ci: summarize TechAPI PR data validation

Author: Seungpyo1007

.github/workflows/techapi-pr-validation-comment.yml

@@ -51,6 +51,13 @@ jobs:
           ref: ${{ env.TECHAPI_HEAD_SHA }}
           path: TechAPI
 
+      - name: Checkout TechAPI main
+        uses: actions/checkout@v4
+        with:
+          repository: GetTechAPI/TechAPI
+          ref: main
+          path: TechAPI-main
+
       - uses: actions/setup-python@v5
         with:
           python-version: "3.12"
@@ -90,6 +97,199 @@ jobs:
           echo "app_status=${app_status:-1}" >> "$GITHUB_OUTPUT"
           echo "integrity_status=${integrity_status:-1}" >> "$GITHUB_OUTPUT"
 
+      - name: Build data quality summary
+        shell: bash
+        run: |
+          python - <<'PY'
+          from __future__ import annotations
+
+          import hashlib
+          import json
+          import re
+          from collections import Counter, defaultdict
+          from pathlib import Path
+          from typing import Any
+
+          HEAD = Path("TechAPI/data")
+          BASE = Path("TechAPI-main/data")
+          CATEGORIES = ("brand", "soc", "smartphone", "gpu", "cpu")
+          MAX_WARNINGS = 20
+
+          def load_json(path: Path) -> dict[str, Any]:
+              return json.loads(path.read_text(encoding="utf-8-sig"))
+
+          def rel_jsons(root: Path, category: str) -> dict[str, Path]:
+              base = root / category
+              if not base.exists():
+                  return {}
+              return {
+                  str(path.relative_to(root)).replace("\\", "/"): path
+                  for path in sorted(base.rglob("*.json"))
+              }
+
+          def digest(path: Path) -> str:
+              return hashlib.sha256(path.read_bytes()).hexdigest()
+
+          def verified_value(record: dict[str, Any]) -> bool | None:
+              value = record.get("verified")
+              return value if isinstance(value, bool) else None
+
+          def has_kaggle_source(record: dict[str, Any]) -> bool:
+              return any(
+                  isinstance(url, str) and "kaggle.com" in url.lower()
+                  for url in record.get("source_urls", [])
+              )
+
+          def name_warnings(category: str, rel: str, record: dict[str, Any]) -> list[str]:
+              warnings: list[str] = []
+              name = record.get("name")
+              if not isinstance(name, str):
+                  return warnings
+              if name != name.strip():
+                  warnings.append("leading/trailing whitespace in name")
+              if "  " in name:
+                  warnings.append("double spaces in name")
+              if "\ufffd" in name:
+                  warnings.append("replacement character in name")
+              if name.count("(") != name.count(")"):
+                  warnings.append("unbalanced parentheses in name")
+              words = re.findall(r"[A-Za-z0-9]+", name.lower())
+              if any(a == b and len(a) > 1 for a, b in zip(words, words[1:])):
+                  warnings.append("repeated adjacent word in name")
+              if re.search(r"\b(unknown|unk|n/a|tbd|null)\b", name, re.I):
+                  warnings.append("placeholder-like token in name")
+              return [f"{category}: {rel}: {warning}" for warning in warnings]
+
+          def value_warnings(category: str, rel: str, record: dict[str, Any]) -> list[str]:
+              warnings: list[str] = []
+              if category == "cpu":
+                  cores = record.get("cores")
+                  threads = record.get("threads")
+                  base = record.get("base_clock_ghz")
+                  boost = record.get("boost_clock_ghz")
+                  if isinstance(cores, int) and isinstance(threads, int) and threads < cores:
+                      warnings.append("threads < cores")
+                  if (
+                      isinstance(base, (int, float))
+                      and isinstance(boost, (int, float))
+                      and boost > 0
+                      and base > 0
+                      and boost < base
+                  ):
+                      warnings.append("boost clock below base clock")
+                  arch = record.get("architecture")
+                  if isinstance(arch, str) and re.search(r"\b(unknown|n/a|tbd|null)\b", arch, re.I):
+                      warnings.append("placeholder-like architecture")
+              if category == "gpu":
+                  base = record.get("base_clock_mhz")
+                  boost = record.get("boost_clock_mhz")
+                  if (
+                      isinstance(base, (int, float))
+                      and isinstance(boost, (int, float))
+                      and boost > 0
+                      and base > 0
+                      and boost < base
+                  ):
+                      warnings.append("boost clock below base clock")
+              return [f"{category}: {rel}: {warning}" for warning in warnings]
+
+          lines: list[str] = []
+          lines.append("## Data summary")
+          lines.append("")
+          lines.append("| Category | Total | Verified | Unverified | Verified % |")
+          lines.append("| --- | ---: | ---: | ---: | ---: |")
+
+          total_all = verified_all = unverified_all = 0
+          by_category: dict[str, dict[str, int]] = {}
+          for category in CATEGORIES:
+              paths = rel_jsons(HEAD, category)
+              verified = unverified = 0
+              for path in paths.values():
+                  value = verified_value(load_json(path))
+                  if value is True:
+                      verified += 1
+                  elif value is False:
+                      unverified += 1
+              total = len(paths)
+              pct = f"{(verified / total * 100):.1f}%" if total else "0.0%"
+              by_category[category] = {
+                  "total": total,
+                  "verified": verified,
+                  "unverified": unverified,
+              }
+              total_all += total
+              verified_all += verified
+              unverified_all += unverified
+              lines.append(f"| {category} | {total} | {verified} | {unverified} | {pct} |")
+          pct_all = f"{(verified_all / total_all * 100):.1f}%" if total_all else "0.0%"
+          lines.append(f"| **all** | **{total_all}** | **{verified_all}** | **{unverified_all}** | **{pct_all}** |")
+
+          lines.append("")
+          lines.append("## PR data delta")
+          lines.append("")
+          lines.append("| Category | Added | Modified | Deleted | Added verified | Added unverified | Added Kaggle-sourced |")
+          lines.append("| --- | ---: | ---: | ---: | ---: | ---: | ---: |")
+
+          all_added: list[tuple[str, str, Path]] = []
+          for category in CATEGORIES:
+              head = rel_jsons(HEAD, category)
+              base = rel_jsons(BASE, category)
+              added_keys = sorted(set(head) - set(base))
+              deleted_keys = sorted(set(base) - set(head))
+              modified_keys = sorted(
+                  key for key in set(head) & set(base) if digest(head[key]) != digest(base[key])
+              )
+              added_verified = added_unverified = added_kaggle = 0
+              for key in added_keys:
+                  record = load_json(head[key])
+                  all_added.append((category, key, head[key]))
+                  if verified_value(record) is True:
+                      added_verified += 1
+                  elif verified_value(record) is False:
+                      added_unverified += 1
+                  if has_kaggle_source(record):
+                      added_kaggle += 1
+              lines.append(
+                  f"| {category} | {len(added_keys)} | {len(modified_keys)} | {len(deleted_keys)} | "
+                  f"{added_verified} | {added_unverified} | {added_kaggle} |"
+              )
+
+          lines.append("")
+          lines.append("## Heuristic review")
+          lines.append("")
+          warnings: list[str] = []
+          manufacturer_counter: Counter[str] = Counter()
+          source_counter: Counter[str] = Counter()
+          for category, rel, path in all_added:
+              record = load_json(path)
+              manufacturer = record.get("manufacturer") or record.get("brand")
+              if isinstance(manufacturer, str):
+                  manufacturer_counter[manufacturer] += 1
+              if has_kaggle_source(record):
+                  source_counter["kaggle"] += 1
+              else:
+                  source_counter["other"] += 1
+              warnings.extend(name_warnings(category, rel, record))
+              warnings.extend(value_warnings(category, rel, record))
+
+          if manufacturer_counter:
+              top = ", ".join(f"{name}: {count}" for name, count in manufacturer_counter.most_common(8))
+              lines.append(f"- Added records by manufacturer/brand: {top}")
+          if source_counter:
+              top = ", ".join(f"{name}: {count}" for name, count in source_counter.most_common())
+              lines.append(f"- Added records by source class: {top}")
+
+          if warnings:
+              lines.append(f"- Heuristic warnings: {len(warnings)} total; showing first {min(MAX_WARNINGS, len(warnings))}.")
+              lines.append("")
+              for warning in warnings[:MAX_WARNINGS]:
+                  lines.append(f"  - {warning}")
+          else:
+              lines.append("- Heuristic warnings: none found.")
+
+          Path("quality-summary.md").write_text("\n".join(lines) + "\n", encoding="utf-8")
+          PY
+
       - name: Build PR comment
         shell: bash
         run: |
@@ -114,6 +314,8 @@ jobs:
             echo "| \`python -m app.validate\` | $([ "${{ steps.validate.outputs.app_status }}" = "0" ] && echo PASS || echo FAIL) |"
             echo "| \`python integrity_check.py TechAPI/data --strict\` | $([ "${{ steps.validate.outputs.integrity_status }}" = "0" ] && echo PASS || echo FAIL) |"
             echo
+            cat quality-summary.md
+            echo
             echo "<details><summary>Validation log</summary>"
             echo
             echo '```text'