From 67bdee105a2477e616bdeeb7777addf6b2781547 Mon Sep 17 00:00:00 2001
From: Celine Mercier <celine.mercier@metabarcoding.org>
Date: Sat, 18 Apr 2020 14:24:08 +0200
Subject: [PATCH] C: build_ref_db: added progress display for each step

---
 src/build_reference_db.c | 29 ++++++++++++++++++++++++-----
 1 file changed, 24 insertions(+), 5 deletions(-)

diff --git a/src/build_reference_db.c b/src/build_reference_db.c
index 14f3ca5..b48fe98 100755
--- a/src/build_reference_db.c
+++ b/src/build_reference_db.c
@@ -157,7 +157,7 @@ int build_reference_db(const char* dms_name,
 	ecotx_t* lca_2 = NULL;
 	ecotx_t* lca = NULL;
 	index_t idx1, idx2;
-	index_t i, j, k;
+	index_t i, j, k, count;
 	int32_t taxid_array_length;
 	int32_t score_array_length;
 	int32_t taxid_array_writable_length;
@@ -185,6 +185,7 @@ int build_reference_db(const char* dms_name,
 	matrix_view_name = strcpy(matrix_view_name, o_view_name);
 	strcat(matrix_view_name, "_matrix");
 
+	fprintf(stderr, "Aligning queries with reference database...\n");
 	if (obi_lcs_align_one_column(dms_name,
 								 refs_view_name,
 								 "",
@@ -320,13 +321,19 @@ int build_reference_db(const char* dms_name,
 		return -1;
 	}
 
+	count = (matrix_with_lca_view->infos)->line_count;
+	fprintf(stderr, "Computing LCAs...\n");
+
 	// Compute all the LCAs
 		// For each pair
-	for (i=0; i<(matrix_with_lca_view->infos)->line_count; i++)
+	for (i=0; i<count; i++)
 	{
 		if (! keep_running)
 			return -1;
 
+		if (i%1000 == 0)
+			fprintf(stderr,"\rDone : %f %%       ", (i / (float) count)*100);
+
 		// Read all taxids associated with the first sequence and compute their LCA
 		// Read line index
 		idx1 = obi_get_int_with_elt_idx_and_col_p_in_view(matrix_with_lca_view, matrix_idx1_column, i, 0);
@@ -363,6 +370,7 @@ int build_reference_db(const char* dms_name,
 			return -1;
 		}
 	}
+	fprintf(stderr,"\rDone : 100 %%           \n");
 
 	// Clone refs view, add 2 arrays columns for lca and score, compute and write them
 
@@ -442,13 +450,18 @@ int build_reference_db(const char* dms_name,
 		return -1;
 	}
 
+	fprintf(stderr, "Building LCA arrays...\n");
+
 	// For each sequence, look for all its alignments in the matrix, and for each different LCA taxid/score, order them and write them
 	// Going through matrix once, filling refs arrays on the go for efficiency
-	for (i=0; i<(matrix_with_lca_view->infos)->line_count; i++)
+	for (i=0; i<count; i++)
 	{
 		if (! keep_running)
 			return -1;
 
+		if (i%1000 == 0)
+			fprintf(stderr,"\rDone : %f %%       ", (i / (float) count)*100);
+
 		// Read ref line indexes
 		idx1 = obi_get_int_with_elt_idx_and_col_p_in_view(matrix_with_lca_view, matrix_idx1_column, i, 0);
 		idx2 = obi_get_int_with_elt_idx_and_col_p_in_view(matrix_with_lca_view, matrix_idx2_column, i, 0);
@@ -769,11 +782,17 @@ int build_reference_db(const char* dms_name,
 			}
 		}
 	}
+	fprintf(stderr,"\rDone : 100 %%           \n");
 
+	fprintf(stderr, "Writing results...\n");
+	count = (o_view->infos)->line_count;
 	// Fill empty LCA informations (because filling from potentially sparse alignment matrix) with the sequence taxid
 	score=1.0;  // technically getting LCA of identical sequences
-	for (i=0; i<(o_view->infos)->line_count; i++)
+	for (i=0; i<count; i++)
 	{
+		if (i%1000 == 0)
+			fprintf(stderr,"\rDone : %f %%       ", (i / (float) count)*100);
+
 		obi_get_array_with_col_p_in_view(o_view, final_lca_taxid_a_column, i, &taxid_array_length);
 		if (taxid_array_length == 0)  // no LCA set
 		{
@@ -799,6 +818,7 @@ int build_reference_db(const char* dms_name,
 			}
 		}
 	}
+	fprintf(stderr,"\rDone : 100 %%           \n");
 
 	// Add information about the threshold used to build the DB
 	snprintf(threshold_str, 5, "%f", threshold);
@@ -858,7 +878,6 @@ int build_reference_db(const char* dms_name,
 	free(matrix_view_name);
 	free(matrix_with_lca_view_name);
 
-	fprintf(stderr,"\rDone : 100 %%           \n");
 	return 0;
 }