cmake_minimum_required(VERSION 3.16)

file(READ "${CMAKE_CURRENT_SOURCE_DIR}/python/markql/_meta.py" MARKQL_META_PY)

string(REGEX MATCH "__version__ = \"([0-9]+\\.[0-9]+\\.[0-9]+)\"" MARKQL_VERSION_MATCH "${MARKQL_META_PY}")

if (MARKQL_VERSION_MATCH)

set(MARKQL_PROJECT_VERSION "${CMAKE_MATCH_1}")

else()

set(MARKQL_PROJECT_VERSION "0.0.0")

endif()

project(markql VERSION ${MARKQL_PROJECT_VERSION} LANGUAGES CXX)

set(_markql_build_agent_default ON)

if (DEFINED XSQL_BUILD_AGENT)

set(_markql_build_agent_default "${XSQL_BUILD_AGENT}")

endif()

set(_markql_agent_fetch_deps_default ON)

if (DEFINED XSQL_AGENT_FETCH_DEPS)

set(_markql_agent_fetch_deps_default "${XSQL_AGENT_FETCH_DEPS}")

endif()

option(MARKQL_BUILD_CLI "Build the MarkQL CLI" ON)

option(MARKQL_BUILD_TESTS "Build tests" ON)

option(MARKQL_BUILD_PYTHON "Build the Python module" OFF)

option(MARKQL_BUILD_AGENT "Build the localhost markql-agent service" ${_markql_build_agent_default})

option(MARKQL_WITH_LIBXML2 "Use libxml2 for HTML parsing" ON)

option(MARKQL_WITH_NLOHMANN_JSON "Use nlohmann/json for CLI output" ON)

option(MARKQL_WITH_CURL "Use libcurl for URL fetching" ON)

option(MARKQL_WITH_ARROW "Enable Apache Arrow for Parquet export" ON)

option(MARKQL_WITH_FLATBUFFERS "Use FlatBuffers for DOCN artifact payloads" ON)

option(MARKQL_ENABLE_KHMER_NUMBER "Enable built-in Khmer number module" ON)

option(MARKQL_AGENT_FETCH_DEPS "Fetch markql-agent deps with FetchContent when missing" ${_markql_agent_fetch_deps_default})

option(MARKQL_OPTIMIZE_FOR_SIZE "Enable size-focused compile/link flags" OFF)

option(MARKQL_STRIP_BINARIES "Strip release-like binaries after build" OFF)

set(CMAKE_CXX_STANDARD 20)

set(CMAKE_CXX_STANDARD_REQUIRED ON)

set(CMAKE_CXX_EXTENSIONS OFF)

if (MARKQL_OPTIMIZE_FOR_SIZE AND CMAKE_CXX_COMPILER_ID MATCHES "GNU|Clang")

add_compile_options(-ffunction-sections -fdata-sections)

add_link_options(-Wl,--gc-sections)

endif()

if (MARKQL_OPTIMIZE_FOR_SIZE)

include(CheckIPOSupported)

check_ipo_supported(RESULT MARKQL_IPO_SUPPORTED OUTPUT MARKQL_IPO_OUTPUT)

if (MARKQL_IPO_SUPPORTED)

set(CMAKE_INTERPROCEDURAL_OPTIMIZATION_MINSIZEREL ON)

set(CMAKE_INTERPROCEDURAL_OPTIMIZATION_RELEASE ON)

else()

message(STATUS "IPO/LTO not enabled: ${MARKQL_IPO_OUTPUT}")

endif()

endif()

function(markql_maybe_strip_target target_name)

if (MARKQL_STRIP_BINARIES AND CMAKE_STRIP AND UNIX)

add_custom_command(

TARGET ${target_name}

POST_BUILD

COMMAND ${CMAKE_STRIP} "$<TARGET_FILE:${target_name}>"

COMMENT "Stripping ${target_name}"

VERBATIM

)

endif()

endfunction()

add_library(markql_core

core/src/version.cpp

core/src/artifacts/artifacts.cpp

core/src/artifacts/artifact_internal.cpp

core/src/artifacts/artifact_document.cpp

core/src/artifacts/artifact_query.cpp

core/src/artifacts/artifact_query_flatbuffers.cpp

core/src/helper/helper_controller.cpp

core/src/helper/helper_json.cpp

core/src/helper/helper_policy.cpp

core/src/helper/helper_result_analysis.cpp

core/src/lang/ast.cpp

core/src/lang/diagnostics.cpp

core/src/lang/diagnostics_parse_support.cpp

core/src/lang/diagnostics_details.cpp

core/src/lang/diagnostics_render.cpp

core/src/lang/markql_parser.cpp

core/src/lang/parser/parser.cpp

core/src/lang/parser/parser_expr.cpp

core/src/lang/parser/parser_expr_scalar.cpp

core/src/lang/parser/parser_query.cpp

core/src/lang/parser/parser_select.cpp

core/src/lang/parser/parser_select_extract.cpp

core/src/lang/parser/parser_source.cpp

core/src/lang/parser/parser_util.cpp

core/src/lang/parser/lexer.cpp

core/src/dom/html_parser.cpp

core/src/dom/backend/parser_naive.cpp

core/src/dom/backend/parser_libxml2.cpp

core/src/runtime/executor/executor.cpp

core/src/runtime/executor/filter.cpp

core/src/runtime/executor/filter_scalar.cpp

core/src/runtime/executor/order.cpp

core/src/util/string_util.cpp

core/src/runtime/engine/execute.cpp

core/src/runtime/engine/execute_common.cpp

core/src/runtime/engine/execute_dom_descendants.cpp

core/src/runtime/engine/execute_dom_projection.cpp

core/src/runtime/engine/execute_dom.cpp

core/src/runtime/engine/execute_fragments.cpp

core/src/runtime/engine/execute_lint.cpp

core/src/runtime/engine/execute_meta.cpp

core/src/runtime/engine/execute_output.cpp

core/src/runtime/engine/execute_relation_join.cpp

core/src/runtime/engine/execute_relation_expr.cpp

core/src/runtime/engine/execute_relation_result.cpp

core/src/runtime/engine/execute_relation.cpp

core/src/runtime/engine/execute_source.cpp

core/src/runtime/engine/query_validation_entry.cpp

core/src/runtime/engine/io.cpp

core/src/runtime/engine/column_names.cpp

core/src/runtime/engine/query_validation_rules.cpp

core/src/runtime/engine/result_builder.cpp

core/src/runtime/engine/table_extract.cpp

core/src/runtime/engine/tfidf.cpp

)

target_include_directories(markql_core

PUBLIC

core/include

PRIVATE

core/src

)

target_compile_features(markql_core PUBLIC cxx_std_20)

execute_process(

COMMAND git rev-parse --short=12 HEAD

WORKING_DIRECTORY ${CMAKE_SOURCE_DIR}

OUTPUT_VARIABLE MARKQL_GIT_COMMIT

OUTPUT_STRIP_TRAILING_WHITESPACE

ERROR_QUIET

)

if (NOT MARKQL_GIT_COMMIT)

set(MARKQL_GIT_COMMIT "unknown")

endif()

execute_process(

COMMAND git diff --quiet --ignore-submodules HEAD --

WORKING_DIRECTORY ${CMAKE_SOURCE_DIR}

RESULT_VARIABLE MARKQL_GIT_DIRTY_RESULT

ERROR_QUIET

)

set(MARKQL_GIT_DIRTY 0)

if (NOT MARKQL_GIT_DIRTY_RESULT EQUAL 0)

set(MARKQL_GIT_DIRTY 1)

endif()

target_compile_definitions(markql_core

PUBLIC

MARKQL_VERSION="${PROJECT_VERSION}"

MARKQL_GIT_COMMIT="${MARKQL_GIT_COMMIT}"

MARKQL_GIT_DIRTY=${MARKQL_GIT_DIRTY}

)

if (MARKQL_WITH_FLATBUFFERS)

find_package(flatbuffers CONFIG QUIET)

if (NOT flatbuffers_FOUND)

message(FATAL_ERROR "flatbuffers not found. Configure with the repo vcpkg manifest/build path or set MARKQL_WITH_FLATBUFFERS=OFF.")

endif()

set(MARKQL_DOCN_SCHEMA "${CMAKE_CURRENT_SOURCE_DIR}/core/src/artifacts/schema/document.fbs")

set(MARKQL_QAST_SCHEMA "${CMAKE_CURRENT_SOURCE_DIR}/core/src/artifacts/schema/prepared_query_payload.fbs")

flatbuffers_generate_headers(

TARGET markql_artifact_docn_flatbuffers

SCHEMAS ${MARKQL_DOCN_SCHEMA}

INCLUDE "${CMAKE_CURRENT_SOURCE_DIR}/core/src/artifacts/schema"

)

flatbuffers_generate_headers(

TARGET markql_artifact_qast_flatbuffers

SCHEMAS ${MARKQL_QAST_SCHEMA}

INCLUDE "${CMAKE_CURRENT_SOURCE_DIR}/core/src/artifacts/schema"

)

add_dependencies(markql_core GENERATE_markql_artifact_docn_flatbuffers)

add_dependencies(markql_core GENERATE_markql_artifact_qast_flatbuffers)

target_link_libraries(markql_core PRIVATE flatbuffers::flatbuffers

markql_artifact_docn_flatbuffers

markql_artifact_qast_flatbuffers)

endif()

if (MARKQL_ENABLE_KHMER_NUMBER)

target_sources(markql_core PRIVATE core/src/khmer_number.cpp)

target_compile_definitions(markql_core PUBLIC MARKQL_ENABLE_KHMER_NUMBER)

endif()

if (MARKQL_WITH_LIBXML2)

find_package(LibXml2)

if (LibXml2_FOUND)

target_include_directories(markql_core PRIVATE ${LIBXML2_INCLUDE_DIR})

target_link_libraries(markql_core PRIVATE ${LIBXML2_LIBRARIES})

target_compile_definitions(markql_core PRIVATE MARKQL_USE_LIBXML2)

else()

message(WARNING "LibXml2 not found; falling back to naive HTML parser.")

endif()

endif()

if (MARKQL_WITH_CURL)

find_package(CURL CONFIG QUIET)

if (TARGET CURL::libcurl)

target_link_libraries(markql_core PRIVATE CURL::libcurl)

target_compile_definitions(markql_core PUBLIC MARKQL_USE_CURL)

else()

find_package(CURL)

if (CURL_FOUND)

target_include_directories(markql_core PRIVATE ${CURL_INCLUDE_DIRS})

target_link_libraries(markql_core PRIVATE ${CURL_LIBRARIES})

target_compile_definitions(markql_core PUBLIC MARKQL_USE_CURL)

else()

message(WARNING "libcurl not found; URL fetching disabled.")

endif()

endif()

endif()

if (MARKQL_WITH_ARROW)

find_package(Arrow CONFIG QUIET)

find_package(Parquet CONFIG QUIET)

if (TARGET Arrow::arrow_shared)

set(MARKQL_ARROW_TARGET Arrow::arrow_shared)

elseif (TARGET Arrow::arrow_static)

set(MARKQL_ARROW_TARGET Arrow::arrow_static)

elseif (TARGET Arrow::arrow)

set(MARKQL_ARROW_TARGET Arrow::arrow)

elseif (TARGET arrow_shared)

set(MARKQL_ARROW_TARGET arrow_shared)

elseif (TARGET arrow_static)

set(MARKQL_ARROW_TARGET arrow_static)

elseif (TARGET arrow)

set(MARKQL_ARROW_TARGET arrow)

endif()

if (TARGET Parquet::parquet_shared)

set(MARKQL_PARQUET_TARGET Parquet::parquet_shared)

elseif (TARGET Parquet::parquet_static)

set(MARKQL_PARQUET_TARGET Parquet::parquet_static)

elseif (TARGET Parquet::parquet)

set(MARKQL_PARQUET_TARGET Parquet::parquet)

elseif (TARGET parquet_shared)

set(MARKQL_PARQUET_TARGET parquet_shared)

elseif (TARGET parquet_static)

set(MARKQL_PARQUET_TARGET parquet_static)

elseif (TARGET parquet)

set(MARKQL_PARQUET_TARGET parquet)

endif()

if (NOT MARKQL_ARROW_TARGET OR NOT MARKQL_PARQUET_TARGET)

message(WARNING "Arrow/Parquet not found; TO PARQUET disabled.")

endif()

endif()

if (MARKQL_BUILD_AGENT)

add_subdirectory(browser_plugin/agent)

endif()

if (MARKQL_BUILD_CLI)

add_executable(markql

cli/main.cpp

cli/cli_args.cpp

cli/cli_utils.cpp

cli/cli_utils_json.cpp

cli/render/query_template_renderer.cpp

cli/explore/dom_explorer.cpp

cli/explore/dom_explorer_run.cpp

cli/explore/dom_explorer_view.cpp

cli/explore/inner_html_search.cpp

cli/explore/markql_suggestor.cpp

cli/script_runner.cpp

cli/repl/core/vim_edit.cpp

cli/repl/core/vim_normal.cpp

cli/repl/core/line_editor.cpp

cli/repl/ui/autocomplete.cpp

cli/repl/ui/sql_keywords.cpp

cli/repl/state/history.cpp

cli/repl/input/terminal.cpp

cli/repl/ui/render.cpp

cli/repl/input/text_util.cpp

cli/repl/core/repl.cpp

cli/repl/config.cpp

cli/repl/commands/registry.cpp

cli/repl/commands/help_command.cpp

cli/repl/commands/set_command.cpp

cli/repl/commands/lint_command.cpp

cli/repl/commands/describe_last_command.cpp

cli/repl/commands/display_mode_command.cpp

cli/repl/commands/mode_command.cpp

cli/repl/commands/max_rows_command.cpp

cli/repl/commands/reload_config_command.cpp

cli/repl/commands/explore_command.cpp

cli/repl/commands/summarize_command.cpp

cli/repl/commands/load_command.cpp

cli/repl/commands/plugin_command.cpp

cli/repl/commands/plugin_install_command.cpp

cli/repl/commands/plugin_registry.cpp

cli/repl/commands/summarize_content_command.cpp

cli/repl/plugin_manager.cpp

cli/ui/color.cpp

cli/export/export_sinks.cpp

cli/render/duckbox_renderer.cpp

)

if (MARKQL_ENABLE_KHMER_NUMBER)

target_sources(markql PRIVATE cli/repl/commands/khmer_number_command.cpp)

endif()

target_link_libraries(markql PRIVATE markql_core)

target_include_directories(markql PRIVATE core/src core/include cli)

target_compile_definitions(markql PRIVATE MARKQL_SOURCE_ROOT="${CMAKE_SOURCE_DIR}")

if (UNIX AND NOT APPLE)

target_link_libraries(markql PRIVATE dl)

endif()

if (MARKQL_WITH_NLOHMANN_JSON)

find_package(nlohmann_json CONFIG)

if (nlohmann_json_FOUND)

target_link_libraries(markql PRIVATE nlohmann_json::nlohmann_json)

target_compile_definitions(markql PRIVATE MARKQL_USE_NLOHMANN_JSON)

else()

message(WARNING "nlohmann_json not found; using built-in JSON output.")

endif()

endif()

if (MARKQL_ARROW_TARGET AND MARKQL_PARQUET_TARGET)

target_link_libraries(markql PRIVATE ${MARKQL_ARROW_TARGET} ${MARKQL_PARQUET_TARGET})

target_compile_definitions(markql PRIVATE MARKQL_USE_ARROW)

endif()

set_target_properties(markql PROPERTIES OUTPUT_NAME markql)

add_custom_command(

TARGET markql

POST_BUILD

COMMAND ${CMAKE_COMMAND} -E copy_if_different

"$<TARGET_FILE:markql>"

"$<TARGET_FILE_DIR:markql>/markql${CMAKE_EXECUTABLE_SUFFIX}"

COMMENT "Creating compatibility binary: markql -> markql"

)

markql_maybe_strip_target(markql)

endif()

if (MARKQL_BUILD_TESTS)

enable_testing()

add_executable(markql_tests

tests/test_main.cpp

tests/test_harness.cpp

tests/test_utils.cpp

tests/test_query_basic.cpp

tests/test_sources_alias.cpp

tests/test_with_join.cpp

tests/test_shorthand.cpp

tests/test_projections.cpp

tests/test_functions.cpp

tests/test_axes.cpp

tests/test_predicates.cpp

tests/test_order_by.cpp

tests/test_duckbox.cpp

tests/test_exports.cpp

tests/test_artifacts.cpp

tests/test_malformed_html.cpp

tests/test_legacy_frames.cpp

tests/test_flatten_text.cpp

tests/test_flatten_extract.cpp

tests/test_repl.cpp

tests/test_fragments.cpp

tests/test_guardrails.cpp

tests/test_meta_commands.cpp

tests/test_cli_utils.cpp

tests/test_lexer_comments.cpp

tests/test_script_runner.cpp

tests/test_cli_args.cpp

tests/test_string_sql.cpp

tests/test_column_names.cpp

tests/test_self_ref.cpp

tests/test_query_template_renderer.cpp

tests/test_helper_core.cpp

tests/test_vim_edit.cpp

tests/test_vim_normal.cpp

tests/test_render.cpp

tests/test_dom_explorer.cpp

cli/explore/dom_explorer.cpp

cli/explore/dom_explorer_run.cpp

cli/explore/dom_explorer_view.cpp

cli/explore/inner_html_search.cpp

cli/explore/markql_suggestor.cpp

tests/test_explore_search.cpp

tests/test_markql_suggestor.cpp

tests/test_table_options.cpp

tests/test_diagnostics.cpp

cli/repl/core/vim_edit.cpp

cli/repl/core/vim_normal.cpp

cli/repl/core/line_editor.cpp

cli/repl/ui/autocomplete.cpp

cli/repl/ui/sql_keywords.cpp

cli/repl/state/history.cpp

cli/repl/input/terminal.cpp

cli/repl/ui/render.cpp

cli/repl/input/text_util.cpp

cli/repl/plugin_manager.cpp

cli/repl/config.cpp

cli/repl/commands/registry.cpp

cli/repl/commands/help_command.cpp

cli/repl/commands/set_command.cpp

cli/repl/commands/lint_command.cpp

cli/repl/commands/describe_last_command.cpp

cli/repl/commands/display_mode_command.cpp

cli/repl/commands/mode_command.cpp

cli/repl/commands/max_rows_command.cpp

cli/repl/commands/reload_config_command.cpp

cli/repl/commands/explore_command.cpp

cli/repl/commands/summarize_command.cpp

cli/repl/commands/load_command.cpp

cli/repl/commands/plugin_command.cpp

cli/repl/commands/plugin_install_command.cpp

cli/repl/commands/plugin_registry.cpp

cli/repl/commands/summarize_content_command.cpp

cli/cli_args.cpp

cli/cli_utils.cpp

cli/cli_utils_json.cpp

cli/render/query_template_renderer.cpp

cli/script_runner.cpp

cli/ui/color.cpp

cli/export/export_sinks.cpp

cli/render/duckbox_renderer.cpp

)

if (MARKQL_ENABLE_KHMER_NUMBER)

target_sources(markql_tests PRIVATE

cli/repl/commands/khmer_number_command.cpp

tests/test_khmer_number.cpp

)

endif()

target_link_libraries(markql_tests PRIVATE markql_core)

target_include_directories(markql_tests PRIVATE cli core/src core/include tests)

target_compile_definitions(markql_tests PRIVATE MARKQL_SOURCE_ROOT="${CMAKE_SOURCE_DIR}")

if (MARKQL_WITH_FLATBUFFERS)

target_link_libraries(markql_tests PRIVATE flatbuffers::flatbuffers

markql_artifact_docn_flatbuffers

markql_artifact_qast_flatbuffers)

endif()

if (MARKQL_ARROW_TARGET AND MARKQL_PARQUET_TARGET)

target_link_libraries(markql_tests PRIVATE ${MARKQL_ARROW_TARGET} ${MARKQL_PARQUET_TARGET})

target_compile_definitions(markql_tests PRIVATE MARKQL_USE_ARROW)

endif()

markql_maybe_strip_target(markql_tests)

add_executable(markql_bench_inner_html tests/bench_inner_html.cpp)

target_link_libraries(markql_bench_inner_html PRIVATE markql_core)

target_include_directories(markql_bench_inner_html PRIVATE core/include)

if (MARKQL_ARROW_TARGET AND MARKQL_PARQUET_TARGET)

target_link_libraries(markql_bench_inner_html PRIVATE ${MARKQL_ARROW_TARGET} ${MARKQL_PARQUET_TARGET})

target_compile_definitions(markql_bench_inner_html PRIVATE MARKQL_USE_ARROW)

endif()

set_target_properties(markql_bench_inner_html PROPERTIES OUTPUT_NAME markql_bench_inner_html)

add_custom_command(

TARGET markql_bench_inner_html

POST_BUILD

COMMAND ${CMAKE_COMMAND} -E copy_if_different

"$<TARGET_FILE:markql_bench_inner_html>"

"$<TARGET_FILE_DIR:markql_bench_inner_html>/markql_bench_inner_html${CMAKE_EXECUTABLE_SUFFIX}"

COMMENT "Creating compatibility binary: markql_bench_inner_html -> markql_bench_inner_html"

)

markql_maybe_strip_target(markql_bench_inner_html)

add_executable(markql_bench_artifacts tests/bench_artifacts.cpp)

target_link_libraries(markql_bench_artifacts PRIVATE markql_core)

target_include_directories(markql_bench_artifacts PRIVATE cli core/src core/include tests)

if (MARKQL_WITH_FLATBUFFERS)

target_link_libraries(markql_bench_artifacts PRIVATE flatbuffers::flatbuffers

markql_artifact_docn_flatbuffers

markql_artifact_qast_flatbuffers)

endif()

set_target_properties(markql_bench_artifacts PROPERTIES OUTPUT_NAME markql_bench_artifacts)

add_custom_command(

TARGET markql_bench_artifacts

POST_BUILD

COMMAND ${CMAKE_COMMAND} -E copy_if_different

"$<TARGET_FILE:markql_bench_artifacts>"

"$<TARGET_FILE_DIR:markql_bench_artifacts>/markql_bench_artifacts${CMAKE_EXECUTABLE_SUFFIX}"

COMMENT "Creating compatibility binary: markql_bench_artifacts -> markql_bench_artifacts"

)

markql_maybe_strip_target(markql_bench_artifacts)

set(MARKQL_TESTS

select_ul_by_id

class_in_matches_token

parent_attribute_filter

multi_tag_select

select_star

class_eq_matches_token

missing_attribute_no_match

invalid_query_throws

text_requires_non_tag_filter

projection_parent_id

projection_attributes

projection_tag_field_list

projection_sibling_pos

inner_html_function

minify_html_basic

minify_html_preserves_attribute_quotes

minify_html_preserves_protected_tags

minify_html_preserves_script_style

inner_html_depth

trim_inner_html

trim_mixed_with_other_projection

inner_html_minified_by_default

raw_inner_html_opt_out

child_axis_direct_only

ancestor_filter_on_a

limit

alias_qualifier

alias_source_only

attr_shorthand_self_and_qualified_aliases

attr_shorthand_axis_paths

parse_alias_field_with_implicit_doc

parse_alias_field_with_explicit_alias

parse_legacy_tag_binding_still_works

duplicate_source_alias_rejected

implicit_doc_alias_matches_explicit_alias_values

doc_identifier_rejected_after_explicit_realias

unknown_identifier_error_mentions_alias_or_legacy_binding

parse_with_single_and_multiple_ctes

parse_derived_table_and_joins

parse_cross_join_lateral

parse_reject_duplicate_cte_name

parse_reject_derived_table_without_alias

parse_reject_join_without_on

parse_reject_cross_join_with_on

parse_reject_lateral_without_alias

lateral_unknown_alias_throws

with_left_join_lateral_baseline_values

with_left_join_lateral_missing_right_value_null

lateral_select_self_equivalent_to_select_alias

with_qualified_parent_axis_and_case_projection

shorthand_attribute_filter

shorthand_qualified_attribute_filter

ancestor_attribute_filter

descendant_attribute_filter

parent_tag_filter

parent_id_filter

attributes_is_null

attributes_is_not_null

node_id_filter

select_exclude_single

select_exclude_list

to_table_flag

to_list_flag

attribute_projection_value

count_aggregate

count_star

summarize_star

summarize_limit

summarize_order_by_count

order_by_tag

order_by_node_id_desc

order_by_multi

not_equal_attribute

is_not_null_attribute

is_null_attribute

text_not_equal

regex_attribute

contains_attribute

contains_all_attribute

contains_any_attribute

sibling_pos_filter

has_direct_text

parenthesized_predicates

exists_child_any

exists_child_tag

exists_child_same_node

duckbox_basic_table

duckbox_truncate_cells

duckbox_maxrows_truncate

duckbox_numeric_alignment

duckbox_null_rendering

csv_escaping

csv_export_integration

json_export_integration

ndjson_export_integration

json_ndjson_stdout_fallback

artifact_cli_args

artifact_html_and_mqd_results_match

artifact_query_and_mqp_results_match

artifact_mqp_on_mqd_matches_direct

artifact_lint_behavior_unchanged

artifact_document_round_trip_is_deterministic

artifact_query_round_trip_is_deterministic

artifact_document_major_version_rejected

artifact_query_major_version_rejected

artifact_corrupted_files_fail_safely

artifact_inspect_reports_metadata

raw_source_literal

fragments_from_raw

fragments_from_inner_html

fragments_non_html_error

fragments_warn_deprecated

parse_from_string_expr

parse_from_subquery

guardrails_limit_cap

guardrails_regex_length_cap

guardrails_fragments_count_cap

malformed_missing_closing_tags

malformed_mismatched_nesting

malformed_junk_bytes_no_throw

legacy_frames_naive_siblings_under_frameset

legacy_frames_fast_node_count

legacy_frames_query_frame_rows_are_siblings

legacy_frames_query_frame_parent_is_frameset

legacy_frames_query_frameset_exists_child_frame

legacy_frames_query_noframes_parent_is_frameset

summarize_content_basic

summarize_content_khmer_requires_plugin

summarize_content_max_tokens

set_colnames_command

lint_command_toggles_on_and_off

lint_command_rejects_invalid_value

describe_last_command_outputs_map

explore_command_uses_active_alias_by_default

explore_command_accepts_direct_target_and_alias_target

parse_show_describe

execute_describe_doc

show_functions_output

describe_language_output

show_input_requires_source

show_inputs_requires_source

show_inputs_fallback_source

auto_include_source_uri

count_table_rows_header

count_table_rows_no_header

count_result_rows

proportional_column_end_maps_to_end

proportional_column_scales_middle

proportional_column_zero_source_len

flatten_extract_basic

flatten_extract_has_direct_text_predicate

flatten_extract_requires_as_pairs

flatten_extract_alias_compatibility

flatten_extract_table_drift_stability

parse_like_predicate

parse_position_with_in

parse_project_nested_string_functions

parse_legacy_has_direct_text

parse_scoped_selector_inside_text

parse_case_expression_in_select

parse_nested_case_expression

parse_parse_source_forms

eval_like_wildcards

eval_string_functions_in_select

eval_length_byte_semantics

eval_position_found_and_not_found

eval_direct_text_excludes_descendants

eval_case_expression_select_and_null_else

project_case_and_selector_positions

project_missing_match_null_and_coalesce

project_selector_scope_and_first_match

project_top_level_comparison_expression

regression_canonical_project_query

normalize_colname_examples

normalize_colname_collision_suffixing

duckbox_uses_normalized_headers_by_default

duckbox_raw_mode_keeps_raw_header

csv_header_normalized_default

csv_header_raw_mode

json_keys_follow_colname_mode

inspect_sql_input_line_comment_only

inspect_sql_input_block_comment_only

inspect_sql_input_unterminated_block_comment

line_comment_before_tokens

line_comment_between_tokens

line_comment_at_eof

block_comment_single_line

block_comment_multi_line

unterminated_block_comment_error

comment_markers_inside_string_literals

query_without_comments_unchanged_behavior

split_script_ignores_empty_statements

split_script_with_comments

split_script_unterminated_block_comment

split_script_semicolon_and_comment_markers_inside_string_literals

run_script_multi_statement_delimiters

run_script_comments_and_empty_statements

run_script_stops_on_first_error_by_default

run_script_continue_on_error

run_script_quiet_suppresses_delimiter

utf8_validation_for_script_file_content

parse_cli_args_accepts_script_flags

parse_cli_args_rejects_missing_value

parse_cli_args_rejects_unknown_argument

parse_cli_args_rejects_query_and_query_file_together

parse_cli_args_accepts_lint_inline_query

parse_cli_args_accepts_lint_json_format

parse_cli_args_rejects_format_without_lint

parse_cli_args_accepts_version_flag

parse_cli_args_accepts_render_flags

parse_cli_args_rejects_vars_without_render

parse_cli_args_rejects_render_without_query_file

parse_cli_args_rejects_unknown_render_mode

parse_cli_args_rejects_render_stdout_with_lint

parse_cli_args_accepts_color_modes

parse_cli_args_rejects_invalid_color_mode

diagnostics_color_policy_modes_and_no_color_precedence

output_color_policy_auto_and_no_color

lint_syntax_diagnostic_has_stable_code_and_span

lint_semantic_diagnostic_has_stable_code

lint_warns_select_from_alias_as_ambiguous_node_value

lint_select_self_has_no_alias_ambiguity_warning

diagnostic_text_renderer_contains_help_and_caret

diagnostic_json_renderer_contains_stable_fields

diagnostic_json_renderer_matches_snapshot

diagnostic_text_renderer_matches_golden_snippet

diagnostic_text_renderer_color_includes_ansi_tokens

diagnostic_json_renderer_never_contains_ansi_sequences

diagnose_query_failure_maps_parse_error

version_string_contains_provenance

query_template_renderer_matches_golden_output

query_template_renderer_writes_exact_output_file

query_template_renderer_requires_defined_variables

query_template_renderer_rejects_invalid_toml

query_template_renderer_rejects_invalid_template_syntax

parse_self_projection_and_where

parse_select_self_node_projection

parse_self_function_predicate

parse_self_text_attr_inner_html_functions

eval_direct_text_self_without_tag_guessing

eval_select_self_returns_current_row_node

eval_self_rebind_inside_exists_descendant

eval_text_and_attr_with_self

eval_inner_html_with_self

eval_raw_inner_html_with_self

select_mixing_tag_and_self_projection_still_errors

flatten_visible_tree_order_and_depth

render_attribute_lines_sorted_format

fuzzy_match_score_basic

fuzzy_search_orders_and_snippet

fuzzy_search_candidate_subset

fuzzy_search_prioritizes_word_level_match

make_inner_html_snippet_context

table_header_normalize_behavior

table_empty_is_semantics

table_trim_empty_rows_semantics

table_trim_empty_cols_semantics

table_sparse_non_empty_cells_only

golden_table_trim_baseline

golden_table_trim_rows

golden_table_trim_trailing_cols

golden_table_trim_rows_and_trailing_cols

golden_table_trim_all_cols

golden_table_stop_after_empty_rows

golden_table_header_normalize

golden_table_sparse_long

golden_table_sparse_wide

)

if (MARKQL_ENABLE_KHMER_NUMBER)

list(APPEND MARKQL_TESTS

khmer_number_digits

khmer_number_scale_boundaries

khmer_number_decimals_negatives

khmer_number_concatenated_input

khmer_number_round_trip_random

khmer_number_golden_vectors

khmer_number_command_to_words

khmer_number_command_compact

)

endif()

if (MARKQL_ARROW_TARGET AND MARKQL_PARQUET_TARGET)

list(APPEND MARKQL_TESTS parquet_export_smoke)

endif()

foreach(test_name IN LISTS MARKQL_TESTS)

add_test(NAME markql_${test_name} COMMAND markql_tests ${test_name})

endforeach()

endif()

if (MARKQL_BUILD_PYTHON)

find_package(Python COMPONENTS Development.Module REQUIRED)

find_package(pybind11 CONFIG REQUIRED)

pybind11_add_module(_core python/markql/_core.cpp)

target_link_libraries(_core PRIVATE markql_core)

target_include_directories(_core PRIVATE core/include)

set_target_properties(_core PROPERTIES OUTPUT_NAME "_core")

markql_maybe_strip_target(_core)

if (DEFINED SKBUILD_PLATLIB_DIR)

set_target_properties(_core PROPERTIES LIBRARY_OUTPUT_DIRECTORY "${SKBUILD_PLATLIB_DIR}/markql")

endif()

endif()

if (MARKQL_BUILD_CLI)

install(TARGETS markql RUNTIME DESTINATION bin)

if (EXISTS "${CMAKE_SOURCE_DIR}/scripts/cli/markqli.sh")

install(PROGRAMS "${CMAKE_SOURCE_DIR}/scripts/cli/markqli.sh" DESTINATION bin)

endif()

endif()

if (MARKQL_BUILD_AGENT)

install(TARGETS markql-agent RUNTIME DESTINATION bin)

endif()

if (EXISTS "${CMAKE_SOURCE_DIR}/README.md")

install(FILES "${CMAKE_SOURCE_DIR}/README.md" DESTINATION share/doc/markql)

endif()

if (EXISTS "${CMAKE_SOURCE_DIR}/LICENSE")

install(FILES "${CMAKE_SOURCE_DIR}/LICENSE" DESTINATION share/doc/markql)

endif()

if (EXISTS "${CMAKE_SOURCE_DIR}/docs/man/markql.1")

install(FILES "${CMAKE_SOURCE_DIR}/docs/man/markql.1" DESTINATION share/man/man1)

endif()

if (EXISTS "${CMAKE_SOURCE_DIR}/completions/markql.bash")

install(FILES "${CMAKE_SOURCE_DIR}/completions/markql.bash"

DESTINATION share/bash-completion/completions

RENAME markql)

install(FILES "${CMAKE_SOURCE_DIR}/completions/markql.bash"

DESTINATION share/bash-completion/completions

RENAME markql)

endif()

set(CPACK_PACKAGE_NAME "markql")

set(CPACK_PACKAGE_VENDOR "MarkQL")

set(CPACK_PACKAGE_VERSION "${PROJECT_VERSION}")

set(CPACK_PACKAGE_CONTACT "maintainers@markql.local")

set(CPACK_PACKAGE_DESCRIPTION_SUMMARY "MarkQL CLI for querying HTML with SQL-like syntax")

set(CPACK_DEBIAN_PACKAGE_SECTION "utils")

set(CPACK_DEBIAN_FILE_NAME DEB-DEFAULT)

set(CPACK_DEBIAN_PACKAGE_SHLIBDEPS ON)

include(CPack)