Improve whitelisted word recognition

Pseurae · Oct 29, 2024 · 95754f1 · 95754f1
1 parent b81b139
commit 95754f1
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 18 deletions.
diff --git a/tools/decap/decap.cpp b/tools/decap/decap.cpp
@@ -36,14 +36,16 @@ int main(int argc, char *argv[])
     f.ProcessStringPointerArray(0x1e79b0, 2, 8); // Boy/Girl
     f.ProcessStringPointerArray(0x1e79c0, 5, 8); // Male preset names
     f.ProcessStringPointerArray(0x1e79e8, 5, 8); // Female preset names
-    f.ProcessStringPointerArray(0x376d04, 6); // Stat labels
-    f.ProcessStringPointerArray(0x3c5564 + 20, 349, 44); // Item descriptions
     f.ProcessStringPointerArray(0x3c1004, 25); // Nature names
     f.ProcessStringPointerArray(0x3c1068, 8); // Summary Screen headers
     f.ProcessStringPointerArray(0x39f494, 10, 8); // Party menu options
     f.ProcessStringPointerArray(0x376d24, 10, 8); // Start menu options
-    f.ProcessStringPointerArray(0x1fa110, 78); // Ability descriptions
     f.ProcessStringPointerArray(0x3e73c4 + 4, 88, 8); // Map names
+
+    f.SetWhitelist(&WhitelistEntries);
+    f.ProcessStringPointerArray(0x376d04, 6); // Stat labels
+    f.ProcessStringPointerArray(0x3c5564 + 20, 349, 44); // Item descriptions
+    f.ProcessStringPointerArray(0x1fa110, 78); // Ability descriptions
     f.ProcessStringPointerArray(0x3c09d8, 354); // Move descriptions
 
     f.Save(argv[2]);

diff --git a/tools/decap/file.cpp b/tools/decap/file.cpp
@@ -36,30 +36,22 @@ void File::Save(const std::string &filename)
     fclose(fp);
 }
 
-struct Word
-{
-    unsigned char text[10];
-    unsigned int len;
-};
-
-static const Word sWhitelistEntries[] = {
+const std::vector<Word> WhitelistEntries = {
     { {0xC2, 0xC7}, 2 }, // HM
     { {0xCE, 0xC7}, 2 }, // TM
     { {0xC2, 0xCA}, 2 }, // HP
     { {0xCA, 0xCA}, 2 }, // PP
 };
 
-bool CheckIfWhitelisted(unsigned char *buffer, unsigned int idx)
+unsigned int CheckIfWhitelisted(const std::vector<Word>& whitelist, unsigned char *buffer, unsigned int idx)
 {
-    for (int i = 0; i < (int)(sizeof(sWhitelistEntries) / sizeof(*sWhitelistEntries)); ++i)
+    for (const auto &word : WhitelistEntries)
     {
-        if (memcmp(&buffer[idx], sWhitelistEntries[i].text, sWhitelistEntries[i].len) == 0)
-        {
-            return true;
-        }
+        if (memcmp(&buffer[idx], word.text, word.len) == 0)
+            return word.len;
     }
 
-    return false;
+    return 0;
 }
 
 // 0x0 is space
@@ -81,8 +73,12 @@ void File::ProcessString(int location, int max_size)
         int idx = location + j;
         if (idx > m_Size) FATAL_ERROR("Index %d is greater than file size (%ld)\n", idx, m_Size);
 
-        if (CheckIfWhitelisted(m_Buffer, idx)) 
+        unsigned int whitelistLength;
+        if (m_Whitelist != nullptr && (whitelistLength = CheckIfWhitelisted(*m_Whitelist, m_Buffer, idx)) > 0) 
+        {
+            j += whitelistLength - 1;
             continue;
+        }
 
         unsigned char c = m_Buffer[idx];
 

diff --git a/tools/decap/file.h b/tools/decap/file.h
@@ -1,6 +1,15 @@
 #pragma once
 
 #include <string>
+#include <vector>
+
+struct Word
+{
+    unsigned char text[10];
+    unsigned int len;
+};
+
+extern const std::vector<Word> WhitelistEntries;
 
 class File
 {
@@ -14,7 +23,9 @@ class File
     void ProcessStringArray(int location, int size, int count);
     void ProcessStringArray(int location, int size, int count, int stride);
     void ProcessStringPointerArray(int location, int count, int stride = 4);
+    inline void SetWhitelist(const std::vector<Word> *whitelist) { m_Whitelist = whitelist; }
 private:
     unsigned char *m_Buffer;
     long m_Pos, m_Size;
+    const std::vector<Word> *m_Whitelist;
 };