From 13aaf6737cecf18cd29e91c87e67a721f1447bd8 Mon Sep 17 00:00:00 2001
From: Marcus Holland-Moritz <github@mhxnet.de>
Date: Tue, 1 Dec 2020 17:50:38 +0100
Subject: [PATCH] Deprecate (chunk|link)_inode_offset

These can be determined at run-time. This also magically fixes
links on file systems created with pre 0.2.3 releases.
---
 src/dwarfs/metadata_v2.cpp | 29 ++++++++++++++++++++++++-----
 src/dwarfs/scanner.cpp     |  2 --
 thrift/metadata.thrift     | 19 ++++++++++++-------
 3 files changed, 36 insertions(+), 14 deletions(-)

diff --git a/src/dwarfs/metadata_v2.cpp b/src/dwarfs/metadata_v2.cpp
index 717d0438..edc8ddbc 100644
--- a/src/dwarfs/metadata_v2.cpp
+++ b/src/dwarfs/metadata_v2.cpp
@@ -101,9 +101,14 @@ class metadata_ : public metadata_v2::impl {
       : data_(data)
       , meta_(map_frozen<thrift::metadata::metadata>(schema, data_))
       , root_(meta_.entries()[meta_.entry_index()[0]], &meta_)
+      , log_(lgr)
       , inode_offset_(inode_offset)
-      , chunk_index_offset_(meta_.chunk_index_offset())
-      , log_(lgr) {}
+      , chunk_index_offset_(
+            find_index_offset(meta_.entry_index().size(),
+                              [](uint16_t mode) { return S_ISREG(mode); }))
+      , link_index_offset_(find_index_offset(
+            chunk_index_offset_, [](uint16_t mode) { return S_ISLNK(mode); })) {
+  }
 
   void dump(std::ostream& os, int detail_level,
             std::function<void(const std::string&, uint32_t)> const& icb)
@@ -154,6 +159,19 @@ class metadata_ : public metadata_v2::impl {
     return make_entry_view(meta_.entry_index()[inode]);
   }
 
+  template <typename Func>
+  size_t find_index_offset(size_t last, Func&& func) const {
+    auto range = boost::irange(size_t(0), last);
+
+    auto it =
+        std::upper_bound(range.begin(), range.end(), 0, [&](int, auto inode) {
+          auto e = make_entry_view_from_inode(inode);
+          return bool(func(e.mode()));
+        });
+
+    return *it;
+  }
+
   directory_view make_directory_view(entry_view entry) const {
     return directory_view(entry, &meta_);
   }
@@ -205,15 +223,16 @@ class metadata_ : public metadata_v2::impl {
 
   std::string_view link_value(entry_view entry) const {
     return meta_
-        .links()[meta_.link_index()[entry.inode() - meta_.link_index_offset()]];
+        .links()[meta_.link_index()[entry.inode() - link_index_offset_]];
   }
 
   folly::ByteRange data_;
   MappedFrozen<thrift::metadata::metadata> meta_;
   entry_view root_;
+  log_proxy<LoggerPolicy> log_;
   const int inode_offset_;
   const int chunk_index_offset_;
-  log_proxy<LoggerPolicy> log_;
+  const int link_index_offset_;
 };
 
 template <typename LoggerPolicy>
@@ -534,7 +553,7 @@ template <typename LoggerPolicy>
 std::optional<chunk_range>
 metadata_<LoggerPolicy>::get_chunks(int inode) const {
   std::optional<chunk_range> rv;
-  inode -= inode_offset_ + meta_.chunk_index_offset();
+  inode -= inode_offset_ + chunk_index_offset_;
   if (inode >= 0 &&
       inode < (static_cast<int>(meta_.chunk_index().size()) - 1)) {
     uint32_t begin = meta_.chunk_index()[inode];
diff --git a/src/dwarfs/scanner.cpp b/src/dwarfs/scanner.cpp
index 9aa5bdd6..203181bb 100644
--- a/src/dwarfs/scanner.cpp
+++ b/src/dwarfs/scanner.cpp
@@ -532,8 +532,6 @@ void scanner_<LoggerPolicy>::scan(filesystem_writer& fsw,
   mv2.names = ge_data.get_names();
   mv2.links = ge_data.get_links();
   mv2.timestamp_base = ge_data.get_timestamp_base();
-  mv2.chunk_index_offset = first_file_inode;
-  mv2.link_index_offset = first_link_inode;
   mv2.block_size = UINT32_C(1) << cfg_.block_size_bits;
   mv2.total_fs_size = prog.original_size;
 
diff --git a/thrift/metadata.thrift b/thrift/metadata.thrift
index 41641186..cc072c43 100644
--- a/thrift/metadata.thrift
+++ b/thrift/metadata.thrift
@@ -63,9 +63,9 @@ struct entry {
     *
     * - For directories, the inode can be used as an index into
     *   metadata.directories.
-    * - For links, (inode - metadata.link_index_offset) can be
+    * - For links, (inode - link_index_offset) can be
     *   used as an index into metadata.links.
-    * - For files, (inode - metadata.chunk_index_offset) can be
+    * - For files, (inode - chunk_index_offset) can be
     *   used as in index into metadata.chunk_index.
     */
    3: required UInt32 inode,
@@ -138,11 +138,16 @@ struct metadata {
    // timestamp base for all entry timestamps
   12: required UInt64          timestamp_base,
 
-   // inode offset for lookups into chunk_index
-  13: required UInt32          chunk_index_offset;
-
-   // inode offset for lookups into link_index
-  14: required UInt32          link_index_offset;
+  /************************ DEPRECATED **********************
+   *
+   * These are redundant and can be determined at run-time
+   * with a simple binary search. Compatibility is not
+   * affected.
+   *
+   *   13: required UInt32          chunk_index_offset;
+   *   14: required UInt32          link_index_offset;
+   *
+   *********************************************************/
 
    // block size
   15: required UInt32          block_size;