Taxonomy handling functions in C. Features: read taxdump, read binary

files, write binary files. Not fully handled yet: *.adx, *.pdx, *.ldx,
merged.dmp and delnodes.dmp files.

python/obitools3/obidms/_obitaxo.pxd

@@ -11,7 +11,7 @@ cdef class OBI_Taxonomy :
     cdef OBIDMS            _dms
     
     cpdef close(self)
-    cpdef _write(self, str prefix)
+    cpdef write(self, str prefix)
 
  
 cdef class OBI_Taxon :

python/obitools3/obidms/_obitaxo.pyx

@@ -3,11 +3,11 @@
 from obitools3.utils   cimport bytes2str, str2bytes
 
 from .capi.obitaxonomy cimport obi_read_taxonomy, \
+                               obi_read_taxdump, \
+                               obi_write_taxonomy, \
                                obi_close_taxonomy, \
-                               obi_taxo_get_taxon_with_taxid, \
+                               obi_taxo_get_taxon_with_taxid                               
-                               write_rankidx, \
+
-                               write_taxonomyidx, \
-                               write_nameidx
 
 from ._obidms cimport OBIDMS
 
@@ -19,10 +19,13 @@ cdef class OBI_Taxonomy :
 
     # TODO function to import taxonomy?
 
-    def __init__(self, OBIDMS dms, str name) :
+    def __init__(self, OBIDMS dms, str name, bint taxdump=False) :
         
         self._dms = dms
         self._name = name
+        if taxdump :
+            self._pointer = obi_read_taxdump(str2bytes(name))
+        else :
             self._pointer = obi_read_taxonomy(dms._pointer, str2bytes(name), True)    # TODO discuss
         # TODO if not found in DMS, try to import?
         
@@ -40,19 +43,31 @@ cdef class OBI_Taxonomy :
             raise Exception("Not implemented")
         
     
+#     def __iter__(self):
+#         
+#         cdef ecotx_t* taxa
+#         cdef ecotx_t* taxon_p
+#         cdef object   taxon_capsule
+# 
+#         taxa = self._pointer.taxa.taxon
+# 
+#         # Yield each taxid
+#         for t in range(self._pointer.taxa.count):
+#             taxon_p = taxa+t    # TODO not compiling for mysterious cython reasons
+#             taxon_capsule = PyCapsule_New(taxon_p, NULL, NULL)
+#             yield OBI_Taxon(taxon_capsule)
+
+
+    cpdef write(self, str prefix) :
+        if obi_write_taxonomy(self._dms._pointer, self._pointer, str2bytes(prefix)) < 0 :
+            raise Exception("Error writing the taxonomy to binary files")
+
+
     cpdef close(self) :
         if (obi_close_taxonomy(self._pointer) < 0) :
             raise Exception("Error closing the taxonomy")
     
     
-    cpdef _write(self, str prefix) :
-        if (write_rankidx(self._dms._pointer, self._pointer, str2bytes(prefix)) < 0) :
-            raise Exception("Error writing the taxonomy rank file")
-        if (write_taxonomyidx(self._dms._pointer, self._pointer, str2bytes(prefix)) < 0) :
-            raise Exception("Error writing the taxonomy taxa file")
-        if (write_nameidx(self._dms._pointer, self._pointer, str2bytes(prefix)) < 0) :
-            raise Exception("Error writing the taxonomy taxa file")
-
 
 cdef class OBI_Taxon :    # TODO dict subclass?
 

python/obitools3/obidms/capi/obitaxonomy.pxd

@@ -7,9 +7,6 @@ from libc.stdint   cimport int32_t
 
 cdef extern from "obidms_taxonomy.h" nogil:
 
-    struct OBIDMS_taxonomy_t
-    ctypedef OBIDMS_taxonomy_t* OBIDMS_taxonomy_p
-    
     struct ecotxnode :
         int32_t           taxid
         int32_t           rank
@@ -20,8 +17,27 @@ cdef extern from "obidms_taxonomy.h" nogil:
     ctypedef ecotxnode ecotx_t
 
 
+    struct ecotxidx_t :   
+        int32_t  count
+        int32_t  max_taxid
+        int32_t  buffer_size
+        ecotx_t* taxon
+
+
+    struct OBIDMS_taxonomy_t :
+#        ecorankidx_t* ranks
+#        econameidx_t* names
+        ecotxidx_t*   taxa
+
+    ctypedef OBIDMS_taxonomy_t* OBIDMS_taxonomy_p
+    
+
     OBIDMS_taxonomy_p obi_read_taxonomy(OBIDMS_p dms, const_char_p taxonomy_name, bint read_alternative_names)
 
+    OBIDMS_taxonomy_p obi_read_taxdump(const_char_p taxdump)
+    
+    int obi_write_taxonomy(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const_char_p tax_name)
+
     int obi_close_taxonomy(OBIDMS_taxonomy_p taxonomy)
     
     ecotx_t* obi_taxo_get_parent_at_rank(ecotx_t* taxon, int32_t rankidx)
@@ -40,7 +56,3 @@ cdef extern from "obidms_taxonomy.h" nogil:
     
     ecotx_t* obi_taxo_get_superkingdom(ecotx_t* taxon, OBIDMS_taxonomy_p taxonomy)
     
-    
-    int write_rankidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const_char_p taxonomy_name)
-    int write_taxonomyidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const_char_p taxonomy_name)
-    int write_nameidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const_char_p taxonomy_name)

src/obidms_taxonomy.h

@@ -17,10 +17,6 @@
 #include "obidms.h"
 
 
-#define SWAPINT32(x)     ((((x) << 24) & 0xFF000000) | (((x) <<  8) & 0xFF0000) | \
-                         (((x) >>  8) & 0xFF00)     | (((x) >> 24) & 0xFF))
-
-
 typedef struct {
 	int32_t  taxid;
 	int32_t  rank;
@@ -109,4 +105,7 @@ int write_rankidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* taxonomy_name
 int write_taxonomyidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* taxonomy_name);
 int write_nameidx(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* taxonomy_name);
 
+int obi_write_taxonomy(OBIDMS_p dms, OBIDMS_taxonomy_p tax, const char* tax_name);
+
+OBIDMS_taxonomy_p obi_read_taxdump(const char* taxdump);