PD-5181: StxTyper ver. 1.0.28

ncbi · Dec 9, 2024 · 4184b32 · 4184b32
1 parent e1e9d05
commit 4184b32
Show file tree

Hide file tree

Showing 9 changed files with 151 additions and 104 deletions.
diff --git a/amrfinder.cpp b/amrfinder.cpp
@@ -33,6 +33,7 @@
 * Dependencies: NCBI BLAST, HMMer, libcurl, gunzip (optional)
 *
 * Release changes:
+*   4.0.6   12/09/2024 PD-5181  StxTyper ver. 1.0.28
 *   4.0.5   11/13/2024 PD-5175  prohibit --database_version and -p or -n
 *   4.0.4   10/29/2024          colorizeDir()
 *                               stxtyper/ -> stx/
@@ -338,7 +339,7 @@ using namespace GFF_sp;
 const string dataVer_min ("2024-08-14.2");
   // 3.12: "2023-12-15.2"
   // 3.11: "2021-02-18.1"  
-const string stxTyperVersion ("1.0.27");  
+const string stxTyperVersion ("1.0.28");  
 
 
 

diff --git a/common.cpp b/common.cpp
@@ -724,6 +724,35 @@ bool strBlank (const string &s)
 
 
 
+bool getScientific (string numberS,
+                    bool &hasPoint,
+                    streamsize &decimals)
+{
+  strUpper (numberS);
+  const size_t ePos     = numberS. find ('E');
+  const size_t pointPos = numberS. find ('.');
+
+  hasPoint = (pointPos != string::npos);  
+  decimals = 0;
+  if (ePos == string::npos)
+  {
+    if (hasPoint)
+      decimals = (streamoff) (numberS. size () - (pointPos + 1));
+  }
+  else
+  {
+    ASSERT (ePos != pointPos);
+    if (hasPoint && ePos < pointPos)
+      hasPoint = false;
+    if (hasPoint)
+      decimals = (streamoff) (ePos - (pointPos + 1));
+  }
+
+  return ePos != string::npos;
+}
+
+
+
 void strUpper (string &s)
 {
   for (char& c : s)
@@ -760,14 +789,23 @@ bool isLower (const string &s)
 
 
 
-string::const_iterator stringInSet (const string &s,
-                    	           	  const string &charSet)
+size_t stringNotInSet (const string &s,
+                   	   const string &charSet,
+                   	   ebool uppercase)
 {
-  CONST_ITER (string, it, s)
-    if (! charInSet (*it, charSet))
-      return it;
-
-  return s. end ();
+  FFOR (size_t, i, s. size ())
+  {
+    char c = s [i];
+    switch (uppercase)
+    {
+      case efalse: c = toLower (c); break;
+      case etrue:  c = toUpper (c); break;
+      case enull: break;
+    }
+    if (! charInSet (c, charSet))
+      return i;
+  }
+  return no_index;
 }
 
 
@@ -4093,7 +4131,12 @@ int Application::run (int argc,
       string logFName (getArg ("log"));
     	ASSERT (! logPtr);
       if (! logFName. empty ())
-    		logPtr = new ofstream (logFName, ios_base::app);
+      {
+    		logPtr = new ofstream (logFName, ios_base::app);    		  
+        time_t rawtime;
+        time (& rawtime);
+    		*logPtr << endl << ctime (& rawtime) << "$ " << getCommandLine () << endl;
+      }
     }
 
     string jsonFName;

diff --git a/common.hpp b/common.hpp
@@ -418,9 +418,9 @@ template <typename To, typename From>
   inline void insertAll (To &to,
                          const From &from)
     { 
-      #pragma GCC diagnostic ignored "-Waggressive-loop-optimizations"
+    #pragma GCC diagnostic ignored "-Waggressive-loop-optimizations"
       to. insert (to. begin (), from. begin (), from. end ()); 
-      #pragma GCC diagnostic warning "-Waggressive-loop-optimizations"
+    #pragma GCC diagnostic warning "-Waggressive-loop-optimizations"
     }
 
 template <typename To, typename From>
@@ -649,6 +649,10 @@ inline string unQuote (const string &s)
 
 bool strBlank (const string &s);
 
+bool getScientific (string numberS,
+                    bool &hasPoint,
+                    streamsize &decimals);
+
 template <typename T>
   string toString (const T &t)
     { ostringstream oss;
@@ -677,28 +681,6 @@ template <typename T>
         catch (...) { return false; } 
     }
 
-inline bool isLeft (const string &s,
-                    const string &left)
-  { return s. substr (0, left. size ()) == left; }
-
-bool isRight (const string &s,
-              const string &right);
-
-bool trimPrefix (string &s,
-                 const string &prefix);
-  // Return: success
-
-bool trimSuffix (string &s,
-                 const string &suffix);
-  // Return: success
-
-void trimSuffixNonAlphaNum (string &s);
-
-bool trimTailAt (string &s,
-                 const string &tailStart);
-  // Return: trimmed
-  // Update: s
-
 void commaize (string &s);
   // ' ' --> ','
 
@@ -742,16 +724,39 @@ inline bool charInSet (char c,
 		                   const string &charSet)
   { return charSet. find (c) != string::npos; }
 
-string::const_iterator stringInSet (const string &s,
-                    	           	  const string &charSet);
-  // Return: != s.end() => *Return is not in charSet
+size_t stringNotInSet (const string &s,
+                   	   const string &charSet,
+                   	   ebool uppercase);
+  // Return: index in s which is not in charSet; may be no_index
 
 size_t strCountSet (const string &s,
 		                const string &charSet);
 
 void strDeleteSet (string &s,
 		               const string &charSet);
 
+inline bool isLeft (const string &s,
+                    const string &left)
+  { return s. substr (0, left. size ()) == left; }
+
+bool isRight (const string &s,
+              const string &right);
+
+bool trimPrefix (string &s,
+                 const string &prefix);
+  // Return: success
+
+bool trimSuffix (string &s,
+                 const string &suffix);
+  // Return: success
+
+void trimSuffixNonAlphaNum (string &s);
+
+bool trimTailAt (string &s,
+                 const string &tailStart);
+  // Return: trimmed
+  // Update: s
+
 void trimLeading (string &s);
 
 void trimTrailing (string &s);
@@ -775,6 +780,17 @@ inline void trim (string &s,
     trimLeading  (s, c); 
   }
 
+inline bool strNull (const string &s)
+  { if (strBlank (s))
+      return true;
+    string s1 (s);
+    trim (s1);
+    strUpper (s1);
+    return    s1 == "NULL"
+           || s1 == "NA"
+           || s1 == "N/A";
+  }
+
 inline bool contains (const string &hay,
                       const string &needle)
   { return hay. find (needle) != string::npos; }
@@ -1117,15 +1133,15 @@ template <typename T>
         { *this << other; }
     bool empty () const
       {
-        #pragma GCC diagnostic ignored "-Wnull-dereference"
+      #pragma GCC diagnostic ignored "-Wnull-dereference"
         return P::empty ();
-        #pragma GCC diagnostic warning "-Wnull-dereference"
+      #pragma GCC diagnostic warning "-Wnull-dereference"
       }
     size_t size () const
       {
-        #pragma GCC diagnostic ignored "-Wnull-dereference"
+      #pragma GCC diagnostic ignored "-Wnull-dereference"
         return P::size ();
-        #pragma GCC diagnostic warning "-Wnull-dereference"
+      #pragma GCC diagnostic warning "-Wnull-dereference"
       }
 
 
@@ -2806,7 +2822,7 @@ template <typename T /* : Root */>
         }
   	void deleteData ()
   	  {	for (const T* t : *this)
-  			  delete t;
+			    delete t;
   			P::clear ();
   	  }
     void erasePtr (size_t index)
@@ -2920,7 +2936,11 @@ template <typename T /* : Root */>
   	    return *this;
   	  }
    ~VectorOwn ()
-      { P::deleteData (); }
+      { 
+      #pragma GCC diagnostic ignored "-Wnull-dereference"
+        P::deleteData (); 
+      #pragma GCC diagnostic warning "-Wnull-dereference"
+      }
 
 
     VectorOwn<T>& operator<< (const T* value)
@@ -3221,9 +3241,7 @@ struct StringVector : Vector<string>
     {}
 
 
-  string toString (const string& sep) const;
-  string toString () const
-    { return toString (noString); }
+  string toString (const string& sep = noString) const;
   bool same (const StringVector &vec,
              const Vector<size_t> &indexes) const;
   void to_xml (Xml::File &f,

diff --git a/seq.cpp b/seq.cpp
@@ -194,9 +194,9 @@ void Seq::qcName () const
 
 void Seq::qcAlphabet () const
 {
-  const string::const_iterator it = stringInSet (seq, getSeqAlphabet ());
-  if (it != seq. end ())	
-    throw runtime_error ("Bad sequence character in " + strQuote (getId ()) + ": " + to_string (int (*it)) + " " + seq. substr ((size_t) (it - seq. begin ())));
+  const size_t i = stringNotInSet (seq, getSeqAlphabet (), etrue);
+  if (i != no_index)	
+    throw runtime_error ("Bad sequence character in " + strQuote (getId ()) + ": ASCII=" + to_string (int (seq [i])) + " pos=" + to_string (i + 1) /*+ "\n" + seq*/);
 }
 
 
@@ -3291,6 +3291,11 @@ AlignScore SubstMat::char2score (char c1,
 { 
   QC_ASSERT (c1 >= '\0');
   QC_ASSERT (c2 >= '\0');
+
+  if (strchr (peptideWildcards, c1))
+    c1 = 'X';
+  if (strchr (peptideWildcards, c2))
+    c2 = 'X';
 
   const size_t i1 = (size_t) c1;
   const size_t i2 = (size_t) c2;
@@ -3304,9 +3309,9 @@ AlignScore SubstMat::char2score (char c1,
 
 
   if (! goodIndex (i1))
-    throw runtime_error ("Bad amino acid: " + string (1, c1) + " (" + to_string (i1));
+    throw runtime_error ("Bad amino acid: " + string (1, c1) + " (" + to_string (i1) + ")");
   if (! goodIndex (i2))
-    throw runtime_error ("Bad amino acid: " + string (1, c2) + " (" + to_string (i2));
+    throw runtime_error ("Bad amino acid: " + string (1, c2) + " (" + to_string (i2) + ")");
 
   const AlignScore s = sim [i1] [i2]; 
   ASSERT (s == s);

diff --git a/seq.hpp b/seq.hpp
@@ -253,7 +253,7 @@ struct Seq : Named
 public:
   virtual Seq* copy () const override = 0;
   void qc () const override
-    { if (qc_on)
+    { if (! qc_on)
         return;
       qcName ();
     	qcAlphabet ();

diff --git a/stx b/stx
+25 −0		METHODS.md
+43 −0		README.md
+123 −69		common.cpp
+108 −84		common.hpp
+7 −33		stxtyper.cpp
+20 −44		tsv.cpp
+12 −8		tsv.hpp
+1 −1		version.txt