libsysprof-analyze: automatically handle compressed streams

If we want /proc/kallsyms and we discover /proc/kallsyms.gz, then use the
/proc/kallsyms.gz instead and transparently decompress it. Also, list the
file in sysprof_document_list_files() as /proc/kallsyms instead of
/proc/kallsyms.gz as that is really the intention (but mark the compressed
bit for decoding the chunks).

src/libsysprof-analyze/sysprof-document-file-private.h

@@ -25,6 +25,7 @@
 G_BEGIN_DECLS
 
 SysprofDocumentFile *_sysprof_document_file_new (const char *path,
-                                                 GPtrArray  *file_chunks);
+                                                 GPtrArray  *file_chunks,
+                                                 gboolean    compressed);
 
 G_END_DECLS

src/libsysprof-analyze/sysprof-document-file.c

@@ -26,9 +26,10 @@
 
 struct _SysprofDocumentFile
 {
-  GObject parent_instance;
+  GObject    parent_instance;
-  char *path;
+  char      *path;
   GPtrArray *file_chunks;
+  guint      compressed : 1;
 };
 
 enum {
@@ -104,7 +105,8 @@ sysprof_document_file_init (SysprofDocumentFile *self)
 
 SysprofDocumentFile *
 _sysprof_document_file_new (const char *path,
-                            GPtrArray  *file_chunks)
+                            GPtrArray  *file_chunks,
+                            gboolean    compressed)
 {
   SysprofDocumentFile *self;
 
@@ -114,6 +116,7 @@ _sysprof_document_file_new (const char *path,
   self = g_object_new (SYSPROF_TYPE_DOCUMENT_FILE, NULL);
   self->path = g_strdup (path);
   self->file_chunks = file_chunks;
+  self->compressed = !!compressed;
 
   return self;
 }
@@ -169,7 +172,7 @@ sysprof_document_file_dup_bytes (SysprofDocumentFile *self)
 GInputStream *
 sysprof_document_file_read (SysprofDocumentFile *self)
 {
-  GInputStream *input;
+  g_autoptr(GInputStream) input = NULL;
 
   g_return_val_if_fail (SYSPROF_IS_DOCUMENT_FILE (self), NULL);
 
@@ -193,6 +196,13 @@ sysprof_document_file_read (SysprofDocumentFile *self)
       g_memory_input_stream_add_bytes (G_MEMORY_INPUT_STREAM (input), bytes);
     }
 
+  if (self->compressed)
+    {
+      g_autoptr(GZlibDecompressor) zlib = g_zlib_decompressor_new (G_ZLIB_COMPRESSOR_FORMAT_GZIP);
+
+      return g_converter_input_stream_new (input, G_CONVERTER (zlib));
+    }
+
   return g_steal_pointer (&input);
 }
 

src/libsysprof-analyze/sysprof-document.c

@@ -895,14 +895,19 @@ SysprofDocumentFile *
 sysprof_document_lookup_file (SysprofDocument *self,
                               const char      *path)
 {
+  g_autofree char *gz_path = NULL;
   gpointer key, value;
 
   g_return_val_if_fail (SYSPROF_IS_DOCUMENT (self), NULL);
   g_return_val_if_fail (path != NULL, NULL);
 
-  if (g_hash_table_lookup_extended (self->files_first_position, path, &key, &value))
+  gz_path = g_strdup_printf ("%s.gz", path);
+
+  if (g_hash_table_lookup_extended (self->files_first_position, path, &key, &value) ||
+      g_hash_table_lookup_extended (self->files_first_position, gz_path, &key, &value))
     {
       g_autoptr(GPtrArray) file_chunks = g_ptr_array_new_with_free_func (g_object_unref);
+      const char *real_path = key;
       EggBitsetIter iter;
       guint target = GPOINTER_TO_SIZE (value);
       guint i;
@@ -913,7 +918,7 @@ sysprof_document_lookup_file (SysprofDocument *self,
             {
               g_autoptr(SysprofDocumentFileChunk) file_chunk = sysprof_document_get_item ((GListModel *)self, i);
 
-              if (g_strcmp0 (path, sysprof_document_file_chunk_get_path (file_chunk)) == 0)
+              if (g_strcmp0 (real_path, sysprof_document_file_chunk_get_path (file_chunk)) == 0)
                 {
                   gboolean is_last = sysprof_document_file_chunk_get_is_last (file_chunk);
 
@@ -926,7 +931,9 @@ sysprof_document_lookup_file (SysprofDocument *self,
           while (egg_bitset_iter_next (&iter, &i));
         }
 
-      return _sysprof_document_file_new (path, g_steal_pointer (&file_chunks));
+      return _sysprof_document_file_new (path,
+                                         g_steal_pointer (&file_chunks),
+                                         g_strcmp0 (real_path, gz_path) == 0);
     }
 
   return NULL;
@@ -957,6 +964,7 @@ sysprof_document_list_files (SysprofDocument *self)
     {
       g_autoptr(SysprofDocumentFile) file = NULL;
       g_autoptr(GPtrArray) file_chunks = g_ptr_array_new_with_free_func (g_object_unref);
+      g_autofree char *no_gz_path = NULL;
       const char *path = key;
       guint target = GPOINTER_TO_SIZE (value);
       guint i;
@@ -980,7 +988,12 @@ sysprof_document_list_files (SysprofDocument *self)
           while (egg_bitset_iter_next (&iter, &i));
         }
 
-      file = _sysprof_document_file_new (path, g_steal_pointer (&file_chunks));
+      if (g_str_has_suffix (path, ".gz"))
+        path = no_gz_path = g_strndup (path, strlen (path) - 3);
+
+      file = _sysprof_document_file_new (path,
+                                         g_steal_pointer (&file_chunks),
+                                         no_gz_path != NULL);
 
       g_list_store_append (model, file);
     }

src/libsysprof-analyze/sysprof-kallsyms-symbolizer.c

@@ -191,10 +191,9 @@ sysprof_kallsyms_symbolizer_prepare_async (SysprofSymbolizer   *symbolizer,
 {
   SysprofKallsymsSymbolizer *self = (SysprofKallsymsSymbolizer *)symbolizer;
   g_autoptr(SysprofDocumentFile) file = NULL;
-  g_autoptr(GZlibDecompressor) decompressor = NULL;
-  g_autoptr(GInputStream) input_gz = NULL;
   g_autoptr(GInputStream) input = NULL;
   g_autoptr(GTask) task = NULL;
+  GInputStream *base_stream;
 
   g_assert (SYSPROF_IS_KALLSYMS_SYMBOLIZER (self));
   g_assert (SYSPROF_IS_DOCUMENT (document));
@@ -203,9 +202,12 @@ sysprof_kallsyms_symbolizer_prepare_async (SysprofSymbolizer   *symbolizer,
   task = g_task_new (self, cancellable, callback, user_data);
   g_task_set_source_tag (task, sysprof_kallsyms_symbolizer_prepare_async);
 
-  if (self->stream == NULL)
+  base_stream = self->stream;
+
+  if (base_stream == NULL)
     {
-      if (!(file = sysprof_document_lookup_file (document, "/proc/kallsyms.gz")))
+
+      if (!(file = sysprof_document_lookup_file (document, "/proc/kallsyms")))
         {
           g_task_return_new_error (task,
                                    G_IO_ERROR,
@@ -214,20 +216,15 @@ sysprof_kallsyms_symbolizer_prepare_async (SysprofSymbolizer   *symbolizer,
           return;
         }
 
-      decompressor = g_zlib_decompressor_new (G_ZLIB_COMPRESSOR_FORMAT_GZIP);
+      base_stream = input = sysprof_document_file_read (file);
-      input_gz = sysprof_document_file_read (file);
+    }
-      input = g_converter_input_stream_new (input_gz, G_CONVERTER (decompressor));
 
-      g_task_set_task_data (task,
+  g_assert (base_stream != NULL);
-                            g_data_input_stream_new (input),
+  g_assert (G_IS_INPUT_STREAM (base_stream));
-                            g_object_unref);
+
-    }
+  g_task_set_task_data (task,
-  else
+                        g_data_input_stream_new (base_stream),
-    {
+                        g_object_unref);
-      g_task_set_task_data (task,
-                            g_steal_pointer (&self->stream),
-                            g_object_unref);
-    }
 
   g_task_run_in_thread (task, sysprof_kallsyms_symbolizer_prepare_worker);
 }
@@ -364,9 +361,7 @@ sysprof_kallsyms_symbolizer_new_for_symbols (GInputStream *symbols)
   g_return_val_if_fail (G_IS_INPUT_STREAM (symbols), NULL);
 
   self = g_object_new (SYSPROF_TYPE_KALLSYMS_SYMBOLIZER, NULL);
-  self->stream = G_INPUT_STREAM (g_data_input_stream_new (symbols));
+  self->stream = symbols;
-
-  g_object_unref (symbols);
 
   return SYSPROF_SYMBOLIZER (self);
 }