commit 2cc3ea0265d275d8be5c5d7d76c49e948591b9ec
parent 8827e0b9239b23f96e7446cc21d99a146c99a11e
Author: Christian Grothoff <christian@grothoff.org>
Date: Fri, 6 May 2005 09:47:28 +0000
draft for py api
Diffstat:
6 files changed, 85 insertions(+), 82 deletions(-)
diff --git a/src/main/Extractor.py b/src/main/Extractor.py
@@ -0,0 +1,28 @@
+import _extractor
+
+class Extractor(object):
+ def __init__(self):
+ self.plugins = _extractor.EXTRACTOR_PY_loadDefaultLibraries()
+ def __del__(self):
+ extractor.EXTRACTOR_PY_removeAll(self.plugins)
+# def load(plugs):
+# self.plugins = _extractor.EXTRACTOR_PY_load(self.plugins, plugs)
+# return None
+# def unload(plugs):
+# self.plugins = _extractor.EXTRACTOR_PY_unload(self.plugins, plugs)
+# return None
+ def extract(self,filename):
+ return _extractor.EXTRACTOR_PY_extract(self.plugins, filename, Keyword)
+
+class Keyword(object):
+ def __init__(self,type,value):
+ self.type = type
+ self.value = value.decode("utf-8")
+ def __repr__(self):
+ return u"%s(%i,%s)" % (self.__class__.__name__,self.type,self.value)
+ def __str__(self):
+ return u"%s: %s" % (self.getType(), self.getValue())
+ def getType(self):
+ return _extractor.EXTRACTOR_PY_getKeywordTypeAsStringType(self.type).decode("utf-8")
+ def getValue(self):
+ return self.value
diff --git a/src/main/Makefile.am b/src/main/Makefile.am
@@ -34,6 +34,8 @@ libextractor_la_DEPENDENCIES = \
EXTRA_DIST = \
winproc.c \
libextractor_python.c \
+ Extractor.py \
+ libextractor_python_setup.py \
extract.py \
iconv.c
@@ -51,3 +53,9 @@ extract_SOURCES = \
getopt.c \
getopt.h \
getopt1.c
+
+
+all-local:
+ python libextractor_python_setup.py build
+install-exec-local:
+ python libextractor_python_setup.py --prefix=$(DESTDIR) install
diff --git a/src/main/extract.py b/src/main/extract.py
@@ -1,13 +1,9 @@
-import extractor
-
-def getKeywordTypeAsString(t):
- return extractor.EXTRACTOR_PY_getKeywordTypeAsString(t)
-
-class Extractor:
- def __init__(self):
- self.plugins = extractor.EXTRACTOR_PY_loadDefaultLibraries(self)
- def __del__(self):
- extractor.EXTRACTOR_PY_removeAll(self, self.plugins)
- def extract(filename):
- extractor.EXTRACTOR_PY_extract(self, self.plugins, filename)
+import Extractor
+import sys
+xtract = Extractor.Extractor()
+for arg in sys.argv:
+ print "Keywords from " + arg
+ keys = xtract.extract(arg);
+ for i in keys:
+ print i
diff --git a/src/main/extractor.i b/src/main/extractor.i
@@ -1,58 +0,0 @@
-/* libextractor interface for SWIG */
-/* extractor.i */
-%module extractor
-%{
-%}
-
-typedef struct EXTRACTOR_Keywords {
- char * keyword;
- EXTRACTOR_KeywordType keywordType;
- struct EXTRACTOR_Keywords * next;
-} EXTRACTOR_KeywordList;
-
-EXTRACTOR_ExtractorList * EXTRACTOR_loadDefaultLibraries();
-
-const char * EXTRACTOR_getKeywordTypeAsString(const EXTRACTOR_KeywordType type);
-
-EXTRACTOR_ExtractorList *
-EXTRACTOR_loadConfigLibraries(EXTRACTOR_ExtractorList * prev,
- const char * config);
-
-EXTRACTOR_ExtractorList *
-EXTRACTOR_addLibrary(EXTRACTOR_ExtractorList * prev,
- const char * library);
-
-EXTRACTOR_ExtractorList *
-EXTRACTOR_addLibraryLast(EXTRACTOR_ExtractorList * prev,
- const char * library);
-
-EXTRACTOR_ExtractorList *
-EXTRACTOR_removeLibrary(EXTRACTOR_ExtractorList * prev,
- const char * library);
-
-void EXTRACTOR_removeAll(EXTRACTOR_ExtractorList * libraries);
-
-EXTRACTOR_KeywordList *
-EXTRACTOR_getKeywords(EXTRACTOR_ExtractorList * extractor,
- const char * filename);
-
-EXTRACTOR_KeywordList *
-EXTRACTOR_removeDuplicateKeywords(EXTRACTOR_KeywordList * list,
- const unsigned int options);
-
-
-EXTRACTOR_KeywordList *
-EXTRACTOR_removeEmptyKeywords (EXTRACTOR_KeywordList * list);
-
-
-void EXTRACTOR_freeKeywords(EXTRACTOR_KeywordList * keywords);
-
-const char * EXTRACTOR_extractLast(const EXTRACTOR_KeywordType type,
- EXTRACTOR_KeywordList * keywords);
-
-const char * EXTRACTOR_extractLastByString (const char * type,
- EXTRACTOR_KeywordList * keywords);
-
-unsigned int EXTRACTOR_countKeywords(EXTRACTOR_KeywordList * keywords);
-
-
diff --git a/src/main/libextractor_python.c b/src/main/libextractor_python.c
@@ -18,8 +18,8 @@
Boston, MA 02111-1307, USA.
*/
-
-#include <python/Python.h>
+#include "extractor.h"
+#include <Python.h>
static PyObject * EXTRACTOR_PY_loadDefaultLibraries(PyObject * self,
PyObject * args) {
@@ -51,23 +51,33 @@ static PyObject * EXTRACTOR_PY_getKeywordTypeAsString(PyObject * self,
static PyObject * EXTRACTOR_PY_extract(PyObject * self,
PyObject * args) {
PyObject * py_exts;
+ PyObject * py_clzz;
+ PyObject * py_elem;
char * filename;
EXTRACTOR_ExtractorList * ex;
EXTRACTOR_KeywordList * keys;
EXTRACTOR_KeywordList * pos;
PyObject * ret;
- PyArg_ParseTuple(args, "Os", &py_exts, &filename);
+ PyArg_ParseTuple(args,
+ "OsO",
+ &py_exts,
+ &filename,
+ &py_clzz);
ex = PyCObject_AsVoidPtr(py_exts);
keys = EXTRACTOR_getKeywords(ex,
filename);
ret = PyList_New(0);
pos = keys;
while (pos != NULL) {
+ py_elem = PyObject_Call(py_clzz,
+ Py_BuildValue("(OO)",
+ PyInt_FromLong((long)pos->keywordType),
+ PyString_FromString(pos->keyword)),
+ NULL);
PyList_Append(ret,
- Py_BuildValue("(OO)",
- PyInt_FromLong((long)pos->keywordType),
- PyString_FromString(pos->keyword)));
+ py_elem);
+ Py_DECREF(py_elem);
pos = pos->next;
}
EXTRACTOR_freeKeywords(keys);
@@ -75,19 +85,27 @@ static PyObject * EXTRACTOR_PY_extract(PyObject * self,
}
static PyMethodDef ExtractorMethods[] = {
- { "getKeywordTypeAsString", EXTRACTOR_PY_getKeywordTypeAsString, METH_VARARGS,
+ { "getKeywordTypeAsString",
+ EXTRACTOR_PY_getKeywordTypeAsString,
+ METH_VARARGS,
"convert a keyword type (int) to the string describing the type" },
- { "loadDefaultLibraries", EXTRACTOR_PY_loadDefaultLibraries, METH_VARARGS,
+ { "loadDefaultLibraries",
+ EXTRACTOR_PY_loadDefaultLibraries,
+ METH_VARARGS,
"load the default set of libextractor plugins (returns the plugins)" },
- { "removeAll", EXTRACTOR_PY_removeAll, METH_VARARGS,
+ { "removeAll",
+ EXTRACTOR_PY_removeAll,
+ METH_VARARGS,
"unload the given set of libextractor plugins (pass plugins as argument)" },
- { "extract", EXTRACTOR_PY_extract, METH_VARARGS,
+ { "extract",
+ EXTRACTOR_PY_extract,
+ METH_VARARGS,
"extract meta data from a file (pass plugins and filename as arguments, returns vector of meta-data)" },
{ NULL, NULL, 0, NULL }
};
PyMODINIT_FUNC
-initextractor() {
- Py_InitModule("extractor", ExtractorMethods);
+init_extractor() {
+ Py_InitModule("_extractor", ExtractorMethods);
}
diff --git a/src/main/libextractor_python_setup.py b/src/main/libextractor_python_setup.py
@@ -0,0 +1,11 @@
+from distutils.core import Extension, setup
+
+cmod = Extension(sources=["libextractor_python.c"],
+ module="_extractor")
+
+setup(name="Extractor",
+ version="0.1",
+ extension=[cmod]
+ sources=["Extractor.py"],
+ author="Christian Grothoff, Heiko Wundram")
+
