Merge pull request #5461 from hannes-steffenhagen-diffblue/cmdline-spell-correction

Add spell correction for CLI options

Author: hannes-steffenhagen-diffblue

regression/cbmc/unknown-argument-suggestion/test.desc

@@ -0,0 +1,12 @@
+CORE
+dummy.c
+--traec
+did you mean --trace
+^EXIT=(1|64)$
+^SIGNAL=0$
+--
+--
+This checks that we get a useful suggestion when we make a typo on the
+commandline.
+
+The error code is 1 on linux/osx and 64 on windows for some reason.

regression/validate-trace-xml-schema/check.py

@@ -33,6 +33,7 @@
     # these test for invalid command line handling
     'bad_option/test_multiple.desc',
     'bad_option/test.desc',
+    'unknown-argument-suggestion/test.desc',
     # this one produces XML intermingled with main XML output when used with --xml-ui
     'graphml_witness2/test.desc',
     # produces intermingled XML on the command line

src/util/Makefile

@@ -11,6 +11,7 @@ SRC = allocate_objects.cpp \
       cout_message.cpp \
       dstring.cpp \
       endianness_map.cpp \
+      edit_distance.cpp \
       expr.cpp \
       expr_initializer.cpp \
       expr_util.cpp \

src/util/cmdline.cpp

@@ -8,6 +8,7 @@ Author: Daniel Kroening, [email protected]
 
 #include "cmdline.h"
 
+#include <util/edit_distance.h>
 #include <util/exception_utils.h>
 #include <util/invariant.h>
 
@@ -250,6 +251,66 @@ cmdlinet::option_namest cmdlinet::option_names() const
   return option_namest{*this};
 }
 
+std::vector<std::string>
+cmdlinet::get_argument_suggestions(const std::string &unknown_argument)
+{
+  struct suggestiont
+  {
+    std::size_t distance;
+    std::string suggestion;
+
+    bool operator<(const suggestiont &other) const
+    {
+      return distance < other.distance;
+    }
+  };
+
+  auto argument_suggestions = std::vector<suggestiont>{};
+  // We allow 3 errors here. This can lead to the output being a bit chatty,
+  // which we mitigate by reducing suggestions to those with the minimum
+  // distance further down below
+  const auto argument_matcher = levenshtein_automatont{unknown_argument, 3};
+  for(const auto &option : options)
+  {
+    if(option.islong)
+    {
+      const auto long_name = "--" + option.optstring;
+      if(auto distance = argument_matcher.get_edit_distance(long_name))
+      {
+        argument_suggestions.push_back({distance.value(), long_name});
+      }
+    }
+    if(!option.islong)
+    {
+      const auto short_name = std::string{"-"} + option.optchar;
+      if(auto distance = argument_matcher.get_edit_distance(short_name))
+      {
+        argument_suggestions.push_back({distance.value(), short_name});
+      }
+    }
+  }
+
+  auto final_suggestions = std::vector<std::string>{};
+  if(!argument_suggestions.empty())
+  {
+    // we only want to keep suggestions with the minimum distance
+    // because otherwise they become quickly too noisy to be useful
+    auto min = std::min_element(
+      argument_suggestions.begin(), argument_suggestions.end());
+    INVARIANT(
+      min != argument_suggestions.end(),
+      "there is a minimum because it's not empty");
+    for(auto const &suggestion : argument_suggestions)
+    {
+      if(suggestion.distance == min->distance)
+      {
+        final_suggestions.push_back(suggestion.suggestion);
+      }
+    }
+  }
+  return final_suggestions;
+}
+
 cmdlinet::option_namest::option_names_iteratort::option_names_iteratort(
   const cmdlinet *command_line,
   std::size_t index)

src/util/cmdline.h

@@ -94,6 +94,9 @@ class cmdlinet
   cmdlinet();
   virtual ~cmdlinet();
 
+  std::vector<std::string>
+  get_argument_suggestions(const std::string &unknown_argument);
+
 protected:
   struct optiont
   {

src/util/edit_distance.cpp

@@ -0,0 +1,73 @@
+/// \file
+/// \author Diffblue Ltd.
+///
+/// Provides a way to compute edit distance between two strings
+
+#include "edit_distance.h"
+
+levenshtein_automatont::levenshtein_automatont(
+  const std::string &string,
+  std::size_t allowed_errors)
+{
+  const std::size_t layer_offset = string.size() + 1;
+  for(std::size_t i = 0; i <= allowed_errors; ++i)
+  {
+    final_states.push_back(string.size() + layer_offset * i);
+  }
+  for(std::size_t string_index = 0; string_index < string.size();
+      ++string_index)
+  {
+    for(std::size_t error_layer = 0; error_layer <= allowed_errors;
+        ++error_layer)
+    {
+      // position string_index matches
+      nfa.add_transition(
+        error_layer * layer_offset + string_index,
+        string[string_index],
+        error_layer * layer_offset + string_index + 1);
+      if(error_layer < allowed_errors)
+      {
+        // insertion, swap or deletion
+        nfa.add_arbitrary_transition(
+          error_layer * layer_offset + string_index,
+          (error_layer + 1) * layer_offset + string_index);
+        nfa.add_epsilon_transition(
+          error_layer * layer_offset + string_index,
+          (error_layer + 1) * layer_offset + string_index + 1);
+        nfa.add_arbitrary_transition(
+          error_layer * layer_offset + string_index,
+          (error_layer + 1) * layer_offset + string_index + 1);
+      }
+    }
+  }
+  for(std::size_t error_layer = 0; error_layer < allowed_errors; ++error_layer)
+  {
+    // arbitrary transitions between error layers
+    nfa.add_arbitrary_transition(
+      error_layer * layer_offset + string.size(),
+      (error_layer + 1) * layer_offset + string.size());
+  }
+}
+
+bool levenshtein_automatont::matches(const std::string &string) const
+{
+  return get_edit_distance(string).has_value();
+}
+
+optionalt<std::size_t>
+levenshtein_automatont::get_edit_distance(const std::string &string) const
+{
+  auto current = nfa.initial_state(0);
+  for(const auto c : string)
+  {
+    current = nfa.next_state(current, c);
+  }
+  for(std::size_t distance = 0; distance < final_states.size(); ++distance)
+  {
+    if(current.contains(final_states[distance]))
+    {
+      return distance;
+    }
+  }
+  return nullopt;
+}

src/util/edit_distance.h

@@ -0,0 +1,46 @@
+/// \file
+/// \author Diffblue Ltd.
+///
+/// Loosely based on this blog post:
+///   http://blog.notdot.net/2010/07/Damn-Cool-Algorithms-Levenshtein-Automata
+/// Provides a way to compute edit distance between two strings
+///
+/// No conversion to DFA or other optimisations are done here because for our
+/// use case (i.e. suggestions for errors in command line specifications) this
+/// is fast enough without them.
+
+#ifndef CPROVER_UTIL_EDIT_DISTANCE_H
+#define CPROVER_UTIL_EDIT_DISTANCE_H
+
+#include "nfa.h"
+
+#include <cstddef>
+#include <string>
+
+#include <util/optional.h>
+
+/// Simple automaton that can detect whether a string can be transformed into
+/// another with a limited number of deletions, insertions or substitutions.
+/// Not a very fast implementation, but should be good enough for small strings.
+struct levenshtein_automatont
+{
+private:
+  nfat<char> nfa;
+  using state_labelt = nfat<char>::state_labelt;
+  std::vector<state_labelt> final_states;
+
+public:
+  levenshtein_automatont(
+    const std::string &string,
+    std::size_t allowed_errors = 2);
+
+  bool matches(const std::string &string) const;
+  optionalt<std::size_t> get_edit_distance(const std::string &string) const;
+
+  void dump_automaton_dot_to(std::ostream &out)
+  {
+    nfa.dump_automaton_dot_to(out);
+  };
+};
+
+#endif // CPROVER_UTIL_EDIT_DISTANCE_H