Support querying files by contained text

ODF files also supported via odt2txt

4 files changed, 192 insertions, 1 deletions

diff --git a/libcaja-private/caja-query.c b/libcaja-private/caja-query.c
index e7baf72f..1064fd16 100644
--- a/libcaja-private/caja-query.c
+++ b/libcaja-private/caja-query.c
@@ -39,6 +39,7 @@ struct CajaQueryDetails
     GList *tags;
     gint64 timestamp;
     gint64 size;
+    char *contained_text;
 };
 
 static void  caja_query_class_init       (CajaQueryClass *class);
@@ -485,3 +486,14 @@ gint64 caja_query_get_size(CajaQuery *query)
 {
     return query->details->size;
 }
+
+void caja_query_set_contained_text (CajaQuery *query, const char *text)
+{
+    g_free (query->details->contained_text);
+    query->details->contained_text = g_strdup (text);
+}
+
+char *caja_query_get_contained_text (CajaQuery *query)
+{
+    return g_strdup (query->details->contained_text);
+}
diff --git a/libcaja-private/caja-query.h b/libcaja-private/caja-query.h
index a1ec2864..117e813b 100644
--- a/libcaja-private/caja-query.h
+++ b/libcaja-private/caja-query.h
@@ -75,4 +75,7 @@ void           caja_query_set_timestamp      (CajaQuery *query, gint64 sec);
 gint64         caja_query_get_size           (CajaQuery *query);
 void           caja_query_set_size           (CajaQuery *query, gint64 size);
 
+char *         caja_query_get_contained_text (CajaQuery *query);
+void           caja_query_set_contained_text (CajaQuery *query, const char *text);
+
 #endif /* CAJA_QUERY_H */
diff --git a/libcaja-private/caja-search-engine-simple.c b/libcaja-private/caja-search-engine-simple.c
index 40abeb3a..f9c6498d 100644
--- a/libcaja-private/caja-search-engine-simple.c
+++ b/libcaja-private/caja-search-engine-simple.c
@@ -38,6 +38,7 @@ typedef struct
     CajaSearchEngineSimple *engine;
     GCancellable *cancellable;
 
+    char *contained_text;
     GList *mime_types;
     GList *tags;
     char **words;
@@ -129,6 +130,7 @@ search_thread_data_new (CajaSearchEngineSimple *engine,
     data->mime_types = caja_query_get_mime_types (query);
     data->timestamp = caja_query_get_timestamp (query);
     data->size = caja_query_get_size (query);
+    data->contained_text = caja_query_get_contained_text (query);
 
     data->cancellable = g_cancellable_new ();
 
@@ -147,6 +149,7 @@ search_thread_data_free (SearchThreadData *data)
     g_list_free_full (data->tags, g_free);
     g_list_free_full (data->mime_types, g_free);
     g_list_free_full (data->uri_hits, g_free);
+    g_free (data->contained_text);
     g_free (data);
 }
 
@@ -342,6 +345,102 @@ file_has_all_tags (GFileInfo *info, GList *tags)
     return TRUE;
 }
 
+static inline gboolean
+check_odt2txt () {
+    gboolean rc = TRUE;
+    int exit = 0;
+    gchar *output = NULL;
+
+    gboolean cmd_rc = g_spawn_command_line_sync ("odt2txt --version", &output, NULL, &exit, NULL);
+
+    if (!cmd_rc || exit != 0 ||
+        !output || !g_str_has_prefix (output, "odt2txt"))
+    {
+        rc = FALSE;
+    }
+
+    g_free (output);
+    return rc;
+}
+
+static inline gchar *
+read_odt (const char *filepath) {
+    gchar *command = g_strdup_printf ("odt2txt \"%s\"", filepath);
+    gchar *output  = NULL;
+    int exit = 0;
+
+    gboolean rc = g_spawn_command_line_sync (command, &output, NULL, &exit, NULL);
+    if (!rc || exit != 0) {
+        g_free (output);
+        g_free (command);
+        return NULL;
+    }
+
+    g_free (command);
+    return output;
+}
+
+static inline gchar *
+utf8_normalize_strdown (const char *str) {
+    gchar* lower = NULL;
+    gchar *normalized = g_utf8_normalize (str, -1, G_NORMALIZE_DEFAULT);
+
+    if (normalized)
+        lower = g_utf8_strdown (normalized, -1);
+
+    g_free (normalized);
+
+    return lower;
+}
+
+static inline gboolean
+is_file_has_str (
+    const char *filepath,
+    const char *str,
+    const char *mime_type,
+    gboolean odt2txt_available)
+{
+    gboolean rc = TRUE;
+    gchar *contents = NULL;
+    gchar *lower_contents = NULL;
+    gchar *lower_str = NULL;
+
+    if (str[0] == '\0') {
+        return TRUE;
+    }
+
+    if (g_content_type_is_mime_type (mime_type, "text/plain")) {
+        rc = g_file_get_contents (filepath, &contents, NULL, NULL);
+    }
+    else {
+        if (!odt2txt_available) {
+            g_warning ("Can't search in file '%s'. odt2txt not found.", filepath);
+            rc = FALSE;
+        }
+        else {
+            contents = read_odt (filepath);
+            if (!contents)
+                rc = FALSE;
+        }
+    }
+
+    if (rc) {
+        lower_str = utf8_normalize_strdown (str);
+        lower_contents = utf8_normalize_strdown (contents);
+
+        if (lower_str && lower_contents && strstr (lower_contents, lower_str))
+            rc = TRUE;
+        else
+            rc = FALSE;
+    }
+
+    g_free (contents);
+    g_free (lower_str);
+    g_free (lower_contents);
+
+    return rc;
+}
+
 static void
 visit_directory (GFile *dir, SearchThreadData *data)
 {
@@ -358,9 +457,11 @@ visit_directory (GFile *dir, SearchThreadData *data)
     GTimeVal result;
     gchar *attributes;
     GString *attr_string;
+    gchar *filepath = NULL;
+    gboolean odt2txt_available = FALSE;
 
     attr_string = g_string_new (STD_ATTRIBUTES);
-    if (data->mime_types != NULL) {
+    if (data->mime_types != NULL || data->contained_text != NULL) {
         g_string_append (attr_string, "," G_FILE_ATTRIBUTE_STANDARD_CONTENT_TYPE);
     }
     if (data->tags != NULL) {
@@ -374,6 +475,10 @@ visit_directory (GFile *dir, SearchThreadData *data)
         g_string_append (attr_string, "," G_FILE_ATTRIBUTE_STANDARD_SIZE);
     }
 
+    if (data->contained_text != NULL) {
+        odt2txt_available = check_odt2txt();
+    }
+
     attributes = g_string_free (attr_string, FALSE);
     enumerator = g_file_enumerate_children (dir, (const char*)attributes, 0,
                                             data->cancellable, NULL);
@@ -456,6 +561,26 @@ visit_directory (GFile *dir, SearchThreadData *data)
 
         child = g_file_get_child (dir, g_file_info_get_name (info));
 
+        if (hit && data->contained_text) {
+            mime_type = g_file_info_get_content_type (info);
+
+            if (g_content_type_is_mime_type (mime_type, "text/plain") ||
+                g_content_type_equals (mime_type, "application/vnd.oasis.opendocument.text") ||
+                g_content_type_equals (mime_type, "application/vnd.oasis.opendocument.text-template") ||
+                g_content_type_equals (mime_type, "application/vnd.oasis.opendocument.spreadsheet") ||
+                g_content_type_equals (mime_type, "application/vnd.oasis.opendocument.spreadsheet-template") ||
+                g_content_type_equals (mime_type, "application/vnd.oasis.opendocument.presentation") ||
+                g_content_type_equals (mime_type, "application/vnd.oasis.opendocument.presentation-template")
+            ) {
+                g_free (filepath);
+                filepath = g_file_get_path (child);
+                hit = is_file_has_str (filepath, data->contained_text, mime_type, odt2txt_available);
+            }
+            else {
+                hit = FALSE;
+            }
+        }
+
         if (hit)
         {
             data->uri_hits = g_list_prepend (data->uri_hits, g_file_get_uri (child));
@@ -495,6 +620,7 @@ next:
         g_object_unref (info);
     }
 
+    g_free (filepath);
     g_object_unref (enumerator);
 }
 
diff --git a/src/caja-query-editor.c b/src/caja-query-editor.c
index 55dff943..5fbfddc9 100644
--- a/src/caja-query-editor.c
+++ b/src/caja-query-editor.c
@@ -57,6 +57,7 @@ typedef enum
     CAJA_QUERY_EDITOR_ROW_TAGS,
     CAJA_QUERY_EDITOR_ROW_TIME_MODIFIED,
     CAJA_QUERY_EDITOR_ROW_SIZE,
+    CAJA_QUERY_EDITOR_ROW_CONTAINED_TEXT,
 
     CAJA_QUERY_EDITOR_ROW_LAST
 } CajaQueryEditorRowType;
@@ -157,6 +158,13 @@ static void         size_row_free_data(CajaQueryEditorRow *row);
 static void         size_add_rows_from_query(CajaQueryEditor *editor,
                                              CajaQuery *query);
 
+static GtkWidget   *contained_text_row_create_widgets(CajaQueryEditorRow *row);
+static void         contained_text_row_add_to_query(CajaQueryEditorRow *row,
+                                             CajaQuery *query);
+static void         contained_text_row_free_data(CajaQueryEditorRow *row);
+static void         contained_text_add_rows_from_query(CajaQueryEditor *editor,
+                                                CajaQuery *query);
+
 static CajaQueryEditorRowOps row_type[] =
 {
     {
@@ -193,6 +201,13 @@ static CajaQueryEditorRowOps row_type[] =
         size_row_add_to_query,
         size_row_free_data,
         size_add_rows_from_query
+    },
+    {
+        N_("Contained text"),
+        contained_text_row_create_widgets,
+        contained_text_row_add_to_query,
+        contained_text_row_free_data,
+        contained_text_add_rows_from_query
     }
 };
 
@@ -1323,6 +1338,41 @@ static void size_add_rows_from_query(CajaQueryEditor *editor, CajaQuery *query)
 {
 }
 
+static GtkWidget *
+contained_text_row_create_widgets (CajaQueryEditorRow *row)
+{
+    GtkWidget *entry = gtk_entry_new();
+    gtk_widget_set_tooltip_text (entry,
+        _("Matches files that contains specified text."));
+
+    gtk_entry_set_placeholder_text (GTK_ENTRY (entry),
+        _("Matches files that contains specified text."));
+
+    gtk_widget_show (entry);
+    gtk_box_pack_start (GTK_BOX (row->hbox), entry, TRUE, TRUE, 0);
+    g_signal_connect (entry, "activate", G_CALLBACK (go_search_cb), row->editor);
+
+    return entry;
+}
+
+static void
+contained_text_row_add_to_query (CajaQueryEditorRow *row, CajaQuery *query)
+{
+    GtkEntry *entry = GTK_ENTRY (row->type_widget);
+    const gchar *text = gtk_entry_get_text (entry);
+
+    caja_query_set_contained_text (query, text);
+}
+
+static void
+contained_text_row_free_data (CajaQueryEditorRow *row)
+{
+}
+
+static void
+contained_text_add_rows_from_query (CajaQueryEditor *editor, CajaQuery *query)
+{
+}
 
 static CajaQueryEditorRowType
 get_next_free_type (CajaQueryEditor *editor)