add -E preprocessing option

6 files changed, 161 insertions, 18 deletions

diff --git a/c.c b/c.c
index 2d3b3cf..58a3147 100644
--- a/c.c
+++ b/c.c
@@ -4017,20 +4017,34 @@ docomp(struct comp *cm)
    }
 }
 
-void
-ccomp(const char *file)
+static void
+initcm(struct comp *cm, const char *file)
 {
    enum { N = 1<<12 };
    static union { char m[sizeof(struct arena) + N]; struct arena *_align; } amem[2];
-   struct comp cm = {0};
-   const char *err = initlexer(&cm.lx, NULL, file);
+   const char *err = initlexer(&cm->lx, NULL, file);
    if (err)
       fatal(NULL, "Cannot open %'s: %s", file, err);
-   cm.fnarena = (void *)amem[0].m;
-   cm.fnarena->cap = N;
-   cm.exarena = (void *)amem[1].m;
-   cm.exarena->cap = N;
+   cm->fnarena = (void *)amem[0].m;
+   cm->fnarena->cap = N;
+   cm->exarena = (void *)amem[1].m;
+   cm->exarena->cap = N;
+}
+
+void
+ccomp(const char *file)
+{
+   struct comp cm = {0};
+   initcm(&cm, file);
    docomp(&cm);
 }
 
+void
+cpp(struct wbuf *out, const char *file)
+{
+   struct comp cm = {0};
+   initcm(&cm, file);
+   lexerdump(&cm.lx, out);
+}
+
 /* vim:set ts=3 sw=3 expandtab: */
diff --git a/common.h b/common.h
index f2fff86..c5b7513 100644
--- a/common.h
+++ b/common.h
@@ -534,7 +534,7 @@ struct wbuf {
    char *buf;
    const uint cap;
    uint len;
-   const int fd;
+   int fd;
    bool err;
 };
 
@@ -578,6 +578,7 @@ void warn(const struct span *, const char *, ...);
 void note(const struct span *, const char *, ...);
 ushort *utf8to16(uint *ulen, struct arena **, const uchar *s, size_t len);
 uint *utf8to32(uint *ulen, struct arena **, const uchar *s, size_t len);
+int utf8enc(char out[4], uint cp);
 
 #endif /* COMMON_H_ */
 
diff --git a/io.c b/io.c
index f73add8..67b23b6 100644
--- a/io.c
+++ b/io.c
@@ -96,14 +96,14 @@ putquoted(struct wbuf *buf, uchar c, uchar qchar, int next)
       case '\v': cseq = 'v'; goto Charseq;
       case '\n': cseq = 'n'; goto Charseq;
       default:
-          if (in_range(next, '0', '7'))
+          if (!next || in_range(next, '0', '7'))
               n += bfmt(buf, "%.3o", c);
           else
               n += bfmt(buf, "%o", c);
       }
       return n;
    }
-   if (c == '?' && next == '?') {
+   if (c == '?' && (!next || next == '?')) {
       return ioputc(buf, c), ioputc(buf, '\\'), 2;
    }
    return ioputc(buf, c), 1;
@@ -473,13 +473,39 @@ vbfmt(struct wbuf *out, const char *fmt, va_list ap)
                if (quote) n += bputc(buf, '\'');
                break;
             case TKCHRLIT:
+               if (tok->wide) n += bputc(buf, tok->wideuni ? tok->wide == 1 ? 'u' : 'U' : 'L');
                n += bputc(buf, '\'');
-               for (int i = 0; i < tok->len; ++i)
-                  n += putquoted(buf, tok->s[i], '\'', i < tok->len - 1 ? tok->s[i+1] : -1);
+               if (tok->wide == 0)
+                  for (int i = 0; i < tok->len; ++i)
+                     n += putquoted(buf, tok->s[i], '\'', i < tok->len - 1 ? tok->s[i+1] : -1);
+               else {
+                  char p[4];
+                  uint c = tok->wide == 1 ? tok->ws16[0] : tok->ws32[0];
+                  int l = utf8enc(p, c);
+                  if (l == 1)
+                     n += putquoted(buf, *p, '\'', -1);
+                  else
+                     n += (iowrite(buf, p, l), l);
+               }
                n += bputc(buf, '\'');
                break;
             case TKSTRLIT:
-               n += bfmt(buf, "%'S", tok->s, tok->len);
+               if (tok->wide == 0)
+                  n += bfmt(buf, "%'S", tok->s, tok->len);
+               else {
+                  n += bputc(buf, tok->wideuni ? tok->wide == 1 ? 'u' : 'U' : 'L');
+                  n += bputc(buf, '\"');
+                  for (int i = 0; i < tok->len; ++i) {
+                     char p[4];
+                     uint c = tok->wide == 1 ? tok->ws16[i] : tok->ws32[i];
+                     int l = utf8enc(p, c);
+                     if (l == 1)
+                        n += putquoted(buf, *p, '\"', 0);
+                     else
+                        n += (iowrite(buf, p, l), l);
+                  }
+                  n += bputc(buf, '\"');
+               }
                break;
             case TKPPMACSTR:
                if (quote) n += bputc(buf, '`');
@@ -937,7 +963,7 @@ utf8to32(uint *ulen, struct arena **arena, const uchar *s, size_t len)
 
    if (!len) return NULL;
 
-   for (p = s; p < s + len; ++n) {
+   for (p = end = s; p < s + len; ++n) {
       end = p;
       if ((*p & 0xF8) == 0xF0) /* 11110xxx */
          p += 4;
@@ -978,5 +1004,28 @@ utf8to32(uint *ulen, struct arena **arena, const uchar *s, size_t len)
    return ret;
 }
 
+int
+utf8enc(char p[4], uint cp)
+{
+   if ((cp & 0xffffff80) == 0) {
+      p[0] = cp;
+      return 1;
+   } else if ((cp & 0xfffff800) == 0) {
+      p[0] = 0xC0 | (cp >> 6 & 0x1F);
+      p[1] = 0x80 | (cp & 0x3F);
+      return 2;
+   } else if ((cp & 0xffff0000) == 0) {
+      p[0] = 0xE0 | (cp >> 12 & 0x0F);
+      p[1] = 0x80 | (cp >>  6 & 0x3F);
+      p[2] = 0x80 | (cp & 0x3F);
+      return 3;
+   } else {
+      p[0] = 0xF0 | (cp >> 18 & 0x07);
+      p[1] = 0x80 | (cp >> 12 & 0x3F);
+      p[2] = 0x80 | (cp >>  6 & 0x3F);
+      p[3] = 0x80 | (cp & 0x3F);
+      return 4;
+   }
+}
 
 /* vim:set ts=3 sw=3 expandtab: */
diff --git a/lex.c b/lex.c
index e668500..605b3d2 100644
--- a/lex.c
+++ b/lex.c
@@ -524,6 +524,7 @@ Begin:
          lexingheadername = 0;
       } else {
    case '\'':
+         tk->wideuni = 0;
          readstrchrlit(lx, tk, c, 0);
       }
       goto End;
@@ -537,6 +538,7 @@ Begin:
       RET(c);
    case 'L':
       if (match(lx, (q = '\'')) || match(lx, (q = '"'))) {
+         tk->wideuni = 0;
          readstrchrlit(lx, tk, q, /* wide */ targ_primsizes[targ_wchartype] == 2 ? 1 : 2);
          goto End;
       }
@@ -629,7 +631,7 @@ tokequ(const struct token *a, const struct token *b)
 static bool /* whitespace separating tokens? */
 wsseparated(const struct token *l, const struct token *r)
 {
-   assert(l->span.sl.file == r->span.sl.file);
+   if (l->span.sl.file != r->span.sl.file) return 1;
    return l->span.sl.off + l->span.sl.len != r->span.sl.off;
 }
 
@@ -1893,4 +1895,42 @@ lexerfreetemps(struct lexer *lx)
    }
 }
 
+void
+lexerdump(struct lexer *lx, struct wbuf *out)
+{
+   struct token prev = {0}, tok;
+   int file = lx->fileid, line = 1, col = 1;
+   bfmt(out, "# %d %'s\n", 1, getfilename(file));
+   while (lex(lx, &tok) != TKEOF) {
+      int tkline, tkcol;
+      getfilepos(&tkline, &tkcol, tok.span.ex.file, tok.span.ex.off);
+      if (tok.span.ex.file != file) {
+         file = tok.span.ex.file;
+         bfmt(out, "\n# %d %'s\n", tkline, getfilename(file));
+         col = 1;
+      } else if (line < tkline && tkline - line < 5) {
+         do
+            ioputc(out, '\n');
+         while (++line != tkline);
+         col = 1;
+      } else if (line != tkline) {
+         bfmt(out, "\n# %d\n", tkline);
+         line = tkline;
+         col = 1;
+      } else if (prev.t && wsseparated(&prev, &tok)) {
+         ioputc(out, ' ');
+         ++col;
+      }
+      if (col == 1)
+         for (; col < tkcol; ++col)
+            ioputc(out, ' ');
+      line = tkline;
+      bfmt(out, "%tk", &tok);
+      col += tok.span.ex.len;
+      prev = tok;
+   }
+   bfmt(out, "\n");
+   ioflush(out);
+}
+
 /* vim:set ts=3 sw=3 expandtab: */
diff --git a/lex.h b/lex.h
index 80898ce..6adbad9 100644
--- a/lex.h
+++ b/lex.h
@@ -52,6 +52,7 @@ struct token {
    short t; /* toktag */
    bool litlit;
    uchar wide : 2; /* for CHRLIT & STRLIT; 1 -> 16bit, 2 -> 32bit */
+   uchar wideuni : 1; /* ditto, 0 -> 'L', 1 -> 'u'/'U' (C11) */
    union {
       uint len;
       struct { ushort macidx, argidx; };
@@ -102,6 +103,7 @@ int lex(struct lexer *, struct token *);
 int lexpeek(struct lexer *, struct token *);
 enum typetag parsenumlit(uvlong *, double *, const struct token *, bool ispp);
 const char *initlexer(struct lexer *, const struct span *span, const char *file);
+void lexerdump(struct lexer *, struct wbuf *out);
 void lexerfreetemps(struct lexer *);
 
 /* vim:set ts=3 sw=3 expandtab: */
diff --git a/main.c b/main.c
index 1f75a50..3d7c846 100644
--- a/main.c
+++ b/main.c
@@ -84,7 +84,7 @@ withext(const char *path, const char *ext)
 }
 
 static struct task {
-   enum outft { OFTexe, OFTdll, OFTobj, OFTasm } outft;
+   enum outft { OFTexe, OFTdll, OFTobj, OFTasm, OFTc } outft;
    const char *out;
    const char *targ;
    const char *inf[64];
@@ -153,6 +153,8 @@ optparse(char **args)
          task.verbose = 1;
       } else if (!strcmp(arg, "c")) {
          task.outft = OFTobj;
+      } else if (!strcmp(arg, "E")) {
+         task.outft = OFTc;
       } else Bad: warn(NULL, "invalid option: %'s", arg-1);
    }
 
@@ -164,9 +166,10 @@ optparse(char **args)
       case OFTexe: task.out = "a.out"; break;
       case OFTasm: task.out = withext(*task.inf, "s"); break;
       case OFTobj: task.out = withext(*task.inf, "o"); break;
+      case OFTc: break;
       }
    }
-   if (!in_range(task.outft, OFTexe, OFTdll) && task.ninf > 1)
+   if (!in_range(task.outft, OFTexe, OFTdll) && task.outft != OFTc && task.ninf > 1)
       fatal(NULL, "too many input files");
 }
 
@@ -309,15 +312,49 @@ prihelp(void)
    );
 }
 
+#include <fcntl.h> /* open */
+
 static int
 driver(void)
 {
+   void cpp(struct wbuf *, const char *);
    if (task.verbose)
       efmt("# Target: %s\n", task.targ);
    if (task.outft == OFTobj) {
       assert(task.ninf == 1);
       assert(*task.inft == IFTc && "nyi");
       return cc1(task.out, *task.inf);
+   } else if (task.outft == OFTc) {
+      static char tmp[1<<12];
+      struct wbuf _buf = FDBUF(tmp, sizeof tmp, 1),
+                  *buf = &_buf;
+      bool ok = 1;
+      if (task.out) {
+         buf->fd = open(task.out, O_CREAT | O_TRUNC | O_WRONLY, 0777);
+         if (buf->fd < 0) {
+            error(NULL, "open(%'s): %s", task.out, strerror(errno));
+            return 1;
+         }
+      }
+      for (int i = 0; i < task.ninf; ++i) {
+         pid_t p;
+         int wstat;
+
+         if ((p = fork()) < 0) {
+            error(NULL, "fork(): %s\n", strerror(errno));
+            ok = 0;
+         } else if (p == 0) {
+            cpp(buf, task.inf[i]);
+            exit(0);
+         }
+         waitpid(p, &wstat, 0);
+         if (!WIFEXITED(wstat)) ok = 0;
+         ok = ok && WEXITSTATUS(wstat) == 0;
+      }
+      if (task.out) {
+         close(buf->fd);
+      }
+      return ok ? 0 : 1;
    } else if (task.outft == OFTexe || task.outft == OFTdll) {
       compileobjs();
       if (ccopt.dbg.any) return 0;