dedup summary log silently omits the fuzzy-merge count when exact_merges == 0

[thejesh23]

Summary

The dedup summary line silently drops the fuzzy-merge count when the run has zero exact merges.

Where

graphify/dedup.py:419-425:

msg = f"[graphify] Deduplicated {total} node(s)"
if exact_merges:
    msg += f" ({exact_merges} exact"
    if fuzzy_merges:
        msg += f", {fuzzy_merges} fuzzy"
    msg += ")"
print(msg + ".", flush=True)

The if fuzzy_merges branch is nested inside if exact_merges, so when exact_merges == 0 and fuzzy_merges > 0 the fuzzy count is never printed. The message reads Deduplicated N node(s). with no breakdown, even though there were N fuzzy merges. Both counters are collected correctly upstream (see lines 263 / 282 for exact, 300 / 394 for fuzzy) — only the summary line is wrong.

Reproduction

Any run where Pass 1 (same-source_file, exact-normalized-label) finds nothing but Pass 2 (Jaro-Winkler over normalized labels across files) does — typical for doc-heavy or semantic-only corpora. Result:

• Actual: [graphify] Deduplicated 12 node(s).
• Expected: [graphify] Deduplicated 12 node(s) (12 fuzzy).

The reverse case (only exact merges) prints correctly.

Suggested fix

Build the breakdown as a list and only wrap in parens when at least one component is present:

msg = f"[graphify] Deduplicated {total} node(s)"
parts: list[str] = []
if exact_merges:
    parts.append(f"{exact_merges} exact")
if fuzzy_merges:
    parts.append(f"{fuzzy_merges} fuzzy")
if parts:
    msg += f" ({', '.join(parts)})"
print(msg + ".", flush=True)

Happy to open a PR.

[thejesh23]

PR up: #1860.