IO
 import System.Directory
@@ -335,8 +336,11 @@ interpretFunction' (FuncName "count")     expr args Nothing = runNGLessIO (execu
 interpretFunction' (FuncName "__check_count") expr args Nothing = runNGLessIO (executeCountCheck expr args)
 interpretFunction' (FuncName "countfile") expr args Nothing = runNGLessIO (executeCountFile expr args)
 interpretFunction' (FuncName "print")     expr args Nothing = executePrint expr args
+interpretFunction' (FuncName "read_int")  expr args Nothing = executeReadInt expr args
+interpretFunction' (FuncName "read_double")  expr args Nothing = executeReadDouble expr args
 interpretFunction' (FuncName "paired")   mate1 args Nothing = runNGLessIO (executePaired mate1 args)
 interpretFunction' (FuncName "select")    expr args (Just b) = executeSelectWBlock expr args b
+interpretFunction' (FuncName "__assert")  expr [] args       = executeAssert expr args
 interpretFunction' fname@(FuncName fname') expr args Nothing = do
     traceExpr ("executing module function: '"++T.unpack fname'++"'") expr
     execF <- findFunction fname
@@ -536,6 +540,24 @@ executePrint :: NGLessObject -> [(T.Text, NGLessObject)] -> InterpretationEnvIO
 executePrint (NGOString s) [] = liftIO (T.putStr s) >> return NGOVoid
 executePrint err  _ = throwScriptError ("Cannot print " ++ show err)
 
+executeReadInt :: NGLessObject -> [(T.Text, NGLessObject)] -> InterpretationEnvIO NGLessObject
+executeReadInt (NGOString "") kwargs = NGOInteger <$> lookupIntegerOrScriptError "read_int" "on_empty_return" kwargs
+executeReadInt (NGOString s) _ = case readEither (T.unpack s) of
+    Right val -> return $! NGOInteger val
+    Left err -> throwDataError ("Could not parse integer from '"++T.unpack s++"'. Error: "++err)
+executeReadInt s _ = throwScriptError ("Cannot parse this object as integer: "++ show s)
+
+executeReadDouble :: NGLessObject -> [(T.Text, NGLessObject)] -> InterpretationEnvIO NGLessObject
+executeReadDouble (NGOString "") kwargs = NGODouble <$> lookupDoubleOrScriptError "read_int" "on_empty_return" kwargs
+executeReadDouble (NGOString s) _ = case readEither (T.unpack s) of
+    Right val -> return $! NGODouble val
+    Left err -> throwDataError ("Could not parse double from '"++T.unpack s++"'. Error: "++err)
+executeReadDouble s _ = throwScriptError ("Cannot parse this object as double: "++ show s)
+
+executeAssert (NGOBool True) _ = return $! NGOVoid
+executeAssert (NGOBool False) _ = throwScriptError "Assert failed"
+executeAssert _ _ = throwShouldNotOccur "Assert did not receive a boolean!"
+
 executeSelectWBlock :: NGLessObject -> [(T.Text, NGLessObject)] -> Block -> InterpretationEnvIO NGLessObject
 executeSelectWBlock input@NGOMappedReadSet{ nglSamFile = isam} args (Block (Variable var) body) = do
         paired <- lookupBoolOrScriptErrorDef (return True) "select" "paired" args
@@ -727,6 +749,8 @@ evalBinary BOpPathAppend a b = case (a,b) of
     (NGOString pa, NGOString pb) -> return . NGOString $! T.pack (T.unpack pa </> T.unpack pb)
     _ -> nglTypeError ("Operator </>: invalid arguments" :: String)
 
+evalBinary BOpEQ (NGOString a) (NGOString b) = return . NGOBool $! a == b
+evalBinary BOpNEQ (NGOString a) (NGOString b) = return . NGOBool $! a /= b
 evalBinary op a b = do
         a' <- asDouble a
         b' <- asDouble b

NGLess/NGLess.hs

@@ -1,3 +1,6 @@
+{- Copyright 2013-2020 NGLess Authors
+ - License: MIT
+ -}
 {-# LANGUAGE OverloadedStrings, FlexibleContexts #-}
 module NGLess
     ( NGLessIO
@@ -23,6 +26,8 @@ module NGLess
     , lookupStringListOrScriptErrorDef
     , lookupIntegerOrScriptError
     , lookupIntegerOrScriptErrorDef
+    , lookupDoubleOrScriptError
+    , lookupDoubleOrScriptErrorDef
     , lookupSymbolOrScriptError
     , lookupSymbolOrScriptErrorDef
     , lookupSymbolListOrScriptError
@@ -93,6 +98,13 @@ lookupIntegerOrScriptErrorDef defval context name args = case lookup name args o
     Just (NGOInteger v) -> return v
     Just other -> throwScriptError ("Expected an integer in argument " ++ T.unpack name ++ " in context '" ++ context ++ "' instead observed: " ++ show other)
 
+lookupDoubleOrScriptError :: (MonadError NGError m) => String-> T.Text -> KwArgsValues -> m Double
+lookupDoubleOrScriptError = requiredLookup lookupDoubleOrScriptErrorDef
+lookupDoubleOrScriptErrorDef defval context name args = case lookup name args of
+    Nothing -> defval
+    Just (NGODouble v) -> return v
+    Just other -> throwScriptError ("Expected a double in argument " ++ T.unpack name ++ " in context '" ++ context ++ "' instead observed: " ++ show other)
+
 lookupSymbolOrScriptError :: (MonadError NGError m) => String-> T.Text -> KwArgsValues -> m T.Text
 lookupSymbolOrScriptError = requiredLookup lookupSymbolOrScriptErrorDef
 lookupSymbolOrScriptErrorDef defval context name args = case lookup name args of

NGLess/Transform.hs

@@ -90,7 +90,8 @@ checkSimple expr = forM_ expr (checkSimple1 . snd) *> return expr
             Condition{} -> throwShouldNotOccur "Non-simple expression (Condition)"
             Assignment{} -> throwShouldNotOccur "Non-simple expression (Assignment)"
             FunctionCall{} -> throwShouldNotOccur "Non-simple expression (FunctionCall)"
-            MethodCall{} -> throwShouldNotOccur "Non-simple expression (MethodCall)"
+            -- Rewriting for MethodCall is not implemented!
+            MethodCall{} -> return () -- throwShouldNotOccur "Non-simple expression (MethodCall)"
 
             ListExpression s -> mapM_ checkSimple0 s
             UnaryOp _ a -> checkSimple0 a
@@ -522,6 +523,12 @@ addTemporaries = addTemporaries' 0
                             put (n + 1)
                             tell [Assignment v e]
                             return (Lookup t v)
+                functionCallTemp e@BinaryOp{} = do
+                            n <- get
+                            let v = Variable (T.pack $ "temp$"++show n)
+                            put (n + 1)
+                            tell [Assignment v e]
+                            return (Lookup Nothing v)
                 functionCallTemp e = return e
 
 {-| Calculation of hashes for output method calls

NGLess/Types.hs

@@ -204,10 +204,17 @@ checkbop BOpGT  a b = checknum a *> checknum b *> return (Just NGLBool)
 checkbop BOpGTE a b = checknum a *> checknum b *> return (Just NGLBool)
 checkbop BOpLT  a b = checknum a *> checknum b *> return (Just NGLBool)
 checkbop BOpLTE a b = checknum a *> checknum b *> return (Just NGLBool)
-checkbop BOpEQ  a b = checknum a *> checknum b *> return (Just NGLBool)
-checkbop BOpNEQ a b = checknum a *> checknum b *> return (Just NGLBool)
-
 checkbop BOpPathAppend a b = softCheck NGLString a *> softCheck NGLString b *> return (Just NGLString)
+checkbop BOpNEQ  a b = checkbop BOpEQ a b
+checkbop BOpEQ  a b = do
+    t <- liftM3 (\a b c -> a <|> b <|> c)
+        (softCheckPair NGLInteger a b)
+        (softCheckPair NGLDouble a b)
+        (softCheckPair NGLString a b)
+    when (isNothing t) $
+        errorInLineC ["Comparison operators (== or !=) must be applied to a pair of strings or numbers"]
+    return (Just NGLBool)
+
 
 softCheck :: NGLType -> Expression -> TypeMSt (Maybe NGLType)
 softCheck expected expr = do

docs/sources/whatsnew.rst

@@ -6,10 +6,16 @@ What's New (History)
 Unreleased development version
 ------------------------------
 
+User-visible improvements
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- Adds conversion from string to numbers (int or double) and back
+
 Bugfixes
 ~~~~~~~~
 - Fix cases where sample names contain ``/`` and ``collect()`` (`issue 141
   <https://github.com/ngless-toolkit/ngless/issues/141>`__)
+ 
 
 
 Version 1.2.0

tests/type-conversions/types.ngl

@@ -0,0 +1,16 @@
+ngless "1.2"
+
+val = 1
+__assert(val.to_string() == "1")
+
+vald = 1.5
+__assert(vald.to_string() == "1.5")
+
+s = "2"
+__assert(read_int(s) == 2)
+__assert(read_int("", on_empty_return=3) == 3)
+
+s = "2.2"
+__assert(read_double(s) == 2.2)
+__assert(read_double("", on_empty_return=3.8) == 3.8)
+