Add and interface a script to move the files generated by the WebExtensions new browser extension into the web input queue

src/index/indexer.cpp

@@ -125,6 +125,7 @@ bool ConfIndexer::index(bool resetbefore, ixType typestorun, int flags)
     }
 #ifndef DISABLE_WEB_INDEXER
     if (m_doweb && (typestorun & IxTWebQueue)) {
+        runWebFilesMoverScript(m_config);
         deleteZ(m_webindexer);
         m_webindexer = new BeagleQueueIndexer(m_config, &m_db, m_updater);
         if (!m_webindexer || !m_webindexer->index()) {

src/index/rclmonprc.cpp

@@ -464,15 +464,22 @@ bool startMonitor(RclConfig *conf, int opts)
     bool timedout;
     time_t lastauxtime = time(0);
     time_t lastixtime = lastauxtime;
+    time_t lastmovetime = 0;
     bool didsomething = false;
     list<string> modified;
     list<string> deleted;
 
-    ;
-    
-    // Set a relatively short timeout for better monitoring of exit requests
     while (true) {
+        time_t now = time(0);
+        if (now - lastmovetime > ixinterval) {
+            lastmovetime = now;
+            runWebFilesMoverScript(conf);
+        }
+
         {
+            // Wait for event or timeout.
+            // Set a relatively short timeout for better monitoring of
+            // exit requests.
             std::unique_lock<std::mutex> lock = rclEQ.wait(2, &timedout);
 
             // x11IsAlive() can't be called from ok() because both
@@ -525,9 +532,9 @@ bool startMonitor(RclConfig *conf, int opts)
             }
         }
 
+        now = time(0);
 	// Process. We don't do this every time but let the lists accumulate
         // a little, this saves processing. Start at once if list is big.
-        time_t now = time(0);
         if (expeditedIndexingRequested(conf) ||
 	    (now - lastixtime > ixinterval) || 
 	    (deleted.size() + modified.size() > 20)) {
@@ -553,8 +560,9 @@ bool startMonitor(RclConfig *conf, int opts)
         }
 
 	// Recreate the auxiliary dbs every hour at most.
-	if (didsomething && time(0) - lastauxtime > auxinterval) {
-	    lastauxtime = time(0);
+        now = time(0);
+	if (didsomething && now - lastauxtime > auxinterval) {
+	    lastauxtime = now;
 	    didsomething = false;
 	    if (!createAuxDbs(conf)) {
 		// We used to bail out on error here. Not anymore,

src/index/recollindex.cpp

@@ -337,6 +337,48 @@ static bool checktopdirs(RclConfig *config, vector<string>& nonexist)
     return true;
 }
 
+bool runWebFilesMoverScript(RclConfig *config)
+{
+    static string downloadsdir;
+    if (downloadsdir.empty()) {
+        if (!config->getConfParam("downloadsdir", downloadsdir)) {
+            downloadsdir = path_tildexpand("~/Downloads");
+        }
+    }
+    static string cmdpath;
+    if (cmdpath.empty()) {
+        cmdpath = config->findFilter("recoll-we-move-files.py");
+        if (cmdpath.empty()) {
+            LOGERR("runWFMoverScript: recoll-we-move-files.py not found\n");
+            return false;
+        }
+    }
+
+    /* Arrange to not actually run the script if the directory did not change */
+    static time_t dirmtime;
+    time_t ndirmtime = 0;
+    struct stat st;
+    if (::stat(downloadsdir.c_str(), &st) == 0) {
+        ndirmtime = st.st_mtime;
+    }
+    /* If stat fails, presumably Downloads does not exist or is not
+       accessible, dirmtime and mdirmtime stay at 0, and we never
+       execute the script, which is the right thing. */
+    if (dirmtime != ndirmtime) {
+        /* The script is going to change the directory, so updating
+           dirmtime before it runs means that we are going to execute
+           it one time too many (it will run without doing anything),
+           but we can't set the mtime to after the run in case files
+           are created during the run. */
+        dirmtime = ndirmtime;
+        ExecCmd cmd;
+        int status = cmd.doexec(cmdpath, {});
+        return status == 0;
+    }
+    return true;
+}
+
+
 static const char *thisprog;
 
 static const char usage [] =

src/index/recollindex.h

@@ -26,6 +26,16 @@ extern bool indexfiles(RclConfig *config, std::list<std::string> &filenames);
 extern bool purgefiles(RclConfig *config, std::list<std::string> &filenames);
 extern bool createAuxDbs(RclConfig *config);
 
+/** 
+ * Helper method for executing the recoll-we (new WebExtensions plugin) helper
+ * script. This moves files from the browser download directory (only
+ * place where the browser accepts to create them), to the web queue
+ * dir). This keeps the c++ code compatible with old and new addon.
+ * The script is executed before a batch pass, or from time to time in
+ *  the monitor, if web processing is enabled.
+ */
+extern bool runWebFilesMoverScript(RclConfig *);
+
 extern int stopindexing;
 
 // Try to explain what went wrong...