bwahh changes!!

6 files changed, 143 insertions, 20 deletions

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..d56a83a
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+pu
+putest
\ No newline at end of file
diff --git a/README.md b/README.md
index 5a8b117..dcc0fbf 100644
--- a/README.md
+++ b/README.md
@@ -13,4 +13,11 @@ ikr,,,
 ## instruction set?
 check the comments in pu.c
 
-i still haven't added docs for the assembler, but ehh it's pretty self explanatory. ask ai to explain it to you if you can't be assed to read.
\ No newline at end of file
+i still haven't added docs for the assembler, but ehh it's pretty self explanatory. ask ai to explain it to you if you can't be assed to read.
+
+## t?
+that file is the instruction encoding table for the assembler. you can add shit to it if you wanna. this is the original:
+```c
+char *R="hlt\0 ar1\2 ar2\3 ar3\4 ar4\5 ar5\6 ar6\7 ar7\x8 ar8\x9 r1a\xA r2a\xB r3a\xC r4a\xD r5a\xE r6a\xF r7a\20 r8a\21 mem\22 aca\23 add\24 sub\25 not\26 clr\27";
+```
+removing it means you can't assemble stuff anymore. though the pu.c wrapper does write it for you if it doesn't exist.
\ No newline at end of file
diff --git a/code.pasm b/code.pasm
new file mode 100644
index 0000000..969f96b
--- /dev/null
+++ b/code.pasm
@@ -0,0 +1,6 @@
+# charcode for c
+con 99
+
+ar4 con 60 ar2 r4a mem con 1 ar4 con 60 ar3 con 0 ar2 con 2 hlt
+
+hlt
diff --git a/pu.c b/pu.c
index a6941fe..4e93a31 100644
--- a/pu.c
+++ b/pu.c
@@ -1,12 +1,16 @@
+#include <stdio.h>
 #include <stdlib.h>
+#include <unistd.h>
 #include <string.h>
-#include <stdio.h>
 #pragma clang diagnostic ignored "-Wreturn-type"
 #pragma clang diagnostic ignored "-Wc23-extensions"
 
+// clang-format off
 #define S 65536
 #define H !--c?
-m[S],c,a;pu(){while((c=m[m[7]++]))H*m=m[m[7]++]:c--<9?a?m[c]=*m:0:8>(c-=8)?*m=m[c]:!(c-=8)?m[m[1]]=*m:H*m=a:H a+=*m:H a-=*m:H a=!a:H a=0:0;}
+#define J while
+m[S],c,a;pu(){J((c=m[m[7]++]))H*m=m[m[7]++]:c--<9?!a?m[c]=*m:0:8>(c-=8)?*m=m[c]:!(c-=8)?m[m[1]]=*m:H*m=a:H a+=*m:H a-=*m:H a=!a:H a=0:0;}
+// clang-format on
 
 // a is acc
 // first 8 memory bytes are registers actually
@@ -15,7 +19,7 @@ m[S],c,a;pu(){while((c=m[m[7]++]))H*m=m[m[7]++]:c--<9?a?m[c]=*m:0:8>(c-=8)?*m=m[
 // --- opcodes ---
 // 0 = halt DONE
 // 1 = const in r0 DONE
-// 2...9 = r0 in rX (only if acc is non-zero) DONE
+// 2...9 = r0 in rX (only if acc is zero) DONE
 // 10...17 = rX in r0 DONE
 // 18 = r0 in m[r1] DONE
 // 19 = acc in r0 DONE
@@ -29,28 +33,127 @@ m[S],c,a;pu(){while((c=m[m[7]++]))H*m=m[m[7]++]:c--<9?a?m[c]=*m:0:8>(c-=8)?*m=m[
 // compiler!
 // the switch skips whitespace. then, the beautiful loop thingymajigy emits bytecode.
 
+// clang-format off
 #define W case
 #define Q goto M;
-char *R="hlt\0 ar1\2 ar2\3 ar3\4 ar4\5 ar5\6 ar6\7 ar7\x8 ar8\x9 r1a\xA r2a\xB r3a\xC r4a\xD r5a\xE r6a\xF r7a\20 r8a\21 mem\22 aca\23 add\24 sub\25 not\26 clr\27";K;as(char*I,int*O){char*V=R;while((K=*I)){V=R;switch(K){case 32:case 10:case 9:I++;Q}while(*V)V+=3,(strncmp(V-3,I,3)?:({*O++=*V;I+=3;Q})),V+=2;strncmp("con ",I,4)?:({*O++=1,*O++=atoi(I+=4);while(*I>47&*I<58)I++;Q});return 1;M:}return 0;}
+char R[]={
+#embed"t"
+};K;char*as(char*I,int*O){char*V=R;J((K=*I)){V=R;switch(K){W 32:W 10:W 9:I++;Q;W 35:J(*++I!=10);Q}strncmp("con ",I,4)?:({*O++=1,*O++=atoi(I+=4);J(*I>47&*I<58)I++;Q});J(*V)V+=3,(strncmp(V-3,I,3)?:({*O++=*V;I+=3;Q})),V+=2;return I;M:}return 0;}
+// clang-format on
 
 #undef Q
 #undef H
 #undef W
 #undef S
-
-// ----------
+#undef J
 
 #ifndef __WRAPPER
-int main() {
-    // __asm__("int $3");
-    m[ 7 ]    = 1024;
-    m[ 1024 ] = 1;
-    m[ 1025 ] = 9;
-    m[ 1026 ] = 1;
-    m[ 1027 ] = 5;
-
-    printf("before: %i\n", m[ 0 ]);
-    pu();
-    printf("after: %i\n", m[ 0 ]);
+
+// runs pu until exit code != 2,
+// that exit code is an interrupt for talking to the emulated hardware
+int emu() {
+    while (1) {
+        pu();
+        int e_code = m[ 0 ];
+
+        if (e_code != 2) return e_code;
+
+        // handle hardware interrupts
+        int int_code = m[ 1 ];
+
+        // 0: output string
+        if (int_code == 0) {
+            int ptr  = m[ 2 ];
+            int size = m[ 3 ], i = 0;
+
+            for (; i < size; i++) putchar(m[ ptr++ ]);
+            write(STDOUT_FILENO, &m[ ptr ], size);
+        }
+
+        // 1: input character
+        else if (int_code == 1) {
+            m[ 2 ] = getchar();
+        }
+
+        else {
+            // unknown interrupt. what to do?
+        }
+
+        m[ 0 ] = 0;
+        m[ 7 ]++;
+    }
+}
+
+char R_backup[] = "hlt\0 ar1\2 ar2\3 ar3\4 ar4\5 ar5\6 ar6\7 ar7\x8 ar8\x9 r1a\xA r2a\xB r3a\xC "
+                  "r4a\xD r5a\xE r6a\xF r7a\20 r8a\21 mem\22 aca\23 add\24 sub\25 not\26 clr\27";
+
+int main(int argc, char **argv) {
+    if (access("t", F_OK)) {
+        // patch the instruction table incase you removed it.
+        printf("you've deleted t! i'll write it again for you.\n\n");
+        FILE *fp = fopen("t", "w+");
+        fwrite(R_backup, 1, sizeof(R_backup), fp);
+        fclose(fp);
+        memcpy(R, R_backup, sizeof(R_backup));
+    }
+
+    if (argc == 1) {
+        printf("Usage:\n    %s <filename>", argv[ 0 ]);
+        exit(1);
+    }
+
+    // read code
+    FILE *fp = fopen(argv[ 1 ], "r");
+    fseek(fp, 0, SEEK_END);
+    int sz = ftell(fp);
+    rewind(fp);
+    char code[ sz + 1 ];
+    fread(code, 1, sz, fp);
+    code[ sz ] = 0;
+    fclose(fp);
+
+    // set pc
+    m[ 7 ] = 1024;
+    // assemble
+    char *fail_point = as(code, &m[ 1024 ]);
+
+    if (fail_point != 0) {
+        int error_pos = (int) (fail_point - code);
+        int i;
+
+        int line = 1, col = 1;
+        for (i = 0; i < error_pos && i < sz; i++) {
+            if (code[ i ] == '\n') {
+                line++;
+                col = 1;
+            } else {
+                col++;
+            }
+        }
+
+        fprintf(
+            stderr, "Assembly failed at line %d, column %d (position %d):\n", line, col, error_pos);
+
+        int line_start = error_pos;
+        while (line_start > 0 && code[ line_start - 1 ] != '\n') { line_start--; }
+
+        int line_end = error_pos;
+        while (line_end < sz && code[ line_end ] != '\n' && code[ line_end ] != '\0') {
+            line_end++;
+        }
+
+        fprintf(stderr, "  ");
+        for (i = line_start; i < line_end; i++) { fputc(code[ i ], stderr); }
+        fprintf(stderr, "\n");
+
+        fprintf(stderr, "  ");
+        for (i = line_start; i < error_pos; i++) { fputc(' ', stderr); }
+        fprintf(stderr, "^ parsing stopped here\n");
+
+        exit(1);
+    }
+
+    // run emulation
+    emu();
 }
 #endif
\ No newline at end of file
diff --git a/pu.out.c b/pu.out.c
index 330e6f0..4d0c9af 100644
--- a/pu.out.c
+++ b/pu.out.c
@@ -3,14 +3,19 @@
 #pragma clang diagnostic ignored "-Wreturn-type"
 #pragma clang diagnostic ignored "-Wc23-extensions"
 
+// clang-format off
 #define S 65536
 #define H !--c?
 #define W case
 #define Q goto M;
-m[S],c,a;pu(){while((c=m[m[7]++]))H*m=m[m[7]++]:c--<9?a?m[c]=*m:0:8>(c-=8)?*m=m[c]:!(c-=8)?m[m[1]]=*m:H*m=a:H a+=*m:H a-=*m:H a=!a:H a=0:0;}
-char *R="hlt\0 ar1\2 ar2\3 ar3\4 ar4\5 ar5\6 ar6\7 ar7\x8 ar8\x9 r1a\xA r2a\xB r3a\xC r4a\xD r5a\xE r6a\xF r7a\20 r8a\21 mem\22 aca\23 add\24 sub\25 not\26 clr\27";K;as(char*I,int*O){char*V=R;while((K=*I)){V=R;switch(K){case 32:case 10:case 9:I++;Q}while(*V)V+=3,(strncmp(V-3,I,3)?:({*O++=*V;I+=3;Q})),V+=2;strncmp("con ",I,4)?:({*O++=1,*O++=atoi(I+=4);while(*I>47&*I<58)I++;Q});return 1;M:}return 0;}
+#define J while
+m[S],c,a;pu(){J((c=m[m[7]++]))H*m=m[m[7]++]:c--<9?!a?m[c]=*m:0:8>(c-=8)?*m=m[c]:!(c-=8)?m[m[1]]=*m:H*m=a:H a+=*m:H a-=*m:H a=!a:H a=0:0;}char R[]={
+#embed"t"
+};K;as(char*I,int*O){char*V=R;J((K=*I)){V=R;switch(K){W 32:W 10:W 9:I++;Q;W 35:J(*++I!=10);Q}strncmp("con ",I,4)?:({*O++=1,*O++=atoi(I+=4);J(*I>47&*I<58)I++;Q});J(*V)V+=3,(strncmp(V-3,I,3)?:({*O++=*V;I+=3;Q})),V+=2;return 1;M:}return 0;}
+// clang-format on
 
 #undef S
 #undef H
+#undef J
 #undef Q
 #undef W
\ No newline at end of file
diff --git a/t b/t
new file mode 100644
index 0000000..7b992ff
--- /dev/null
+++ b/t