Merge branch 'master' of github.com:smithlabcode/smithlab_cpp

Author: andrewdavidsmith

GenomicRegion.cpp

@@ -39,8 +39,7 @@ unordered_map<chrom_id_type, string> SimpleGenomicRegion::fw_table_out;
 
 chrom_id_type
 SimpleGenomicRegion::assign_chrom(const std::string &c) {
-  unordered_map<string, chrom_id_type>::const_iterator
-    chr_id(fw_table_in.find(c));
+  auto chr_id(fw_table_in.find(c));
   if (chr_id == fw_table_in.end()) {
     const chrom_id_type r = fw_table_in.size();
     fw_table_in[c] = r;
@@ -54,38 +53,15 @@ SimpleGenomicRegion::assign_chrom(const std::string &c) {
 
 string
 SimpleGenomicRegion::retrieve_chrom(chrom_id_type i) {
-  unordered_map<chrom_id_type, string>::const_iterator chr_name(fw_table_out.find(i));
-  assert(chr_name != fw_table_out.end());
+  auto chr_name(fw_table_out.find(i));
+  // assert(chr_name != fw_table_out.end());
   return chr_name->second;
 }
 
 
 SimpleGenomicRegion::SimpleGenomicRegion(const GenomicRegion &r) :
   chrom(assign_chrom(r.get_chrom())), start(r.get_start()), end(r.get_end()) {}
 
-// SimpleGenomicRegion::SimpleGenomicRegion(string string_representation) {
-//   vector<string> parts = smithlab::split_whitespace_quoted(string_representation);
-
-//   // make sure there is the minimal required info
-//   if (parts.size() < 3)
-//     throw runtime_error("Invalid string representation: " +
-//                                  string_representation);
-//   // set the chromosome name
-//   chrom = assign_chrom(parts[0]);
-
-//   // set the start position
-//   const int checkChromStart = atoi(parts[1].c_str());
-//   if (checkChromStart < 0)
-//     throw runtime_error("Invalid start: " + parts[1]);
-//   else start = static_cast<size_t>(checkChromStart);
-
-//   // set the end position
-//   const int checkChromEnd = atoi(parts[2].c_str());
-//   if (checkChromEnd < 0)
-//     throw runtime_error("Invalid end: " + parts[2]);
-//   else end = static_cast<size_t>(checkChromEnd);
-// }
-
 SimpleGenomicRegion::SimpleGenomicRegion(const char *s, const size_t len) {
   size_t i = 0;
 
@@ -182,7 +158,7 @@ unordered_map<chrom_id_type, string> GenomicRegion::fw_table_out;
 
 chrom_id_type
 GenomicRegion::assign_chrom(const std::string &c) {
-  unordered_map<string, chrom_id_type>::const_iterator chr_id(fw_table_in.find(c));
+  auto chr_id(fw_table_in.find(c));
   if (chr_id == fw_table_in.end()) {
     const chrom_id_type r = fw_table_in.size();
     fw_table_in[c] = r;
@@ -200,38 +176,6 @@ GenomicRegion::retrieve_chrom(chrom_id_type i) {
 }
 
 
-// GenomicRegion::GenomicRegion(string string_representation) : strand('+') {
-//   vector<string> parts(smithlab::split_whitespace_quoted(string_representation));
-
-//   // make sure there is the minimal required info
-//   if (parts.size() < 3)
-//     throw runtime_error("Invalid string representation: " +
-//                                  string_representation);
-//   // set the chromosome name
-//   chrom = assign_chrom(parts[0]);
-
-//   // set the start position
-//   const int checkChromStart = atoi(parts[1].c_str());
-//   if (checkChromStart < 0)
-//     throw runtime_error("Invalid start: " + parts[1]);
-//   else start = static_cast<size_t>(checkChromStart);
-
-//   // set the end position
-//   const int checkChromEnd = atoi(parts[2].c_str());
-//   if (checkChromEnd < 0)
-//     throw runtime_error("Invalid end: " + parts[2]);
-//   else end = static_cast<size_t>(checkChromEnd);
-
-//   if (parts.size() > 3)
-//     name = parts[3];
-
-//   if (parts.size() > 4)
-//     score = atof(parts[4].c_str());
-
-//   if (parts.size() > 5)
-//     strand = parts[5][0];
-// }
-
 GenomicRegion::GenomicRegion(const char *s, const size_t len) {
   size_t i = 0;
 
@@ -300,33 +244,6 @@ GenomicRegion::tostring() const {
   return s.str();
 }
 
-// std::ostream&
-// operator<<(std::ostream& s, const GenomicRegion& region) {
-//   return s << region.tostring();
-// }
-
-// std::istream&
-// operator>>(std::istream& s, GenomicRegion& region) {
-//   string chrom, name;
-//   size_t start = 0ul, end = 0ul;
-//   double score = 0.0;
-//   char strand = '\0';
-
-//   if (s >> chrom >> start >> end >> name >> score >> strand)
-//     region = GenomicRegion(chrom, start, end, name, score, strand);
-//   else region = GenomicRegion();
-
-//   char c;
-//   while ((c = s.get()) != '\n' && s);
-
-//   if (c != '\n')
-//     s.setstate(std::ios::badbit);
-
-//   if (s.eof())
-//     s.setstate(std::ios::badbit);
-
-//   return s;
-// }
 
 bool
 GenomicRegion::contains(const GenomicRegion& other) const {
@@ -438,8 +355,6 @@ separate_chromosomes(const vector<GenomicRegion>& regions,
 }
 
 
-
-
 static bool
 is_header_line(const string& line) {
   static const char *browser_label = "browser";

GenomicRegion.hpp

@@ -39,20 +39,12 @@ class GenomicRegion;
 
 class SimpleGenomicRegion {
 public:
-  // Cannonical
   SimpleGenomicRegion() : chrom(assign_chrom("(null)")), start(0), end(0) {}
   void swap(SimpleGenomicRegion &rhs) {
     std::swap(chrom, rhs.chrom);
     std::swap(start, rhs.start);
     std::swap(end, rhs.end);
   }
-  // SimpleGenomicRegion &operator=(const SimpleGenomicRegion &rhs) {
-  //   SimpleGenomicRegion tmp(rhs);
-  //   swap(tmp);
-  //   return *this;
-  // }
-  // SimpleGenomicRegion(const SimpleGenomicRegion &rhs) :
-  //   chrom(rhs.chrom), start(rhs.start), end(rhs.end) {}
 
   // Other constructors
   SimpleGenomicRegion(std::string c, size_t sta, size_t e) :
@@ -136,14 +128,6 @@ class GenomicRegion {
     std::swap(score, rhs.score);
     std::swap(strand, rhs.strand);
   }
-  // GenomicRegion(const GenomicRegion &other) :
-  //   chrom(other.chrom), name(other.name), start(other.start), end(other.end),
-  //   score(other.score), strand(other.strand) {}
-  // GenomicRegion &operator=(const GenomicRegion &rhs) {
-  //   GenomicRegion tmp(rhs);
-  //   swap(tmp);
-  //   return *this;
-  // }
 
   // Other constructors
   GenomicRegion(std::string c, size_t sta, size_t e,
@@ -432,5 +416,4 @@ assemble_region_name(const T &region, const std::string sep) {
           smithlab::toa(region.get_end()));
 }
 
-
 #endif

Makefile.am

@@ -18,7 +18,7 @@ EXTRA_DIST = README.md
 
 ACLOCAL_AMFLAGS = -I m4
 
-CXXFLAGS = -O3
+CXXFLAGS = -Wall -O3
 
 lib_LIBRARIES = libsmithlab_cpp.a
 

OptionParser.cpp

@@ -32,7 +32,6 @@
 #include <cstring>
 #include <cctype>
 #include <functional>
-#include <regex>
 #include <iterator>
 
 #include "smithlab_utils.hpp"
@@ -41,7 +40,6 @@ using std::vector;
 using std::string;
 using std::endl;
 using std::runtime_error;
-using std::regex;
 using std::begin;
 using std::end;
 
@@ -346,13 +344,23 @@ bool valid_option_char(char ch) {
   return std::isalnum(ch) || ch == '_';
 }
 
+static void
+fix_whitespace(string &s) {
+  std::istringstream iss(s);
+  string token;
+  s.clear();
+  while (iss >> token) {
+    if (!s.empty())
+      s += ' ';
+    s += token;
+  }
+}
+
 static void
 read_config_file(const string &config_filename,
                  vector<string> &config_file_options) {
   static const char comment_character = '#';
   static const char separator_character = ':';
-  static const string outer_space = "^[:space:]+|[:space:]+$";
-  static const string inner_space = "([:space:])[:space:]+";
 
   config_file_options.clear();
 
@@ -368,8 +376,7 @@ read_config_file(const string &config_filename,
       throw runtime_error("failed to config line from " + config_filename);
 
     // remove leading and trailing space
-    line = regex_replace(line, regex(outer_space), "");
-    line = regex_replace(line, regex(inner_space), " ");
+    fix_whitespace(line);
 
     if (!line.empty() && line.front() != comment_character) {
 
@@ -386,7 +393,7 @@ read_config_file(const string &config_filename,
 
       string option_value(line.substr(sep_pos + 1));
       // remove leading space
-      option_value = regex_replace(line, regex(outer_space), "");
+      fix_whitespace(option_value);
 
       if (!all_of(begin(option_value), end(option_value), valid_option_char))
         throw runtime_error("bad option label: " + line);

bisulfite_utils.hpp

@@ -22,6 +22,7 @@
 
 #include <string>
 #include <random>
+#include <cstdint>
 
 #include "sam_record.hpp"
 

dna_four_bit.hpp

@@ -118,24 +118,35 @@ struct genome_four_bit_itr {
     offset = (offset - 1) & 15ul;
     return *this;
   }
+
   genome_four_bit_itr operator--(int) {
     genome_four_bit_itr tmp(*this);
     itr -= (offset == 0);
     offset = (offset - 1) & 15ul;
     return tmp;
   }
+
   genome_four_bit_itr operator+(const size_t step) const {
     // whether the sum of offsets is >= 16
     const bool shift_one_pos =
-      (((offset + (static_cast<int>(step) & 15)) & 16) >> 4);
+      ((offset + (static_cast<int>(step) & 15)) & 16) >> 4;
 
     const int new_offset = (offset + step) & 15;
-    return genome_four_bit_itr(itr + step/16 + shift_one_pos,
-                               new_offset);
+    return genome_four_bit_itr(itr + step / 16 + shift_one_pos, new_offset);
   }
+
   bool operator!=(const genome_four_bit_itr &rhs) const {
     return itr != rhs.itr || offset != rhs.offset;
   }
+
+  bool operator<(const genome_four_bit_itr &rhs) const {
+    return itr < rhs.itr || (itr == rhs.itr && offset < rhs.offset);
+  }
+
+  bool operator<=(const genome_four_bit_itr &rhs) const {
+    return itr < rhs.itr || (itr == rhs.itr && offset <= rhs.offset);
+  }
+
   std::vector<size_t>::const_iterator itr;
   int offset;
 };