Taxonomy: added functions to check if a taxonomy already exists in a

DMS, and added taxdump import from a compressed file

python/obitools3/dms/capi/obitaxonomy.pxd

@@ -37,7 +37,9 @@ cdef extern from "obidms_taxonomy.h" nogil:
 
     ctypedef OBIDMS_taxonomy_t* OBIDMS_taxonomy_p
     
-
+    
+    int obi_taxonomy_exists(OBIDMS_p dms, const char* taxonomy_name)
+    
     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)

python/obitools3/dms/taxo/taxo.pyx

@@ -2,7 +2,8 @@
 
 from obitools3.utils cimport str2bytes, bytes2str, tobytes, tostr
 
-from ..capi.obitaxonomy cimport obi_read_taxonomy, \
+from ..capi.obitaxonomy cimport obi_taxonomy_exists, \
+                                obi_read_taxonomy, \
                                 obi_read_taxdump, \
                                 obi_write_taxonomy, \
                                 obi_close_taxonomy, \
@@ -17,6 +18,8 @@ from ..capi.obitaxonomy cimport obi_read_taxonomy, \
                                            
 from cpython.pycapsule cimport PyCapsule_New, PyCapsule_GetPointer
 
+import tarfile
+
 
 cdef class Taxonomy(OBIWrapper) :
     # TODO function to import taxonomy?
@@ -25,6 +28,16 @@ cdef class Taxonomy(OBIWrapper) :
         return <OBIDMS_taxonomy_p>(self._pointer)        
 
 
+    @staticmethod
+    def exists(DMS dms, object name) :
+        e = obi_taxonomy_exists(dms.pointer(), tobytes(name))
+        if e < 0:
+            raise RuntimeError("Error : Cannot check if taxonomy %s exists" 
+                               % tostr(name))
+        else:
+            return e
+        
+
     @staticmethod
     def open(DMS dms, object name) :
         
@@ -46,27 +59,45 @@ cdef class Taxonomy(OBIWrapper) :
         taxo._ranks = []
         for r in range((<OBIDMS_taxonomy_p>pointer).ranks.count) :
             taxo._ranks.append(obi_taxo_rank_index_to_label(r, (<OBIDMS_taxonomy_p>pointer).ranks))
-                
+
         return taxo
 
 
     @staticmethod
-    def open_taxdump(DMS dms, object name) :
+    def open_taxdump(DMS dms, object path) :
         
         cdef void*    pointer
         cdef Taxonomy taxo
-                
-        pointer = <void*>obi_read_taxdump(tobytes(name))        
+        cdef bytes    path_b
+        cdef int      idx
+        
+        path_b = tobytes(path)
+        folder_path = path_b
+
+        if path_b.endswith(b"tar.gz") or path_b.endswith(b"tar"):
+            idx = path_b.index(b".tar")
+            folder_path = path_b[:idx]
+
+        if path_b.endswith(b"tar.gz"):
+            tar = tarfile.open(path_b, "r:gz")
+            tar.extractall(path=tostr(folder_path))
+            tar.close()
+        elif path_b.endswith(b"tar"):
+            tar = tarfile.open(path_b, "r:")
+            tar.extractall(path=tostr(folder_path))
+            tar.close()
+            
+        pointer = <void*>obi_read_taxdump(folder_path)        
         if pointer == NULL :
             raise RuntimeError("Error : Cannot read taxonomy %s" 
-                               % tostr(name))
+                               % tostr(folder_path))
         
         taxo = OBIWrapper.new_wrapper(Taxonomy, pointer)
 
         dms.register(taxo)
 
         taxo._dms = dms
-        taxo._name = tobytes(name)
+        taxo._name = folder_path
         
         taxo._ranks = []
         for r in range((<OBIDMS_taxonomy_p>pointer).ranks.count) :