Universal mixed-file batch

Drop PDFs, Office documents, images, archives and data files together. MakeFileWork routes each file to a real validator or conservative repair path and rejects anything it cannot verify safely.

Reopen and verify a downloaded batch ZIP

A modified ZIP and matching forged report can pass: this is not a digital signature or proof of authorship.

Use a SHA-256 received separately from a source you trust. Copying the hash from this ZIP does not prove who created it.

Local only: files are processed in this browser and are not uploaded.

Up to 99 files, 25 MB each and 60 MB total. Processing and verification stay in your browser. Valid CSV/TSV files keep their exact bytes and CSV delimiters; incomplete CSV rows can be padded without changing existing cells. Spreadsheet formulas are rejected. Valid .tar.gz, .tgz and .gz downloads keep their original compression suffix.

Inspect a mixed batch without changing its files

Use Diagnose all files when you need an inventory before attempting any repair. The read-only diagnosis checks file signatures, runs available bounded structural validators, records a SHA-256 fingerprint for nonempty files and separates checked, review and rejected results. A damaged ZIP can retain its source fingerprint without being presented as a valid archive.

Export the JSON report for machine-readable evidence or the CSV audit for a spreadsheet review. The CSV neutralizes spreadsheet formula prefixes in filenames and findings. Neither report contains the original file bytes, and neither proves that a document is malware-free. A warning is not a successful repair; run the processing step separately only for files you choose to handle.

Evidence, not optimistic success

Every file receives a detected type, action and verified, repaired or rejected result. The download contains only valid outputs plus a JSON evidence report. Unsafe or colliding output filenames are normalized and recorded in the report without changing file contents.

You can cancel a running batch between processing steps; cancellation stops the next step and never publishes a partial ZIP. Validated TAR.GZ, TGZ and GZIP files retain their respective extensions, so downloaded archives remain recognizable by extraction tools. Progress distinguishes file validation from ZIP packaging. A very large synchronous operation may take time to finish before cancellation takes effect.

Valid CSV and TSV files are preserved byte-for-byte, including their original CSV delimiter and encoding. If CSV rows are short, the repair adds only missing empty cells and verifies all original cell values. Potential spreadsheet formulas are rejected rather than silently rewritten.

CSV line endings: why a repaired file may look different

Mixed line endings are valid text: one CSV record can end in CRLF, the next in LF, and another in CR. The batch verifier reads records with quoted fields, preserves the original delimiter and UTF-8 BOM, and leaves every record terminator in its original position when it must pad a short row. An embedded newline inside a quoted cell is part of the value and is never treated as a record boundary.

How to verify a repaired CSV

For example, a file with id,name, 1,Ana and a final 2 without a newline gains only the missing comma on its last record; the missing final newline stays missing. Reopen the output and compare row counts and cell values. If you instead want a standardized comma-separated file with CRLF endings, use the dedicated CSV repair tool. Neither tool can reconstruct values absent from the source.

What happens when every file fails validation?

Drag up to 99 individual files onto the batch area, or use the file picker. Folders are rejected rather than silently skipped. You can remove a file before processing; the 25 MB per-file and 60 MB total limits are checked before starting. No file leaves your browser.

If every file is rejected, download an evidence-only ZIP containing the JSON report and the input SHA-256 fingerprints, not a misleading repaired file. When only some files pass, download verified outputs individually or as one ZIP with the rejection report. Open the report to compare source names, detected types, rejection reasons, output names and SHA-256 hashes. A rejected file has no output hash or downloadable content.

For a specific invalid spreadsheet, use CSV repair; for damaged compressed files, see archive inspection. Neither tool guarantees recovery of missing bytes.

Retry only rejected files without changing verified outputs

For example, if a valid CSV passes but a malformed JSON file is rejected, select Retry rejected files only. The JSON is checked again against the same input fingerprint; the verified CSV is not processed again. The report records each attempt and still refuses to produce a download for JSON that cannot pass validation. Repeated retries cannot reconstruct missing bytes or repair unsupported formats.

Use Download JSON evidence report to inspect the results without extracting the ZIP. Each accepted file has a SHA-256 output hash; a rejected file has only its input fingerprint and a reason. Changing the file or the selection starts a new batch instead of reusing old verification evidence. For deeper problems, try File Doctor or JSON validation.

How the final ZIP is verified before download

The ZIP download is withheld if the JSON evidence report disagrees with a file result, a verified output no longer matches its original SHA-256 fingerprint, or an output changes type before packaging. Each included output is hashed again, and the ZIP is reopened to verify entry integrity. This detects inconsistencies; it does not restore lost source data.

XML is validated as XML, not just UTF-8 text

An XML file can contain readable text and still be unusable because of unclosed elements, undeclared namespace prefixes, duplicate attributes or undefined entities. The batch now parses the complete bounded XML document before marking it verified, preserves its original bytes and rejects external DTD/ENTITY expansion. For deliberate formatting, use XML validation and formatting. Unsupported XML is rejected rather than silently rewritten.

Office files with missing internal parts

DOCX, XLSX and PPTX are ZIP-based packages. The batch rejects encrypted, macro-enabled, path-ambiguous and structurally inconsistent packages rather than trying to bypass security checks. A workbook missing a worksheet, a presentation missing a slide or a document missing an embedded image cannot be called losslessly repaired: use File Doctor to review partial recovery evidence. Remaining Office text, values and slides are not silently reclassified as a complete original.

Fix one rejected file without rerunning the batch

Suppose a five-file batch contains three valid documents, one malformed JSON file and one unsupported binary. Choose Retry this rejected file beside the JSON entry to retry only that item. If you corrected it locally, choose Replace and retry this file and select the new JSON. The batch first rechecks the original input hashes and every retained successful output, then validates only the replacement. Other results and their attempt counts remain unchanged.

If the replacement still fails, it remains rejected with its own input SHA-256 and another attempt recorded; no fake repaired file is offered. Cancelling a retry keeps the last verified ZIP and report available. Use Download verified CSV audit for a spreadsheet-readable status table: its cells are escaped against formula injection and it is derived only after reopening the verified ZIP. The JSON evidence in the ZIP remains authoritative. For a more specific diagnosis, try File Doctor.

Check actual formats before running a mixed batch

After selecting files, the browser inspects format signatures and shows compatible, review and unsupported entries. A DOCX, XLSX or PPTX filename that actually contains a different package is flagged before processing; it will not be silently downloaded as a verified ZIP. Remove unsupported files with one action or keep them to receive explicit rejection evidence. This preliminary check is not a malware scan or proof that a document will open: full validation still happens during processing.

Reopen a downloaded batch ZIP without the original session

Use the independent ZIP verifier to select a batch archive saved earlier. It reads the ZIP directory and CRC values, compares the embedded JSON evidence with each output size and SHA-256, checks output format signatures and offers verified files and JSON/CSV audit exports again. A rejected source still has no downloadable output. The check works locally even after closing the original tab; no original source files are required.

This verifies internal consistency, not authenticity: someone who can alter both the ZIP contents and its evidence report can fabricate matching hashes. Keep the original ZIP SHA-256 in a separate trusted location if you need to compare later. The importer rejects nested or ambiguous names, unexpected files, inconsistent hashes and unsupported ZIP structures.

Verify a saved ZIP against an independently trusted SHA-256

Save the ZIP checksum sidecar when you download a batch. If you keep that fingerprint separately in a trusted place, paste its 64 hexadecimal characters into the importer before selecting the ZIP. The browser compares the entire archive against that fingerprint before reading the internal report and checks it again before every recovered download. A matching checksum confirms the bytes match your trusted reference; it does not identify the author, scan for malware or prove that a conversion retained every detail.

Continue a saved batch after correcting rejected files

Reopen the saved batch ZIP and select corrected files only for entries previously marked rejected. For example, a rejected JSON file can be replaced while the previously verified CSV and PDF outputs retain their original SHA-256 fingerprints. Several rejected files can be corrected in one operation. A replacement that still fails remains rejected; it is never described as repaired.

The browser independently checks the old ZIP, validates each replacement with the same processing engine, builds a new evidence report with updated attempt counts, then reopens and verifies the new ZIP before offering a download. Keep the original ZIP and its independently stored checksum: the resumed archive has a different SHA-256 because its contents and report changed. A saved ZIP does not contain rejected source bytes, so you must supply corrected files yourself.

Recover individual files after a partial batch resume

When corrected replacements are processed, you can download each verified output directly, export the new JSON evidence or a spreadsheet-safe CSV audit, or save the rebuilt ZIP and its SHA-256. If another replacement is rejected, continue correcting the remaining entries from the new verified archive without uploading the original source files again. Accepted output hashes are checked across recovery rounds. An independent .sha256 file can also be loaded before verifying an existing ZIP; its recorded filename must match the chosen archive. Keep that checksum separate from the ZIP if you need independent provenance evidence.

Compare a previous diagnosis with today's source files

Keep the Batch Doctor JSON report and optionally save its SHA-256 separately. Later select the current source files and import that report: MakeFileWork recomputes hashes and reruns structural validators rather than trusting saved verdicts. It distinguishes byte changes even when sizes match, renamed files, MIME changes, new diagnostic findings, reordered inputs and missing entries.

Export the comparison as JSON or spreadsheet-safe CSV. A matching comparison establishes consistency with current files, not authorship, malware safety or a cryptographic signature. The original files must be available for a full recheck.

Carry a Batch Doctor report as a portable evidence ZIP

Batch Doctor can export a four-entry ZIP containing the JSON diagnosis, a spreadsheet-safe CSV audit, a SHA-256 checksum file and a manifest. No original input bytes are packaged. The archive is reopened before download to verify ZIP CRCs, safe entry names, exact membership, checksum values and a byte-for-byte JSON-to-CSV reconstruction.

To review a later version, select the current files and import the previous evidence ZIP. The tool independently recalculates file hashes and repeats structural checks to flag changed bytes, metadata and diagnostic findings. A SHA-256 obtained separately from a trusted source can establish that the whole ZIP matches that reference; internal checksums alone do not establish authenticity or malware safety.

Export only the current files that actually changed

After comparing an earlier diagnosis with the current selection, you can explicitly export a ZIP containing only files with changed bytes, metadata or structural findings. Files that merely moved to another position are excluded. Safe numbered ZIP paths and a manifest preserve original filenames, reasons and SHA-256 values. The resulting archive is reopened to verify CRCs, checksums and source bytes; an empty change set does not create a misleading ZIP.

Privacy: unlike the evidence-only ZIP, this download contains original file contents and may include sensitive data. It is created only on request and capped at 30 MB. Use JSON or CSV comparison exports when you need to share findings without distributing source documents.

Reopen changed-file archives and recover individual originals

A changed-files ZIP is different from a diagnosis-only evidence ZIP: it includes the exact changed source bytes. Select the ZIP in Batch Doctor, optionally provide a SHA-256 saved independently, and inspect the restored file list before downloading any individual original. Each download reopens the ZIP and checks CRC, entry names, sizes and SHA-256. Unsafe source paths become safe download names, while the manifest retains the original filename for audit.

You can compare the archive against the current selection to distinguish identical bytes, missing files, changed names and changed content. This is not a repair, a virus scan or proof of who created the archive. Do not share source archives when a JSON/CSV evidence package is sufficient.

Process only files flagged by Batch Doctor

After a read-only diagnosis, choose REVIEW, REJECTED or both and process only those inputs. Already checked documents stay untouched: the browser verifies the original filename, size, MIME, signature and SHA-256 for every source before running the processing engine. A changed source requires a new diagnosis rather than an outdated recovery decision.

The resulting ZIP is reopened and its CRC, report and output hashes are checked before download. A rejected input never receives a fake repaired file; you can download accepted results individually and save a source-indexed JSON or spreadsheet-safe CSV audit. A structurally accepted file is not a malware scan or proof of full content fidelity.

Download one complete batch after selective recovery

Once Batch Doctor has processed the files marked for review or rejection, use the consolidated ZIP download to collect both the originally checked files and the newly accepted outputs in their original selection order. Each checked source is re-hashed and structurally validated before its original bytes are retained. Previously rejected files that were not selected, or that still fail, remain explicitly rejected in the ZIP report and never appear as fake repaired downloads.

The browser reopens the consolidated ZIP and checks CRC, SHA-256, output count and the full source-indexed report before releasing it. Keep the separately downloaded ZIP SHA-256 if you need to compare it later. This is a convenience for complete delivery, not a malware scan or proof that all original content was preserved.

Compare a portable source preflight before reprocessing

Before the universal batch runs, each selected source receives a SHA-256 fingerprint, detected file type and a routing warning if its extension or browser MIME differs. Export an evidence-only JSON or spreadsheet-safe CSV inventory. This inspection is a lightweight routing check, not a structural validation, antivirus scan or proof of authorship.

Later import the earlier JSON and select the current files again. The comparison distinguishes changed bytes, renamed files, new and missing sources, and files moved to another position. Choose only changed, renamed or new files to prepare a smaller batch; no source bytes are included in the comparison report. Before processing or exporting evidence, the tool rechecks the current source SHA-256 values to prevent stale selections. Keep a trusted external hash if you need to authenticate an imported report.

Choose a different conversion for each structured file

The universal queue can now verify an Office document, retain an archive and convert CSV, TSV, JSON or XML in the same batch. Select one of eight explicit data operations for an individual source; leaving it unchanged runs diagnosis and conservative verification instead. Conversions are limited to 12 files and 24 MB of input, with 8 MB per source and result. Each converted output has its own detected output type, SHA-256 and a distinct converted status rather than a misleading repair claim.

Before downloading, the converted file is parsed again, included in the ZIP with a type-aware evidence record and reopened independently. A rejected conversion produces no downloadable output. JSON-to-CSV or XML-to-JSON may change the original data model; inspect the output before relying on it. All work stays local.

Change the operation when retrying a rejected conversion

A rejected CSV-to-TSV conversion does not have to be repeated unchanged. Choose the rejected row, select a different compatible operation such as CSV-to-JSON, and retry only that file. A spreadsheet formula remains inert JSON text instead of becoming an executable spreadsheet cell. Alternatively, choose diagnosis only if you want to validate the source without converting it.

When you reopen a previous batch ZIP, supply the corrected original for the rejected row and select the replacement operation before resuming. Every retained accepted file keeps its original bytes and SHA-256. The new operation is validated against the replacement file and the conversion budget; unsupported formats or oversized inputs are rejected rather than silently processed. A retry is not a guarantee that missing source data can be reconstructed.

Extract mixed archives inside the 99-file batch

Choose Extract ZIP / TAR / GZIP to verified ZIP for a ZIP, TAR, TAR.GZ, TGZ or GZIP source. Unlike the default archive validation action, extraction reads the contained regular files, checks archive integrity and safe names, and creates a separate ZIP containing the extracted files and its JSON/CSV audit. A GZIP file with a single payload becomes one extracted file; a TAR.GZ yields its regular entries. Office packages are not offered as generic archives.

Each explicit extraction is limited to 8 MB of source data, 8 MB of verified output and 50 extracted entries; the universal queue permits up to 12 selected conversions and 24 MB combined input. A corrupt archive, unsafe path, link, incompatible signature or expansion beyond limits is rejected rather than producing a partial success. The outer ZIP is verified and can be reopened; its nested extracted ZIP is also rechecked before individual download or import. After correcting a rejected source, you can retry only that archive and preserve all previously accepted output bytes. These checks do not authenticate an archive publisher or scan for malware.

Convert JPEG, PNG and WebP within a mixed-file batch

Select a target JPEG, PNG or WebP format separately for each image in the universal queue. The browser checks the original image signature, PNG CRC or WebP container, dimensions and decoded canvas result before accepting the conversion. The generated file is reopened and its encoded signature, dimensions and container structure are verified again. Images keep their pixel dimensions, but conversion can change pixels, metadata, transparency and size; JPEG uses lossy compression and flattens transparency to white. Inputs and outputs are limited to 8 MB per selected conversion. Failed image conversions can be retried with a different target or corrected original without touching previously accepted files.

Recode text to UTF-8 without silently guessing

For TXT, Markdown and LOG files, choose strict UTF-8 recoding, explicitly interpret Windows-1252 bytes, or explicitly repair reversible UTF-8 mojibake. UTF-16 files require a byte-order mark. Invalid sequences, binary NULs, suspicious controls and results over 8 MB are rejected rather than replaced with question marks. The original line endings and decoded characters are retained unless the selected mojibake repair changes them. A Windows-1252 interpretation is a user decision, not an automatic claim that the encoding was identified correctly.

Every converted text output is reopened with strict UTF-8 decoding before ZIP packaging and checked again on import and individual download. A failed recoding can be retried with a corrected file or a different explicit action; accepted outputs remain unchanged.

Review identical content before omitting duplicate files

After the 99-file SHA-256 preflight, inspect redundant copies and case- or Unicode-equivalent filename collisions. Files are never removed automatically. Choose the unique-file action only when you want to omit redundant originals. Identical bytes with a different extension, detected type or selected conversion remain separate jobs. Download the JSON or formula-safe CSV audit first if you need a record of omitted names. SHA-256 comparison does not establish publisher authenticity.

Correct misleading file extensions in a batch without rewriting content

A valid PNG named photo.jpg or DOCX named report.doc may fail upload checks despite intact bytes. Batch diagnosis identifies the real signature. Select up to twelve reviewable files for an explicit filename correction. The tool rechecks source SHA-256 and format structure, preserves the original bytes, and changes only the download name and declared MIME. Corrupt or unverified inputs are rejected, not presented as repaired.

Download a ZIP containing corrected files, a JSON manifest and a formula-safe CSV audit. Reopen it later to recheck CRC32, SHA-256 and the format-specific download gates, then retrieve individual files. A separately stored ZIP hash can establish file identity; internal hashes alone cannot prove origin. For damaged data use File Doctor.

Download checked originals together with corrected filenames

After fixing misleading extensions, the consolidated batch keeps the originally checked files and adds the independently verified filename corrections. Every source is rechecked against its SHA-256; changed bytes stop consolidation and require a new diagnosis. Unselected and still-invalid files remain explicitly rejected in the report, without fake downloads.

You can also import an earlier correction ZIP and consolidate it with the current originals only when the source diagnosis matches. Before download, the combined ZIP is reopened to check CRC32, signatures, structure and SHA-256. Renaming an extension does not reconstruct damaged content or certify that a file is malware-free.

Recover a rejected file without losing the accepted documents

After consolidating Batch Doctor results, select a replacement only for a rejected row. If a damaged ZIP failed while a DOCX and JSON passed, supply a corrected ZIP for that row. The browser checks original fingerprints, reopens the prior evidence archive, validates the replacement format and structure, then creates a new ZIP. Accepted outputs retain their exact bytes; a replacement that still fails remains rejected. For later recovery rounds, reopen the recovered ZIP. The report records original and replacement SHA-256 separately; internal hashes establish consistency, not authorship.

Convert an image or extract an archive even when its extension is wrong

A valid PNG named picture.bin can now be explicitly converted to WebP, and a ZIP named backup.bin can be explicitly extracted. The selection follows the detected binary signature rather than trusting the extension or browser MIME. The original filename, bytes and SHA-256 remain in the report; only a temporary processing view receives a compatible suffix. Output signatures, image dimensions, archive entry safety, CRC and SHA-256 are checked again before download and after importing the batch ZIP. Unknown content is not guessed, and Office declarations cannot bypass Office-specific safety checks. Explicit conversion does not imply pixel-perfect image fidelity or publisher authenticity.

Reuse verified operation choices after reopening a batch

A portable operation plan saves each selected conversion together with the source filename, detected type, byte count and SHA-256. If you reopen the same files in a different order, the plan restores each operation to its matching source instead of applying it to the wrong row. A uniquely renamed file can also match by its unchanged bytes; changed or missing sources, ambiguous duplicates and incompatible conversions stop the import without partially applying actions.

Export the JSON plan before leaving the page, then reselect your original files and import it. No document contents are stored in the plan, and the browser rehashes every file before applying it. Selecting only changed files, removing unsupported entries or deduplicating a batch retains the remaining files' conversion choices. A matching SHA-256 confirms byte identity, not authorship or freedom from malware. For malformed files, see File Doctor.

Export selectable PDF text without guessing scanned pages

Choose PDF → UTF-8 selectable text (no OCR) for a digitally generated PDF. The browser reads up to 50 pages, preserves their order with explicit page markers and checks every page contains extractable text. A scanned, blank, encrypted or unreadable page blocks the conversion instead of delivering an incomplete transcript. The 8 MB input and output limits apply, and this operation does not invoke OCR or download language packs. PDF text extraction is not a faithful layout or table conversion: inspect the resulting TXT before using it.

Reuse saved operations in a changed selection

Strict import still requires every original to match. The separate Merge matching saved operations control restores only SHA-256-matched files and preserves existing choices for newly added files. Unique renames can match by content; changed originals with the same filename and ambiguous duplicates block the merge without applying any choices. Saved “verify only” choices explicitly clear a previous conversion. This is a local planning convenience, not a digital signature.

Export readable Office text without pretending to preserve the document

Choose Office DOCX/XLSX/PPTX → UTF-8 readable text for a genuine Word document, Excel workbook or PowerPoint presentation. The browser checks the OOXML ZIP structure and external/active references before reading content. Word paragraphs, worksheet cell coordinates and slide text are exported in their original document order with section labels. A workbook containing formulas is refused because cached formula results could be outdated; tracked Word changes, unhandled headers and presentation speaker notes are also refused rather than silently omitted.

The TXT output deliberately does not preserve fonts, layout, images or formulas. Each output is rechecked for UTF-8 validity, ordered sections and SHA-256 before ZIP packaging, individual download and ZIP reimport. A file renamed with a misleading extension is accepted only if its Office signature is genuine. This local extraction is not a malware scan or a substitute for opening the original in Office. For structural damage, use File Doctor.

Extract text from OpenDocument ODT and ODP safely

Choose ODT/ODP → UTF-8 readable text for a genuine OpenDocument text document or presentation. The browser verifies every ZIP component, its CRC32 and the internal MIME type before reading the document body. Paragraphs remain in order; presentation slides have numbered section markers and speaker notes are identified when present. A renamed ODF file can be recognized by its real MIME signature, but damaged archives, macros, unsafe links and invalid XML are rejected rather than converted optimistically.

The resulting TXT preserves readable Unicode but deliberately omits layout, images, charts and formatting. Empty scanned or image-only documents cannot be OCRed by this operation. Reopen the verified ZIP to compare output hashes and sections, or use File Doctor to investigate a rejected original.