Skip to content

Commit

Permalink
Add Lexer for NDISASM (#933)
Browse files Browse the repository at this point in the history 
The Netwide Disassembler outputs text that is assembly prefixed with
memory offsets and opcode bytes.
  • Loading branch information
@unjello
unjello authored Feb 27, 2024
1 parent 6dd9f26 commit fe5dde8
Show file tree
Hide file tree
Showing 3 changed files with 243 additions and 0 deletions.
123 changes: 123 additions & 0 deletions lexers/embedded/ndisasm.xml
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,123 @@
<lexer>
<config>
<name>NDISASM</name>
<alias>ndisasm</alias>
<mime_type>text/x-disasm</mime_type>
<case_insensitive>true</case_insensitive>
<priority>0.5</priority> <!-- Lower than NASM -->
</config>
<rules>
<state name="root">
<rule pattern="^[0-9A-Za-z]+">
<token type="CommentSpecial"/>
<push state="offset"/>
</rule>
</state>
<state name="offset">
<rule pattern="[0-9A-Za-z]+">
<token type="CommentSpecial"/>
<push state="assembly"/>
</rule>
<rule>
<include state="whitespace"/>
</rule>
</state>
<state name="punctuation">
<rule pattern="[,():\[\]]+">
<token type="Punctuation"/>
</rule>
<rule pattern="[&amp;|^&lt;&gt;+*/%~-]+">
<token type="Operator"/>
</rule>
<rule pattern="[$]+">
<token type="KeywordConstant"/>
</rule>
<rule pattern="seg|wrt|strict">
<token type="OperatorWord"/>
</rule>
<rule pattern="byte|[dq]?word">
<token type="KeywordType"/>
</rule>
</state>
<state name="assembly">
<rule>
<include state="whitespace"/>
</rule>
<rule pattern="[a-z$._?][\w$.?#@~]*:">
<token type="NameLabel"/>
</rule>
<rule pattern="([a-z$._?][\w$.?#@~]*)(\s+)(equ)">
<bygroups>
<token type="NameConstant"/>
<token type="KeywordDeclaration"/>
<token type="KeywordDeclaration"/>
</bygroups>
<push state="instruction-args"/>
</rule>
<rule pattern="BITS|USE16|USE32|SECTION|SEGMENT|ABSOLUTE|EXTERN|GLOBAL|ORG|ALIGN|STRUC|ENDSTRUC|COMMON|CPU|GROUP|UPPERCASE|IMPORT|EXPORT|LIBRARY|MODULE">
<token type="Keyword"/>
<push state="instruction-args"/>
</rule>
<rule pattern="(?:res|d)[bwdqt]|times">
<token type="KeywordDeclaration"/>
<push state="instruction-args"/>
</rule>
<rule pattern="[a-z$._?][\w$.?#@~]*">
<token type="NameFunction"/>
<push state="instruction-args"/>
</rule>
<rule pattern="[\r\n]+">
<token type="Text"/>
<pop depth="2"/>
</rule>
</state>
<state name="instruction-args">
<rule pattern="&#34;(\\&#34;|[^&#34;\n])*&#34;|&#39;(\\&#39;|[^&#39;\n])*&#39;|`(\\`|[^`\n])*`">
<token type="LiteralString"/>
</rule>
<rule pattern="(?:0x[0-9a-f]+|$0[0-9a-f]*|[0-9]+[0-9a-f]*h)">
<token type="LiteralNumberHex"/>
</rule>
<rule pattern="[0-7]+q">
<token type="LiteralNumberOct"/>
</rule>
<rule pattern="[01]+b">
<token type="LiteralNumberBin"/>
</rule>
<rule pattern="[0-9]+\.e?[0-9]+">
<token type="LiteralNumberFloat"/>
</rule>
<rule pattern="[0-9]+">
<token type="LiteralNumberInteger"/>
</rule>
<rule>
<include state="punctuation"/>
</rule>
<rule pattern="r[0-9][0-5]?[bwd]|[a-d][lh]|[er]?[a-d]x|[er]?[sb]p|[er]?[sd]i|[c-gs]s|st[0-7]|mm[0-7]|cr[0-4]|dr[0-367]|tr[3-7]">
<token type="NameBuiltin"/>
</rule>
<rule pattern="[a-z$._?][\w$.?#@~]*">
<token type="NameVariable"/>
</rule>
<rule pattern="[\r\n]+">
<token type="Text"/>
<pop depth="3"/>
</rule>
<rule>
<include state="whitespace"/>
</rule>
</state>
<state name="whitespace">
<rule pattern="\n">
<token type="Text"/>
<pop depth="2"/>
</rule>
<rule pattern="[ \t]+">
<token type="Text"/>
</rule>
<rule pattern=";.*">
<token type="CommentSingle"/>
</rule>
</state>
</rules>
</lexer>
12 changes: 12 additions & 0 deletions lexers/testdata/ndisasm.actual
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,12 @@
00000000 B013 mov al,0x13
00000002 CD10 int 0x10
00000004 B9027D mov cx,0x7d02
00000007 51 push cx
00000008 6800A0 push word 0xa000
0000000B 07 pop es
0000000C B00F mov al,0xf
0000000E F3AA rep stosb
00000010 59 pop cx
00000011 B004 mov al,0x4
00000013 F3AA rep stosb
00000015 CD16 int 0x16
108 changes: 108 additions & 0 deletions lexers/testdata/ndisasm.expected
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,108 @@
[
{"type":"CommentSpecial","value":"00000000"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"B013"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"mov"},
{"type":"Text","value":" "},
{"type":"NameBuiltin","value":"al"},
{"type":"Punctuation","value":","},
{"type":"LiteralNumberHex","value":"0x13"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"00000002"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"CD10"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"int"},
{"type":"Text","value":" "},
{"type":"LiteralNumberHex","value":"0x10"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"00000004"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"B9027D"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"mov"},
{"type":"Text","value":" "},
{"type":"NameBuiltin","value":"cx"},
{"type":"Punctuation","value":","},
{"type":"LiteralNumberHex","value":"0x7d02"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"00000007"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"51"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"push"},
{"type":"Text","value":" "},
{"type":"NameBuiltin","value":"cx"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"00000008"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"6800A0"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"push"},
{"type":"Text","value":" "},
{"type":"KeywordType","value":"word"},
{"type":"Text","value":" "},
{"type":"LiteralNumberHex","value":"0xa000"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"0000000B"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"07"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"pop"},
{"type":"Text","value":" "},
{"type":"NameBuiltin","value":"es"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"0000000C"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"B00F"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"mov"},
{"type":"Text","value":" "},
{"type":"NameBuiltin","value":"al"},
{"type":"Punctuation","value":","},
{"type":"LiteralNumberHex","value":"0xf"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"0000000E"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"F3AA"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"rep"},
{"type":"Text","value":" "},
{"type":"NameVariable","value":"stosb"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"00000010"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"59"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"pop"},
{"type":"Text","value":" "},
{"type":"NameBuiltin","value":"cx"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"00000011"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"B004"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"mov"},
{"type":"Text","value":" "},
{"type":"NameBuiltin","value":"al"},
{"type":"Punctuation","value":","},
{"type":"LiteralNumberHex","value":"0x4"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"00000013"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"F3AA"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"rep"},
{"type":"Text","value":" "},
{"type":"NameVariable","value":"stosb"},
{"type":"Text","value":"\n"},
{"type":"CommentSpecial","value":"00000015"},
{"type":"Text","value":" "},
{"type":"CommentSpecial","value":"CD16"},
{"type":"Text","value":" "},
{"type":"NameFunction","value":"int"},
{"type":"Text","value":" "},
{"type":"LiteralNumberHex","value":"0x16"},
{"type":"Text","value":"\n"}
]

0 comments on commit fe5dde8

Please sign in to comment.