WIP: Supporting XML v0.4

.github/workflows/ci.yml

@@ -1,6 +1,6 @@
 name: CI
 on:
-#  workflow_dispatch:  # <-- Add this to allow manual execution
+ workflow_dispatch:  # <-- Add this to allow manual execution
   pull_request:
     branches:
       - master

.gitignore

@@ -5,3 +5,4 @@ Manifest.toml
 docs/Manifest.toml
 .vscode
 *.gz
+.claude

Project.toml

@@ -8,7 +8,6 @@ repo = "https://github.com/juliadata/XLSX.jl.git"
 [deps]
 Colors = "5ae59095-9a9b-59fe-a467-6f913c188581"
 Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
-OrderedCollections = "bac558e1-5e72-5ebc-8fee-abe8a469f55d"
 PrecompileTools = "aea7be01-6a6a-4083-8856-8a6e6704d82a"
 Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
@@ -28,12 +27,11 @@ StyledStringsSstsExt = "StyledStrings"
 CSV = "0.10.15"
 Colors = "0.12, 0.13"
 Distributions = "0.25.0"
-OrderedCollections = "1"
 PrecompileTools = "1"
 StyledStrings = "1.0.3"
 Tables = "1"
 UUIDs = "1.8"
-XML = "0.3.8"
+XML = "0.4"
 ZipArchives = "2.5"
 julia = "1.8"
 [extras]

src/XLSX.jl

@@ -11,7 +11,6 @@ import Tables
 import Unicode
 import UUIDs
 import XML
-using OrderedCollections: OrderedDict
 import ZipArchives
 
 import PrecompileTools as PCT    # this is a small dependency.
@@ -40,6 +39,9 @@ export
     getMergedCells, isMergedCell, getMergedBaseCell, mergeCells
 
 const SPREADSHEET_NAMESPACE_XPATH_ARG = "http://schemas.openxmlformats.org/spreadsheetml/2006/main"
+
+xml_elements(node) = filter(n -> XML.nodetype(n) == XML.Element, XML.children(node))
+xml_root_element(doc) = last(xml_elements(doc))
 const EXCEL_MAX_COLS = 16_384     # total columns supported by Excel per sheet
 const EXCEL_MAX_ROWS = 1_048_576  # total rows supported by Excel per sheet (including headers)
 const ROW_CHUNKSIZE = 1000        # number of rows to be processed in each thread

src/cell.jl

@@ -188,7 +188,7 @@ end
 =#
 
 # Parses a style string to (UInt32, Int) for use as style and num_style.
-function _parse_style(s::String)
+function _parse_style(s::AbstractString)
     isempty(s) && return UInt32(0), 0
     n = parse(Int, s)
     return UInt32(n), n
@@ -235,8 +235,8 @@ function Cell(c::XML.LazyNode, ws::Worksheet, sst_pfx::String; mylock::Union{Ree
             if tag == "v"
                 ch = XML.children(child)
                 isempty(ch) && continue
-                raw = XML.value(ch[1])
-                v = occursin('&', raw) ? XLSX.unescape(raw) : raw
+                # `child` is a `LazyNode`; XML.jl 0.4 already unescapes its value.
+                v = XML.value(ch[1])
                 datatype, value = process_tv(wb, t, v, num_style; mylock)
             elseif tag == "f"
                 f = parse_formula_from_element(wb,child)
@@ -253,12 +253,15 @@ function parse_formula_from_element(wb, c_child_element)::AbstractFormula
     localname(c_child_element) == "f" ||
         throw(XLSXError("Expected nodename `f`. Found: `$(localname(c_child_element))`"))
 
-    # Extract formula string
-    formula_string = if XML.is_simple(c_child_element)
-        XLSX.unescape(XML.simple_value(c_child_element))
-    else
-        text_nodes = filter(x -> XML.nodetype(x) == XML.Text, XML.children(c_child_element))
-        isempty(text_nodes) ? "" : XLSX.unescape(XML.value(text_nodes[1]))
+    # Extract formula string — `<f>` may have attributes (t="shared", si=…, ref=…)
+    # which makes it non-"simple", so collect the first Text child.
+    formula_string = ""
+    for ch in XML.children(c_child_element)
+        if XML.nodetype(ch) === XML.Text
+            v = XML.value(ch)
+            isnothing(v) || (formula_string = v)
+            break
+        end
     end
 
     a = XML.attributes(c_child_element)
@@ -312,7 +315,7 @@ function _parse_excel_datetime_raw(v::AbstractString)
     end
 end
 
-function process_tv(wb::Workbook, t::String, v::String, num_style::Int; mylock::Union{ReentrantLock,Nothing}=nothing)
+function process_tv(wb::Workbook, t::AbstractString, v::AbstractString, num_style::Int; mylock::Union{ReentrantLock,Nothing}=nothing)
     datatype::CellValueType = CT_EMPTY
     value::UInt64           = UInt64(0)
     isempty(v) && return datatype, value
@@ -479,29 +482,21 @@ end
 # Extract cells from a <row> LazyNode and push them (in place) into a Dict(column -> Cell)
 function get_rowcells!(rowcells::Dict{Int,Cell}, row::XML.LazyNode, ws::Worksheet, sst_pfx::String; mylock::Union{ReentrantLock,Nothing}=nothing)
 
-    # unthreaded cell extraction is (exceedingly marginally) slower but no lock conflicts introduced.
-
     # debug
-    # @assert row.tag == "row" "Not a row node"
+    # @assert localname(row) == "row" "Not a row node"
 
     sst_count = 0
 
-    d = row.depth
-
-    cellnode = XML.next(row)
-
-    while !isnothing(cellnode) && cellnode.depth > d
-        if localname(cellnode) == "c" # This is a cell
-            cell = Cell(cellnode, ws, sst_pfx; mylock) # construct an XLSX.Cell from an XML.LazyNode
+    for child in XML.children(row)
+        if localname(child) == "c" # This is a cell
+            cell = Cell(child, ws, sst_pfx; mylock) # construct an XLSX.Cell from an XML.LazyNode
             sst_count += cell.datatype == CT_STRING ? 1 : 0
             rowcells[column_number(cell)] = cell
         end
-        cellnode = XML.next(cellnode)
-    end
-    if !isnothing(cellnode) && localname(cellnode) == "row" # have reached the beginning of next row
-        return cellnode, sst_count
-    else                                             # no more rows
-        return nothing, sst_count
     end
 
+    # Row nodes are materialised up-front (see `_collect_row_nodes`), so callers
+    # advance by index; the first tuple element is unused but kept for the
+    # `next, sst_count = get_rowcells!(...)` call contract.
+    return nothing, sst_count
 end

src/cellformat-helpers.jl

@@ -113,8 +113,9 @@ const EXCEL_COLUMN_WIDTH_PADDING = 0.7109375
 #
 
 function copynode(o::XML.Node)
-    n = XML.Node(o.nodetype, o.tag, o.attributes, o.value, isnothing(o.children) ? nothing : [copynode(x) for x in o.children])
-    return n
+    attrs = isnothing(o.attributes) ? nothing : copy(o.attributes)
+    children = isnothing(o.children) ? nothing : XML.Node{String}[copynode(x) for x in o.children]
+    return XML.Node{String}(o.nodetype, o.tag, attrs, o.value, children)
 end
 function do_sheet_names_match(ws::Worksheet, rng::T) where {T<:Union{SheetCellRef,AbstractSheetCellRange}}
     if ws.name == rng.sheet
@@ -125,8 +126,8 @@ function do_sheet_names_match(ws::Worksheet, rng::T) where {T<:Union{SheetCellRe
 end
 
 function make_child_node(tag::String, name::String, pfx::String)::XML.Node
-    children = tag ∈ ("border", "fill") ? Vector{XML.Node}() : nothing
-    return XML.Node(XML.Element, pfx*name, OrderedDict{String,String}(), nothing, children)
+    children = tag ∈ ("border", "fill") ? Vector{XML.Node{String}}() : nothing
+    return XML.Node{String}(XML.Element, pfx*name, Pair{String,String}[], nothing, children)
 end
 
 function build_font_child!(new_node::XML.Node, tag::String, name::String, attrs::Union{Nothing,Dict{String,String}}, pfx::String)
@@ -216,7 +217,7 @@ end
                 if isnothing(attributes[a])
                     cnode = XML.Element(a)
                 else
-                    cnode = XML.Node(XML.Element, a, OrderedDict{String,String}(), nothing, tag ∈ ["border", "fill"] ? Vector{XML.Node}() : nothing)
+                    cnode = XML.Node{String}(XML.Element, a, Pair{String,String}[], nothing, tag ∈ ["border", "fill"] ? Vector{XML.Node{String}}() : nothing)
                     for (k, v) in attributes[a]
                         cnode[k] = v
                     end
@@ -228,7 +229,7 @@ end
                 if isnothing(attributes[a])
                     cnode = XML.Element(a)
                 else
-                    cnode = XML.Node(XML.Element, a, OrderedDict{String,String}(), nothing, tag ∈ ["border", "fill"] ? Vector{XML.Node}() : nothing)
+                    cnode = XML.Node{String}(XML.Element, a, Pair{String,String}[], nothing, tag ∈ ["border", "fill"] ? Vector{XML.Node{String}}() : nothing)
                     color = XML.Element("color")
                     for (k, v) in attributes[a]
                         if k == "style" && v != "none"
@@ -255,7 +256,7 @@ end
                 if isnothing(attributes[a])
                     cnode = XML.Element(a)
                 else
-                    cnode = XML.Node(XML.Element, a, OrderedDict{String,String}(), nothing, tag ∈ ["border", "fill"] ? Vector{XML.Node}() : nothing)
+                    cnode = XML.Node{String}(XML.Element, a, Pair{String,String}[], nothing, tag ∈ ["border", "fill"] ? Vector{XML.Node{String}}() : nothing)
                     patternfill = XML.Element("patternFill")
                     fgcolor = XML.Element("fgColor")
                     bgcolor = XML.Element("bgColor")
@@ -475,14 +476,14 @@ function get_new_formatId(wb::Workbook, format::String)::Int
         if isnothing(j) # There are no existing custom formats
             return styles_add_numFmt(wb, format)
         else
-            existing_elements_count = length(XML.children(xroot[i][j]))
+            existing_elements_count = length(xml_elements(xroot[i][j]))
             if parse(Int, xroot[i][j]["count"]) != existing_elements_count
                 throw(XLSXError("Wrong number of font elements found: $existing_elements_count. Expected $(parse(Int, xroot[i][j]["count"]))."))
             end
 
             format_node = XML.Element("numFmt";
                 numFmtId=string(existing_elements_count + PREDEFINED_NUMFMT_COUNT),
-                formatCode=XLSX.escape(format)
+                formatCode=format
             )
 
             return styles_add_cell_attribute(wb, format_node, "numFmts") + PREDEFINED_NUMFMT_COUNT
@@ -529,7 +530,7 @@ function update_template_xf(ws::Worksheet, allXfNodes::Vector{XML.Node}, existin
     old_cell_xf = styles_cell_xf(allXfNodes, Int(existing_style.id))
     new_cell_xf = copynode(old_cell_xf)
     if isnothing(new_cell_xf.children)
-        new_cell_xf=XML.Node(new_cell_xf, alignment)
+        new_cell_xf = XML.Node{String}(XML.Element, XML.tag(new_cell_xf), new_cell_xf.attributes, nothing, XML.Node{String}[alignment])
     elseif length(XML.children(new_cell_xf)) == 0
         push!(new_cell_xf, alignment)
     else
@@ -550,14 +551,14 @@ end
 function styles_add_cell_attribute(wb::Workbook, new_att::XML.Node, att::String)::Int
     xroot = styles_xmlroot(wb)
     i, j = get_idces(xroot, "styleSheet", att)
-    existing_elements_count = length(XML.children(xroot[i][j]))
+    existing_elements_count = length(xml_elements(xroot[i][j]))
     if parse(Int, xroot[i][j]["count"]) != existing_elements_count
         throw(XLSXError("Wrong number of elements elements found: $existing_elements_count. Expected $(parse(Int, xroot[i][j]["count"]))."))
     end
 
     # Check new_att doesn't duplicate any existing att. If yes, use that rather than create new.
-    for (k, node) in enumerate(XML.children(xroot[i][j]))
-        if localname(new_att) == "numFmt" # mustn't compare numFmtId attribute for formats
+    for (k, node) in enumerate(xml_elements(xroot[i][j]))
+        if XML.tag(new_att) == "numFmt" # mustn't compare numFmtId attribute for formats
             if node["formatCode"] == new_att["formatCode"]
                 return k - 1 # CellDataFormat is zero-indexed
             end
@@ -1110,7 +1111,7 @@ function process_uniform_core(f::Function, ws::Worksheet, allXfNodes::Vector{XML
     if first                           # Get the attribute of the first cell in the range.
         newid = f(ws, cellref; kw...)
         new_alignment = getAlignment(ws, cellref).alignment["alignment"]
-        alignment_node = XML.Node(XML.Element, "alignment", new_alignment, nothing, nothing)
+        alignment_node = XML.Node{String}(XML.Element, "alignment", isnothing(new_alignment) ? Pair{String,String}[] : Pair{String,String}[k => v for (k,v) in new_alignment], nothing, nothing)
         first = false
     else                               # Apply the same attribute to the rest of the cells in the range.
         if cell.style == UInt64(0)
@@ -1320,12 +1321,12 @@ function update_sharedString_font(ws::Worksheet, cell::Cell;
 
     is = parse(str_formatted, XML.Node)[1] # Convert to XML.Node for ease of handling
 
-    all_r = filter(z -> z.tag == "r", XML.children(is))
+    all_r = filter(z -> localname(z) == "r", XML.children(is))
     run_elements = reduce(vcat, [XML.children(z) for z in all_r])
-    rPr_elements=filter(z -> z.tag == "rPr", run_elements) # rPr elements
+    rPr_elements = filter(z -> localname(z) == "rPr", run_elements) # rPr elements
 
     t=String[] # text elements
-    for i in filter(z -> z.tag == "t", run_elements)
+    for i in filter(z -> localname(z) == "t", run_elements)
         push!(t, XML.is_simple(i[1]) ? XML.simple_value(i[1]) : XML.value(i[1]))
     end
 
@@ -1337,7 +1338,7 @@ function update_sharedString_font(ws::Worksheet, cell::Cell;
 
         for att in XML.children(rPr) # first copy existing attributes
             for i in 1:length(atts)
-                if att.tag == atts[i]
+                if XML.tag(att) == atts[i]
                     new_rPr[i] = att
                 end
             end
@@ -1372,7 +1373,7 @@ function update_sharedString_font(ws::Worksheet, cell::Cell;
         if !isnothing(rPr.children)
             empty!(rPr.children)
             foreach(new_rPr) do element 
-                element.tag != "DeleteMe" && push!(rPr.children, element)
+                XML.tag(element) != "DeleteMe" && push!(rPr.children, element)
             end
         end
     end
@@ -1420,7 +1421,7 @@ function update_sharedString_font(ws::Worksheet, cell::Cell;
         for r in 1:length(all_r)
             if t[r] != ")___DeleteMe___(" # signals a merged <r> element to be skipped
                 write(new_r, "  <r>\n")
-                r > inc_first && write(new_r, XML.write(rPr_elements[r-inc_first];depth=3) * "\n")
+                r > inc_first && write(new_r, XML.write(rPr_elements[r-inc_first]) * "\n")
                 write(new_r, "    <t" * (needs_preserve(t[r]) ? " xml:space=\"preserve\"" : "") * ">" *t[r] * "</t>\n")
                 write(new_r, "  </r>\n")
             end