[tracker/wip/sam/resource: 10/15] Enable experimental --output-format=json-ld for TrackerExtract

[Sam Thursfield <sthursfield src gnome org>]

commit fdb034f991c512e7d273b7ad86642f6532ca30ca
Author: Sam Thursfield <sam afuera me uk>
Date:   Thu Apr 7 18:34:44 2016 +0100

    Enable experimental --output-format=json-ld for TrackerExtract

 src/tracker-extract/tracker-extract.c |   88 ++++++++++++++++++++++----------
 src/tracker/tracker-extract.c         |    8 ++--
 2 files changed, 64 insertions(+), 32 deletions(-)
---
diff --git a/src/tracker-extract/tracker-extract.c b/src/tracker-extract/tracker-extract.c
index 9ff3c31..d420d7d 100644
--- a/src/tracker-extract/tracker-extract.c
+++ b/src/tracker-extract/tracker-extract.c
@@ -795,45 +795,77 @@ tracker_extract_get_metadata_by_cmdline (TrackerExtract             *object,
        while (task->cur_func) {
                if (!filter_module (object, task->cur_module) &&
                    get_file_metadata (task, &info)) {
-                       const gchar *preupdate_str, *postupdate_str, *statements_str, *where;
-                       TrackerSparqlBuilder *builder;
-
                        no_data_or_modules = FALSE;
-                       preupdate_str = statements_str = postupdate_str = NULL;
 
-                       builder = tracker_extract_info_get_metadata_builder (info);
+                       /* This will break in one case or the other depending on what extractor was used... 
FIXME */
+                       if (output_format == TRACKER_SERIALIZATION_FORMAT_SPARQL) {
+                               /* Only works for old-style extractors using TrackerSparqlBuilder */
+                               const gchar *preupdate_str, *postupdate_str, *statements_str, *where;
+                               TrackerSparqlBuilder *builder;
 
-                       if (tracker_sparql_builder_get_length (builder) > 0) {
-                               statements_str = tracker_sparql_builder_get_result (builder);
-                       }
+                               preupdate_str = statements_str = postupdate_str = NULL;
 
-                       builder = tracker_extract_info_get_preupdate_builder (info);
+                               builder = tracker_extract_info_get_metadata_builder (info);
 
-                       if (tracker_sparql_builder_get_length (builder) > 0) {
-                               preupdate_str = tracker_sparql_builder_get_result (builder);
-                       }
+                               if (tracker_sparql_builder_get_length (builder) > 0) {
+                                       statements_str = tracker_sparql_builder_get_result (builder);
+                               }
 
-                       builder = tracker_extract_info_get_postupdate_builder (info);
+                               builder = tracker_extract_info_get_preupdate_builder (info);
 
-                       if (tracker_sparql_builder_get_length (builder) > 0) {
-                               postupdate_str = tracker_sparql_builder_get_result (builder);
-                       }
+                               if (tracker_sparql_builder_get_length (builder) > 0) {
+                                       preupdate_str = tracker_sparql_builder_get_result (builder);
+                               }
+
+                               builder = tracker_extract_info_get_postupdate_builder (info);
 
-                       where = tracker_extract_info_get_where_clause (info);
+                               if (tracker_sparql_builder_get_length (builder) > 0) {
+                                       postupdate_str = tracker_sparql_builder_get_result (builder);
+                               }
 
-                       g_print ("\n");
+                               where = tracker_extract_info_get_where_clause (info);
 
-                       g_print ("SPARQL pre-update:\n--\n%s--\n\n",
-                                preupdate_str ? preupdate_str : "");
-                       g_print ("SPARQL item:\n--\n%s--\n\n",
-                                statements_str ? statements_str : "");
-                       g_print ("SPARQL where clause:\n--\n%s--\n\n",
-                                where ? where : "");
-                       g_print ("SPARQL post-update:\n--\n%s--\n\n",
-                                postupdate_str ? postupdate_str : "");
+                               g_print ("\n");
 
-                       tracker_extract_info_unref (info);
-                       break;
+                               g_print ("SPARQL pre-update:\n--\n%s--\n\n",
+                                        preupdate_str ? preupdate_str : "");
+                               g_print ("SPARQL item:\n--\n%s--\n\n",
+                                        statements_str ? statements_str : "");
+                               g_print ("SPARQL where clause:\n--\n%s--\n\n",
+                                        where ? where : "");
+                               g_print ("SPARQL post-update:\n--\n%s--\n\n",
+                                        postupdate_str ? postupdate_str : "");
+
+                               tracker_extract_info_unref (info);
+                               break;
+                       } else {
+                               /* For experimental new TrackerResource extractors */
+                               GList *resources = tracker_extract_info_get_resources (info);
+                               GError *error = NULL;
+                               /* FIXME: we should return a JSON-LD @graph if there are multiple resources,
+                                * not a sequence of JSON documents!!!
+                                */
+                               for (GList *l = resources; l; l = l->next) {
+                                       char *json;
+
+                                       /* If this was going into the tracker-store we'd generate a unique ID
+                                        * here, so that the data persisted across file renames.
+                                        */
+                                       tracker_resource_set_identifier (TRACKER_RESOURCE (l->data), uri);
+
+                                       json = tracker_resource_print_jsonld (TRACKER_RESOURCE (l->data), 
&error);
+                                       if (json) {
+                                               g_print ("%s\n", json);
+                                               g_free (json);
+                                       } else {
+                                               g_printerr ("%s\n", error->message);
+                                               g_error_free (error);
+                                       }
+                               }
+
+                               tracker_extract_info_unref (info);
+                               break;
+                       }
                } else {
                        if (!tracker_mimetype_info_iter_next (task->mimetype_handlers)) {
                                break;
diff --git a/src/tracker/tracker-extract.c b/src/tracker/tracker-extract.c
index 5f5b838..6abcf4e 100644
--- a/src/tracker/tracker-extract.c
+++ b/src/tracker/tracker-extract.c
@@ -30,20 +30,20 @@
 #include "tracker-config.h"
 #include "tracker-extract.h"
 
-static gchar *output_format = "json-ld";
 static gchar *verbosity;
+static gchar *output_format = "json-ld";
 static gchar **filenames;
 
 #define EXTRACT_OPTIONS_ENABLED()        \
        ((filenames && g_strv_length (filenames) > 0))
 
 static GOptionEntry entries[] = {
-       { "output-format", 'o', 0, G_OPTION_ARG_STRING, &output_format,
-         N_("Output results format: 'json' or 'sparql'"),
-         N_("FORMAT") },
        { "verbosity", 'v', 0, G_OPTION_ARG_STRING, &verbosity,
          N_("Sets the logging verbosity to LEVEL ('debug', 'detailed', 'minimal', 'errors') for all 
processes"),
          N_("LEVEL") },
+       { "output-format", 'o', 0, G_OPTION_ARG_STRING, &output_format,
+         N_("Output results format: 'json' or 'sparql'"),
+         N_("FORMAT") },
        { G_OPTION_REMAINING, 0, 0, G_OPTION_ARG_FILENAME_ARRAY, &filenames,
          N_("FILE"),
          N_("FILE") },