yetangitu/recoll
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

python: make sure that all ways to get a doc attr use the same code

Browse source
This commit is contained in:
Jean-Francois Dockes 2016-08-12 18:01:37 +02:00
parent 01d0f4b734
commit aff4bd10a5
1 changed files with 90 additions and 107 deletions
Download patch file Download diff file Expand all files Collapse all files

197
src/python/recoll/pyrecoll.cpp
View file

@ -38,6 +38,7 @@ using namespace std;
#include "pathut.h" #include "pathut.h"
#include "plaintorich.h" #include "plaintorich.h"
#include "hldata.h" #include "hldata.h"
#include "smallut.h"
#include "pyrecoll.h" #include "pyrecoll.h"
@ -428,19 +429,82 @@ Doc_items(recoll_DocObject *self)
return pdict; return pdict;
} }
static bool docget(recoll_DocObject *self, const string& key, string& value) static bool idocget(recoll_DocObject *self, const string& key, string& value)
{ {
// switch (key.at(0)) {
if (!key.compare("xdocid")) { case 'u':
char cid[30]; if (!key.compare(Rcl::Doc::keyurl)) {
sprintf(cid, "%lu", (unsigned long)self->doc->xdocid); value = self->doc->url;
value = cid; return true;
return true; }
} else { break;
if (self->doc->getmeta(key, 0)) { case 'f':
value = self->doc->meta[key]; if (!key.compare(Rcl::Doc::keyfs)) {
value = self->doc->fbytes;
return true;
} else if (!key.compare(Rcl::Doc::keyfmt)) {
value = self->doc->fmtime;
return true;
}
break;
case 'd':
if (!key.compare(Rcl::Doc::keyds)) {
value = self->doc->dbytes;
return true;
} else if (!key.compare(Rcl::Doc::keydmt)) {
value = self->doc->dmtime;
return true;
}
break;
case 'i':
if (!key.compare(Rcl::Doc::keyipt)) {
value = self->doc->ipath;
return true;
}
break;
case 'm':
if (!key.compare(Rcl::Doc::keytp)) {
value = self->doc->mimetype;
return true;
} else if (!key.compare(Rcl::Doc::keymt)) {
value = self->doc->dmtime.empty() ? self->doc->fmtime :
self->doc->dmtime;
return true;
}
break;
case 'o':
if (!key.compare(Rcl::Doc::keyoc)) {
value = self->doc->origcharset;
return true;
}
break;
case 's':
if (!key.compare(Rcl::Doc::keysig)) {
value = self->doc->sig;
return true;
} else if (!key.compare(Rcl::Doc::keysz)) {
value = self->doc->dbytes.empty() ? self->doc->fbytes :
self->doc->dbytes;
return true;
}
break;
case 't':
if (!key.compare("text")) {
value = self->doc->text;
return true;
}
break;
case 'x':
if (!key.compare("xdocid")) {
ulltodecstr(self->doc->xdocid, value);
return true; return true;
} }
break;
}
if (self->doc->getmeta(key, 0)) {
value = self->doc->meta[key];
return true;
} }
return false; return false;
} }
@ -452,7 +516,7 @@ PyDoc_STRVAR(doc_Doc_get,
static PyObject * static PyObject *
Doc_get(recoll_DocObject *self, PyObject *args) Doc_get(recoll_DocObject *self, PyObject *args)
{ {
LOGDEB0("Doc_get\n" ); LOGDEB1("Doc_get\n" );
if (self->doc == 0 || the_docs.find(self->doc) == the_docs.end()) { if (self->doc == 0 || the_docs.find(self->doc) == the_docs.end()) {
PyErr_SetString(PyExc_AttributeError, "doc??"); PyErr_SetString(PyExc_AttributeError, "doc??");
return 0; return 0;
@ -465,10 +529,8 @@ Doc_get(recoll_DocObject *self, PyObject *args)
PyMem_Free(sutf8); PyMem_Free(sutf8);
string value; string value;
bool found = docget(self, key, value); if (idocget(self, key, value)) {
return PyUnicode_Decode(value.c_str(), value.size(), "UTF-8","replace");
if (found) {
return PyUnicode_Decode(value.c_str(), value.size(), "UTF-8", "replace");
} }
Py_RETURN_NONE; Py_RETURN_NONE;
@ -495,14 +557,15 @@ Doc_getattro(recoll_DocObject *self, PyObject *nameobj)
return 0; return 0;
} }
bool found = false; PyObject *meth = PyObject_GenericGetAttr((PyObject*)self, nameobj);
string value; if (meth) {
string key; return meth;
}
PyErr_Clear();
char *name = 0; char *name = 0;
PyObject* utf8o = 0;
if (PyUnicode_Check(nameobj)) { if (PyUnicode_Check(nameobj)) {
utf8o = PyUnicode_AsUTF8String(nameobj); PyObject* utf8o = PyUnicode_AsUTF8String(nameobj);
if (utf8o == 0) { if (utf8o == 0) {
LOGERR("Doc_getattro: encoding name to utf8 failed\n" ); LOGERR("Doc_getattro: encoding name to utf8 failed\n" );
PyErr_SetString(PyExc_AttributeError, "name??"); PyErr_SetString(PyExc_AttributeError, "name??");
@ -517,90 +580,12 @@ Doc_getattro(recoll_DocObject *self, PyObject *nameobj)
Py_RETURN_NONE; Py_RETURN_NONE;
} }
key = rclconfig->fieldQCanon(string(name)); string key = rclconfig->fieldQCanon(string(name));
LOGDEB1("Doc_getattro: key: " << key << endl); string value;
if (idocget(self, key, value)) {
switch (key.at(0)) {
case 'u':
if (!key.compare(Rcl::Doc::keyurl)) {
value = self->doc->url; found = true;
}
break;
case 'f':
if (!key.compare(Rcl::Doc::keyfs)) {
value = self->doc->fbytes; found = true;
} else if (!key.compare(Rcl::Doc::keyfmt)) {
value = self->doc->fmtime; found = true;
}
break;
case 'd':
if (!key.compare(Rcl::Doc::keyds)) {
value = self->doc->dbytes; found = true;
} else if (!key.compare(Rcl::Doc::keydmt)) {
value = self->doc->dmtime; found = true;
}
break;
case 'i':
if (!key.compare(Rcl::Doc::keyipt)) {
value = self->doc->ipath; found = true;
}
break;
case 'm':
if (!key.compare(Rcl::Doc::keytp)) {
value = self->doc->mimetype; found = true;
} else if (!key.compare(Rcl::Doc::keymt)) {
value = self->doc->dmtime.empty() ? self->doc->fmtime :
self->doc->dmtime; found = true;
}
break;
case 'o':
if (!key.compare(Rcl::Doc::keyoc)) {
value = self->doc->origcharset; found = true;
}
break;
case 's':
if (!key.compare(Rcl::Doc::keysig)) {
value = self->doc->sig; found = true;
} else if (!key.compare(Rcl::Doc::keysz)) {
value = self->doc->dbytes.empty() ? self->doc->fbytes :
self->doc->dbytes; found = true;
}
break;
case 't':
if (!key.compare("text")) {
value = self->doc->text; found = true;
}
break;
case 'x':
if (!key.compare("xdocid")) {
char cid[30];
sprintf(cid, "%lu", (unsigned long)self->doc->xdocid);
value = cid;
found = true;
}
break;
}
if (!found) {
// This will look up a method name (we have no other standard
// attributes)
PyObject *meth = PyObject_GenericGetAttr((PyObject*)self, nameobj);
if (meth) {
return meth;
}
PyErr_Clear();
// Else look for another attribute
if (self->doc->getmeta(key, 0)) {
value = self->doc->meta[key];
found = true;
}
}
if (found) {
LOGDEB1("Doc_getattro: [" << key << "] -> [" << value << "]\n"); LOGDEB1("Doc_getattro: [" << key << "] -> [" << value << "]\n");
// Return a python unicode object // Return a python unicode object
return PyUnicode_Decode(value.c_str(), value.size(), "utf-8", return PyUnicode_Decode(value.c_str(), value.size(), "utf-8","replace");
"replace");
} }
Py_RETURN_NONE; Py_RETURN_NONE;
@ -739,10 +724,8 @@ Doc_subscript(recoll_DocObject *self, PyObject *key)
string skey = rclconfig->fieldQCanon(string(name)); string skey = rclconfig->fieldQCanon(string(name));
string value; string value;
bool found = docget(self, skey, value); if (idocget(self, skey, value)) {
return PyUnicode_Decode(value.c_str(), value.size(), "UTF-8","replace");
if (found) {
return PyUnicode_Decode(value.c_str(), value.size(), "UTF-8", "replace");
} }
Py_RETURN_NONE; Py_RETURN_NONE;