Repository navigation
feat(export): analysis files in the study ZIP (JSONL + CSV for AI-assisted analysis) - #101
Merged
Merged
Conversation
interviews.jsonl (exact text, the quotation source), interviews.csv, profile_fields.csv and per-interview turn CSVs, plus a README that repeats the Markdown export's consent/provider warning from the shared constant. CSVs: UTF-8 BOM, CRLF, every cell through csvCell. Turn indexes match the analysis citation numbering. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
…orts The ZIP writer accepts chunked entries with incremental CRC and size checks, so the Worker streams the combined files instead of holding rows; turns are sharded per interview (a single table's bound is ~830 MiB at the 500-interview cap). The Durable path rereads the same export sequence and still verifies it before the closing records. Per-interview Markdown dates are now ISO UTC. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The formula guard prefixed an apostrophe to negative numbers (offset_s can be negative), so spreadsheets read them as text. A finite number cannot carry a formula; text cells are still guarded. From the Kimi review. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
… files The ST-08 journey counted every .md in the archive; analysis/README.md is now one more. Assert the analysis entries exist instead, with one turn shard and one JSONL line per interview, through the built Worker. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
linxule
force-pushed
the
feat/analysis-export
branch
from
October 10, 2026 13:29
46a5f67 to
18a94ed
Compare
menotbobbybrown
pushed a commit
to menotbobbybrown/openinterviewer
that referenced
this pull request
Oct 11, 2026
On-device voice input (linxule#100), analysis files in the study export (linxule#101), OpenNext 1.20.10 / wrangler 4.148.0 with Next held at 16.3 (linxule#98), provider SDKs (linxule#95). Code-only release; rollback to 5.3 refuses device-mode studies. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
For researchers who analyse exports with Claude Code or another AI tool, or in a spreadsheet (Yinghua's use case). Every study ZIP, single-study and all-studies, on Node and Cloudflare, now also contains
analysis/:interviews.jsonl: one object per interview (openinterviewer.analysis.v1) with study context from the interview's own collection snapshot, consent/provider provenance, analysis, profile fields with their historical labels, and turns with exact text. This is the file to quote from.interviews.csv: one row per interview; oneprofile:<fieldId>column per field seen.profile_fields.csv: long form, safe across revisions.turns/NNN.csv: one per interview. A single turns table could reach ~830 MiB at the 500-interview cap, too much for a Worker.README.md: a data dictionary, plus the Markdown export's consent/provider warning from the same constant.Details
csvCell. Numbers stay numbers, a fix from the Kimi review.Cloudflare path. The ZIP writer takes chunked entries, with incremental CRC and size checks. The Worker streams the combined files by re-walking the same export snapshot; every page read re-checks row fingerprints, and
verifyExportSequencestill runs before the closing records. The cost is about 5× snapshot reads per export, bounded.Not included: no adapter for third-party pipelines. shipaleks/ux-research-pipeline is "all rights reserved"; its documented input is a per-interview JSON array that a researcher can derive from
interviews.jsonl.Reviews: Kimi approved, no blockers. Accepted: numbers in CSV. Noted, not changed: the 5× read cost, and the replay identity check relying on the Durable page fingerprints.
npm run check:cloudflare: 20/20 lanes on a clean tree at the head commit, including the artifact journey that now asserts the analysis entries through the built Worker.🤖 Generated with Claude Code