diff --git a/geomagio/Url.py b/geomagio/Url.py
new file mode 100644
index 0000000000000000000000000000000000000000..636a2dd656d93250d76918e68775377ce50c425e
--- /dev/null
+++ b/geomagio/Url.py
@@ -0,0 +1,239 @@
+"""Class to read a file from a URL given a template"""
+import os
+import urllib2
+from TimeseriesFactoryException import TimeseriesFactoryException
+
+
+class URL():
+ """URL class to allow reading of files using the urllib2 class
+
+ Parameters
+ ----------
+ urlTemplate : str
+ A string that contains any of the following replacement patterns:
+ - '%(i)s' : interval abbreviation
+ - '%(interval)s' interval name
+ - '%(julian)s' julian date
+ - '%(obs)s' lowercase observatory code
+ - '%(OBS)s' uppercase observatory code
+ - '%(t)s' type abbreviation
+ - '%(type)s' type name
+ - '%(year)s' year formatted as YYYY
+ - '%(ymd)s' time formatted as YYYYMMDD
+ """
+
+ def __init__(self, urlTemplate):
+ self.urlTemplate = urlTemplate
+
+ def get_file_from_url(self, url):
+ """Get a file for writing.
+
+ Ensures parent directory exists.
+
+ Parameters
+ ----------
+ url : str
+ path to file
+
+ Returns
+ -------
+ str
+ path to file without file:// prefix
+
+ Raises
+ ------
+ TimeseriesFactoryException
+ if url does not start with file://
+ """
+ if not url.startswith('file://'):
+ raise TimeseriesFactoryException(
+ 'Only file urls are supported for writing')
+ filename = url.replace('file://', '')
+ parent = os.path.dirname(filename)
+ if not os.path.exists(parent):
+ os.makedirs(parent)
+ return filename
+
+ def get_url(self, observatory, date, type='variation', interval='minute'):
+ """Get the url for a specified file.
+
+ Replaces patterns (described in class docstring) with values based on
+ parameter values.
+
+ Parameters
+ ----------
+ observatory : str
+ observatory code.
+ date : obspy.core.UTCDateTime
+ day to fetch (only year, month, day are used)
+ type : {'variation', 'quasi-definitive', 'definitive'}
+ data type.
+ interval : {'minute', 'second', 'hourly', 'daily'}
+ data interval.
+
+ Raises
+ ------
+ TimeseriesFactoryException
+ if type or interval are not supported.
+ """
+ return self.urlTemplate % {
+ 'i': self._get_interval_abbreviation(interval),
+ 'interval': self._get_interval_name(interval),
+ 'julian': date.strftime("%j"),
+ 'obs': observatory.lower(),
+ 'OBS': observatory.upper(),
+ 't': self._get_type_abbreviation(type),
+ 'type': self._get_type_name(type),
+ 'year': date.strftime("%Y"),
+ 'ymd': date.strftime('%Y%m%d')}
+
+ def read_url(self, url):
+ """Open and read url contents.
+
+ Parameters
+ ----------
+ url : str
+ A urllib2 compatible url, such as http:// or file://.
+
+ Returns
+ -------
+ str
+ contents returned by url.
+
+ Raises
+ ------
+ urllib2.URLError
+ if any occurs
+ """
+ response = urllib2.urlopen(url)
+ content = None
+ try:
+ content = response.read()
+ except urllib2.URLError, e:
+ print e.reason
+ raise
+ finally:
+ response.close()
+ return content
+
+ def _get_interval_abbreviation(self, interval):
+ """Get abbreviation for a data interval.
+
+ Used by ``_get_url`` to replace ``%(i)s`` in urlTemplate.
+
+ Parameters
+ ----------
+ interval : {'daily', 'hourly', 'minute', 'monthly', 'second'}
+
+ Returns
+ -------
+ abbreviation for ``interval``.
+
+ Raises
+ ------
+ TimeseriesFactoryException
+ if ``interval`` is not supported.
+ """
+ interval_abbr = None
+ if interval == 'daily':
+ interval_abbr = 'day'
+ elif interval == 'hourly':
+ interval_abbr = 'hor'
+ elif interval == 'minute':
+ interval_abbr = 'min'
+ elif interval == 'monthly':
+ interval_abbr = 'mon'
+ elif interval == 'second':
+ interval_abbr = 'sec'
+ else:
+ raise TimeseriesFactoryException(
+ 'Unexpected interval "%s"' % interval)
+ return interval_abbr
+
+ def _get_interval_name(self, interval):
+ """Get name for a data interval.
+
+ Used by ``_get_url`` to replace ``%(interval)s`` in urlTemplate.
+
+ Parameters
+ ----------
+ interval : {'minute', 'second'}
+
+ Returns
+ -------
+ name for ``interval``.
+
+ Raises
+ ------
+ TimeseriesFactoryException
+ if ``interval`` is not supported.
+ """
+ interval_name = None
+ if interval == 'minute':
+ interval_name = 'OneMinute'
+ elif interval == 'second':
+ interval_name = 'OneSecond'
+ else:
+ raise TimeseriesFactoryException(
+ 'Unsupported interval "%s"' % interval)
+ return interval_name
+
+ def _get_type_abbreviation(self, type):
+ """Get abbreviation for a data type.
+
+ Used by ``_get_url`` to replace ``%(t)s`` in urlTemplate.
+
+ Parameters
+ ----------
+ type : {'definitive', 'provisional', 'quasi-definitive', 'variation'}
+
+ Returns
+ -------
+ name for ``type``.
+
+ Raises
+ ------
+ TimeseriesFactoryException
+ if ``type`` is not supported.
+ """
+ type_abbr = None
+ if type == 'definitive':
+ type_abbr = 'd'
+ elif type == 'provisional':
+ type_abbr = 'p'
+ elif type == 'quasi-definitive':
+ type_abbr = 'q'
+ elif type == 'variation':
+ type_abbr = 'v'
+ else:
+ raise TimeseriesFactoryException(
+ 'Unexpected type "%s"' % type)
+ return type_abbr
+
+ def _get_type_name(self, type):
+ """Get name for a data type.
+
+ Used by ``_get_url`` to replace ``%(type)s`` in urlTemplate.
+
+ Parameters
+ ----------
+ type : {'variation', 'quasi-definitive'}
+
+ Returns
+ -------
+ name for ``type``.
+
+ Raises
+ ------
+ TimeseriesFactoryException
+ if ``type`` is not supported.
+ """
+ type_name = None
+ if type == 'variation':
+ type_name = ''
+ elif type == 'quasi-definitive':
+ type_name = 'QuasiDefinitive'
+ else:
+ raise TimeseriesFactoryException(
+ 'Unsupported type "%s"' % type)
+ return type_name
diff --git a/geomagio/__init__.py b/geomagio/__init__.py
index 74c7d9ea771a19bcb02cd7068c234e885425d0f8..47c82746e0de2803d10b49eb2c8907202e162587 100644
--- a/geomagio/__init__.py
+++ b/geomagio/__init__.py
@@ -11,6 +11,7 @@ from TimeseriesFactory import TimeseriesFactory
from TimeseriesFactoryException import TimeseriesFactoryException
import TimeseriesUtility
import Util
+import Url
from XYZAlgorithm import XYZAlgorithm
__all__ = [
@@ -24,5 +25,6 @@ __all__ = [
'TimeseriesFactoryException',
'TimeseriesUtility',
'Util',
+ 'Url',
'XYZAlgorithm'
]
diff --git a/geomagio/iaga2002/IAGA2002Factory.py b/geomagio/iaga2002/IAGA2002Factory.py
index 1017cc4e21a258beae7539248c6ad6ea047f72dc..43c94f7679b5c4af4df850ca730706bc9bd50719 100644
--- a/geomagio/iaga2002/IAGA2002Factory.py
+++ b/geomagio/iaga2002/IAGA2002Factory.py
@@ -1,11 +1,10 @@
"""Factory that loads IAGA2002 Files."""
import obspy.core
-import os
-import urllib2
from .. import ChannelConverter
from ..TimeseriesFactory import TimeseriesFactory
from ..TimeseriesFactoryException import TimeseriesFactoryException
+from ..Url import URL
from IAGA2002Parser import IAGA2002Parser
from IAGA2002Writer import IAGA2002Writer
@@ -14,36 +13,6 @@ from IAGA2002Writer import IAGA2002Writer
IAGA_FILE_PATTERN = '%(obs)s%(ymd)s%(t)s%(i)s.%(i)s'
-def read_url(url):
- """Open and read url contents.
-
- Parameters
- ----------
- url : str
- A urllib2 compatible url, such as http:// or file://.
-
- Returns
- -------
- str
- contents returned by url.
-
- Raises
- ------
- urllib2.URLError
- if any occurs
- """
- response = urllib2.urlopen(url)
- content = None
- try:
- content = response.read()
- except urllib2.URLError, e:
- print e.reason
- raise
- finally:
- response.close()
- return content
-
-
class IAGA2002Factory(TimeseriesFactory):
"""TimeseriesFactory for IAGA 2002 formatted files.
@@ -103,9 +72,10 @@ class IAGA2002Factory(TimeseriesFactory):
interval = interval or self.interval
days = self._get_days(starttime, endtime)
timeseries = obspy.core.Stream()
+ url = URL(self.urlTemplate)
for day in days:
- url = self._get_url(observatory, day, type, interval)
- iagaFile = read_url(url)
+ url_id = url.get_url(observatory, day, type, interval)
+ iagaFile = url.read_url(url_id)
timeseries += self.parse_string(iagaFile)
# merge channel traces for multiple days
timeseries.merge()
@@ -147,159 +117,6 @@ class IAGA2002Factory(TimeseriesFactory):
stream += obspy.core.Trace(data[channel], stats)
return stream
- def _get_url(self, observatory, date, type='variation', interval='minute'):
- """Get the url for a specified IAGA2002 file.
-
- Replaces patterns (described in class docstring) with values based on
- parameter values.
-
- Parameters
- ----------
- observatory : str
- observatory code.
- date : obspy.core.UTCDateTime
- day to fetch (only year, month, day are used)
- type : {'variation', 'quasi-definitive'}
- data type.
- interval : {'minute', 'second'}
- data interval.
-
- Raises
- ------
- TimeseriesFactoryException
- if type or interval are not supported.
- """
- return self.urlTemplate % {
- 'i': self._get_interval_abbreviation(interval),
- 'interval': self._get_interval_name(interval),
- 'obs': observatory.lower(),
- 'OBS': observatory.upper(),
- 't': self._get_type_abbreviation(type),
- 'type': self._get_type_name(type),
- 'ymd': date.strftime("%Y%m%d")}
-
- def _get_interval_abbreviation(self, interval):
- """Get abbreviation for a data interval.
-
- Used by ``_get_url`` to replace ``%(i)s`` in urlTemplate.
-
- Parameters
- ----------
- interval : {'daily', 'hourly', 'minute', 'monthly', 'second'}
-
- Returns
- -------
- abbreviation for ``interval``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``interval`` is not supported.
- """
- interval_abbr = None
- if interval == 'daily':
- interval_abbr = 'day'
- elif interval == 'hourly':
- interval_abbr = 'hor'
- elif interval == 'minute':
- interval_abbr = 'min'
- elif interval == 'monthly':
- interval_abbr = 'mon'
- elif interval == 'second':
- interval_abbr = 'sec'
- else:
- raise TimeseriesFactoryException(
- 'Unexpected interval "%s"' % interval)
- return interval_abbr
-
- def _get_interval_name(self, interval):
- """Get name for a data interval.
-
- Used by ``_get_url`` to replace ``%(interval)s`` in urlTemplate.
-
- Parameters
- ----------
- interval : {'minute', 'second'}
-
- Returns
- -------
- name for ``interval``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``interval`` is not supported.
- """
- interval_name = None
- if interval == 'minute':
- interval_name = 'OneMinute'
- elif interval == 'second':
- interval_name = 'OneSecond'
- else:
- raise TimeseriesFactoryException(
- 'Unsupported interval "%s"' % interval)
- return interval_name
-
- def _get_type_abbreviation(self, type):
- """Get abbreviation for a data type.
-
- Used by ``_get_url`` to replace ``%(t)s`` in urlTemplate.
-
- Parameters
- ----------
- type : {'definitive', 'provisional', 'quasi-definitive', 'variation'}
-
- Returns
- -------
- name for ``type``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``type`` is not supported.
- """
- type_abbr = None
- if type == 'definitive':
- type_abbr = 'd'
- elif type == 'provisional':
- type_abbr = 'p'
- elif type == 'quasi-definitive':
- type_abbr = 'q'
- elif type == 'variation':
- type_abbr = 'v'
- else:
- raise TimeseriesFactoryException(
- 'Unexpected type "%s"' % type)
- return type_abbr
-
- def _get_type_name(self, type):
- """Get name for a data type.
-
- Used by ``_get_url`` to replace ``%(type)s`` in urlTemplate.
-
- Parameters
- ----------
- type : {'variation', 'quasi-definitive'}
-
- Returns
- -------
- name for ``type``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``type`` is not supported.
- """
- type_name = None
- if type == 'variation':
- type_name = ''
- elif type == 'quasi-definitive':
- type_name = 'QuasiDefinitive'
- else:
- raise TimeseriesFactoryException(
- 'Unsupported type "%s"' % type)
- return type_name
-
def _get_days(self, starttime, endtime):
"""Get days between (inclusive) starttime and endtime.
@@ -382,42 +199,14 @@ class IAGA2002Factory(TimeseriesFactory):
starttime = starttime or stats.starttime
endtime = endtime or stats.endtime
days = self._get_days(starttime, endtime)
+ url = URL(self.urlTemplate)
for day in days:
- day_filename = self._get_file_from_url(
- self._get_url(observatory, day, type, interval))
+ day_filename = url.get_file_from_url(
+ url.get_url(observatory, day, type, interval))
day_timeseries = self._get_slice(timeseries, day, interval)
with open(day_filename, 'wb') as fh:
self.write_file(fh, day_timeseries, channels)
- def _get_file_from_url(self, url):
- """Get a file for writing.
-
- Ensures parent directory exists.
-
- Parameters
- ----------
- url : str
- Url path to IAGA2002
-
- Returns
- -------
- str
- path to file without file:// prefix
-
- Raises
- ------
- TimeseriesFactoryException
- if url does not start with file://
- """
- if not url.startswith('file://'):
- raise TimeseriesFactoryException(
- 'Only file urls are supported for writing')
- filename = url.replace('file://', '')
- parent = os.path.dirname(filename)
- if not os.path.exists(parent):
- os.makedirs(parent)
- return filename
-
def _get_slice(self, timeseries, day, interval):
"""Get the first and last time for a day
diff --git a/geomagio/imfv283/IMFV283Factory.py b/geomagio/imfv283/IMFV283Factory.py
index 530ca8e6518a53059db9b7fe7607ab3fb80e412b..2a33a40e7c9fc22f60e40e5985774c16e359f4a0 100644
--- a/geomagio/imfv283/IMFV283Factory.py
+++ b/geomagio/imfv283/IMFV283Factory.py
@@ -1,64 +1,17 @@
"""Factory that loads IAGA2002 Files."""
import obspy.core
-import os
-import urllib2
import numpy
from .. import ChannelConverter
from ..TimeseriesFactory import TimeseriesFactory
from ..TimeseriesFactoryException import TimeseriesFactoryException
+from ..Url import URL
from IMFV283Parser import IMFV283Parser
-# pattern for IMFV283 file names
-IMFV283_FILE_PATTERN = 'dcpmsgs.txt'
-
-
-def read_url(url):
- """Open and read url contents.
-
- Parameters
- ----------
- url : str
- A urllib2 compatible url, such as http:// or file://.
-
- Returns
- -------
- str
- contents returned by url.
-
- Raises
- ------
- urllib2.URLError
- if any occurs
- """
- response = urllib2.urlopen(url)
- content = None
- try:
- content = response.read()
- except urllib2.URLError, e:
- print e.reason
- raise
- finally:
- response.close()
- return content
-
-
class IMFV283Factory(TimeseriesFactory):
"""TimeseriesFactory for IMFV283 formatted files.
- Parameters
- ----------
- urlTemplate : str
- A string that contains any of the following replacement patterns:
- - '%(i)s' : interval abbreviation
- - '%(interval)s' interval name
- - '%(obs)s' lowercase observatory code
- - '%(OBS)s' uppercase observatory code
- - '%(t)s' type abbreviation
- - '%(type)s' type name
- - '%(ymd)s' time formatted as YYYYMMDD
-
See Also
--------
IMFV283Parser
@@ -108,10 +61,11 @@ class IMFV283Factory(TimeseriesFactory):
type = type or self.type
interval = interval or self.interval
timeseries = obspy.core.Stream()
-
- url = self._get_url(observatory, obspy.core.UTCDateTime(),
+ url = URL(self.urlTemplate)
+ url_id = url.get_url(observatory, obspy.core.UTCDateTime(),
type, interval)
- imfV283File = read_url(url)
+
+ imfV283File = url.read_url(url_id)
timeseries += self.parse_string(imfV283File)
# merge channel traces for multiple days
timeseries.merge()
@@ -151,195 +105,6 @@ class IMFV283Factory(TimeseriesFactory):
return stream
- def _get_url(self, observatory, date, type='variation', interval='minute'):
- """Get the url for a specified IMFV283 file.
-
- Replaces patterns (described in class docstring) with values based on
- parameter values.
-
- Parameters
- ----------
- observatory : str
- observatory code.
- date : obspy.core.UTCDateTime
- day to fetch (only year, month, day are used)
- type : {'variation', 'quasi-definitive'}
- data type.
- interval : {'minute', 'second'}
- data interval.
-
- Raises
- ------
- TimeseriesFactoryException
- if type or interval are not supported.
- """
- return self.urlTemplate % {
- 'i': self._get_interval_abbreviation(interval),
- 'interval': self._get_interval_name(interval),
- 'obs': observatory.lower(),
- 'OBS': observatory.upper(),
- 't': self._get_type_abbreviation(type),
- 'type': self._get_type_name(type),
- 'ymd': date.strftime("%Y%m%d")}
-
- def _get_interval_abbreviation(self, interval):
- """Get abbreviation for a data interval.
-
- Used by ``_get_url`` to replace ``%(i)s`` in urlTemplate.
-
- Parameters
- ----------
- interval : {'daily', 'hourly', 'minute', 'monthly', 'second'}
-
- Returns
- -------
- abbreviation for ``interval``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``interval`` is not supported.
- """
- print interval
- interval_abbr = None
- if interval == 'daily':
- interval_abbr = 'day'
- elif interval == 'hourly':
- interval_abbr = 'hor'
- elif interval == 'minute':
- interval_abbr = 'min'
- elif interval == 'monthly':
- interval_abbr = 'mon'
- elif interval == 'second':
- interval_abbr = 'sec'
- else:
- raise TimeseriesFactoryException(
- 'Unexpected interval "%s"' % interval)
- return interval_abbr
-
- def _get_interval_name(self, interval):
- """Get name for a data interval.
-
- Used by ``_get_url`` to replace ``%(interval)s`` in urlTemplate.
-
- Parameters
- ----------
- interval : {'minute', 'second'}
-
- Returns
- -------
- name for ``interval``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``interval`` is not supported.
- """
- interval_name = None
- if interval == 'minute':
- interval_name = 'OneMinute'
- elif interval == 'second':
- interval_name = 'OneSecond'
- else:
- raise TimeseriesFactoryException(
- 'Unsupported interval "%s"' % interval)
- return interval_name
-
- def _get_type_abbreviation(self, type):
- """Get abbreviation for a data type.
-
- Used by ``_get_url`` to replace ``%(t)s`` in urlTemplate.
-
- Parameters
- ----------
- type : {'definitive', 'provisional', 'quasi-definitive', 'variation'}
-
- Returns
- -------
- name for ``type``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``type`` is not supported.
- """
- type_abbr = None
- if type == 'definitive':
- type_abbr = 'd'
- elif type == 'provisional':
- type_abbr = 'p'
- elif type == 'quasi-definitive':
- type_abbr = 'q'
- elif type == 'variation':
- type_abbr = 'v'
- else:
- raise TimeseriesFactoryException(
- 'Unexpected type "%s"' % type)
- return type_abbr
-
- def _get_type_name(self, type):
- """Get name for a data type.
-
- Used by ``_get_url`` to replace ``%(type)s`` in urlTemplate.
-
- Parameters
- ----------
- type : {'variation', 'quasi-definitive'}
-
- Returns
- -------
- name for ``type``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``type`` is not supported.
- """
- type_name = None
- if type == 'variation':
- type_name = ''
- elif type == 'quasi-definitive':
- type_name = 'QuasiDefinitive'
- else:
- raise TimeseriesFactoryException(
- 'Unsupported type "%s"' % type)
- return type_name
-
- def _get_days(self, starttime, endtime):
- """Get days between (inclusive) starttime and endtime.
-
- Parameters
- ----------
- starttime : obspy.core.UTCDateTime
- the start time
- endtime : obspy.core.UTCDateTime
- the end time
-
- Returns
- -------
- array_like
- list of times, one per day, for all days between and including
- ``starttime`` and ``endtime``.
-
- Raises
- ------
- TimeseriesFactoryException
- if starttime is after endtime
- """
- if starttime > endtime:
- raise TimeseriesFactoryException(
- 'starttime must be before endtime')
- days = []
- day = starttime
- lastday = (endtime.year, endtime.month, endtime.day)
- while True:
- days.append(day)
- if lastday == (day.year, day.month, day.day):
- break
- # move to next day
- day = obspy.core.UTCDateTime(day.timestamp + 86400)
- return days
-
def write_file(self, fh, timeseries, channels):
"""writes timeseries data to the given file object.
@@ -352,32 +117,3 @@ class IMFV283Factory(TimeseriesFactory):
list of channels to store
"""
raise TimeseriesFactoryException('IAF write_file not implemented.')
-
- def _get_file_from_url(self, url):
- """Get a file for writing.
-
- Ensures parent directory exists.
-
- Parameters
- ----------
- url : str
- Url path to IMFV283
-
- Returns
- -------
- str
- path to file without file:// prefix
-
- Raises
- ------
- TimeseriesFactoryException
- if url does not start with file://
- """
- if not url.startswith('file://'):
- raise TimeseriesFactoryException(
- 'Only file urls are supported for writing')
- filename = url.replace('file://', '')
- parent = os.path.dirname(filename)
- if not os.path.exists(parent):
- os.makedirs(parent)
- return filename
diff --git a/geomagio/pcdcp/PCDCPFactory.py b/geomagio/pcdcp/PCDCPFactory.py
index 8d6c0e82be2686887847574ed2b6b90a353c517a..d6b961240904b51b4a893f72fbc576bae6b91ff0 100644
--- a/geomagio/pcdcp/PCDCPFactory.py
+++ b/geomagio/pcdcp/PCDCPFactory.py
@@ -1,11 +1,10 @@
"""Factory that loads PCDCP Files."""
import obspy.core
-import os
-import urllib2
from .. import ChannelConverter
from ..TimeseriesFactory import TimeseriesFactory
from ..TimeseriesFactoryException import TimeseriesFactoryException
+from ..Url import URL
from PCDCPParser import PCDCPParser
from PCDCPWriter import PCDCPWriter
@@ -14,36 +13,6 @@ from PCDCPWriter import PCDCPWriter
PCDCP_FILE_PATTERN = '%(obs)s%(y)s%(j)s.%(i)s'
-def read_url(url):
- """Open and read url contents.
-
- Parameters
- ----------
- url : str
- A urllib2 compatible url, such as http:// or file://.
-
- Returns
- -------
- str
- contents returned by url.
-
- Raises
- ------
- urllib2.URLError
- if any occurs
- """
- response = urllib2.urlopen(url)
- content = None
- try:
- content = response.read()
- except urllib2.URLError, e:
- print e.reason
- raise
- finally:
- response.close()
- return content
-
-
class PCDCPFactory(TimeseriesFactory):
"""TimeseriesFactory for PCDCP formatted files.
@@ -105,10 +74,10 @@ class PCDCPFactory(TimeseriesFactory):
interval = interval or self.interval
days = self._get_days(starttime, endtime)
timeseries = obspy.core.Stream()
-
+ url = URL(self.urlTemplate)
for day in days:
- url = self._get_url(observatory, day, type, interval)
- pcdcpFile = read_url(url)
+ url_id = url.get_url(observatory, day, type, interval)
+ pcdcpFile = url.read_url(url_id)
timeseries += self.parse_string(pcdcpFile)
# merge channel traces for multiple days
@@ -174,162 +143,6 @@ class PCDCPFactory(TimeseriesFactory):
return stream
- def _get_url(self, observatory, date, type='variation', interval='minute'):
- """Get the url for a specified PCDCP file.
-
- Replaces patterns (described in class docstring) with values based on
- parameter values.
-
- Parameters
- ----------
- observatory : str
- observatory code.
- date : obspy.core.UTCDateTime
- day to fetch (only year, month, day are used)
- type : {'variation', 'quasi-definitive'}
- data type.
- interval : {'minute', 'second'}
- data interval.
-
- Raises
- ------
- TimeseriesFactoryException
- If type or interval are not supported.
- """
- return self.urlTemplate % {
- 'i': self._get_interval_abbreviation(interval),
- 'interval': self._get_interval_name(interval),
- 'julian': date.strftime("%j"),
- 'obs': observatory.lower(),
- 'OBS': observatory.upper(),
- 't': self._get_type_abbreviation(type),
- 'type': self._get_type_name(type),
- 'year': date.strftime("%Y"),
- 'ymd': date.strftime("%Y%m%d")
- }
-
- def _get_interval_abbreviation(self, interval):
- """Get abbreviation for a data interval.
-
- Used by ``_get_url`` to replace ``%(i)s`` in urlTemplate.
-
- Parameters
- ----------
- interval : {'daily', 'hourly', 'minute', 'monthly', 'second'}
-
- Returns
- -------
- abbreviation for ``interval``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``interval`` is not supported.
- """
- interval_abbr = None
- if interval == 'daily':
- interval_abbr = 'day'
- elif interval == 'hourly':
- interval_abbr = 'hor'
- elif interval == 'minute':
- interval_abbr = 'min'
- elif interval == 'monthly':
- interval_abbr = 'mon'
- elif interval == 'second':
- interval_abbr = 'sec'
- else:
- raise TimeseriesFactoryException(
- 'Unexpected interval "%s"' % interval)
- return interval_abbr
-
- def _get_interval_name(self, interval):
- """Get name for a data interval.
-
- Used by ``_get_url`` to replace ``%(interval)s`` in urlTemplate.
-
- Parameters
- ----------
- interval : {'minute', 'second'}
-
- Returns
- -------
- name for ``interval``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``interval`` is not supported.
- """
- interval_name = None
- if interval == 'minute':
- interval_name = 'OneMinute'
- elif interval == 'second':
- interval_name = 'OneSecond'
- else:
- raise TimeseriesFactoryException(
- 'Unsupported interval "%s"' % interval)
- return interval_name
-
- def _get_type_abbreviation(self, type):
- """Get abbreviation for a data type.
-
- Used by ``_get_url`` to replace ``%(t)s`` in urlTemplate.
-
- Parameters
- ----------
- type : {'definitive', 'provisional', 'quasi-definitive', 'variation'}
-
- Returns
- -------
- name for ``type``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``type`` is not supported.
- """
- type_abbr = None
- if type == 'definitive':
- type_abbr = 'd'
- elif type == 'provisional':
- type_abbr = 'p'
- elif type == 'quasi-definitive':
- type_abbr = 'q'
- elif type == 'variation':
- type_abbr = 'v'
- else:
- raise TimeseriesFactoryException(
- 'Unexpected type "%s"' % type)
- return type_abbr
-
- def _get_type_name(self, type):
- """Get name for a data type.
-
- Used by ``_get_url`` to replace ``%(type)s`` in urlTemplate.
-
- Parameters
- ----------
- type : {'variation', 'quasi-definitive'}
-
- Returns
- -------
- name for ``type``.
-
- Raises
- ------
- TimeseriesFactoryException
- if ``type`` is not supported.
- """
- type_name = None
- if type == 'variation':
- type_name = ''
- elif type == 'quasi-definitive':
- type_name = 'QuasiDefinitive'
- else:
- raise TimeseriesFactoryException(
- 'Unsupported type "%s"' % type)
- return type_name
-
def _get_days(self, starttime, endtime):
"""Get days between (inclusive) starttime and endtime.
@@ -414,47 +227,15 @@ class PCDCPFactory(TimeseriesFactory):
starttime = starttime or stats.starttime
endtime = endtime or stats.endtime
days = self._get_days(starttime, endtime)
-
+ url = URL(self.urlTemplate)
for day in days:
- day_filename = self._get_file_from_url(
- self._get_url(observatory, day, type, interval))
+ day_filename = url.get_file_from_url(
+ url.get_url(observatory, day, type, interval))
day_timeseries = self._get_slice(timeseries, day, interval)
with open(day_filename, 'wb') as fh:
self.write_file(fh, day_timeseries, channels)
- def _get_file_from_url(self, url):
- """Get a file for writing.
-
- Ensures parent directory exists.
-
- Parameters
- ----------
- url : str
- Url path to PCDCP
-
- Returns
- -------
- str
- path to file without file:// prefix
-
- Raises
- ------
- TimeseriesFactoryException
- if url does not start with file://
- """
- if not url.startswith('file://'):
- raise TimeseriesFactoryException(
- 'Only file urls are supported for writing')
-
- filename = url.replace('file://', '')
- parent = os.path.dirname(filename)
-
- if not os.path.exists(parent):
- os.makedirs(parent)
-
- return filename
-
def _get_slice(self, timeseries, day, interval):
"""Get the first and last time for a day