utils.py: Merge checkValidHeader and validateHeaderMetadata

src/tja2fumen/parsers.py

@@ -292,13 +292,9 @@ def readFumen(fumenFile, byteOrder=None, debug=False):
         file = fumenFile
     size = os.fstat(file.fileno()).st_size
 
-    # Check for valid fumen header (first 432 bytes) using valid byte substrings
+    # Check for valid fumen header (first 512 bytes) using valid byte substrings
-    fumenHeader = file.read(432)
+    fumenHeader = file.read(512)
-    if not checkValidHeader(fumenHeader):
+    checkValidHeader(fumenHeader)
-        debugPrint(f"Invalid header!")
-    # Read the next 80 bytes, which contains unknown information
-    fumenHeaderUnknown = file.read(80)
-    # validateHeaderMetadata(fumenHeaderUnknown)
 
     # Determine:
     #   - The byte order (big or little endian)
@@ -319,8 +315,8 @@ def readFumen(fumenFile, byteOrder=None, debug=False):
 
     # Initialize the dict that will contain the chart information
     song = {'measures': []}
-    song['headerPadding'] = fumenHeader
+    song['headerPadding'] = fumenHeader[:432]
-    song['headerMetadata'] = fumenHeaderUnknown
+    song['headerMetadata'] = fumenHeader[-80:]
     song['order'] = order
     song["length"] = totalMeasures
 

src/tja2fumen/utils.py

@@ -5,28 +5,30 @@ import math
 from tja2fumen.constants import simpleHeaders, byte_strings
 
 
-def checkValidHeader(header):
+def checkValidHeader(headerBytes):
-    # These simple headers (substrings repeated 36 times) are used for many Gen2 systems (AC, Wii, etc.)
+    # Fumen headers should contain 512 bytes.
-    if header in simpleHeaders:
+    assert len(headerBytes) == 512
-        return True
+    # The header for fumens can be split into two groups: The first 432 bytes, and the last 80 bytes.
-    # Starting with Gen3, they began using unique headers for every song. (3DS and PSPDX are the big offenders.)
+    headerPadding = headerBytes[:432]
-    #   They seem to be some random combination of b_x00 + one of the non-null byte substrings.
+    headerMetadata = headerBytes[-80:]
-    #   To avoid enumerating every combination of 432 bytes, we do a lazy check instead.
+
-    elif (byte_strings['x00'] in header and
+    # 1. Check the header's padding bytes for several possible combinations
-            any(b in header for b in [byte_strings[key] for key in ['431', '432', '433', '434', 'V1', 'V2', 'V3']])):
+    # 1a. These simple headers (substrings repeated 36 times) are used for many Gen2 systems (AC, Wii, etc.)
-        return True
+    cond1 = headerPadding in simpleHeaders
-    # The PS4 song 'wii5op' is a special case: It throws in this odd 'g1' string in combo with other substrings.
+    # 1b. Starting with Gen3, they began using unique headers for every song. (3DS and PSPDX are the big offenders.)
-    elif (byte_strings['g1'] in header and
+    #   - They seem to be some random combination of b_x00 + one of the non-null byte substrings.
-            any(b in header for b in [byte_strings[key] for key in ['431', 'V2']])):
+    #   - To avoid enumerating every combination of 432 bytes, we do a lazy check instead.
-        return True
+    cond2 = (byte_strings['x00'] in headerPadding and
+             any(b in headerPadding for b in
+                 [byte_strings[key] for key in ['431', '432', '433', '434', 'V1', 'V2', 'V3']]))
+    # 1c. The PS4 song 'wii5op' is a special case: It throws in this odd 'g1' string in combo with 2 other substrings.
+    cond3 = (byte_strings['g1'] in headerPadding and
+             any(b in headerPadding for b in [byte_strings[key] for key in ['431', 'V2']]))
     # Otherwise, this is some unknown header we haven't seen before.
-    # Typically, these will be tja2bin.exe converted files with a completely invalid header.
+    assert cond1 or cond2 or cond3, "Header padding bytes do not match expected fumen byte substrings."
-    else:
-        return False
 
-
+    # 2. Check the header's metadata bytes
-def validateHeaderMetadata(headerBytes):
+    for idx, val in enumerate(headerMetadata):
-    for idx, val in enumerate(headerBytes):
         # 0. Unknown
         # Notes:
         #   * Breakdown of distribution of different byte combinations: