Add scripts to work with racer.tab

README.md

@@ -6,5 +6,7 @@ This is a collection of small scripts to extract files from the 1999 Game "Star
 - out_splineblock.py: Extract out_splineblock.bin to Wavefront OBJ files
 - out_spriteblock.py: Extract out_spriteblock.bin to PNG files
 - out_textureblock.py: Extract out_textureblock.bin to PNG files
+- extract-racer-tab.py: Extracts strings from swep1rcr.exe which are typcially translated
+- parse-racer-tab.py: Validates racer.tab translation files
 
 Licensed under GPL version 2 or later.

extract-racer-tab.py

@@ -0,0 +1,49 @@
+#!/usr/bin/env python3
+
+import sys
+
+strings = {}
+
+for path in sys.argv[1:]:
+  with open(path, 'rb') as in_file:
+    data = in_file.read()
+
+    identifiers = [
+      b"/LANGID",
+      b"/CREDITS_H_",
+      b"/MONDOTEXT_H_",
+      b"/SCREENTEXT_"
+    ]
+
+    for identifier in identifiers:
+      cursor = 0
+      while True:
+        start = data.find(identifier, cursor)
+        if start == -1:
+          break
+        sep = data.find(b"/", start + 1)
+        assert(sep > start)
+        end = data.find(b"\0", sep + 1)
+        assert(end > sep)
+
+        key = data[start + 1:sep]
+        value = data[sep + 1:end]
+
+        tmp = value.decode('windows-1252')
+        tmp = tmp.translate(str.maketrans({"\n": r"\n",
+                                           "\t": r"\t",
+                                           "\r": r"\r",
+                                           "\"": r"\"",
+                                           "\\": r"\\"}))
+        value = tmp.encode('windows-1252')
+
+        if key in strings:
+          print("Warning: duplicate key " + str(key), file=sys.stderr)
+        strings[key] = value
+
+        cursor = end + 1
+
+for key in strings:
+  value = strings[key].decode('windows-1252')
+  key = key.decode('ascii')
+  print(key + "\t" + value + "\r\n", end='', flush=True)

parse-racer-tab.py

@@ -0,0 +1,40 @@
+#!/usr/bin/env python3
+
+import sys
+
+strings = {}
+
+for path in sys.argv[1:]:
+  with open(path, 'rb') as in_file:
+    data = in_file.read()
+
+    # FIXME: Make sure first 4 bytes are not 'ENCR' or decrypt file
+
+    cursor = 0
+    while(cursor < len(data)):
+      start = cursor
+
+      # Find end of this entry
+      while(cursor < len(data)):
+        if (data[cursor] == 10 or data[cursor] == 13):
+          break
+        cursor += 1
+      end = cursor
+
+      sep = data.find(9, start)
+      key = data[start:sep]
+      value = data[sep + 1:end]
+      if key in strings:
+        print("Warning: duplicate key " + str(key), file=sys.stderr)
+      strings[key] = value
+
+      # Move to next entry
+      while(cursor < len(data)):
+        if (data[cursor] != 10 and data[cursor] != 13):
+          break
+        cursor += 1
+
+for key in strings:
+  value = strings[key].decode('windows-1252')
+  key = key.decode('ascii')
+  print("'" + key + "'  ->  '" + value + "'")