[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[lmi-commits] [lmi] master aac9256 3/4: Separate input-sequence implemen
|
From: |
Greg Chicares |
|
Subject: |
[lmi-commits] [lmi] master aac9256 3/4: Separate input-sequence implementation into two TUs |
|
Date: |
Mon, 13 Feb 2017 17:57:38 -0500 (EST) |
branch: master
commit aac925671994c84584689d8bf86e739f8327fa16
Author: Gregory W. Chicares <address@hidden>
Commit: Gregory W. Chicares <address@hidden>
Separate input-sequence implementation into two TUs
Incidentally regularized one-line comments at top of all files.
Running
git commit --all --dry-run
beforehand revealed that git sees part of this commit as a rename:
renamed: input_sequence.cpp -> input_sequence_parser.cpp
It would be possible to prevent that, but it's not worth the trouble:
git log --follow
tracks history across renames, and git-2.6.0 offers
git config log.follow true
to make that desirable behavior automatic.
---
Makefile.am | 5 +
input_seq_test.cpp | 2 +-
input_sequence.cpp | 697 +-----------------------
input_sequence.hpp | 118 +---
input_sequence_interval.hpp | 60 ++
input_sequence.cpp => input_sequence_parser.cpp | 392 +------------
input_sequence_parser.hpp | 121 ++++
objects.make | 3 +
8 files changed, 203 insertions(+), 1195 deletions(-)
diff --git a/Makefile.am b/Makefile.am
index f852293..d0e9df6 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -323,6 +323,7 @@ liblmi_common_sources = \
input_realization.cpp \
input_seq_helpers.cpp \
input_sequence.cpp \
+ input_sequence_parser.cpp \
input_xml_io.cpp \
interest_rates.cpp \
ledger.cpp \
@@ -719,6 +720,7 @@ test_input_seq_SOURCES = \
$(common_test_objects) \
input_seq_test.cpp \
input_sequence.cpp \
+ input_sequence_parser.cpp \
miscellany.cpp
test_input_seq_CXXFLAGS = $(AM_CXXFLAGS)
@@ -744,6 +746,7 @@ test_input_SOURCES = \
input_realization.cpp \
input_seq_helpers.cpp \
input_sequence.cpp \
+ input_sequence_parser.cpp \
input_test.cpp \
input_xml_io.cpp \
mc_enum.cpp \
@@ -1163,6 +1166,8 @@ noinst_HEADERS = \
input_seq_helpers.hpp \
input_sequence.hpp \
input_sequence_entry.hpp \
+ input_sequence_interval.hpp \
+ input_sequence_parser.hpp \
interest_rates.hpp \
istream_to_string.hpp \
ledger.hpp \
diff --git a/input_seq_test.cpp b/input_seq_test.cpp
index 6d9492a..624069e 100644
--- a/input_seq_test.cpp
+++ b/input_seq_test.cpp
@@ -1,4 +1,4 @@
-// Input sequences e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0... : unit test.
+// Input sequences (e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0...): unit test
//
// Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011,
2012, 2013, 2014, 2015, 2016, 2017 Gregory W. Chicares.
//
diff --git a/input_sequence.cpp b/input_sequence.cpp
index a043c34..db94ecb 100644
--- a/input_sequence.cpp
+++ b/input_sequence.cpp
@@ -1,5 +1,4 @@
-// Input sequences e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0....
-// Extract the grammar from lines matching the regexp _// GRAMMAR_ .
+// Input sequences (e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0...)
//
// Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011,
2012, 2013, 2014, 2015, 2016, 2017 Gregory W. Chicares.
//
@@ -20,6 +19,8 @@
// email: <address@hidden>
// snail: Chicares, 186 Belle Woods Drive, Glastonbury CT 06033, USA
+// Extract the grammar from lines matching the regexp _// GRAMMAR_ .
+
#include "pchfile.hpp"
#include "input_sequence.hpp"
@@ -27,659 +28,14 @@
#include "alert.hpp"
#include "assert_lmi.hpp"
#include "contains.hpp"
-#include "miscellany.hpp"
+#include "input_sequence_parser.hpp"
#include "value_cast.hpp"
-#include <algorithm>
-#include <cctype>
-#include <ios>
-#include <istream>
-#include <iterator> // std::ostream_iterator
-#include <ostream>
+#include <algorithm> // std::fill()
#include <sstream>
#include <stdexcept>
#include <type_traits>
-SequenceParser::SequenceParser
- (std::string const& input_expression
- ,int a_years_to_maturity
- ,int a_issue_age
- ,int a_retirement_age
- ,int a_inforce_duration
- ,int a_effective_year
- ,std::vector<std::string> const& a_allowed_keywords
- ,bool a_keywords_only
- )
- :input_stream_ (input_expression.c_str())
- ,years_to_maturity_ (a_years_to_maturity)
- ,issue_age_ (a_issue_age)
- ,retirement_age_ (a_retirement_age)
- ,inforce_duration_ (a_inforce_duration)
- ,effective_year_ (a_effective_year)
- ,allowed_keywords_ (a_allowed_keywords)
- ,keywords_only_ (a_keywords_only)
-{
- sequence();
-}
-
-SequenceParser::~SequenceParser() = default;
-
-std::string SequenceParser::diagnostics() const
-{
- return diagnostics_.str();
-}
-
-std::vector<ValueInterval> const& SequenceParser::intervals() const
-{
- return intervals_;
-}
-
-std::string SequenceParser::token_type_name(SequenceParser::token_type t)
-{
- switch(t)
- {
- case 0:
- {
- return "end of input";
- }
- case e_major_separator:
- case e_minor_separator:
- case e_begin_incl:
- case e_begin_excl:
- case e_end_incl:
- case e_end_excl:
- case e_age_prefix:
- case e_cardinal_prefix:
- {
- std::string s;
- s += static_cast<char>(t);
- return s;
- }
- case e_startup:
- {
- return "beginning of input";
- }
- case e_number:
- {
- return "number";
- }
- case e_keyword:
- {
- return "keyword";
- }
- default:
- {
- return std::string("unknown: ") + static_cast<char>(t);
- }
- }
-}
-
-// GRAMMAR interval-begin: one of [ (
-// GRAMMAR interval-end: one of ] )
-
-// GRAMMAR duration-constant: one of inforce retirement maturity
-// SOMEDAY !! 'inforce' not yet implemented
-
-// GRAMMAR duration-scalar: integer
-// GRAMMAR duration-scalar: @ integer
-// GRAMMAR duration-scalar: # integer
-// GRAMMAR duration-scalar: duration-constant
-// SOMEDAY !! calendar year not yet implemented
-
-void SequenceParser::duration_scalar()
-{
- switch(current_token_type_)
- {
- case e_age_prefix:
- {
- current_duration_scalar_mode_ = e_attained_age;
- match(current_token_type_);
- }
- break;
- case e_cardinal_prefix:
- {
- current_duration_scalar_mode_ = e_number_of_years;
- match(current_token_type_);
- }
- break;
- case e_number:
- {
- current_duration_scalar_mode_ = e_duration;
- }
- break;
- case e_keyword:
- {
- if("retirement" == current_keyword_)
- {
- current_duration_scalar_mode_ = e_retirement;
- current_duration_scalar_ = retirement_age_ - issue_age_;
- match(e_keyword);
- return;
- }
- else if("maturity" == current_keyword_)
- {
- current_duration_scalar_mode_ = e_maturity;
- current_duration_scalar_ = years_to_maturity_;
- match(e_keyword);
- return;
- }
- else
- {
- diagnostics_
- << "Expected keyword "
- << "'retirement' "
- << " or 'maturity'. "
- ;
- mark_diagnostic_context();
- }
- }
- break;
- default:
- {
- diagnostics_ << "Expected keyword, number, '@', or '#'. ";
- mark_diagnostic_context();
- }
- }
-
- current_duration_scalar_ = static_cast<int>(current_number_);
- match(e_number);
-
- switch(current_duration_scalar_mode_)
- {
- case e_duration:
- {
- ; // Do nothing.
- }
- break;
- case e_attained_age:
- {
- current_duration_scalar_ -= issue_age_;
- }
- break;
- case e_number_of_years:
- {
- current_duration_scalar_ += last_input_duration_;
- }
- break;
- default:
- {
- diagnostics_ << "Expected number, '@', or '#'. ";
- mark_diagnostic_context();
- }
- }
-}
-
-void SequenceParser::null_duration()
-{
- int tentative_begin_duration = last_input_duration_++;
- duration_mode tentative_begin_duration_mode =
previous_duration_scalar_mode_;
- int tentative_end_duration = last_input_duration_;
- duration_mode tentative_end_duration_mode = e_duration;
- validate_duration
- (tentative_begin_duration
- ,tentative_begin_duration_mode
- ,tentative_end_duration
- ,tentative_end_duration_mode
- );
-}
-
-// GRAMMAR duration: duration-scalar
-
-void SequenceParser::single_duration()
-{
- duration_scalar();
- int tentative_begin_duration = last_input_duration_;
-// last_input_duration_ += static_cast<int>(current_duration_scalar_);
-// last_input_duration_ = static_cast<int>(current_duration_scalar_);
-// duration_mode tentative_begin_duration_mode = e_duration;
- duration_mode tentative_begin_duration_mode =
previous_duration_scalar_mode_;
- int tentative_end_duration = current_duration_scalar_;
-// int tentative_end_duration = last_input_duration_ +
static_cast<int>(current_duration_scalar_);
- duration_mode tentative_end_duration_mode =
current_duration_scalar_mode_;
- validate_duration
- (tentative_begin_duration
- ,tentative_begin_duration_mode
- ,tentative_end_duration
- ,tentative_end_duration_mode
- );
-}
-
-// GRAMMAR duration: interval-begin duration-scalar , duration-scalar
interval-end
-
-void SequenceParser::intervalic_duration()
-{
- bool begin_excl = (e_begin_excl == current_token_type_);
- match(current_token_type_);
- duration_scalar();
- // Add one to the interval-beginning if it was expressed
- // as exclusive, because we store [begin, end).
- int tentative_begin_duration =
- static_cast<int>(current_duration_scalar_)
- + begin_excl
- ;
- duration_mode tentative_begin_duration_mode =
current_duration_scalar_mode_;
- int tentative_end_duration = -1;
- duration_mode tentative_end_duration_mode = e_invalid_mode;
- match(e_minor_separator); // TODO ?? Require this?
- duration_scalar();
- if
- ( e_end_incl == current_token_type_
- || e_end_excl == current_token_type_
- )
- {
- bool end_incl = (e_end_incl == current_token_type_);
- match(current_token_type_);
- // Add one to the interval-end if it was expressed
- // as inclusive, because we store [begin, end).
- tentative_end_duration = current_duration_scalar_ + end_incl;
- tentative_end_duration_mode = current_duration_scalar_mode_;
- }
- else
- {
- diagnostics_ << "Expected ')' or ']'. ";
- mark_diagnostic_context();
- return;
- }
- validate_duration
- (tentative_begin_duration
- ,tentative_begin_duration_mode
- ,tentative_end_duration
- ,tentative_end_duration_mode
- );
-}
-
-void SequenceParser::validate_duration
- (int tentative_begin_duration
- ,duration_mode tentative_begin_duration_mode
- ,int tentative_end_duration
- ,duration_mode tentative_end_duration_mode
- )
-{
- if(tentative_begin_duration < 0)
- {
- current_interval_.insane = true;
- diagnostics_
- << "Interval "
- << "[ " << tentative_begin_duration << ", "
- << tentative_end_duration << " )"
- << " is improper: it "
- << "begins before the first possible duration. "
- ;
- mark_diagnostic_context();
- return;
- }
- // Remember that our paradigm is [begin, end), so that
- // [X, X) truly is improper: it
- // starts at X, and includes X, and
- // ends at X, and excludes X
- // so it both includes and excludes X. Thus, an interval
- // [B, E) is improper if B == E.
- else if(tentative_end_duration <= tentative_begin_duration)
- {
- current_interval_.insane = true;
- diagnostics_
- << "Interval "
- << "[ " << tentative_begin_duration << ", "
- << tentative_end_duration << " )"
- << " is improper: it "
- << "ends before it begins. "
- ;
- mark_diagnostic_context();
- return;
- }
- else if(years_to_maturity_ < tentative_end_duration)
- {
- current_interval_.insane = true;
- diagnostics_
- << "Interval "
- << "[ " << tentative_begin_duration << ", "
- << tentative_end_duration << " )"
- << " is improper: it "
- << "ends after the last possible duration. "
- ;
- mark_diagnostic_context();
- return;
- }
- else if
- ( e_invalid_mode == tentative_begin_duration_mode
- || e_invalid_mode == tentative_end_duration_mode
- )
- {
- current_interval_.insane = true;
- diagnostics_
- << "Interval "
- << "[ " << tentative_begin_duration << ", "
- << tentative_end_duration << " )"
- << " has an invalid duration mode. "
- ;
- mark_diagnostic_context();
- return;
- }
- else
- {
- current_interval_.begin_duration = tentative_begin_duration ;
- current_interval_.end_duration = tentative_end_duration ;
- current_interval_.begin_mode = tentative_begin_duration_mode;
- current_interval_.end_mode = tentative_end_duration_mode ;
- last_input_duration_ = current_interval_.end_duration;
- }
-}
-
-void SequenceParser::duration()
-{
- switch(current_token_type_)
- {
- case e_eof:
- case e_major_separator:
- {
- null_duration();
- }
- break;
- case e_number:
- case e_age_prefix:
- case e_cardinal_prefix:
- case e_keyword:
- {
- single_duration();
- }
- break;
- case e_begin_incl:
- case e_begin_excl:
- {
- intervalic_duration();
- }
- break;
- default:
- {
- diagnostics_
- << "Expected number, '[', '(', 'retirement', or 'maturity'. "
- ;
- mark_diagnostic_context();
- }
- }
- previous_duration_scalar_mode_ = current_duration_scalar_mode_;
-}
-
-// GRAMMAR value: floating-point-number
-// GRAMMAR value: value-constant
-
-void SequenceParser::value()
-{
- switch(current_token_type_)
- {
- case e_number:
- {
- current_interval_.value_is_keyword = false;
- if(keywords_only_)
- {
- diagnostics_
- << "Expected keyword chosen from { "
- ;
- std::copy
- (allowed_keywords_.begin()
- ,allowed_keywords_.end()
- ,std::ostream_iterator<std::string>(diagnostics_, " ")
- );
- diagnostics_ << "}. ";
- mark_diagnostic_context();
- break;
- }
- current_interval_.value_number = current_number_;
- match(current_token_type_);
- }
- break;
- case e_keyword:
- {
- current_interval_.value_is_keyword = true;
- if(allowed_keywords_.empty())
- {
- diagnostics_ << "Expected number. ";
- mark_diagnostic_context();
- break;
- }
- if(contains(allowed_keywords_, current_keyword_))
- {
- current_interval_.value_keyword = current_keyword_;
- match(current_token_type_);
- }
- else
- {
- diagnostics_
- << "Expected keyword chosen from { "
- ;
- std::copy
- (allowed_keywords_.begin()
- ,allowed_keywords_.end()
- ,std::ostream_iterator<std::string>(diagnostics_, " ")
- );
- diagnostics_ << "}. ";
- mark_diagnostic_context();
- }
- }
- break;
- default:
- {
- diagnostics_ << "Expected number or keyword. ";
- mark_diagnostic_context();
- }
- }
-}
-
-// GRAMMAR span: value
-// GRAMMAR span: value , duration
-
-void SequenceParser::span()
-{
- // Assume interval is sane until shown otherwise.
- current_interval_.insane = false;
- value();
- switch(current_token_type_)
- {
- case e_minor_separator:
- {
- match(e_minor_separator);
- duration();
- }
- break;
- case e_eof:
- case e_major_separator:
- case e_begin_incl:
- case e_begin_excl:
- case e_age_prefix:
- case e_cardinal_prefix:
- case e_number:
- case e_keyword:
- {
- duration();
- }
- break;
- default:
- {
- diagnostics_
- << "Expected ';', ',', '@', '#', '[', '(',"
- << " number, 'retirement', or 'maturity'. "
- ;
- mark_diagnostic_context();
- }
- }
- if(!current_interval_.insane)
- {
- intervals_.push_back(current_interval_);
- }
-}
-
-// GRAMMAR sequence: span
-// GRAMMAR sequence: span ; span
-// GRAMMAR sequence: series ;
-
-void SequenceParser::sequence()
-{
- current_token_type_ = e_startup;
- previous_duration_scalar_mode_ = e_inception;
- current_duration_scalar_mode_ = e_inception;
- last_input_duration_ = 0;
-
- // All ValueInterval members should be set explicitly, so
- // initialize them now to recognizable, implausible values,
- // such that the interval is insane until modified.
- current_interval_.value_number = -999999999999.999;
- current_interval_.value_keyword = "daft";
- current_interval_.value_is_keyword = true;
- current_interval_.begin_duration = -1;
- current_interval_.begin_mode = e_invalid_mode;
- current_interval_.end_duration = -2;
- current_interval_.end_mode = e_invalid_mode;
- current_interval_.insane = true;
-
- for(;;)
- {
- switch(current_token_type_)
- {
- case e_eof:
- {
- return;
- }
- case e_major_separator:
- {
- match(current_token_type_);
- if(e_eof == current_token_type_)
- {
- return;
- }
- span();
- }
- break;
-// TODO ?? Superfluous. Same comment elsewhere. Fall through instead.
- case e_startup:
- {
- match(current_token_type_);
- if(e_eof == current_token_type_)
- {
- return;
- }
- span();
- }
- break;
- default:
- {
- diagnostics_ << "Expected ';'. ";
- mark_diagnostic_context();
- return;
- }
- }
- }
-}
-
-SequenceParser::token_type SequenceParser::get_token()
-{
- char c = '\0';
- do
- {
- if(!input_stream_.get(c))
- {
- // COMPILER !! bc++5.02 puts garbage into 'c': reset to '\0'.
- // I believe [27.6.1.3] doesn't allow the garbage.
- c = '\0';
- break;
- }
- }
- while(is_ok_for_cctype(c) && std::isspace(c));
-
- switch(c)
- {
- case 0:
- {
- return current_token_type_ = e_eof;
- }
- case e_major_separator:
- case e_minor_separator:
- case e_begin_incl:
- case e_begin_excl:
- case e_end_incl:
- case e_end_excl:
- case e_age_prefix:
- case e_cardinal_prefix:
- {
- return current_token_type_ = token_type(c);
- }
- case '0': case '1': case '2': case '3': case '4':
- case '5': case '6': case '7': case '8': case '9':
- case '.': case '-':
- {
- input_stream_.putback(c);
- // TODO ?? Assigning 0.0 here at least gives a predictable
- // value upon failure. We can't read both '.' and the
- // following character and then reliably put both back.
- current_number_ = 0.0;
- input_stream_ >> current_number_;
- if(input_stream_.fail())
- {
- diagnostics_ << "Invalid number '" << c << "'. ";
- mark_diagnostic_context();
- }
- return current_token_type_ = e_number;
- }
- // An arbitrary rule must be selected for keyword names;
- // we choose the regexp '[a-z][a-z0-9_]*'.
- case 'a': case 'b': case 'c': case 'd': case 'e':
- case 'f': case 'g': case 'h': case 'i': case 'j':
- case 'k': case 'l': case 'm': case 'n': case 'o':
- case 'p': case 'q': case 'r': case 's': case 't':
- case 'u': case 'v': case 'w': case 'x': case 'y':
- case 'z':
- {
- current_keyword_ = c;
- while
- ( input_stream_.get(c)
- && ((is_ok_for_cctype(c) && std::isalnum(c)) || '_' == c)
- )
- {
- // COMPILER !! 21.3/6 gives push_back() as a member of the
- // string class, and Stroustrup uses it in 6.1.3 of TC++PL3,
- // whence this and other ideas are borrowed, but gcc-2.95.x
- // seems not to support it; operator+=() is likely to be
- // implemented the same way anyway.
- current_keyword_ += c;
- }
- input_stream_.putback(c);
- return current_token_type_ = e_keyword;
- }
- default:
- {
- diagnostics_ << "Unknown token '" << c << "'. ";
- mark_diagnostic_context();
- return current_token_type_ = token_type(0);
- }
- }
-}
-
-void SequenceParser::match(SequenceParser::token_type t)
-{
- if(current_token_type_ == t)
- {
- current_token_type_ = get_token();
- }
- else
- {
- diagnostics_
- << "Expected '"
- << token_type_name(t)
- << "' . "
- ;
- mark_diagnostic_context();
- }
-}
-
-void SequenceParser::mark_diagnostic_context()
-{
- diagnostics_
- << "Current token '"
- << token_type_name(current_token_type_)
- << "' at position " << input_stream_.tellg()
- << ".\n"
- ;
-}
-
InputSequence::InputSequence
(std::string const& input_expression
,int a_years_to_maturity
@@ -1053,46 +409,3 @@ void InputSequence::realize_intervals()
keyword_result_ = s;
}
-/// Extract first substring from a '\n'-delimited exception::what().
-///
-/// SequenceParser::diagnostics() returns a '\n'-delimited string
-/// describing all the anomalies diagnosed while parsing an input
-/// sequence. When that string is not empty, it is reasonable to throw
-/// an exception constructed from it--most generally, in its entirety.
-///
-/// In the important special case where diagnostics are to be shown to
-/// end users, to whom the full multiline set may be overwhelming, use
-/// this function where the exception is caught. It extracts only the
-/// first line, which is presumably the most helpful element, removing
-/// that line's terminal '\n'.
-///
-/// It then strips anything SequenceParser::mark_diagnostic_context()
-/// may have added at the end (and any blanks preceding it), because
-/// end users are unlikely to know what a "token" is, or to care about
-/// the (origin-zero) offset of the error.
-///
-/// Precondition: the argument is not empty; throws otherwise.
-/// Postcondition: the return value is not empty; throws otherwise.
-
-std::string abridge_diagnostics(char const* what)
-{
- std::string s(what);
- LMI_ASSERT(!s.empty());
-
- std::string::size_type z0(s.find('\n'));
- if(std::string::npos != z0)
- {
- s.erase(z0);
- }
-
- std::string::size_type z1(s.find("Current token"));
- if(std::string::npos != z1)
- {
- s.erase(z1);
- }
- rtrim(s, " ");
-
- LMI_ASSERT(!s.empty());
- return s;
-}
-
diff --git a/input_sequence.hpp b/input_sequence.hpp
index f997b31..bf6adb4 100644
--- a/input_sequence.hpp
+++ b/input_sequence.hpp
@@ -1,4 +1,4 @@
-// Input sequences e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0....
+// Input sequences (e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0...)
//
// Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012,
2013, 2014, 2015, 2016, 2017 Gregory W. Chicares.
//
@@ -42,7 +42,7 @@
// editing each life, which was unacceptably tedious.
// Extract the grammar from lines matching the regexp _// GRAMMAR_ in
-// the implementation file.
+// the parser TU.
//
// Index origin is always zero. This is actually natural for end users,
// who think in terms of endpoints--inception implicitly being duration
@@ -131,126 +131,14 @@
#include "config.hpp"
+#include "input_sequence_interval.hpp"
#include "obstruct_slicing.hpp"
#include "so_attributes.hpp"
#include "uncopyable_lmi.hpp"
#include <string>
-#include <sstream>
#include <vector>
-enum duration_mode
- {e_invalid_mode
- ,e_duration
- ,e_attained_age
- ,e_number_of_years
- ,e_inception
- ,e_inforce
- ,e_retirement
- ,e_maturity
- };
-
-/// A single 'span' in the input-sequence grammar.
-///
-/// value_{number,keyword} applies throughout the interval
-/// [begin_duration, end_duration)
-
-struct ValueInterval
-{
- ValueInterval() = default;
-
- double value_number = 0.0;
- std::string value_keyword;
- bool value_is_keyword = false;
- int begin_duration = 0;
- duration_mode begin_mode = e_inception;
- int end_duration = 0;
- duration_mode end_mode = e_maturity;
- bool insane = false;
-};
-
-class SequenceParser
- : private lmi::uncopyable <SequenceParser>
- ,virtual private obstruct_slicing<SequenceParser>
-{
- public:
- SequenceParser
- (std::string const& input_expression
- ,int a_years_to_maturity
- ,int a_issue_age
- ,int a_retirement_age
- ,int a_inforce_duration
- ,int a_effective_year
- ,std::vector<std::string> const& a_allowed_keywords
- ,bool a_keywords_only
- );
-
- ~SequenceParser();
-
- std::string diagnostics() const;
- std::vector<ValueInterval> const& intervals() const;
-
- private:
- enum token_type
- {e_eof = 0
- ,e_major_separator = ';'
- ,e_minor_separator = ','
- ,e_begin_incl = '['
- ,e_begin_excl = '('
- ,e_end_incl = ']'
- ,e_end_excl = ')'
- ,e_age_prefix = '@'
- ,e_cardinal_prefix = '#'
- ,e_number
- ,e_keyword
- ,e_startup
- };
- std::string token_type_name(token_type);
-
- void duration_scalar();
- void null_duration();
- void single_duration();
- void intervalic_duration();
- void validate_duration
- (int tentative_begin_duration
- ,duration_mode tentative_begin_duration_mode
- ,int tentative_end_duration
- ,duration_mode tentative_end_duration_mode
- );
- void duration();
- void value();
- void span();
- void sequence();
- token_type get_token();
- void match(token_type);
-
- void mark_diagnostic_context();
-
- std::istringstream input_stream_;
-
- // Copies of ctor args that are identical to class InputSequence's.
- int years_to_maturity_;
- int issue_age_;
- int retirement_age_;
- int inforce_duration_;
- int effective_year_;
- std::vector<std::string> allowed_keywords_;
- bool keywords_only_;
-
- token_type current_token_type_;
- double current_number_;
- std::string current_keyword_;
- int current_duration_scalar_;
- duration_mode previous_duration_scalar_mode_;
- duration_mode current_duration_scalar_mode_;
- ValueInterval current_interval_;
- int last_input_duration_;
-
- std::ostringstream diagnostics_;
-
- std::vector<ValueInterval> intervals_;
-};
-
class LMI_SO InputSequence
: private lmi::uncopyable <InputSequence>
,virtual private obstruct_slicing<InputSequence>
diff --git a/input_sequence_interval.hpp b/input_sequence_interval.hpp
new file mode 100644
index 0000000..1310e1b
--- /dev/null
+++ b/input_sequence_interval.hpp
@@ -0,0 +1,60 @@
+// Input sequences (e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0...): interval
+//
+// Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012,
2013, 2014, 2015, 2016, 2017 Gregory W. Chicares.
+//
+// This program is free software; you can redistribute it and/or modify
+// it under the terms of the GNU General Public License version 2 as
+// published by the Free Software Foundation.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program; if not, write to the Free Software Foundation,
+// Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
+//
+// http://savannah.nongnu.org/projects/lmi
+// email: <address@hidden>
+// snail: Chicares, 186 Belle Woods Drive, Glastonbury CT 06033, USA
+
+#ifndef input_sequence_interval_hpp
+#define input_sequence_interval_hpp
+
+#include "config.hpp"
+
+#include <string>
+
+enum duration_mode
+ {e_invalid_mode
+ ,e_duration
+ ,e_attained_age
+ ,e_number_of_years
+ ,e_inception
+ ,e_inforce
+ ,e_retirement
+ ,e_maturity
+ };
+
+/// A single 'span' in the input-sequence grammar.
+///
+/// value_{number,keyword} applies throughout the interval
+/// [begin_duration, end_duration)
+
+struct ValueInterval
+{
+ ValueInterval() = default;
+
+ double value_number = 0.0;
+ std::string value_keyword;
+ bool value_is_keyword = false;
+ int begin_duration = 0;
+ duration_mode begin_mode = e_inception;
+ int end_duration = 0;
+ duration_mode end_mode = e_maturity;
+ bool insane = false;
+};
+
+#endif // input_sequence_interval_hpp
+
diff --git a/input_sequence.cpp b/input_sequence_parser.cpp
similarity index 62%
copy from input_sequence.cpp
copy to input_sequence_parser.cpp
index a043c34..bd87d9f 100644
--- a/input_sequence.cpp
+++ b/input_sequence_parser.cpp
@@ -1,5 +1,4 @@
-// Input sequences e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0....
-// Extract the grammar from lines matching the regexp _// GRAMMAR_ .
+// Input sequences (e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0...): parser
//
// Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011,
2012, 2013, 2014, 2015, 2016, 2017 Gregory W. Chicares.
//
@@ -22,23 +21,15 @@
#include "pchfile.hpp"
-#include "input_sequence.hpp"
+#include "input_sequence_parser.hpp"
-#include "alert.hpp"
#include "assert_lmi.hpp"
#include "contains.hpp"
-#include "miscellany.hpp"
-#include "value_cast.hpp"
+#include "miscellany.hpp" // is_ok_for_cctype(), rtrim()
-#include <algorithm>
-#include <cctype>
-#include <ios>
-#include <istream>
+#include <algorithm> // std::copy()
+#include <cctype> // std::isalnum, std::isspace()
#include <iterator> // std::ostream_iterator
-#include <ostream>
-#include <sstream>
-#include <stdexcept>
-#include <type_traits>
SequenceParser::SequenceParser
(std::string const& input_expression
@@ -680,379 +671,6 @@ void SequenceParser::mark_diagnostic_context()
;
}
-InputSequence::InputSequence
- (std::string const& input_expression
- ,int a_years_to_maturity
- ,int a_issue_age
- ,int a_retirement_age
- ,int a_inforce_duration
- ,int a_effective_year
- ,std::vector<std::string> const& a_allowed_keywords
- ,bool a_keywords_only
- ,std::string const& a_default_keyword
- )
- :years_to_maturity_ (a_years_to_maturity)
- ,issue_age_ (a_issue_age)
- ,retirement_age_ (a_retirement_age)
- ,inforce_duration_ (a_inforce_duration)
- ,effective_year_ (a_effective_year)
- ,allowed_keywords_ (a_allowed_keywords)
- ,keywords_only_ (a_keywords_only)
- ,default_keyword_ (a_default_keyword)
-{
- // A default keyword should be specified (i.e., nonempty) only for
- // keyword-only sequences (otherwise, the default is numeric), and
- // it should always be allowable even though other keywords may be
- // disallowed in context. As this is written in 2017-02, the only
- // UDTs with default keywords are:
- // mode_sequence::default_keyword() // "annual"
- // dbo_sequence::default_keyword() // "a"
- // This assertion will provide useful guidance if, e.g., a new
- // policy form that forbids annual mode is implemented.
- LMI_ASSERT
- ( a_default_keyword.empty()
- || a_keywords_only && contains(a_allowed_keywords, a_default_keyword)
- );
-
- SequenceParser parser
- (input_expression
- ,a_years_to_maturity
- ,a_issue_age
- ,a_retirement_age
- ,a_inforce_duration
- ,a_effective_year
- ,a_allowed_keywords
- ,a_keywords_only
- );
-
- std::string const parser_diagnostics = parser.diagnostics();
- if(!parser_diagnostics.empty())
- {
- throw std::runtime_error(parser_diagnostics);
- }
-
- intervals_ = parser.intervals();
-
- // Inception and maturity endpoints exist, so the interval they
- // define must exist. However, parsing an empty expression
- // constructs zero intervals, so a default one must be created
- // to make the physical reality meet the conceptual requirement.
- if(intervals_.empty())
- {
- intervals_.push_back(ValueInterval());
- }
-
- // Extend the last interval's endpoint to maturity, replicating
- // the last element. (This doesn't need to be done by the ctors
- // that take vector arguments, because those arguments specify
- // each value in [inception, maturity) and deduce the terminal
- // (maturity) duration from size().)
-
- // This invariant has not yet been established, whether or not the
- // sequence was empty.
- intervals_.back().end_duration = a_years_to_maturity;
- // This invariant is established by realize_intervals(), but it
- // does no harm to repeat it here, and it would be confusing not
- // to do so in conjunction with the line above.
- intervals_.back().end_mode = e_maturity;
-
- realize_intervals();
-}
-
-/// Construct from vector: e.g, 1 1 1 2 2 --> 1[0,3); 2[3,4).
-///
-/// This is used, e.g., when interest rates obtained from an external
-/// source vary from one year to the next, and it is desired to use
-/// them as lmi input. It might seem that inserting semicolons between
-/// elements would produce acceptable input, and that the only benefit
-/// is run-length encoding. However, if the imported vector is of
-/// length 20, with the last 19 elements the same, then pasting it
-/// into lmi with semicolon delimiters would be an input error if
-/// there are only 15 years until retirement.
-
-InputSequence::InputSequence(std::vector<double> const& v)
- :years_to_maturity_(v.size())
-{
- initialize_from_vector(v);
- realize_intervals();
-}
-
-/// Construct from vector: e.g, a a a b b --> a[0,3); b[3,4).
-///
-/// No actual need for this particular ctor has yet been found, but
-/// one might be, someday.
-
-InputSequence::InputSequence(std::vector<std::string> const& v)
- :years_to_maturity_(v.size())
-{
- initialize_from_vector(v);
- realize_intervals();
-}
-
-namespace
-{
-// Naturally {value_number, value_keyword} constitute a discriminated
-// union: perhaps std::variant when lmi someday requires C++17. See:
-// http://lists.nongnu.org/archive/html/lmi/2017-02/msg00025.html
-// Until then...
-
-void set_value(ValueInterval& v, double d)
-{
- LMI_ASSERT(!v.value_is_keyword);
- v.value_number = d;
-}
-
-void set_value(ValueInterval& v, std::string const& s)
-{
- LMI_ASSERT(v.value_is_keyword);
- v.value_keyword = s;
-}
-} // Unnamed namespace.
-
-// Constructors taking only one (vector) argument are used to convert
-// flat vectors with one value per year to input sequences, compacted
-// with run-length encoding.
-//
-// The control constructs may appear nonobvious. This design treats
-// the push_back operation as fundamental: push_back is called exactly
-// when we know that a new interval must be added. This avoids special
-// handling
-// when the vectors are of length zero, and
-// for the last interval.
-// As a consequence, we always push_back a dummy interval exactly when
-// we know that it will be needed, and then write to intervals_.back().
-//
-// An alternative design would work with a temporary interval and
-// call push_back as needed. I tried that and concluded that this
-// design is simpler.
-//
-// Strings in input vectors are not validated against a map of
-// permissible strings: these constructors are designed for use only
-// with vectors of strings generated by the program from known-valid
-// input, and should not be used in any other situation.
-// SOMEDAY !! Ideally, therefore, they should be protected from
-// unintended use.
-
-template<typename T>
-void InputSequence::initialize_from_vector(std::vector<T> const& v)
-{
- bool const T_is_double = std::is_same<T,double >::value;
- bool const T_is_string = std::is_same<T,std::string>::value;
- static_assert(T_is_double || T_is_string, "");
-
- ValueInterval dummy;
- dummy.value_is_keyword = T_is_string;
-
- T prior_value = v.empty() ? T() : v.front();
- T current_value = prior_value;
-
- intervals_.push_back(dummy);
- set_value(intervals_.back(), current_value);
-
- for(auto const& vi : v)
- {
- current_value = vi;
- if(prior_value == current_value)
- {
- ++intervals_.back().end_duration;
- }
- else
- {
- int value_change_duration = intervals_.back().end_duration;
- intervals_.push_back(dummy);
- set_value(intervals_.back(), current_value);
- intervals_.back().begin_duration = value_change_duration;
- intervals_.back().end_duration = ++value_change_duration;
- prior_value = current_value;
- }
- }
-}
-
-InputSequence::~InputSequence() = default;
-
-std::vector<double> const& InputSequence::linear_number_representation() const
-{
- return number_result_;
-}
-
-std::vector<std::string> const& InputSequence::linear_keyword_representation()
const
-{
- return keyword_result_;
-}
-
-/// Regularized representation in [x,y) interval notation.
-///
-/// If there's only one interval, it must span all years, so depict it
-/// as the simple scalar that it is, specifying no interval.
-///
-/// Use keyword 'maturity' for the last duration. This avoids
-/// gratuitous differences between lives, e.g.
-/// '10000 [20,55); 0' for a 45-year-old
-/// and
-/// '10000 [20,65); 0' for a 35-year-old
-/// which the census GUI would treat as varying across cells, whereas
-/// '10000 [20,65); maturity'
-/// expresses the same sequence uniformly.
-///
-/// TODO ?? For the same reason, this representation should preserve
-/// duration keywords such as 'retirement'.
-
-std::string InputSequence::mathematical_representation() const
-{
- std::ostringstream oss;
- for(auto const& interval_i : intervals_)
- {
- if(interval_i.value_is_keyword)
- {
- oss << interval_i.value_keyword;
- }
- else
- {
- oss << value_cast<std::string>(interval_i.value_number);
- }
-
- if(1 == intervals_.size())
- {
- break;
- }
-
- if(interval_i.end_duration != years_to_maturity_)
- {
- oss
- << " ["
- << interval_i.begin_duration
- << ", "
- << interval_i.end_duration
- << "); "
- ;
- }
- else
- {
- oss
- << " ["
- << interval_i.begin_duration
- << ", "
- << "maturity"
- << ")"
- ;
- }
- }
- return oss.str();
-}
-
-std::vector<ValueInterval> const& InputSequence::interval_representation()
const
-{
- return intervals_;
-}
-
-void InputSequence::realize_intervals()
-{
- // Post-construction invariants.
- // Every ctor must already have established this...
- LMI_ASSERT(!intervals_.empty());
- // ...and this:
- LMI_ASSERT(years_to_maturity_ == intervals_.back().end_duration);
- // It cannot be assumed that all ctors have yet established this...
- intervals_.back().end_mode = e_maturity;
- // ...though now of course it has been established:
- LMI_ASSERT(e_maturity == intervals_.back().end_mode );
-
- std::vector<double> r(years_to_maturity_);
- std::vector<std::string> s(years_to_maturity_, default_keyword_);
- number_result_ = r;
- keyword_result_ = s;
-
- int prior_begin_duration = 0;
- for(auto const& interval_i : intervals_)
- {
- if(interval_i.insane)
- {
- fatal_error()
- << "Untrapped parser error."
- << LMI_FLUSH
- ;
- }
- if(e_invalid_mode == interval_i.begin_mode)
- {
- fatal_error()
- << "Interval "
- << "[ " << interval_i.begin_duration << ", "
- << interval_i.end_duration << " )"
- << " has invalid begin_mode."
- << LMI_FLUSH
- ;
- }
- if(e_invalid_mode == interval_i.end_mode)
- {
- fatal_error()
- << "Interval "
- << "[ " << interval_i.begin_duration << ", "
- << interval_i.end_duration << " )"
- << " has invalid end_mode."
- << LMI_FLUSH
- ;
- }
- if(interval_i.value_is_keyword && "daft" == interval_i.value_keyword)
- {
- fatal_error()
- << "Interval "
- << "[ " << interval_i.begin_duration << ", "
- << interval_i.end_duration << " )"
- << " has invalid value_keyword."
- << LMI_FLUSH
- ;
- }
- if(interval_i.begin_duration < prior_begin_duration)
- {
- fatal_error()
- << "Previous interval began at duration "
- << prior_begin_duration
- << "; current interval "
- << "[ " << interval_i.begin_duration << ", "
- << interval_i.end_duration << " )"
- << " would begin before that."
- << LMI_FLUSH
- ;
- return;
- }
- prior_begin_duration = interval_i.begin_duration;
- bool interval_is_ok =
- 0 <= interval_i.begin_duration
- && interval_i.begin_duration <= interval_i.end_duration
- && interval_i.end_duration <= years_to_maturity_
- ;
- if(!interval_is_ok)
- {
- fatal_error()
- << "Interval "
- << "[ " << interval_i.begin_duration << ", "
- << interval_i.end_duration << " )"
- << " not valid."
- << LMI_FLUSH
- ;
- }
- if(interval_i.value_is_keyword)
- {
- std::fill
- (s.begin() + interval_i.begin_duration
- ,s.begin() + interval_i.end_duration
- ,interval_i.value_keyword
- );
- }
- else
- {
- std::fill
- (r.begin() + interval_i.begin_duration
- ,r.begin() + interval_i.end_duration
- ,interval_i.value_number
- );
- }
- }
-
- number_result_ = r;
- keyword_result_ = s;
-}
-
/// Extract first substring from a '\n'-delimited exception::what().
///
/// SequenceParser::diagnostics() returns a '\n'-delimited string
diff --git a/input_sequence_parser.hpp b/input_sequence_parser.hpp
new file mode 100644
index 0000000..8f76afc
--- /dev/null
+++ b/input_sequence_parser.hpp
@@ -0,0 +1,121 @@
+// Input sequences (e.g. 1 3; 7 5;0; --> 1 1 1 7 7 0...): parser
+//
+// Copyright (C) 2002, 2003, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012,
2013, 2014, 2015, 2016, 2017 Gregory W. Chicares.
+//
+// This program is free software; you can redistribute it and/or modify
+// it under the terms of the GNU General Public License version 2 as
+// published by the Free Software Foundation.
+//
+// This program is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+// GNU General Public License for more details.
+//
+// You should have received a copy of the GNU General Public License
+// along with this program; if not, write to the Free Software Foundation,
+// Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
+//
+// http://savannah.nongnu.org/projects/lmi
+// email: <address@hidden>
+// snail: Chicares, 186 Belle Woods Drive, Glastonbury CT 06033, USA
+
+// See documentation in main input-sequence header.
+
+#ifndef input_sequence_parser_hpp
+#define input_sequence_parser_hpp
+
+#include "config.hpp"
+
+#include "input_sequence_interval.hpp"
+#include "obstruct_slicing.hpp"
+#include "so_attributes.hpp"
+#include "uncopyable_lmi.hpp"
+
+#include <string>
+#include <sstream>
+#include <vector>
+
+class SequenceParser
+ : private lmi::uncopyable <SequenceParser>
+ ,virtual private obstruct_slicing<SequenceParser>
+{
+ public:
+ SequenceParser
+ (std::string const& input_expression
+ ,int a_years_to_maturity
+ ,int a_issue_age
+ ,int a_retirement_age
+ ,int a_inforce_duration
+ ,int a_effective_year
+ ,std::vector<std::string> const& a_allowed_keywords
+ ,bool a_keywords_only
+ );
+
+ ~SequenceParser();
+
+ std::string diagnostics() const;
+ std::vector<ValueInterval> const& intervals() const;
+
+ private:
+ enum token_type
+ {e_eof = 0
+ ,e_major_separator = ';'
+ ,e_minor_separator = ','
+ ,e_begin_incl = '['
+ ,e_begin_excl = '('
+ ,e_end_incl = ']'
+ ,e_end_excl = ')'
+ ,e_age_prefix = '@'
+ ,e_cardinal_prefix = '#'
+ ,e_number
+ ,e_keyword
+ ,e_startup
+ };
+ std::string token_type_name(token_type);
+
+ void duration_scalar();
+ void null_duration();
+ void single_duration();
+ void intervalic_duration();
+ void validate_duration
+ (int tentative_begin_duration
+ ,duration_mode tentative_begin_duration_mode
+ ,int tentative_end_duration
+ ,duration_mode tentative_end_duration_mode
+ );
+ void duration();
+ void value();
+ void span();
+ void sequence();
+ token_type get_token();
+ void match(token_type);
+
+ void mark_diagnostic_context();
+
+ std::istringstream input_stream_;
+
+ // Copies of ctor args that are identical to class InputSequence's.
+ int years_to_maturity_;
+ int issue_age_;
+ int retirement_age_;
+ int inforce_duration_;
+ int effective_year_;
+ std::vector<std::string> allowed_keywords_;
+ bool keywords_only_;
+
+ token_type current_token_type_;
+ double current_number_;
+ std::string current_keyword_;
+ int current_duration_scalar_;
+ duration_mode previous_duration_scalar_mode_;
+ duration_mode current_duration_scalar_mode_;
+ ValueInterval current_interval_;
+ int last_input_duration_;
+
+ std::ostringstream diagnostics_;
+
+ std::vector<ValueInterval> intervals_;
+};
+
+#endif // input_sequence_parser_hpp
+
diff --git a/objects.make b/objects.make
index 7eb950d..c83585d 100644
--- a/objects.make
+++ b/objects.make
@@ -212,6 +212,7 @@ common_common_objects := \
input_realization.o \
input_seq_helpers.o \
input_sequence.o \
+ input_sequence_parser.o \
input_xml_io.o \
interest_rates.o \
ledger.o \
@@ -637,6 +638,7 @@ input_seq_test$(EXEEXT): \
$(common_test_objects) \
input_seq_test.o \
input_sequence.o \
+ input_sequence_parser.o \
miscellany.o \
input_test$(EXEEXT): \
@@ -662,6 +664,7 @@ input_test$(EXEEXT): \
input_realization.o \
input_seq_helpers.o \
input_sequence.o \
+ input_sequence_parser.o \
input_test.o \
input_xml_io.o \
lmi.o \