fix: line accounting is less buggy

4 files changed, 118 insertions, 58 deletions

diff --git a/sys/cmd/cc/cc.c b/sys/cmd/cc/cc.c
index 6bc363c..87d9a35 100644
--- a/sys/cmd/cc/cc.c
+++ b/sys/cmd/cc/cc.c
@@ -99,30 +99,28 @@ END:
 Io*
 openio(byte *path)
 {
-    Io     *it;
+    string *it, *end;
     Stream *f;
 
     intern(&path);
 
     // See if we have already opened file;
     // If so, and it hasn't been flagged return it
-    for (it = C.iostk; it != C.io; ++it) {
-        if ((uintptr)it->path == (uintptr)path) {
-            if (it->kind & IOonce) {
-                return nil;
-            }
-            return it;
-        }
+    for (it = C.omit.path, end = it + C.omit.len; it < end; ++it) {
+        if ((uintptr)(*it) == (uintptr)(path))
+            return nil;
     }
+
     printf("OPENING PATH %s\n", path);
 
     if ((C.io - C.iostk) >= arrlen(C.iostk)-1) 
         panicf("out of I/O space!");
 
-    C.io->f    = io·open(path, "r");
-    if (!C.io->f) {
+    C.io->f = io·open(path, "r");
+    if (!C.io->f)
         panicf("file %s not found", path);
-    }
+
+    C.io->kind = IOfile;
     C.io->path = path;
     bufio·initreader(&C.io->buf, asrdr(io·read), C.io->f);
 
@@ -151,21 +149,28 @@ makeio()
 } 
 #undef asrdr
 
-// TODO: Think about if this is always at the _end_ of the stack.
-//       Right now we don't have access to it.
 void
 freeio(Io *io)
 {
-    if (io->kind & ~IOmac) {
-        free(io->b);
-    } else {
+    if (io->kind & IOfile) {
         io·close(io->f);
     }
+    io->kind  = 0;
     io->link  = nil;
     io->path  = nil;
     io->store = (Pos){ 0 };
 }
 
+void
+pushomit(string omit)
+{
+    if (C.omit.len == C.omit.cap) {
+        C.omit.cap += 20;
+        C.omit.path = realloc(C.omit.path, C.omit.cap*sizeof(*C.omit.path));
+    }
+    C.omit.path[C.omit.len++] = omit;
+}
+
 // -----------------------------------------------------------------------
 // universal compiler builtins
 
@@ -234,8 +239,13 @@ init(void)
     C.inc.dir = calloc(C.inc.cap, sizeof(*C.inc.dir));
     C.inc.dir[C.inc.len++] = ".";
 
-    C.outfile = nil;
-    C.io      = C.iostk;
+    C.omit.len  = 0;
+    C.omit.cap  = 100;
+    C.omit.path = calloc(C.omit.cap, sizeof(*C.inc.dir));
+
+    C.outfile  = nil;
+    C.io       = C.iostk;
+    C.io->link = nil;
     memset(C.iostk, 0, sizeof(C.iostk));
 
     C.lxr     = (Lexer){ 0 };
@@ -248,7 +258,9 @@ compile(byte *path)
     Token tok;
     byte *p, out[400];
 
+    intern(&path);
     strcpy(out, path);
+
     p = utf8·findrrune(out, '/');
     if (p) 
         *p++ = '\0';
@@ -268,7 +280,12 @@ compile(byte *path)
         }
     }
 
-    C.lxr.io = openio(path);
+    C.lxr.io  = openio(path);
+    C.lxr.pos = (Pos){
+        .path = path,
+        .line = 1,
+        .col  = 1,
+    };
     while (tok = lex(&C.lxr), tok.kind > Aeof) {
         puttok(tok);
     }
diff --git a/sys/cmd/cc/cc.h b/sys/cmd/cc/cc.h
index 84f173f..0e525be 100644
--- a/sys/cmd/cc/cc.h
+++ b/sys/cmd/cc/cc.h
@@ -235,8 +235,9 @@ struct SymTab
   Sym    **vals;
 };
 
-Sym  *lookup(SymTab *tab, string ident);
 Sym  *define(SymTab *tab, string ident, int kind);
+Sym  *lookup(SymTab *tab, string ident);
+error forget(SymTab *tab, string ident);
 
 struct Lexer
 {
@@ -363,8 +364,8 @@ struct Decl
 enum
 {
     IOnil  = iota(0),
-    IOonce = iota(1),
-    IOmac  = iota(2),
+    IOfile = iota(1),
+    IObuff = iota(2),
 };
 
 struct Io
@@ -381,6 +382,7 @@ struct Io
     struct Io *link;
 };
 
+/* cc.c io functions */
 Io*  openio(byte *path);
 Io*  makeio();
 void freeio(Io *io);
@@ -396,6 +398,7 @@ struct StrTab
   int32  *vals;
 };
 
+/* cc.c string functions */
 int32  intern(byte **str);
 string internview(byte* beg, byte *end);
 
@@ -413,6 +416,11 @@ struct Compiler
 
     Io     *io;
     Io      iostk[100];
+    struct {
+        int      cap;
+        int      len;
+        string  *path;
+    } omit;
 
     string outfile;
 
@@ -420,6 +428,8 @@ struct Compiler
 }; 
 extern Compiler C;
 
-void   init();
+/* cc.c compiler functions */
+void pushomit(string omit);
+void init();
 
 #undef iota
diff --git a/sys/cmd/cc/lex.c b/sys/cmd/cc/lex.c
index 90f282a..20c6f9c 100644
--- a/sys/cmd/cc/lex.c
+++ b/sys/cmd/cc/lex.c
@@ -39,10 +39,8 @@ getnsbyte(Lexer *lx)
     for (;;) {
         if (b >= RuneSelf || !isspace(b))
             return b;
-        if (b == '\n') {
-            lx->pos.line++;
+        if (b == '\n')
             return b;
-        }
         b = getbyte(lx);
     }
     return b;
@@ -257,9 +255,12 @@ Dispatch:
         goto TNum;
 
     switch (b) {
-    case ' ': case '\n': case '\r': case '\t': case '\v': case '\f':
+    case '\n': 
+        lx->pos.line++;
+    case ' ': case '\r': case '\t': case '\v': case '\f':
         while (b = getbyte(lx), isspace(b)) 
-            if (b == '\n') lx->pos.line++;
+            if (b == '\n') 
+                lx->pos.line++;
         goto Dispatch;
 
     case '\'':
@@ -363,7 +364,6 @@ Dispatch:
         else if (b == '/') { 
             while (b != EOF && b != '\n') 
                 b = getbyte(lx);
-            lx->pos.line++;
             goto Dispatch;
         } else if (b == '*') {
             int level = 1;
@@ -378,7 +378,8 @@ Dispatch:
                     if (b == '/') 
                         level--;
                 } 
-                if (b == '\n') lx->pos.line++;
+                if (b == '\n') 
+                    lx->pos.line++;
                 b = getbyte(lx);
             }
             goto Dispatch;
@@ -396,7 +397,11 @@ Dispatch:
     break;
 
     case EOF:
-        panicf("need to implement popio");
+        popio(lx);
+        if (lx->io)
+            goto GetByte;
+        tok.kind = Aeof;
+        return tok;
 
     CASE1('(', Alparen)
     CASE1(')', Arparen)
@@ -593,8 +598,8 @@ pushio(Lexer *lx, Io *new)
     lx->io        = new;
 
     lx->pos = (Pos){
-        .line = 0,
-        .col  = 0,
+        .line = 1,
+        .col  = 1,
         .path = new->path,
     };
 }
@@ -608,6 +613,11 @@ popio(Lexer *lx)
     if (!prev) {
         panicf("no buffer left");
     }
+    /* 
+     * NOTE: should we copy over unget bytes here?
+     *       would modify the buffer...
+     */
+    freeio(lx->io);
 
     lx->pos = prev->store;
     lx->io  = prev;
@@ -619,18 +629,6 @@ popio(Lexer *lx)
 #define PTR_HASH(p)        (uintptr)(p)
 #define PTR_EQUAL(p1, p2) ((uintptr)(p1) == (uintptr)(p2))
 
-Sym*
-lookup(SymTab *tab, string ident)
-{
-    int idx;
-    MAP_GET(idx, tab, ident, PTR_HASH, PTR_EQUAL);
-
-    if (idx < tab->n_buckets)
-        return tab->vals[idx];
-
-    return nil;
-}
-
 static
 int
 moresymtab(SymTab *tab, int n)
@@ -662,6 +660,32 @@ define(SymTab *tab, string name, int kind)
     return sym;
 }
 
+Sym*
+lookup(SymTab *tab, string ident)
+{
+    int idx;
+    MAP_GET(idx, tab, ident, PTR_HASH, PTR_EQUAL);
+
+    if (idx < tab->n_buckets)
+        return tab->vals[idx];
+
+    return nil;
+}
+
+
+error
+forget(SymTab *tab, string ident)
+{
+    int idx;
+    MAP_GET(idx, tab, ident, PTR_HASH, PTR_EQUAL);
+
+    if (idx < tab->n_buckets) {
+        MAP_DEL(tab, idx);
+        return 0;
+    }
+    return 1;
+}
+
 // -----------------------------------------------------------------------
 // error reporting
 
@@ -671,7 +695,7 @@ errorat(Pos x, byte *fmt, ...)
     va_list args;
     va_start(args, fmt);
 
-    printf("error %d: ", x.line);
+    printf("error:%s:%d:%d: ", os·basename(x.path), x.line, x.col);
     vprintf(fmt, args);
     printf("\n");
 
diff --git a/sys/cmd/cc/pp.c b/sys/cmd/cc/pp.c
index ceb9d66..30bb3d4 100644
--- a/sys/cmd/cc/pp.c
+++ b/sys/cmd/cc/pp.c
@@ -62,7 +62,7 @@ defmacro(Lexer *lx, string name, string macro)
 {
     Sym *mac;
 
-    printf("DEFINING MACRO %s\n", name);
+    printf("DEFINING MACRO %s, ON LINE %d\n", name, lx->pos.line);
     mac = define(&lx->sym, name, Smacro);
     mac->macro = macro;
 
@@ -400,6 +400,9 @@ ppend(Lexer *lx)
         b = getnsbyte(lx);
     } while (b > 0 && b != '\n');
 
+    if (b == '\n')
+        lx->pos.line++;
+
     return 0;
 }
 
@@ -409,15 +412,17 @@ static
 error
 ppund(Lexer *lx)
 {
-    Sym   *sym;
     string s;
+    error err;
 
     s = ident(lx);
     intern(&s);
 
-    sym = lookup(&lx->sym, s);
-    if (!sym) {
+    printf("FORGETTING %s\n", s);
+    err = forget(&lx->sym, s);
+    if (err) {
         errorat(lx->pos, "attempting to undefine unrecognized symbol '%s'", s);
+        return err;
     }
     ppend(lx);
     return 0;
@@ -443,7 +448,7 @@ ppdef(Lexer *lx)
     sym = lookup(&lx->sym, s);
     if (sym) {
         errorat(lx->pos, "macro redefined: '%s'", sym->name);
-        goto Bad;
+        // goto Bad;
     }
 
     n = 0;
@@ -537,6 +542,7 @@ ppdef(Lexer *lx)
             b = getbyte(lx);
             /* unix */
             if (b == '\n') {
+                lx->pos.line++;
                 b = getbyte(lx);
                 continue;
             }
@@ -544,23 +550,24 @@ ppdef(Lexer *lx)
             if (b == '\r') {
                 b = getbyte(lx);
                 if (b == '\n') {
+                    lx->pos.line++;
                     b = getbyte(lx);
                     continue;
                 }
             }
             str·appendbyte(&buf, '\\');
         }
-        if (b == '\n')
+        if (b == '\n') {
+            lx->pos.line++;
             break;
+        }
 
         if (b == '#' && n > 0) {
-            panicf("needs implementation");
+            panicf("tokenizer needs implementation");
         }
 
         str·appendbyte(&buf, b);
         b = getbyte(lx);
-        if (b == '\n')
-            lx->pos.line++;
         if (b == EOF) {
             errorat(lx->pos, "eof found in macro '%s'", s);
             goto Bad;
@@ -797,7 +804,7 @@ ppinc(Lexer *lx)
             lx->buf[0] = '\0';
         strcat(lx->buf, s);
 
-        if (io·exists(lx->buf, ReadOK)) {
+        if (os·exists(lx->buf, ReadOK)) {
             break;
         }
     }
@@ -833,7 +840,7 @@ ppprag(Lexer *lx)
         goto Bad;
     }
     if (strcmp(s, "once") == 0) {
-        lx->io->kind |= IOonce;
+        pushomit(lx->io->path);
         return 0;
     }
 Bad:
@@ -882,8 +889,10 @@ Skip:
         if (c != '#') {
             if (!isspace(c))
                 b = 0;
-            if (c == '\n')
+            if (c == '\n') {
+                lx->pos.line++;
                 b = 1;
+            }
             continue;
         }
         if (!b)