From b9e672abda7f680ae66195b64ef589a1136a75fe Mon Sep 17 00:00:00 2001
From: Jean-Francois Dockes <jfd@recoll.org>
Date: Mon, 11 Jul 2016 18:13:39 +0200
Subject: [PATCH] Allow execm input handlers to set arbitrary data fields

---
 src/filters/rclexecm.py     | 40 +++++++++++++++++++++++++------------
 src/filters/rclzip          | 11 ++++++++++
 src/internfile/mh_execm.cpp | 10 +++++++++-
 3 files changed, 47 insertions(+), 14 deletions(-)

diff --git a/src/filters/rclexecm.py b/src/filters/rclexecm.py
index 26c9764e..4bb86390 100644
--- a/src/filters/rclexecm.py
+++ b/src/filters/rclexecm.py
@@ -66,6 +66,8 @@ class RclExecM:
             self.myname = "???"
         self.mimetype = b""
 
+        self.fields = {}
+        
         if os.environ.get("RECOLL_FILTER_MAXMEMBERKB"):
             self.maxmembersize = \
             int(os.environ.get("RECOLL_FILTER_MAXMEMBERKB"))
@@ -125,6 +127,9 @@ class RclExecM:
     def setmimetype(self, mt):
         self.mimetype = makebytes(mt)
 
+    def setfield(self, nm, value):
+        self.fields[nm] = value
+
     # Read single parameter from process input: line with param name and size
     # followed by data. The param name is returned as str/unicode, the data
     # as bytes
@@ -160,35 +165,44 @@ class RclExecM:
         return (paramname, paramdata)
 
     if PY3:
-        def senditem(self, nm, len, data):
-            sys.stdout.buffer.write(makebytes("%s: %d\n" % (nm, len)))
-            self.breakwrite(sys.stdout.buffer, makebytes(data))
+        def senditem(self, nm, data):
+            data = makebytes(data)
+            l = len(data)
+            sys.stdout.buffer.write(makebytes("%s: %d\n" % (nm, l)))
+            self.breakwrite(sys.stdout.buffer, data)
     else:
-        def senditem(self, nm, len, data):
-            sys.stdout.write(makebytes("%s: %d\n" % (nm, len)))
-            self.breakwrite(sys.stdout, makebytes(data))
+        def senditem(self, nm, data):
+            data = makebytes(data)
+            l = len(data)
+            sys.stdout.write(makebytes("%s: %d\n" % (nm, l)))
+            self.breakwrite(sys.stdout, data)
         
     # Send answer: document, ipath, possible eof.
     def answer(self, docdata, ipath, iseof = noteof, iserror = noerror):
 
         if iserror != RclExecM.fileerror and iseof != RclExecM.eofnow:
-            self.senditem("Document", len(docdata), docdata)
+            self.senditem("Document", docdata)
 
             if len(ipath):
-                self.senditem("Ipath", len(ipath), ipath)
+                self.senditem("Ipath", ipath)
 
             if len(self.mimetype):
-                self.senditem("Mimetype", len(self.mimetype), self.mimetype)
+                self.senditem("Mimetype", self.mimetype)
 
+            for nm,value in self.fields.iteritems():
+                #self.rclog("Senditem: [%s] -> [%s]" % (nm, value))
+                self.senditem("%s:"%nm, value)
+            self.fields = {}
+            
         # If we're at the end of the contents, say so
         if iseof == RclExecM.eofnow:
-            self.senditem("Eofnow", 0, b'')
+            self.senditem("Eofnow", b'')
         elif iseof == RclExecM.eofnext:
-            self.senditem("Eofnext", 0, b'')
+            self.senditem("Eofnext", b'')
         if iserror == RclExecM.subdocerror:
-            self.senditem("Subdocerror", 0, b'')
+            self.senditem("Subdocerror", b'')
         elif iserror == RclExecM.fileerror:
-            self.senditem("Fileerror", 0, b'')
+            self.senditem("Fileerror", b'')
   
         # End of message
         print()
diff --git a/src/filters/rclzip b/src/filters/rclzip
index 82974e54..2131bf1d 100755
--- a/src/filters/rclzip
+++ b/src/filters/rclzip
@@ -21,6 +21,7 @@
 from __future__ import print_function
 
 import os
+import posixpath
 import fnmatch
 import rclexecm
 from zipfile import ZipFile
@@ -89,6 +90,16 @@ class ZipExtractor:
                 #raise BadZipfile()
             else:
                 docdata = self.zip.read(ipath)
+            try:
+                # We are assuming here that the zip uses forward slash
+                # separators, which is not necessarily the case. At
+                # worse, we'll get a wrong or no file name, which is
+                # no big deal (the ipath is the important data
+                # element).
+                filename = posixpath.basename(ipath)
+                self.em.setfield("filename", filename)
+            except:
+                pass
             ok = True
         except Exception as err:
             self.em.rclog("extractone: failed: [%s]" % err)
diff --git a/src/internfile/mh_execm.cpp b/src/internfile/mh_execm.cpp
index 05ecdce0..2e294d7e 100644
--- a/src/internfile/mh_execm.cpp
+++ b/src/internfile/mh_execm.cpp
@@ -169,6 +169,8 @@ bool MimeHandlerExecMultiple::next_document()
         return false;
     }
 
+    m_metaData.clear();
+    
     // Send request to child process. This maybe the first/only
     // request for a given file, or a continuation request. We send an
     // empty file name in the latter case.
@@ -260,8 +262,14 @@ bool MimeHandlerExecMultiple::next_document()
         } else if (!stringlowercmp("mimetype:", name)) {
             mtype = data;
             LOGDEB(("MHExecMultiple: got mimetype [%s]\n", data.c_str()));
+        } else {
+            string nm = stringtolower((const string&)name);
+            trimstring(nm, ":");
+            LOGDEB(("MHExecMultiple: got [%s] -> [%s]\n", nm.c_str(),
+                    data.c_str()));
+            m_metaData[nm] += data;
         }
-        if (loop == 10) {
+        if (loop == 20) {
             // ?? 
             LOGERR(("MHExecMultiple: filter sent too many parameters\n"));
             return false;