fixed styles, new functions to view tables, more modular functions

Project.toml

@@ -1,7 +1,7 @@
 name = "HTMLTables"
 uuid = "b1afcece-b80e-4563-b90e-36b4cc56d3fa"
 authors = ["Ceco E. Maples <[email protected]>"]
-version = "0.3.0"
+version = "0.3.1"
 
 [deps]
 Cascadia = "54eefc05-d75b-58de-a785-1a3403f0919f"

README.md

@@ -1,12 +1,18 @@
 # HTMLTables
 
-Julia package for reading and writing HTML tables.
+Julia package for reading, writing, and viewing HTML tables.
 
-## Reading HTML tables
-- `HTMLTables.read`: extracts data from HTML tables.
+Reading HTML tables:
+- `HTMLTables.get` reads an HTML table as a string.
+- `HTMLTables.read` extracts data from HTML tables.
 
-## Writing HTML tables
-- `HTMLTables.write`: uses the Tables.jl interface to write an HTML table.
+Writing HTML tables:
+- `HTMLTables.table` uses the Tables.jl interface to write an HTML table as a string.
+- `HTMLTables.write` uses the Tables.jl interface to write an HTML table in a file.
+
+Viewing HTML tables:
+- `HTMLTables.display` displays a julia table as an HTML table in julia.
+- `HTMLTables.open` opens a julia table as an HTML table in the browser.
 
 ## License
 Copyright © 2024 Ceco Elijah Maples

src/HTMLTables.jl

@@ -1,21 +1,33 @@
 """
     HTMLTables
 
-Julia package for reading and writing HTML tables.
+Julia package for reading, writing, and viewing HTML tables.
 
 Reading HTML tables:
+- `HTMLTables.get` reads an HTML table as a string.
 - `HTMLTables.read` extracts data from HTML tables.
 
 Writing HTML tables:
-- `HTMLTables.write` uses the Tables.jl interface to write an HTML table.
+- `HTMLTables.table` uses the Tables.jl interface to write an HTML table as a string.
+- `HTMLTables.write` uses the Tables.jl interface to write an HTML table in a file.
+
+Viewing HTML tables:
+- `HTMLTables.display` displays a julia table as an HTML table in julia.
+- `HTMLTables.open` opens a julia table as an HTML table in the browser.
 """
 module HTMLTables
 
 using Cascadia, Colors, ColorSchemes, Gumbo, HTTP, Tables
 
-export read, write
+export get, read, table, write, display, open
 
+include("get.jl")
 include("read.jl")
+
+include("table.jl")
 include("write.jl")
 
+include("display.jl")
+include("open.jl")
+
 end

src/display.jl

@@ -0,0 +1,10 @@
+"""
+    HTMLTables.display(tbl; kwargs...)
+
+Displays a julia table as an HTML table in julia.
+"""
+function display(tbl; kwargs...)
+    html_table::String = table(tbl; kwargs...)
+
+    Base.display("image/svg+xml", html_table)
+end

src/get.jl

@@ -0,0 +1,65 @@
+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
+
+"""
+    HTMLTables.get(source::String; id::String="", classes::Union{Vector{String},String}="", index::Int=1)
+
+Returns an HTML table a source as a string.
+
+## Arguments
+
+- `source::String`: URL or path to the HTML table.
+- `id::String`: The id of the HTML table.
+- `classes::Union{Vector{String},String}`: The classes of the HTML table.
+- `index::Int`: The index of the HTML table in the HTML document.
+
+"""
+function get(
+    source::String;
+    id::String="",
+    classes::Union{Vector{String},String}="",
+    index::Int=1
+)
+    if index <= 0
+        Base.throw(Base.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_document::Gumbo.HTMLDocument = Gumbo.parsehtml(html_content)
+
+    selector::String = ""
+    if Base.isempty(id)
+        if Base.isempty(classes)
+            selector *= "table"
+        elseif !Base.isempty(classes) && Base.isa(classes, String)
+            selector *= "table.$classes"
+        elseif !Base.isempty(classes) && Base.isa(classes, Vector{String})
+            selector *= "table." * Base.join(classes, ".")
+        end
+    elseif !Base.isempty(id)
+        selector *= "#$id"
+    end
+
+    tables::Vector{Gumbo.HTMLNode} = Base.eachmatch(Cascadia.Selector(selector), html_document.root)
+
+    number_of_tables::Int = Base.length(tables)
+
+    if number_of_tables == 0
+        Base.throw(Base.ArgumentError("No HTML tables found"))
+    elseif index > number_of_tables
+        Base.throw(Base.ArgumentError("$number_of_tables table(s) found. Index $index does not exist."))
+    end
+
+    table::Gumbo.HTMLNode = tables[index]
+
+    return table
+end

src/open.jl

@@ -0,0 +1,16 @@
+"""
+    HTMLTables.open(tbl; kwargs...)
+
+Opens a julia table as an HTML table in the browser.
+"""
+function open(tbl; kwargs...)
+    path::String = write(tbl; kwargs...)
+
+    if Base.Sys.iswindows()
+        Base.run(`start $path`)
+    elseif Base.Sys.islinux()
+        Base.run(`xdg-open $path`)
+    elseif Base.Sys.isapple()
+        Base.run(`open $path`)
+    end
+end

src/read.jl

@@ -1,15 +1,7 @@
-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
-    )
+    cells::Vector{Gumbo.HTMLNode} = Base.eachmatch(Cascadia.Selector("td,th"), row)
 
-    return [Cascadia.nodeText(cell) for cell in cell_matches]
+    return [Cascadia.nodeText(cell) for cell in cells]
 end
 
 """
@@ -23,6 +15,7 @@ Reads a HTML table into a sink function such as `DataFrame`.
 - `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.
+- `index::Int`: The index of the HTML table in the HTML document.
 
 ## Examples
 
@@ -31,8 +24,6 @@ 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(
@@ -42,39 +33,7 @@ function read(
     classes::Union{Vector{String},String}="",
     index::Int=1
 )
-    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_document::Gumbo.HTMLDocument = Gumbo.parsehtml(html_content)
-
-    selector::String = ""
-    if Base.isempty(id)
-        if Base.isempty(classes)
-            selector *= "table"
-        elseif !Base.isempty(classes) && Base.isa(classes, String)
-            selector *= "table.$classes"
-        elseif !Base.isempty(classes) && Base.isa(classes, Vector{String})
-            selector *= "table." * Base.join(classes, ".")
-        end
-    elseif !Base.isempty(id)
-        selector *= "#$id"
-    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]
+    table = get(source, id=id, classes=classes, index=index)
 
     rows::Vector{Gumbo.HTMLNode} = Base.eachmatch(Cascadia.Selector("tr"), table)
     headers::Vector = []