Implement shell globbing, tilde expansion, and brace expansion (issues #13, #18)

Unquoted tokens are now expanded through a tilde → brace → glob pipeline
in parse_argv. Quoted tokens (single, double, or backslash) suppress all
expansion. Uses glob(3) with GLOB_NOCHECK for wildcard matching and manual
implementations for tilde (~→$HOME) and brace ({a,b,c}) expansion.

lcmd.c

@@ -10,6 +10,7 @@
 #include "lprefix.h"
 
 #include <errno.h>
+#include <glob.h>
 #include <signal.h>
 #include <stdlib.h>
 #include <string.h>
@@ -25,16 +26,303 @@
 
 /* ===== argv parser ===== */
 
+#define PA_QUOTED 1  /* suppress glob/tilde expansion for this token */
+
 typedef struct {
   char **argv;
   char *buf;
   int argc;
+  int *flags;  /* per-token flags (PA_QUOTED etc.) */
 } ParsedArgs;
 
 
 static void free_argv (ParsedArgs *pa) {
   free(pa->buf);
   free(pa->argv);
+  free(pa->flags);
+}
+
+
+/* ===== glob/tilde/brace expansion ===== */
+
+/*
+** Expand ~ at the start of a token to the home directory.
+** Returns a malloc'd string (caller must free), or NULL on failure.
+** If no tilde prefix, returns a strdup of the original.
+*/
+static char *expand_tilde (const char *token) {
+  const char *home;
+  size_t homelen, restlen;
+  char *result;
+  if (token[0] != '~')
+    return strdup(token);
+  /* only handle plain ~ and ~/path (not ~user for now) */
+  if (token[1] != '\0' && token[1] != '/')
+    return strdup(token);
+  home = getenv("HOME");
+  if (home == NULL)
+    return strdup(token);
+  homelen = strlen(home);
+  restlen = strlen(token + 1);  /* everything after ~ */
+  result = (char *)malloc(homelen + restlen + 1);
+  if (result == NULL) return NULL;
+  memcpy(result, home, homelen);
+  memcpy(result + homelen, token + 1, restlen + 1);
+  return result;
+}
+
+
+/*
+** Brace expansion: expand {a,b,c} patterns into multiple strings.
+** Handles prefix{a,b,c}suffix → prefixasuffix prefixbsuffix prefixcsuffix.
+** Does NOT handle nested braces (kept literal).
+** Writes results into out[] (caller provides), returns count.
+** out[] entries are malloc'd strings. max_out limits array size.
+** Returns 0 if no braces found (token unchanged).
+*/
+#define MAX_BRACE_RESULTS 256
+
+static int expand_braces (const char *token, char **out, int max_out) {
+  const char *open, *close, *p;
+  size_t prefix_len, suffix_len;
+  int depth, ncommas, count;
+  const char *alts[MAX_BRACE_RESULTS];
+  size_t altlens[MAX_BRACE_RESULTS];
+  int nalts = 0;
+
+  /* find first '{' */
+  open = strchr(token, '{');
+  if (open == NULL) return 0;
+
+  /* find matching '}' at depth 0, counting commas */
+  depth = 1; ncommas = 0;
+  for (p = open + 1; *p != '\0' && depth > 0; p++) {
+    if (*p == '{') depth++;
+    else if (*p == '}') depth--;
+    else if (*p == ',' && depth == 1) ncommas++;
+  }
+  if (depth != 0 || ncommas == 0)
+    return 0;  /* no valid brace expression */
+
+  close = p - 1;  /* points to '}' */
+  prefix_len = (size_t)(open - token);
+  suffix_len = strlen(close + 1);
+
+  /* split alternatives on ',' at depth 0 */
+  p = open + 1;
+  alts[0] = p;
+  for (; p < close; p++) {
+    if (*p == '{') depth++;
+    else if (*p == '}') depth--;
+    else if (*p == ',' && depth == 0) {
+      altlens[nalts] = (size_t)(p - alts[nalts]);
+      nalts++;
+      if (nalts >= MAX_BRACE_RESULTS) return 0;
+      alts[nalts] = p + 1;
+    }
+  }
+  altlens[nalts] = (size_t)(close - alts[nalts]);
+  nalts++;
+
+  /* generate prefix + alt + suffix for each alternative */
+  count = 0;
+  for (int i = 0; i < nalts && count < max_out; i++) {
+    size_t total = prefix_len + altlens[i] + suffix_len + 1;
+    char *s = (char *)malloc(total);
+    if (s == NULL) {
+      for (int j = 0; j < count; j++) free(out[j]);
+      return -1;
+    }
+    memcpy(s, token, prefix_len);
+    memcpy(s + prefix_len, alts[i], altlens[i]);
+    memcpy(s + prefix_len + altlens[i], close + 1, suffix_len + 1);
+    out[count++] = s;
+  }
+  return count;
+}
+
+
+/*
+** Growing list of tokens used during expansion.
+** Each item is a separately malloc'd string.
+*/
+typedef struct {
+  char **argv;
+  int *flags;
+  int argc;
+  int capacity;
+} TokenList;
+
+
+static int tklist_init (TokenList *tl, int capacity) {
+  tl->argv = (char **)malloc((size_t)capacity * sizeof(char *));
+  tl->flags = (int *)calloc((size_t)capacity, sizeof(int));
+  tl->argc = 0;
+  tl->capacity = capacity;
+  if (tl->argv == NULL || tl->flags == NULL) {
+    free(tl->argv); free(tl->flags);
+    return -1;
+  }
+  return 0;
+}
+
+
+static void tklist_free (TokenList *tl) {
+  int i;
+  for (i = 0; i < tl->argc; i++)
+    free(tl->argv[i]);
+  free(tl->argv);
+  free(tl->flags);
+}
+
+
+/*
+** Add a token to the list. Takes ownership of 's' on success.
+** Returns 0 on success, -1 on allocation failure (caller still owns 's').
+*/
+static int tklist_add (TokenList *tl, char *s, int flag) {
+  if (tl->argc >= tl->capacity) {
+    int newcap = tl->capacity * 2;
+    char **a = (char **)realloc(tl->argv, (size_t)newcap * sizeof(char *));
+    int *f = (int *)realloc(tl->flags, (size_t)newcap * sizeof(int));
+    if (a == NULL || f == NULL) {
+      /* if one succeeded, keep the original pointer valid */
+      if (a) tl->argv = a;
+      if (f) tl->flags = f;
+      return -1;
+    }
+    tl->argv = a;
+    tl->flags = f;
+    tl->capacity = newcap;
+  }
+  tl->argv[tl->argc] = s;
+  tl->flags[tl->argc] = flag;
+  tl->argc++;
+  return 0;
+}
+
+
+/*
+** Glob a single token and append results to 'tl'.
+** If the token has no metacharacters, it is added as-is (takes ownership).
+** Otherwise, glob results are added and 'token' is freed.
+** Returns 0 on success, -1 on failure (token is freed either way).
+*/
+static int glob_token (TokenList *tl, char *token) {
+  const char *s;
+  int has_meta = 0;
+  for (s = token; *s != '\0'; s++) {
+    if (*s == '*' || *s == '?' || *s == '[') { has_meta = 1; break; }
+  }
+  if (!has_meta)
+    return tklist_add(tl, token, 0);  /* no glob needed */
+  {
+    glob_t g;
+    int ret = glob(token, GLOB_NOCHECK, NULL, &g);
+    free(token);
+    if (ret != 0)
+      return 0;  /* GLOB_NOCHECK means this shouldn't happen, but be safe */
+    {
+      size_t k;
+      for (k = 0; k < g.gl_pathc; k++) {
+        char *dup = strdup(g.gl_pathv[k]);
+        if (dup == NULL || tklist_add(tl, dup, 0) != 0) {
+          free(dup);
+          globfree(&g);
+          return -1;
+        }
+      }
+    }
+    globfree(&g);
+    return 0;
+  }
+}
+
+
+/*
+** Expand unquoted tokens: tilde → brace → glob.
+** Quoted tokens (PA_QUOTED) are passed through unchanged.
+** Replaces pa->argv, pa->buf, and pa->flags in place.
+** Returns 0 on success, -1 on allocation failure.
+*/
+static int expand_argv (ParsedArgs *pa) {
+  int i;
+  TokenList tl;
+  size_t total_buflen = 0;
+  char *new_buf;
+  size_t bp = 0;
+
+  if (tklist_init(&tl, pa->argc + 16) != 0)
+    return -1;
+
+  for (i = 0; i < pa->argc; i++) {
+    if (pa->flags[i] & PA_QUOTED) {
+      char *dup = strdup(pa->argv[i]);
+      if (dup == NULL || tklist_add(&tl, dup, PA_QUOTED) != 0) {
+        free(dup); goto fail;
+      }
+    } else {
+      char *tilded;
+      char *brace_results[MAX_BRACE_RESULTS];
+      int nbrace;
+
+      /* step 1: tilde expansion */
+      tilded = expand_tilde(pa->argv[i]);
+      if (tilded == NULL) goto fail;
+
+      /* step 2: brace expansion */
+      nbrace = expand_braces(tilded, brace_results, MAX_BRACE_RESULTS);
+      if (nbrace < 0) { free(tilded); goto fail; }
+
+      if (nbrace == 0) {
+        /* no braces — glob the tilde-expanded token */
+        if (glob_token(&tl, tilded) != 0) goto fail;
+      } else {
+        int j;
+        free(tilded);
+        /* glob each brace alternative */
+        for (j = 0; j < nbrace; j++) {
+          if (glob_token(&tl, brace_results[j]) != 0) {
+            /* free remaining brace results */
+            for (j++; j < nbrace; j++) free(brace_results[j]);
+            goto fail;
+          }
+        }
+      }
+    }
+  }
+
+  /* pack all tokens into a single contiguous buffer */
+  for (i = 0; i < tl.argc; i++)
+    total_buflen += strlen(tl.argv[i]) + 1;
+
+  new_buf = (char *)malloc(total_buflen > 0 ? total_buflen : 1);
+  if (new_buf == NULL) goto fail;
+
+  for (i = 0; i < tl.argc; i++) {
+    char *old = tl.argv[i];
+    size_t slen = strlen(old);
+    memcpy(new_buf + bp, old, slen + 1);
+    tl.argv[i] = new_buf + bp;
+    bp += slen + 1;
+    free(old);
+  }
+
+  /* replace old arrays */
+  free(pa->argv);
+  free(pa->buf);
+  free(pa->flags);
+  pa->argv = (char **)realloc(tl.argv, ((size_t)tl.argc + 1) * sizeof(char *));
+  if (pa->argv == NULL) pa->argv = tl.argv;
+  pa->argv[tl.argc] = NULL;
+  pa->buf = new_buf;
+  pa->flags = tl.flags;
+  pa->argc = tl.argc;
+  return 0;
+
+fail:
+  tklist_free(&tl);
+  return -1;
 }
 
 
@@ -49,17 +337,21 @@ static int parse_argv (const char *cmd, ParsedArgs *result) {
   char *buf = (char *)malloc(len + 1);
   int capacity = 8;
   char **argv = (char **)malloc((size_t)capacity * sizeof(char *));
+  int *flags = (int *)calloc((size_t)capacity, sizeof(int));
   int argc = 0;
   size_t bp = 0;  /* position in buf */
   const char *p = cmd;
 
-  if (buf == NULL || argv == NULL) {
+  if (buf == NULL || argv == NULL || flags == NULL) {
     free(buf);
     free(argv);
+    free(flags);
     return -1;
   }
 
   while (*p != '\0') {
+    int quoted = 0;
+
     /* skip whitespace */
     while (*p == ' ' || *p == '\t' || *p == '\n' || *p == '\r')
       p++;
@@ -70,9 +362,13 @@ static int parse_argv (const char *cmd, ParsedArgs *result) {
     if (argc >= capacity - 1) {
       capacity *= 2;
       argv = (char **)realloc(argv, (size_t)capacity * sizeof(char *));
-      if (argv == NULL) { free(buf); return -1; }
+      flags = (int *)realloc(flags, (size_t)capacity * sizeof(int));
+      if (argv == NULL || flags == NULL) {
+        free(buf); free(argv); free(flags);
+        return -1;
+      }
     }
-    argv[argc++] = buf + bp;
+    argv[argc] = buf + bp;
 
     while (*p != '\0') {
       if (*p == ' ' || *p == '\t' || *p == '\n' || *p == '\r')
@@ -80,14 +376,16 @@ static int parse_argv (const char *cmd, ParsedArgs *result) {
 
       if (*p == '\'') {
         /* single quote: literal until closing quote */
+        quoted = 1;
         p++;
         while (*p != '\0' && *p != '\'')
           buf[bp++] = *p++;
         if (*p == '\'') p++;
-        else { free(buf); free(argv); return -1; }
+        else { free(buf); free(argv); free(flags); return -1; }
       }
       else if (*p == '"') {
         /* double quote: with escape sequences */
+        quoted = 1;
         p++;
         while (*p != '\0' && *p != '"') {
           if (*p == '\\' && p[1] != '\0') {
@@ -107,10 +405,11 @@ static int parse_argv (const char *cmd, ParsedArgs *result) {
           }
         }
         if (*p == '"') p++;
-        else { free(buf); free(argv); return -1; }
+        else { free(buf); free(argv); free(flags); return -1; }
       }
       else if (*p == '\\' && p[1] != '\0') {
         /* backslash escape outside quotes */
+        quoted = 1;
         p++;
         buf[bp++] = *p++;
       }
@@ -120,12 +419,23 @@ static int parse_argv (const char *cmd, ParsedArgs *result) {
     }
 
     buf[bp++] = '\0';
+    flags[argc] = quoted ? PA_QUOTED : 0;
+    argc++;
   }
 
   argv[argc] = NULL;
   result->argv = argv;
   result->buf = buf;
   result->argc = argc;
+  result->flags = flags;
+
+  /* expand globs and tildes on unquoted tokens */
+  if (expand_argv(result) != 0) {
+    free(buf);
+    free(argv);
+    free(flags);
+    return -1;
+  }
   return 0;
 }
 

testes/lush/globbing.lua

@@ -0,0 +1,135 @@
+-- testes/lush/globbing.lua
+-- Tests for shell globbing (#13) and tilde expansion (#18).
+
+print "testing globbing and tilde expansion"
+
+-- === wildcard expansion ===
+
+-- *.lua should expand to matching files (we know lush test files exist)
+do
+  local r = `echo *.c`
+  local out = r.stdout:gsub("\n$", "")
+  -- should NOT be the literal "*.c" since .c files exist in the project
+  assert(out ~= "*.c", "expected *.c to expand, got literal: " .. out)
+  assert(out:find("lcmd.c"), "expected lcmd.c in expansion, got: " .. out)
+end
+
+-- no match: keeps literal pattern (GLOB_NOCHECK)
+do
+  local r = `echo *.nonexistent_xyz_9876`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == "*.nonexistent_xyz_9876",
+         "expected literal, got: " .. out)
+end
+
+-- ? single-char wildcard
+do
+  -- create temp files to test ? pattern
+  os.execute("mkdir -p /tmp/_lush_glob_test")
+  os.execute("touch /tmp/_lush_glob_test/a1 /tmp/_lush_glob_test/b2")
+  local r = `echo /tmp/_lush_glob_test/?1`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == "/tmp/_lush_glob_test/a1",
+         "expected ?1 match, got: " .. out)
+  os.execute("rm -rf /tmp/_lush_glob_test")
+end
+
+
+-- === quoting suppresses expansion ===
+
+-- double-quoted glob stays literal
+do
+  local r = `echo "*.c"`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == "*.c", "double-quoted glob expanded: " .. out)
+end
+
+-- single-quoted glob stays literal
+do
+  local r = `echo '*.c'`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == "*.c", "single-quoted glob expanded: " .. out)
+end
+
+-- backslash-escaped glob stays literal
+do
+  local r = `echo \*.c`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == "*.c", "backslash-escaped glob expanded: " .. out)
+end
+
+
+-- === brace expansion ===
+
+do
+  local r = `echo {a,b,c}`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == "a b c", "brace expansion failed, got: " .. out)
+end
+
+-- brace with prefix
+do
+  local r = `echo hello{world,there}`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == "helloworld hellothere",
+         "brace prefix expansion failed, got: " .. out)
+end
+
+-- space inside braces: no expansion (treated as separate tokens)
+do
+  local r = `echo {a, b}`
+  local out = r.stdout:gsub("\n$", "")
+  -- with space after comma, the shell splits into tokens "{a," and "b}"
+  assert(out == "{a, b}", "space-in-braces should not expand, got: " .. out)
+end
+
+
+-- === tilde expansion ===
+
+-- ~ expands to $HOME
+do
+  local home = os.getenv("HOME")
+  local r = `echo ~`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == home, "~ did not expand to HOME, got: " .. out)
+end
+
+-- ~/path expands to $HOME/path
+do
+  local home = os.getenv("HOME")
+  local r = `echo ~/foo`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == home .. "/foo",
+         "~/foo did not expand, got: " .. out)
+end
+
+-- quoted ~ stays literal
+do
+  local r = `echo "~"`
+  local out = r.stdout:gsub("\n$", "")
+  assert(out == "~", "quoted ~ expanded: " .. out)
+end
+
+-- cd ~ should work (builtin receives expanded path)
+do
+  local before = `pwd`.stdout:gsub("\n$", "")
+  local home = os.getenv("HOME")
+  local r = `cd ~`
+  assert(r.code == 0, "cd ~ failed: " .. r.stderr)
+  local after = `pwd`.stdout:gsub("\n$", "")
+  assert(after == home, "cd ~ did not go home, got: " .. after)
+  `cd ${before}`
+end
+
+
+-- === expansion in pipelines ===
+
+do
+  local r = `echo *.c | head -1`
+  local out = r.stdout:gsub("\n$", "")
+  -- pipeline should have expanded *.c before piping
+  assert(out ~= "*.c", "glob not expanded in pipeline, got: " .. out)
+end
+
+
+print "OK"