From e2a506e54a297dcbd13cdb881936e50bbefceb28 Mon Sep 17 00:00:00 2001 From: ridiculousfish Date: Tue, 11 Jun 2013 09:37:51 -0700 Subject: [PATCH] More work on new parser --- fish.xcodeproj/project.pbxproj | 2 + parse_exec.cpp | 147 +++++++++++++++++++++++++++++++++ parse_exec.h | 15 +++- parse_tree.cpp | 121 ++------------------------- parse_tree.h | 90 ++++++++++++++++++++ 5 files changed, 257 insertions(+), 118 deletions(-) diff --git a/fish.xcodeproj/project.pbxproj b/fish.xcodeproj/project.pbxproj index 165fb2785..6f6c83ba1 100644 --- a/fish.xcodeproj/project.pbxproj +++ b/fish.xcodeproj/project.pbxproj @@ -76,6 +76,7 @@ D0A564FE168D23D800AF6161 /* man in CopyFiles */ = {isa = PBXBuildFile; fileRef = D0A564F1168D0BAB00AF6161 /* man */; }; D0A56501168D258300AF6161 /* man in Copy Files */ = {isa = PBXBuildFile; fileRef = D0A564F1168D0BAB00AF6161 /* man */; }; D0C52F371765284C00BFAB82 /* parse_tree.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D0C52F351765284C00BFAB82 /* parse_tree.cpp */; }; + D0C52F381765720600BFAB82 /* parse_exec.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D0C52F331765281F00BFAB82 /* parse_exec.cpp */; }; D0CBD587159EF0E10024809C /* launch_fish.scpt in Resources */ = {isa = PBXBuildFile; fileRef = D0CBD586159EF0E10024809C /* launch_fish.scpt */; }; D0D02A67159837AD008E62BD /* complete.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D0A0853713B3ACEE0099B651 /* complete.cpp */; }; D0D02A69159837B2008E62BD /* env.cpp in Sources */ = {isa = PBXBuildFile; fileRef = D0A0853A13B3ACEE0099B651 /* env.cpp */; }; @@ -1112,6 +1113,7 @@ D0D02A7B15983928008E62BD /* env_universal_common.cpp in Sources */, D0D02A89159839DF008E62BD /* fish.cpp in Sources */, D0C52F371765284C00BFAB82 /* parse_tree.cpp in Sources */, + D0C52F381765720600BFAB82 /* parse_exec.cpp in Sources */, ); runOnlyForDeploymentPostprocessing = 0; }; diff --git a/parse_exec.cpp b/parse_exec.cpp index 5f4cf2d1d..30503857e 100644 --- a/parse_exec.cpp +++ b/parse_exec.cpp @@ -1 +1,148 @@ #include "parse_exec.h" +#include + + +struct exec_node_t +{ + node_offset_t parse_node_idx; + + exec_node_t(size_t pni) : parse_node_idx(pni) + { + } + + virtual ~exec_node_t(); +}; + +exec_node_t::~exec_node_t() +{ +} + +struct exec_redirection_t : public exec_node_t +{ + +}; + +struct exec_argument_t : public exec_node_t +{ + +}; + +struct exec_statement_t +{ + enum + { + decoration_plain, + decoration_command, + decoration_builtin + } decoration; + + std::vector arguments; + std::vector redirections; +}; + +class parse_exec_t +{ + parse_node_tree_t parse_tree; + wcstring src; + std::vector exec_nodes; + + parse_exec_t(const parse_node_tree_t &tree, const wcstring &s) : parse_tree(tree), src(s) + { + } + + void pop_push(uint32_t child_idx) + { + exec_node_t &top = exec_nodes.back(); + const parse_node_t &parse_node = parse_tree.at(top.parse_node_idx); + PARSE_ASSERT(child_idx < parse_node.child_count); + node_offset_t child_node_idx = parse_node.child_start + child_idx; + exec_nodes.pop_back(); + exec_nodes.push_back(child_node_idx); + + } + + void simulate(void); + void enter_parse_node(size_t idx); + void run_top_node(void); +}; + +void parse_exec_t::run_top_node() +{ + PARSE_ASSERT(! exec_nodes.empty()); + exec_node_t &top = exec_nodes.back(); + const parse_node_t &parse_node = parse_tree.at(top.parse_node_idx); + + switch (parse_node.type) + { + case symbol_statement_list: + PARSE_ASSERT(parse_node.child_count == 0 || parse_node.child_count == 2); + if (parse_node.child_count == 0) + { + // Statement list done + exec_nodes.pop_back(); + } + else + { + // First child is a statement, next is the rest of the list + node_offset_t head = parse_node.child_start; + node_offset_t tail = parse_node.child_start + 1; + exec_nodes.pop_back(); + exec_nodes.push_back(tail); + exec_nodes.push_back(head); + } + break; + + case symbol_statement: + PARSE_ASSERT(parse_node.child_count == 1); + pop_push(0); + break; + + case decorated_statement: + PARSE_ASSERT(parse_node.child_count == 1 || parse_node.child_count == 2 ); + pop_push(0); + break; + + default: + fprintf(stderr, "Unhandled token type %ld\n", (long)parse_node.type); + PARSER_DIE(); + break; + + } +} + +void parse_exec_t::enter_parse_node(size_t idx) +{ + PARSE_ASSERT(idx < parse_tree.size()); + exec_node_t exec(idx); + exec_nodes.push_back(exec); +} + +wcstring parse_exec_t::simulate(void) +{ + PARSE_ASSERT(exec_nodes.empty()); + assemble_statement_list(0); + enter_parse_node(0); + run_node(); +} + +wcstring parse_execution_context_t::simulate() +{ + if (parse_tree.empty()) + return L"(empty!)"; + + PARSE_ASSERT(node_idx < nodes.size()); + PARSE_ASSERT(nodes.at(node_idx).type == symbol_statement_list); + + wcstring result; + +} + +parse_execution_context_t::parse_execution_context_t(const parse_node_tree_t &n, const wcstring &s) +{ + ctx = new parse_exec_t(n, s); +} + +wcstring parse_execution_context_t::simulate(void) +{ + return ctx->simulate(); +} diff --git a/parse_exec.h b/parse_exec.h index 50a0ddd68..635ebb836 100644 --- a/parse_exec.h +++ b/parse_exec.h @@ -3,9 +3,20 @@ Programmatic execution of a parse tree */ -#ifndef FISH_PARSE_TREE_H -#define FISH_PARSE_TREE_H +#ifndef FISH_PARSE_EXEC_H +#define FISH_PARSE_EXEC_H #include "parse_tree.h" +class parse_exec_t; +class parse_execution_context_t +{ + parse_exec_t *ctx; + + public: + parse_execution_context_t(const parse_node_tree_t &n, const wcstring &s); + wcstring simulate(void); +}; + + #endif diff --git a/parse_tree.cpp b/parse_tree.cpp index bdc3cf2bd..8c38ff0cd 100644 --- a/parse_tree.cpp +++ b/parse_tree.cpp @@ -1,48 +1,10 @@ -#include "expression.h" +#include "parse_tree.h" #include "tokenizer.h" #include -struct parse_node_t; -typedef size_t node_offset_t; -typedef std::vector parse_node_tree_t; - -#define PARSE_ASSERT(a) assert(a) - -#define PARSER_DIE() assert(0) class parse_command_t; -enum parse_token_type_t -{ - token_type_invalid, - - // Non-terminal tokens - symbol_statement_list, - symbol_statement, - symbol_block_statement, - symbol_block_header, - symbol_if_header, - symbol_for_header, - symbol_while_header, - symbol_begin_header, - symbol_function_header, - symbol_boolean_statement, - symbol_decorated_statement, - symbol_plain_statement, - symbol_arguments_or_redirections_list, - symbol_argument_or_redirection, - - // Terminal types - parse_token_type_string, - parse_token_type_pipe, - parse_token_type_redirection, - parse_token_background, - parse_token_type_end, - parse_token_type_terminate, - - FIRST_PARSE_TOKEN_TYPE = parse_token_type_string -}; - static wcstring token_type_description(parse_token_type_t type) { switch (type) @@ -75,25 +37,11 @@ static wcstring token_type_description(parse_token_type_t type) } } -enum parse_keyword_t +wcstring parse_node_t::describe(void) const { - parse_keyword_none, - parse_keyword_if, - parse_keyword_else, - parse_keyword_for, - parse_keyword_in, - parse_keyword_while, - parse_keyword_begin, - parse_keyword_function, - parse_keyword_switch, - parse_keyword_end, - parse_keyword_and, - parse_keyword_or, - parse_keyword_not, - parse_keyword_command, - parse_keyword_builtin -}; - + wcstring result = token_type_description(type); + return result; +} struct parse_token_t { @@ -135,41 +83,6 @@ static parse_token_t parse_token_from_tokenizer_token(enum token_type tokenizer_ return result; } -/** Base class for nodes of a parse tree */ -class parse_node_t -{ - public: - - /* Start in the source code */ - size_t source_start; - - /* Length of our range in the source code */ - size_t source_length; - - /* Children */ - node_offset_t child_start; - node_offset_t child_count; - - /* Type-dependent data */ - uint32_t tag; - - /* Type of the node */ - enum parse_token_type_t type; - - - /* Description */ - wcstring describe(void) const - { - wcstring result = token_type_description(type); - return result; - } - - /* Constructor */ - explicit parse_node_t(parse_token_type_t ty) : type(ty), source_start(0), source_length(0), child_start(0), child_count(0), tag(0) - { - } -}; - static void dump_tree_recursive(const parse_node_tree_t &nodes, const wcstring &src, size_t start, size_t indent, wcstring *result, size_t *line) { assert(start < nodes.size()); @@ -222,30 +135,6 @@ struct parse_stack_element_t } }; -class parse_execution_context_t -{ - wcstring src; - const parse_node_tree_t nodes; - size_t node_idx; - - public: - parse_execution_context_t(const parse_node_tree_t &n, const wcstring &s) : src(s), nodes(n), node_idx(0) - { - } - - wcstring simulate(void); -}; - -wcstring parse_execution_context_t::simulate() -{ - if (nodes.empty()) - return L"(empty!"); - - PARSE_ASSERT(node_idx < nodes.size()); - PARSE_ASSERT(nodes.at(node_idx).type == symbol_statement_list); - -} - class parse_ll_t { friend class parse_t; diff --git a/parse_tree.h b/parse_tree.h index 9536479bd..892c36cdd 100644 --- a/parse_tree.h +++ b/parse_tree.h @@ -12,6 +12,10 @@ #include "util.h" #include "common.h" #include "tokenizer.h" +#include + +#define PARSE_ASSERT(a) assert(a) +#define PARSER_DIE() assert(0) class parse_ll_t; @@ -25,6 +29,92 @@ class parse_t void parse(const wcstring &str); }; +class parse_node_t; +typedef std::vector parse_node_tree_t; +typedef size_t node_offset_t; + + +enum parse_token_type_t +{ + token_type_invalid, + + // Non-terminal tokens + symbol_statement_list, + symbol_statement, + symbol_block_statement, + symbol_block_header, + symbol_if_header, + symbol_for_header, + symbol_while_header, + symbol_begin_header, + symbol_function_header, + symbol_boolean_statement, + symbol_decorated_statement, + symbol_plain_statement, + symbol_arguments_or_redirections_list, + symbol_argument_or_redirection, + + // Terminal types + parse_token_type_string, + parse_token_type_pipe, + parse_token_type_redirection, + parse_token_background, + parse_token_type_end, + parse_token_type_terminate, + + FIRST_PARSE_TOKEN_TYPE = parse_token_type_string +}; + +enum parse_keyword_t +{ + parse_keyword_none, + parse_keyword_if, + parse_keyword_else, + parse_keyword_for, + parse_keyword_in, + parse_keyword_while, + parse_keyword_begin, + parse_keyword_function, + parse_keyword_switch, + parse_keyword_end, + parse_keyword_and, + parse_keyword_or, + parse_keyword_not, + parse_keyword_command, + parse_keyword_builtin +}; + +/** Base class for nodes of a parse tree */ +class parse_node_t +{ + public: + + /* Type of the node */ + enum parse_token_type_t type; + + /* Start in the source code */ + size_t source_start; + + /* Length of our range in the source code */ + size_t source_length; + + /* Children */ + node_offset_t child_start; + node_offset_t child_count; + + /* Type-dependent data */ + uint32_t tag; + + + /* Description */ + wcstring describe(void) const; + + /* Constructor */ + explicit parse_node_t(parse_token_type_t ty) : type(ty), source_start(0), source_length(0), child_start(0), child_count(0), tag(0) + { + } +}; + /* Fish grammar: