Changeset 2401 for trunk/gsdl/src

Timestamp:

2001-05-11T15:58:47+12:00 (23 years ago)

Author:

sjboddie

Message:

Made some changes to the collector so that paths/URLs/ftp addresses are
checked (pinged for HTTP/FTP) when leaving the "source data" page.
Functionality is complete I think but the interface still needs a
little work.

Location:

trunk/gsdl/src/recpt

Files:

• collectoraction.cpp (modified) (18 diffs)
• collectoraction.h (modified) (3 diffs)

trunk/gsdl/src/recpt/collectoraction.cpp

@@ -52 +52 @@
   disabled = true;
   do_mkcol = false;
+  badsources = false;
+  failedsources.erase(failedsources.begin(), failedsources.end());
   gsdlosc = NULL;
   gsdlhomec = NULL;
@@ -140 +142 @@
   argsinfo.addarginfo (NULL, arg_ainfo);
 
-  arg_ainfo.shortname = "bc1inputdir1";
-  arg_ainfo.longname = "collector specific";
-  arg_ainfo.multiplechar = true;
-  arg_ainfo.defaultstatus = cgiarginfo::weak;
-  arg_ainfo.argdefault = "";
-  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
-  argsinfo.addarginfo (NULL, arg_ainfo);
-
-  arg_ainfo.shortname = "bc1inputdir2";
-  arg_ainfo.longname = "collector specific";
-  arg_ainfo.multiplechar = true;
-  arg_ainfo.defaultstatus = cgiarginfo::weak;
-  arg_ainfo.argdefault = "";
-  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
-  argsinfo.addarginfo (NULL, arg_ainfo);
-
-  arg_ainfo.shortname = "bc1inputdir3";
-  arg_ainfo.longname = "collector specific";
-  arg_ainfo.multiplechar = true;
-  arg_ainfo.defaultstatus = cgiarginfo::weak;
-  arg_ainfo.argdefault = "";
-  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
-  argsinfo.addarginfo (NULL, arg_ainfo);
-
-  arg_ainfo.shortname = "bc1inputdir4";
-  arg_ainfo.longname = "collector specific";
-  arg_ainfo.multiplechar = true;
-  arg_ainfo.defaultstatus = cgiarginfo::weak;
-  arg_ainfo.argdefault = "";
-  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
-  argsinfo.addarginfo (NULL, arg_ainfo);
-
   // only set when one of the fields was changed in
   // the "collection info" page
@@ -226 +196 @@
   arg_ainfo.argdefault = "0";
   arg_ainfo.savedarginfo = cgiarginfo::must;
+  argsinfo.addarginfo (NULL, arg_ainfo);
+
+  arg_ainfo.shortname = "bc1filenum";
+  arg_ainfo.longname = "collector specific";
+  arg_ainfo.multiplechar = true;
+  arg_ainfo.defaultstatus = cgiarginfo::weak;
+  arg_ainfo.argdefault = "1";
+  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
+  argsinfo.addarginfo (NULL, arg_ainfo);
+
+  arg_ainfo.shortname = "bc1httpnum";
+  arg_ainfo.longname = "collector specific";
+  arg_ainfo.multiplechar = true;
+  arg_ainfo.defaultstatus = cgiarginfo::weak;
+  arg_ainfo.argdefault = "1";
+  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
+  argsinfo.addarginfo (NULL, arg_ainfo);
+
+  arg_ainfo.shortname = "bc1ftpnum";
+  arg_ainfo.longname = "collector specific";
+  arg_ainfo.multiplechar = true;
+  arg_ainfo.defaultstatus = cgiarginfo::weak;
+  arg_ainfo.argdefault = "1";
+  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
+  argsinfo.addarginfo (NULL, arg_ainfo);
+
+  arg_ainfo.shortname = "bc1fileinput";
+  arg_ainfo.longname = "collector specific";
+  arg_ainfo.multiplechar = true;
+  arg_ainfo.multiplevalue = true;
+  arg_ainfo.defaultstatus = cgiarginfo::weak;
+  arg_ainfo.argdefault = "";
+  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
+  argsinfo.addarginfo (NULL, arg_ainfo);
+
+  arg_ainfo.shortname = "bc1httpinput";
+  arg_ainfo.longname = "collector specific";
+  arg_ainfo.multiplechar = true;
+  arg_ainfo.multiplevalue = true;
+  arg_ainfo.defaultstatus = cgiarginfo::weak;
+  arg_ainfo.argdefault = "";
+  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
+  argsinfo.addarginfo (NULL, arg_ainfo);
+
+  arg_ainfo.shortname = "bc1ftpinput";
+  arg_ainfo.longname = "collector specific";
+  arg_ainfo.multiplechar = true;
+  arg_ainfo.multiplevalue = true;
+  arg_ainfo.defaultstatus = cgiarginfo::weak;
+  arg_ainfo.argdefault = "";
+  arg_ainfo.savedarginfo = cgiarginfo::mustnot; // saved on disk
+  argsinfo.addarginfo (NULL, arg_ainfo);
+
+  // will be set when we've just come from the "source data" page
+  arg_ainfo.shortname = "bc1fromsrce";
+  arg_ainfo.longname = "collector specific";
+  arg_ainfo.multiplechar = false;
+  arg_ainfo.defaultstatus = cgiarginfo::weak;
+  arg_ainfo.argdefault = "0";
+  arg_ainfo.savedarginfo = cgiarginfo::mustnot;
   argsinfo.addarginfo (NULL, arg_ainfo);
 }
@@ -323 +353 @@
     enum source_t {default_arg, compressed_arg, cgi_arg, program_arg};
     text_tmap saved_args;
-    saved_args["bc1inputdir1"] = "";
-    saved_args["bc1inputdir2"] = "";
-    saved_args["bc1inputdir3"] = "";
-    saved_args["bc1inputdir4"] = "";
     saved_args["bc1fullname"] = "";
     saved_args["bc1contactemail"] = "";
     saved_args["bc1aboutdesc"] = "";
     saved_args["bc1clone"] = "";
+    saved_args["bc1clonecol"] = "";
+    saved_args["bc1filenum"] = "";
+    saved_args["bc1httpnum"] = "";
+    saved_args["bc1ftpnum"] = "";
+    saved_args["bc1fileinput"] = "";
+    saved_args["bc1httpinput"] = "";
+    saved_args["bc1ftpinput"] = "";
 
     // update the argdb database with any arguments that were set
@@ -465 +498 @@
       }
       delete fbldc;
+    }
+  }
+
+  if (args["bc1fromsrce"] == "1") {
+
+    // we've just come from the "source data" page so we need to check that
+    // input sources are valid
+    if (!check_sources(args, logout)) {
+      args["p"] = "srce";
     }
   }
@@ -701 +743 @@
 }
 
-// set the _fullnamemenu_ macro (and _warnindex_ if we're on the "srce" page)
+// set the _fullnamemenu_ macro (and _warnindex_ and _selectedindex_ if
+// we're on the "srce" page)
 void collectoraction::set_fullnamemenu (displayclass &disp, cgiargsclass &args, 
                     recptprotolistclass *protos, ostream &logout) {
@@ -719 +762 @@
   text_tarray fullnames;
   vector<bool> write_protected;
+  bool is_selected = false;
   int selected_index = 0;
   int index = 0;
@@ -727 +771 @@
     if ((*rprotolist_here).p != NULL) {
 
-      // don't include z39.50 collection
+      // don't include z39.50 collections
       comerror_t err = noError;
       if ((*rprotolist_here).p->get_protocol_name (err) == "z3950proto") {
@@ -760 +804 @@
         else write_protected.push_back(false);
 
-        if (*collist_here == currentname) selected_index = index;       
+        if (*collist_here == currentname) {
+          is_selected = true;
+          selected_index = index;       
+        }
         index ++;
       }
@@ -773 +820 @@
   text_t warnindex;
   text_t fullnamemenu = "<select name=\"bc1dirname\">\n";
-  if (current_page == "srce") fullnamemenu = "<select name=\"bc1clonecol\" onChange=\"menuchange();\">\n";
+  if (current_page == "srce") {
+    fullnamemenu = "<select name=\"bc1clonecol\" onChange=\"menuchange();\">\n";
+    fullnamemenu += "<option value=defaultstructure";
+    if (!is_selected) fullnamemenu += " selected>";
+    else fullnamemenu.push_back('>');
+    fullnamemenu += "default structure\n";
+  }
   for (int i = 0; i < index; i ++) {
     // don't want write protected collections in list on "change existing
@@ -779 +832 @@
     if (write_protected[i] && current_page == "existing") continue;
     fullnamemenu += "<option value=\"" + dirnames[i] + "\"";
-    if (i == selected_index) fullnamemenu += " selected";
+    if ((i == 0 && !is_selected && current_page != "srce") || 
+    (is_selected && i == selected_index)) {
+      fullnamemenu += " selected";
+      selected_index++;
+      is_selected = false;
+    }
     fullnamemenu.push_back ('>');
     fullnamemenu += fullnames[i];
@@ -787 +845 @@
     // (if creating clone collection list)
     if (current_page == "srce") {
-      if (!first) warnindex.push_back(',');
+      if (first) warnindex += "0,";
+      else warnindex.push_back(',');
       if (uses_weird_plugin (dirnames[i])) {
     warnindex += text_t (1);
@@ -798 +857 @@
   fullnamemenu += "</select>\n";
 
-  if (!first) {
-    disp.setmacro ("fullnamemenu", "collector", fullnamemenu);
-    if (current_page == "srce")
-      disp.setmacro ("warnindex", "collector", warnindex);
-  }
+  disp.setmacro ("fullnamemenu", "collector", fullnamemenu);
+  if (current_page == "srce") {
+    disp.setmacro ("warnindex", "collector", warnindex);
+    disp.setmacro ("selectedindex", "collector", text_t(selected_index));
+  }
+}
+
+// set _filelist_, _httplist_, _ftplist_ and _badsources_ macros
+void collectoraction::set_inputsourceboxes (displayclass &disp, cgiargsclass &args, 
+                        ostream &logout) {
+  
+  if (badsources) disp.setmacro ("badsources", "collector", "1");
+
+  int numfileboxes = args["bc1filenum"].getint();
+  int numhttpboxes = args["bc1httpnum"].getint();
+  int numftpboxes = args["bc1ftpnum"].getint();
+
+  text_t filelist = get_source_box(args["bc1fileinput"], numfileboxes, "file");
+  text_t httplist = get_source_box(args["bc1httpinput"], numhttpboxes, "http");
+  text_t ftplist = get_source_box(args["bc1ftpinput"], numftpboxes, "ftp");
+
+  disp.setmacro("filelist", "collector", dm_safe(filelist));
+  disp.setmacro("httplist", "collector", dm_safe(httplist));
+  disp.setmacro("ftplist", "collector", dm_safe(ftplist));
+
+  // reset badsources and failedsources variables
+  badsources = false;
+  failedsources.erase(failedsources.begin(), failedsources.end());
+}
+
+text_t collectoraction::get_source_box (text_t inputarg, int numboxes, text_t boxtype) {
+
+  text_tarray files;
+  // this relies on filenames/URLs/ftp addresses not being able to contain
+  // commas
+  splitchar (inputarg.begin(), inputarg.end(), ',', files);
+
+  text_t rv = "<table border=0>\n";
+  for (int i = 0; i < numboxes; i++) {
+    rv += "<tr><td><input type=text name=\"bc1" + boxtype + "input\" value=\"";
+    if (i < files.size()) {
+      rv += files[i];
+    }
+    rv += "\" size=50></td>";
+    if (badsources) {
+      if ((i < files.size()) && (!files[i].empty())) {
+    if (failedsources[files[i]] == "1") {
+      rv += "<td width=40>no</td>";
+    } else {
+      rv += "<td width=40>yes</td>";
+    }
+      } else {
+    rv += "<td width=40></td>";
+      }
+    }
+    if (i+1 == numboxes) {
+      rv += "<td><a href=\"javascript:more('" + boxtype + "')\">more</a></td></tr>\n";
+    } else {
+      rv += "<td></td></tr>\n";
+    }
+  }
+  rv += "</table>\n";
+  
+  return rv;
 }
 
@@ -836 +954 @@
     
     // define it as a macro
-    disp.setmacro("cfgfile","collector",dm_safe(cfgtext));
+    disp.setmacro("cfgfile", "collector", dm_safe(cfgtext));
 
   } else {
@@ -887 +1005 @@
   // _gsdlhome_ - the gsdlhome path (dm_safe)
 
+  // _filelist_ -- these three display "input source" text boxes
+  // _httplist_ -- for the "source data" page 
+  // _ftplist_  --
+
+  // _badsources_ -- will be set to "1" if we've come from the 
+  //                 "source data" page and there's a problem 
+  //                 with the input sources
+
   text_t &collector_page = args["p"];
   int esrce = args["bc1esrce"].getint();
@@ -1030 +1156 @@
   if (collector_page == "bildstatus")
     set_statusline (disp, args, logout);
+  if (collector_page == "srce") {
+    set_inputsourceboxes (disp, args, logout);
+  }
 
   disp.setmacro ("gsdlhome", "collector", dm_safe(gsdlhome));
@@ -1507 +1636 @@
   }
 
-  if (!args["bc1inputdir1"].empty())
-    options += " -download \"" + args["bc1inputdir1"] + "\"";
-  if (!args["bc1inputdir2"].empty())
-    options += " -download \"" + args["bc1inputdir2"] + "\"";
-  if (!args["bc1inputdir3"].empty())
-    options += " -download \"" + args["bc1inputdir3"] + "\"";
-  if (!args["bc1inputdir4"].empty())
-    options += " -download \"" + args["bc1inputdir4"] + "\"";
+  text_tarray files;
+  text_t filevals = args["bc1fileinput"];
+  // this relies on filenames not being able to contain commas
+  splitchar (filevals.begin(), filevals.end(), ',', files);
+  text_tarray::const_iterator here = files.begin();
+  text_tarray::const_iterator end = files.end();
+  while (here != end) {
+    options += " -download \"" + *here + "\"";
+    here ++;
+  }
+  text_tarray https;
+  text_t httpvals = args["bc1httpinput"];
+  // this relies on URLs not being able to contain commas
+  splitchar (httpvals.begin(), httpvals.end(), ',', https);
+  here = https.begin();
+  end = https.end();
+  while (here != end) {
+    options += " -download \"" + *here + "\"";
+    here ++;
+  }
+  text_tarray ftps;
+  text_t ftpvals = args["bc1ftpinput"];
+  // this relies on ftp addresses not being able to contain commas
+  splitchar (ftpvals.begin(), ftpvals.end(), ',', ftps);
+  here = ftps.begin();
+  end = ftps.end();
+  while (here != end) {
+    options += " -download \"" + *here + "\"";
+    here ++;
+  }
 
   if (logevents) {
@@ -1707 +1858 @@
   return header;
 }
+
+bool collectoraction::check_sources (cgiargsclass &args, ostream &logout) {
+
+  bool found = false;
+
+  // check local filesystem for file:// sources
+  text_tarray files;
+  text_t filevals = args["bc1fileinput"];
+  // this relies on filenames not being able to contain commas
+  splitchar (filevals.begin(), filevals.end(), ',', files);
+  text_tarray::const_iterator here = files.begin();
+  text_tarray::const_iterator end = files.end();
+  while (here != end) {
+    text_t file = format_url(*here);
+    if (!file_exists(file) && !directory_exists(file)) {
+      failedsources[*here] = "1";
+      badsources = true;
+    }
+    found = true;
+    here ++;
+  }
+
+  // ping http:// sources
+  files.erase(files.begin(), files.end());
+  text_t httpvals = args["bc1httpinput"];
+  // this relies on URLs not being able to contain commas
+  splitchar (httpvals.begin(), httpvals.end(), ',', files);
+  here = files.begin();
+  end = files.end();
+  while (here != end) {
+    if (gsdl_system ("perl -S ping.pl -quiet http://" + format_url(*here), true, logout)) {
+      failedsources[*here] = "1";
+      badsources = true;
+    }
+    found = true;
+    here++;
+  }
+
+  // ping ftp:// sources
+  files.erase(files.begin(), files.end());
+  text_t ftpvals = args["bc1ftpinput"];
+  // this relies on FTP addresses not being able to contain commas
+  splitchar (ftpvals.begin(), ftpvals.end(), ',', files);
+  here = files.begin();
+  end = files.end();
+  while (here != end) {
+    if (gsdl_system ("perl -S ping.pl -quiet ftp://" + format_url(*here), true, logout)) {
+      failedsources[*here] = "1";
+      badsources = true;
+    }
+    found = true;
+    here++;
+  }
+
+  // set badsources if there weren't any sources at all
+  if (!found) badsources = true;
+
+  if (badsources) return false;
+  return true;
+}
+
+// format_url simply strips "http://", "ftp://", or "file://" off the
+// beginning of url if they're there
+text_t collectoraction::format_url (const text_t &url) {
+  text_t::const_iterator begin = url.begin();
+  text_t::const_iterator end = url.end();
+
+  if (url.size() >= 7) {
+    text_t prefix = substr(begin, begin+7);
+    if (prefix == "http://" || prefix == "file://") {
+      return substr(begin+7, end);
+    }
+  }
+  if (url.size() >= 6) {
+    if (substr(begin, begin+6) == "ftp://") {
+      return substr(begin+6, end);
+    }
+  }
+  return url;
+}

trunk/gsdl/src/recpt/collectoraction.h

@@ -45 +45 @@
   bool do_mkcol;
 
+  bool badsources;
+  text_tmap failedsources;
+
   text_t get_button (const text_t &thispage, const text_t &color, 
              const text_t &type, bool enabled);
@@ -50 +53 @@
   void set_fullnamemenu (displayclass &disp, cgiargsclass &args,
              recptprotolistclass *protos, ostream &logout);
+
+  void set_inputsourceboxes (displayclass &disp, cgiargsclass &args, 
+                 ostream &logout);
+
+  text_t get_source_box (text_t inputarg, int numboxes, text_t boxtype);
 
   void set_cfgfile (displayclass &disp, cgiargsclass &args, ostream &logout);
@@ -88 +96 @@
   text_t get_event_header (cgiargsclass &args);
 
+  bool check_sources (cgiargsclass &args, ostream &logout);
+
+  text_t format_url (const text_t &url);
+
 public:
   collectoraction ();