Kelsidavis-WoWee/tools/editor/cli_tree_summary_md.cpp

#include "cli_tree_summary_md.hpp"
#include "cli_arg_parse.hpp"
#include "cli_format_table.hpp"

#include <cstdint>
#include <cstdio>
#include <cstring>
#include <filesystem>
#include <fstream>
#include <map>
#include <sstream>
#include <string>
#include <vector>

namespace wowee {
namespace editor {
namespace cli {

namespace {

namespace fs = std::filesystem;

struct FileRow {
    fs::path relPath;
    const FormatMagicEntry* fmt;
    uint32_t version;
    uint32_t entryCount;
    std::string catalogName;
    bool hasHeader;
    uintmax_t bytes;
};

// Read the standard catalog header (magic+version+name+
// entryCount). World/asset formats don't have this layout —
// for those we leave the header fields blank.
bool peekHeader(const fs::path& path, char magic[4],
                uint32_t& version, std::string& catalogName,
                uint32_t& entryCount) {
    std::ifstream is(path, std::ios::binary);
    if (!is) return false;
    if (!is.read(magic, 4) || is.gcount() != 4) return false;
    if (!is.read(reinterpret_cast<char*>(&version), 4)) return false;
    uint32_t nameLen = 0;
    if (!is.read(reinterpret_cast<char*>(&nameLen), 4)) return false;
    if (nameLen > (1u << 20)) return false;
    catalogName.resize(nameLen);
    if (nameLen > 0) {
        if (!is.read(catalogName.data(), nameLen)) return false;
    }
    if (!is.read(reinterpret_cast<char*>(&entryCount), 4)) return false;
    return true;
}

int handleSummary(int& i, int argc, char** argv) {
    std::string dir = argv[++i];
    std::string outPath;
    if (parseOptArg(i, argc, argv)) outPath = argv[++i];
    if (!fs::exists(dir) || !fs::is_directory(dir)) {
        std::fprintf(stderr,
            "tree-summary-md: not a directory: %s\n", dir.c_str());
        return 1;
    }
    std::vector<FileRow> rows;
    std::map<std::string, uint64_t> formatFileCounts;
    std::map<std::string, uint64_t> formatEntryCounts;
    uint64_t totalFiles = 0;
    uint64_t totalRecognized = 0;
    uint64_t totalBytes = 0;
    for (const auto& entry : fs::recursive_directory_iterator(dir)) {
        if (!entry.is_regular_file()) continue;
        ++totalFiles;
        char magic[4];
        std::ifstream is(entry.path(), std::ios::binary);
        if (!is.read(magic, 4) || is.gcount() != 4) continue;
        const FormatMagicEntry* fmt = findFormatByMagic(magic);
        if (!fmt) continue;
        ++totalRecognized;
        FileRow r;
        r.relPath = fs::relative(entry.path(), dir);
        r.fmt = fmt;
        r.version = 0; r.entryCount = 0;
        r.bytes = entry.file_size();
        // Re-open and parse standard header for catalog formats.
        // Skip for world/asset formats (infoFlag == nullptr).
        if (fmt->infoFlag != nullptr) {
            char headerMagic[4];
            uint32_t v = 0, ec = 0;
            std::string cn;
            if (peekHeader(entry.path(), headerMagic, v, cn, ec)) {
                r.version = v; r.catalogName = cn;
                r.entryCount = ec; r.hasHeader = true;
                formatEntryCounts[fmt->magic] += ec;
            } else {
                r.hasHeader = false;
            }
        } else {
            r.hasHeader = false;
        }
        formatFileCounts[fmt->magic] += 1;
        totalBytes += r.bytes;
        rows.push_back(std::move(r));
    }
    // Build the markdown.
    std::ostringstream md;
    md << "# Wowee open-format inventory: " << dir << "\n\n";
    md << "Generated by `--tree-summary-md`. "
       << "Walks the directory recursively and identifies every "
       << "Wowee open-format file by its 4-byte magic.\n\n";
    md << "## Summary\n\n";
    md << "| Stat | Value |\n";
    md << "|------|-------|\n";
    md << "| Total files scanned | " << totalFiles << " |\n";
    md << "| Recognized .w* files | " << totalRecognized << " |\n";
    md << "| Unrecognized files | "
       << (totalFiles - totalRecognized) << " |\n";
    md << "| Total recognized bytes | " << totalBytes << " |\n\n";
    if (!formatFileCounts.empty()) {
        md << "## Per-format breakdown\n\n";
        md << "| Magic | Extension | Files | Total entries | Description |\n";
        md << "|-------|-----------|-------|---------------|-------------|\n";
        for (const auto& kv : formatFileCounts) {
            std::string magicStr = kv.first;
            const FormatMagicEntry* fmt = nullptr;
            for (const FormatMagicEntry* p = formatTableBegin();
                 p != formatTableEnd(); ++p) {
                if (std::memcmp(p->magic, magicStr.data(), 4) == 0) {
                    fmt = p; break;
                }
            }
            uint64_t entries = formatEntryCounts.count(magicStr)
                ? formatEntryCounts[magicStr] : 0;
            md << "| `" << magicStr << "` | `"
               << (fmt ? fmt->extension : "?") << "` | "
               << kv.second << " | " << entries << " | "
               << (fmt ? fmt->description : "?") << " |\n";
        }
        md << "\n";
    }
    if (!rows.empty()) {
        md << "## Per-file detail\n\n";
        md << "| Path | Magic | Version | Catalog name | Entries | Bytes |\n";
        md << "|------|-------|---------|--------------|---------|-------|\n";
        for (const auto& r : rows) {
            char ms[5] = {r.fmt->magic[0], r.fmt->magic[1],
                          r.fmt->magic[2], r.fmt->magic[3], 0};
            md << "| `" << r.relPath.string() << "` | `"
               << ms << "` | ";
            if (r.hasHeader) {
                md << r.version << " | `" << r.catalogName << "` | "
                   << r.entryCount << " | ";
            } else {
                md << "- | - | - | ";
            }
            md << r.bytes << " |\n";
        }
        md << "\n";
    }
    md << "## How to inspect\n\n";
    md << "Use `--info-magic <path>` to identify any file by magic, "
       << "or the per-format `--info-*` flag listed in `--list-formats`.\n";
    std::string output = md.str();
    if (outPath.empty()) {
        std::printf("%s", output.c_str());
    } else {
        std::ofstream out(outPath);
        if (!out) {
            std::fprintf(stderr,
                "tree-summary-md: cannot write %s\n",
                outPath.c_str());
            return 1;
        }
        out << output;
        std::printf("Wrote %s\n", outPath.c_str());
        std::printf("  scanned files     : %llu\n",
                    static_cast<unsigned long long>(totalFiles));
        std::printf("  recognized .w*    : %llu\n",
                    static_cast<unsigned long long>(totalRecognized));
        std::printf("  unique formats    : %zu\n",
                    formatFileCounts.size());
    }
    return 0;
}

} // namespace

bool handleTreeSummaryMd(int& i, int argc, char** argv, int& outRc) {
    if (std::strcmp(argv[i], "--tree-summary-md") == 0 &&
        i + 1 < argc) {
        outRc = handleSummary(i, argc, argv); return true;
    }
    return false;
}

} // namespace cli
} // namespace editor
} // namespace wowee