From 3b9cf88179737563acfdb83b516c0b5219cc531e Mon Sep 17 00:00:00 2001
From: Milot Mirdita <milot@mirdita.de>
Date: Fri, 29 Jul 2022 13:20:28 +0900
Subject: [PATCH] Add URIs as allowed parameter inputs

---
 src/commons/Command.cpp    |  1 +
 src/commons/Command.h      |  1 +
 src/commons/Parameters.cpp | 18 +++++++++++++++---
 src/commons/Parameters.h   |  2 ++
 4 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/src/commons/Command.cpp b/src/commons/Command.cpp
index 68596e88a..677f46a03 100644
--- a/src/commons/Command.cpp
+++ b/src/commons/Command.cpp
@@ -52,6 +52,7 @@ std::vector<int> DbValidator::directory = {Parameters::DBTYPE_DIRECTORY};
 std::vector<int> DbValidator::flatfile = {Parameters::DBTYPE_FLATFILE};
 std::vector<int> DbValidator::flatfileAndStdin = {Parameters::DBTYPE_FLATFILE, Parameters::DBTYPE_STDIN};
 std::vector<int> DbValidator::flatfileStdinAndGeneric = {Parameters::DBTYPE_FLATFILE, Parameters::DBTYPE_STDIN, Parameters::DBTYPE_GENERIC_DB};
+std::vector<int> DbValidator::flatfileStdinGenericUri = {Parameters::DBTYPE_FLATFILE, Parameters::DBTYPE_STDIN, Parameters::DBTYPE_GENERIC_DB, Parameters::DBTYPE_URI};
 std::vector<int> DbValidator::resultDb =  {Parameters::DBTYPE_ALIGNMENT_RES, Parameters::DBTYPE_PREFILTER_RES, Parameters::DBTYPE_PREFILTER_REV_RES, Parameters::DBTYPE_CLUSTER_RES};
 std::vector<int> DbValidator::ppResultDb =  {Parameters::DBTYPE_ALIGNMENT_RES, Parameters::DBTYPE_PREFILTER_RES, Parameters::DBTYPE_PREFILTER_REV_RES, Parameters::DBTYPE_CLUSTER_RES, Parameters::DBTYPE_INDEX_DB};
 std::vector<int> DbValidator::taxonomyReportInput =  {Parameters::DBTYPE_ALIGNMENT_RES, Parameters::DBTYPE_PREFILTER_RES, Parameters::DBTYPE_PREFILTER_REV_RES, Parameters::DBTYPE_CLUSTER_RES, Parameters::DBTYPE_TAXONOMICAL_RESULT, Parameters::DBTYPE_NUCLEOTIDES, Parameters::DBTYPE_HMM_PROFILE, Parameters::DBTYPE_AMINO_ACIDS};
diff --git a/src/commons/Command.h b/src/commons/Command.h
index a291f96bc..5c5fd1ae4 100644
--- a/src/commons/Command.h
+++ b/src/commons/Command.h
@@ -68,6 +68,7 @@ struct DbValidator {
     static std::vector<int> flatfile;
     static std::vector<int> flatfileAndStdin;
     static std::vector<int> flatfileStdinAndGeneric;
+    static std::vector<int> flatfileStdinGenericUri;
     static std::vector<int> empty;
 };
 
diff --git a/src/commons/Parameters.cpp b/src/commons/Parameters.cpp
index 67c3e4949..4208b272b 100644
--- a/src/commons/Parameters.cpp
+++ b/src/commons/Parameters.cpp
@@ -2007,9 +2007,15 @@ void Parameters::checkIfDatabaseIsValid(const Command& command, int argc, const
                 }
 
                 if (filenames[fileIdx] != "stdin" && FileUtil::fileExists((filenames[fileIdx]).c_str()) == false && FileUtil::fileExists((filenames[fileIdx] + ".dbtype").c_str()) == false) {
-                    printParameters(command.cmd, argc, argv, *command.params);
-                    Debug(Debug::ERROR) << "Input " << filenames[fileIdx] << " does not exist\n";
-                    EXIT(EXIT_FAILURE);
+                    regex_t regex;
+                    compileRegex(&regex, "[a-zA-Z][a-zA-Z0-9+-.]*:\\/\\/");
+                    int nomatch = regexec(&regex, filenames[fileIdx].c_str(), 0, NULL, 0);
+                    regfree(&regex);
+                    if (nomatch) {
+                        printParameters(command.cmd, argc, argv, *command.params);
+                        Debug(Debug::ERROR) << "Input " << filenames[fileIdx] << " does not exist\n";
+                        EXIT(EXIT_FAILURE);
+                    }
                 }
                 int dbtype = FileUtil::parseDbType(filenames[fileIdx].c_str());
                 if (db.specialType & DbType::NEED_HEADER && FileUtil::fileExists((filenames[fileIdx] + "_h.dbtype").c_str()) == false && Parameters::isEqualDbtype(dbtype, Parameters::DBTYPE_INDEX_DB) == false) {
@@ -2035,6 +2041,12 @@ void Parameters::checkIfDatabaseIsValid(const Command& command, int argc, const
                     int validatorDbtype = db.validator->at(i);
                     if (validatorDbtype == Parameters::DBTYPE_STDIN) {
                         dbtypeFound = (filenames[fileIdx] == "stdin");
+                    } else if (validatorDbtype == Parameters::DBTYPE_URI) {
+                        regex_t regex;
+                        compileRegex(&regex, "[a-zA-Z][a-zA-Z0-9+-.]*:\\/\\/");
+                        int nomatch = regexec(&regex, filenames[fileIdx].c_str(), 0, NULL, 0);
+                        regfree(&regex);
+                        dbtypeFound = nomatch == false;
                     } else if (validatorDbtype == Parameters::DBTYPE_FLATFILE) {
                         dbtypeFound = (FileUtil::fileExists(filenames[fileIdx].c_str()) == true &&
                                        FileUtil::directoryExists(filenames[fileIdx].c_str()) == false);
diff --git a/src/commons/Parameters.h b/src/commons/Parameters.h
index fe5eb480e..74aeedd0c 100644
--- a/src/commons/Parameters.h
+++ b/src/commons/Parameters.h
@@ -82,6 +82,7 @@ class Parameters {
     static const int DBTYPE_FLATFILE = 17; // needed for verification
     static const int DBTYPE_SEQTAXDB = 18; // needed for verification
     static const int DBTYPE_STDIN = 19; // needed for verification
+    static const int DBTYPE_URI = 20; // needed for verification
 
     static const unsigned int DBTYPE_EXTENDED_COMPRESSED = 1;
     static const unsigned int DBTYPE_EXTENDED_INDEX_NEED_SRC = 2;
@@ -1152,6 +1153,7 @@ class Parameters {
             case DBTYPE_DIRECTORY: return "Directory";
             case DBTYPE_FLATFILE: return "Flatfile";
             case DBTYPE_STDIN: return "stdin";
+            case DBTYPE_URI: return "uri";
 
             default: return "Unknown";
         }