From 36a8aaa92e29b97f7addae3259685c241bc63843 Mon Sep 17 00:00:00 2001
From: Celine Mercier <celine.mercier@metabarcoding.org>
Date: Wed, 10 Jun 2020 16:57:42 +0200
Subject: [PATCH] grep: now creating empty views instead of displaying an error
 when selecting on an unexisting column/tag

---
 python/obitools3/commands/grep.pyx | 44 +++++++++++++++++++-----------
 1 file changed, 28 insertions(+), 16 deletions(-)

diff --git a/python/obitools3/commands/grep.pyx b/python/obitools3/commands/grep.pyx
index 357a690..c8bd68c 100644
--- a/python/obitools3/commands/grep.pyx
+++ b/python/obitools3/commands/grep.pyx
@@ -161,8 +161,7 @@ def obi_eval(compiled_expr, loc_env, line):
     return obi_eval_result
     
 
-def Filter_generator(options, tax_filter):
-    #taxfilter = taxonomyFilterGenerator(options)
+def Filter_generator(options, tax_filter, i_view):
 
     # Initialize conditions
     predicates = None
@@ -171,6 +170,9 @@ def Filter_generator(options, tax_filter):
     attributes = None
     if "attributes" in options and len(options["attributes"]) > 0:
         attributes = options["attributes"]
+        for attribute in attributes:
+            if attribute not in i_view:
+                return None
     lmax = None
     if "lmax" in options:
         lmax = options["lmax"]
@@ -196,6 +198,8 @@ def Filter_generator(options, tax_filter):
     if "attribute_patterns" in options and len(options["attribute_patterns"]) > 0:
         for p in options["attribute_patterns"]:
             attribute, pattern = p.split(":", 1)
+            if attribute not in i_view:
+                return None
             attribute_patterns[tobytes(attribute)] = re.compile(tobytes(pattern))
     
     def filter(line, loc_env):
@@ -324,21 +328,29 @@ def run(config):
  
     # Apply filter
     tax_filter = Taxonomy_filter_generator(taxo, config["grep"])
-    filter = Filter_generator(config["grep"], tax_filter)
+    filter = Filter_generator(config["grep"], tax_filter, i_view)        
     selection = Line_selection(i_view)
-    for i in range(len(i_view)):
-        PyErr_CheckSignals()
-        pb(i)
-        line = i_view[i]
-                 
-        loc_env = {"sequence": line, "line": line, "taxonomy": taxo, "obi_eval_result": False}
-        
-        good = filter(line, loc_env)
- 
-        if good :
-            selection.append(i)
-
-    pb(i, force=True)
+    
+    if filter is None and config["grep"]["invert_selection"]: # all sequences are selected: filter is None if no line will be selected because some columns don't exist
+        for i in range(len(i_view)):
+            PyErr_CheckSignals()
+            pb(i)
+            selection.append(i)        
+    
+    elif filter is not None :   # filter is None if no line will be selected because some columns don't exist
+        for i in range(len(i_view)):
+            PyErr_CheckSignals()
+            pb(i)
+            line = i_view[i]
+                     
+            loc_env = {"sequence": line, "line": line, "taxonomy": taxo, "obi_eval_result": False}
+            
+            good = filter(line, loc_env)
+     
+            if good :
+                selection.append(i)
+    
+    pb(len(i_view), force=True)
     print("", file=sys.stderr)
 
     # Create output view with the line selection