diff --git a/src/filters/rclexec1.py b/src/filters/rclexec1.py
index 13970048..ffa68c53 100644
--- a/src/filters/rclexec1.py
+++ b/src/filters/rclexec1.py
@@ -34,12 +34,14 @@ import rclexecm
# the object which we receive as a parameter, which in turn is defined
# in the actual executable filter (e.g. rcldoc.py)
class Executor:
+ opt_ignxval = 1
+
def __init__(self, em, flt):
self.em = em
self.flt = flt
self.currentindex = 0
- def runCmd(self, cmd, filename, postproc):
+ def runCmd(self, cmd, filename, postproc, opt):
''' Substitute parameters and execute command, process output
with the specific postprocessor and return the complete text.
We expect cmd as a list of command name + arguments'''
@@ -60,7 +62,7 @@ class Executor:
postproc.takeLine(line.strip())
proc.wait()
- if proc.returncode:
+ if (opt & self.opt_ignxval) == 0 and proc.returncode:
self.em.rclog("extractone: [%s] returncode %d" % \
(filename, proc.returncode))
return False, postproc.wrapData()
@@ -78,9 +80,12 @@ class Executor:
fn = params["filename:"]
while True:
- cmd, postproc = self.flt.getCmd(fn)
+ cmdseq = self.flt.getCmd(fn)
+ cmd = cmdseq[0]
+ postproc = cmdseq[1]
+ opt = cmdseq[2] if len(cmdseq) == 3 else 0
if cmd:
- ok, data = self.runCmd(cmd, fn, postproc)
+ ok, data = self.runCmd(cmd, fn, postproc, opt)
if ok:
break
else:
diff --git a/src/filters/rclppt.py b/src/filters/rclppt.py
new file mode 100755
index 00000000..0d2bf661
--- /dev/null
+++ b/src/filters/rclppt.py
@@ -0,0 +1,58 @@
+#!/usr/bin/env python
+
+import rclexecm
+import rclexec1
+import re
+import sys
+import os
+
+# Processing the output from unrtf
+class PPTProcessData:
+ def __init__(self, em):
+ self.em = em
+ self.out = ""
+ self.gotdata = 0
+
+ # Some versions of unrtf put out a garbled charset line.
+ # Apart from this, we pass the data untouched.
+ def takeLine(self, line):
+ if not self.gotdata:
+ self.out += '''''' + \
+ '''''' + \
+ ''''''
+ self.gotdata = True
+ self.out += self.em.htmlescape(line)
+
+ def wrapData(self):
+ return self.out + '''
'''
+
+class PPTFilter:
+ def __init__(self, em):
+ self.em = em
+ self.ntry = 0
+
+ def reset(self):
+ self.ntry = 0
+ pass
+
+ def getCmd(self, fn):
+ if self.ntry:
+ return ([], None)
+ self.ntry = 1
+ cmd = rclexecm.which("ppt-dump.py")
+ if cmd:
+ # ppt-dump.py often exits 1 with valid data. Ignore exit value
+ return ([cmd, "--no-struct-output", "--dump-text"],
+ PPTProcessData(self.em), rclexec1.Executor.opt_ignxval)
+ else:
+ return ([], None)
+
+if __name__ == '__main__':
+ if not rclexecm.which("ppt-dump.py"):
+ print("RECFILTERROR HELPERNOTFOUND ppt-dump.py")
+ sys.exit(1)
+ proto = rclexecm.RclExecM()
+ filter = PPTFilter(proto)
+ extract = rclexec1.Executor(proto, filter)
+ rclexecm.main(proto, extract)
diff --git a/src/filters/rclrtf.py b/src/filters/rclrtf.py
index 81af1eab..c7031030 100755
--- a/src/filters/rclrtf.py
+++ b/src/filters/rclrtf.py
@@ -35,16 +35,20 @@ class RTFProcessData:
class RTFFilter:
def __init__(self, em):
self.em = em
+ self.ntry = 0
def reset(self):
- pass
+ self.ntry = 0
def getCmd(self, fn):
+ if self.ntry:
+ return ([], None)
+ self.ntry = 1
cmd = rclexecm.which("unrtf")
if cmd:
return ([cmd, "--nopict", "--html"], RTFProcessData(self.em))
else:
- return ([],None)
+ return ([], None)
if __name__ == '__main__':
if not rclexecm.which("unrtf"):
diff --git a/src/filters/rclxls.py b/src/filters/rclxls.py
new file mode 100755
index 00000000..806efb34
--- /dev/null
+++ b/src/filters/rclxls.py
@@ -0,0 +1,64 @@
+#!/usr/bin/env python
+
+import rclexecm
+import rclexec1
+import xlsxmltocsv
+import re
+import sys
+import os
+import xml.sax
+
+# Processing the output from unrtf
+class XLSProcessData:
+ def __init__(self, em):
+ self.em = em
+ self.out = ""
+ self.gotdata = 0
+ self.xmldata = ""
+
+ # Some versions of unrtf put out a garbled charset line.
+ # Apart from this, we pass the data untouched.
+ def takeLine(self, line):
+ if not self.gotdata:
+ self.out += '''''' + \
+ '''''' + \
+ ''''''
+ self.gotdata = True
+ self.xmldata += line
+
+ def wrapData(self):
+ handler = xlsxmltocsv.XlsXmlHandler()
+ data = xml.sax.parseString(self.xmldata, handler)
+ self.out += self.em.htmlescape(handler.output)
+ return self.out + '''
'''
+
+class XLSFilter:
+ def __init__(self, em):
+ self.em = em
+ self.ntry = 0
+
+ def reset(self):
+ self.ntry = 0
+ pass
+
+ def getCmd(self, fn):
+ if self.ntry:
+ return ([], None)
+ self.ntry = 1
+ cmd = rclexecm.which("xls-dump.py")
+ if cmd:
+ # xls-dump.py often exits 1 with valid data. Ignore exit value
+ return ([cmd, "--dump-mode=canonical-xml", "--utf-8", "--catch"],
+ XLSProcessData(self.em), rclexec1.Executor.opt_ignxval)
+ else:
+ return ([], None)
+
+if __name__ == '__main__':
+ if not rclexecm.which("ppt-dump.py"):
+ print("RECFILTERROR HELPERNOTFOUND ppt-dump.py")
+ sys.exit(1)
+ proto = rclexecm.RclExecM()
+ filter = XLSFilter(proto)
+ extract = rclexec1.Executor(proto, filter)
+ rclexecm.main(proto, extract)
diff --git a/src/filters/xlsxmltocsv.py b/src/filters/xlsxmltocsv.py
index f8a6d654..72850d3a 100755
--- a/src/filters/xlsxmltocsv.py
+++ b/src/filters/xlsxmltocsv.py
@@ -15,10 +15,13 @@ else:
dquote = '"'
class XlsXmlHandler(xml.sax.handler.ContentHandler):
+ def __init__(self):
+ self.output = ""
+
def startElement(self, name, attrs):
if name == "worksheet":
if "name" in attrs:
- print("%s" % attrs["name"].encode("UTF-8"))
+ self.output += "%s\n" % attrs["name"].encode("UTF-8")
elif name == "row":
self.cells = dict()
elif name == "label-cell" or name == "number-cell":
@@ -30,7 +33,7 @@ class XlsXmlHandler(xml.sax.handler.ContentHandler):
self.cells[int(attrs["col"])] = value
else:
#??
- sys.stdout.write("%s%s"%(value.encode("UTF-8"),sepstring))
+ self.output += "%s%s" % (value.encode("UTF-8"), sepstring)
elif name == "formula-cell":
if "formula-result" in attrs and "col" in attrs:
self.cells[int(attrs["col"])] = \
@@ -40,17 +43,21 @@ class XlsXmlHandler(xml.sax.handler.ContentHandler):
if name == "row":
curidx = 0
for idx, value in self.cells.iteritems():
- sys.stdout.write(sepstring * (idx - curidx))
- sys.stdout.write('%s%s%s' % (dquote, value, dquote))
+ self.output += sepstring * (idx - curidx)
+ self.output += "%s%s%s" % (dquote, value, dquote)
curidx = idx
- sys.stdout.write("\n")
+ self.output += "\n"
elif name == "worksheet":
- print("")
+ self.output += "\n"
-try:
- xml.sax.parse(sys.stdin, XlsXmlHandler())
-except BaseException as err:
- error("xml-parse: %s\n" % (str(sys.exc_info()[:2]),))
- sys.exit(1)
-sys.exit(0)
+if __name__ == '__main__':
+ try:
+ handler = XlsXmlHandler()
+ xml.sax.parse(sys.stdin, handler)
+ print(handler.output)
+ except BaseException as err:
+ error("xml-parse: %s\n" % (str(sys.exc_info()[:2]),))
+ sys.exit(1)
+
+ sys.exit(0)