Hash utilities: Check for and report on junk files.

[u/mdw/catacomb] / cc-hash.c

diff --git a/cc-hash.c b/cc-hash.c
index 9e23f79..00dcb39 100644 (file)
--- a/cc-hash.c
+++ b/cc-hash.c
@@ -32,11 +32,17 @@
 #include "config.h"
 
 #include <ctype.h>
+#include <errno.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
+#include <sys/types.h>
+#include <dirent.h>
+
+#include <mLib/alloc.h>
 #include <mLib/dstr.h>
+#include <mLib/report.h>
 #include <mLib/str.h>
 
 #include <mLib/hex.h>
@@ -166,6 +172,18 @@ const encodeops *getencoding(const char *ename)
 
 /*----- File hashing ------------------------------------------------------*/
 
+enum {
+  FHETY_DIR,
+  FHETY_FILE
+};
+
+typedef struct fhent {
+  struct fhent *next;
+  unsigned ty;
+  struct fhent *sub;
+  char name[1];
+} fhdir;
+
 /* --- @gethash@ --- *
  *
  * Arguments:  @const char *name@ = pointer to name string
@@ -193,6 +211,27 @@ const gchash *gethash(const char *name)
   return (gg);
 }
 
+/* --- @describefile@ --- *
+ *
+ * Arguments:  @const struct stat *st@ = pointer to file state
+ *
+ * Returns:    A snappy one-word description of the file.
+ */
+
+const char *describefile(const struct stat *st)
+{
+  switch (st->st_mode & S_IFMT) {
+    case S_IFBLK: return ("block-special");
+    case S_IFCHR: return ("char-special");
+    case S_IFIFO: return ("fifo");
+    case S_IFREG: return ("file");
+    case S_IFLNK: return ("symlink");
+    case S_IFDIR: return ("directory");
+    case S_IFSOCK: return ("socket");
+    default: return ("unknown");
+  }
+}
+
 /* --- @fhash_init@ ---*
  *
  * Arguments:  @fhashstate *fh@ = pointer to fhash state to initialize
@@ -205,7 +244,7 @@ const gchash *gethash(const char *name)
  */
 
 void fhash_init(fhashstate *fh, const gchash *gch, unsigned f)
-  { fh->f = f; fh->gch = gch; }
+  { fh->f = f; fh->gch = gch; fh->ents = 0; }
 
 /* --- @fhash_free@ --- *
  *
@@ -216,7 +255,19 @@ void fhash_init(fhashstate *fh, const gchash *gch, unsigned f)
  * Use:                Frees an fhash state.
  */
 
-void fhash_free(fhashstate *fh) { return; }
+static void freefhents(struct fhent *fhe)
+{
+  struct fhent *ffhe;
+
+  for (; fhe; fhe = ffhe) {
+    ffhe = fhe->next;
+    freefhents(fhe->sub);
+    xfree(fhe);
+  }
+}
+
+void fhash_free(fhashstate *fh)
+  { freefhents(fh->ents); }
 
 /* --- @fhash@ --- *
  *
@@ -236,6 +287,9 @@ int fhash(fhashstate *fh, const char *file, void *buf)
   size_t sz;
   ghash *h;
   int rc = 0;
+  struct fhent *fhe, **ffhe;
+  const char *p, *q;
+  size_t n;
   fprogress ff;
 
   if (!file || strcmp(file, "-") == 0)
@@ -247,6 +301,35 @@ int fhash(fhashstate *fh, const char *file, void *buf)
     if (fprogress_init(&ff, file, fp)) return (-1);
   }
 
+  if (fh->f & FHF_JUNK) {
+    p = file;
+    if (strncmp(p, "./", 2) == 0) p += 2;
+    q = p;
+    ffhe = &fh->ents;
+    for (;;) {
+      if (*q == '/' || *q == 0) {
+       n = q - p;
+       for (; *ffhe; ffhe = &(*ffhe)->next) {
+         fhe = *ffhe;
+         if (strncmp(p, fhe->name, n) == 0 && fhe->name[n] == 0)
+           goto found;
+       }
+       fhe = xmalloc(offsetof(struct fhent, name) + n + 1);
+       fhe->next = 0;
+       fhe->ty = *q == '/' ? FHETY_DIR : FHETY_FILE;
+       fhe->sub = 0;
+       *ffhe = fhe;
+       memcpy(fhe->name, p, n); fhe->name[n] = 0;
+      found:
+       if (!*q) break;
+       while (*++q == '/');
+       p = q;
+       ffhe = &fhe->sub;
+      } else
+       q++;
+    }
+  }
+
   h = GH_INIT(fh->gch);
   while ((sz = fread(fbuf, 1, sizeof(fbuf), fp)) > 0) {
     GH_HASH(h, fbuf, sz);
@@ -260,6 +343,99 @@ int fhash(fhashstate *fh, const char *file, void *buf)
   return (rc);
 }
 
+/* --- @fhash_junk@ --- *
+ *
+ * Arguments:  @fhashstate *fh@ = pointer to fhash state
+ *             @void (*func)(const char *, const struct stat *, void *)@
+ *             @void *p@ = pointer to pass to function
+ *
+ * Returns:    Positive if any junk was found, negative on error, zero if
+ *             everything was fine.
+ *
+ * Use:                Reports junk files in any directories covered by the hash
+ *             state.
+ */
+
+struct fhjunk {
+  int (*func)(const char *, const struct stat *, void *);
+  void *p;
+  dstr *d;
+};
+
+static int fhjunk(struct fhjunk *fhj, struct fhent *ents)
+{
+  DIR *dp;
+  int rc = 0, rrc;
+  struct stat st;
+  struct dirent *d;
+  const char *dname;
+  size_t n = fhj->d->len;
+  struct fhent *fhe;
+
+  dname = n ? fhj->d->buf : ".";
+  if ((dp = opendir(dname)) == 0) {
+    moan("failed to open directory `%s': %s", dname, strerror(errno));
+    rc = -1;
+    goto subs;
+  }
+  if (n) {
+    dstr_putc(fhj->d, '/');
+    n++;
+  }
+  while (errno = 0, (d = readdir(dp)) != 0) {
+    if (strcmp(d->d_name, ".") == 0 || strcmp(d->d_name, "..") == 0)
+      continue;
+    for (fhe = ents; fhe; fhe = fhe->next) {
+      if (strcmp(d->d_name, fhe->name) == 0) goto found;
+    }
+    fhj->d->len = n;
+    dstr_puts(fhj->d, d->d_name);
+    if (!lstat(fhj->d->buf, &st)) {
+      if (!rc) rc = 1;
+      rrc = fhj->func(fhj->d->buf, &st, fhj->p);
+    } else {
+      rc = -1;
+      rrc = fhj->func(fhj->d->buf, 0, fhj->p);
+    }
+    if (rrc < 0) rc = -1;
+  found:;
+  }
+  closedir(dp);
+  if (errno) {
+    moan("failed to read directory `%s': %s", dname, strerror(errno));
+    rc = -1;
+  }
+
+subs:
+  for (fhe = ents; fhe; fhe = fhe->next) {
+    if (fhe->ty == FHETY_DIR) {
+      fhj->d->len = n;
+      dstr_puts(fhj->d, fhe->name);
+      rrc = fhjunk(fhj, fhe->sub);
+      if (rrc < 0) rc = -1;
+      else if (!rc) rc = rrc;
+    }
+  }
+
+  return (rc);
+}
+
+int fhash_junk(fhashstate *fh,
+              int (*func)(const char *, const struct stat *, void *),
+              void *p)
+{
+  dstr d = DSTR_INIT;
+  struct fhjunk fhj;
+  int rc;
+
+  fhj.func = func;
+  fhj.p = p;
+  fhj.d = &d;
+  rc = fhjunk(&fhj, fh->ents);
+  dstr_destroy(&d);
+  return (rc);
+}
+
 /* --- @hfparse@ --- *
  *
  * Arguments:  @hfpctx *hfp@ = pointer to the context structure