armandli/advent-cookiecutter

Workflow

Step 1 — Parse Arguments

Split $ARGUMENTS on whitespace:

• Token 1 = output .cpp filename (required)
• Token 2 = example input filename to analyze (required)
• Token 3 = coordinate specifier (optional)

Valid specifiers: p2, pd2, pdpd2, p3, pd3, pdpd3, none

If token 1 or 2 is missing, stop and ask the user. If token 3 is present and not in the valid list, stop and ask the user.

---

Step 2 — Classify the Example Input

Read token 2 with the Read tool. Examine all lines, then pick one classification:

| Signal | Classification | |---|---| | All lines same length, only printable ASCII, no digit-only lines | grid | | Lines contain only digits, spaces, minus signs | int-tokens | | Any line contains a comma | comma-separated | | Lines match a repeated template like "key: val" or "p=%d,%d v=%d,%d" | structured-sscanf | | Blank lines divide the file into distinct groups of lines | multi-group | | Lines contain a mix of words and integers | space-separated | | Unrecognized or file unreadable | stub |

If the file cannot be read, use stub and warn the user.

---

Step 3 — Determine Struct Sections

Look up the exact code in references/struct-templates.md. Paste each listed section in order when assembling the file. D must always be emitted before P when both are present (P-2D-full / P-3D-full use D inline).

| Specifier | Sections to include (in order) | |---|---| | p2 | P-2D-simple, Comparators-2D, hash-P-2D | | pd2 | D-2D, P-2D-full, Comparators-2D, Free-Operators-2D, hash-P-2D, Direction-Constants-2D, hashing-turn-helpers | | pdpd2 | All of pd2 + PD-2D, hash-PD-2D | | p3 | P-3D-simple, Comparators-3D, hash-P-3D | | pd3 | D-3D, P-3D-full, Comparators-3D, Free-Operators-3D, hash-P-3D, Direction-Constants-3D | | pdpd3 | All of pd3 + PD-3D, hash-PD-3D | | absent / none | (nothing) |

---

Step 4 — Assemble the Output File

Construct the file text in this exact order:

1. Includes (always emit these):

#include <iostream>
#include <vector>
#include <string>
#include <cassert>
#include <array>
#include <unordered_set>
#include <unordered_map>
#include <algorithm>
#include <numeric>

Add extra includes based on classification:

• comma-separated, space-separated → also add #include <sstream>
• int-tokens with multiple values per line → also add #include <sstream>
• structured-sscanf → also add #include <cstdio>

2. Type aliases (always):

using namespace std;
using ll   = long long;
using ull  = unsigned long long;
using uint = unsigned;
template <typename T>             using uset = unordered_set<T>;
template <typename K, typename V> using umap = unordered_map<K, V>;

3. Coordinate structs — paste the sections from references/struct-templates.md identified in Step 3. Separate each section with one blank line.

4. Parsing function — paste the matching section from references/parsing-templates.md:

| Classification | Section to paste | |---|---| | grid | Grid | | int-tokens | Integer-Tokens | | comma-separated | Comma-Separated | | space-separated | Space-Separated | | structured-sscanf | Structured-sscanf | | multi-group | Multi-Group | | stub | Stub |

5. main() (always):

int main(){
  auto input = parse();
  // Part 1
  cout << 0 << endl;
  // Part 2
  cout << 0 << endl;
}

---

Step 5 — Write and Report

Use the Write tool to create the file at the exact path from token 1.

Then report:

• Output file path written
• Structs included (or "none")
• Parsing pattern detected

---

Additional Resources

• Coordinate struct code → references/struct-templates.md
• Parsing template code → references/parsing-templates.md

---

Final Step — Record Usage

After the skill's primary task completes, run:

python3 ${PWD}/.claude/skills/skill-stat/scripts/record-stat.py "advent-cookiecutter"