A Python library for crawling THREDDS servers


License
MIT
Install
pip install thredds_crawler==1.5.4

Documentation

thredds_crawler

Build Status

A simple crawler/parser for THREDDS catalogs

Installation

pip install thredds_crawler

or

conda install -c conda-forge thredds_crawler

Usage

Select

You can select datasets based on their THREDDS ID using the 'select' parameter. Python regex is supported.

from thredds_crawler.crawl import Crawl
c = Crawl('http://tds.maracoos.org/thredds/MODIS.xml', select=[".*-Agg"])
print c.datasets
[
  <LeafDataset id: MODIS-Agg, name: MODIS-Complete Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2009-Agg, name: MODIS-2009 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2010-Agg, name: MODIS-2010 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2011-Agg, name: MODIS-2011 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2012-Agg, name: MODIS-2012 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2013-Agg, name: MODIS-2013 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-One-Agg, name: 1-Day-Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-Three-Agg, name: 3-Day-Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-Seven-Agg, name: 7-Day-Aggregation, services: ['OPENDAP', 'ISO']>
]

Skip

You can skip datasets based on their name and catalogRefs based on their xlink:title. By default, the crawler uses some common regular expressions to skip lists of thousands upon thousands of individual files that are part of aggregations or FMRCs:

  • .*files.*
  • .*Individual Files.*
  • .*File_Access.*
  • .*Forecast Model Run.*
  • .*Constant Forecast Offset.*
  • .*Constant Forecast Date.*

By setting the skip parameter to anything other than a superset of the default you run the risk of having some angry system admins after you.

You can access the default skip list through the Crawl.SKIPS class variable

from thredds_crawler.crawl import Crawl
print Crawl.SKIPS
[
  '.*files.*',
  '.*Individual Files.*',
  '.*File_Access.*',
  '.*Forecast Model Run.*',
  '.*Constant Forecast Offset.*',
  '.*Constant Forecast Date.*'
]

If you need to remove or add a new skip, it is strongly encouraged you use the SKIPS class variable as a starting point!

from thredds_crawler.crawl import Crawl
skips = Crawl.SKIPS + [".*-Day-Aggregation"]
c = Crawl(
  'http://tds.maracoos.org/thredds/MODIS.xml',
  select=[".*-Agg"],
  skip=skips
)
print c.datasets

[
  <LeafDataset id: MODIS-Agg, name: MODIS-Complete Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2009-Agg, name: MODIS-2009 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2010-Agg, name: MODIS-2010 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2011-Agg, name: MODIS-2011 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2012-Agg, name: MODIS-2012 Aggregation, services: ['OPENDAP', 'ISO']>,
  <LeafDataset id: MODIS-2013-Agg, name: MODIS-2013 Aggregation, services: ['OPENDAP', 'ISO']>,
]

Workers

By default there are 4 worker threads used in the crawling. You can change this by specifying a workers parameter.

import time
from contextlib import contextmanager
from thredds_crawler.crawl import Crawl

@contextmanager
def timeit(name):
    startTime = time.time()
    yield
    elapsedTime = time.time() - startTime
    print('[{}] finished in {} ms'.format(name, int(elapsedTime * 1000)))

for x in range(1, 11):
    with timeit('{} workers'.format(x)):
        Crawl("http://tds.maracoos.org/thredds/MODIS.xml", workers=x)

[1 workers] finished in 872 ms
[2 workers] finished in 397 ms
[3 workers] finished in 329 ms
[4 workers] finished in 260 ms
[5 workers] finished in 264 ms
[6 workers] finished in 219 ms
[7 workers] finished in 212 ms
[8 workers] finished in 185 ms
[9 workers] finished in 217 ms
[10 workers] finished in 205 ms

Modified Time

You can select data by the THREDDS modified_time by using a the before and after parameters. Keep in mind that the modified time is only avaiable for individual files hosted in THREDDS (not aggregations).

import pytz
from thredds_crawler.crawl import Crawl

bf = datetime(2016, 1, 5, 0, 0)
af = datetime(2015, 12, 30, 0, 0, tzinfo=pytz.utc)
url = 'http://tds.maracoos.org/thredds/catalog/MODIS-Chesapeake-Salinity/raw/2016/catalog.xml'

# after
c = Crawl(url, after=af)
assert len(c.datasets) == 3

# before
c = Crawl(url, before=bf)
assert len(c.datasets) == 3

# both
af = datetime(2016, 1, 20, 0, 0)
bf = datetime(2016, 2, 1, 0, 0)
c = Crawl(url, before=bf, after=af)
assert len(c.datasets) == 11

Authentication

You can pass an auth parameter as needed. It needs to be a requests compatible auth object.

from thredds_crawler.crawl import Crawl
auth = ('user', 'password')
c = Crawl(
  'http://tds.maracoos.org/thredds/MODIS.xml',
  select=['.*-Agg'],
  skip=Crawl.SKIPS,
  auth=auth
)

Debugging

You can pass in a debug=True parameter to Crawl to log to STDOUT what is actually happening.

from thredds_crawler.crawl import Crawl
skips = Crawl.SKIPS + [".*-Day-Aggregation"]
c = Crawl(
  'http://tds.maracoos.org/thredds/MODIS.xml',
  select=['.*-Agg'],
  skip=skips,
  debug=True
)

Crawling: http://tds.maracoos.org/thredds/MODIS.xml
Skipping catalogRef based on 'skips'.  Title: MODIS Individual Files
Skipping catalogRef based on 'skips'.  Title: 1-Day Individual Files
Skipping catalogRef based on 'skips'.  Title: 3-Day Individual Files
Skipping catalogRef based on 'skips'.  Title: 8-Day Individual Files
Processing MODIS-Agg
Processing MODIS-2009-Agg
Processing MODIS-2010-Agg
Processing MODIS-2011-Agg
Processing MODIS-2012-Agg
Processing MODIS-2013-Agg
Skipping dataset based on 'skips'.  Name: 1-Day-Aggregation

Logging

If you are using thredds_crawler from inside of another program, you can access its logs using the named logger thredds_crawler to control the log level. If you access to the named logger, do not include debug=True when initializing the Crawl object.

import logging
crawl_log = logging.getLogger('thredds_crawler')
crawl_log.setLevel(logging.WARNING)

Dataset

You can get some basic information about a LeafDataset, including the services available.

from thredds_crawler.crawl import Crawl
c = Crawl('http://tds.maracoos.org/thredds/MODIS.xml', select=['.*-Agg'])
dataset = c.datasets[0]
print dataset.id
MODIS-Agg
print dataset.name
MODIS-Complete Aggregation
print dataset.services
[
  {
    'url': 'http://tds.maracoos.org/thredds/dodsC/MODIS-Agg.nc',
    'name': 'odap',
    'service': 'OPENDAP'
  },
  {
    'url': 'http://tds.maracoos.org/thredds/iso/MODIS-Agg.nc',
    'name': 'iso',
    'service': 'ISO'
  }
]

If you have a list of datasets you can easily return all endpoints of a certain type:

from thredds_crawler.crawl import Crawl
c = Crawl('http://tds.maracoos.org/thredds/MODIS.xml', select=['.*-Agg'])
urls = [s.get("url") for d in c.datasets for s in d.services if s.get("service").lower() == "opendap"]
print urls
[
  'http://tds.maracoos.org/thredds/dodsC/MODIS-Agg.nc',
  'http://tds.maracoos.org/thredds/dodsC/MODIS-2009-Agg.nc',
  'http://tds.maracoos.org/thredds/dodsC/MODIS-2010-Agg.nc',
  'http://tds.maracoos.org/thredds/dodsC/MODIS-2011-Agg.nc',
  'http://tds.maracoos.org/thredds/dodsC/MODIS-2012-Agg.nc',
  'http://tds.maracoos.org/thredds/dodsC/MODIS-2013-Agg.nc',
  'http://tds.maracoos.org/thredds/dodsC/MODIS-One-Agg.nc',
  'http://tds.maracoos.org/thredds/dodsC/MODIS-Three-Agg.nc',
  'http://tds.maracoos.org/thredds/dodsC/MODIS-Seven-Agg.nc'
]

You can also obtain the dataset size. This returns the size on disk if the informaton is available in the TDS catalog. If it is not available and a DAP endpoint is available, it returns the theoretical size of all of thh variables. This isn't necessarialy the size on disk, because it does not account for missing_value and _FillValue space.

from thredds_crawler.crawl import Crawl
c = Crawl(
  'http://thredds.axiomalaska.com/thredds/catalogs/cencoos.html',
  select=['MB_.*']
)
sizes = [d.size for d in c.datasets]
print sizes
[29247.410283999998, 72166.289680000002]

Metadata

The entire THREDDS catalog metadata record is saved along with the dataset object. It is an etree Element object ready for you to pull information out of. See the THREDDS metadata spec

from thredds_crawler.crawl import Crawl
c = Crawl('http://tds.maracoos.org/thredds/MODIS.xml', select=['.*-Agg'])
dataset = c.datasets[0]
print dataset.metadata.find("{http://www.unidata.ucar.edu/namespaces/thredds/InvCatalog/v1.0}documentation").text
Ocean Color data are provided as a service to the broader community, and can be
influenced by sensor degradation and or algorithm changes. We make efforts to keep
this dataset updated and calibrated. The products in these files are experimental.
Aggregations are simple means of available data over the specified time frame. Use at
your own discretion.

Use Case

Below is a python script that can be used to harvest THEDDS catalogs and save the ISO metadata files to a local directory

import os
import urllib
from thredds_crawler.crawl import Crawl

import logging
import logging.handlers
logger = logging.getLogger('thredds_crawler')
fh = logging.handlers.RotatingFileHandler('/var/log/iso_harvest/iso_harvest.log', maxBytes=1024*1024*10, backupCount=5)
fh.setLevel(logging.DEBUG)
ch = logging.StreamHandler()
ch.setLevel(logging.DEBUG)
formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
fh.setFormatter(formatter)
ch.setFormatter(formatter)
logger.addHandler(fh)
logger.addHandler(ch)
logger.setLevel(logging.DEBUG)

SAVE_DIR="/srv/http/iso"

THREDDS_SERVERS = {
    "aoos":      "http://thredds.axiomalaska.com/thredds/catalogs/aoos.html",
    "cencoos":   "http://thredds.axiomalaska.com/thredds/catalogs/cencoos.html",
    "maracoos" : "http://tds.maracoos.org/thredds/catalog.html",
    "glos":      "http://tds.glos.us/thredds/catalog.html"
}

for subfolder, thredds_url in THREDDS_SERVERS.items():
  logger.info("Crawling %s (%s)" % (subfolder, thredds_url))
  crawler = Crawl(thredds_url, debug=True)
  isos = [(d.id, s.get("url")) for d in crawler.datasets for s in d.services if s.get("service").lower() == "iso"]
  filefolder = os.path.join(SAVE_DIR, subfolder)
  if not os.path.exists(filefolder):
    os.makedirs(filefolder)
  for iso in isos:
    try:
      filename = iso[0].replace("/", "_") + ".iso.xml"
      filepath = os.path.join(filefolder, filename)
      logger.info("Downloading/Saving %s" % filepath)
      urllib.urlretrieve(iso[1], filepath)
    except BaseException:
      logger.exception("Error!")