Neatbo.

Recorded file guide: subtitles, EML, HAR and byte comparisons

Check subtitle reading rules, recover recorded email attachments or HAR bodies, and compare ZIP payloads or binary offsets. Inspect complete downloads before delivery.

Updated 5 min read

Check subtitle cues against explicit rules

Open SRT or supported WebVTT and set the reading, duration, line, overlap and gap rules for your delivery. Visible Unicode code points include spaces; supported format tags and line breaks do not contribute to the count. Unsupported markup is rejected. Review cue identifiers and issue codes in the complete JSON and CSV. These checks do not certify translation quality or a publisher’s approval.

Choose the stored layer
InputQuestionEvidence to inspect
EMLWhich attachments were saved?Decoded payload ZIP and MIME filename map
HARWas the response body captured?Recovered body or explicit omission
Two ZIP filesWhich expanded documents changed?Full paths and actual-content hashes
Two binary filesWhich same offsets differ?Both hex values and missing tails
SRT / WebVTTWhich visible cues exceed the rules?Cue identifiers, durations and issue codes

Recover attachments from complete local messages

Select up to 500 small EML files within the total byte limit. Attachment decoding follows supported MIME transfer encodings and filename declarations. The map preserves the source message, MIME part, original name, safe ZIP name and byte count. Open the ZIP and compare the recovered bytes with the original MIME payload.

A malformed or unsupported selected attachment fails the whole batch. No-attachment messages produce a complete zero-attachment report. Embedded messages are not recursively unpacked as another mailbox. Keep source messages if you need a different recovery scope.

Distinguish recorded HTTP bodies from missing responses

Open a local HAR. Base64 content recovers exact captured bytes; a text body becomes UTF-8 and is labelled as that conversion. Missing body fields and 304 responses are reported as omitted. URLs are provenance, and the tool does not request them.

Use the JSON map to identify the recovered numbered files, byte counts and omitted entries. A zero-byte captured body can be a valid file. A request URL or content size alone cannot reconstruct a response that was never saved.

Compare the correct file layer

ZIP comparison expands both supported archives, validates entry integrity and compares full case-sensitive file paths and actual bytes. Explicit directory entries and packaging order, timestamps or compression are outside this content comparison. Check added, removed and changed paths in the complete report.

Binary comparison aligns the same zero-based offset in both files. It records both hexadecimal byte values, including a missing side after a shorter file ends. It does not search for insertions, decode a file format or realign content. Use the full CSV or JSON when the on-page preview is bounded.

Inspect the handoff

  • Keep the original and the complete JSON report.
  • Use filename maps to connect safe output names to their exact source.
  • Reopen recovered payloads with the intended reader.
  • Review omitted entries and unsupported input errors before claiming recovery is complete.
  • Select text or archive comparison according to the layer the recipient needs.

Recover actual Office notes with their context

PPTX speaker notes are linked to individual slides and differ from the visible slide text. Select a batch of ordinary presentations; the notes export preserves slide order and empty positions. Hidden slide status and source names remain in the index. Header, footer, date, slide-image and page-number placeholders are outside the notes text.

Word comments need the body selection they discuss. The export records classic main-document anchors across runs and paragraphs with comment text, author and stored date. Text offsets describe logical UTF-16 text, not rendered page positions. Modern threads, revisions and unsupported anchoring fail explicitly rather than producing an incomplete review list.

Keep displayed footnote numbers distinct from internal IDs

For the supported continuous decimal numbering, footnotes and endnotes receive their visible numbers separately in body reference order. Internal XML IDs are diagnostic values and can be out of numeric order. The report retains the referring paragraph beside the note so the recipient can locate it.

Section restarts, custom marks and other numbering formats require a different workflow and are rejected here. Retain the DOCX if those constructs are present; a plausible sequential list would not establish its original displayed numbering.

Recover only embedded source text

A Source Map can name a source without storing its content. Recovery uses actual sourcesContent strings, including an empty string, and reports null or missing content as unavailable. It produces UTF-8 files in a safe ZIP with a source-index and original-path map.

Ordinary v3 maps and supported explicit inline data URIs are accepted. Source URLs and sourceRoot do not trigger requests. The tool does not execute the original or generated code, and mappings alone cannot reconstruct absent source text. Indexed sections and ambiguous directives are rejected.

References

Tools in this category

Expand a tool to see its steps, options and supported formats, then open its workspace.

Subtitle quality reportFind fast, crowded or overlapping subtitle cues using your own readable-text rules.

Review SRT or WebVTT cue durations, Unicode codepoint lengths, reading speeds and timing gaps. Supported formatting is excluded from the counts. Download the complete per-cue JSON and CSV report without changing the subtitles.

Steps

  1. Choose an SRT or WebVTT file, or paste its text.
  2. Set your reading speed, line, duration, gap and overlap rules.
  3. Review cue IDs, codepoint counts and issue codes; download the complete reports. The table previews up to 200 cues.

Available options

Maximum codepoints per second
20
Maximum codepoints per line
42
Maximum lines per cue
2
Minimum cue duration (seconds)
0.5
Maximum cue duration (seconds)
7
Minimum gap (milliseconds)
80
Overlap tolerance (milliseconds)
0

Capabilities and limits

  • One UTF-8 subtitle file up to 10 MiB or pasted text up to 1 MiB; up to 10,000 cues and 100,000 overlap pairs. Counts include spaces and exclude formatting and line breaks; emoji and Chinese characters count by Unicode codepoint, rather than grapheme or word.
  • Paired b/i/u tags, SRT font color tags and WebVTT c.class tags are supported. Common amp/lt/gt/nbsp/lrm/rlm and valid numeric entities are decoded for counting. Unknown or mismatched markup, cue settings, NOTE, STYLE and REGION are rejected. HTML is never rendered. Rules are your own thresholds, not platform certification.
Open Subtitle quality report →
Extract EML attachmentsRecover original MIME attachment bytes from up to 500 local EML files with a complete filename map.

Choose a batch of saved email files and recover their attachments without opening an inbox. Nested MIME, inline files and supported encodings are parsed locally. A ZIP and reports map every recovered file to its source email and MIME part.

Steps

  1. Choose your saved .eml files as one bounded batch.
  2. Extract and review mail number, MIME part, original filename and safe ZIP filename.
  3. Download the ZIP containing all attachments and the complete mapping reports. Keep the original emails.

Capabilities and limits

  • Up to 500 EML files and 30 MiB total input; up to 1,000 attachments, 10 MiB per decoded attachment and 30 MiB total decoded payload. MIME nesting is limited to 20 levels and 5,000 total parts.
  • Supports Base64, quoted-printable, binary, 7bit and 8bit payloads; filename/name parameters, RFC2231 continuations and UTF-8/ASCII/ISO-8859-1 encoded filenames. TNEF and attached message/rfc822 recursion are unsupported. A malformed selected attachment fails the whole batch: no incomplete success is shown. Email HTML is never rendered or executed; no mailbox connection occurs.
Open Extract EML attachments →
Recover HAR response filesRecover response bodies already captured in a local HAR file, with a clear report of missing content.

Turn captured HAR response bodies into local files and a ZIP. Base64 payloads recover their exact recorded bytes; ordinary response text is saved as UTF-8. Missing bodies and bodyless statuses stay visible in the report.

Steps

  1. Choose a HAR export that includes response bodies, or paste its JSON.
  2. Review recovered and omitted entries, byte provenance and safe filenames.
  3. Download the resource ZIP and complete reports. Duplicate URLs remain separate entries.

Capabilities and limits

  • One UTF-8 HAR file up to 10 MiB or pasted JSON up to 1 MiB; up to 5,000 entries, 2,000 recovered bodies, 10 MiB per body and 30 MiB total recovered bytes. Duplicate JSON keys and corrupt declared Base64 are rejected.
  • Only content.text already in the recording is used. No URL is fetched. 204/304, missing content.text and unsupported encodings are reported as omitted. Plain text is explicitly UTF-8 reserialization and may differ from original wire bytes. Report URLs omit credentials, query and fragment; headers and cookies are excluded. Reports are still private local artifacts.
Open Recover HAR response files →
Compare ZIP contentsCompare actual expanded files inside two ZIP archives while ignoring packaging timestamps, order and compression.

Find added, removed or changed paths between two ZIP versions. Every supported file is decompressed and CRC-checked, then compared by its actual bytes with SHA-256 recorded in the report. Packaging-only differences do not count as file changes.

Steps

  1. Choose the two ZIP versions in left-to-right order.
  2. Compare and inspect added, removed, changed and unchanged path counts.
  3. Download JSON/CSV with full paths, expanded sizes and SHA-256 values. No files are extracted onto your filesystem automatically.

Capabilities and limits

  • Exactly two ZIP files, up to 50 MiB each and 100 MiB total; up to 1,000 entries and 50 MiB expanded data per archive, with a 100:1 maximum expansion ratio. Store/Deflate and UTF-8 or ASCII paths are supported.
  • Encrypted ZIP, ZIP64, symlinks, unsafe paths and duplicate NFC paths are rejected. Paths are case-sensitive; explicit directory entries are ignored while empty regular files are compared. Preview shows at most 200 entries; JSON/CSV include the complete comparison. CSV formula-looking strings are prefixed for spreadsheet safety.
Open Compare ZIP contents →
Binary byte differencesLocate exact differing offsets and both hexadecimal byte values in two local binary files.

Compare arbitrary bytes at the same offsets, including empty files and missing tail bytes. The complete JSON and CSV list every difference; the on-screen table previews the first 200. Inputs are never decoded as text.

Steps

  1. Choose the left and right binary files in order.
  2. Compare same-offset bytes and inspect hexadecimal offsets and the two values.
  3. Download complete JSON/CSV to review all differences. Reduce the files if the complete report exceeds its limit.

Capabilities and limits

  • Exactly two files up to 10 MiB each and 20 MiB total; at most 100,000 differences and 20 MiB per complete report. A larger difference report is rejected rather than silently truncated.
  • Comparison uses equal offsets. It does not realign inserted or deleted blocks. Offsets are zero-based hexadecimal; a missing side is null in JSON and missing in CSV. Both empty files compare equal.
Open Binary byte differences →
Export PPTX speaker notesExport the actual speaker notes from up to 100 local PPTX presentations, preserving every slide position.

Recover a script from presentation notes in slide order. Slides with no notes or empty notes remain in the index, and hidden slides are identified. Notes stay plain text; each presentation gets TXT and Markdown files in a ZIP.

Steps

  1. Choose the local PPTX files.
  2. Export and inspect slide order, notes status and hidden-slide flags.
  3. Download speaker-notes.zip plus full JSON/CSV indexes.

Capabilities and limits

  • Up to 100 PPTX files, 10 MiB per file and 30 MiB total, 5,000 slides and 64 MiB expanded contents across the batch; 20 MiB notes text and bounded reports.
  • Ordinary unsigned, unencrypted PPTX only. Old PPT, macros, broken/external notes or slide references and unsupported notes text constructs fail the entire batch. Slide number, header, footer, date and slide-image placeholders are excluded; the slide body is not used as notes.
Open Export PPTX speaker notes →
Export Word comments with contextExport classic DOCX review comments with their exact selected body text, author and paragraph context.

Give reviewers a comment list that includes the original passage. Classic comment ranges are reconstructed across runs and paragraphs, including overlapping comments. The original document is preserved; outputs are text, JSON and spreadsheet-safe CSV.

Steps

  1. Choose the local DOCX.
  2. Review each comment's author, selected text and paragraph context.
  3. Download the full JSON, CSV and text reports.

Capabilities and limits

  • One DOCX up to 20 MiB, 64 MiB expanded content and 5,000 comments; output comment/context text is limited to 8 MiB and reports to 20 MiB each.
  • Classic main-document comments, including tables and point comments. Modern threads, tracked changes, fields, text boxes and unsupported anchoring fail explicitly. Paragraph/offset coordinates describe logical text in UTF-16 code units, with paragraph separators; they do not identify rendered page positions.
Open Export Word comments with context →
Export Word footnote reference indexExport DOCX footnotes and endnotes with their actual continuous decimal reference numbers and body context.

Match each note to the body passage that refers to it. Supported automatic decimal numbering is calculated from the document's real settings and reference order, independently for footnotes and endnotes. Internal IDs remain separate diagnostic fields.

Steps

  1. Choose your DOCX with supported automatic note numbering.
  2. Check visible number, note kind, referring paragraph and note text.
  3. Download the complete reference index as JSON, CSV and text.

Capabilities and limits

  • One DOCX up to 20 MiB, 64 MiB expanded content and 10,000 notes; combined exported note/context text up to 8 MiB, each report up to 20 MiB.
  • Only continuous automatic decimal numbering starting at 1, including absent OOXML defaults or equivalent explicit settings. Section/page restart, other formats such as Roman endnotes, custom marks, reused note references, fields, revisions or unsupported stories are rejected. Internal IDs are not displayed note numbers. Offsets use logical UTF-16 text.
Open Export Word footnote reference index →
Recover embedded Source Map sourcesRecover actual sourcesContent text from a local v3 source map or an explicit inline data URI, with a complete availability index.

Restore source files that were embedded in a build's source map. Every source index stays distinct, including duplicate names and empty files. Null or missing sourcesContent is reported as unavailable; mappings are never used to invent source code.

Steps

  1. Paste a v3 map or choose a local map/JS/CSS file.
  2. Recover and inspect which sources are present, null or unavailable.
  3. Download the available-source ZIP and full mapping reports.

Capabilities and limits

  • One input up to 5 MiB, pasted text up to 1 MiB, 2,000 source records and 30 MiB recovered UTF-8 text. Reports are bounded to 20 MiB each.
  • Ordinary source-map v3 with sources, names and mappings; indexed sections are unsupported. JS/CSS input must end with one explicit inline application/json sourceMappingURL data URI. No code execution or remote requests. Recovered text is encoded as UTF-8, without claiming the original file's wire encoding. Safe ZIP names include source indices; source URLs have credentials/query/fragment removed from reports.
Open Recover embedded Source Map sources →

Tools used in this article

Subtitle quality report →Find fast, crowded or overlapping subtitle cues using your own readable-text rules.Extract EML attachments →Recover original MIME attachment bytes from up to 500 local EML files with a complete filename map.Recover HAR response files →Recover response bodies already captured in a local HAR file, with a clear report of missing content.Compare ZIP contents →Compare actual expanded files inside two ZIP archives while ignoring packaging timestamps, order and compression.Binary byte differences →Locate exact differing offsets and both hexadecimal byte values in two local binary files.Export PPTX speaker notes →Export the actual speaker notes from up to 100 local PPTX presentations, preserving every slide position.Export Word comments with context →Export classic DOCX review comments with their exact selected body text, author and paragraph context.Export Word footnote reference index →Export DOCX footnotes and endnotes with their actual continuous decimal reference numbers and body context.Recover embedded Source Map sources →Recover actual sourcesContent text from a local v3 source map or an explicit inline data URI, with a complete availability index.