From f6f4d8426af5befabbb7a3f2ad5fbdae0e5bef5f Mon Sep 17 00:00:00 2001
From: Jean-Francois Dockes <jf@dockes.org>
Date: Fri, 28 Dec 2018 10:32:01 +0100
Subject: [PATCH] comments + really compute md5 on uncompressed data

---
 src/internfile/mh_xslt.cpp |  6 ++++++
 src/utils/readfile.cpp     | 19 +++++++++----------
 src/utils/readfile.h       |  7 +++----
 3 files changed, 18 insertions(+), 14 deletions(-)

diff --git a/src/internfile/mh_xslt.cpp b/src/internfile/mh_xslt.cpp
index 8afbc83a..56d6429a 100644
--- a/src/internfile/mh_xslt.cpp
+++ b/src/internfile/mh_xslt.cpp
@@ -32,6 +32,12 @@
 
 using namespace std;
 
+// Do we need this? It would need to be called from recollinit
+// Call once, not reentrant
+// xmlInitParser();
+// LIBXML_TEST_VERSION;
+// Probably not:    xmlCleanupParser();
+        
 
 class FileScanXML : public FileScanDo {
 public:
diff --git a/src/utils/readfile.cpp b/src/utils/readfile.cpp
index ae2edf16..d51da8b2 100644
--- a/src/utils/readfile.cpp
+++ b/src/utils/readfile.cpp
@@ -511,14 +511,6 @@ bool file_scan(const string& fn, FileScanDo* doer, int64_t startoffs,
     FileScanSourceFile source(doer, fn, startoffs, cnttoread, reason);
     FileScanUpstream *up = &source;
 
-    // We compute the MD5 on the uncompressed data, so insert this
-    // right at the source.
-    string digest;
-    FileScanMd5 md5filter(digest);
-    if (md5p) {
-        md5filter.insertAtSink(doer, up);
-        up = &md5filter;
-    }
     
 #if defined(READFILE_ENABLE_ZLIB)
     GzFilter gzfilter;
@@ -528,6 +520,15 @@ bool file_scan(const string& fn, FileScanDo* doer, int64_t startoffs,
     }
 #endif
 
+    // We compute the MD5 on the uncompressed data, so insert this
+    // right at the source (after the decompressor).
+    string digest;
+    FileScanMd5 md5filter(digest);
+    if (md5p) {
+        md5filter.insertAtSink(doer, up);
+        up = &md5filter;
+    }
+    
     bool ret = source.scan();
 
     if (md5p) {
@@ -572,8 +573,6 @@ bool string_scan(const char *data, size_t cnt, FileScanDo* doer,
     FileScanSourceBuffer source(doer, data, cnt, reason);
     FileScanUpstream *up = &source;
 
-    // We compute the MD5 on the uncompressed data, so insert this
-    // right at the source.
     string digest;
     FileScanMd5 md5filter(digest);
     if (md5p) {
diff --git a/src/utils/readfile.h b/src/utils/readfile.h
index 64323965..854fcec1 100644
--- a/src/utils/readfile.h
+++ b/src/utils/readfile.h
@@ -65,14 +65,13 @@ public:
 bool file_scan(const std::string& fn, FileScanDo* doer, int64_t startoffs,
                int64_t cnttoread, std::string *reason, std::string *md5p);
 
-/** Same as file_scan, from a memory buffer */
-bool string_scan(const char *data, size_t cnt, FileScanDo* doer, 
-                 std::string *reason, std::string *md5p);
-
 /** Same as above, not offset/cnt/md5 */
 bool file_scan(const std::string& filename, FileScanDo* doer,
                std::string *reason);
 
+/** Same as file_scan, from a memory buffer. No libz processing */
+bool string_scan(const char *data, size_t cnt, FileScanDo* doer, 
+                 std::string *reason, std::string *md5p);
 
 #if defined(READFILE_ENABLE_MINIZ)
 /* Process a zip archive member */