fixed themes and read function. version 0.3

Project.toml

@@ -1,13 +1,12 @@
 name = "HTMLTables"
 uuid = "b1afcece-b80e-4563-b90e-36b4cc56d3fa"
 authors = ["Ceco E. Maples <[email protected]>"]
-version = "0.2.0"
+version = "0.3.0"
 
 [deps]
 Cascadia = "54eefc05-d75b-58de-a785-1a3403f0919f"
 ColorSchemes = "35d6a980-a343-548e-a6ea-1d62b119f2f4"
 Colors = "5ae59095-9a9b-59fe-a467-6f913c188581"
-DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
 Gumbo = "708ec375-b3d6-5a57-a7ce-8257bf98657a"
 HTTP = "cd3eb016-35fb-5094-929b-558a96fad6f3"
 Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
@@ -16,7 +15,6 @@ Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
 Cascadia = "1"
 ColorSchemes = "3"
 Colors = "0.12"
-DataFrames = "1"
 Gumbo = "0.8"
 HTTP = "1"
 Tables = "1.11"

README.md

@@ -3,10 +3,10 @@
 Julia package for reading and writing HTML tables.
 
 ## Reading HTML tables
-- `HTMLTables.read`: read HTML table into a `DataFrame`.
+- `HTMLTables.read`: extracts data from HTML tables.
 
 ## Writing HTML tables
-- `HTMLTables.write`: write a `DataFrame` into an HTML table.
+- `HTMLTables.write`: uses the Tables.jl interface to write an HTML table.
 
 ## License
 Copyright © 2024 Ceco Elijah Maples

src/HTMLTables.jl

@@ -4,23 +4,17 @@
 Julia package for reading and writing HTML tables.
 
 Reading HTML tables:
-- `HTMLTables.read` reads HTML table element `<table></table>` into a `DataFrame`.
+- `HTMLTables.read` extracts data from HTML tables.
 
 Writing HTML tables:
-- `HTMLTables.write` writes a `DataFrame` into an HTML table element `<table></table>`.
+- `HTMLTables.write` uses the Tables.jl interface to write an HTML table.
 """
 module HTMLTables
 
-using Cascadia, Colors, ColorSchemes, DataFrames, Gumbo, HTTP, Tables
+using Cascadia, Colors, ColorSchemes, Gumbo, HTTP, Tables
 
 export read, write
 
-function isurl(source::String)::Bool
-    url_pattern::Regex = r"(?i)\b((?:https?|ftp):\/\/[\w-]+(\.[\w-]+)+([\w.,@?^=%&:/~+#-]*[\w@?^=%&/~+#-])?)\b"
-
-    return Base.occursin(url_pattern, source)
-end
-
 include("read.jl")
 include("write.jl")
 

src/read.jl

@@ -1,66 +1,95 @@
+function isurl(source::String)::Bool
+    url_pattern ::Regex = r"(?i)\b((?:https?|ftp):\/\/[\w-]+(\.[\w-]+)+([\w.,@?^=%&:/~+#-]*[\w@?^=%&/~+#-])?)\b"
+
+    return Base.occursin(url_pattern, source)
+end
+
+function extractrowdata(row::Gumbo.HTMLNode)::Vector
+    cell_matches::Vector{Gumbo.HTMLNode} = Base.eachmatch(
+        Cascadia.Selector("td,th"), row
+    )
+
+    return [Cascadia.nodeText(cell) for cell in cell_matches]
+end
+
 """
     HTMLTables.read(source::String; id::String="", classes::Union{Vector{String},String}="")
 
-Reads a HTML table into a `DataFrame`.
+Reads a HTML table into a sink function such as `DataFrame`.
 
 ## Arguments
 
 - `source::String`: URL or path to the HTML table.
+- `sink`: The function that materializes the table data.
 - `id::String`: The id of the HTML table.
 - `classes::Union{Vector{String},String}`: The classes of the HTML table.
 
+## Examples
+
+```julia
+using HTMLTables, DataFrames
+
+# read an HTML table into a DataFrame
+df = HTMLTables.read("https://www.w3schools.com/html/html_tables.asp", DataFrame)
+println(df)
+
+```
 """
 function read(
-    source::String;
-    id::String="",
-    classes::Union{Vector{String},String}=""
+    source::String,
+    sink; 
+    id::String="", 
+    classes::Union{Vector{String},String}="",
+    index::Int=1
 )
-    if isurl(source)
+    if index <= 0
+        throw(ArgumentError("Index must be a positive integer"))
+    end
+
+    if isurl(source) == true
         response::HTTP.Response = HTTP.get(source)
         html_content = Base.String(response.body)
     else
         html_content = Base.read(source, String)
     end
 
-    html::Gumbo.HTMLDocument = Gumbo.parsehtml(html_content)
+    html_document::Gumbo.HTMLDocument = Gumbo.parsehtml(html_content)
 
     selector::String = ""
-
-    if id == ""
-        if classes !== ""
-            if Base.isa(classes, String)
-                selector *= ".$classes"
-            elseif Base.isa(classes, Vector{String})
-                selector *= Base.join([".$cls" for cls in classes], ",")
-            else
-                Base.throw(Base.ArgumentError("classes must be a String or Vector{String}"))
-            end
+    if Base.isempty(id)
+        if Base.isempty(classes)
+            selector *= "table"
+        elseif !Base.isempty(classes) && Base.isa(classes, String)
+            selector *= "table.$classes"
+        elseif !Base.isempty(classes) && Base.isa(classes, Vector{String})
+            selector *= "table." * Base.join(classes, ".")
         end
-    elseif id !== ""
+    elseif !Base.isempty(id)
         selector *= "#$id"
-    else
-        Base.throw(Base.ArgumentError("id must be of type String"))
     end
 
-    tables = Base.eachmatch(Cascadia.Selector(selector), html.root)
-
-    if !Base.isempty(tables)
-        table = tables[1]
-        rows = Base.eachmatch(Cascadia.Selector("tr"), table)
-        headers = []
-        data = []
-
-        for (i, row) in Base.Iterators.enumerate(rows)
-            cells = Base.eachmatch(Cascadia.Selector("td,th"), row)
-            if i == 1 && Base.isempty(headers)
-                headers = [Cascadia.nodeText(cell) for cell in cells]
-            else
-                Base.push!(data, [Cascadia.nodeText(cell) for cell in cells])
-            end
+    tables::Vector{Gumbo.HTMLNode} = Base.eachmatch(Cascadia.Selector(selector), html_document.root)
+
+    if Base.isempty(tables) == true
+        throw(ArgumentError("No HTML tables found"))
+    end
+
+    table::Gumbo.HTMLNode = tables[index]
+
+    rows::Vector{Gumbo.HTMLNode} = Base.eachmatch(Cascadia.Selector("tr"), table)
+    headers::Vector = []
+    data::Vector{Vector} = []
+
+    for (i, row) in Base.enumerate(rows)
+        rowdata = extractrowdata(row)
+        if (i == 1 && Base.isempty(headers)) == true
+            headers = rowdata
+        else
+            Base.push!(data, rowdata)
         end
-
-        return DataFrames.DataFrame(data, Symbol.(headers))
-    else
-        return DataFrames.DataFrame()
     end
+
+    tuples::Vector = [Base.Tuple(row) for row in data]
+
+    return sink(tuples, headers)
 end

src/themes/00-default.css

@@ -6,6 +6,7 @@ table {
   border-collapse: collapse;
   border-top: 1px solid black;
   border-bottom: 1px solid black;
+  background-color: white;
 }
 
 thead {

src/themes/01-red.css

@@ -4,21 +4,20 @@ table {
   color: black;
   width: 100%;
   border-collapse: collapse;
+  border-top: 1px solid black;
+  border-bottom: 1px solid black;
 }
 
 thead {
   font-weight: bold;
   background-color: hsl(0, 100%, 60%);
+  border-bottom: 1px solid black;
 }
 
 tfoot {
   height: 20px;
   background-color: hsl(0, 100%, 65%);
-}
-
-thead, tfoot {
   border-top: 1px solid black;
-  border-bottom: 1px solid black;
 }
 
 tbody tr:nth-child(even) {

src/themes/02-orange.css

@@ -4,21 +4,20 @@ table {
   color: black;
   width: 100%;
   border-collapse: collapse;
+  border-top: 1px solid black;
+  border-bottom: 1px solid black;
 }
 
 thead {
   font-weight: bold;
   background-color: hsl(25, 100%, 60%);
+  border-bottom: 1px solid black;
 }
 
 tfoot {
   height: 20px;
   background-color: hsl(25, 100%, 65%);
-}
-
-thead, tfoot {
   border-top: 1px solid black;
-  border-bottom: 1px solid black;
 }
 
 tbody tr:nth-child(even) {

src/themes/03-yellow.css

@@ -4,21 +4,20 @@ table {
   color: black;
   width: 100%;
   border-collapse: collapse;
+  border-top: 1px solid black;
+  border-bottom: 1px solid black;
 }
 
 thead {
   font-weight: bold;
   background-color: hsl(60, 100%, 60%);
+  border-bottom: 1px solid black;
 }
 
 tfoot {
   height: 20px;
   background-color: hsl(60, 100%, 65%);
-}
-
-thead, tfoot {
   border-top: 1px solid black;
-  border-bottom: 1px solid black;
 }
 
 tbody tr:nth-child(even) {

src/themes/04-green.css

@@ -4,21 +4,20 @@ table {
   color: black;
   width: 100%;
   border-collapse: collapse;
+  border-top: 1px solid black;
+  border-bottom: 1px solid black;
 }
 
 thead {
   font-weight: bold;
   background-color: hsl(115, 100%, 60%);
+  border-bottom: 1px solid black;
 }
 
 tfoot {
   height: 20px;
   background-color: hsl(115, 100%, 65%);
-}
-
-thead, tfoot {
   border-top: 1px solid black;
-  border-bottom: 1px solid black;
 }
 
 tbody tr:nth-child(even) {

src/themes/05-blue.css

@@ -4,21 +4,20 @@ table {
   color: black;
   width: 100%;
   border-collapse: collapse;
+  border-top: 1px solid black;
+  border-bottom: 1px solid black;
 }
 
 thead {
   font-weight: bold;
   background-color: hsl(205, 100%, 60%);
+  border-bottom: 1px solid black;
 }
 
 tfoot {
   height: 20px;
   background-color: hsl(205, 100%, 65%);
-}
-
-thead, tfoot {
   border-top: 1px solid black;
-  border-bottom: 1px solid black;
 }
 
 tbody tr:nth-child(even) {

src/themes/06-violet.css

@@ -4,21 +4,20 @@ table {
   color: black;
   width: 100%;
   border-collapse: collapse;
+  border-top: 1px solid black;
+  border-bottom: 1px solid black;
 }
 
 thead {
   font-weight: bold;
   background-color: hsl(260, 100%, 60%);
+  border-bottom: 1px solid black;
 }
 
 tfoot {
   height: 20px;
   background-color: hsl(260, 100%, 65%);
-}
-
-thead, tfoot {
   border-top: 1px solid black;
-  border-bottom: 1px solid black;
 }
 
 tbody tr:nth-child(even) {

src/themes/07-magenta.css

@@ -4,21 +4,20 @@ table {
   color: black;
   width: 100%;
   border-collapse: collapse;
+  border-top: 1px solid black;
+  border-bottom: 1px solid black;
 }
 
 thead {
   font-weight: bold;
   background-color: hsl(320, 100%, 60%);
+  border-bottom: 1px solid black;
 }
 
 tfoot {
   height: 20px;
   background-color: hsl(320, 100%, 65%);
-}
-
-thead, tfoot {
   border-top: 1px solid black;
-  border-bottom: 1px solid black;
 }
 
 tbody tr:nth-child(even) {