update

src/util.jl

@@ -770,29 +770,106 @@ function extract_triple_backtick_text(input::String)::Vector{String}
 end
 
 
+wordwindow(word::String, startindex::Integer)::UnitRange = startindex:startindex + length(word) -1
+
+function cuttext(range, text)
+  # check whether range is outside text boundary
+  if range.start > length(text) || range.stop > length(text)
+    return nothing
+  else
+    return text[range]
+  end
+end
+
 """
-    detect_keyword(keywords::AbstractVector{String}, text::String) -> Dict{String, Integer}
+    detect_keyword(keywords::AbstractVector{String}, text::String; mode::Union{String, Nothing}=nothing, delimiter::AbstractVector=[' ', '\n', '.']) -> Dict{String, Integer}
 
 Detects and counts occurrences of multiple keywords in the text in different case variations (lowercase, uppercase first letter, or all uppercase).
 
-# Arguments:
+# Arguments
 - `keywords::AbstractVector{String}` Vector of keywords to search for
 - `text::String` The text to search in
 
-# Returns:
+# Keyword Arguments
+- `mode::Union{String, Nothing}` When set to "individual", only counts matches that are individual words (default: nothing)
+- `delimiter::AbstractVector` Characters used to determine word boundaries when mode="individual" (default: [' ', '\n', '.'])
+
+# Returns
 - `Dict{String, Integer}` Returns a dictionary mapping each keyword to its count in the text (0 if not found)
 
-# Examples:
+# Examples
   ```jldoctest
-  julia> detect_keyword(["test", "error", "case"], "This is a Test case with ERRORS case")
+  julia> detect_keyword(["test", "example"], "This is a Test EXAMPLE")
-  Dict{String, Integer}("test" => 1, "error" => 1, "case" => 2)
+  Dict{String, Integer}("test" => 1, "example" => 1)
 
-  julia> detect_keyword(["warning", "missing"], "Warning: data is complete")
+  julia> detect_keyword(["cat"], "cats and category", mode="individual")
-  Dict{String, Integer}("warning" => 1, "missing" => 0)
+  Dict{String, Integer}("cat" => 0)
 
+  julia> detect_keyword(["error"], "No ERRORS found!")
+  Dict{String, Integer}("error" => 1)
+  ```
   
 # Signature
 """
+# function detect_keyword(keywords::T1, text::String; 
+#   mode::Union{String, Nothing}=nothing, delimiter::T2=[' ', '\n', '.']
+#   )::Dict{String, Integer} where {T1<:AbstractVector, T2<:AbstractVector}
+#   # Initialize dictionary to store keyword counts
+#   kwdict = Dict{String, Integer}()
+#   for i in keywords
+#     kwdict[i] = 0
+#   end
+  
+#   startindex = 1
+#   # Iterate through each keyword and search for matches in text
+#   for kw in keywords
+#     # Check each possible starting position in the text
+#     for startindex in 1:1:length(text)
+#       # Get the window range for current keyword at current position
+#       wordwindows = wordwindow(kw, startindex)
+#       # Extract the text slice for comparison
+#       cuttexts = cuttext(wordwindows, text)
+#       if cuttexts !== nothing
+#         # Try to detect keyword in current text slice
+#         detected_kw = detect_keyword(kw, cuttexts)
+#         if detected_kw !== nothing && mode === nothing
+#           # Increment count if keyword found and no mode restrictions
+#           kwdict[kw] +=1
+#         elseif detected_kw !== nothing && mode === "individual"
+#           # For individual word mode, check word boundaries
+#           # Check if character before keyword is a delimiter or start of text
+#           checkbefore = 
+#             if wordwindows.start > 1 && 
+#               text[wordwindows.start-1] ∈ delimiter
+#               true
+#             elseif wordwindows.start == 1
+#               true
+#             else
+#               false
+#             end
+
+#           # Check if character after keyword is a delimiter or end of text
+#           checkafter =
+#             if wordwindows.stop < length(text) && 
+#               text[wordwindows.stop+1] ∈ delimiter
+#               true
+#             elseif wordwindows.stop == length(text)
+#               true
+#             else
+#               false
+#             end
+#           # Only count keyword if it's a complete word
+#           if checkbefore && checkafter
+#             kwdict[kw] +=1
+#           end
+#         end
+#       end
+#     end
+#   end
+#   return kwdict
+# end
+
+
 function detect_keyword(keywords::T, text::String)::Dict{String, Integer} where {T<:AbstractVector}
   kw = Dict{String, Integer}()
   splittext = string.(split(text, " "))

test/runtests.jl

@@ -1,15 +1,35 @@
 using Test
-using GeneralUtils
+using GeneralUtils: detect_keyword
 
-@testset "ealierElementsIndex" begin
+@testset "detect_keyword tests" begin
-  @test GeneralUtils.ealierElementsIndex([1,2,3,4,5], 2) == 1:3
+    @test detect_keyword(["test"], "this is a test") == Dict("test" => 1)
-  @test GeneralUtils.ealierElementsIndex([1,2,3], 0) == 1:3
+    
-  @test GeneralUtils.ealierElementsIndex([1], 1) == 1:0
+    @test detect_keyword(["hello", "world"], "hello world hello") == Dict("hello" => 2, "world" => 1)
-  @test GeneralUtils.ealierElementsIndex([], 0) == 1:0
+    
-  @test GeneralUtils.ealierElementsIndex([1,2,3,4], 4) == 1:0
+    @test detect_keyword(["cat"], "category") == Dict("cat" => 1)
-  @test GeneralUtils.ealierElementsIndex([1,2,3,4], 5) == 1:0
+    
-  @test GeneralUtils.ealierElementsIndex(collect(1:10), 3) == 1:7
+    @test detect_keyword(["cat"], "category"; mode="individual") == Dict("cat" => 0)
-  @test_throws ErrorException GeneralUtils.ealierElementsIndex([1,2,3], -1)
+    
+    @test detect_keyword(["dog"], "dogs and cats"; mode="individual", delimiter=[' ']) == Dict("dog" => 0)
+    
+    @test detect_keyword(["test"], "test.case"; mode="individual", delimiter=['.']) == Dict("test" => 1)
+    
+    @test detect_keyword(["word"], "") == Dict("word" => 0)
+    
+    @test detect_keyword(String[], "some text") == Dict{String, Integer}()
+    
+    @test detect_keyword(["a", "b"], "a.b\nc"; delimiter=['.', '\n']) == Dict("a" => 1, "b" => 1)
+    
+    multiline_text = """
+    first line
+    second line
+    first word
+    """
+    @test detect_keyword(["first"], multiline_text) == Dict("first" => 2)
+    
+    @test detect_keyword(["word"], "word"; mode="individual") == Dict("word" => 1)
+    
+    @test detect_keyword(["test"], "testing.test.tester"; mode="individual", delimiter=['.']) == Dict("test" => 1)
 end