AnotherReplayReader/tools/expand_knowledge.py

#!/usr/bin/env python3
"""
Extract game knowledge strings from AIAnalyze.cs, expand [MOD:] tags,
and write mod-specific knowledge files.

Usage: python expand_knowledge.py [--cs-path PATH]

Outputs:
  knowledge_default.md  — base game (vanilla) knowledge
  knowledge_corona.md   — Corona mod knowledge
"""

import re
import sys
import os
from pathlib import Path


# ── 1. Extract @"" verbatim strings from C# source ─────────────────────────

def extract_verbatim_string(text: str, start: int) -> tuple[str, int]:
    """
    Extract a C# @"" verbatim string starting after '@"'.
    Returns (content, end_position).
    Handles "" as escaped double-quote.
    """
    chars = []
    i = start
    while i < len(text):
        c = text[i]
        if c == '"':
            # Escaped quote "" → one literal "
            if i + 1 < len(text) and text[i + 1] == '"':
                chars.append('"')
                i += 2
                continue
            # End of verbatim string
            i += 1
            break
        chars.append(c)
        i += 1
    return ''.join(chars), i


def find_strings(cs_path: str) -> dict[str, str]:
    """Find all known knowledge string variables in the .cs file and return
    {variable_name: content}."""
    with open(cs_path, 'r', encoding='utf-8') as f:
        source = f.read()

    known_vars = [
        'generalDescriptions',
        'alliedDescriptions',
        'celestialDescriptions',
        'infinityIsleVanilla',
        'infinityIsleCorona',
    ]

    results = {}
    for var in known_vars:
        # Look for: var {name} = @"
        pattern = f'var {var} = @"'
        idx = source.find(pattern)
        if idx == -1:
            print(f'[WARN] Could not find "{var}" in source')
            continue
        content, end = extract_verbatim_string(source, idx + len(pattern))
        # content still has the leading newline from @"\n...
        results[var] = content
        print(f'[OK]   {var}: {len(content)} chars')

    return results


# ── 2. [MOD:] tag expansion ────────────────────────────────────────────────

def expand_mod_text(text: str, mod_name: str) -> str:
    """Expand [MOD:...] and [MOD:NO:...] tags for the given mod.
    Mimics the logic in AIAnalyze.Process()."""
    lines = text.replace('\r', '').split('\n')
    result_lines = []
    skip_next = False

    for line in lines:
        if skip_next:
            skip_next = False
            continue

        stripped = line.strip()

        # Line-level [MOD:xxx] (must be at start of line, possibly with leading spaces)
        # Check for [MOD:NO:xxx] first
        no_match = re.match(r'^(\s*)\[MOD:NO:([^\]]+)\]$', line)
        if no_match:
            indent, denied_mod = no_match.groups()
            if denied_mod.lower() == mod_name.lower():
                # [MOD:NO:corona] when mod is corona → skip content line
                skip_next = True
            # Either way, skip the tag line itself
            continue

        # Check for [MOD:xxx]
        mod_match = re.match(r'^(\s*)\[MOD:([^\]]+)\]$', line)
        if mod_match:
            indent, entry_mod = mod_match.groups()
            if entry_mod.lower() != mod_name.lower():
                # This mod's content doesn't apply → skip content line
                skip_next = True
            # Skip the tag line itself
            continue

        # Inline tags: process [MOD:xxx]content[/MOD] and [MOD:NO:xxx]content[/MOD]
        # Multiple inline tags can appear on one line (e.g. line 662)
        processed = line
        while True:
            # Find next [MOD: or [MOD:NO:
            tag_match = re.search(
                r'\[MOD:(NO:)?([^\]]+)\](.*?)\[/MOD\]',
                processed,
                re.IGNORECASE
            )
            if not tag_match:
                break

            is_no = tag_match.group(1) is not None
            entry_mod = tag_match.group(2)
            inner_text = tag_match.group(3)
            before = processed[:tag_match.start()]
            after = processed[tag_match.end():]

            include = False
            if is_no:
                # [MOD:NO:corona] → include if mod != corona
                if entry_mod.lower() != mod_name.lower():
                    include = True
            else:
                # [MOD:corona] → include if mod == corona
                if entry_mod.lower() == mod_name.lower():
                    include = True

            if include:
                processed = before + inner_text + after
            else:
                processed = before + after

        result_lines.append(processed)

    return '\n'.join(result_lines)


# ── 3. Assemble per-mod knowledge ──────────────────────────────────────────

def assemble_knowledge(strings: dict[str, str], mod_name: str) -> str:
    """Assemble the full knowledge text for a given mod."""
    parts = []

    # generalDescriptions — most is global rules, but has one inline [MOD:CORONA] tag
    # on the faction list line; process it to expand that tag.
    parts.append(expand_mod_text(strings.get('generalDescriptions', ''), mod_name).strip())

    # alliedDescriptions
    allied = strings.get('alliedDescriptions', '')
    parts.append(expand_mod_text(allied, mod_name).strip())

    # celestialDescriptions
    celestial = strings.get('celestialDescriptions', '')
    parts.append(expand_mod_text(celestial, mod_name).strip())

    # Map description — mod-specific
    map_key = f'infinityIsle{mod_name.capitalize()}'
    if map_key in strings:
        parts.append(strings[map_key].strip())
    else:
        # Fallback: default map
        default_map = strings.get('infinityIsleVanilla', '')
        parts.append(expand_mod_text(default_map, mod_name).strip())

    return '\n\n\n'.join(p for p in parts if p)


# ── 4. Main ────────────────────────────────────────────────────────────────

def main():
    # Determine paths
    script_dir = Path(__file__).resolve().parent
    repo_root = script_dir.parent  # tools/ is directly under repo root
    cs_path = repo_root / 'Utils' / 'AIAnalyze.cs'
    output_dir = repo_root

    # Override via CLI
    if '--cs-path' in sys.argv:
        idx = sys.argv.index('--cs-path')
        if idx + 1 < len(sys.argv):
            cs_path = Path(sys.argv[idx + 1])

    print(f'Reading: {cs_path}')
    if not cs_path.exists():
        print(f'[ERROR] File not found: {cs_path}')
        sys.exit(1)

    # Extract all strings
    strings = find_strings(str(cs_path))
    if not strings:
        print('[ERROR] No strings extracted, aborting.')
        sys.exit(1)

    # Generate per-mod files
    for mod_name in ('default', 'corona'):
        knowledge = assemble_knowledge(strings, mod_name)
        out_path = output_dir / f'knowledge_{mod_name}.md'
        with open(out_path, 'w', encoding='utf-8') as f:
            f.write(knowledge)
        lines = knowledge.count('\n') + 1
        print(f'[OK]   {out_path.name}: {lines} lines, {len(knowledge)} chars')

    print('\nDone. Files written to:', output_dir)


if __name__ == '__main__':
    main()