Under oldstyle, mangled echoed unknown text to simulate old behavior.

[open-adventure.git] / misc.c

diff --git a/misc.c b/misc.c
index 2b6a58b212a7561eab784dad1a151566319cab4f..d1c68ab2d42641c0e5a36928dc1f46d6aaf21fd3 100644 (file)
--- a/misc.c
+++ b/misc.c
@@ -23,7 +23,7 @@ static void* xmalloc(size_t size)
     return (ptr);
 }
 
-void packed_to_token(long packed, char token[TOKLEN+1])
+void packed_to_token(long packed, char token[TOKLEN + 1])
 {
     // The advent->ascii mapping.
     const char advent_to_ascii[] = {
@@ -81,7 +81,7 @@ long token_to_packed(const char token[])
 
     size_t t_len = strlen(token);
     if (t_len > TOKLEN)
-       t_len = TOKLEN;
+        t_len = TOKLEN;
     long packed = 0;
     for (size_t i = 0; i < t_len; ++i) {
         char mapped = ascii_to_advent[(int) toupper(token[i])];
@@ -94,12 +94,36 @@ void tokenize(char* raw, struct command_t *cmd)
 {
     memset(cmd, '\0', sizeof(struct command_t));
 
-    /* FIXME: put a bound prefix on the %s to prevent buffer overflow */
-    int word_count = sscanf(raw, "%s%s", cmd->raw1, cmd->raw2);
+    /* Bound prefix on the %s would be needed to prevent buffer
+     * overflow.  but we shortstop this more simply by making each
+     * raw-input buffer as long as the enrire inout buffer. */
+    sscanf(raw, "%s%s", cmd->raw1, cmd->raw2);
 
     // pack the substrings
     cmd->wd1  = token_to_packed(cmd->raw1);
     cmd->wd2  = token_to_packed(cmd->raw2);
+
+    /* (ESR) In oldstyle mode, simulate the uppercasing and truncating
+     * effect on raw tokens of packing them into sixbit characters, 5
+     * to a 32-bit word.  This is something the FORTRAN version did
+     * becuse archaic FORTRAN had no string types.  Don Wood's
+     * mechanical translation of 2.5 to C retained the packing and
+     * thus this misfeature.
+     *
+     * It's philosophically questionable whether this is the right
+     * thing to do even in oldstyle mode.  On one hand, the text
+     * mangling was not authorial intent, but a result of limitations
+     * in their tools. On the other, not simulating this misbehavior
+     * goes against the goal of making oldstyle as accurate as
+     * possible an emulation of the original UI.
+     */
+    if (settings.oldstyle) {
+       cmd->raw1[TOKLEN+TOKLEN] = cmd->raw1[TOKLEN+TOKLEN] = '\0';
+       for (int i = 0; i < strlen(cmd->raw1); i++)
+           cmd->raw1[i] = toupper(cmd->raw1[i]);
+       for (int i = 0; i < strlen(cmd->raw2); i++)
+           cmd->raw2[i] = toupper(cmd->raw2[i]);
+    }
 }
 
 /* Hide the fact that wods are corrently packed longs */
@@ -274,16 +298,16 @@ int word_count(char* str)
     int inblanks = true;
 
     for (char *s = str; *s; s++)
-       if (inblanks) {
-           if (strchr(delims, *s) == 0) {
-               ++count;
-               inblanks = false;
-           }
-       } else {
-           if (strchr(delims, *s) != 0) {
-               inblanks = true;
-           }
-       }
+        if (inblanks) {
+            if (strchr(delims, *s) == 0) {
+                ++count;
+                inblanks = false;
+            }
+        } else {
+            if (strchr(delims, *s) != 0) {
+                inblanks = true;
+            }
+        }
 
     return (count);
 }
@@ -483,6 +507,10 @@ int get_special_vocab_id(const char* word)
 long get_vocab_id(const char* word)
 // Search the vocab categories in order for the supplied word.
 {
+    /* Check for an empty string */
+    if (strncmp(word, "", sizeof("")) == 0)
+        return (WORD_EMPTY);
+
     long ref_num;
 
     /* FIXME: Magic numbers related to vocabulary */
@@ -645,7 +673,7 @@ long randrange(long range)
     return range * get_next_lcg_value() / game.lcg_m;
 }
 
-void make_zzword(char zzword[TOKLEN+1])
+void make_zzword(char zzword[TOKLEN + 1])
 {
     for (int i = 0; i < 5; ++i) {
         zzword[i] = 'A' + randrange(26);