How does the tool decide which element becomes a row?

It counts the element paths in the file and takes the one that repeats most, breaking a tie in favour of the shallower path. For a file of orders that is /orders/order, and the paths it considered sit under the box as buttons, up to four of them. Type your own path to overrule it: an absolute path such as /orders/order, or just the element name, which matches that name anywhere in the tree. A namespace prefix never has to be typed, so /orders/order also selects ns:order.

What happens when one record is missing a field?

The field is written empty and the columns stay where they are. In the sample, the second order has no qty, so its line is 2,Hammer, with a trailing empty field rather than a row that is one column short. The tool also says so: a warning names the column and how many of the rows it was missing from, because a converter that silently shortens a row is the reason CSV imports go wrong two steps later.

Is it safe to paste XML that came from somewhere else?

That is what this converter is built around. The XML is read as characters and turned into plain objects, so nothing in it is constructed, evaluated or resolved. DTD processing is off: a DOCTYPE that declares an entity, or that names an external DTD, stops the conversion with a message naming it, and nothing at that address is read or fetched. That single rule is also what bounds an entity bomb, because a declaration is never expanded. Elements nested more than 400 deep are refused rather than read part way, and nothing is uploaded.

How are commas and quotes inside a value handled?

By RFC 4180. A field carrying a comma, a double quote or a line break is wrapped in double quotes, and a double quote inside such a field is doubled. A product called Wrench, 8mm comes out as the quoted field "Wrench, 8mm", so the comma stays inside the value instead of splitting the row. Records are separated by a carriage return and a line feed, which is the line ending the same document specifies.

What if one column path matches several nodes in the same row?

You choose. By default the values are joined into one field, separated by a semicolon and a space, so two tags come out as red; blue, and a warning says how many rows that happened in. The other two settings keep only the first match, or write the number of matches instead, which is useful when you want a count of line items rather than their contents. Whichever you pick, the row keeps exactly one field per column.

How large a file can it take, and is anything uploaded?

One paste can contain up to 4 MiB of UTF-8 XML (4,194,304 bytes), with at most 400 nesting levels and 256 requested columns. Conversion runs in this tab. Larger input is refused with both sizes given; it is not partly exported. The preview shows the first 40 records while Copy and Download include all accepted records. A draft up to 200,000 text units and your settings are saved in this browser. Start over clears them.

How do I extract XML attributes or nested elements into CSV columns?

Choose the row element first. Use @id for an attribute on that row, address/city for a nested child, item/@sku for an attribute on a child, and . for the row element’s own text. List the column paths separated by commas. These are simple paths, not full XPath: predicates, wildcards, functions and attribute-based row filters are not supported. Unprefixed names can match prefixed XML names, but the converter does not distinguish namespace URIs; use the exact prefixed name when the local names overlap.

Can I convert an XML file to Excel or XLSX?

Open the XML file as text and paste it here, choose the rows and columns, then download rows.csv. Import that CSV into Excel with the relevant columns set to Text for codes, dates or long IDs that must preserve their characters. For an XLSX file, open the CSV in CSV to XLSX and inspect its column types before downloading. This is a two-step value conversion; it does not reproduce a workbook’s styles, formulas or XML mappings.

Can I extract a sitemap or an RSS feed?

For a standard sitemap, select /urlset/url and choose loc, lastmod, changefreq or priority as columns. A sitemap index uses /sitemapindex/sitemap. For an RSS feed, choose /rss/channel/item and fields such as title, link and pubDate. Paste the downloaded XML as text; this page does not fetch feed URLs or follow sitemap links. Missing elements stay empty, and mixed content is treated as text rather than preserved markup.

Can I convert CSV back to XML or process multiple XML files?

The output is CSV only. There is no CSV-to-XML or Excel-to-XML writer, batch input, scheduled feed conversion, XSD mapping or native SpreadsheetML workbook reader. A generic XML export from another product needs row and column paths suited to that export; compatibility with Apple Health, WordPress, Nmap and other product-specific schemas has not been certified.

XML to CSV

Paste XML, select the repeating element that becomes a record and choose the paths that become columns. A row path such as /orders/order can pair with @id, name and address/city; a child missing from one record becomes an empty field. Repeated matches can be joined, reduced to the first value or counted. The preview shows 40 rows, while CSV copy and download contain every accepted record. The XML is read locally without resolving external entities.

xml2 records

csv3 columns

row path /orders/ordercolumns @id, name, qtyid,name,qty1,"Wrench, 8mm",22,Hammer,

Row path

Columns

Repeated nodes

Rows detected at /orders/order, columns detected from the records. When one column path matches several nodes in a row, they are joined with "; ".

Row path

/orders/order

Rows

Columns

Entities

Blocked

What the reader noticed

The column "qty" is missing from 1 of the 2 rows, so those fields are empty and the columns stay aligned.

2 rows from /orders/order, 3 columns, 1 warnings, 0 network requests made.

A comma inside a valueWrench, 8mm
id,name,qty 1,"Wrench, 8mm",2
qty missing on row 2same row path
id,name,qty 1,"Wrench, 8mm",2 2,Hammer,The column "qty" is missing from 1 of the 2 rows, so those fields are empty and the columns stay aligned.
A DOCTYPE with an external entityany XML
The DOCTYPE points at an external DTD (http://example.com/orders.dtd). DTD processing is off here, so the file is refused rather than resolved, and nothing at that address is read or fetched. Delete the DOCTYPE and convert the elements.

Accuracy. The XML is parsed inert with DTD processing and entity expansion switched off, so an external entity is refused rather than resolved and no file or URL it names is fetched. Rows come from the element path you select and each column from its own path: a missing child becomes an empty field so the columns never shift, and values are quoted by RFC 4180. Nodes that match nothing are listed as warnings.

Up to 4 MiB of XML in one pass, read here in your browser with nothing uploaded and no account. Leave both boxes empty and the tool picks the element that repeats and the columns those records carry; type a path to overrule it. A column path is a child name, an attribute like @id, a nested path like address/city, or a dot for the text of the row element itself, and a namespace prefix never has to be typed. Up to 256 columns, elements nested up to 400 deep, and the panel prints the first 40 rows while the CSV you copy or download holds every one. Your three choices and your draft are kept in this browser alone, drafts up to 200,000 characters, and the Start over button above the tool forgets both.

Common questions

How does the tool decide which element becomes a row?: It counts the element paths in the file and takes the one that repeats most, breaking a tie in favour of the shallower path. For a file of orders that is /orders/order, and the paths it considered sit under the box as buttons, up to four of them. Type your own path to overrule it: an absolute path such as /orders/order, or just the element name, which matches that name anywhere in the tree. A namespace prefix never has to be typed, so /orders/order also selects ns:order.
What happens when one record is missing a field?: The field is written empty and the columns stay where they are. In the sample, the second order has no qty, so its line is 2,Hammer, with a trailing empty field rather than a row that is one column short. The tool also says so: a warning names the column and how many of the rows it was missing from, because a converter that silently shortens a row is the reason CSV imports go wrong two steps later.
Is it safe to paste XML that came from somewhere else?: That is what this converter is built around. The XML is read as characters and turned into plain objects, so nothing in it is constructed, evaluated or resolved. DTD processing is off: a DOCTYPE that declares an entity, or that names an external DTD, stops the conversion with a message naming it, and nothing at that address is read or fetched. That single rule is also what bounds an entity bomb, because a declaration is never expanded. Elements nested more than 400 deep are refused rather than read part way, and nothing is uploaded.
How are commas and quotes inside a value handled?: By RFC 4180. A field carrying a comma, a double quote or a line break is wrapped in double quotes, and a double quote inside such a field is doubled. A product called Wrench, 8mm comes out as the quoted field "Wrench, 8mm", so the comma stays inside the value instead of splitting the row. Records are separated by a carriage return and a line feed, which is the line ending the same document specifies.
What if one column path matches several nodes in the same row?: You choose. By default the values are joined into one field, separated by a semicolon and a space, so two tags come out as red; blue, and a warning says how many rows that happened in. The other two settings keep only the first match, or write the number of matches instead, which is useful when you want a count of line items rather than their contents. Whichever you pick, the row keeps exactly one field per column.
How large a file can it take, and is anything uploaded?: One paste can contain up to 4 MiB of UTF-8 XML (4,194,304 bytes), with at most 400 nesting levels and 256 requested columns. Conversion runs in this tab. Larger input is refused with both sizes given; it is not partly exported. The preview shows the first 40 records while Copy and Download include all accepted records. A draft up to 200,000 text units and your settings are saved in this browser. Start over clears them.
How do I extract XML attributes or nested elements into CSV columns?: Choose the row element first. Use @id for an attribute on that row, address/city for a nested child, item/@sku for an attribute on a child, and . for the row element’s own text. List the column paths separated by commas. These are simple paths, not full XPath: predicates, wildcards, functions and attribute-based row filters are not supported. Unprefixed names can match prefixed XML names, but the converter does not distinguish namespace URIs; use the exact prefixed name when the local names overlap.
Can I convert an XML file to Excel or XLSX?: Open the XML file as text and paste it here, choose the rows and columns, then download rows.csv. Import that CSV into Excel with the relevant columns set to Text for codes, dates or long IDs that must preserve their characters. For an XLSX file, open the CSV in <a href="/spreadsheet-converter/csv-to-xlsx">CSV to XLSX</a> and inspect its column types before downloading. This is a two-step value conversion; it does not reproduce a workbook’s styles, formulas or XML mappings.
Can I extract a sitemap or an RSS feed?: For a standard sitemap, select /urlset/url and choose loc, lastmod, changefreq or priority as columns. A sitemap index uses /sitemapindex/sitemap. For an RSS feed, choose /rss/channel/item and fields such as title, link and pubDate. Paste the downloaded XML as text; this page does not fetch feed URLs or follow sitemap links. Missing elements stay empty, and mixed content is treated as text rather than preserved markup.
Can I convert CSV back to XML or process multiple XML files?: The output is CSV only. There is no CSV-to-XML or Excel-to-XML writer, batch input, scheduled feed conversion, XSD mapping or native SpreadsheetML workbook reader. A generic XML export from another product needs row and column paths suited to that export; compatibility with Apple Health, WordPress, Nmap and other product-specific schemas has not been certified.

The XML is parsed inert with DTD processing and entity expansion switched off, so an external entity is refused rather than resolved and no file or URL it names is fetched. Rows come from the element path you select and each column from its own path: a missing child becomes an empty field so the columns never shift, and values are quoted by RFC 4180. Nodes that match nothing are listed as warnings.