#include "vdm/rules/filename.hpp"

#include <string>

#include "vtest.hpp"

using vdm::rules::sanitize_filename;

VT_TEST(filename_passthrough_when_already_clean) {
    VT_CHECK_EQ(sanitize_filename("report.pdf"), std::string("report.pdf"));
}

VT_TEST(filename_strips_path_separators) {
    VT_CHECK_EQ(sanitize_filename("a/b\\c.txt"), std::string("abc.txt"));
}

VT_TEST(filename_collapses_dotdot_after_separator_strip) {
    // "../../etc/passwd" -> separators stripped, then the resulting ".." runs collapse to
    // a single '.', which strip-trailing-dot then removes entirely.
    auto out = sanitize_filename("../../etc/passwd");
    VT_CHECK(out.find("..") == std::string::npos);
}

VT_TEST(filename_strips_control_bytes) {
    std::string raw = "bad";
    raw.push_back('\0');
    raw += "name.txt";
    auto out = sanitize_filename(raw);
    VT_CHECK_EQ(out, std::string("badname.txt"));
}

VT_TEST(filename_replaces_ntfs_illegal_chars) {
    VT_CHECK_EQ(sanitize_filename("a<b>c:d\"e|f?g*h.txt"), std::string("a_b_c_d_e_f_g_h.txt"));
}

VT_TEST(filename_strips_trailing_dot_and_space) {
    VT_CHECK_EQ(sanitize_filename("name.   "), std::string("name"));
}

VT_TEST(filename_empty_falls_back_to_download) {
    VT_CHECK_EQ(sanitize_filename(""), std::string("download"));
}

VT_TEST(filename_all_stripped_falls_back_to_download) {
    VT_CHECK_EQ(sanitize_filename("/\\"), std::string("download"));
}

VT_TEST(filename_reserved_device_name_bare) {
    VT_CHECK_EQ(sanitize_filename("CON"), std::string("CON_"));
    VT_CHECK_EQ(sanitize_filename("con"), std::string("con_"));
}

VT_TEST(filename_reserved_device_name_with_extension) {
    VT_CHECK_EQ(sanitize_filename("NUL.txt"), std::string("NUL_.txt"));
    VT_CHECK_EQ(sanitize_filename("com3.tar.gz"), std::string("com3_.tar.gz"));
}

VT_TEST(filename_reserved_device_name_not_a_false_positive) {
    // "CONTEST" is not "CON" — must not get mangled.
    VT_CHECK_EQ(sanitize_filename("CONTEST.txt"), std::string("CONTEST.txt"));
    VT_CHECK_EQ(sanitize_filename("COM99.txt"), std::string("COM99.txt"));  // not COM1-9
}

VT_TEST(filename_truncates_long_name_keeping_extension) {
    std::string stem(500, 'a');
    auto out = sanitize_filename(stem + ".txt", 255);
    VT_CHECK(out.size() <= 255);
    VT_CHECK(out.ends_with(".txt"));
}

VT_TEST(filename_truncation_is_utf8_safe) {
    // Each "é" is 2 bytes (C3 A9); a 5-byte budget can fit 2 whole codepoints (4 bytes) but
    // not a 3rd (needs 6) -- an unguarded byte-length cut at 5 would split the 3rd
    // codepoint's C3 from its A9, leaving a dangling lead byte.
    std::string stem;
    for (int i = 0; i < 20; ++i) stem += "\xC3\xA9";
    auto out = sanitize_filename(stem, 5);
    VT_CHECK_EQ(out, std::string("\xC3\xA9\xC3\xA9"));  // 2 whole codepoints, 4 bytes
}

VT_TEST(filename_preserves_non_ascii) {
    VT_CHECK_EQ(sanitize_filename("caf\xC3\xA9.pdf"), std::string("caf\xC3\xA9.pdf"));
}