LanguageOps extracts translation-ready content from office files, publishing packages, structured data, bilingual formats, websites, repositories and media workflows. We preserve the original structure wherever the format supports a round trip, and give you previews or filter controls when extraction needs judgement.
Files containing millions of words are supported. Format retention is always our priority.
The extension tells us what the file is. Its structure, your extraction choices and the required output determine the safest route through the platform.
DOC · DOCX · RTF · ODT · IDML · PDF
Office and publishing filters protect formatting runs, inline objects and document structure. IDML adds layer analysis and layout-aware context; scanned or visually complex PDFs can use OCR.
Example: A multilingual product manual keeps headings, emphasis, links, tables and inline formatting attached to the translated output.
PPT · PPTX · ODP
Slide text is extracted through structure-aware office filters so translated content can be merged back into the presentation rather than delivered as a flat text list.
Example: Translate slide titles, body copy and supported notes while retaining the deck as the delivery format.
XLS · XLSX · XLSM · ODS · CSV · TSV
Use a standard spreadsheet filter or the native column-pair engine for multilingual columns, repeated source/target pairs, row rules, colours, notes, context and character limits.
Example: Map C→D for product titles and E→F for descriptions, translate only blank targets, and write each result back to its original cell.
HTML · HTM · XML · JSON · YAML · YML · PO · POT · PROPERTIES
Structure-aware filters separate translatable values from keys, markup and code. JSON key rules, HTML element previews and custom include/exclude rules help constrain extraction.
Example: Translate visible HTML copy and selected attributes while scripts, styles, element structure and protected values remain untouched.
TXT · LOG · MD · MARKDOWN · RST · TEX · LATEX · SRT · VTT · SUB · SBV · ASS · SSA
Plain and marked-up text can be segmented directly or routed through a specialist filter. LaTeX commands are protected as inline codes; subtitle timing remains connected to the text workflow.
Example: Translate a LaTeX chapter while commands, references and protected arguments stay available as non-translatable tags.
XLIFF · XLF · XLIF · SDLXLIFF · MXLIFF · MQXLIFF
Existing bilingual files are imported directly, including source and target content, segment state and supported vendor metadata. The original bilingual is retained as the export base.
Example: Review an SDLXLIFF in the browser, move formatting tags with the translated text, and return a valid SDLXLIFF to the Trados workflow.
ZIP · SDLPPX · SDLRPX · SCORM ZIP
Package-aware inspection identifies the translatable files and supporting project metadata before processing. SCORM workflows retain course structure and connect text, subtitles and embedded media.
Example: Open an SDLPPX, detect the language direction from SDLPROJ, translate only the target-side bilingual files, and receive a rebuilt SDLPPX for delivery.
PNG · JPG · JPEG · GIF · WEBP · TIFF · BMP · MP4 · MOV · AVI · WEBM · MKV · MP3 · WAV · M4A · AAC · OGG · FLAC
OCR recovers text from images and complex PDFs. Dedicated audiovisual workflows handle transcription, subtitle translation, review, text-to-speech and dubbed delivery.
Example: Generate a timestamped transcript from video, translate it in the CAT editor, then export subtitles or a dubbed video.
Some legacy or unusually structured variants may require a filter choice or sample inspection. We confirm the extraction and delivery route rather than assuming that identical extensions contain identical structures.
LanguageOps recognises standalone SDLXLIFF and ZIP-backed SDLPPX or SDLRPX packages. Package inspection reads the embedded SDLPROJ to establish the real source and target variants and identify the target-side bilingual files.
SDLPPX → SDLPROJ → SDLXLIFF
SDL <g> formatting spans and <x/> standalone codes appear as protected editor tags. They are placed with the translation and reconstructed as real XML during export.
Source and existing target, confirmation state, match percentage, origin, lock information and other supported metadata are imported at marker level.
A missing, substituted or unbalanced SDL tag blocks export instead of silently flattening the bilingual. The original SDLXLIFF remains the reconstruction base.
Rules discovered in SDLPROJ can be reviewed for import into project prompts or deterministic QA. Embedded resources and remote TM or termbase references are reported separately.
Package delivery. LanguageOps retains the original SDLPPX or SDLRPX, replaces its target-side SDLXLIFF files with the completed bilinguals, and returns the rebuilt package with SDLPROJ, source bilinguals, reports and other package members preserved. Remote Trados memories or termbases referenced by SDLPROJ still require access credentials or a separately supplied TMX/TBX payload.
Inspect layers, text-story counts, representative text and applied-language hints. Exclude hidden, alternate-language or non-deliverable layers on a working copy while the untouched IDML remains available for final merge.
Review sample text by element (headings, paragraphs, lists and table cells) before processing. Scripts and styles are excluded, while the round-trip filter protects markup.
Verify the actual sample segments produced by selected sheets, source/target columns, column pairs, blank-target rules, row conditions and colour filters before committing.
Inspect which keys and values match include/exclude rules, whether standalone strings are extracted and whether HTML-like values should be skipped.
Preview sample paragraphs and supported header/footer choices for Word-family files before the complete structure-aware conversion.
Review recovered content as rendered HTML before moving complex PDF or image text into translation and export workflows.
A filter determines what becomes translatable and what must remain protected. The resulting bilingual is then used by the editor, memory, terminology and QA workflow.
Known extensions select an established round-trip filter for the file type. This is the quickest route for a conventional DOCX, PPTX, HTML, PO or similar source.
DOCX → Office filter → tagged bilingual → translated DOCXChoose supported format-specific behaviour such as Office comments and hidden text, PO bilingual mode, JSON key handling, OpenDocument notes or IDML extraction preferences.
DOCX + translate comments, but leave hidden text untouchedUse purpose-built extraction where ordinary filters are too limited: complex sheets, CSV column pairs, colour/row rules, IDML layer control, OCR and very large structured files.
XLSX + three column pairs + two target languages + blank targets onlyDescribe the content you need, for example “only rows where Status is Ready” or “exclude internal_note values”. LanguageOps resolves that into a reviewable structured rule instead of running arbitrary code.
JSON + exclude keys matching debug_* and internal_noteSave a successfully applied extraction configuration for an organisation or project so recurring files use the same reviewed rules.
Reuse the approved supplier-catalogue mapping every monthConnected content is converted into the same controlled segment workflow, with language, context and external identity retained for review and delivery.
Import projects and language content from Lokalise, work with translation memory, terminology and QA in LanguageOps, and keep the imported project relationship visible.
Extract titles, body content and supported fields from WordPress into bilingual segments, then publish approved translations through the connected workflow.
Select localisation files from a repository, process supported structured, text and subtitle formats, and return translated content to the development workflow.
Figma and other localisation-system connectors can bring content into the editor with external context. XLIFF, TMX and TBX remain available for standards-based interchange.
LanguageOps is designed to keep unusually large jobs together instead of forcing you to split the source into artificial chunks.
Large JSON is streamed in bounded batches, and spreadsheet extraction retains physical row and cell locations for write-back. The editor loads work incrementally while server-side search, assignments, memory and progress operate across the complete project.
Read the large-file guideWhere the source format supports a round trip, LanguageOps keeps the original file or bilingual skeleton and changes only the approved target content. Tags, placeholders, keys, cell locations, timing and vendor metadata are protected according to the format. If a faithful export cannot be produced, we prefer an explicit error over a deceptively successful flattened file.
Send us a representative sample and the required delivery format. We will confirm the extraction route, preview options and round-trip expectations before you commit the full project.