Feat: globbing

Introduced shell globbing

3 files changed, 206 insertions, 26 deletions

diff --git a/src/cmd/rc/exec.c b/src/cmd/rc/exec.c
index a551d52..62375df 100644
--- a/src/cmd/rc/exec.c
+++ b/src/cmd/rc/exec.c
@@ -103,8 +103,7 @@ freelist(Word *w)
 
 }
 
-static
-void
+static void
 pushlist(void)
 {
     List *stack = emalloc(sizeof(*stack));
@@ -115,8 +114,7 @@ pushlist(void)
     runner->args = stack;
 }
 
-static
-void
+static void
 poplist(void)
 {
     List *stack = runner->args;
@@ -128,6 +126,24 @@ poplist(void)
     efree(stack);
 }
 
+static void
+globargs(void)
+{
+    Word *arg;
+
+    globmatch = nil;
+    globlist(runner->args->word);
+    poplist();
+    pushlist();
+    if(globmatch){
+        for(arg=globmatch; arg->link; arg = arg->link)
+            ;
+        arg->link= runner->args->word;
+        runner->args->word = globmatch;
+    }
+}
+
+
 /* system interop */
 static
 Word*
@@ -792,7 +808,7 @@ Xdollar(void)
         return;
     }
     s = runner->args->word->str;
-    // deglob(s);
+    deglob(s);
     n = 0;
 
     for(t = s;'0'<=*t && *t<='9';t++) 
@@ -846,7 +862,7 @@ getindex(Word *array, int len, Word *index, Word *tail)
     tail = getindex(array, len, index->link, tail);
 
     s = index->str;
-    //deglob(s)
+    deglob(s);
 
     m = 0, n = 0;
     while('0' <= *s && *s <= '9')
@@ -881,7 +897,7 @@ Xindex(void)
         return;
     }
     s = runner->args->word->str;
-    //deglob(s)
+    deglob(s);
     val = var(s)->val;
     poplist();
 
@@ -935,6 +951,12 @@ Xjoin(void)
 }
 
 void
+Xglob(void)
+{
+    globargs();
+}
+
+void
 Xassign(void)
 {
     Var *v;
@@ -943,11 +965,11 @@ Xassign(void)
         Xerror("variable name not singleton!\n");
         return;
     }
-    //deglob(runq->argv->words->word);
+    deglob(runner->args->word->str);
     v = var(runner->args->word->str);
     poplist();
 
-    //globlist();
+    globargs();
     freewords(v->val);
     v->val = runner->args->word;
     v->new = 1;
@@ -991,7 +1013,7 @@ Xlocal(void)
         Xerror("variable name must be singleton\n");
         return;
     }
-    //deglob(shell->args->word->str);
+    deglob(runner->args->word->str);
 
     runner->local      = makevar(strdup(runner->args->word->str), runner->local);
     runner->local->val = copywords(runner->args->link->word, nil);
@@ -1141,6 +1163,7 @@ Xbasic(void)
     int pid, status;
     struct Builtin *b;
 
+    globargs();
     arg = runner->args->word;
     if(!arg){
         Xerror("empty argument list\n");
@@ -1189,6 +1212,8 @@ Xcount(void)
     }
 
     str = runner->args->word->str;
+    deglob(str);
+
     arg = var(str)->val;
     poplist();
 
@@ -1209,6 +1234,8 @@ Xflat(void)
     }
 
     str = runner->args->word->str;
+    deglob(str);
+
     arg = var(str)->val;
     poplist();
 
diff --git a/src/cmd/rc/glob.c b/src/cmd/rc/glob.c
index 9d88b59..eee0318 100644
--- a/src/cmd/rc/glob.c
+++ b/src/cmd/rc/glob.c
@@ -1,25 +1,16 @@
 #include "rc.h"
+#include <dirent.h>
 
-/* removes glob marks in place */
-void
-deglob(char *s)
-{
-    char *t = s;
-    do{
-        if(*t==GLOB)
-            t++;
-        *s++=*t;
-    }while(*t++);
-}
+Word *globmatch = nil;
+static char *globbuf = nil;
 
 int
 match(char *s, char *p,  int stop)
 {
     rune rs, rp, lo, hi;
-    int neg, hit;
+    int np, ns, neg, hit;
 
-    utf8·decode(p, &rp), utf8·decode(s, &rs);
-    for(; *p && *p != stop; s+=utf8·decode(s,&rs),p+=utf8·decode(p,&rp)){
+    for(np=utf8·decode(p, &rp), ns=utf8·decode(s, &rs); *p && *p != stop; s+=ns, p+=np, ns=utf8·decode(s,&rs), np=utf8·decode(p,&rp)){
         /* fast path: normal character */
         if(*p != GLOB){
             if(rp != rs)
@@ -34,10 +25,12 @@ match(char *s, char *p,  int stop)
                 return 0;
             break;
         case '*':
+            p += utf8·decode(p, &rp);
             for(;;){
-                p += utf8·decode(p, &rp);
                 if(match(s, p, stop))
                     return 1;
+                if(!*s)
+                    break;
                 s += utf8·decode(s, &rs);
             }
             return 0;
@@ -80,3 +73,158 @@ match(char *s, char *p,  int stop)
 
     return *s == 0;
 }
+
+static int
+matchpath(char *s, char *p)
+{
+    if(s[0] == '.' && (s[1] == 0 || s[1] == '.' && s[2] == 0) && p[0] != '.')
+        return 0;
+    return match(s, p, '/');
+}
+
+/* removes glob marks in place */
+void
+deglob(char *s)
+{
+    char *t = s;
+    do{
+        if(*t==GLOB)
+            t++;
+        *s++=*t;
+    }while(*t++);
+}
+
+#define DIRLEN 256
+static int
+globsize(char *p)
+{
+    ulong is=0, len=DIRLEN+1;
+    for(;*p;p++){
+        if(*p!=GLOB)
+            len++;
+        else{
+            p++;
+            if(*p != GLOB)
+                is++;
+            len += (*p == '*') ? DIRLEN : 1;
+        }
+    }
+    return is ? len : 0;
+}
+#undef DIRLEN
+
+static void
+globpath(char *p, char *buf)
+{
+    int fd;
+    char *s, *base;
+    DIR *dir;
+    struct dirent *sub;
+
+    if(*p == 0){
+        globmatch = makeword(globbuf, globmatch);
+        return;
+    }
+
+    s=buf, base=p;
+    while(*base){
+        if(*base==GLOB)
+            break;
+        *s = *base++;
+        if(*s++ == '/'){
+            buf=s, p=base;
+        }
+    }
+
+    /* we ran out of pattern, no glob */
+    if(*base == 0){
+        *s = 0;
+        if(os·exists(globbuf, 0))
+            globmatch = makeword(globbuf, globmatch);
+        return;
+    }
+
+    *buf = 0;
+    /* read the directory and recurse for any entry that matches */
+    /* TODO: replace ugly C interface with something else */
+    fd = openat(AT_FDCWD, globbuf[0] ? globbuf : ".", O_RDONLY | O_CLOEXEC | O_DIRECTORY);
+    if(fd < 0)
+        return;
+    if(!(dir = fdopendir(fd))){
+        close(fd);
+        return;
+    }
+
+    while(*base != '/' && *base != 0)
+        base++;
+
+    while((sub=readdir(dir))){
+        strcpy(buf, sub->d_name);
+        if(matchpath(buf, p)){
+            for(s=buf; *s; s++)
+                ;
+            globpath(base, s);
+        }
+    }
+    closedir(dir);
+}
+
+static void
+sort(Word *left, Word *right)
+{
+    int n,f;
+    Word *arg;
+    char **arr, *buf[256];
+
+    for(arg=left,n=0; arg != right; arg=arg->link)
+        n++;
+
+    if(n<arrlen(buf))
+        arr=buf, f=0;
+    else
+        arr=emalloc(n*sizeof(*arr)), f=1;
+
+    for(arg=left,n=0; arg!=right; arg=arg->link,n++)
+        arr[n] = arg->str;
+
+    sort·string(n,arr);
+
+    for(arg=left,n=0; arg!=right; arg=arg->link, n++)
+        arg->str = arr[n];
+
+    if(f)
+        efree(arr);
+}
+
+static void
+glob(char *p)
+{
+    Word *old = globmatch;
+    int nglob = globsize(p);
+    if(!nglob)
+        goto nomatch;
+
+    globbuf = emalloc(nglob);
+    globbuf[0] = 0;
+    globpath(p, globbuf);
+    efree(globbuf);
+    if(globmatch == old)
+        goto nomatch;
+    /* fallthrough */
+match:
+    sort(globmatch, old);
+    return;
+nomatch:
+    deglob(p);
+    globmatch = makeword(p, globmatch);
+}
+
+void
+globlist(Word *g)
+{
+    if(g){
+        globlist(g->link);
+        glob(g->str);
+    }
+}
+
diff --git a/src/cmd/rc/rc.h b/src/cmd/rc/rc.h
index f7533a0..b51b78b 100644
--- a/src/cmd/rc/rc.h
+++ b/src/cmd/rc/rc.h
@@ -1,3 +1,4 @@
+#pragma once
 
 #include <u.h>
 #include <base.h>
@@ -265,4 +266,8 @@ void  freecode(Code *c);
 
 /* glob.c */
 #define GLOB ((char)(0x01))
-int match(char *, char *, int);
+extern Word *globmatch;
+
+void deglob(char *);
+void globlist(Word *);
+int  match(char *, char *, int);