|
a/src/mediaserver/cdplugins/uprcl/uprclfolders.py |
|
b/src/mediaserver/cdplugins/uprcl/uprclfolders.py |
|
... |
|
... |
44 |
#
|
44 |
#
|
45 |
# Entry 0 in _dirvec is special: it holds the 'topdirs' from the recoll
|
45 |
# Entry 0 in _dirvec is special: it holds the 'topdirs' from the recoll
|
46 |
# configuration. The entries are paths instead of simple names, and
|
46 |
# configuration. The entries are paths instead of simple names, and
|
47 |
# the docidx is 0. The diridx points to a dirvec entry.
|
47 |
# the docidx is 0. The diridx points to a dirvec entry.
|
48 |
|
48 |
|
49 |
|
|
|
50 |
import os
|
49 |
import os
|
51 |
import shlex
|
50 |
import shlex
|
52 |
import urllib
|
51 |
import urllib
|
53 |
import sys
|
52 |
import sys
|
54 |
import time
|
53 |
import time
|
|
... |
|
... |
58 |
from uprclutils import docarturi, audiomtypes, rcldirentry, \
|
57 |
from uprclutils import docarturi, audiomtypes, rcldirentry, \
|
59 |
rcldoctoentry, cmpentries
|
58 |
rcldoctoentry, cmpentries
|
60 |
from recoll import recoll
|
59 |
from recoll import recoll
|
61 |
from recoll import rclconfig
|
60 |
from recoll import rclconfig
|
62 |
|
61 |
|
63 |
_foldersIdPfx = '0$uprcl$folders'
|
62 |
class Folders(object):
|
64 |
|
63 |
|
|
|
64 |
# Initialize (read recoll data and build tree).
|
|
|
65 |
def __init__(self, confdir, httphp, pathprefix):
|
|
|
66 |
self._idprefix = '0$uprcl$folders'
|
|
|
67 |
self._httphp = httphp
|
|
|
68 |
self._pprefix = pathprefix
|
65 |
# Debug : limit processed recoll entries for speed
|
69 |
# Debug : limit processed recoll entries for speed
|
66 |
_maxrclcnt = 0
|
70 |
self._maxrclcnt = 0
|
|
|
71 |
self._fetchalldocs(confdir)
|
|
|
72 |
self._rcl2folders(confdir)
|
67 |
|
73 |
|
68 |
_dirvec = []
|
74 |
def rcldocs(self):
|
69 |
|
75 |
return self._rcldocs
|
70 |
|
76 |
|
71 |
# Create new directory entry: insert in father and append dirvec slot
|
77 |
# Create new directory entry: insert in father and append dirvec slot
|
72 |
# (with ".." entry)
|
78 |
# (with ".." entry)
|
73 |
def _createdir(dirvec, fathidx, docidx, nm):
|
79 |
def _createdir(self, fathidx, docidx, nm):
|
74 |
dirvec.append({})
|
80 |
self._dirvec.append({})
|
75 |
dirvec[fathidx][nm] = (len(dirvec) - 1, docidx)
|
81 |
self._dirvec[fathidx][nm] = (len(self._dirvec) - 1, docidx)
|
76 |
dirvec[-1][".."] = (fathidx, -1)
|
82 |
self._dirvec[-1][".."] = (fathidx, -1)
|
77 |
return len(dirvec) - 1
|
83 |
return len(self._dirvec) - 1
|
78 |
|
84 |
|
79 |
|
85 |
|
80 |
# Walk the recoll docs array and split the URLs paths to build the
|
86 |
# Walk the recoll docs array and split the URLs paths to build the
|
81 |
# [folders] data structure
|
87 |
# [folders] data structure
|
82 |
def _rcl2folders(docs, confdir, httphp, pathprefix):
|
88 |
def _rcl2folders(self, confdir):
|
83 |
global dirvec
|
|
|
84 |
dirvec = []
|
89 |
self._dirvec = []
|
85 |
start = timer()
|
90 |
start = timer()
|
86 |
|
91 |
|
87 |
rclconf = rclconfig.RclConfig(confdir)
|
92 |
rclconf = rclconfig.RclConfig(confdir)
|
88 |
topdirs = [os.path.expanduser(d) for d in
|
93 |
topdirs = [os.path.expanduser(d) for d in
|
89 |
shlex.split(rclconf.getConfParam('topdirs'))]
|
94 |
shlex.split(rclconf.getConfParam('topdirs'))]
|
90 |
topdirs = [d.rstrip('/') for d in topdirs]
|
95 |
topdirs = [d.rstrip('/') for d in topdirs]
|
91 |
|
96 |
|
92 |
# Create the 1st entry. This is special because it holds the
|
97 |
# Create the 1st entry. This is special because it holds the
|
93 |
# recoll topdirs, which are paths instead of simple names. There
|
98 |
# recoll topdirs, which are paths instead of simple names. There
|
94 |
# does not seem any need to build the tree between a topdir and /
|
99 |
# does not seem any need to build the tree between a topdir and /
|
95 |
dirvec.append({})
|
100 |
self._dirvec.append({})
|
96 |
dirvec[0][".."] = (0, -1)
|
101 |
self._dirvec[0][".."] = (0, -1)
|
97 |
for d in topdirs:
|
102 |
for d in topdirs:
|
98 |
dirvec.append({})
|
103 |
self._dirvec.append({})
|
99 |
dirvec[0][d] = (len(dirvec)-1, -1)
|
104 |
self._dirvec[0][d] = (len(self._dirvec)-1, -1)
|
100 |
dirvec[-1][".."] = (0, -1)
|
105 |
self._dirvec[-1][".."] = (0, -1)
|
101 |
|
106 |
|
102 |
# Walk the doc list and update the directory tree according to the
|
107 |
# Walk the doc list and update the directory tree according to the
|
103 |
# url: create intermediary directories if needed, create leaf
|
108 |
# url: create intermediary directories if needed, create leaf
|
104 |
# entry.
|
109 |
# entry.
|
105 |
for docidx in range(len(docs)):
|
110 |
for docidx in range(len(self._rcldocs)):
|
106 |
doc = docs[docidx]
|
111 |
doc = self._rcldocs[docidx]
|
107 |
|
112 |
|
108 |
# Possibly enrich the doc entry with a cover art uri.
|
113 |
# Possibly enrich the doc entry with a cover art uri.
|
109 |
arturi = docarturi(doc, httphp, pathprefix)
|
114 |
arturi = docarturi(doc, self._httphp, self._pprefix)
|
110 |
if arturi:
|
115 |
if arturi:
|
111 |
# The uri is quoted, so it's ascii and we can just store
|
116 |
# The uri is quoted, so it's ascii and we can just store
|
112 |
# it as a doc attribute
|
117 |
# it as a doc attribute
|
113 |
doc.albumarturi = arturi
|
118 |
doc.albumarturi = arturi
|
114 |
|
119 |
|
115 |
# No need to include non-audio types in the visible tree.
|
120 |
# No need to include non-audio types in the visible tree.
|
116 |
if doc.mtype not in audiomtypes:
|
121 |
if doc.mtype not in audiomtypes:
|
117 |
continue
|
122 |
continue
|
118 |
|
123 |
|
119 |
url = doc.getbinurl()
|
124 |
url = doc.getbinurl()
|
120 |
url = url[7:]
|
125 |
url = url[7:]
|
|
|
126 |
try:
|
|
|
127 |
decoded = url.decode('utf-8')
|
|
|
128 |
except:
|
|
|
129 |
decoded = urllib.quote(url).decode('utf-8')
|
|
|
130 |
|
|
|
131 |
# Determine the root entry (topdirs element). Special because
|
|
|
132 |
# its path is not a simple name.
|
|
|
133 |
fathidx = -1
|
|
|
134 |
for rtpath,idx in self._dirvec[0].iteritems():
|
|
|
135 |
if url.startswith(rtpath):
|
|
|
136 |
fathidx = idx[0]
|
|
|
137 |
break
|
|
|
138 |
if fathidx == -1:
|
|
|
139 |
uplog("No parent in topdirs: %s" % decoded)
|
|
|
140 |
continue
|
|
|
141 |
|
|
|
142 |
# Compute rest of path
|
|
|
143 |
url1 = url[len(rtpath):]
|
|
|
144 |
if len(url1) == 0:
|
|
|
145 |
continue
|
|
|
146 |
|
|
|
147 |
# If there is a contentgroup field, just add it as a virtual
|
|
|
148 |
# directory in the path. This only affects the visible tree,
|
|
|
149 |
# not the 'real' URLs of course.
|
|
|
150 |
if doc.contentgroup:
|
|
|
151 |
a = os.path.dirname(url1).decode('utf-8', errors='replace')
|
|
|
152 |
b = os.path.basename(url1).decode('utf-8', errors='replace')
|
|
|
153 |
url1 = os.path.join(a, doc.contentgroup, b)
|
|
|
154 |
|
|
|
155 |
# Split path, then walk the vector, possibly creating
|
|
|
156 |
# directory entries as needed
|
|
|
157 |
path = url1.split('/')[1:]
|
|
|
158 |
#uplog("%s"%path, file=sys.stderr)
|
|
|
159 |
for idx in range(len(path)):
|
|
|
160 |
elt = path[idx]
|
|
|
161 |
if elt in self._dirvec[fathidx]:
|
|
|
162 |
# This path element was already seen
|
|
|
163 |
# If this is the last entry in the path, maybe update
|
|
|
164 |
# the doc idx (previous entries were created for
|
|
|
165 |
# intermediate elements without a Doc).
|
|
|
166 |
if idx == len(path) -1:
|
|
|
167 |
self._dirvec[fathidx][elt] = (self._dirvec[fathidx][elt][0], docidx)
|
|
|
168 |
#uplog("updating docidx for %s" % decoded)
|
|
|
169 |
# Update fathidx for next iteration
|
|
|
170 |
fathidx = self._dirvec[fathidx][elt][0]
|
|
|
171 |
else:
|
|
|
172 |
# Element has no entry in father directory (hence no
|
|
|
173 |
# self._dirvec entry either).
|
|
|
174 |
if idx != len(path) -1:
|
|
|
175 |
# This is an intermediate element. Create a
|
|
|
176 |
# Doc-less directory
|
|
|
177 |
fathidx = self._createdir(fathidx, -1, elt)
|
|
|
178 |
else:
|
|
|
179 |
# Last element. If directory, needs a self._dirvec entry
|
|
|
180 |
if doc.mtype == 'inode/directory':
|
|
|
181 |
fathidx = self._createdir(fathidx, docidx, elt)
|
|
|
182 |
#uplog("Setting docidx for %s" % decoded)
|
|
|
183 |
else:
|
|
|
184 |
self._dirvec[fathidx][elt] = (-1, docidx)
|
|
|
185 |
|
|
|
186 |
if False:
|
|
|
187 |
for ent in self._dirvec:
|
|
|
188 |
uplog("%s" % ent)
|
|
|
189 |
|
|
|
190 |
end = timer()
|
|
|
191 |
uplog("_rcl2folders took %.2f Seconds" % (end - start))
|
|
|
192 |
|
|
|
193 |
# Fetch all the docs by querying Recoll with [mime:*], which is
|
|
|
194 |
# guaranteed to match every doc without overflowing the query size
|
|
|
195 |
# (because the number of mime types is limited). Something like
|
|
|
196 |
# title:* would overflow. This creates the main doc array, which is
|
|
|
197 |
# then used by all modules.
|
|
|
198 |
def _fetchalldocs(self, confdir):
|
|
|
199 |
start = timer()
|
|
|
200 |
|
|
|
201 |
rcldb = recoll.connect(confdir=confdir)
|
|
|
202 |
rclq = rcldb.query()
|
|
|
203 |
rclq.execute("mime:*", stemming=0)
|
|
|
204 |
uplog("Estimated alldocs query results: %d" % (rclq.rowcount))
|
|
|
205 |
|
|
|
206 |
totcnt = 0
|
|
|
207 |
self._rcldocs = []
|
|
|
208 |
while True:
|
|
|
209 |
docs = rclq.fetchmany()
|
|
|
210 |
for doc in docs:
|
|
|
211 |
self._rcldocs.append(doc)
|
|
|
212 |
totcnt += 1
|
|
|
213 |
if (self._maxrclcnt > 0 and totcnt >= self._maxrclcnt) or \
|
|
|
214 |
len(docs) != rclq.arraysize:
|
|
|
215 |
break
|
|
|
216 |
time.sleep(0)
|
|
|
217 |
end = timer()
|
|
|
218 |
uplog("Retrieved %d docs in %.2f Seconds" % (totcnt,end - start))
|
|
|
219 |
|
|
|
220 |
|
|
|
221 |
##############
|
|
|
222 |
# Browsing the initialized [folders] hierarchy
|
|
|
223 |
|
|
|
224 |
# Extract dirvec index from objid, according to the way we generate them.
|
|
|
225 |
def _objidtodiridx(self, pid):
|
|
|
226 |
if not pid.startswith(self._idprefix):
|
|
|
227 |
raise Exception("folders.browse: bad pid %s" % pid)
|
|
|
228 |
|
|
|
229 |
if len(self._rcldocs) == 0:
|
|
|
230 |
raise Exception("folders:browse: no docs")
|
|
|
231 |
|
|
|
232 |
diridx = pid[len(self._idprefix):]
|
|
|
233 |
if not diridx:
|
|
|
234 |
diridx = 0
|
|
|
235 |
else:
|
|
|
236 |
if diridx[1] != 'd':
|
|
|
237 |
raise Exception("folders:browse: called on non dir objid %s" %
|
|
|
238 |
pid)
|
|
|
239 |
diridx = int(diridx[2:])
|
|
|
240 |
|
|
|
241 |
if diridx >= len(self._dirvec):
|
|
|
242 |
raise Exception("folders:browse: bad pid %s" % pid)
|
|
|
243 |
|
|
|
244 |
return diridx
|
|
|
245 |
|
|
|
246 |
|
|
|
247 |
# Tell the top module what entries we define in the root
|
|
|
248 |
def rootentries(self, pid):
|
|
|
249 |
return [rcldirentry(pid + 'folders', pid, '[folders]'),]
|
|
|
250 |
|
|
|
251 |
|
|
|
252 |
# Look all non-directory docs inside directory, and return the cover
|
|
|
253 |
# art we find.
|
|
|
254 |
def _arturifordir(self, diridx):
|
|
|
255 |
for nm,ids in self._dirvec[diridx].iteritems():
|
|
|
256 |
if ids[1] >= 0:
|
|
|
257 |
doc = self._rcldocs[ids[1]]
|
|
|
258 |
if doc.mtype != 'inode/directory' and doc.albumarturi:
|
|
|
259 |
return doc.albumarturi
|
|
|
260 |
|
|
|
261 |
|
|
|
262 |
# Folder hierarchy browse method.
|
|
|
263 |
# objid is like folders$index
|
|
|
264 |
# flag is meta or children.
|
|
|
265 |
def browse(self, pid, flag):
|
|
|
266 |
|
|
|
267 |
diridx = self._objidtodiridx(pid)
|
|
|
268 |
|
|
|
269 |
# If there is only one entry in root, skip it. This means that 0
|
|
|
270 |
# and 1 point to the same dir, but this does not seem to be an
|
|
|
271 |
# issue
|
|
|
272 |
if diridx == 0 and len(self._dirvec[0]) == 2:
|
|
|
273 |
diridx = 1
|
|
|
274 |
|
|
|
275 |
entries = []
|
|
|
276 |
|
|
|
277 |
# The basename call is just for diridx==0 (topdirs). Remove it if
|
|
|
278 |
# this proves a performance issue
|
|
|
279 |
for nm,ids in self._dirvec[diridx].iteritems():
|
|
|
280 |
if nm == "..":
|
|
|
281 |
continue
|
|
|
282 |
thisdiridx = ids[0]
|
|
|
283 |
thisdocidx = ids[1]
|
|
|
284 |
if thisdocidx >= 0:
|
|
|
285 |
doc = self._rcldocs[thisdocidx]
|
|
|
286 |
else:
|
|
|
287 |
uplog("No doc for %s" % pid)
|
|
|
288 |
doc = None
|
|
|
289 |
|
|
|
290 |
if thisdiridx >= 0:
|
|
|
291 |
# Skip empty directories
|
|
|
292 |
if len(self._dirvec[thisdiridx]) == 1:
|
|
|
293 |
continue
|
|
|
294 |
id = self._idprefix + '$' + 'd' + str(thisdiridx)
|
|
|
295 |
if doc and doc.albumarturi:
|
|
|
296 |
arturi = doc.albumarturi
|
|
|
297 |
else:
|
|
|
298 |
arturi = self._arturifordir(thisdiridx)
|
|
|
299 |
entries.append(rcldirentry(id, pid, os.path.basename(nm),
|
|
|
300 |
arturi=arturi))
|
|
|
301 |
else:
|
|
|
302 |
# Not a directory. docidx had better been set
|
|
|
303 |
if thisdocidx == -1:
|
|
|
304 |
uplog("folders:docidx -1 for non-dir entry %s"%nm)
|
|
|
305 |
continue
|
|
|
306 |
doc = self._rcldocs[thisdocidx]
|
|
|
307 |
id = self._idprefix + '$i' + str(thisdocidx)
|
|
|
308 |
e = rcldoctoentry(id, pid, self._httphp, self._pprefix, doc)
|
|
|
309 |
if e:
|
|
|
310 |
entries.append(e)
|
|
|
311 |
|
|
|
312 |
return sorted(entries, cmp=cmpentries)
|
|
|
313 |
|
|
|
314 |
# Return path for objid, which has to be a container.This is good old
|
|
|
315 |
# pwd... It is called from the search module for generating a 'dir:'
|
|
|
316 |
# recoll filtering directive.
|
|
|
317 |
def dirpath(self, objid):
|
|
|
318 |
# We may get called from search, on the top dir (above
|
|
|
319 |
# [folders]). Return empty in this case
|
121 |
try:
|
320 |
try:
|
122 |
decoded = url.decode('utf-8')
|
321 |
diridx = self._objidtodiridx(objid)
|
123 |
except:
|
322 |
except:
|
124 |
decoded = urllib.quote(url).decode('utf-8')
|
323 |
return ""
|
125 |
|
324 |
|
126 |
# Determine the root entry (topdirs element). Special because
|
|
|
127 |
# its path is not a simple name.
|
|
|
128 |
fathidx = -1
|
|
|
129 |
for rtpath,idx in dirvec[0].iteritems():
|
|
|
130 |
if url.startswith(rtpath):
|
|
|
131 |
fathidx = idx[0]
|
|
|
132 |
break
|
|
|
133 |
if fathidx == -1:
|
325 |
if diridx == 0:
|
134 |
uplog("No parent in topdirs: %s" % decoded)
|
326 |
return "/"
|
135 |
continue
|
|
|
136 |
|
|
|
137 |
# Compute rest of path
|
|
|
138 |
url1 = url[len(rtpath):]
|
|
|
139 |
if len(url1) == 0:
|
|
|
140 |
continue
|
|
|
141 |
|
|
|
142 |
# If there is a contentgroup field, just add it as a virtual
|
|
|
143 |
# directory in the path. This only affects the visible tree,
|
|
|
144 |
# not the 'real' URLs of course.
|
|
|
145 |
if doc.contentgroup:
|
|
|
146 |
a = os.path.dirname(url1).decode('utf-8', errors='replace')
|
|
|
147 |
b = os.path.basename(url1).decode('utf-8', errors='replace')
|
|
|
148 |
url1 = os.path.join(a, doc.contentgroup, b)
|
|
|
149 |
|
|
|
150 |
# Split path, then walk the vector, possibly creating
|
|
|
151 |
# directory entries as needed
|
|
|
152 |
path = url1.split('/')[1:]
|
|
|
153 |
#uplog("%s"%path, file=sys.stderr)
|
|
|
154 |
for idx in range(len(path)):
|
|
|
155 |
elt = path[idx]
|
|
|
156 |
if elt in dirvec[fathidx]:
|
|
|
157 |
# This path element was already seen
|
|
|
158 |
# If this is the last entry in the path, maybe update
|
|
|
159 |
# the doc idx (previous entries were created for
|
|
|
160 |
# intermediate elements without a Doc).
|
|
|
161 |
if idx == len(path) -1:
|
|
|
162 |
dirvec[fathidx][elt] = (dirvec[fathidx][elt][0], docidx)
|
|
|
163 |
#uplog("updating docidx for %s" % decoded)
|
|
|
164 |
# Update fathidx for next iteration
|
|
|
165 |
fathidx = dirvec[fathidx][elt][0]
|
|
|
166 |
else:
|
|
|
167 |
# Element has no entry in father directory (hence no
|
|
|
168 |
# dirvec entry either).
|
|
|
169 |
if idx != len(path) -1:
|
|
|
170 |
# This is an intermediate element. Create a
|
|
|
171 |
# Doc-less directory
|
|
|
172 |
fathidx = _createdir(dirvec, fathidx, -1, elt)
|
|
|
173 |
else:
|
|
|
174 |
# Last element. If directory, needs a dirvec entry
|
|
|
175 |
if doc.mtype == 'inode/directory':
|
|
|
176 |
fathidx = _createdir(dirvec, fathidx, docidx, elt)
|
|
|
177 |
#uplog("Setting docidx for %s" % decoded)
|
|
|
178 |
else:
|
|
|
179 |
dirvec[fathidx][elt] = (-1, docidx)
|
|
|
180 |
|
|
|
181 |
if False:
|
|
|
182 |
for ent in dirvec:
|
|
|
183 |
uplog("%s" % ent)
|
|
|
184 |
|
|
|
185 |
end = timer()
|
|
|
186 |
uplog("_rcl2folders took %.2f Seconds" % (end - start))
|
|
|
187 |
return dirvec
|
|
|
188 |
|
|
|
189 |
# Fetch all the docs by querying Recoll with [mime:*], which is
|
|
|
190 |
# guaranteed to match every doc without overflowing the query size
|
|
|
191 |
# (because the number of mime types is limited). Something like
|
|
|
192 |
# title:* would overflow. This creates the main doc array, which is
|
|
|
193 |
# then used by all modules.
|
|
|
194 |
def _fetchalldocs(confdir):
|
|
|
195 |
start = timer()
|
|
|
196 |
allthedocs = []
|
|
|
197 |
|
|
|
198 |
rcldb = recoll.connect(confdir=confdir)
|
|
|
199 |
rclq = rcldb.query()
|
|
|
200 |
rclq.execute("mime:*", stemming=0)
|
|
|
201 |
uplog("Estimated alldocs query results: %d" % (rclq.rowcount))
|
|
|
202 |
|
|
|
203 |
totcnt = 0
|
|
|
204 |
while True:
|
|
|
205 |
docs = rclq.fetchmany()
|
|
|
206 |
for doc in docs:
|
|
|
207 |
allthedocs.append(doc)
|
|
|
208 |
totcnt += 1
|
|
|
209 |
if (_maxrclcnt > 0 and totcnt >= _maxrclcnt) or \
|
|
|
210 |
len(docs) != rclq.arraysize:
|
|
|
211 |
break
|
|
|
212 |
time.sleep(0)
|
|
|
213 |
end = timer()
|
|
|
214 |
uplog("Retrieved %d docs in %.2f Seconds" % (totcnt,end - start))
|
|
|
215 |
return allthedocs
|
|
|
216 |
|
|
|
217 |
|
|
|
218 |
# Initialize (read recoll data and build tree). This is called by
|
|
|
219 |
# uprcl-app init
|
|
|
220 |
def inittree(confdir, httphp, pathprefix):
|
|
|
221 |
global g_alldocs, _dirvec
|
|
|
222 |
|
327 |
|
223 |
g_alldocs = _fetchalldocs(confdir)
|
328 |
lpath = []
|
224 |
_dirvec = _rcl2folders(g_alldocs, confdir, httphp, pathprefix)
|
329 |
while True:
|
225 |
return g_alldocs
|
330 |
fathidx = self._dirvec[diridx][".."][0]
|
226 |
|
|
|
227 |
|
|
|
228 |
|
|
|
229 |
##############
|
|
|
230 |
# Browsing the initialized [folders] hierarchy
|
|
|
231 |
|
|
|
232 |
|
|
|
233 |
# Extract dirvec index from objid, according to the way we generate them.
|
|
|
234 |
def _objidtodiridx(pid):
|
|
|
235 |
if not pid.startswith(_foldersIdPfx):
|
|
|
236 |
raise Exception("folders.browse: bad pid %s" % pid)
|
|
|
237 |
|
|
|
238 |
if len(g_alldocs) == 0:
|
|
|
239 |
raise Exception("folders:browse: no docs")
|
|
|
240 |
|
|
|
241 |
diridx = pid[len(_foldersIdPfx):]
|
|
|
242 |
if not diridx:
|
|
|
243 |
diridx = 0
|
|
|
244 |
else:
|
|
|
245 |
if diridx[1] != 'd':
|
|
|
246 |
raise Exception("folders:browse: called on non dir objid %s" % pid)
|
|
|
247 |
diridx = int(diridx[2:])
|
|
|
248 |
|
|
|
249 |
if diridx >= len(_dirvec):
|
|
|
250 |
raise Exception("folders:browse: bad pid %s" % pid)
|
|
|
251 |
|
|
|
252 |
return diridx
|
|
|
253 |
|
|
|
254 |
|
|
|
255 |
# Tell the top module what entries we define in the root
|
|
|
256 |
def rootentries(pid):
|
|
|
257 |
return [rcldirentry(pid + 'folders', pid, '[folders]'),]
|
|
|
258 |
|
|
|
259 |
|
|
|
260 |
# Look all non-directory docs inside directory, and return the cover
|
|
|
261 |
# art we find.
|
|
|
262 |
def _arturifordir(diridx):
|
|
|
263 |
for nm,ids in _dirvec[diridx].iteritems():
|
331 |
for nm, ids in self._dirvec[fathidx].iteritems():
|
264 |
if ids[1] >= 0:
|
332 |
if ids[0] == diridx:
|
265 |
doc = g_alldocs[ids[1]]
|
333 |
lpath.append(nm)
|
266 |
if doc.mtype != 'inode/directory' and doc.albumarturi:
|
334 |
break
|
267 |
return doc.albumarturi
|
335 |
diridx = fathidx
|
268 |
|
336 |
if diridx == 0:
|
|
|
337 |
break
|
269 |
|
338 |
|
270 |
# Folder hierarchy browse method.
|
339 |
if not lpath:
|
271 |
# objid is like folders$index
|
340 |
path = "/"
|
272 |
# flag is meta or children.
|
|
|
273 |
# httphp and pathprefix are used to generate URIs
|
|
|
274 |
def browse(pid, flag, httphp, pathprefix):
|
|
|
275 |
|
|
|
276 |
diridx = _objidtodiridx(pid)
|
|
|
277 |
|
|
|
278 |
# If there is only one entry in root, skip it. This means that 0
|
|
|
279 |
# and 1 point to the same dir, but this does not seem to be an
|
|
|
280 |
# issue
|
|
|
281 |
if diridx == 0 and len(dirvec[0]) == 2:
|
|
|
282 |
diridx = 1
|
|
|
283 |
|
|
|
284 |
entries = []
|
|
|
285 |
|
|
|
286 |
# The basename call is just for diridx==0 (topdirs). Remove it if
|
|
|
287 |
# this proves a performance issue
|
|
|
288 |
for nm,ids in _dirvec[diridx].iteritems():
|
|
|
289 |
if nm == "..":
|
|
|
290 |
continue
|
|
|
291 |
thisdiridx = ids[0]
|
|
|
292 |
thisdocidx = ids[1]
|
|
|
293 |
if thisdocidx >= 0:
|
|
|
294 |
doc = g_alldocs[thisdocidx]
|
|
|
295 |
else:
|
341 |
else:
|
296 |
uplog("No doc for %s" % pid)
|
|
|
297 |
doc = None
|
|
|
298 |
|
|
|
299 |
if thisdiridx >= 0:
|
|
|
300 |
# Skip empty directories
|
|
|
301 |
if len(dirvec[thisdiridx]) == 1:
|
|
|
302 |
continue
|
|
|
303 |
id = _foldersIdPfx + '$' + 'd' + str(thisdiridx)
|
|
|
304 |
if doc and doc.albumarturi:
|
|
|
305 |
arturi = doc.albumarturi
|
|
|
306 |
else:
|
|
|
307 |
arturi = _arturifordir(thisdiridx)
|
|
|
308 |
entries.append(rcldirentry(id, pid, os.path.basename(nm),
|
|
|
309 |
arturi=arturi))
|
|
|
310 |
else:
|
|
|
311 |
# Not a directory. docidx had better been set
|
|
|
312 |
if thisdocidx == -1:
|
|
|
313 |
uplog("folders:docidx -1 for non-dir entry %s"%nm)
|
|
|
314 |
continue
|
|
|
315 |
doc = g_alldocs[thisdocidx]
|
|
|
316 |
id = _foldersIdPfx + '$i' + str(thisdocidx)
|
|
|
317 |
e = rcldoctoentry(id, pid, httphp, pathprefix, doc)
|
|
|
318 |
if e:
|
|
|
319 |
entries.append(e)
|
|
|
320 |
|
|
|
321 |
return sorted(entries, cmp=cmpentries)
|
|
|
322 |
|
|
|
323 |
# Return path for objid, which has to be a container.This is good old
|
|
|
324 |
# pwd... It is called from the search module for generating a 'dir:'
|
|
|
325 |
# recoll filtering directive.
|
|
|
326 |
def dirpath(objid):
|
|
|
327 |
# We may get called from search, on the top dir (above [folders]). Return
|
|
|
328 |
# empty in this case
|
|
|
329 |
try:
|
|
|
330 |
diridx = _objidtodiridx(objid)
|
|
|
331 |
except:
|
|
|
332 |
return ""
|
|
|
333 |
|
|
|
334 |
if diridx == 0:
|
|
|
335 |
return "/"
|
|
|
336 |
|
|
|
337 |
lpath = []
|
|
|
338 |
while True:
|
|
|
339 |
fathidx = _dirvec[diridx][".."][0]
|
|
|
340 |
for nm, ids in _dirvec[fathidx].iteritems():
|
|
|
341 |
if ids[0] == diridx:
|
|
|
342 |
lpath.append(nm)
|
|
|
343 |
break
|
|
|
344 |
diridx = fathidx
|
|
|
345 |
if diridx == 0:
|
|
|
346 |
break
|
|
|
347 |
|
|
|
348 |
if not lpath:
|
|
|
349 |
path = "/"
|
|
|
350 |
else:
|
|
|
351 |
path = ""
|
342 |
path = ""
|
352 |
for elt in reversed(lpath):
|
343 |
for elt in reversed(lpath):
|
353 |
path += elt + "/"
|
344 |
path += elt + "/"
|
354 |
|
345 |
|
355 |
return path
|
346 |
return path
|