Get uncompression to work and fix a few other issues

src/common/rclconfig.cpp

@@ -1369,7 +1369,21 @@ bool RclConfig::getUncompressor(const string &mtype, vector<string>& cmd) const
     if (stringlowercmp("uncompress", *it++)) 
 	return false;
     cmd.clear();
-    cmd.push_back(findFilter(*it++));
+    cmd.push_back(findFilter(*it));
+
+    // Special-case python and perl on windows: we need to also locate the
+    // first argument which is the script name "python somescript.py". 
+    // On Unix, thanks to #!, we usually just run "somescript.py", but need
+    // the same change if we ever want to use the same cmdling as windows
+    if (!stringlowercmp("python", *it) || !stringlowercmp("perl", *it)) {
+        it++;
+        if (tokens.size() < 3) {
+            LOGERR(("getUncpressor: python/perl cmd: no script?. [%s]\n", mtype.c_str()));
+        } else {
+            *it = findFilter(*it);
+        }
+    }
+    
     cmd.insert(cmd.end(), it, tokens.end());
     return true;
 }

src/filters/rclmpdf.py

@@ -70,6 +70,7 @@ class PDFExtractor:
         self.pdftk = ""
         self.em = em
         self.attextractdone = False
+        self.attachlist = []
         
     # Extract all attachments if any into temporary directory
     def extractAttach(self):
@@ -79,7 +80,8 @@ class PDFExtractor:
 
         global tmpdir
         if not tmpdir or not self.pdftk:
-            return False
+            # no big deal
+            return True
 
         try:
             vacuumdir(tmpdir)
@@ -89,7 +91,8 @@ class PDFExtractor:
             return True
         except Exception, e:
             self.em.rclog("extractAttach: failed: %s" % e)
-            return False
+            # Return true anyway, pdf attachments are no big deal
+            return True
 
     def extractone(self, ipath):
         #self.em.rclog("extractone: [%s]" % ipath)
@@ -186,6 +189,8 @@ class PDFExtractor:
 
         if not self.pdftotext:
             self.pdftotext = rclexecm.which("pdftotext")
+            if not self.pdftotext:
+                self.pdftotext = rclexecm.which("poppler/pdftotext")
             if not self.pdftotext:
                 print("RECFILTERROR HELPERNOTFOUND pdftotext")
                 sys.exit(1);
@@ -209,7 +214,8 @@ class PDFExtractor:
                 # eofnext error instead of waiting for actual eof,
                 # which avoids a bug in recollindex up to 1.20
                 self.extractAttach()
-
+        else:
+            self.attextractdone = True
         return True
 
     def getipath(self, params):
@@ -218,8 +224,8 @@ class PDFExtractor:
         return (ok, data, ipath, eof)
         
     def getnext(self, params):
+        # self.em.rclog("getnext: current %d" % self.currentindex)
         if self.currentindex == -1:
-            #self.em.rclog("getnext: current -1")
             self.currentindex = 0
             return self._selfdoc()
         else:

src/filters/rcluncomp.py

@@ -9,30 +9,44 @@ import platform
 import subprocess
 import glob
 
-sysplat = platform.system()
+ftrace = sys.stderr
+#ftrace = open("C:/Users/Bill/log-uncomp.txt", "w")
 
+sysplat = platform.system()
 if sysplat != "Windows":
-    print("rcluncomp.py: only for Windows", file = sys.stderr)
+    print("rcluncomp.py: only for Windows", file = ftrace)
+    sys.exit(1)
+
+try:
+    import msvcrt
+    msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
+except Exception as err:
+    print("setmode binary failed: %s" % str(err), file = ftrace)
 
 sevenz = rclexecm.which("7z")
 if not sevenz:
     print("rcluncomp.py: can't find 7z exe. Maybe set recollhelperpath " \
-          "in recoll.conf ?", file=sys.stderr)
+          "in recoll.conf ?", file=ftrace)
-    sys.exit(1)
+    sys.exit(2)
-#print("rcluncomp.py: 7z is %s" % sevenz, file = sys.stderr)
 
 # Params: uncompression program, input file name, temp directory.
 # We ignore the uncomp program, and always use 7z on Windows
 
 infile = sys.argv[2]
 outdir = sys.argv[3]
+# print("rcluncomp.py infile [%s], outdir [%s]" % (infile, outdir), file = ftrace)
 
 # There is apparently no way to suppress 7z output. Hopefully the
 # possible deadlock described by the subprocess module doc can't occur
 # here because there is little data printed. AFAIK nothing goes to stderr anyway
-subprocess.check_output([sevenz, "e", "-bd", "-y", "-o" + outdir, infile],
+try:
-                        stderr = subprocess.PIPE)
+    cmd = [sevenz, "e", "-bd", "-y", "-o" + outdir, infile]
+    subprocess.check_output(cmd, stderr = subprocess.PIPE)
+    outputname = glob.glob(os.path.join(outdir, "*"))
+    # There should be only one file in there..
+    print(outputname[0])
+except Exception as err:
+    print("%s" % (str(err),), file = ftrace)
+    sys.exit(4)
 
-outputname = glob.glob(os.path.join(outdir, "*"))
+sys.exit(0)
-# There should be only one file in there..
-print(outputname[0])

src/utils/pathut.cpp

@@ -145,6 +145,7 @@ string path_wingettempfilename(TCHAR *pref)
     // Windows will have created a temp file, we delete it.
     string filename = path_tchartoutf8(buf);
     unlink(filename.c_str());
+    path_slashize(filename);
     return filename;
 }
 #endif

src/windows/mimeconf

@@ -149,8 +149,8 @@ text/x-python = exec python rclpython
 text/x-shellscript = internal text/plain
 text/x-srt = internal text/plain
 
-application/xml = execm python rclxml
+application/xml = execm python rclxml.py
-text/xml = execm python rclxml
+text/xml = execm python rclxml.py
 # Using these instead of the two above would index all parameter and tag
 # names, attribute values etc, instead of just the text content.
 #application/xml = internal text/plain

src/windows/mkinstdir.sh

@@ -62,9 +62,9 @@ RCLS=$RCLW/build-rclstartw-${QTA}-${qtsdir}/${qtsdir}/rclstartw.exe
 
 
 # Needed for a VS build (which we did not ever complete because of
-# missing Qt VS2015 support). Needed for unrtf
+# missing Qt VS2015 support).
-CONFIGURATION=Release
+#CONFIGURATION=Release
-PLATFORM=Win32
+#PLATFORM=Win32
 
 ################
 # Script:
@@ -82,12 +82,14 @@ copyqt()
 {
     cd $DESTDIR
     $QTBIN/windeployqt recoll.exe
+    chkcp $QTBIN/libwinpthread-1.dll $DESTDIR
 }
 
 copyxapian()
 {
     chkcp $LIBXAPIAN $DESTDIR
 }
+
 copyzlib()
 {
     chkcp $ZLIB/zlib1.dll $DESTDIR
@@ -133,7 +135,8 @@ copyantiword()
 
 copyunrtf()
 {
-    bindir=$UNRTF/Windows/$PLATFORM/$CONFIGURATION
+#    bindir=$UNRTF/Windows/$PLATFORM/$CONFIGURATION
+     bindir=$UNRTF/Windows/
 
     test -d $FILTERS/Share || mkdir -p $FILTERS/Share || exit 1
     chkcp  $bindir/unrtf.exe               $FILTERS
@@ -158,8 +161,7 @@ copyepub()
 copypyexiv2()
 {
     cp -rp $PYEXIV2/pyexiv2 $FILTERS
-    # Check
+    chkcp $PYEXIV2/libexiv2python.pyd $FILTERS/
-    chkcp $PYEXIV2/pyexiv2/exif.py $FILTERS/pyexiv2
 }
 
 copyxslt()
@@ -170,10 +172,12 @@ copyxslt()
 
 copypoppler()
 {
+    test -d $FILTERS/poppler || mkdir $FILTERS/poppler || \
+        fatal cant create poppler dir
     for f in pdftotext.exe libpoppler.dll freetype6.dll jpeg62.dll \
              libpng16-16.dll zlib1.dll libtiff3.dll \
              libgcc_s_dw2-1.dll libstdc++-6.dll; do
-        chkcp $POPPLER/bin/$f $FILTERS/
+        chkcp $POPPLER/bin/$f $FILTERS/poppler
     done
 }
 

src/windows/qmkrecoll/recollindex.pro

@@ -32,6 +32,5 @@ windows {
     C:/recoll/src/windows/build-librecoll-Desktop_Qt_5_5_0_MinGW_32bit-Debug/debug/librecoll.dll \
     -lshlwapi -lpsapi -lkernel32
 
-  INCLUDEPATH += ../../windows \
+  INCLUDEPATH += ../../windows
-            C:/temp/xapian-core-1.2.8/include
 }

src/windows/qmkrecoll/recollq.pro

@@ -30,6 +30,5 @@ windows {
     C:/recoll/src/windows/build-librecoll-Desktop_Qt_5_5_0_MinGW_32bit-Debug/debug/librecoll.dll \
     -lshlwapi -lpsapi -lkernel32
 
-  INCLUDEPATH += ../../windows \
+  INCLUDEPATH += ../../windows
-            C:/recolldeps/xapian/xapian-core-1.2.8/include
 }