Sorting optionsHEADmaster

1 files changed, 55 insertions, 1 deletions

diff --git a/tabulate.cpp b/tabulate.cpp
index 7b262da..3099636 100644
--- a/tabulate.cpp
+++ b/tabulate.cpp
@@ -4,12 +4,14 @@
 #include <string_view>
 #include <unordered_set>
 #include <unordered_map>
+#include <algorithm>
 #include <map>
 #include <tuple>
 #include <memory>
 #include <cstdlib>
 #include <cstring>
 #include <cerrno>
+#include <cassert>
 
 
 static void usage(const char *argv0) {
@@ -30,7 +32,9 @@ static void usage(const char *argv0) {
     "\n"
     "Options:\n"
     "  -h        Show help.\n"
-    "  -s DELIM  The character that separates fields. Default space (' ').\n";
+    "  -s DELIM  The character that separates fields. Default space (' ').\n"
+    "  -S        Sort table, row and column labels lexicographically."
+    "  -n        Sort table, row and column labels numerically.";
 }
 
 struct Range {
@@ -141,8 +145,29 @@ std::ostream& operator<<(std::ostream &os, AlignBase<T, Dir::right> al) {
   return os << Spaces(al.outw - al.sw) << al.s;
 }
 
+enum class Sort {
+  no, lex, num
+};
+
+bool sort_lessthan(Sort type, const std::string &a, const std::string &b) {
+  switch (type) {
+    case Sort::no: assert(false);
+    case Sort::num: {
+      int an = 0, bn = 0;
+      try { an = std::stoi(a); } catch (...) {}
+      try { bn = std::stoi(b); } catch (...) {}
+      if (an != bn) return an < bn;
+    }
+    // fallthrough
+    case Sort::lex:
+      return a < b;
+  }
+  assert(false);
+}
+
 int main(int argc, char **argv) {
   char sepchar = ' ';
+  Sort opt_sort = Sort::no;
   std::vector<Fieldspec> fieldspecs;
   fieldspecs.reserve(4);
 
@@ -167,6 +192,8 @@ int main(int argc, char **argv) {
               return 1;
             }
             break;
+          case 'S': opt_sort = Sort::lex; break;
+          case 'n': opt_sort = Sort::num; break;
           default:
             std::cerr << "Invalid option '-" << argv[i][j] << "'" << std::endl;
             return 1;
@@ -256,6 +283,33 @@ int main(int argc, char **argv) {
     }
   }
 
+  if (opt_sort != Sort::no) {
+    std::vector<int> renumber[3];
+    for (int key = 0; key < 3; key++) {
+      const size_t N = labels[key].size();
+      renumber[key].resize(N);
+      for (size_t i = 0; i < N; i++) renumber[key][i] = i;
+      std::sort(renumber[key].begin(), renumber[key].end(), [&](int a, int b) {
+        return sort_lessthan(opt_sort, *labels[key][a], *labels[key][b]);
+      });
+
+      std::vector<std::unique_ptr<std::string>> newlabels;
+      newlabels.reserve(N);
+      for (size_t i = 0; i < N; i++)
+        newlabels.push_back(std::move(labels[key][renumber[key][i]]));
+      labels[key] = std::move(newlabels);
+    }
+
+    std::map<std::tuple<int, int, int>, std::string> newvalues;
+    for (auto &pair : values) {
+      auto [i0, i1, i2] = pair.first;
+      std::tuple<int, int, int> k = std::make_tuple(
+        renumber[0][i0], renumber[1][i1], renumber[2][i2]);
+      newvalues.emplace(k, std::move(pair.second));
+    }
+    values = std::move(newvalues);
+  }
+
   int leftwid = 0;  // does not include table name
   for (const std::unique_ptr<std::string> &rowname : labels[ROW])
     leftwid = std::max<int>(leftwid, swidth(*rowname));