autfilt, dstar2tgba: add CSV input

Fixes #91.

* bin/autfilt.cc, bin/dstar2tgba.cc: Implement reading CSV files.
* bin/common_finput.cc: Fix comments.
* bin/common_aoutput.cc: Show %<, %> in help text.
* NEWS, doc/org/csv.org: Document it.
* tests/core/readsave.test: Add a short test case.

bin/autfilt.cc

@@ -24,6 +24,8 @@
 #include <limits>
 #include <set>
 #include <memory>
+#include <sys/stat.h>
+#include <unistd.h>
 
 #include <argp.h>
 #include "error.h"
@@ -874,6 +876,90 @@ namespace
       SPOT_UNREACHABLE();
     }
 
+    int process_string(const std::string& input, const char* filename,
+                       int linenum) override
+    {
+      std::ostringstream loc;
+      loc << filename << ':' << linenum;
+      std::string locstr = loc.str();
+      return process_automaton_stream
+        (spot::automaton_stream_parser(input.c_str(), locstr, opt_parse),
+         locstr.c_str());
+    }
+
+    int
+    aborted(const spot::const_parsed_aut_ptr& h, const char* filename)
+    {
+      std::cerr << filename << ':' << h->loc << ": aborted input automaton\n";
+      return 2;
+    }
+
+    int
+    process_file(const char* filename) override
+    {
+      // If we have a filename like "foo/NN" such
+      // that:
+      // ① foo/NN is not a file,
+      // ② NN is a number,
+      // ③ foo is a file,
+      // then it means we want to open foo as
+      // a CSV file and process column NN.
+      if (const char* slash = strrchr(filename, '/'))
+        {
+          char* end;
+          errno = 0;
+          long int col = strtol(slash + 1, &end, 10);
+          if (errno == 0 && !*end && col != 0)
+            {
+              struct stat buf;
+              if (stat(filename, &buf) != 0)
+                {
+                  col_to_read = col;
+                  if (real_filename)
+                    free(real_filename);
+                  real_filename = strndup(filename, slash - filename);
+
+                  // Special case for stdin.
+                  if (real_filename[0] == '-' && real_filename[1] == 0)
+                    return process_stream(std::cin, real_filename);
+
+                  std::ifstream input(real_filename);
+                  if (input)
+                    return process_stream(input, real_filename);
+
+                  error(2, errno, "cannot open '%s' nor '%s'",
+                        filename, real_filename);
+                }
+            }
+        }
+
+      return process_automaton_stream(spot::automaton_stream_parser(filename,
+                                                                    opt_parse),
+                                      filename);
+    }
+
+    int process_automaton_stream(spot::automaton_stream_parser&& hp,
+                                 const char* filename)
+    {
+      int err = 0;
+      while (!abort_run)
+        {
+          auto haut = hp.parse(opt->dict);
+          if (!haut->aut && haut->errors.empty())
+            break;
+          if (haut->format_errors(std::cerr))
+            err = 2;
+          if (!haut->aut)
+            error(2, 0, "failed to read automaton from %s", filename);
+          else if (haut->aborted)
+            err = std::max(err, aborted(haut, filename));
+          else
+            process_automaton(haut, filename);
+        }
+      return err;
+    }
+
+
     int
     process_automaton(const spot::const_parsed_aut_ptr& haut,
                       const char* filename)
@@ -1111,42 +1197,14 @@ namespace
 
       ++match_count;
 
-      printer.print(aut, nullptr, filename, -1, conversion_time, haut);
+      printer.print(aut, nullptr, filename, -1, conversion_time, haut,
+                    prefix, suffix);
 
       if (opt_max_count >= 0 && match_count >= opt_max_count)
         abort_run = true;
 
       return 0;
     }
-
-    int
-    aborted(const spot::const_parsed_aut_ptr& h, const char* filename)
-    {
-      std::cerr << filename << ':' << h->loc << ": aborted input automaton\n";
-      return 2;
-    }
-
-    int
-    process_file(const char* filename) override
-    {
-      auto hp = spot::automaton_stream_parser(filename, opt_parse);
-      int err = 0;
-      while (!abort_run)
-        {
-          auto haut = hp.parse(opt->dict);
-          if (!haut->aut && haut->errors.empty())
-            break;
-          if (haut->format_errors(std::cerr))
-            err = 2;
-          if (!haut->aut)
-            error(2, 0, "failed to read automaton from %s", filename);
-          else if (haut->aborted)
-            err = std::max(err, aborted(haut, filename));
-          else
-            process_automaton(haut, filename);
-        }
-      return err;
-    }
   };
 }
 
@@ -1155,7 +1213,7 @@ main(int argc, char** argv)
 {
   setup(argv);
 
-  const argp ap = { options, parse_opt, "[FILENAMES...]",
+  const argp ap = { options, parse_opt, "[FILENAME[/COL]...]",
                     argp_program_doc, children, nullptr, nullptr };
 
   try

bin/common_aoutput.cc

@@ -181,6 +181,12 @@ static const argp_option io_options[] =
       "one word accepted by the output automaton", 0 },
     { "%%", 0, nullptr, OPTION_DOC | OPTION_NO_USAGE,
       "a single %", 0 },
+    { "%<", 0, nullptr, OPTION_DOC | OPTION_NO_USAGE,
+      "the part of the line before the automaton if it "
+      "comes from a column extracted from a CSV file", 4 },
+    { "%>", 0, nullptr, OPTION_DOC | OPTION_NO_USAGE,
+      "the part of the line after the automaton if it "
+      "comes from a column extracted from a CSV file", 4 },
     { nullptr, 0, nullptr, 0, nullptr, 0 }
   };
 

bin/common_finput.cc

@@ -305,7 +305,7 @@ job_processor::process_file(const char* filename)
   // If we have a filename like "foo/NN" such
   // that:
   // ① foo/NN is not a file (already the case),
-  // ② NN is a number > 0,
+  // ② NN is a number,
   // ③ foo is a file,
   // then it means we want to open foo as
   // a CSV file and process column NN.
@@ -315,7 +315,6 @@ job_processor::process_file(const char* filename)
       char* end;
       errno = 0;
       long int col = strtol(slash + 1, &end, 10);
-      // strtol ate all remaining characters and NN is positive
       if (errno == 0 && !*end && col != 0)
         {
           col_to_read = col;

bin/dstar2tgba.cc

@@ -22,6 +22,8 @@
 #include <string>
 #include <iostream>
 #include <memory>
+#include <sys/stat.h>
+#include <unistd.h>
 
 #include <argp.h>
 #include "error.h"
@@ -123,19 +125,15 @@ namespace
       SPOT_UNREACHABLE();
     }
 
-    int
-    process_automaton(const spot::const_parsed_aut_ptr& haut,
-                      const char* filename)
+    int process_string(const std::string& input, const char* filename,
+                       int linenum) override
     {
-      spot::stopwatch sw;
-      sw.start();
-      auto nba = spot::to_generalized_buchi(haut->aut);
-      auto aut = post.run(nba, nullptr);
-      const double conversion_time = sw.stop();
-
-      printer.print(aut, nullptr, filename, -1, conversion_time, haut);
-      flush_cout();
-      return 0;
+      std::ostringstream loc;
+      loc << filename << ':' << linenum;
+      std::string locstr = loc.str();
+      return process_automaton_stream
+        (spot::automaton_stream_parser(input.c_str(), locstr, opt_parse),
+         locstr.c_str());
     }
 
     int
@@ -148,7 +146,50 @@ namespace
     int
     process_file(const char* filename) override
     {
-      auto hp = spot::automaton_stream_parser(filename, opt_parse);
+      // If we have a filename like "foo/NN" such
+      // that:
+      // ① foo/NN is not a file,
+      // ② NN is a number,
+      // ③ foo is a file,
+      // then it means we want to open foo as
+      // a CSV file and process column NN.
+      if (const char* slash = strrchr(filename, '/'))
+        {
+          char* end;
+          errno = 0;
+          long int col = strtol(slash + 1, &end, 10);
+          if (errno == 0 && !*end && col != 0)
+            {
+              struct stat buf;
+              if (stat(filename, &buf) != 0)
+                {
+                  col_to_read = col;
+                  if (real_filename)
+                    free(real_filename);
+                  real_filename = strndup(filename, slash - filename);
+
+                  // Special case for stdin.
+                  if (real_filename[0] == '-' && real_filename[1] == 0)
+                    return process_stream(std::cin, real_filename);
+
+                  std::ifstream input(real_filename);
+                  if (input)
+                    return process_stream(input, real_filename);
+
+                  error(2, errno, "cannot open '%s' nor '%s'",
+                        filename, real_filename);
+                }
+            }
+        }
+
+      return process_automaton_stream(spot::automaton_stream_parser(filename,
+                                                                    opt_parse),
+                                      filename);
+    }
+
+    int process_automaton_stream(spot::automaton_stream_parser&& hp,
+                                 const char* filename)
+    {
       int err = 0;
       while (!abort_run)
         {
@@ -166,6 +207,22 @@ namespace
         }
       return err;
     }
+
+
+    int
+    process_automaton(const spot::const_parsed_aut_ptr& haut,
+                      const char* filename)
+    {
+      spot::stopwatch sw;
+      sw.start();
+      auto nba = spot::to_generalized_buchi(haut->aut);
+      auto aut = post.run(nba, nullptr);
+      const double conversion_time = sw.stop();
+
+      printer.print(aut, nullptr, filename, -1, conversion_time, haut);
+      flush_cout();
+      return 0;
+    }
   };
 }
 
@@ -174,7 +231,7 @@ main(int argc, char** argv)
 {
   setup(argv);
 
-  const argp ap = { options, parse_opt, "[FILENAMES...]",
+  const argp ap = { options, parse_opt, "[FILENAME[/COL]...]",
                     argp_program_doc, children, nullptr, nullptr };
 
   if (int err = argp_parse(&ap, argc, argv, ARGP_NO_HELP, nullptr, nullptr))