fileformer
search⌘KtranslateENexpand_moreupload_fileConvert
Converters8 file families · 44 direct conversions
What changes in a conversionarrow_forward
home/tools/document-tools

The right format depends on what the other person needs to do.

Sending a document to read is different from sending one to edit, sign or print. Pick the app you wrote it in and what they need to do. We'll say which format to send, what they'll lose, and the converter that gets you there.

you wrote it indescriptionMicrosoft Word.docxcloudGoogle Docsonlinelaptop_macApple Pages.pagesedit_noteLibreOffice.odttable_chartExcel.xlsxslideshowPowerPoint.pptx
arrow_forward
they need tovisibilityRead iteditEdit itdrawSign itprintPrint itdatasetUse the datamenu_bookRead on an e-reader
send themPDFfixed layoutA PDF looks identical on every phone, laptop and printer, and can’t be changed by accident. It’s what people expect when they only need to read.
check_circleFonts, layout and images exactly as you see themcheck_circleOpens on any device without the original appinfoComments and tracked changes are left out unless you print them
Convert Word to PDFWord → .pdfarrow_forward
~/there-and-back

Word → PDF → Word is not a round trip.

A PDF records where things appear on the page, not what they are. Converting back rebuilds the document from those positions, so some features come back as imitations. Pick a feature to follow it through each step.

titleHeadingsedit_noteTracked changestableTablescommentCommentsfunctionsEquationstext_fieldsFonts
1 · in Wordwith history
rent of £1,150rent of £1,195
Every edit is recorded with who made it and when.arrow_forward
arrow_forwardarrow_forward
2 · as PDFfinal text only
rent of £1,195
The PDF shows the final text. The edit history isn’t stored at all.arrow_forward
arrow_forwardarrow_forward
3 · back in Wordhistory gone
rent of £1,195no history
Nothing to rebuild from: the history is gone for good. Keep the original DOCX.arrow_forward
Tracked changes can’t survive the trip. If you might need to edit again, keep the original and send the PDF alongside it.
~/what-the-file-remembers

A .docx is a zip of XML files. Some of them remember more than you think.

DOCX, XLSX and PPTX, along with ODT, ODS, ODP and EPUB, are ZIP archives of XML files. Drop one in and your browser unzips it, then reads the parts that travel with every copy you send: authors, editing time, deleted wording kept as tracked changes, comments, hidden text, hidden sheets and slides, speaker notes, macros and external links. Everything happens on your device. Or open the sample, a tenancy agreement built in your browser just now.

descriptionSample: tenancy-agreement.docxdownloadDownload it, open it in Word
tenancy-agreement.docx6.3 KB · 8 parts · 5.4 KB unzipped
parts inside · unzipped size · method
[Content_Types].xml930 Bstored
_rels/.rels589 Bstored
word/document.xml1.4 KBstored
word/_rels/document.xml.rels626 Bstored
word/comments.xml342 Bstored
word/settings.xml402 Bstored
docProps/core.xml682 Bstored
docProps/app.xml415 Bstored
Word documentOffice Open XML · WordprocessingML
Authorcore.xmlJ. PatelLast saved bycore.xmlAna RuizCreatedcore.xml2026-09-02 09:12 UTCLast savedcore.xml2026-09-14 16:41 UTCLast printedcore.xml2026-09-13 08:15 UTCRevision numbercore.xml14Total editing timeapp.xml5 h 12 minCompanyapp.xmlNorthwind Lettings LtdManagerapp.xmlR. OkaforTemplateapp.xmlNorthwind Lettings.dotxSaved withapp.xmlMicrosoft Office Word 16.0000
travels with every copy
edit_note4 tracked changesword/document.xml · <w:del>, <w:ins>Deleted wording stays in the file until the changes are accepted. Anyone can switch on Show Markup and read it.“£1,150” → “£1,195” “1 October” → “15 October”
comment1 commentword/comments.xmlMargin comments, with their author and timestamp.Can we push this to six weeks? The last tenant left damage.
visibility_offHidden textword/document.xml · <w:vanish/>Formatted as hidden, so it doesn’t show or print, but it’s plain text in the XML.Internal note: the landlord would accept £1,100 if pushed.
folder_sharedLocal file paths_rels · TargetMode="External"Relationships point to files on the author’s computer or network, revealing usernames and folder names.file:///C:/Users/jpatel/Templates/Northwind Lettings.dotx
history7 editing sessions recordedword/settings.xml · <w:rsids>Word tags each save session with a random revision ID (rsid). Comparing them across copies can show which edits came from the same session.
track_changesTrack changes is still onword/settings.xmlEvery edit the recipient makes will be recorded too.
group2 names in the filecore.xml, comments, revisionsEvery person recorded as an author, editor or commenter.J. Patel, Ana Ruiz
before you send it4 things to clear. Word to PDF: send what you see, nothing elsearrow_forward
~/working-with-the-words

Tools for the text itself.

Count it, compare versions, fix the capitals, or write it in Markdown. Each runs in the browser as you type.

~/the-simplest-format-that-isnt

CSV has a standard. Spreadsheets don't follow it.

RFC 4180 only defines the commas, quotes and line breaks. It says nothing about character encoding, dates, leading zeros or which character separates the columns. Edit the file below. The left grid is a strict RFC 4180 parse. The right grid shows what a spreadsheet typically does when you double-click the file. The encoding panel shows the actual bytes.

Customer exportGene listFormula injectionSave with UTF-8 BOMGerman-locale spreadsheet
BYTES ON DISK · UTF-8211 bytes69 64 2C 6E 61 6D 65 2C 63 69 74 79 2C 70 68 6F 6E 65 2C 73 69 67 6E 75 70 2C 6E 65 77 73 6C 65 74 74 65 72 0A 30 30 31 …DECODED AS WINDOWS-1252 (no BOM, so Excel guesses)id,name,city,phone,signup,newsletter00123,Zoë Müller,Köln,+44 20 7946 0958,03/04/2026,TRUE00124,"Ruiz, Ana",São Paulo,0049 30 1234567,1-2,FALSE00125,José Café,Málaga,1234567890123456789,10/11/2026,TRUEé is C3 A9 in UTF-8. Windows-1252 maps C3 to à and A9 to ©, so Café becomes Café. This decode really ran in your browser.
RFC 4180 parse · 4 rows × 6 columns, all text
idnamecityphonesignupnewsletter
00123Zoë MüllerKöln+44 20 7946 095803/04/2026TRUE
00124Ruiz, AnaSão Paulo0049 30 12345671-2FALSE
00125José CaféMálaga123456789012345678910/11/2026TRUE
spreadsheet, double-clicked · en-US
idnamecityphonesignupnewsletter
123Zoë MüllerKöln+44 20 7946 095803/04/2026TRUE
124Ruiz, AnaSão Paulo0049 30 12345672-JanFALSE
125José CaféMálaga1.23457E+1810/11/2026TRUE
translateAccents garbledWithout a byte-order mark, Excel on Windows reads the file as Windows-1252, so each UTF-8 accent turns into two wrong characters.
cancelData silently lost · 4Leading zeros and digits past the 15th are dropped. Saving the file again writes the damaged values back. Here: 00123, 00124, 00125, 1234567890123456789
change_circleConverted to something else · 4Dates, booleans and scientific numbers replace the text. This is how gene names like SEPT2 became dates. Here: TRUE, 1-2, FALSE, TRUE
calendar_monthAmbiguous dates · 2Day-month or month-day depends on the computer’s region, not the file. ISO 8601 (2026-04-03) avoids this. Here: 03/04/2026, 10/11/2026

The spreadsheet column models Excel's default behaviour when it opens a .csv by double-click. It isn't Excel itself. Recent versions let you turn some of these conversions off, and the Data → From Text/CSV import avoids most of them. Opening the file with CSV to XLSX sets each column's type explicitly.

~/every-document-converter

19 converters, grouped by the kind of document.

Each page explains what changes on the way, with a live demo of the tricky parts. The family converters handle the less common pairs.

descriptionWord processingDOCX · ODT · RTF · PAGES · TXT
table_chartSpreadsheetsXLSX · CSV
slideshowPresentationsPPTX · KEY · ODP
notesPlain textTXT
menu_bookEbooksEPUB · MOBI
~/reference

Twelve office formats, by what's really inside.

Most modern office files are ZIP archives and start with the same four bytes, 50 4B 03 04. What tells them apart is a part inside: [Content_Types].xml for Microsoft, or a mimetype file for OpenDocument and EPUB. The older binary formats share a different signature, the OLE2 compound file header.

formatMIME typefirst bytestells it apartstandard
DOCXapplication/vnd.openxmlformats-officedocument.wordprocessingml.document50 4B 03 04word/document.xmlECMA-376 · ISO/IEC 29500 · 2006
XLSXapplication/vnd.openxmlformats-officedocument.spreadsheetml.sheet50 4B 03 04xl/workbook.xmlECMA-376 · ISO/IEC 29500 · 2006
PPTXapplication/vnd.openxmlformats-officedocument.presentationml.presentation50 4B 03 04ppt/presentation.xmlECMA-376 · ISO/IEC 29500 · 2006
ODTapplication/vnd.oasis.opendocument.text50 4B 03 04mimetype first, uncompressedOASIS ODF 1.3 · ISO/IEC 26300
ODSapplication/vnd.oasis.opendocument.spreadsheet50 4B 03 04mimetype …spreadsheetOASIS ODF 1.3 · ISO/IEC 26300
DOCapplication/mswordD0 CF 11 E0 A1 B1 1A E1WordDocument streamMS-DOC (open spec) · 1997 binary
XLSapplication/vnd.ms-excelD0 CF 11 E0 A1 B1 1A E1Workbook stream (BIFF8)MS-XLS (open spec)
PPTapplication/vnd.ms-powerpointD0 CF 11 E0 A1 B1 1A E1PowerPoint Document streamMS-PPT (open spec)
RTFapplication/rtf7B 5C 72 74 66 (“{\rtf”)plain text control wordsMicrosoft RTF 1.9.1 · 2008
CSVtext/csvnone: plain textcommas and quotesRFC 4180 · 2005
PAGESapplication/vnd.apple.pages50 4B 03 04Index/Document.iwaApple, unpublished
EPUBapplication/epub+zip50 4B 03 04mimetype at byte 38W3C EPUB 3.3 · 2023

From binary files to zipped XML

1983Word 1.0Microsoft Word ships for MS-DOS, with its own binary format.
1987RTFRich Text Format arrives as a plain-text way to move formatted documents between programs.
1997Office 97DOC, XLS and PPT settle on the OLE2 compound file format used for the next decade.
2005ODF and RFC 4180OASIS approves OpenDocument, and CSV finally gets a written specification.
2006Office Open XMLECMA-376 publishes the zipped-XML DOCX, XLSX and PPTX formats.
2007Office 2007DOCX becomes Word’s default format.
2008Binary specs openedMicrosoft publishes the DOC, XLS and PPT binary specifications under its Open Specification Promise.
2013iWork goes binaryPages, Numbers and Keynote switch from XML to Snappy-compressed protobuf .iwa files.