Add Smali lexer (#765)

Converted from the pygments Smali lexer. Add a basic test case and update the 'Supported languages' table in the README.
2025-03-17 20:58:08 +02:00 · 2023-03-23 11:59:08 -07:00 · 2023-03-23 11:59:08 -07:00 · 2672d3c4c3
commit 2672d3c4c3
parent b9e3758120
4 changed files with 383 additions and 1 deletions
--- a/README.md
+++ b/README.md
@ -55,7 +55,7 @@ O | Objective-C, OCaml, Octave, OnesEnterprise, OpenEdge ABL, OpenSCAD, Org Mode
 P | PacmanConf, Perl, PHP, PHTML, Pig, PkgConfig, PL/pgSQL, plaintext, Pony, PostgreSQL SQL dialect, PostScript, POVRay, PowerShell, Prolog, PromQL, Properties, Protocol Buffer, PSL, Puppet, Python 2, Python
 Q | QBasic
 R | R, Racket, Ragel, Raku, react, ReasonML, reg, reStructuredText, Rexx, Ruby, Rust
-S | SAS, Sass, Scala, Scheme, Scilab, SCSS, Sed, Smalltalk, Smarty, Snobol, Solidity, SPARQL, SQL, SquidConf, Standard ML, stas, Stylus, Svelte, Swift, SYSTEMD, systemverilog
+S | SAS, Sass, Scala, Scheme, Scilab, SCSS, Sed, Smali, Smalltalk, Smarty, Snobol, Solidity, SPARQL, SQL, SquidConf, Standard ML, stas, Stylus, Svelte, Swift, SYSTEMD, systemverilog
 T | TableGen, TASM, Tcl, Tcsh, Termcap, Terminfo, Terraform, TeX, Thrift, TOML, TradingView, Transact-SQL, Turing, Turtle, Twig, TypeScript, TypoScript, TypoScriptCssData, TypoScriptHtmlData
 V | VB.net, verilog, VHDL, VHS, VimL, vue
 W | WDTE
--- a/lexers/embedded/smali.xml
+++ b/lexers/embedded/smali.xml
@ -0,0 +1,73 @@
+<!--
+  Generated from https://github.com/pygments/pygments/blob/15f222adefd2bf7835bfd74a12d720028ae68d29/pygments/lexers/dalvik.py.
+-->
+<lexer>
+  <config>
+    <name>Smali</name>
+    <alias>smali</alias>
+    <filename>*.smali</filename>
+    <mime_type>text/smali</mime_type>
+  </config>
+  <rules>
+    <state name="root">
+      <rule><include state="comment"/></rule>
+      <rule><include state="label"/></rule>
+      <rule><include state="field"/></rule>
+      <rule><include state="method"/></rule>
+      <rule><include state="class"/></rule>
+      <rule><include state="directive"/></rule>
+      <rule><include state="access-modifier"/></rule>
+      <rule><include state="instruction"/></rule>
+      <rule><include state="literal"/></rule>
+      <rule><include state="punctuation"/></rule>
+      <rule><include state="type"/></rule>
+      <rule><include state="whitespace"/></rule>
+    </state>
+    <state name="directive">
+      <rule pattern="^([ \t]*)(\.(?:class|super|implements|field|subannotation|annotation|enum|method|registers|locals|array-data|packed-switch|sparse-switch|catchall|catch|line|parameter|local|prologue|epilogue|source))"><bygroups><token type="TextWhitespace"/><token type="Keyword"/></bygroups></rule>
+      <rule pattern="^([ \t]*)(\.end)( )(field|subannotation|annotation|method|array-data|packed-switch|sparse-switch|parameter|local)"><bygroups><token type="TextWhitespace"/><token type="Keyword"/><token type="TextWhitespace"/><token type="Keyword"/></bygroups></rule>
+      <rule pattern="^([ \t]*)(\.restart)( )(local)"><bygroups><token type="TextWhitespace"/><token type="Keyword"/><token type="TextWhitespace"/><token type="Keyword"/></bygroups></rule>
+    </state>
+    <state name="access-modifier">
+      <rule pattern="(public|private|protected|static|final|synchronized|bridge|varargs|native|abstract|strictfp|synthetic|constructor|declared-synchronized|interface|enum|annotation|volatile|transient)"><token type="Keyword"/></rule>
+    </state>
+    <state name="whitespace">
+      <rule pattern="\n"><token type="TextWhitespace"/></rule>
+      <rule pattern="\s+"><token type="TextWhitespace"/></rule>
+    </state>
+    <state name="instruction">
+      <rule pattern="\b[vp]\d+\b"><token type="NameBuiltin"/></rule>
+      <rule pattern="(\b[a-z][A-Za-z0-9/-]+)(\s+)"><bygroups><token type="Text"/><token type="TextWhitespace"/></bygroups></rule>
+    </state>
+    <state name="literal">
+      <rule pattern="&quot;.*&quot;"><token type="LiteralString"/></rule>
+      <rule pattern="0x[0-9A-Fa-f]+t?"><token type="LiteralNumberHex"/></rule>
+      <rule pattern="[0-9]*\.[0-9]+([eE][0-9]+)?[fd]?"><token type="LiteralNumberFloat"/></rule>
+      <rule pattern="[0-9]+L?"><token type="LiteralNumberInteger"/></rule>
+    </state>
+    <state name="field">
+      <rule pattern="(\$?\b)([\w$]*)(:)"><bygroups><token type="Punctuation"/><token type="NameVariable"/><token type="Punctuation"/></bygroups></rule>
+    </state>
+    <state name="method">
+      <rule pattern="&lt;(?:cl)?init&gt;"><token type="NameFunction"/></rule>
+      <rule pattern="(\$?\b)([\w$]*)(\()"><bygroups><token type="Punctuation"/><token type="NameFunction"/><token type="Punctuation"/></bygroups></rule>
+    </state>
+    <state name="label">
+      <rule pattern=":\w+"><token type="NameLabel"/></rule>
+    </state>
+    <state name="class">
+      <rule pattern="(L)((?:[\w$]+/)*)([\w$]+)(;)"><bygroups><token type="KeywordType"/><token type="Text"/><token type="NameClass"/><token type="Text"/></bygroups></rule>
+    </state>
+    <state name="punctuation">
+      <rule pattern="-&gt;"><token type="Punctuation"/></rule>
+      <rule pattern="[{},():=.-]"><token type="Punctuation"/></rule>
+    </state>
+    <state name="type">
+      <rule pattern="[ZBSCIJFDV\[]+"><token type="KeywordType"/></rule>
+    </state>
+    <state name="comment">
+      <rule pattern="#.*?\n"><token type="Comment"/></rule>
+    </state>
+  </rules>
+</lexer>
+
--- a/lexers/testdata/smali.actual
+++ b/lexers/testdata/smali.actual
@ -0,0 +1,55 @@
+.class LExample;
+.super Ljava/lang/Object;
+.source "Example.java"
+
+
+# instance fields
+.field protected count:I
+
+.field private label:Ljava/lang/String;
+
+
+# direct methods
+.method constructor <init>()V
+    .registers 1
+
+    .line 1
+    invoke-direct {p0}, Ljava/lang/Object;-><init>()V
+
+    return-void
+.end method
+
+.method public static main([Ljava/lang/String;)V
+    .registers 2
+
+    .line 16
+    sget-object p0, Ljava/lang/System;->out:Ljava/io/PrintStream;
+
+    const-string v0, "Hello world!"
+
+    invoke-virtual {p0, v0}, Ljava/io/PrintStream;->println(Ljava/lang/String;)V
+
+    .line 17
+    return-void
+.end method
+
+
+# virtual methods
+.method public getLabel()Ljava/lang/String;
+    .registers 2
+
+    .line 8
+    iget-object v0, p0, LExample;->label:Ljava/lang/String;
+
+    return-object v0
+.end method
+
+.method public setLabel(Ljava/lang/String;)V
+    .registers 2
+
+    .line 12
+    iput-object p1, p0, LExample;->label:Ljava/lang/String;
+
+    .line 13
+    return-void
+.end method
--- a/lexers/testdata/smali.expected
+++ b/lexers/testdata/smali.expected
@ -0,0 +1,254 @@
+[
+  {"type":"Keyword","value":".class"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"L"},
+  {"type":"NameClass","value":"Example"},
+  {"type":"Text","value":";"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Keyword","value":".super"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"Object"},
+  {"type":"Text","value":";"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Keyword","value":".source"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralString","value":"\"Example.java\""},
+  {"type":"TextWhitespace","value":"\n\n\n"},
+  {"type":"Comment","value":"# instance fields\n"},
+  {"type":"Keyword","value":".field"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"protected"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameVariable","value":"count"},
+  {"type":"Punctuation","value":":"},
+  {"type":"KeywordType","value":"I"},
+  {"type":"TextWhitespace","value":"\n\n"},
+  {"type":"Keyword","value":".field"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"private"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameVariable","value":"label"},
+  {"type":"Punctuation","value":":"},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"String"},
+  {"type":"Text","value":";"},
+  {"type":"TextWhitespace","value":"\n\n\n"},
+  {"type":"Comment","value":"# direct methods\n"},
+  {"type":"Keyword","value":".method"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"constructor"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameFunction","value":"\u003cinit\u003e"},
+  {"type":"Punctuation","value":"()"},
+  {"type":"KeywordType","value":"V"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Keyword","value":".registers"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"1"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Keyword","value":".line"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"1"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Text","value":"invoke-direct"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"NameBuiltin","value":"p0"},
+  {"type":"Punctuation","value":"},"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"Object"},
+  {"type":"Text","value":";"},
+  {"type":"Punctuation","value":"-\u003e"},
+  {"type":"NameFunction","value":"\u003cinit\u003e"},
+  {"type":"Punctuation","value":"()"},
+  {"type":"KeywordType","value":"V"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Text","value":"return-void"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Keyword","value":".end"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"method"},
+  {"type":"TextWhitespace","value":"\n\n"},
+  {"type":"Keyword","value":".method"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"static"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameFunction","value":"main"},
+  {"type":"Punctuation","value":"("},
+  {"type":"KeywordType","value":"[L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"String"},
+  {"type":"Text","value":";"},
+  {"type":"Punctuation","value":")"},
+  {"type":"KeywordType","value":"V"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Keyword","value":".registers"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"2"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Keyword","value":".line"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"16"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Text","value":"sget-object"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameBuiltin","value":"p0"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"System"},
+  {"type":"Text","value":";"},
+  {"type":"Punctuation","value":"-\u003e"},
+  {"type":"NameVariable","value":"out"},
+  {"type":"Punctuation","value":":"},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/io/"},
+  {"type":"NameClass","value":"PrintStream"},
+  {"type":"Text","value":";"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Text","value":"const-string"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameBuiltin","value":"v0"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralString","value":"\"Hello world!\""},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Text","value":"invoke-virtual"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"NameBuiltin","value":"p0"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameBuiltin","value":"v0"},
+  {"type":"Punctuation","value":"},"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/io/"},
+  {"type":"NameClass","value":"PrintStream"},
+  {"type":"Text","value":";"},
+  {"type":"Punctuation","value":"-\u003e"},
+  {"type":"NameFunction","value":"println"},
+  {"type":"Punctuation","value":"("},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"String"},
+  {"type":"Text","value":";"},
+  {"type":"Punctuation","value":")"},
+  {"type":"KeywordType","value":"V"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Keyword","value":".line"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"17"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Text","value":"return-void"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Keyword","value":".end"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"method"},
+  {"type":"TextWhitespace","value":"\n\n\n"},
+  {"type":"Comment","value":"# virtual methods\n"},
+  {"type":"Keyword","value":".method"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameFunction","value":"getLabel"},
+  {"type":"Punctuation","value":"()"},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"String"},
+  {"type":"Text","value":";"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Keyword","value":".registers"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"2"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Keyword","value":".line"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"8"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Text","value":"iget-object"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameBuiltin","value":"v0"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameBuiltin","value":"p0"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"L"},
+  {"type":"NameClass","value":"Example"},
+  {"type":"Text","value":";"},
+  {"type":"Punctuation","value":"-\u003e"},
+  {"type":"NameVariable","value":"label"},
+  {"type":"Punctuation","value":":"},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"String"},
+  {"type":"Text","value":";"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Text","value":"return-object"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameBuiltin","value":"v0"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Keyword","value":".end"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"method"},
+  {"type":"TextWhitespace","value":"\n\n"},
+  {"type":"Keyword","value":".method"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameFunction","value":"setLabel"},
+  {"type":"Punctuation","value":"("},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"String"},
+  {"type":"Text","value":";"},
+  {"type":"Punctuation","value":")"},
+  {"type":"KeywordType","value":"V"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Keyword","value":".registers"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"2"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Keyword","value":".line"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"12"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Text","value":"iput-object"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameBuiltin","value":"p1"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameBuiltin","value":"p0"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"L"},
+  {"type":"NameClass","value":"Example"},
+  {"type":"Text","value":";"},
+  {"type":"Punctuation","value":"-\u003e"},
+  {"type":"NameVariable","value":"label"},
+  {"type":"Punctuation","value":":"},
+  {"type":"KeywordType","value":"L"},
+  {"type":"Text","value":"java/lang/"},
+  {"type":"NameClass","value":"String"},
+  {"type":"Text","value":";"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Keyword","value":".line"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"13"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Text","value":"return-void"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Keyword","value":".end"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"method"},
+  {"type":"TextWhitespace","value":"\n"}
+]