Merge pull request #25 from tk3369/tk/caseinsensitivecol

Case-insensitive match for column symbols in include/exclude_columns argument

Author: tk3369

README.md

@@ -126,6 +126,8 @@ julia> head(df,5)
 
 ### Inclusion/Exclusion of Columns
 
+**Column Inclusion**
+
 It is always faster to read only the columns that you need.  The `include_columns` argument comes in handy:
 
 ```
@@ -142,6 +144,8 @@ Read data set of size 1440 x 4 in 0.004 seconds
 │ 6   │ 948.0  │ 1993-06-01 │ SOFA    │ 1993.0 │
 ```
 
+**Column Exclusion**
+
 Likewise, you can read all columns except the ones you don't want as specified in `exclude_columns` argument:
 
 ```
@@ -158,6 +162,23 @@ Read data set of size 1440 x 6 in 0.031 seconds
 │ 6   │ CANADA  │ EDUCATION │ 486.0   │ FURNITURE │ 2.0     │ EAST   │
 ```
 
+**Case Sensitivity and Column Number**
+
+Column symbols are matched in a case insensitive manner with SAS column names.  
+
+Both `include_columns` and `exclude_columns` accept column number.  In fact, you can mixed column symbols and column numbers as such:
+
+```
+julia> readsas("productsales.sas7bdat", include_columns=[:actual, :predict, 8, 9, 10])[:column_names]
+Read productsales.sas7bdat with size 1440 x 5 in 0.00511 seconds
+5-element Array{String,1}:
+ "ACTUAL" 
+ "PREDICT"
+ "QUARTER"
+ "YEAR"   
+ "MONTH"  
+```
+
 ### Incremental Reading
 
 If you need to read files incrementally, you can do so as such:

src/SASLib.jl

@@ -1691,26 +1691,49 @@ logdebug = println
 #     return d
 # end
 
+# case insensitive column mapping
+Base.lowercase(s::Symbol) = Symbol(lowercase(String(s)))
+case_insensitive_in(s::Symbol, ar::AbstractArray) = 
+    lowercase(s) in [x isa Symbol ? lowercase(x) : x for x in ar]
+
 # fill column indices as a dictionary (key = column index, value = column symbol)
 function _fill_column_indices(handler)
     handler.column_indices = Vector{Tuple{Int64, Symbol, UInt8}}()
     inflag = length(handler.config.include_columns) > 0
     exflag = length(handler.config.exclude_columns) > 0
     inflag && exflag && throw(ConfigError("You can specify either include_columns or exclude_columns but not both."))
+    processed = []
     for j in 1:length(handler.column_symbols)
         name = handler.column_symbols[j]
         if inflag 
-            if j in handler.config.include_columns || name in handler.config.include_columns
+            if j in handler.config.include_columns || 
+                    case_insensitive_in(name, handler.config.include_columns)
                 push!(handler.column_indices, (j, name, handler.column_types[j]))
+                push!(processed, lowercase(name))
             end
         elseif exflag 
-            if !(j in handler.config.exclude_columns || name in handler.config.exclude_columns)
+            if !(j in handler.config.exclude_columns || 
+                    case_insensitive_in(name, handler.config.exclude_columns))
                 push!(handler.column_indices, (j, name, handler.column_types[j]))
+            else
+                push!(processed, lowercase(name))
             end
         else
             push!(handler.column_indices, (j, name, handler.column_types[j]))
         end
     end
+    if inflag && length(processed) != length(handler.config.include_columns) 
+        diff = setdiff(handler.config.include_columns, processed)
+        for c in diff
+            warn("Unknown include column $c")
+        end
+    end
+    if exflag && length(processed) != length(handler.config.exclude_columns) 
+        diff = setdiff(handler.config.exclude_columns, processed)
+        for c in diff
+            warn("Unknown exclude column $c")
+        end
+    end
     # println2(handler, "column_indices = $(handler.column_indices)")
 end
 

test/runtests.jl

@@ -113,6 +113,16 @@ openfile(dir, file; kwargs...)  = SASLib.open(getpath(dir, file), kwargs...)
         @test result[:ncols] == 1
         @test sort(result[:column_symbols]) == sort([:ACTUAL])
 
+        # case insensitive include/exclude
+        result = readsas(fname, include_columns=[:month, :Year])
+        @test result[:ncols] == 2
+        result = readsas(fname, exclude_columns=[:diVisiON])
+        @test result[:ncols] == 9
+
+        # bad include/exclude param
+        @test_warn "Unknown include column" readsas(fname, include_columns=[:blah, :Year])
+        @test_warn "Unknown exclude column" readsas(fname, exclude_columns=[:blah, :Year])
+
         # error handling
         @test_throws SASLib.ConfigError readsas(fname, 
             include_columns=[1], exclude_columns=[1])