From a012320a9a77dcb60803b5a5f9a5e61a65370770 Mon Sep 17 00:00:00 2001
From: ridiculousfish <corydoras@ridiculousfish.com>
Date: Sun, 7 Jan 2018 13:34:04 -0800
Subject: [PATCH] Add grammar in type system

---
 fish.xcodeproj/project.pbxproj |   2 +
 src/parse_grammar.h            | 163 +++++++++++++++++++++++++++++++++
 src/parse_grammar_elements.inc |  32 +++++++
 src/parse_productions.cpp      |   1 +
 4 files changed, 198 insertions(+)
 create mode 100644 src/parse_grammar.h
 create mode 100644 src/parse_grammar_elements.inc

diff --git a/fish.xcodeproj/project.pbxproj b/fish.xcodeproj/project.pbxproj
index 9c365ccbc..1067cbeb8 100644
--- a/fish.xcodeproj/project.pbxproj
+++ b/fish.xcodeproj/project.pbxproj
@@ -703,6 +703,7 @@
 		D025C02815D1FEA100B9DB63 /* functions */ = {isa = PBXFileReference; lastKnownFileType = folder; name = functions; path = share/functions; sourceTree = "<group>"; };
 		D025C02915D1FEA100B9DB63 /* tools */ = {isa = PBXFileReference; lastKnownFileType = folder; name = tools; path = share/tools; sourceTree = "<group>"; };
 		D02960E51FBD726100CA3985 /* builtin_wait.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = builtin_wait.cpp; sourceTree = "<group>"; };
+		D0301C1D2002B90500B1F463 /* parse_grammar.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; path = parse_grammar.h; sourceTree = "<group>"; };
 		D031890915E36D9800D9CC39 /* base */ = {isa = PBXFileReference; lastKnownFileType = text; path = base; sourceTree = BUILT_PRODUCTS_DIR; };
 		D03238891849D1980032CF2C /* pager.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = pager.cpp; sourceTree = "<group>"; };
 		D032388A1849D1980032CF2C /* pager.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = pager.h; sourceTree = "<group>"; };
@@ -1232,6 +1233,7 @@
 				D03238891849D1980032CF2C /* pager.cpp */,
 				D0A0851B13B3ACEE0099B651 /* parse_util.h */,
 				D0A0855213B3ACEE0099B651 /* parse_util.cpp */,
+				D0301C1D2002B90500B1F463 /* parse_grammar.h */,
 				D0A0851C13B3ACEE0099B651 /* parser_keywords.h */,
 				D0A0855313B3ACEE0099B651 /* parser_keywords.cpp */,
 				D0A0851D13B3ACEE0099B651 /* parser.h */,
diff --git a/src/parse_grammar.h b/src/parse_grammar.h
new file mode 100644
index 000000000..23e79bd79
--- /dev/null
+++ b/src/parse_grammar.h
@@ -0,0 +1,163 @@
+// Programmatic representation of fish grammar
+#ifndef FISH_PARSE_GRAMMAR_H
+#define FISH_PARSE_GRAMMAR_H
+
+#include "parse_constants.h"
+#include "tokenizer.h"
+
+namespace grammar {
+using production_element_t = uint8_t;
+
+// Forward declarations.
+#define ELEM(T) struct T;
+#include "parse_grammar_elements.inc"
+
+// A production is a sequence of production elements.
+template <int WHICH, uint8_t count = 0>
+struct production_t {
+    production_element_t elems[count];
+};
+
+template <int TOKEN>
+struct prim {};
+
+using tok_end = prim<TOK_END>;
+using tok_string = prim<TOK_STRING>;
+
+template <int WHICH>
+struct keyword {};
+
+// A producer holds various productions.
+template <typename T>
+struct producer {};
+
+// Empty produces nothing.
+struct empty : public producer<production_t<0>> {};
+
+// Not sure if we need this.
+template <class A>
+struct single {};
+
+template <class A>
+using produces_single = producer<single<A>>;
+
+// Alternative represents a choice.
+template <class A1, class A2, class A3 = empty, class A4 = empty, class A5 = empty>
+struct alternative {};
+
+template <class... Args>
+using produces_alternative = producer<alternative<Args...>>;
+
+// Sequence represents a list of productions.
+template <class A1, class A2, class A3 = empty, class A4 = empty, class A5 = empty,
+          class A6 = empty>
+struct seq {};
+
+template <class... Args>
+using produces_sequence = producer<seq<Args...>>;
+
+// Following are the grammar productions.
+#define DEF(T) struct T : public
+
+// A job_list is a list of jobs, separated by semicolons or newlines
+DEF(job_list)
+produces_alternative<empty,               //
+                     seq<job, job_list>,  //
+                     seq<job, job_list>>{};
+
+// A job is a non-empty list of statements, separated by pipes. (Non-empty is useful for cases
+// like if statements, where we require a command). To represent "non-empty", we require a
+// statement, followed by a possibly empty job_continuation, and then optionally a background
+// specifier '&'
+DEF(job) produces_sequence<statement, job_continuation, optional_background>{};
+
+DEF(job_continuation)
+produces_alternative<empty,  //
+                     seq<prim<TOK_PIPE>, statement, job_continuation>>{};
+
+// A statement is a normal command, or an if / while / and etc
+DEF(statement)
+produces_alternative<boolean_statement,  //
+                     block_statement,    //
+                     if_statement,       //
+                     switch_statement,   //
+                     decorated_statement>{};
+
+// A block is a conditional, loop, or begin/end
+DEF(if_statement)
+produces_sequence<if_clause, else_clause, end_command, arguments_or_redirections_list>{};
+
+DEF(if_clause)
+produces_sequence<keyword<parse_keyword_if>, job, tok_end, andor_job_list, job_list>{};
+
+DEF(else_clause) produces_alternative<empty, seq<keyword<parse_keyword_else>, else_continuation>>{};
+DEF(else_continuation)
+produces_alternative<seq<if_clause, else_clause>,  //
+                     seq<tok_end, job_list>>{};
+
+DEF(switch_statement)
+produces_sequence<keyword<parse_keyword_switch>, argument, tok_end, case_item_list, end_command,
+                  arguments_or_redirections_list>{};
+
+DEF(case_item_list)
+produces_alternative<empty,                           //
+                     seq<case_item, case_item_list>,  //
+                     seq<tok_end, case_item_list>>{};
+
+DEF(case_item) produces_sequence<keyword<parse_keyword_case>, argument_list, tok_end, job_list>{};
+
+DEF(block_statement)
+produces_sequence<block_header, job_list, end_command, arguments_or_redirections_list>{};
+DEF(block_header) produces_alternative<for_header, while_header, function_header, begin_header>{};
+
+DEF(for_header)
+produces_sequence<keyword<parse_keyword_for>, tok_string, keyword<parse_keyword_in>, argument_list,
+                  tok_end>{};
+
+DEF(while_header) produces_sequence<keyword<parse_keyword_while>, job, tok_end, andor_job_list>{};
+
+struct begin_header : produces_single<keyword<parse_keyword_begin>> {};
+
+// Functions take arguments, and require at least one (the name). No redirections allowed.
+DEF(function_header)
+produces_sequence<keyword<parse_keyword_function>, argument, argument_list, tok_end>{};
+
+//  A boolean statement is AND or OR or NOT
+DEF(boolean_statement)
+produces_alternative<seq<keyword<parse_keyword_and>, statement>,  //
+                     seq<keyword<parse_keyword_or>, statement>,   //
+                     seq<keyword<parse_keyword_not>, statement>>{};
+// An andor_job_list is zero or more job lists, where each starts with an `and` or `or` boolean
+// statement.
+DEF(andor_job_list)
+produces_alternative<empty,                     //
+                     seq<job, andor_job_list>,  //
+                     seq<tok_end, andor_job_list>>{};
+
+// A decorated_statement is a command with a list of arguments_or_redirections, possibly with
+// "builtin" or "command" or "exec"
+DEF(decorated_statement)
+produces_alternative<plain_statement,                                       //
+                     seq<keyword<parse_keyword_command>, plain_statement>,  //
+                     seq<keyword<parse_keyword_builtin>, plain_statement>,  //
+                     seq<keyword<parse_keyword_exec>, plain_statement>>{};
+
+DEF(plain_statement) produces_sequence<tok_string, arguments_or_redirections_list>{};
+
+DEF(argument_list) produces_alternative<empty, seq<argument, argument_list>>{};
+DEF(arguments_or_redirections_list)
+produces_alternative<empty, seq<argument_or_redirection, arguments_or_redirections_list>>{};
+
+DEF(argument_or_redirection) produces_alternative<argument, redirection>{};
+DEF(argument) produces_single<tok_string>{};
+DEF(optional_background) produces_alternative<empty, prim<TOK_BACKGROUND>>{};
+DEF(end_command) produces_single<keyword<parse_keyword_end>>{};
+
+// A freestanding_argument_list is equivalent to a normal argument list, except it may contain
+// TOK_END (newlines, and even semicolons, for historical reasons)
+DEF(freestanding_argument_list)
+produces_alternative<empty,                                      //
+                     seq<argument, freestanding_argument_list>,  //
+                     seq<tok_end, freestanding_argument_list>>{};
+}
+#endif
diff --git a/src/parse_grammar_elements.inc b/src/parse_grammar_elements.inc
new file mode 100644
index 000000000..6f38cf99d
--- /dev/null
+++ b/src/parse_grammar_elements.inc
@@ -0,0 +1,32 @@
+// Define ELEM before including this file.
+ELEM(job_list)
+ELEM(job)
+ELEM(job_continuation)
+ELEM(statement)
+ELEM(if_statement)
+ELEM(if_clause)
+ELEM(else_clause)
+ELEM(else_continuation)
+ELEM(switch_statement)
+ELEM(case_item_list)
+ELEM(case_item)
+ELEM(block_statement)
+ELEM(block_header)
+ELEM(for_header)
+ELEM(while_header)
+ELEM(begin_header)
+ELEM(function_header)
+ELEM(boolean_statement)
+ELEM(andor_job_list)
+ELEM(decorated_statement)
+ELEM(plain_statement)
+ELEM(argument_list)
+ELEM(arguments_or_redirections_list)
+ELEM(argument_or_redirection)
+ELEM(argument)
+ELEM(redirection)
+ELEM(optional_background)
+ELEM(end_command)
+ELEM(freestanding_argument_list)
+#undef ELEM
+
diff --git a/src/parse_productions.cpp b/src/parse_productions.cpp
index 7a66f9d7d..ae0df3303 100644
--- a/src/parse_productions.cpp
+++ b/src/parse_productions.cpp
@@ -4,6 +4,7 @@
 
 #include "common.h"
 #include "parse_constants.h"
+#include "parse_grammar.h"
 #include "parse_productions.h"
 #include "parse_tree.h"