# Archive Import and Knowledge City Planning Archive import is a staged, auditable operation. Uploading an archive never writes Pages directly. ## Pipeline 1. Store the uploaded ZIP or tar-family archive in private import storage. 2. Queue `import.analyze`; foreground requests always take scheduler priority. 3. Validate entry names, link types, file count and expanded byte limits. 4. Produce an immutable dry-run manifest containing source path, destination Path, detected content type and initial Quartier/Street/Spot suggestions. 5. Allow a curator to edit, merge, exclude and approve suggestions. 6. Queue import execution in bounded batches. Markdown is transpiled to the canonical HTML Document while its source representation remains available. 7. Record created UUIDs and topology operations in an import journal so a session can be reviewed, retried or rolled back. ## Safety limits - archive upload: 512 MiB; - expanded content: 2 GiB; - entries: 10,000; - absolute paths, `..`, NUL, hard links and symbolic links are rejected; - supported staging formats: ZIP, tar, tar.gz/tgz and tar.xz/txz. ## Initial suggestion rules The first directory component proposes a Quartier, the second proposes a Street, and the nearest containing directory proposes a Spot. Markdown receives higher initial confidence than opaque assets. Later scoring incorporates front matter, headings, README/index portals, internal links, vocabulary and existing City boundaries. Suggestions never mutate City topology until a curator confirms the plan.