[businessarshgoyal]

Description

Adds a WPS writer producing documents for the word processor of Microsoft Works 7 and 8, ie an OLE compound file whose CONTENTS stream holds the CHNKWKS chunks of the document, as described by the format notes linked in the issue and implemented by libwps (WPS8Parser).

$writer = IOFactory::createWriter($phpWord, 'WPS');
$writer->save(__DIR__ . '/sample.wps');

What is added:

• Shared\OLEWrite: a small compound file (CFB v3, 512 bytes sectors) writer, counterpart of the existing Shared\OLERead, building the FAT, the mini FAT, the mini stream and the red-black directory from a set of named streams. No new dependency.
• Writer\WPS and Writer\WPS\Part\Contents, following the structure of the RTF writer: the part writes the stream header, the chunk index and the TEXT and FONT chunks; the text is UTF-16LE, paragraphs are terminated by \r and page breaks by \f.
• Registration of WPS in IOFactory::createWriter().

First iteration scope, as suggested in the issue: text and basic paragraph output. Text, text runs, titles, links, list items, text breaks, page breaks and the text of the cells of a table are written as paragraphs. Not supported yet (silently ignored, no error): character and paragraph styles, table and list structure, images, objects, headers, footers, footnotes, endnotes, comments, table of contents, document properties, form fields, and reading .wps files.

About #2769

I looked at #2769 first and did not build on it: it writes a ZIP package of ODF-like XML parts (content.xml, styles.xml, META-INF/manifest.xml), which is not the Microsoft Works binary format the issue asks for — such a file is not an OLE document and is not recognised by Works or by libwps-based readers (LibreOffice). This PR writes the CONTENTS/CHNKWKS stream instead, so the two approaches do not share code.

Validation

Besides the unit tests, the generated files were checked against the reference implementation, wps2text (libwps 0.4.12), which extracts the expected text, and against olefile/file(1), which validate the compound file structure.

Fixes #69

/claim #69

Checklist:

• My CI is 🟢
• I have covered by unit tests my new code (check build/coverage for coverage report)
• I have updated the documentation to describe the changes
• I have updated the changelog

[coveralls]

coverage: 96.636% (+0.01%) from 96.622% — businessarshgoyal:devin/1788015204-wps-writer into PHPOffice:master

[Progi1984]

Rather than adding file src/PhpWord/Shared/OLEWrite.php, could use PhpOffice/Common ?

[businessarshgoyal]

Happy to, but PhpOffice/Common only has Common\Microsoft\OLERead on master (1.0.5), there is no compound file writer in it, and PHPWord dropped the phpoffice/common requirement back in 67b1598 (Jan 2021), so pulling it in for this would both add a dependency and still need the writer to be written somewhere.

Two ways I can go, tell me which you prefer:

1. I open a PR on PhpOffice/Common adding Common\Microsoft\OLEWrite as the counterpart of OLERead, then this PR requires phpoffice/common and drops src/PhpWord/Shared/OLEWrite.php. Cleanest long term since PHPPresentation and PHPWord could share it, but it blocks this PR on a Common release
2. It stays in PHPWord as Shared\OLEWrite, marked @internal, and moves to Common later without a breaking change for users

Default is 2 unless you say otherwise

Written by Devin

[Progi1984]

Go to 1

[businessarshgoyal]

Done: PHPOffice/Common#65 adds Common\Microsoft\OLEWrite (round-trip tested against OLERead; phpunit, phpstan level 7 and phpmd pass).

Once that lands and Common tags a release, I'll update this PR to require phpoffice/common and delete src/PhpWord/Shared/OLEWrite.php, pointing Writer\MsDoc at the Common class. I've left the copy in place here for now so this PR keeps building — say the word if you'd rather I switch it to dev-master immediately instead of waiting for the tag.

Written by Devin

[Progi1984]

@businessarshgoyal It's published : https://github.com/PHPOffice/Common/releases/tag/1.1.0

[businessarshgoyal]

Thanks @Progi1984. Switched in e274d83e7: composer.json now requires phpoffice/common ^1.1, Writer\\WPS uses PhpOffice\\Common\\Microsoft\\OLEWrite, and src/PhpWord/Shared/OLEWrite.php is gone. Also merged current master. tests/PhpWordTests/Writer/WPSTest.php is 7 passed locally; phpstan and php-cs-fixer clean on the touched file.

Written by Devin

[Progi1984]

@businessarshgoyal ⚠️ the CI is 🔴

[businessarshgoyal]

Thanks @Progi1984 — fixed in a84bae6. The static-analysis failure was a side effect of requiring phpoffice/common: it pulls in pclzip/pclzip, so PclZip becomes a known class to phpstan and the previously-baselined "unknown class PclZip" ignores in src/PhpWord/Shared/ZipArchive.php no longer matched. Added the 5 corresponding undefined method entries to phpstan-baseline.neon; phpstan is clean locally.

[Progi1984]

@businessarshgoyal It's merged for the writer. Thanks. Do you plan a reader too ?