tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

rclepub: use elt ids instead of hrefs + debug traces

Browse Source
This commit is contained in:
Jean-Francois Dockes 2012-10-11 15:35:15 +02:00
parent 9f4fed8bb6
commit d4edbbaedb
2 changed files with 16 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
src/filters/rclepub
View File

@ -25,7 +25,7 @@ class rclEPUB:
self.em = em self.em = em
self.em.setmimetype(rclepub_html_mtype) self.em.setmimetype(rclepub_html_mtype)
def extractone(self, path): def extractone(self, id):
"""Extract one path-named internal file from the EPUB file""" """Extract one path-named internal file from the EPUB file"""
#self.em.rclog("extractone: [%s]"%(path)) #self.em.rclog("extractone: [%s]"%(path))
@ -34,13 +34,17 @@ class rclEPUB:
iseof = rclexecm.RclExecM.eofnext iseof = rclexecm.RclExecM.eofnext
try: try:
doc = self.book.read(path) item = self.book.get_item(id)
if item is None:
raise Exception("Item not found for id %s" % (id,))
doc = self.book.read_item(item)
doc = re.sub('''</[hH][eE][aA][dD]''', doc = re.sub('''</[hH][eE][aA][dD]''',
'''<meta name="rclaptg" content="epub"></head>''', doc) '''<meta name="rclaptg" content="epub"></head>''', doc)
return (True, doc, path, iseof) self.em.setmimetype(rclepub_html_mtype)
return (True, doc, id, iseof)
except Exception, err: except Exception, err:
self.em.rclog("openfile: failed: [%s]" % err) self.em.rclog("extractone: failed: [%s]" % err)
return (False, "", path, iseof) return (False, "", id, iseof)
def openfile(self, params): def openfile(self, params):
"""Open the EPUB file""" """Open the EPUB file"""
@ -49,12 +53,12 @@ class rclEPUB:
try: try:
self.book = epub.open(params["filename:"]) self.book = epub.open(params["filename:"])
except Exception, err: except Exception, err:
self.em.rclog("openfile: failed: [%s]" % err) self.em.rclog("openfile: epub.open failed: [%s]" % err)
return False return False
for id, item in self.book.opf.manifest.iteritems(): for id, item in self.book.opf.manifest.iteritems():
# print item.__dict__ # print item.__dict__
if item.media_type == u'application/xhtml+xml': if item.media_type == u'application/xhtml+xml':
self.contents.append(item.href) self.contents.append(id)
return True return True

5
src/internfile/mh_execm.cpp
View File

@ -120,7 +120,7 @@ bool MimeHandlerExecMultiple::readDataElement(string& name, string &data)
ibuf.c_str())); ibuf.c_str()));
return false; return false;
} }
LOGDEB(("MHExecMultiple: got name [%s] len: %d\n", name.c_str(), len));
if (len / 1024 > m_maxmemberkb) { if (len / 1024 > m_maxmemberkb) {
LOGERR(("MHExecMultiple: data len > maxmemberkb\n")); LOGERR(("MHExecMultiple: data len > maxmemberkb\n"));
return false; return false;
@ -143,6 +143,9 @@ bool MimeHandlerExecMultiple::readDataElement(string& name, string &data)
len, datap->length())); len, datap->length()));
return false; return false;
} }
LOGDEB1(("MHExecMe:rdDtElt got: name [%s] len %d value [%s]\n",
name.c_str(), len, datap->size() > 100 ?
(datap->substr(0, 100) + " ...").c_str() : datap->c_str()));
return true; return true;
} }