Revert "Push variant-wise for-loop to C extension"

This reverts commit b73b02a.
sgkit-dev · Nov 6, 2024 · ab19bfd · ab19bfd
1 parent 2dbed01
commit ab19bfd
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 56 deletions.
diff --git a/vcztools/_vcztoolsmodule.c b/vcztools/_vcztoolsmodule.c
@@ -476,57 +476,6 @@ VcfEncoder_encode(VcfEncoder *self, PyObject *args)
     return ret;
 }
 
-static PyObject *
-VcfEncoder_encode_all(VcfEncoder *self, PyObject *args)
-{
-	bool allowed_threads = false;
-
-    if (VcfEncoder_check_state(self) != 0) {
-        goto out;
-    }
-
-    Py_BEGIN_ALLOW_THREADS
-	allowed_threads = true;
-	const size_t num_variants = self->vcf_encoder->num_variants;
-	size_t bufsize = 1024;
-
-	for (size_t row = 0; row < num_variants; row++) {
-		while (true) {
-			char* const buf = PyMem_RawMalloc(bufsize);
-
-			if (buf == NULL) {
-				PyErr_NoMemory();
-				goto out;
-			}
-
-			const int64_t line_length = vcz_variant_encoder_encode(
-				self->vcf_encoder, row, buf, bufsize);
-
-			if (line_length < 0) {
-				PyMem_RawFree(buf);
-
-				if (line_length == VCZ_ERR_BUFFER_OVERFLOW) {
-					bufsize *= 2;
-				} else {
-					handle_library_error((int) line_length);
-					goto out;
-				}
-			} else {
-				puts(buf);
-				PyMem_RawFree(buf);
-				break;
-			} // if (line_length < 0)
-		} // while (true)
-	}
-
-out:
-	if (allowed_threads) {
-    	Py_END_ALLOW_THREADS
-    }
-
-	Py_RETURN_NONE;
-}
-
 static PyObject *
 VcfEncoder_print_state(VcfEncoder *self, PyObject *args)
 {
@@ -597,10 +546,6 @@ static PyMethodDef VcfEncoder_methods[] = {
         .ml_meth = (PyCFunction) VcfEncoder_encode,
         .ml_flags = METH_VARARGS,
         .ml_doc = "Return the specified row of VCF text" },
-    { .ml_name = "encode_all",
-        .ml_meth = (PyCFunction) VcfEncoder_encode_all,
-        .ml_flags = METH_VARARGS,
-        .ml_doc = "Print all rows of VCF text" },
     { NULL } /* Sentinel */
 };
 

diff --git a/vcztools/vcf_writer.py b/vcztools/vcf_writer.py
@@ -416,7 +416,19 @@ def c_chunk_to_vcf(
     if preceding_future:
         concurrent.futures.wait((preceding_future,))
 
-    encoder.encode_all()
+    # TODO: (1) make a guess at this based on number of fields and samples,
+    # and (2) log a DEBUG message when we have to double.
+    buflen = 1024
+    for j in range(num_variants):
+        failed = True
+        while failed:
+            try:
+                line = encoder.encode(j, buflen)
+                failed = False
+            except _vcztools.VczBufferTooSmall:
+                buflen *= 2
+                # print("Bumping buflen to", buflen)
+        print(line, file=output)
 
 
 def _generate_header(ds, original_header, sample_ids, *, no_version: bool = False):