first commit

backend/app/parser/record_decoder.py

@@ -0,0 +1,78 @@
+"""Bounds-checked property record reader; keeps physical indexes including binary records."""
+import re
+import base64
+import struct
+from dataclasses import dataclass
+
+MAX_RECORDS = 100_000
+MAX_RECORD_BYTES = 4 * 1024 * 1024
+
+class ParseError(ValueError):
+    pass
+
+@dataclass
+class Record:
+    index: int
+    stream: str
+    properties: dict[str,str]
+    binary: bool = False
+    @property
+    def id(self):
+        return f'{self.stream}:{self.index}'
+
+
+def properties(payload: bytes):
+    result = {}
+    # UTF8 companion fields take precedence irrespective of source order.
+    utf8 = {}
+    for item in payload.rstrip(b'\x00\r\n').split(b'|'):
+        if b'=' not in item:
+            continue
+        k,v = item.split(b'=',1)
+        key = k.decode('ascii',errors='replace').upper()
+        if key.startswith('%UTF8%'):
+            utf8[key[6:]] = v.decode('utf-8',errors='replace')
+        else:
+            result[key] = v.decode('cp1252',errors='replace')
+    result.update(utf8)
+    return result
+
+
+def binary_records(data: bytes, stream: str):
+    offset=0
+    records=[]
+    while offset < len(data):
+        if len(records)>MAX_RECORDS:
+            raise ParseError('Record count limit exceeded')
+        if len(data)-offset<4:
+            if not any(data[offset:]): break
+            raise ParseError(f'{stream}: truncated length at byte {offset}')
+        word=struct.unpack_from('<I',data,offset)[0]
+        length, kind=word&0xffffff, word>>24
+        if not 0 < length <= MAX_RECORD_BYTES or length>len(data)-offset-4:
+            raise ParseError(f'{stream}: invalid record length {length} at byte {offset}')
+        payload=data[offset+4:offset+4+length]
+        # Header is outside OWNERINDEX numbering; first object is index zero.
+        records.append(Record(len(records)-1,stream,properties(payload) if kind==0 else {'BINARY_KIND':str(kind),'BINARY_BASE64':base64.b64encode(payload).decode('ascii')},kind!=0))
+        offset+=4+length
+    return records
+
+
+def ascii_records(data: bytes):
+    lines=data.lstrip(b'\xef\xbb\xbf').splitlines()
+    records=[]
+    for line in lines:
+        if not line.strip(): continue
+        if len(line)>MAX_RECORD_BYTES: raise ParseError('Text record size limit exceeded')
+        if len(records)>MAX_RECORDS: raise ParseError('Record count limit exceeded')
+        p=properties(line)
+        if not p or not ('HEADER' in p or 'RECORD' in p):
+            raise ParseError('Malformed ASCII schematic record')
+        records.append(Record(len(records)-1,'FileHeader',p))
+    return records
+
+
+def detect_format(data):
+    if data.startswith(b'\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1'): return 'ole'
+    if re.match(br'\|HEADER=',data.lstrip(b'\xef\xbb\xbf \r\n'),re.I): return 'ascii'
+    raise ParseError('Not a supported SchDoc: expected OLE container or |HEADER= ASCII schematic')