From d4edbbaedbf78bda7c835f33e63f31631666393c Mon Sep 17 00:00:00 2001
From: Jean-Francois Dockes <jfd@recoll.org>
Date: Thu, 11 Oct 2012 15:35:15 +0200
Subject: [PATCH] rclepub: use elt ids instead of hrefs + debug traces

---
 src/filters/rclepub         | 20 ++++++++++++--------
 src/internfile/mh_execm.cpp |  5 ++++-
 2 files changed, 16 insertions(+), 9 deletions(-)

diff --git a/src/filters/rclepub b/src/filters/rclepub
index d1b9b39c..802abb3b 100755
--- a/src/filters/rclepub
+++ b/src/filters/rclepub
@@ -25,7 +25,7 @@ class rclEPUB:
         self.em = em
         self.em.setmimetype(rclepub_html_mtype)
 
-    def extractone(self, path):
+    def extractone(self, id):
         """Extract one path-named internal file from the EPUB file"""
 
         #self.em.rclog("extractone: [%s]"%(path))
@@ -34,13 +34,17 @@ class rclEPUB:
             iseof = rclexecm.RclExecM.eofnext
 
         try:
-            doc = self.book.read(path)
+            item = self.book.get_item(id)
+            if item is None:
+                raise Exception("Item not found for id %s" % (id,))
+            doc = self.book.read_item(item)
             doc = re.sub('''</[hH][eE][aA][dD]''',
                          '''<meta name="rclaptg" content="epub"></head>''', doc)
-            return (True, doc, path, iseof)
+            self.em.setmimetype(rclepub_html_mtype)
+            return (True, doc, id, iseof)
         except Exception, err:
-            self.em.rclog("openfile: failed: [%s]" % err)
-            return (False, "", path, iseof)
+            self.em.rclog("extractone: failed: [%s]" % err)
+            return (False, "", id, iseof)
 
     def openfile(self, params):
         """Open the EPUB file"""
@@ -49,12 +53,12 @@ class rclEPUB:
         try:
             self.book = epub.open(params["filename:"])
         except Exception, err:
-            self.em.rclog("openfile: failed: [%s]" % err)
+            self.em.rclog("openfile: epub.open failed: [%s]" % err)
             return False
         for id, item in self.book.opf.manifest.iteritems():
-#            print item.__dict__
+            # print item.__dict__
             if item.media_type == u'application/xhtml+xml':
-                self.contents.append(item.href)
+                self.contents.append(id)
         return True
         
 
diff --git a/src/internfile/mh_execm.cpp b/src/internfile/mh_execm.cpp
index 10109eae..befd7c34 100644
--- a/src/internfile/mh_execm.cpp
+++ b/src/internfile/mh_execm.cpp
@@ -120,7 +120,7 @@ bool MimeHandlerExecMultiple::readDataElement(string& name, string &data)
                 ibuf.c_str()));
         return false;
     }
-    LOGDEB(("MHExecMultiple: got name [%s] len: %d\n", name.c_str(), len));
+
     if (len / 1024 > m_maxmemberkb) {
         LOGERR(("MHExecMultiple: data len > maxmemberkb\n"));
         return false;
@@ -143,6 +143,9 @@ bool MimeHandlerExecMultiple::readDataElement(string& name, string &data)
                 len, datap->length()));
         return false;
     }
+    LOGDEB1(("MHExecMe:rdDtElt got: name [%s] len %d value [%s]\n", 
+	     name.c_str(), len, datap->size() > 100 ? 
+	     (datap->substr(0, 100) + " ...").c_str() : datap->c_str()));
     return true;
 }