Split some large files (fs_nonlocal_tools.cpp, onsite_proj.cpp, to_wannier90_lcao.cpp, etc.) into smaller ones (deepmodeling#7873)

* refactor(onsite_proj): split init/init_proj/read_abacus_orb into onsite_proj_init.cpp

Step 1 of file split. Move init(), init_proj(), read_abacus_orb() from
onsite_proj.cpp to a new onsite_proj_init.cpp. Switch the main file
from whole-class 'template class' instantiation to method-level
explicit instantiation so the moved methods are emitted from
onsite_proj_init.cpp only. README comment block moved along with init.

CMakeLists.txt updated to compile the new file. Verified by
'make -j 30' in build_max_para_test (built abacus_max_para successfully).

* refactor(onsite_proj): split tabulate_atomic/overlap_proj_psi/cal_occupations into onsite_proj_overlap.cpp

Step 2 of file split. Move the three overlap/occupation methods to
onsite_proj_overlap.cpp and emit them via method-level instantiation.
Slim the main onsite_proj.cpp includes (drop onsite_proj_print.h,
cell_tools.h, parallel_reduce.h, parameter.h, timer.h, <vector>) that
were only needed by the moved methods.

Verified by 'make -j 30' in build_max_para_test (built abacus_max_para).

* refactor(onsite_proj): split force/stress wrappers into onsite_proj_force_stress.cpp

Step 3 of file split. Move the four cal_{force,stress}_onsite_{dftu,dspin}
wrappers to onsite_proj_force_stress.cpp. Main onsite_proj.cpp now
contains only the singleton accessor and destructor, with method-level
instantiation for those two. Drop the now-unused dftu_base.h and
<complex> includes from the main file.

Verified by 'make -j 30' in build_max_para_test (built abacus_max_para).

* Refactor: split cal_becp from onsite_proj_tools.cpp into onsite_proj_tools_becp.cpp

Move the cal_becp method definition from onsite_proj_tools.cpp to a new
file onsite_proj_tools_becp.cpp with method-level explicit instantiation
for the CPU specialization. Add required includes (math_polyint.h,
math_ylmreal.h) for Nonlocal_maths dependencies. Update CMakeLists.txt.

Build verified: make -j 30 in build_max_para_test succeeds.

* Refactor: split dbecp/save_vkb/revert_vkb/transfer_gcar from onsite_proj_tools.cpp

Move cal_dbecp_s, cal_dbecp_f, save_vkb, revert_vkb, and transfer_gcar
method definitions from onsite_proj_tools.cpp to a new file
onsite_proj_tools_dbecp.cpp with method-level explicit instantiation
for the CPU specialization. Add required includes (math_polyint.h,
math_ylmreal.h, kernels/force_op.h for cal_vkb1_nl_op). Update CMakeLists.

Build verified: make -j 30 in build_max_para_test succeeds.

* Refactor: split force/stress methods from onsite_proj_tools.cpp

Move cal_force_dftu and cal_force_dspin to onsite_proj_tools_force.cpp,
and cal_stress_dftu and cal_stress_dspin to onsite_proj_tools_stress.cpp.
Both new files use method-level explicit instantiation for the CPU
specialization. Update CMakeLists.txt.

Build verified: make -j 30 in build_max_para_test succeeds.

* Refactor: split fs_nonlocal_tools.cpp by functionality

Split the 827-line fs_nonlocal_tools.cpp into four functionally-cohesive
files:
- fs_nonlocal_tools.cpp (core): constructor, destructor, allocate_memory,
  delete_memory, whole-class template instantiation
- fs_nonlocal_tools_vkb.cpp: cal_vkb, cal_becp, reduce_pool_becp
- fs_nonlocal_tools_stress.cpp: cal_vkb_deri_s, cal_dbecp_s, cal_stress
- fs_nonlocal_tools_force.cpp: cal_vkb_deri_f, cal_dbecp_f, save_vkb,
  revert_vkb, transfer_gcar, cal_force

Each new file uses method-level explicit instantiation for the CPU
specialization. Update CMakeLists.txt.

Build verified: make -j 30 in build_max_para_test succeeds.
Tests verified: MODULE_PW_pw_test, MODULE_PW_pwdft_soc, onsite_op_test,
dftu_pw/core/operator, deltaspin_pw/core all pass.

* docs(agents): add incremental verification rule for multi-step refactors

Add a bullet to the Required Baseline section: for multi-step refactors
such as splitting a large .cpp into several files, build and commit after
each step rather than batching all changes before verification.

* Refactor: split to_wannier90_lcao.cpp by functionality

Split the 1202-line to_wannier90_lcao.cpp into four functionally-cohesive
files (non-template class, no explicit instantiation needed):
- to_wannier90_lcao.cpp (core): constructor, destructor, calculate
- to_wannier90_lcao_io.cpp: cal_Mmn, cal_Amn, out_unk (file output)
- to_wannier90_lcao_setup.cpp: initialize_orb_table, set_R_coor,
  count_delta_k, unkdotkb, produce_basis_orb, produce_trial_in_lcao,
  construct_overlap_table_project
- to_wannier90_lcao_overlap.cpp: cal_orbA_overlap_R, unkdotA

All files wrapped in #ifdef __LCAO. Update CMakeLists.txt.

Build verified: make -j 30 in build_max_para_test succeeds.

* Refactor: split xc_grad.cpp by functionality

Split the 1051-line xc_grad.cpp into three files:
- xc_grad.cpp (gradcorr): the monolithic gradient correction method
  (single 798-line method, cannot be mechanically split further)
- xc_grad_wfc.cpp: grad_wfc template method + explicit instantiation
- xc_grad_utils.cpp: grad_rho, grad_dot, laplacian_rho, noncolin_rho

Update CMakeLists.txt for xc_ library and test targets that compile
xc_grad.cpp directly (test_xc3/5/7, surchem cal_vcav/cal_vel).

Build verified: make -j 30 in build_max_para_test succeeds, including
all test executables.

* Refactor: extract gradcorr_prepare_rho from gradcorr (step 1/3)

Extract the rho preparation phase (FFT, array allocation, gradient and
laplacian computation) from the 798-line gradcorr method into a separate
gradcorr_prepare_rho function (26 parameters, all by reference).

gradcorr shrinks from 826 to 657 lines. The orchestrator now declares
the shared arrays, calls prepare_rho, then proceeds to the XC kernel
and potential assembly (to be extracted in steps 2-3).

Update xc_functional.h, CMakeLists.txt, and test CMakeLists that
compile xc_grad.cpp directly.

Build verified: make -j 30 in build_max_para_test succeeds.

* Refactor: extract gradcorr_xc_kernel from gradcorr (step 2/3)

Extract the main XC computation loop (#pragma omp parallel + #ifdef __LIBXC
nspin branches + OpenMP reduction) from gradcorr into a separate
gradcorr_xc_kernel function (27 parameters, all by reference/pointer).

gradcorr shrinks from 657 to 355 lines. The orchestrator now:
1. Guards + setup
2. Calls prepare_rho (step 1)
3. Calls xc_kernel (this step)
4. [Laplacian stress + assembly — to be extracted in step 3]
5. Cleanup

Update xc_functional.h, CMakeLists.txt, and test CMakeLists.

Build verified: make -j 30 in build_max_para_test succeeds.

* Refactor: extract gradcorr_assemble_vxc from gradcorr (step 3/3)

Extract the potential assembly phase (Laplacian stress contribution,
vxc from vtxcgc + dh + laplacian, noncolinear rotation back) from
gradcorr into a separate gradcorr_assemble_vxc function (24 parameters).

gradcorr is now a 183-line orchestrator that:
1. Guards + early returns
2. Setup (nspin0, fac, need_laplacian, stress_gga init)
3. Variable declarations
4. Calls prepare_rho (step 1)
5. Calls xc_kernel (step 2)
6. Calls assemble_vxc (this step)
7. Cleanup (delete arrays)

The original 798-line monolith is now split across:
- xc_grad.cpp (183 lines, orchestrator)
- xc_grad_prepare.cpp (222 lines, rho preparation)
- xc_grad_kernel.cpp (351 lines, XC computation loop)
- xc_grad_assemble.cpp (222 lines, potential assembly)

Build verified: make -j 30 in build_max_para_test succeeds.

* Replace raw new/delete with std::vector in gradcorr

- Replace 9 raw pointer allocations (rhotmp1/2, rhogsum1/2, gdr1/2, h1/2, neg) with std::vector, removing all manual delete[] in the orchestrator
- Flatten 2D arrays vsave/vgg (double**) into 1D vectors (vsave[is*nrxx+ir] index pattern), eliminating nested heap allocations
- prepare_rho now resizes vectors instead of new; kernel receives raw pointers via .data() (signature unchanged)
- assemble_vxc signature updated: double** vsave/vgg -> double* (flattened 1D)
- Build verified: make -j 30, 5 surchem tests passed

* fix: add GPU method-level explicit instantiation for split template files

After splitting fs_nonlocal_tools.cpp and onsite_proj_tools.cpp into
multiple .cpp files with method-level explicit instantiation, only CPU
instantiations were added to the split files. The whole-class explicit
instantiation in the main file does not instantiate methods defined in
other translation units, causing undefined reference errors for
DEVICE_GPU specializations during GPU linking.

Add #if ((defined __CUDA) || (defined __ROCM)) guarded GPU method-level
explicit instantiations mirroring the existing CPU ones in:
- fs_nonlocal_tools_vkb.cpp (cal_vkb/cal_becp/reduce_pool_becp)
- fs_nonlocal_tools_stress.cpp (cal_vkb_deri_s/cal_dbecp_s/cal_stress)
- fs_nonlocal_tools_force.cpp (cal_vkb_deri_f/cal_dbecp_f/save_vkb/
  revert_vkb/transfer_gcar/cal_force)
- onsite_proj_tools_dbecp.cpp (cal_dbecp_s/cal_dbecp_f/save_vkb/
  revert_vkb/transfer_gcar)
- onsite_proj_tools_force.cpp (cal_force_dftu/cal_force_dspin)
- onsite_proj_tools_stress.cpp (cal_stress_dftu/cal_stress_dspin)

Verified with local GPU build (USE_CUDA=ON): abacus_pw_gpu links cleanly.

* Replace raw new/delete with std::vector in module_xc

- xc_grad_assemble.cpp: replace 'double* dh = new double[]' with
  'std::vector<double> dh', pass .data() to grad_dot
- xc_grad_utils.cpp (grad_rho): replace gdrtmp raw pointer with
  std::vector<std::complex<double>>, use .data() for recip2real
- xc_grad_utils.cpp (grad_dot): replace aux and gaux raw pointers
  with std::vector<std::complex<double>>, use .data() for FFT calls
- xc_functional.cpp: minor comment typo fix (GGA, plane-wave)

All replacements are zero-overhead (vector uses contiguous memory
identical to new[]), exception-safe (auto-dealloc on scope exit), and
consistent with the previous RAII migration in xc_grad.cpp.

Verified: make -j 30 passes; ctest -R 'GRADCORR|VXC|PBE|SCAN|LAPL|HSE' 7/7 passed.

* split to_wannier90_pw.cpp (1/4): extract IO cal_Mmn/cal_Amn/out_unk to to_wannier90_pw_io.cpp

* split to_wannier90_pw.cpp (2/4): extract setup gen_radial/produce_trial/get_trial/integral to to_wannier90_pw_setup.cpp

* split to_wannier90_pw.cpp (3/4): extract overlap unkdotkb/unkdotW_A to to_wannier90_pw_overlap.cpp, reduce original to 77 lines (ctor/dtor/calculate/set_tpiba)

* split symm_basic.cpp (1/4): extract atom_ordering_new/test_atom_ordering to symm_basic_order.cpp

* split symm_basic.cpp (2/4): extract matrigen/setgroup to symm_basic_setgroup.cpp

* split symm_basic.cpp (3/4): extract subgroup/pointgroup to symm_basic_pointgroup.cpp

* split symm_basic.cpp (4/4): finalize base file, remove unused formatter include

* split numerical_basis.cpp (1/2): extract output_overlap+cal_overlap_Q/Sq/V+cal_flq+cal_ylm+cal_gpow to numerical_basis_overlap.cpp

* split numerical_basis.cpp (2/2): extract output_info+output_k+output_overlap_Q/Sq/V to numerical_basis_output.cpp

* Refactor xc_grad pipeline: pack stage arguments into internal structs

The gradcorr pipeline stages (gradcorr_prepare_rho / gradcorr_xc_kernel /
gradcorr_assemble_vxc) took 20+ mostly same-typed scalar parameters
(pointer/array pairs for per-spin buffers, plus several int/bool flags
with fixed ordering conventions). This made silent argument swaps easy
to introduce and hard to review.

Move the three stage functions out of the XC_Functional class into
internal free functions declared in a new xc_grad_internal.h, which is
included only by the xc_grad*.cpp implementation files. Their parameters
are packed into two structs:

- GradCorrParams: read-only inputs (charge, basis, ucell, spin flags,
  hybrid parameters, functional metadata), built once by
  XC_Functional::gradcorr and passed by const reference;
- GradCorrBuffers: per-spin scratch arrays (rhotmp/rhogsum/gdr/h/lapl/
  vlapl/vsave/vgg/neg) allocated by the prepare stage and consumed by
  the kernel and assemble stages.

The static-members use_libxc/func_type/func_id, which the stage
functions previously read implicitly across translation units, are now
explicit fields of GradCorrParams. The public entry point
XC_Functional::gradcorr keeps its exact signature, so no caller changes.
Also drop the unused leftover epsg constant and need_laplacian unpacking
in the kernel stage.

Verification: xc_ target builds (ENABLE_LIBXC=OFF, gnu++14 + OpenMP);
g++ -fsyntax-only -D__LIBXC over the xc_grad*.cpp files covers the libxc
branch; nm confirms XC_Functional::gradcorr signature unchanged and a
single definition of each stage function; MODULE_ESTATE_elecstate_pw
builds, links and passes under OMP_NUM_THREADS=1; agent governance
check reports no findings.

* Rename module_wannier wannier90 naming to w90

- git mv to_wannier90* -> to_w90* in source/source_io/module_wannier
- Rename classes toWannier90, toWannier90_PW, toWannier90_LCAO,
  toWannier90_LCAO_IN_PW to toW90, toW90_PW, toW90_LCAO, toW90_LCAO_IN_PW
- Update include guards TOWannier90*_H -> TO_W90*_H and all includes
- Update call sites in ctrl_scf_lcao.cpp / ctrl_output_pw.cpp and the
  developer comment in read_inp_postproc.cpp
- Update source_io/CMakeLists.txt and source/Makefile.Objects
- Keep INPUT parameter 'towannier90' and user-facing Wannier90 strings
  unchanged (they refer to the external Wannier90 code)

Verification:
- grep -rn "toWannier90|to_wannier90|TOWannier90" source/: no residue
- cmake --build build --target io_advanced io_input -j 16: passed
- cmake --build build --target abacus_basic_para -j 16: linked OK
- python3 tools/03_code_analysis/agent_governance_check.py --staged: no
  errors (warnings are rename-induced include-line false positives)

* update AGENTS.md

* Use std::vector for toW90 member arrays

Replace raw new/delete member arrays (R_centre, L, m, rvalue, alfa,
spin_qaxis etc.) with std::vector and drop the manual destructor cleanup.

* Clean up naming and allocations in to_w90_lcao_setup

Rename local ORB_gaunt_table MGT to mgt_, replace raw new/delete arrays
with std::vector, and wrap overlong lines.

* Use std::vector and unique_ptr in FR_overlap

Replace raw new/delete grid buffers with std::vector, own Leb_grid and
FR_container via std::unique_ptr, keep C++11 compatibility (reset(new)
instead of std::make_unique), fix tab indentation and overlong lines.

* Replace raw new with std::vector in to_w90_pw_setup

Convert local new/delete buffers (gk, sf, orbital arrays, radial
integration buffers) to std::vector, drop leftover delete[] statements,
and wrap overlong lines.

* Clean up allocations and formatting in to_w90_pw_io

Replace new/delete buffers (porter, zpiece) with std::vector, use
nullptr instead of NULL, and wrap overlong output lines.

* Replace raw new with std::vector in to_w90_pw_overlap

Convert phase/psir FFT buffers to std::vector with data() at call
sites, drop manual delete[], and wrap the overlong line.

* Own unk_inLcao via unique_ptr in to_w90_lcao_pw

Return std::unique_ptr from get_unk_from_lcao instead of a raw owning
pointer, drop the caller-side delete, and wrap overlong lines.

* Own psi_initer_ and psi rebuild via unique_ptr transfer

Allocate psi_init_nao and the new Psi through local std::unique_ptr and
release() into the owning members so ownership transfer stays explicit.

* Pass nspin explicitly to toW90 and drop redundant __MPI guard

- Add nspin parameter to toW90 constructor chain (toW90, toW90_PW,
  toW90_LCAO, toW90_LCAO_IN_PW) and store it in member nspin_,
  replacing all 19 PARAM.inp.nspin references in module_wannier;
  callers in ctrl_output_pw.cpp and ctrl_scf_lcao.cpp pass inp.nspin
- Remove unneeded #ifdef __MPI around the GlobalV::MY_RANK check in
  toW90::out_eig, consistent with other MY_RANK checks in the module
- Verified: incremental build passes with no errors and all touched
  TUs rebuilt; grep confirms no PARAM.inp.nspin left in module_wannier

* Pass nbands, nqx, dq and npol explicitly to toW90

- Add nbands/nqx/dq/npol parameters to the toW90 constructor chain
  and store them in members nbands_/nqx_/dq_/npol_, removing all
  PARAM.inp.nbands and PARAM.globalv.nqx/dq/npol references in
  module_wannier (to_w90, pw_setup, lcao_setup, lcao_overlap,
  lcao_pw); ctrl callers pass PARAM values at construction
- Verified: incremental build passes with no errors; grep confirms
  no PARAM.inp.nbands / PARAM.globalv.nqx/dq/npol left in the module;
  quality score now 13/16 files passing (avg 78.8), pw_setup and
  lcao_overlap cross the pass line

* Drop redundant __MPI guards around internally-guarded wrappers

Remove #ifdef __MPI/#endif around 12 Parallel_Reduce::reduce_all and
Parallel_Common::bcast_bool calls in module_wannier: these wrappers
are declared unconditionally and behave as no-op serially, so calling
them without preprocessor guards is safe and consistent with the
governance rule of using guarded wrappers instead of direct MPI.
Guards around MPI_Barrier, ScalapackConnector::gemm and the pool
partition logic in out_unk are kept.
- Verified: incremental build passes with no errors; 5 guards remain
  in the module, all wrapping MPI/ScaLAPACK-specific symbols or logic

* Move __MPI guards inside Parallel_Common bcast functions

Restructure parallel_common.cpp following parallel_reduce.cpp style:
the file-level #ifdef around all 11 bcast_* definitions is replaced
by function-body guards, so serial builds get no-op definitions
instead of link errors. Callers no longer need #ifdef __MPI around
bcast calls (consistent with Parallel_Reduce::reduce_* usage); the
guard previously restored around bcast_bool in to_w90.cpp is removed
again as the net change there is zero.

This fixes the "build without MPI" link failure reported by CI:
undefined reference to Parallel_Common::bcast_bool(bool&).

- Verified: incremental build with MPI passes with no errors;
  g++ -fsyntax-only of parallel_common.cpp without __MPI passes

* Fix uninitialized radial_c pointer in toW90_PW trial orbital setup

Commit 385e5d5 replaced the inline radial_in_q[wannier_index].c
expressions with a radial_c variable but left it self-initialized,
so produce_trial_in_pw dereferenced an indeterminate pointer during
Amn/Mmn trial orbital interpolation and crashed (101_PW_W90 segfault).

Initialize radial_c from radial_in_q[wannier_index].c. Verified by
rebuilding abacus_max_para and rerunning the 101_PW_W90 case with
4 MPI processes: etotref, etotperatomref, CompareAMN, CompareMMN and
CompareEIG all pass; agent_governance_check reports no findings.

* Loosen stress threshold of 15_KP_HSE_SOC_symm to 15 kbar

Same-binary reruns at np=4/OMP=2 showed the stress-sum deviation
swinging between +1.9 and -12.7 kbar run-to-run, matching the
documented non-reproducible MPI reduction order in the EXX/LibRI
sums of this ill-conditioned metallic case. Also record the
observed stress spread in the threshold comment.

* Print EXX stress in LCAO PARTS OF STRESS breakdown

The test_stress parts printout in Force_Stress_LCAO::getForceStress
omitted the EXX term, unlike the PW side (Stress_PW::cal_stress).
Add the EXX stress entry so hybrid runs can directly inspect this
contribution, which is a major component in LCAO HSE calculations.

* Add navigation header comments to split xc_grad translation units

The xc_grad.cpp header comment was updated in the split commit, but the
five remaining xc_grad*.cpp files carried no file-level comment. Add one
to each so the pipeline layout (prepare -> kernel -> assemble) and the
auxiliary utilities stay discoverable without reading xc_functional.h.

* Trim copied include blocks in split xc_grad translation units

The split TUs carried a verbatim copy of the original xc_functional.cpp
include block regardless of actual use. Trim each file to what it needs:
- ATen headers and xc_functional_op.h stay only in xc_grad_wfc.cpp,
  the only TU using ct::Tensor and the operator kernels
- libxc_abacus.h stays only in xc_grad_kernel.cpp, the only TU calling
  XC_Functional_Libxc entry points (XC_GGA_C_LYP in xc_grad.cpp comes
  from xc.h/xc_ids.h already included by xc_functional.h)
- exx_info.h, source_io/module_parameter/parameter.h and unused
  timer/constants/pw_basis_k includes are removed everywhere

Verified with the GNU Make build (non-__LIBXC path) and the CMake xc_
target (__LIBXC path), both compile cleanly.

* Fix Makefile build: register new sources and resolve VPATH conflicts

Makefile.Objects:
- Register the new .cpp files split in this PR in their object groups
- Fix dangling entries: xc_funct_hcth.o -> xc_hcth.o, remove
  onsite_proj_pw.o, memory.o -> memory_recorder.o, dftu.o ->
  dftu_lcao.o and other renamed leftovers
- Update VPATH for renamed directories (source_lcao/module_hcontainer
  and module_gint moved to source_hamilt, add source_io subdirectories)
- Add an explicit symm_rho_charge.o rule target for
  source_estate/module_charge/symm_rho.cpp: two symm_rho.cpp files
  share the name and VPATH resolves to the source_cell one, leaving
  Symmetry_rho undefined (pre-existing on develop)
- Remove the duplicate setup_dftu_pw.o from OBJS_DFTU and drop the
  orphan dmr_complex.o, which no CMakeLists.txt compiles and which
  duplicates a DensityMatrix::cal_DMR specialization

Makefile:
- Set .DEFAULT_GOAL to abacus and pass -f to the sub-make so the
  default target actually builds instead of stopping after the first
  rule (the Intel Makefile CI job has been a 0-second no-op because
  the first rule only generated build_info.h)

generate_build_info.sh:
- Locate build_info.h.in relative to the script directory so the
  header is generated regardless of the caller's cwd

Verified: full GNU Make build with mpicxx links successfully; ABACUS.mpi
--version reports v3.11.0-beta8; tests/01_PW/101_PW_W90 runs to
completion with results matching result.ref; CMake xc_ target still
compiles on both libxc paths.

---------

Co-authored-by: abacus_fixer <mohanchen@pku.eud.cn>

Author: mohanchen

AGENTS.md

@@ -34,6 +34,10 @@ rules. Read the complete governance document before making or reviewing changes:
   is required.
 - Report the exact verification performed. Do not claim completion without
   fresh test or check output.
+- For multi-step refactors (e.g., splitting a large `.cpp` into several
+  files), build and commit after each step rather than batching all changes
+  before verification. This keeps the blast radius small when a step
+  surfaces a missing include or instantiation error.
 - Prefer `std::vector` over raw `new`/`delete` for dynamic arrays; before
   converting class members, confirm no external code consumes them as raw
   pointers (e.g., `std::vector<bool>` has no `.data()`), and use
@@ -106,6 +110,8 @@ rules. Read the complete governance document before making or reviewing changes:
 python3 tools/03_code_analysis/agent_governance_check.py --staged
 python3 tools/03_code_analysis/agent_governance_check.py --base upstream/develop --head HEAD --format text
 pre-commit run abacus-agent-governance --all-files
+# Score changed C++ files for quality debt (pass line is 60):
+python3 tools/03_code_analysis/code_quality_score.py $(git diff --name-only upstream/develop...HEAD | grep -E '\.(cpp|h)$')
 ```
 
 The repository text files have been normalized to LF once. Day-to-day line

generate_build_info.sh

@@ -257,7 +257,8 @@ if [ "${USE_CUDA}" == "ON" ]; then
 fi
 # --- Final File Generation ---
 
-INPUT_FILE="source_io/build_info.h.in"
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+INPUT_FILE="${SCRIPT_DIR}/source/source_io/build_info.h.in"
 
 # Use sed to replace all placeholders with detected values
 # Note the use of different delimiters (#) for paths to avoid conflicts with /

source/Makefile

@@ -6,6 +6,10 @@ OBJ_DIR = $(BUILD_DIR)/obj
 BIN_DIR = $(BUILD_DIR)/bin
 BUILD_INFO_DIR = $(BUILD_DIR)/source_io
 
+# The first rule in this file only generates build_info.h, so make would
+# stop after generating that header unless the default goal is set here.
+.DEFAULT_GOAL := abacus
+
 include $(ABACUS_ROOT)Makefile.vars
 
 #==========================
@@ -252,18 +256,25 @@ ${OBJ_DIR}/parse_args.o: $(ABACUS_ROOT)source_io/parse_args.cpp $(BUILD_INFO_DIR
 	@mkdir -p $(dir $@)
 	${CXX} ${OPTS} ${OPTS_MPI} -c ${HONG} $< -o $@
 
+# Explicit rule: source_cell/module_symmetry and source_estate/module_charge
+# both contain a symm_rho.cpp. VPATH resolves symm_rho.o to the source_cell
+# one (listed first), so the Symmetry_rho implementation needs an explicit
+# object name to be compiled from the correct source.
+${OBJ_DIR}/symm_rho_charge.o: $(ABACUS_ROOT)source_estate/module_charge/symm_rho.cpp
+	@mkdir -p $(dir $@)
+	${CXX} ${OPTS} ${OPTS_MPI} -c ${HONG} $< -o $@
+
 ###### END of ABACUS INFO PART ######
 
 #==========================
 # MAKING OPTIONS
 #==========================
 abacus:
-	@ if [ ! -d $(OBJ_DIR) ]; then mkdir $(OBJ_DIR); fi
-	@ if [ ! -d $(BIN_DIR) ]; then mkdir $(BIN_DIR); fi
-	@ $(MAKE) $(BIN_DIR)/${VERSION}.$(suffix)
+	@ mkdir -p $(OBJ_DIR) $(BIN_DIR)
+	@ $(MAKE) -f $(firstword $(MAKEFILE_LIST)) $(BIN_DIR)/${VERSION}.$(suffix)
 
 test:
-	@ $(MAKE) abacus
+	@ $(MAKE) -f $(firstword $(MAKEFILE_LIST)) abacus
 	@ cd $(ABACUS_ROOT)../tests/integrate/;sh Autotest.sh -a $(realpath $(BIN_DIR))/ABACUS.mpi -n $(TESTNP)
 
 pw $(BIN_DIR)/${VERSION}-PW.x:
@@ -280,7 +291,8 @@ $(BIN_DIR)/${VERSION}.$(suffix) : ${FP_OBJS} ${PDIAG_OBJS} ${HEADERS}
 #==========================
 # Note: The specific rule for parse_args.o above is more precise.
 # This generic rule will apply to all other .cpp files.
-${OBJ_DIR}/%.o:$(realpath $(BIN_DIR))%.cpp
+${OBJ_DIR}/%.o:%.cpp
+	@ mkdir -p $(dir $@)
 	${CXX} ${OPTS} ${OPTS_MPI} -c ${HONG} $< -o $@
 
 .PHONY:clean test