From d7799a390ed6dccb569b3571f4c38b7a88ac66ad Mon Sep 17 00:00:00 2001
From: Daniel-VM <danielvmillares@gmail.com>
Date: Tue, 9 Apr 2024 12:22:23 +0200
Subject: [PATCH] add reviewer sugestions #258

---
 .../assets/pipeline_utils/viralrecon.py       | 19 ++++++++++++++-----
 relecov_tools/conf/bioinfo_config.json        |  4 ++--
 2 files changed, 16 insertions(+), 7 deletions(-)

diff --git a/relecov_tools/assets/pipeline_utils/viralrecon.py b/relecov_tools/assets/pipeline_utils/viralrecon.py
index 07240c05..5d001c22 100644
--- a/relecov_tools/assets/pipeline_utils/viralrecon.py
+++ b/relecov_tools/assets/pipeline_utils/viralrecon.py
@@ -136,17 +136,26 @@ def convert_to_json(self, samp_dict):
         j_list = []
         # Grab date from filename
         result_regex = re.search(
-            "variants_long_table_(.*).csv", os.path.basename(self.file_path)
+            "variants_long_table(?:_\d{8})?\.csv", os.path.basename(self.file_path)
         )
+        stderr.print(result_regex.group(0))
         if result_regex is None:
-            log.error("Analysis date not found in filename, aborting")
             stderr.print(
-                "[red]Error: filename must include analysis date in format YYYYMMDD"
+                "[EROR]\tWARN: Couldn't find variants long table file. Expected file name is:\n\t" + 
+                "variants_long_table.csv or variants_long_table_YYYYMMDD.csv. Aborting..."
             )
-            stderr.print("[red]e.g. variants_long_table_20220830.csv")
             sys.exit(1)
+        else:
+            date_regex = re.search(r"(\d{8})", result_regex.group())
+            if date_regex is not None:
+                analysis_date = date_regex
+                stderr.print(f"[green]\tDate {analysis_date.group()} found in {self.file_path}")
+            else:
+                analysis_date = "Not Provided [GENEPIO:0001668]"
+                stderr.print(f"[yellow]\tWARN:No analysis date found in long table: {self.file_path}")
+
         for key, values in samp_dict.items():
-            j_dict = {"sample_name": key, "analysis_date": result_regex.group(1)}
+            j_dict = {"sample_name": key, "analysis_date": analysis_date.group()}
             j_dict["variants"] = values
             j_list.append(j_dict)
         return j_list
diff --git a/relecov_tools/conf/bioinfo_config.json b/relecov_tools/conf/bioinfo_config.json
index 80a052b4..8789f387 100644
--- a/relecov_tools/conf/bioinfo_config.json
+++ b/relecov_tools/conf/bioinfo_config.json
@@ -1,7 +1,7 @@
 {
     "viralrecon": {
         "mapping_stats": {
-            "fn": "mapping_illumina_stats.tab",
+            "fn": "mapping_illumina_stats(?:_\\d{8})?\\.tab",
             "sample_col_idx": 5,
             "header_row_idx": 1,
             "required": true,
@@ -37,7 +37,7 @@
             }
         },
         "variants_long_table": {
-            "fn": "variants_long_table(?:_\\d{8})?\\.csv$",
+            "fn": "variants_long_table(?:_\\d{8})?\\.csv",
             "sample_col_idx": 1,
             "header_row_idx": 1,
             "required": true,