wip

build.py

@@ -1,7 +1,3 @@
-import cython
-import os
-import glob
-import shutil
 from setuptools import Extension, Distribution
 from typing import List
 
@@ -14,17 +10,20 @@
 BUILD_DIR = Path("cython_build")
 compile_args = []
 link_args = []
-# Relative to project root directory
-include_dirs = ["bwa", "bwapy"]
+include_dirs = ["bwa"]
 libraries = ['m', 'z', 'pthread']
 library_dirs=['bwa']
 extra_objects = []  #glob.glob(os.path.join('bwa', '*.o'))
+h_files = []
+c_files = []
+for root_dir in ["bwa", "bwapy"]:
+    h_files.extend(str(x) for x in Path(root_dir).rglob("*.h"))
+    c_files.extend(str(x) for x in Path(root_dir).rglob("*.c") if x.name not in ['example.c', 'main.c'])
 
-c_files = [str(x) for x in Path("bwa").rglob("*.c") if x.name not in ['example.c', 'main.c']]
 extension_module = Extension(
     name='bwapy.libbwapy',
     sources=['bwapy/libbwapy.pyx'] + c_files,
-    depends=['bwa/bwtaln.h'],
+    depends=h_files,
     extra_compile_args=compile_args,
     extra_link_args=link_args,
     extra_objects=extra_objects,
@@ -35,7 +34,6 @@
 )
 
 
-
 def cythonize_helper(extension_modules: List[Extension]) -> List[Extension]:
     """Cythonize all Python extensions"""
 

bwapy/__init__.py

@@ -1,3 +0,0 @@
-import bwapy.libbwapy as libbwapy
-
-__all__ = (libbwapy.__all__)

bwapy/libbwapy.pxd

@@ -0,0 +1,113 @@
+# cython: language_level=3
+
+from libc.stdint cimport uint8_t, uint64_t, uint16_t, uint32_t, int64_t, int32_t
+from libc.stdio cimport FILE
+
+cdef extern from "libbwapy_utils.h":
+    void bwa_cal_pac_pos_with_bwt(const bntseq_t *bns, int n_seqs, bwa_seq_t *seqs, int max_mm,
+                                  float fnr, bwt_t *bwt)
+
+cdef extern from "utils.h":
+    int err_fseek(FILE *stream, long offset, int whence)
+    size_t err_fread_noeof(void *ptr, size_t size, size_t nmemb, FILE *stream)
+
+cdef extern from "bntseq.h":
+    unsigned char nst_nt4_table[256]
+    int bns_cnt_ambi(const bntseq_t *bns, int64_t pos_f, int len, int *ref_id)
+
+cdef extern from "bwa.h":
+    char * bwa_idx_infer_prefix(const char * hint)
+
+cdef extern from "bwt.h":
+    ctypedef struct bwt_t:
+        int sa_intv
+
+    bwt_t *bwt_restore_bwt(const char *fn)
+    void bwt_restore_sa(const char *fn, bwt_t *bwt);
+    void bwt_destroy(bwt_t *bwt)
+
+cdef extern from "bwtaln.h":
+    int BWA_TYPE_NO_MATCH
+
+    int __cigar_op(uint16_t __cigar)
+    int __cigar_len(uint16_t __cigar)
+
+    ctypedef struct gap_opt_t:
+        int trim_qual
+        int s_mm
+        int s_gapo
+        int s_gape
+        int mode # bit 24-31 are the barcode length
+        int indel_end_skip
+        int max_del_occ
+        int max_entries
+        float fnr
+        int max_diff
+        int max_gapo
+        int max_gape
+        int max_seed_diff
+        int seed_len
+        int n_threads
+        int max_top2
+        int trim_qual
+        int sam
+        char *rg_line
+        int n_occ
+        int interactive_mode
+        int with_md
+
+    gap_opt_t *gap_init_opt()
+
+    void seq_reverse(int len, unsigned char *seq, int is_comp)
+
+    ctypedef struct bwt_aln1_t:
+        pass
+
+cdef extern from "bntseq.h":
+    ctypedef  struct bntann1_t:
+        int64_t offset
+        int32_t len
+        char *name
+
+    ctypedef struct bntseq_t:
+        int64_t l_pac
+        bntann1_t *anns
+        FILE * fp_pac
+
+    bntseq_t * bns_restore(const char * prefix)
+    void bns_destroy(bntseq_t *bns)
+
+cdef extern from "kseq.h":
+    ctypedef struct kstring_t:
+        pass
+
+cdef extern from "bwase.h":
+    void bwa_aln2seq_core(int n_aln, const bwt_aln1_t *aln, bwa_seq_t *s, int set_main, int n_multi)
+    int64_t pos_end(const bwa_seq_t *p)
+    void bwa_refine_gapped(const bntseq_t *bns, int n_seqs, bwa_seq_t *seqs, unsigned char *_pacseq)
+    char *bwa_cal_md1(int n_cigar, uint16_t *cigar, int len, uint64_t pos, unsigned char *seq, uint64_t l_pac, unsigned char *pacseq, kstring_t *str, int *_nm)
+    void bwase_initialize()
+
+cdef extern from "bwtaln.h":
+    ctypedef struct bwa_seq_t:
+
+        char *name
+        uint8_t *seq
+        uint8_t *rseq
+        uint8_t *qual
+        uint32_t len
+        uint32_t strand
+        uint32_t type
+        int mapQ
+        int clip_len
+        bwt_aln1_t *aln
+        int n_aln
+        uint16_t pos
+        uint16_t *cigar
+        int n_cigar
+        int tid
+
+
+    void bwa_free_read_seq(int n_seqs, bwa_seq_t *seqs)
+
+    void bwa_cal_sa_reg_gap(int tid, bwt_t *const bwt, int n_seqs, bwa_seq_t *seqs, const gap_opt_t *opt)

bwapy/libbwapy.pyi

@@ -0,0 +1,21 @@
+from typing import List
+
+from pysam import AlignedSegment
+from pysam import AlignmentHeader
+from pysam import FastxRecord
+
+class BwaOptions:
+    def __init__(self) -> None: ...
+    @property
+    def trim_qual(self) -> int: ...
+
+ERROR_HANDLER: str
+TEXT_ENCODING: str
+
+class BwaIndex:
+    header: AlignmentHeader
+    def __init__(self, prefix: str) -> None: ...
+
+class Bwa:
+    def __init__(self, prefix: str) -> None: ...
+    def align(self, opt: BwaOptions, max_hits: int, query: FastxRecord) -> List[AlignedSegment]: ...