v0.3.190613ac

wgetGenBankWGS.sh

@@ -33,7 +33,11 @@
 # = VERSIONS =                                                                                              #
 # ============                                                                                              #
 #                                                                                                           #
-  VERSION=0.2.190228ac                                                                                      #
+  VERSION=0.3.190613ac                                                                                      #
+# + no test between ftp and http protocols; use directly http                                               #
+# + fixed bug when the specified pattern has no match                                                       #
+#                                                                                                           #
+# VERSION=0.2.190228ac                                                                                      #
 # + option -d for downloading from either genbank or refseq                                                 #
 # + option -t to get the type strain name(s) for each selected species                                      #
 #                                                                                                           #
@@ -210,20 +214,12 @@ ASSEMBLY_REPORT=ftp.ncbi.nlm.nih.gov/genomes/ASSEMBLY_REPORTS/assembly_summary_$
 if [ ! -e $OUTDIR ];          then echo "creating output directory: $OUTDIR" ; mkdir $OUTDIR ;                                            fi
 
 SUMMARY=$OUTDIR/summary.txt;
-t=$(( $(date +%s%N) / 1000000 ));
-while [ 1 ]
-do
-  wget $WGETOPT -O - ftp://$ASSEMBLY_REPORT   | sed -n '2p' ;
-  [ $? == 0 ] && break || sleep 1 ;
-done > $SUMMARY 
-f=$(( $(date +%s%N) / 1000000 - $t ));
+PROTOCOL="https:";
 while [ 1 ]
 do
-  wget $WGETOPT -O - https://$ASSEMBLY_REPORT | sed -n '2p' ;
+  wget $WGETOPT -O - $PROTOCOL"//"$ASSEMBLY_REPORT > $SUMMARY ;
   [ $? == 0 ] && break || sleep 1 ;
-done > $SUMMARY 
-h=$(( $(date +%s%N) / 1000000 - $f - $t ));
-[ $h -lt $f ] && PROTOCOL="https:" || PROTOCOL="ftp:";
+done 
 
 
 #############################################################################################################
@@ -234,12 +230,10 @@ h=$(( $(date +%s%N) / 1000000 - $f - $t ));
 echo "selection criterion: $INCLUDE_PATTERN" ;
 if [ "$EXCLUDE_PATTERN" != "^#" ]; then echo "exclusion criterion: $EXCLUDE_PATTERN" ; fi
 tmp=$(randomfile $SUMMARY);
-while [ 1 ]
-do
-  wget $WGETOPT -O - $PROTOCOL"//"$ASSEMBLY_REPORT | grep -E "$INCLUDE_PATTERN" | grep -v -E "$EXCLUDE_PATTERN" | grep -F "ftp://ftp.ncbi.nlm.nih.gov" > $tmp ;
-  if [ $? == 0 ]; then cat $tmp >> $SUMMARY ; rm -f $tmp ; break; fi
-  sleep 1 ;
-done
+mv $SUMMARY $tmp ;
+sed -n '2p' $tmp > $SUMMARY ;
+sed '1,2d' $tmp | grep -E "$INCLUDE_PATTERN" | grep -v -E "$EXCLUDE_PATTERN" | grep -F "ftp://ftp.ncbi.nlm.nih.gov" >> $SUMMARY ;
+rm -f $tmp ;
 n=$(grep -v -c "^#" $SUMMARY);
 echo "$REPOSITORY: $n WGS nucleotide sequence FASTA files" ;
 if [ $n -eq 0 ]; then exit 0 ; fi