Add Postman Collection import command

[vedssharma]

Adds apicli import postman <file> to load a Postman Collection v2.0/v2.1 JSON file and save all its requests as an apicli collection.

Key details:

• Handles both URL formats: plain string and structured object
• Recursively flattens nested folders using "Folder / Request" prefixes
• Appends to an existing collection if the name already exists
• --collection flag overrides the Postman collection name as the target
• Skips non-raw body modes (form-data, urlencoded) gracefully

https://claude.ai/code/session_01PyL8EW4HoiuaCszqviyfGt

[claude]

Code Review: Add Postman Collection import command

Overall this is a solid implementation. The polymorphic URL handling with json.RawMessage is exactly right for Postman's v2.0/v2.1 compatibility requirement, and the recursive folder flattening is clean and readable. A few issues worth addressing:

---

Bug: Disabled headers are imported as active

postmanHeader has no disabled field, so headers marked disabled in Postman will be silently imported and sent on every request.

// Current
type postmanHeader struct {
    Key   string `json:"key"`
    Value string `json:"value"`
}

// Should be
type postmanHeader struct {
    Key      string `json:"key"`
    Value    string `json:"value"`
    Disabled bool   `json:"disabled"`
}

// And in convertPostmanRequest:
for _, h := range pr.Header {
    if h.Key != "" && !h.Disabled {
        headers[h.Key] = h.Value
    }
}

---

Performance: One transaction per request

runImportPostman calls store.AddToCollection() in a loop, and each call opens its own transaction. For large Postman collections this will be noticeably slow. Consider batching the inserts into a single transaction, or adding a AddManyToCollection(name string, reqs []model.SavedRequest) error method on the storage layer.

---

Minor: Query params dropped in URL reconstruction fallback

In resolvePostmanURL, when falling back to reconstructing the URL from parts, query params are not included:

// port and query params are both missing from the fallback
return fmt.Sprintf("%s://%s%s", scheme, host, path)

In practice Postman always populates raw, so this is a low-risk edge case, but worth a // Note: query params and port not reconstructed comment so future readers don't wonder.

---

Minor: Misleading #nosec comment

data, err := os.ReadFile(filePath) // #nosec G304 – path comes from CLI argument, not user-controlled input

CLI arguments are user-controlled; the suppression is valid, but the rationale should be "intentional user-supplied path" rather than "not user-controlled."

---

No tests

The repo has no test files at all, so this isn't a regression, but flattenPostmanItems and resolvePostmanURL are pure functions that are easy to unit-test and cover important edge cases (deeply nested folders, string vs object URL, missing URL, disabled headers). These would be good candidates for the project's first _test.go.

---

Positive notes

• The json.RawMessage approach for the polymorphic URL field is the right call and avoids a custom UnmarshalJSON.
• Appending to an existing collection rather than overwriting is the right UX default.
• Partial-import error handling (continue + count) is consistent with the rest of the codebase.
• The --collection flag to override the target name is a useful escape hatch.

🤖 Generated with Claude Code