- fix classpath use

Author: chrismattmann

README.md

@@ -126,20 +126,29 @@ running and omit printing the check messages.
 Changing the Tika Classpath
 ---------------------------
 You can update the classpath that Tika server uses by
-setting
-```python
-import tika
-tika.TikaServerClasspath = '/tmp/keys'
-```
+setting the classpath as a set of ':' delimited strings.
+For example if you want to get Tika-Python working with 
+[GeoTopicParsing](http://wiki.apache.org/tika/GeoTopicParser),
+you can do this, replace paths below with your own paths, as
+identified [here](http://wiki.apache.org/tika/GeoTopicParser) 
+and make sure that you have done this:
 
-Then, kill Tika server:
+kill Tika server (if already running):
 
 ```bash
 ps aux | grep java | grep Tika
 kill -9 PID
 ```
 
-Then, try Tika, and you should see the new classpath.
+```python
+import tika.tika
+import os
+from tika import parser
+home = os.getenv('HOME')
+tika.tika.TikaServerClasspath = home + '/git/geotopicparser-utils/mime:'+home+'/git/geotopicparser-utils/models/polar'
+parsed = parser.from_file(home + '/git/geotopicparser-utils/geotopics/polar.geot')
+print parsed["metadata"]```
+
 
 New Command Line Client Tool
 ============================

tika/tika.py

@@ -125,8 +125,7 @@
     'TIKA_TRANSLATOR',
     "org.apache.tika.language.translate.Lingo24Translator")
 TikaClientOnly = os.getenv('TIKA_CLIENT_ONLY', False)
-
-TikaServerClasspath = os.getenv('TIKA_SERVER_CLASSPATH', "")
+TikaServerClasspath = os.getenv('TIKA_SERVER_CLASSPATH', '')
 
 Verbose = 0
 EncodeUtf8 = 0
@@ -311,12 +310,15 @@ def getConfig(option, serverEndpoint=ServerEndpoint, verbose=Verbose, tikaServer
 
 
 def callServer(verb, serverEndpoint, service, data, headers, verbose=Verbose, tikaServerJar=TikaServerJar, 
-               httpVerbs={'get': requests.get, 'put': requests.put, 'post': requests.post}, classpath = TikaServerClasspath):
+               httpVerbs={'get': requests.get, 'put': requests.put, 'post': requests.post},classpath=None):
     """Call the Tika Server, do some error checking, and return the response."""
 
     parsedUrl = urlparse(serverEndpoint) 
     serverHost = parsedUrl.hostname
     port = parsedUrl.port
+    if classpath is None:
+        classpath = TikaServerClasspath
+    
     global TikaClientOnly
     if not TikaClientOnly:
         serverEndpoint = checkTikaServer(serverHost, port, tikaServerJar, classpath)
@@ -343,8 +345,10 @@ def callServer(verb, serverEndpoint, service, data, headers, verbose=Verbose, ti
     return (resp.status_code, resp.text)
 
 
-def checkTikaServer(serverHost=ServerHost, port = Port, tikaServerJar=TikaServerJar, classpath = TikaServerClasspath):
+def checkTikaServer(serverHost=ServerHost, port = Port, tikaServerJar=TikaServerJar,classpath=None):
     """Check that tika-server is running.  If not, download JAR file and start it up."""
+    if classpath is None:
+        classpath = TikaServerClasspath
     urlp = urlparse(tikaServerJar)
     serverEndpoint = 'http://%s:%s' % (serverHost, port)
     jarPath = os.path.join(TikaJarPath, 'tika-server.jar')
@@ -374,14 +378,17 @@ def checkJarSig(tikaServerJar, jarPath):
             return existingContents == m.hexdigest()
 
 
-def startServer(tikaServerJar, serverHost = ServerHost, port = Port, classpath = TikaServerClasspath):
+def startServer(tikaServerJar, serverHost = ServerHost, port = Port, classpath=None):
+    if classpath is None:
+        classpath = TikaServerClasspath
+    
     host = "localhost"
     if Windows:
         host = "0.0.0.0"
 
-    if classpath :
+    if classpath:
         classpath += ":" + tikaServerJar
-    else :
+    else:
         classpath = tikaServerJar
 
     cmd = 'java -cp %s org.apache.tika.server.TikaServerCli --port %i --host %s &' % (classpath, port, host)