From edb4b967f0ea8edf23dcea0f047a04f7bf955121 Mon Sep 17 00:00:00 2001
From: Andrew D Smith <andrewds@usc.edu>
Date: Sat, 8 Jun 2024 16:52:54 -0700
Subject: [PATCH] sym and xcounts: adding verbose output to indicating
 chromosome being processed

---
 src/utils/symmetric-cpgs.cpp | 12 +++++++-----
 src/utils/xcounts.cpp        | 14 ++++++++------
 2 files changed, 15 insertions(+), 11 deletions(-)

diff --git a/src/utils/symmetric-cpgs.cpp b/src/utils/symmetric-cpgs.cpp
index 3c4190cb..b1d0cd9f 100644
--- a/src/utils/symmetric-cpgs.cpp
+++ b/src/utils/symmetric-cpgs.cpp
@@ -76,7 +76,7 @@ get_first_site(T &in, T &out) {
 }
 
 template<class T> static bool
-process_sites(T &in, T &out) {
+process_sites(const bool verbose, T &in, T &out) {
 
   // get the first site while dealing with the header
   auto [prev_site, prev_is_cpg] = get_first_site(in, out);
@@ -101,6 +101,8 @@ process_sites(T &in, T &out) {
       }
     }
     else {
+      if (verbose)
+        cerr << "processing: " << curr_site.chrom << endl;
       if (chroms_seen.find(curr_site.chrom) != cend(chroms_seen)) return false;
       chroms_seen.insert(curr_site.chrom);
 
@@ -125,7 +127,7 @@ main_symmetric_cpgs(int argc, const char **argv) {
   try {
     // file types from HTSlib use "-" for the filename to go to stdout
     string outfile{"-"};
-    // (not used) bool VERBOSE = false;
+    bool verbose = false;
     bool compress_output = false;
     int32_t n_threads = 1;
 
@@ -139,7 +141,7 @@ main_symmetric_cpgs(int argc, const char **argv) {
                       outfile);
     opt_parse.add_opt("threads", 't', "number of threads", false, n_threads);
     opt_parse.add_opt("zip", 'z', "output gzip format", false, compress_output);
-    // opt_parse.add_opt("verbose", 'v', "print more run info", false, VERBOSE);
+    opt_parse.add_opt("verbose", 'v', "print more run info", false, verbose);
     std::vector<string> leftover_args;
     opt_parse.parse(argc, argv, leftover_args);
     if (argc == 1 || opt_parse.help_requested()) {
@@ -179,11 +181,11 @@ main_symmetric_cpgs(int argc, const char **argv) {
       tp.set_io(out);
     }
 
-    const bool sites_are_sorted = process_sites(in, out);
+    const bool sites_are_sorted = process_sites(verbose, in, out);
 
     if (!sites_are_sorted) {
-      namespace fs = std::filesystem;
       cerr << "sites are not sorted in: " << filename << endl;
+      namespace fs = std::filesystem;
       const fs::path outpath{outfile};
       if (fs::exists(outpath)) fs::remove(outpath);
       return EXIT_FAILURE;
diff --git a/src/utils/xcounts.cpp b/src/utils/xcounts.cpp
index 2c77a49a..c9ddc3c9 100644
--- a/src/utils/xcounts.cpp
+++ b/src/utils/xcounts.cpp
@@ -82,6 +82,7 @@ fill_output_buffer(const uint32_t offset, const MSite &s, T &buf) {
 int
 main_xcounts(int argc, const char **argv) {
   try {
+    bool verbose = false;
     bool require_coverage = false;
     size_t n_threads = 1;
     string genome_file;
@@ -104,6 +105,7 @@ main_xcounts(int argc, const char **argv) {
                       false, header_file);
     opt_parse.add_opt("threads", 't', "threads for compression (use few)",
                       false, n_threads);
+    opt_parse.add_opt("verbose", 'v', "print more run info", false, verbose);
     std::vector<string> leftover_args;
     opt_parse.parse(argc, argv, leftover_args);
     if (argc == 1 || opt_parse.help_requested()) {
@@ -132,20 +134,18 @@ main_xcounts(int argc, const char **argv) {
       const int ret =
         get_chrom_sizes_for_counts_header(n_threads, genome_file,
                                           chrom_names, chrom_sizes);
-      if (ret) throw dnmt_error("failed to get chrom sizes from: " +
-                                genome_file);
+      if (ret)
+        throw dnmt_error{"failed to get chrom sizes from: " + genome_file};
     }
 
-
     bamxx::bam_tpool tpool(n_threads);
-
     bgzf_file in(filename, "r");
-    if (!in) throw dnmt_error("could not open file: " + filename);
+    if (!in) throw dnmt_error{"could not open file: " + filename};
 
     const auto outfile_mode = in.is_compressed() ? "w" : "wu";
 
     bgzf_file out(outfile, outfile_mode);
-    if (!out) throw dnmt_error("error opening output file: " + outfile);
+    if (!out) throw dnmt_error{"error opening output file: " + outfile};
 
     if (n_threads > 1) {
       if (in.is_bgzf())
@@ -184,6 +184,8 @@ main_xcounts(int argc, const char **argv) {
       if (!status_ok || !found_header) break;
 
       if (site.chrom != prev_chrom) {
+        if (verbose)
+          cerr << "processing: " << site.chrom << endl;
         prev_chrom = site.chrom;
         offset = 0;