Merged in smattr/pygments-main/1575ad37-9b95-4885-9b49-352f0b1c860d (pull request #579)

Add support for the CapDL language.

3 files changed, 124 insertions, 1 deletions

diff --git a/pygments/lexers/_mapping.py b/pygments/lexers/_mapping.py
index 0e197abd..fec72ca3 100644
--- a/pygments/lexers/_mapping.py
+++ b/pygments/lexers/_mapping.py
@@ -70,6 +70,7 @@ LEXERS = {
     'CSharpLexer': ('pygments.lexers.dotnet', 'C#', ('csharp', 'c#'), ('*.cs',), ('text/x-csharp',)),
     'Ca65Lexer': ('pygments.lexers.asm', 'ca65 assembler', ('ca65',), ('*.s',), ()),
     'CadlLexer': ('pygments.lexers.archetype', 'cADL', ('cadl',), ('*.cadl',), ()),
+    'CapDLLexer': ('pygments.lexers.esoteric', 'CapDL', ('capdl',), ('*.cdl',), ()),
     'CbmBasicV2Lexer': ('pygments.lexers.basic', 'CBM BASIC V2', ('cbmbas',), ('*.bas',), ()),
     'CeylonLexer': ('pygments.lexers.jvm', 'Ceylon', ('ceylon',), ('*.ceylon',), ('text/x-ceylon',)),
     'Cfengine3Lexer': ('pygments.lexers.configs', 'CFEngine3', ('cfengine3', 'cf3'), ('*.cf',), ()),
diff --git a/pygments/lexers/esoteric.py b/pygments/lexers/esoteric.py
index c9db26b5..c99f7392 100644
--- a/pygments/lexers/esoteric.py
+++ b/pygments/lexers/esoteric.py
@@ -13,7 +13,8 @@ from pygments.lexer import RegexLexer, include, words
 from pygments.token import Text, Comment, Operator, Keyword, Name, String, \
     Number, Punctuation, Error
 
-__all__ = ['BrainfuckLexer', 'BefungeLexer', 'RedcodeLexer', 'CAmkESLexer']
+__all__ = ['BrainfuckLexer', 'BefungeLexer', 'RedcodeLexer', 'CAmkESLexer',
+    'CapDLLexer']
 
 
 class BrainfuckLexer(RegexLexer):
@@ -143,6 +144,63 @@ class CAmkESLexer(RegexLexer):
     }
 
 
+class CapDLLexer(RegexLexer):
+    """
+    Basic lexer for
+    `CapDL <https://ssrg.nicta.com.au/publications/nictaabstracts/Kuz_KLW_10.abstract.pml>`_.
+
+    The source of the primary tool that reads such specifications is available
+    at https://github.com/seL4/capdl/tree/master/capDL-tool. Note that this
+    lexer only supports a subset of the grammar. For example, identifiers can
+    shadow type names, but these instances are currently incorrectly
+    highlighted as types. Supporting this would need a stateful lexer that is
+    considered unnecessarily complex for now.
+    """
+    name = 'CapDL'
+    aliases = ['capdl']
+    filenames = ['*.cdl']
+
+    tokens = {
+        'root':[
+
+            # C pre-processor directive
+            (r'^\s*#.*\n', Comment.Preproc),
+
+            # Whitespace, comments
+            (r'\s+', Text),
+            (r'/\*(.|\n)*?\*/', Comment),
+            (r'(//|--).*\n', Comment),
+
+            (r'[<>\[\(\)\{\},:;=\]]', Punctuation),
+            (r'\.\.', Punctuation),
+
+            (words(('arch', 'arm11', 'caps', 'child_of', 'ia32', 'irq', 'maps',
+                    'objects'), suffix=r'\b'), Keyword),
+
+            (words(('aep', 'asid_pool', 'cnode', 'ep', 'frame', 'io_device',
+                    'io_ports', 'io_pt', 'notification', 'pd', 'pt', 'tcb',
+                    'ut', 'vcpu'), suffix=r'\b'), Keyword.Type),
+
+            # Properties
+            (words(('asid', 'addr', 'badge', 'cached', 'dom', 'domainID', 'elf',
+                    'fault_ep', 'G', 'guard', 'guard_size', 'init', 'ip',
+                    'prio', 'sp', 'R', 'RG', 'RX', 'RW', 'RWG', 'RWX', 'W',
+                    'WG', 'WX', 'level', 'masked', 'master_reply', 'paddr',
+                    'ports', 'reply', 'uncached'), suffix=r'\b'),
+                    Keyword.Reserved),
+
+            # Literals
+            (r'0[xX][\da-fA-F]+', Number.Hex),
+            (r'\d+(\.\d+)?(k|M)?', Number),
+            (words(('bits',), suffix=r'\b'), Number),
+            (words(('cspace', 'vspace', 'reply_slot', 'caller_slot',
+                    'ipc_buffer_slot'), suffix=r'\b'), Number),
+
+            # Identifiers
+            (r'[a-zA-Z_][-_@\.\w]*', Name),
+        ],
+    }
+
 class RedcodeLexer(RegexLexer):
     """
     A simple Redcode lexer based on ICWS'94.
diff --git a/tests/examplefiles/capdl_example.cdl b/tests/examplefiles/capdl_example.cdl
new file mode 100644
index 00000000..050e56a6
--- /dev/null
+++ b/tests/examplefiles/capdl_example.cdl
@@ -0,0 +1,64 @@
+#ifdef ARCH_ARM
+arch arm11
+#else
+arch ia32
+#endif
+
+objects {
+  my_ep = ep /* A synchronous endpoint */
+
+  /* Two thread control blocks */
+  tcb1 = tcb
+  tcb2 = tcb
+
+  /* Four frames of physical memory */
+  frame1 = frame (4k)
+  frame2 = frame (4k)
+  frame3 = frame (4k)
+  frame4 = frame (4k)
+
+  /* Two page tables */
+  pt1 = pt
+  pt2 = pt
+
+  /* Two page directories */
+  pd1 = pd
+  pd2 = pd
+
+  /* Two capability nodes */
+  cnode1 = cnode (2 bits)
+  cnode2 = cnode (3 bits)
+}
+caps {
+  cnode1 {
+    0x1: frame1 (RW) /* read/write */
+    0x2: my_ep (R)   /* read-only */
+  }
+  cnode2 {
+    0x1: my_ep (W)   /* write-only */
+  }
+  tcb1 {
+    vspace: pd1
+    ipc_buffer_slot: frame1
+    cspace: cnode1
+  }
+  pd1 {
+    0x10: pt1
+  }
+  pt1 {
+    0x8: frame1 (RW)
+    0x9: frame2 (R)
+  }
+  tcb2 {
+    vspace: pd2
+    ipc_buffer_slot: frame3
+    cspace: cnode2
+  }
+  pd2 {
+    0x10: pt2
+  }
+  pt2 {
+    0x10: frame3 (RW)
+    0x12: frame4 (R)
+  }
+}