pdf/engine/tests/document_load_test.cpp

#include "document_test_helpers.hpp"

namespace pdfengine {

TEST(DocumentLoadTest, NonExistentFileReturnsFileNotFound) {
    SKIP_IF_NO_PDFIUM();
    auto result = PdfDocument::loadFromFile("nonexistent_file_12345.pdf");
    ASSERT_FALSE(result.has_value());
    EXPECT_EQ(result.error(), EngineError::FileNotFound);
}

TEST(DocumentLoadTest, InvalidFileReturnsInvalidFormat) {
    SKIP_IF_NO_PDFIUM();
    std::string path = "invalid_format_test.pdf";
    {
        std::ofstream out(path, std::ios::binary);
        out << "NOT A PDF FILE!";
    }
    
    auto result = PdfDocument::loadFromFile(path);
    std::filesystem::remove(path);
    
    ASSERT_FALSE(result.has_value());
    EXPECT_EQ(result.error(), EngineError::InvalidFormat);
}

TEST(DocumentLoadTest, EncryptedPdfRequiresPassword) {
    SKIP_IF_NO_PDFIUM();
    auto path = getCorpusPath("edge-cases", "encrypted.pdf");
    if (!std::filesystem::exists(path)) {
        GTEST_SKIP() << "encrypted.pdf not found in corpus.";
    }

    auto result = PdfDocument::loadFromFile(path.string());
    ASSERT_FALSE(result.has_value());
    EXPECT_EQ(result.error(), EngineError::PasswordRequired);
}

TEST(DocumentLoadTest, EncryptedPdfInvalidPassword) {
    SKIP_IF_NO_PDFIUM();
    auto path = getCorpusPath("edge-cases", "encrypted.pdf");
    if (!std::filesystem::exists(path)) {
        GTEST_SKIP() << "encrypted.pdf not found in corpus.";
    }

    auto result = PdfDocument::loadFromFile(path.string(), "wrong_password");
    ASSERT_FALSE(result.has_value());
    EXPECT_EQ(result.error(), EngineError::InvalidPassword);
}

TEST(DocumentLoadTest, EncryptedPdfCorrectPassword) {
    SKIP_IF_NO_PDFIUM();
    auto path = getCorpusPath("edge-cases", "encrypted.pdf");
    if (!std::filesystem::exists(path)) {
        GTEST_SKIP() << "encrypted.pdf not found in corpus.";
    }

    std::vector<std::string> candidates = {"tessy", "test", "password", "123456", "1234", "foobar", "user", "owner", ""};
    bool success = false;
    for (const auto& pw : candidates) {
        auto result = PdfDocument::loadFromFile(path.string(), pw);
        if (result.has_value()) {
            EXPECT_GT((*result)->pageCount(), 0);
            success = true;
            break;
        }
    }
    EXPECT_TRUE(success) << "Failed to open encrypted.pdf with any of the candidate passwords.";
}

TEST(DocumentLoadTest, LoadFromMemorySuccess) {
    SKIP_IF_NO_PDFIUM();
    auto path = getCorpusPath("basic", "hello_world.pdf");
    if (!std::filesystem::exists(path)) {
        GTEST_SKIP() << "hello_world.pdf not found in corpus.";
    }

    auto buffer = readFile(path);
    ASSERT_FALSE(buffer.empty());

    auto result = PdfDocument::loadFromMemory(buffer);
    ASSERT_TRUE(result.has_value());
    EXPECT_EQ((*result)->pageCount(), 1);
}

TEST(DocumentLoadTest, LoadFromMemoryEmptyBuffer) {
    SKIP_IF_NO_PDFIUM();
    std::vector<uint8_t> empty_buf;
    auto result = PdfDocument::loadFromMemory(empty_buf);
    ASSERT_FALSE(result.has_value());
    EXPECT_EQ(result.error(), EngineError::InvalidFormat);
}

TEST(PageCountTest, CorpusPageCounts) {
    SKIP_IF_NO_PDFIUM();

    struct ExpectedPageCount {
        std::string folder;
        std::string file;
        int count;
    };

    std::vector<ExpectedPageCount> targets = {
        {"basic", "about_blank.pdf", 1},
        {"basic", "black.pdf", 1},
        {"basic", "hello_world.pdf", 1},
        {"basic", "hello_world_2_pages.pdf", 2},
        {"basic", "rectangles_multi_pages.pdf", 5}
    };

    for (const auto& t : targets) {
        auto path = getCorpusPath(t.folder, t.file);
        if (!std::filesystem::exists(path)) {
            continue;
        }
        auto docRes = PdfDocument::loadFromFile(path.string());
        ASSERT_TRUE(docRes.has_value()) << "Failed to open " << t.file;
        EXPECT_EQ((*docRes)->pageCount(), t.count) << "Mismatch for " << t.file;
    }

    auto verifyCorpusFolder = [](const std::string& folder, const std::vector<std::string>& files) {
        for (const auto& f : files) {
            auto path = getCorpusPath(folder, f);
            if (!std::filesystem::exists(path)) {
                ADD_FAILURE() << "Missing corpus file: " << path.string();
                continue;
            }

            std::vector<std::string> passwords = {""};
            if (f == "encrypted.pdf") {
                passwords = {"tessy", "test", "password", "123456", "1234", "foobar", "user", "owner"};
            }

            bool success = false;
            EngineError lastErr = EngineError::Unknown;
            for (const auto& pw : passwords) {
                auto docRes = PdfDocument::loadFromFile(path.string(), pw);
                if (docRes.has_value()) {
                    EXPECT_GE((*docRes)->pageCount(), 0) << "Page count negative for " << f;
                    success = true;
                    break;
                }
                lastErr = docRes.error();
            }
            EXPECT_TRUE(success) << "Failed to open " << f << " error: " << static_cast<int>(lastErr);
        }
    };

    verifyCorpusFolder("basic", corpus_basic);
    verifyCorpusFolder("fonts", corpus_fonts);
    verifyCorpusFolder("edge-cases", corpus_edge);
}
}