added support for xml schemas

example4.xml

@@ -0,0 +1,68 @@
+<schema>
+    <structure id="main" bits="32">
+        <range start="28" end="31" name="cond"></range>
+        <range start="27" end="27" name="0"></range>
+        <range start="26" end="26" name="1"></range>
+        <range start="25" end="25" name="I"></range>
+        <range start="24" end="24" name="P">
+            <description>pre / post indexing bit</description>
+            <values>
+                <case value="0">post, add offset after transfer</case>
+                <case value="1">pre, add offset before transfer</case>
+            </values>
+        </range>
+        <range start="23" end="23" name="U">
+            <description>up / down bit</description>
+            <values>
+                <case value="0">down, subtract offset from base</case>
+                <case value="1">up, addition offset to base</case>
+            </values>
+        </range>
+        <range start="22" end="22" name="B">
+            <description>byte / word bit</description>
+            <values>
+                <case value="0">transfer word quantity</case>
+                <case value="1">transfer byte quantity</case>
+            </values>
+        </range>
+        <range start="21" end="21" name="W">
+            <description>write-back bit</description>
+            <values>
+                <case value="0">no write-back</case>
+                <case value="1">write address into base</case>
+            </values>
+        </range>
+        <range start="20" end="20" name="L">
+            <description>load / store bit</description>
+            <values>
+                <case value="0">store to memory</case>
+                <case value="1">load from memory</case>
+            </values>
+        </range>
+        <range start="16" end="19" name="Rn">
+            <description>base register</description>
+        </range>
+        <range start="12" end="15" name="Rd">
+            <description>source / destination register</description>
+        </range>
+        <range start="0" end="11" name="offset" depends-on="25">
+            <values>
+                <case value="0" structure="immediateOffset">offset is an immediate value</case>
+                <case value="1" structure="registerOffset">offset is a register</case>
+            </values>
+        </range>
+    </structure>
+    <structure id="immediateOffset" bits="12">
+        <range start="0" end="11" name="12-bit immediate offset">
+            <description>unsigned number</description>
+        </range>
+    </structure>
+    <structure id="registerOffset" bits="12">
+        <range start="4" end="11" name="shift">
+            <description>shift applied to Rm</description>
+        </range>
+        <range start="0" end="3" name="Rm">
+            <description>offset register</description>
+        </range>
+    </structure>
+</schema>

schema.py

@@ -5,12 +5,13 @@ import yaml
 from config import Config
 from renderer import Renderer
 from structure import Structure
+from xml_loader import XMLLoader
 
 class UnsupportedFormatException(Exception):
     ...
 
 class InstructionSetSchema:
-    VALID_EXTENSIONS = ("yaml", "json")
+    VALID_EXTENSIONS = ("yaml", "json", "xml")
     
     def __init__(self, path: str, configPath: str = "config.json") -> None:
         self.config = Config(configPath)
@@ -31,6 +32,9 @@ class InstructionSetSchema:
         
             elif ext == "json":
                 schema = json.load(f)
+            
+            elif ext == "xml":
+                schema = XMLLoader.load(f)
 
         self.structures = {}
         

structure.py

@@ -28,7 +28,7 @@ class Structure:
                     values[v] = d
                 range_.values = values
         
-        return Structure(id_, data["bits"], ranges, data.get("start", 0))
+        return Structure(id_, int(data["bits"]), ranges, data.get("start", 0))
 
     def getSortedRanges(self) -> list[Range]:
         ranges = self.ranges.values()

xml_loader.py

@@ -0,0 +1,65 @@
+from __future__ import annotations
+from bs4 import BeautifulSoup
+from typing import TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from io import TextIOWrapper
+
+class XMLLoader:
+    def load(file_: TextIOWrapper) -> dict:
+        schema = {}
+        bs = BeautifulSoup(file_.read(), "xml")
+        structElmts = bs.findAll("structure")
+        structures = {}
+        
+        for structElmt in structElmts:
+            structures[structElmt.get("id")] = XMLLoader.parseStructure(structElmt)
+        
+        schema["structures"] = structures
+        return schema
+    
+    def parseStructure(structElmt: any) -> dict:
+        struct = {}
+        struct["bits"] = structElmt.get("bits")
+        ranges = {}
+        rangeElmts = structElmt.findAll("range")
+        for rangeElmt in rangeElmts:
+            span = rangeElmt.get("end") + "-" + rangeElmt.get("start")
+            ranges[span] = XMLLoader.parseRange(rangeElmt)
+        
+        struct["ranges"] = ranges
+        return struct
+
+    def parseRange(rangeElmt: any) -> dict:
+        range_ = {}
+        range_["name"] = rangeElmt.get("name")
+        desc = rangeElmt.find("description")
+        if desc is not None: range_["description"] = desc.getText()
+        
+        valuesElmt = rangeElmt.find("values")
+        if valuesElmt is not None:
+            range_["values"] = XMLLoader.parseValues(valuesElmt)
+        
+        if rangeElmt.get("depends-on"):
+            range_["depends-on"] = rangeElmt.get("depends-on")
+        
+        return range_
+
+    def parseValues(valuesElmt: any) -> dict:
+        values = {}
+        caseElmts = valuesElmt.findAll("case")
+        
+        for caseElmt in caseElmts:
+            val = caseElmt.get("value")
+            desc = caseElmt.getText()
+            struct = caseElmt.get("structure")
+            
+            if struct:
+                values[val] = {
+                    "description": desc,
+                    "structure": struct
+                }
+            else:
+                values[val] = desc
+
+        return values