scripts/complexity_estimator.py
7.1 KB| 1 | #!/usr/bin/env python3 |
| 2 | """complexity_estimator.py — Recommend full-4-section vs compressed output. |
| 3 | |
| 4 | Stdlib-only. Counts non-empty items in a dump, detects clustering signal |
| 5 | (repeated keywords across items), and recommends one of: |
| 6 | |
| 7 | format=full → use the full Projects/Tasks/Connections/How-I-Can-Help |
| 8 | 4-section format (8+ items AND clustering signal) |
| 9 | format=compressed → use the compressed What-I-heard / How-I-can-help |
| 10 | format (≤5 items OR no clustering signal) |
| 11 | |
| 12 | The recommendation is HEURISTIC. The capture skill applies judgment on top |
| 13 | based on full dump context. Use this as the seed. |
| 14 | |
| 15 | NO LLM CALLS. Pure word counting + frequency analysis. |
| 16 | |
| 17 | Usage: |
| 18 | python complexity_estimator.py path/to/dump.txt |
| 19 | python complexity_estimator.py path/to/dump.txt --output json |
| 20 | python complexity_estimator.py --sample |
| 21 | """ |
| 22 | |
| 23 | import argparse |
| 24 | import json |
| 25 | import re |
| 26 | import sys |
| 27 | from collections import Counter |
| 28 | from pathlib import Path |
| 29 | from typing import Any, Dict, List |
| 30 | |
| 31 | |
| 32 | SAMPLE_DUMP_LARGE = """Ok dump time. |
| 33 | Q3 launch needs pricing nailed down. |
| 34 | Draft the Q3 launch email. |
| 35 | Brief Sarah on the Q3 marketing angle. |
| 36 | Decide: launch July 15 or August 1? |
| 37 | Ferret app idea keeps nagging me. |
| 38 | Should I talk to my cofounder about ferret app? |
| 39 | Sketch the ferret matching algorithm if serious. |
| 40 | Decide: ferret app serious or shelf? |
| 41 | Fix the auth bug. |
| 42 | Rewrite the login form (ugly). |
| 43 | Write tests for auth + login. |
| 44 | Add 2fa module to auth. |
| 45 | Do my Q3 OKRs before launch. |
| 46 | """ |
| 47 | |
| 48 | SAMPLE_DUMP_SMALL = """Email Sarah. |
| 49 | Fix the flaky test. |
| 50 | Decide: Postgres or Mongo for new service. |
| 51 | Dentist appointment. |
| 52 | Finish reading the RAG article. |
| 53 | """ |
| 54 | |
| 55 | |
| 56 | # Stop-words to exclude from clustering detection |
| 57 | STOP_WORDS = { |
| 58 | "a", "an", "the", "is", "are", "was", "were", "be", "been", "being", |
| 59 | "to", "of", "in", "on", "at", "for", "with", "by", "from", "up", "down", |
| 60 | "and", "or", "but", "if", "then", "else", "so", "as", |
| 61 | "i", "you", "he", "she", "it", "we", "they", "me", "my", "your", "our", |
| 62 | "this", "that", "these", "those", "do", "did", "have", "has", "had", |
| 63 | "will", "would", "should", "could", "can", "may", "might", |
| 64 | "what", "when", "where", "why", "how", "who", "which", |
| 65 | "go", "get", "got", "make", "made", "let", "let's", "yeah", "ok", "well", |
| 66 | "just", "really", "very", "much", "more", "most", "some", "any", |
| 67 | "not", "no", "yes", "now", "before", "after", |
| 68 | } |
| 69 | |
| 70 | |
| 71 | def extract_items(text: str) -> List[str]: |
| 72 | """Return non-empty stripped lines (each line = one item).""" |
| 73 | return [line.strip() for line in text.splitlines() if line.strip()] |
| 74 | |
| 75 | |
| 76 | def extract_keywords(items: List[str]) -> List[str]: |
| 77 | """Return all alphabetic tokens >= 3 chars, lowercased, stop-words removed.""" |
| 78 | tokens: List[str] = [] |
| 79 | for it in items: |
| 80 | for tok in re.findall(r"\b[A-Za-z][a-zA-Z]{2,}\b", it): |
| 81 | t = tok.lower() |
| 82 | if t in STOP_WORDS: |
| 83 | continue |
| 84 | tokens.append(t) |
| 85 | return tokens |
| 86 | |
| 87 | |
| 88 | def detect_clusters(items: List[str], min_cluster_size: int) -> List[Dict[str, Any]]: |
| 89 | """A 'cluster' is a keyword that appears in min_cluster_size+ different items.""" |
| 90 | keyword_to_item_indices: Dict[str, List[int]] = {} |
| 91 | for i, it in enumerate(items): |
| 92 | seen_in_item: set = set() |
| 93 | for tok in re.findall(r"\b[A-Za-z][a-zA-Z]{2,}\b", it): |
| 94 | t = tok.lower() |
| 95 | if t in STOP_WORDS or t in seen_in_item: |
| 96 | continue |
| 97 | seen_in_item.add(t) |
| 98 | keyword_to_item_indices.setdefault(t, []).append(i) |
| 99 | clusters: List[Dict[str, Any]] = [] |
| 100 | for kw, idxs in keyword_to_item_indices.items(): |
| 101 | if len(idxs) >= min_cluster_size: |
| 102 | clusters.append({"keyword": kw, "item_indices": idxs, "size": len(idxs)}) |
| 103 | clusters.sort(key=lambda c: (-c["size"], c["keyword"])) |
| 104 | return clusters |
| 105 | |
| 106 | |
| 107 | def estimate(text: str, min_cluster_size: int = 3) -> Dict[str, Any]: |
| 108 | items = extract_items(text) |
| 109 | item_count = len(items) |
| 110 | clusters = detect_clusters(items, min_cluster_size) |
| 111 | cluster_count = len(clusters) |
| 112 | |
| 113 | # Decision logic per references/complexity_matching.md |
| 114 | if item_count >= 8 and cluster_count >= 1: |
| 115 | recommendation = "full" |
| 116 | rationale = f"{item_count} items with {cluster_count} cluster(s) of {min_cluster_size}+ → full 4-section format" |
| 117 | elif item_count >= 8 and cluster_count == 0: |
| 118 | recommendation = "compressed" |
| 119 | rationale = f"{item_count} items but no clustering signal → compressed (with 'no clusters' note)" |
| 120 | elif 5 <= item_count <= 7 and cluster_count >= 1: |
| 121 | recommendation = "full" |
| 122 | rationale = f"{item_count} items with clustering signal → judgment call, defaulting full" |
| 123 | elif 5 <= item_count <= 7 and cluster_count == 0: |
| 124 | recommendation = "compressed" |
| 125 | rationale = f"{item_count} items, no clustering → compressed" |
| 126 | elif item_count <= 5: |
| 127 | recommendation = "compressed" |
| 128 | rationale = f"{item_count} items (small dump) → compressed" |
| 129 | else: |
| 130 | recommendation = "compressed" |
| 131 | rationale = "fallback → compressed" |
| 132 | |
| 133 | return { |
| 134 | "item_count": item_count, |
| 135 | "cluster_count": cluster_count, |
| 136 | "clusters": clusters[:5], # top 5 only in output for readability |
| 137 | "recommendation": recommendation, |
| 138 | "rationale": rationale, |
| 139 | } |
| 140 | |
| 141 | |
| 142 | def render_human(result: Dict[str, Any]) -> str: |
| 143 | out: List[str] = [] |
| 144 | out.append(f"Item count: {result['item_count']}") |
| 145 | out.append(f"Cluster count: {result['cluster_count']}") |
| 146 | out.append(f"Recommendation: format={result['recommendation']}") |
| 147 | out.append(f"Rationale: {result['rationale']}") |
| 148 | if result["clusters"]: |
| 149 | out.append("") |
| 150 | out.append("Top clusters (keyword → items containing it):") |
| 151 | for c in result["clusters"]: |
| 152 | out.append(f" - '{c['keyword']}' in {c['size']} items: lines {c['item_indices']}") |
| 153 | return "\n".join(out) |
| 154 | |
| 155 | |
| 156 | def main(argv: List[str]) -> int: |
| 157 | parser = argparse.ArgumentParser(description=__doc__.split("\n")[0]) |
| 158 | parser.add_argument("path", nargs="?", help="Path to dump file") |
| 159 | parser.add_argument("--sample", choices=["large", "small"], help="Estimate the embedded sample dump (large or small)") |
| 160 | parser.add_argument("--min-cluster-size", type=int, default=3, help="Minimum items sharing a keyword to count as a cluster (default: 3)") |
| 161 | parser.add_argument("--output", choices=["human", "json"], default="human") |
| 162 | args = parser.parse_args(argv) |
| 163 | |
| 164 | if args.sample: |
| 165 | text = SAMPLE_DUMP_LARGE if args.sample == "large" else SAMPLE_DUMP_SMALL |
| 166 | elif args.path: |
| 167 | p = Path(args.path) |
| 168 | if not p.exists(): |
| 169 | print(f"error: {args.path} not found", file=sys.stderr) |
| 170 | return 2 |
| 171 | text = p.read_text(encoding="utf-8") |
| 172 | else: |
| 173 | parser.print_help() |
| 174 | return 0 |
| 175 | |
| 176 | result = estimate(text, args.min_cluster_size) |
| 177 | if args.output == "json": |
| 178 | print(json.dumps(result, indent=2)) |
| 179 | else: |
| 180 | print(render_human(result)) |
| 181 | return 0 |
| 182 | |
| 183 | |
| 184 | if __name__ == "__main__": |
| 185 | sys.exit(main(sys.argv[1:])) |
文件内容是平台检查时读出的文本，只供查看。平台不会运行包里的任何脚本或程序。
