[evolution-rss] better handle enclosures obey filesize if present

[Lucian Langa <lucilanga src gnome org>]

commit 5a2f831431c032387db3523bd56d710093a28640
Author: Lucian Langa <lucilanga gnome org>
Date:   Mon Feb 23 01:17:46 2015 +0100

    better handle enclosures
    obey filesize if present

 src/parser.c |   24 +++++++++++----
 src/rss.c    |   94 ++++++++++++++++++++++++++++++++++++++++++++++++---------
 src/rss.h    |    5 ++-
 3 files changed, 100 insertions(+), 23 deletions(-)
---
diff --git a/src/parser.c b/src/parser.c
index 1d7847d..e8b4fe6 100644
--- a/src/parser.c
+++ b/src/parser.c
@@ -1080,7 +1080,8 @@ parse_channel_line(xmlNode *top, gchar *feed_name, RDF *r, gchar **article_uid)
        gchar *qsafe, *tcat;
        GList *category = NULL;
        create_feed *CF;
-       GList *attachments = NULL;
+       GList *attachments = NULL, *attsizes = NULL, *l = NULL;
+       GHashTable *attlengths = NULL;
        gchar *base = NULL, *main_date = NULL;
 
        if (r) {
@@ -1220,10 +1221,20 @@ parse_channel_line(xmlNode *top, gchar *feed_name, RDF *r, gchar **article_uid)
                }
                //handle attatchments (can be multiple)
                attachments = layer_find_tag_prop(top, "media", "url");
-               if (!attachments)
+               attsizes = layer_find_tag_prop(top, "media", "fileSize");
+               if (!attachments) {
                        attachments = layer_query_find_all_prop (top,
                                "link", (xmlChar *)"rel",
                                "enclosure", (xmlChar *)"href");
+                       attsizes = layer_query_find_all_prop (top,
+                               "link", (xmlChar *)"rel",
+                               "enclosure", (xmlChar *)"length");
+               }
+               attlengths = g_hash_table_new_full(g_str_hash, g_str_equal, g_free, g_free);
+               for (l = g_list_first(attsizes); l != NULL; l = l->next) {
+                       gchar *c = get_url_basename(g_list_first(attachments)->data);
+                       g_hash_table_insert(attlengths, g_strdup(c), g_strdup(l->data));
+               }
 
 //                char *comments = g_strdup(layer_find (top, "comments", NULL));       //RSS,
                comments = (gchar *)layer_find_ns_tag(top, "wfw", "commentRss", NULL); //add slash:comments
@@ -1262,6 +1273,7 @@ parse_channel_line(xmlNode *top, gchar *feed_name, RDF *r, gchar **article_uid)
                CF->website     = g_strdup(link);
                CF->encl        = g_strdup(encl);
                CF->attachments = attachments;
+               CF->attlengths  = attlengths;
                CF->comments    = g_strdup(comments);
                CF->feed_fname  = g_strdup(feed_name);  //feed file name
                CF->feed_uri    = g_strdup(feed);       //feed uri (uid!)
@@ -1369,11 +1381,11 @@ display_channel_items_sync(AsyncData *asyncr)
                if (g_settings_get_boolean(rss_settings, CONF_DOWNLOAD_ENCLOSURES)) {
 #endif
                        if (CF->encl) {
-                               process_enclosure(CF);
-                               goto done;
+                               if (process_enclosure(CF))
+                                       goto done;
                        } else if (g_list_length(CF->attachments)) {
-                               process_attachments(CF);
-                               goto done;
+                               if (process_attachments(CF))
+                                       goto done;
                        }
                }
 
diff --git a/src/rss.c b/src/rss.c
index 983434a..e80c549 100644
--- a/src/rss.c
+++ b/src/rss.c
@@ -3916,6 +3916,7 @@ create_mail(create_feed *CF)
        GList *p, *l;
        gchar *time_str, *buf;
        gint offset;
+       guint c = 0;
 
        mail_folder = check_feed_folder(CF->full_path);
        if (!mail_folder)
@@ -4052,7 +4053,25 @@ create_mail(create_feed *CF)
 #else
                        camel_object_unref(part);
 #endif
+#if EVOLUTION_VERSION < 30304
+               GConfClient *client = gconf_client_get_default();
+#else
+               rss_settings = g_settings_new(RSS_CONF_SCHEMA);
+#endif
+               gdouble encl_max_size = g_settings_get_double(
+                                       rss_settings, CONF_ENCLOSURE_SIZE)*1024;
                for (l = g_list_first(CF->attachedfiles); l != NULL; l = l->next) {
+                       gdouble emax;
+                       gchar *emaxstr = g_hash_table_lookup(CF->attlengths,
+                                               get_url_basename(l->data));
+                       if (emaxstr)
+                               emax = atof(emaxstr);
+                       else
+                               emax = 0;
+                       if (emax > encl_max_size) {
+                               continue;
+                       }
+                       c++;
                        msgp = file_to_message(l->data);
                        if (msgp) {
                                camel_multipart_add_part(mp, msgp);
@@ -4063,6 +4082,8 @@ create_mail(create_feed *CF)
 #endif
                        }
                }
+               if (!c)
+                       goto out;
 #if EVOLUTION_VERSION >= 23100
                camel_medium_set_content((CamelMedium *)new, (CamelDataWrapper *)mp);
 #else
@@ -4112,6 +4133,7 @@ create_mail(create_feed *CF)
                camel_object_unref(mp);
 #endif
        } else
+out:
 #if EVOLUTION_VERSION >= 23100
                camel_medium_set_content(CAMEL_MEDIUM(new), CAMEL_DATA_WRAPPER(rtext));
 #else
@@ -4292,11 +4314,14 @@ finish_attachment (
        cfl *user_data);
 #endif
 
-void
+gboolean
 process_attachments(create_feed *CF)
 {
        cfl *CFL;
        GList *l = g_list_first(CF->attachments);
+       gchar *emaxstr = NULL;
+       gdouble emax;
+       guint proc = 0;
 
        g_return_if_fail(CF->attachments != NULL);
 
@@ -4306,6 +4331,25 @@ process_attachments(create_feed *CF)
                if (g_list_find_custom(rf->enclist, l->data,
                        (GCompareFunc)strcmp))
                        continue;
+               //don't queue download if it exceeds max allowed size
+#if EVOLUTION_VERSION < 30304
+               GConfClient *client = gconf_client_get_default();
+#else
+               rss_settings = g_settings_new(RSS_CONF_SCHEMA);
+#endif
+               gdouble encl_max_size = g_settings_get_double(
+                                       rss_settings, CONF_ENCLOSURE_SIZE)*1024;
+               if (CF->encl) {
+                       emaxstr = g_hash_table_lookup(CF->attlengths, get_url_basename(CF->encl));
+               }
+               if (emaxstr)
+                       emax = atof(emaxstr);
+               else
+                       emax = 0;
+               if (emax > encl_max_size) {
+                       continue;
+               }
+               proc++;
                CFL = g_new0(cfl, 1);
                CFL->url = l->data;
                CFL->CF = CF;
@@ -4320,6 +4364,9 @@ process_attachments(create_feed *CF)
                        1,
                        NULL);
        } while ((l = l->next));
+       if (proc)
+               return TRUE;
+       return FALSE;
 }
 
 
@@ -4378,27 +4425,44 @@ out:    if (user_data->file)
                                NULL);
 }
 
-void
+gboolean
 process_enclosure(create_feed *CF)
 {
        cfl *CFL;
+       gdouble emax;
+       gchar *emaxstr;
 
        if (g_list_find_custom(rf->enclist, CF->encl,
                        (GCompareFunc)strcmp)) {
-               return;
+               return TRUE; //assume true for now
        }
-       d("enclosure file:%s\n", CF->encl)
-       CFL = g_new0(cfl, 1);
-       CFL->url = CF->encl;
-       CFL->CF = CF;
-       download_unblocking(
-               CF->encl,
-               download_chunk,
-               CFL,
-               (gpointer)finish_enclosure,
-               CFL,
-               1,
-               NULL);
+#if EVOLUTION_VERSION < 30304
+       GConfClient *client = gconf_client_get_default();
+#else
+       rss_settings = g_settings_new(RSS_CONF_SCHEMA);
+#endif
+       gdouble encl_max_size = g_settings_get_double(
+                               rss_settings, CONF_ENCLOSURE_SIZE)*1024;
+       emaxstr = g_hash_table_lookup(CF->attlengths, get_url_basename(CF->encl));
+       if (emaxstr)
+               emax = atof(emaxstr);
+       else
+               emax = 0;
+       if (emax <= encl_max_size) {
+               d("enclosure file:%s\n", CF->encl)
+               CFL = g_new0(cfl, 1);
+               CFL->url = CF->encl;
+               CFL->CF = CF;
+               download_unblocking(
+                       CF->encl,
+                       download_chunk,
+                       CFL,
+                       (gpointer)finish_enclosure,
+                       CFL,
+                       1,
+                       NULL);
+       } else
+               return FALSE;
 }
 
 void
diff --git a/src/rss.h b/src/rss.h
index ae59d68..7b9ff42 100644
--- a/src/rss.h
+++ b/src/rss.h
@@ -339,6 +339,7 @@ typedef struct CREATE_FEED {        /* used by create_mail function when called by unbl
        gchar *encl;            //feed enclosure
        gchar *enclurl;
        GList *attachments;     //feed media files
+       GHashTable *attlengths; //feed media files
        GList *attachedfiles;   //list of downloaded media files
        guint attachmentsqueue; //list of downloaded media files
        FILE *efile;            //enclosure file
@@ -500,8 +501,8 @@ void download_chunk(
        gpointer statusdata,
        gpointer data);
 
-void process_enclosure(create_feed *CF);
-void process_attachments(create_feed *CF);
+gboolean process_enclosure(create_feed *CF);
+gboolean process_attachments(create_feed *CF);
 
 #ifdef HAVE_GECKO
 void rss_mozilla_init(void);