From 5baa63ab502c98f6f62368302d92b6b90f9fcd26 Mon Sep 17 00:00:00 2001
From: Nicholas <nbnoll@eml.cc>
Date: Mon, 15 Nov 2021 18:43:39 -0800
Subject: Feat: globbing

Introduced shell globbing
---
 src/cmd/rc/glob.c | 178 +++++++++++++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 163 insertions(+), 15 deletions(-)

(limited to 'src/cmd/rc/glob.c')

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);
+    }
+}
+
-- 
cgit v1.2.1