X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~mdw/git/disorder/blobdiff_plain/fbcfb25715ddbb522c8efac3faec1eca4811683d..9f28e855aad11d25c0a69a7e59825c07bdefcafe:/server/trackdb.c

diff --git a/server/trackdb.c b/server/trackdb.c
index b4e0de9..bc526ca 100644
--- a/server/trackdb.c
+++ b/server/trackdb.c
@@ -17,6 +17,8 @@
  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
  * USA
  */
+/** @file server/trackdb.c
+ * @brief Track database */
 
 #include <config.h>
 #include "types.h"
@@ -34,6 +36,7 @@
 #include <sys/resource.h>
 #include <time.h>
 #include <arpa/inet.h>
+#include <sys/wait.h>
 
 #include "event.h"
 #include "mem.h"
@@ -144,9 +147,17 @@ static int compare(DB attribute((unused)) *db_,
   return compare_path_raw(a->data, a->size, b->data, b->size);
 }
 
-/* open environment */
-void trackdb_init(int recover) {
+/** @brief Open database environment
+ * @param flags Flags word
+ *
+ * Flags should be one of:
+ * - @ref TRACKDB_NO_RECOVER
+ * - @ref TRACKDB_NORMAL_RECOVER
+ * - @ref TRACKDB_FATAL_RECOVER
+ */
+void trackdb_init(int flags) {
   int err;
+  const int recover = flags & TRACKDB_RECOVER_MASK;
   static int recover_type[] = { 0, DB_RECOVER, DB_RECOVER_FATAL };
 
   /* sanity checks */
@@ -207,7 +218,8 @@ static pid_t subprogram(ev_source *ev, const char *prog,
   /* If we're in the background then trap subprocess stdout/stderr */
   if(!(pid = xfork())) {
     exitfn = _exit;
-    ev_signal_atfork(ev);
+    if(ev)
+      ev_signal_atfork(ev);
     signal(SIGPIPE, SIG_DFL);
     if(outputfd != -1) {
       xdup2(outputfd, 1);
@@ -285,9 +297,17 @@ static DB *open_db(const char *path,
   return db;
 }
 
-/* open track databases */
-void trackdb_open(void) {
+/** @brief Open track databases
+ * @param Flags flags word
+ *
+ * @p flags should be one of:
+ * - @p TRACKDB_NO_UPGRADE, if no upgrade should be attempted
+ * - @p TRACKDB_CAN_UPGRADE, if an upgrade may be attempted
+ * - @p TRACKDB_OPEN_FOR_UPGRADE, if this is disorder-dbupgrade
+ */
+void trackdb_open(int flags) {
   int newdb, err;
+  pid_t pid;
 
   /* sanity checks */
   assert(opened == 0);
@@ -300,19 +320,51 @@ void trackdb_open(void) {
     long oldversion;
 
     s = trackdb_get_global("_dbversion");
-    oldversion = s ? atol(s) : 1;
-    if(oldversion != config->dbversion) {
-      /* This database needs upgrading.  This isn't implemented yet so we just
-       * fail. */
-      fatal(0, "database needs upgrading from %ld to %ld",
-            oldversion, config->dbversion);
-    }
-    newdb = 0;
     /* Close the database again,  we'll open it property below */
     if((err = trackdb_globaldb->close(trackdb_globaldb, 0)))
       fatal(0, "error closing global.db: %s", db_strerror(err));
     trackdb_globaldb = 0;
+    /* Convert version string to an integer */
+    oldversion = s ? atol(s) : 1;
+    if(oldversion > config->dbversion) {
+      /* Database is from the future; we never allow this. */
+      fatal(0, "this version of DisOrder is too old for database version %ld",
+            oldversion);
+    }
+    if(oldversion < config->dbversion) {
+      /* Database version is out of date */
+      switch(flags & TRACKDB_UPGRADE_MASK) {
+      case TRACKDB_NO_UPGRADE:
+        /* This database needs upgrading but this is not permitted */
+        fatal(0, "database needs upgrading from %ld to %ld",
+              oldversion, config->dbversion);
+      case TRACKDB_CAN_UPGRADE:
+        /* This database needs upgrading */
+        info("invoking disorder-dbupgrade to upgrade from %ld to %ld",
+             oldversion, config->dbversion);
+        pid = subprogram(0, "disorder-dbupgrade", -1);
+        while(waitpid(pid, &err, 0) == -1 && errno == EINTR)
+          ;
+        if(err)
+          fatal(0, "disorder-dbupgrade %s", wstat(err));
+        info("disorder-dbupgrade succeeded");
+        break;
+      case TRACKDB_OPEN_FOR_UPGRADE:
+        break;
+      default:
+        abort();
+      }
+    }
+    if(oldversion == config->dbversion && (flags & TRACKDB_OPEN_FOR_UPGRADE)) {
+      /* This doesn't make any sense */
+      fatal(0, "database is already at current version");
+    }
+    newdb = 0;
   } else {
+    if(flags & TRACKDB_OPEN_FOR_UPGRADE) {
+      /* Cannot upgrade a new database */
+      fatal(0, "cannot upgrade a database that does not exist");
+    }
     /* This is a brand new database */
     newdb = 1;
   }
@@ -327,10 +379,11 @@ void trackdb_open(void) {
   trackdb_globaldb = open_db("global.db", 0, DB_HASH, DB_CREATE, 0666);
   trackdb_noticeddb = open_db("noticed.db",
                              DB_DUPSORT, DB_BTREE, DB_CREATE, 0666);
-  /* Stash the database version */
   if(newdb) {
+    /* Stash the database version */
     char buf[32];
 
+    assert(!(flags & TRACKDB_OPEN_FOR_UPGRADE));
     snprintf(buf, sizeof buf, "%ld", config->dbversion);
     trackdb_set_global("_dbversion", buf, 0);
   }
@@ -584,6 +637,22 @@ static int tailor_underscore_Word_Break_Other(uint32_t c) {
   }
 }
 
+/** @brief Remove all combining characters in-place
+ * @param s Pointer to start of string
+ * @param ns Length of string
+ * @return New, possiblby reduced, length
+ */
+static size_t remove_combining_chars(uint32_t *s, size_t ns) {
+  uint32_t *start = s, *t = s, *end = s + ns;
+
+  while(s < end) {
+    const uint32_t c = *s++;
+    if(!utf32_combining_class(c))
+      *t++ = c;
+  }
+  return t - start;
+}
+
 /** @brief Normalize and split a string using a given tailoring */
 static void word_split(struct vector *v,
                        const char *s,
@@ -597,6 +666,8 @@ static void word_split(struct vector *v,
   /* Erase case distinctions */
   if(!(t32 = utf32_casefold_compat(t32, nt32, &nt32)))
     return;
+  /* Drop combining characters */
+  nt32 = remove_combining_chars(t32, nt32);
   /* Split into words, treating _ as a space */
   w32 = utf32_word_split(t32, nt32, &nw, pt);
   /* Convert words back to UTF-8 and append to result */
@@ -1763,11 +1834,20 @@ char **trackdb_search(char **wordlist, int nwordlist, int *ntracks) {
   const char *dbname;
 
   *ntracks = 0;				/* for early returns */
-  /* casefold all the words */
+  /* normalize all the words */
   w = xmalloc(nwordlist * sizeof (char *));
   for(n = 0; n < nwordlist; ++n) {
+    uint32_t *w32;
+    size_t nw32;
+    
     w[n] = utf8_casefold_compat(wordlist[n], strlen(wordlist[n]), 0);
     if(checktag(w[n])) ++ntags;         /* count up tags */
+    /* Strip out combining characters (AFTER checking whether it's a tag) */
+    if(!(w32 = utf8_to_utf32(w[n], strlen(w[n]), &nw32)))
+      return 0;
+    nw32 = remove_combining_chars(w32, nw32);
+    if(!(w[n] = utf32_to_utf8(w32, nw32, 0)))
+      return 0;
   }
   /* find the longest non-stopword */
   for(n = 0; n < nwordlist; ++n)
@@ -1928,9 +2008,9 @@ static int reap_rescan(ev_source attribute((unused)) *ev,
                        void attribute((unused)) *u) {
   if(pid == rescan_pid) rescan_pid = -1;
   if(status)
-    error(0, "disorderd-rescan: %s", wstat(status));
+    error(0, RESCAN": %s", wstat(status));
   else
-    D(("disorderd-rescan terminate: %s", wstat(status)));
+    D((RESCAN" terminated: %s", wstat(status)));
   /* Our cache of file lookups is out of date now */
   cache_clean(&cache_files_type);
   eventlog("rescanned", (char *)0);
@@ -1938,14 +2018,22 @@ static int reap_rescan(ev_source attribute((unused)) *ev,
 }
 
 void trackdb_rescan(ev_source *ev) {
+  int w;
+
   if(rescan_pid != -1) {
     error(0, "rescan already underway");
     return;
   }
   rescan_pid = subprogram(ev, RESCAN, -1);
-  ev_child(ev, rescan_pid, 0, reap_rescan, 0);
-  D(("started rescanner"));
-  
+  if(ev) {
+    ev_child(ev, rescan_pid, 0, reap_rescan, 0);
+    D(("started rescanner"));
+  } else {
+    /* This is the first rescan, we block until it is complete */
+    while(waitpid(rescan_pid, &w, 0) < 0 && errno == EINTR)
+      ;
+    reap_rescan(0, rescan_pid, w, 0, 0);
+  }
 }
 
 int trackdb_rescan_cancel(void) {