From a90cb6dcf56683d7d72ac17bebcc9817b5d0615a Mon Sep 17 00:00:00 2001
From: Jan Travnicek <jan.travnicek@.fit.cvut.cz>
Date: Fri, 22 Nov 2019 12:50:41 +0100
Subject: [PATCH] operators Lexing redesign
---
alib2cli/src/lexer/Lexer.cpp | 158 ++++++++++++++++++++++++++++++++---
alib2cli/src/lexer/Lexer.h | 34 ++++++--
2 files changed, 177 insertions(+), 15 deletions(-)
diff --git a/alib2cli/src/lexer/Lexer.cpp b/alib2cli/src/lexer/Lexer.cpp
index 732a3b97b0..79c758ac46 100644
--- a/alib2cli/src/lexer/Lexer.cpp
+++ b/alib2cli/src/lexer/Lexer.cpp
@@ -33,13 +33,13 @@ q0: if ( m_source.getCharacter ( ) == EOF ) {
res.m_raw += m_source.getCharacter ( );
m_source.advance ( readNextLine );
res.m_type = TokenType::LESS_SIGN;
- return res;
+ goto q10;
}
if ( m_source.getCharacter ( ) == '>' ) {
res.m_raw += m_source.getCharacter ( );
m_source.advance ( readNextLine );
res.m_type = TokenType::MORE_SIGN;
- return res;
+ goto q11;
}
if ( m_source.getCharacter ( ) == '(' ) {
res.m_raw += m_source.getCharacter ( );
@@ -93,13 +93,13 @@ q0: if ( m_source.getCharacter ( ) == EOF ) {
res.m_raw += m_source.getCharacter ( );
m_source.advance ( readNextLine );
res.m_type = TokenType::AMPERSAND_SIGN;
- return res;
+ goto q8;
}
if ( m_source.getCharacter ( ) == '|' ) {
res.m_raw += m_source.getCharacter ( );
m_source.advance ( readNextLine );
res.m_type = TokenType::PIPE_SIGN;
- return res;
+ goto q9;
}
if ( m_source.getCharacter ( ) == '^' ) {
res.m_raw += m_source.getCharacter ( );
@@ -122,8 +122,8 @@ q0: if ( m_source.getCharacter ( ) == EOF ) {
if ( m_source.getCharacter ( ) == '=' ) {
res.m_raw += m_source.getCharacter ( );
m_source.advance ( readNextLine );
- res.m_type = TokenType::EQUAL_SIGN;
- return res;
+ res.m_type = TokenType::ASSIGN_OPERATOR;
+ goto q12;
}
if ( m_source.getCharacter ( ) == '#' ) {
res.m_raw += m_source.getCharacter ( );
@@ -147,7 +147,7 @@ q0: if ( m_source.getCharacter ( ) == EOF ) {
res.m_raw += m_source.getCharacter ( );
m_source.advance ( readNextLine );
res.m_type = TokenType::PLUS_SIGN;
- return res;
+ goto q7;
}
if ( m_source.getCharacter ( ) == '/' ) {
res.m_raw += m_source.getCharacter ( );
@@ -171,7 +171,7 @@ q0: if ( m_source.getCharacter ( ) == EOF ) {
res.m_raw += m_source.getCharacter ( );
m_source.advance ( readNextLine );
res.m_type = TokenType::EXCLAMATION_SIGN;
- return res;
+ goto q13;
}
if ( m_source.getCharacter ( ) == '%' ) {
res.m_raw += m_source.getCharacter ( );
@@ -230,12 +230,18 @@ q1: if ( m_source.getCharacter ( ) == '\0' ) {
res.m_type = TokenType::INTEGER;
return res;
}
-
-q2: if ( m_source.getCharacter ( ) == '\0' ) {
+q2:
+ if ( m_source.getCharacter ( ) == '\0' ) {
res.m_value = "";
res.m_type = TokenType::MINUS_SIGN;
return res;
}
+ if ( m_source.getCharacter ( ) == '-' ) {
+ res.m_raw += m_source.getCharacter ( );
+ res.m_value += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ goto q2Prime;
+ }
if ( ( m_source.getCharacter ( ) >= '0' && m_source.getCharacter ( ) <= '9' ) ) {
res.m_raw += m_source.getCharacter ( );
res.m_value += m_source.getCharacter ( );
@@ -260,6 +266,36 @@ q2: if ( m_source.getCharacter ( ) == '\0' ) {
res.m_type = TokenType::MINUS_SIGN;
return res;
+q2Prime:
+ if ( m_source.getCharacter ( ) == '\0' ) {
+ res.m_value = "";
+ res.m_type = TokenType::DEC_OPERATOR;
+ return res;
+ }
+ if ( ( m_source.getCharacter ( ) >= '0' && m_source.getCharacter ( ) <= '9' ) ) {
+ res.m_raw += m_source.getCharacter ( );
+ res.m_value += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ goto q1;
+ }
+
+ if ( ( m_source.getCharacter ( ) >= 'a' && m_source.getCharacter ( ) <= 'z' )
+ || ( m_source.getCharacter ( ) >= 'A' && m_source.getCharacter ( ) <= 'Z' )
+ || ( m_source.getCharacter ( ) == ':' ) ) {
+ res.m_raw += m_source.getCharacter ( );
+ res.m_value += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ goto q3;
+ } else if ( m_source.getCharacter ( ) == '\\' ) {
+ res.m_raw += m_source.getCharacter ( );
+ m_source.advance ( true );
+ goto q3Escape;
+ }
+
+ res.m_value = "";
+ res.m_type = TokenType::DEC_OPERATOR;
+ return res;
+
q3: if ( m_source.getCharacter ( ) == '\0' ) {
res.m_type = is_kw ( res.m_value );
return res;
@@ -350,6 +386,108 @@ q6: if ( m_source.getCharacter ( ) == '\0' ) {
return res;
}
+q7: if ( m_source.getCharacter ( ) == EOF ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '\0' ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '+' ) {
+ res.m_raw += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ res.m_type = TokenType::INC_OPERATOR;
+ } else {
+ return res;
+ }
+
+q8: if ( m_source.getCharacter ( ) == EOF ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '\0' ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '&' ) {
+ res.m_raw += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ res.m_type = TokenType::AND_OPERATOR;
+ } else {
+ return res;
+ }
+
+q9: if ( m_source.getCharacter ( ) == EOF ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '\0' ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '|' ) {
+ res.m_raw += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ res.m_type = TokenType::OR_OPERATOR;
+ } else {
+ return res;
+ }
+
+q10:
+ if ( m_source.getCharacter ( ) == EOF ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '\0' ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '=' ) {
+ res.m_raw += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ res.m_type = TokenType::LESS_OR_EQUAL_OPERATOR;
+ } else {
+ return res;
+ }
+
+q11:
+ if ( m_source.getCharacter ( ) == EOF ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '\0' ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '=' ) {
+ res.m_raw += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ res.m_type = TokenType::MORE_OR_EQUAL_OPERATOR;
+ } else {
+ return res;
+ }
+
+q12:
+ if ( m_source.getCharacter ( ) == EOF ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '\0' ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '=' ) {
+ res.m_raw += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ res.m_type = TokenType::EQUAL_OPERATOR;
+ } else {
+ return res;
+ }
+
+q13:
+ if ( m_source.getCharacter ( ) == EOF ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '\0' ) {
+ return res;
+ }
+ if ( m_source.getCharacter ( ) == '=' ) {
+ res.m_raw += m_source.getCharacter ( );
+ m_source.advance ( readNextLine );
+ res.m_type = TokenType::NOT_EQUAL_OPERATOR;
+ } else {
+ return res;
+ }
+
qFile:
if ( m_source.getCharacter ( ) == EOF ) {
res.m_type = TokenType::EOT;
diff --git a/alib2cli/src/lexer/Lexer.h b/alib2cli/src/lexer/Lexer.h
index 3e796624ec..d373532db9 100644
--- a/alib2cli/src/lexer/Lexer.h
+++ b/alib2cli/src/lexer/Lexer.h
@@ -31,7 +31,9 @@ public:
INTEGER,
STRING,
LESS_SIGN,
+ LESS_OR_EQUAL_OPERATOR,
MORE_SIGN,
+ MORE_OR_EQUAL_OPERATOR,
LEFT_PAREN,
RIGHT_PAREN,
LEFT_BRACE,
@@ -41,18 +43,24 @@ public:
DOLAR_SIGN,
AT_SIGN,
AMPERSAND_SIGN,
+ AND_OPERATOR,
PIPE_SIGN,
+ OR_OPERATOR,
CARET_SIGN,
COLON_SIGN,
SEMICOLON_SIGN,
MINUS_SIGN,
+ DEC_OPERATOR,
PLUS_SIGN,
+ INC_OPERATOR,
SLASH_SIGN,
STAR_SIGN,
TYLDE_SIGN,
EXCLAMATION_SIGN,
PERCENTAGE_SIGN,
- EQUAL_SIGN,
+ ASSIGN_OPERATOR,
+ EQUAL_OPERATOR,
+ NOT_EQUAL_OPERATOR,
HASH_SIGN,
COMMA,
DOT,
@@ -76,9 +84,13 @@ public:
case TokenType::STRING :
return "string";
case TokenType::LESS_SIGN :
- return "in_redirect";
+ return "less_sign";
+ case TokenType::LESS_OR_EQUAL_OPERATOR :
+ return "less_or_equal_operator";
case TokenType::MORE_SIGN :
- return "out_redirect";
+ return "more_sign";
+ case TokenType::MORE_OR_EQUAL_OPERATOR :
+ return "more_or_equal_operator";
case TokenType::LEFT_PAREN :
return "left_paren";
case TokenType::RIGHT_PAREN :
@@ -97,8 +109,12 @@ public:
return "dolar_sign";
case TokenType::AMPERSAND_SIGN :
return "ampersand_sign";
+ case TokenType::AND_OPERATOR :
+ return "and_operator";
case TokenType::PIPE_SIGN :
return "pipe_sign";
+ case TokenType::OR_OPERATOR :
+ return "or_operator";
case TokenType::CARET_SIGN :
return "caret_sign";
case TokenType::COLON_SIGN :
@@ -107,8 +123,12 @@ public:
return "semicolon_sign";
case TokenType::MINUS_SIGN :
return "minus_sign";
+ case TokenType::DEC_OPERATOR :
+ return "dec_operator";
case TokenType::PLUS_SIGN :
return "plus_sign";
+ case TokenType::INC_OPERATOR :
+ return "inc_operator";
case TokenType::SLASH_SIGN :
return "slash_sign";
case TokenType::STAR_SIGN :
@@ -119,8 +139,12 @@ public:
return "exclemation_sign";
case TokenType::PERCENTAGE_SIGN :
return "percentage_sign";
- case TokenType::EQUAL_SIGN :
- return "equal_sign";
+ case TokenType::ASSIGN_OPERATOR :
+ return "assign_operator";
+ case TokenType::EQUAL_OPERATOR :
+ return "equal_operator";
+ case TokenType::NOT_EQUAL_OPERATOR :
+ return "not_equal_operator";
case TokenType::HASH_SIGN :
return "hash_sign";
case TokenType::COMMA :
--
GitLab