akara.conf.template

# -*- mode: python -*-

# This is the main Akara server configuration file. It contains
# settings for the server and for its extension modules.

# This file is preprocessed by setup.py to create akara.conf, which
# replaces tokens in this file with values from akara.ini. akara.ini
# takes the standard python ConfigParser format of: 
#
#     [Section]
#     Key = value 
# 
# Tokens in this file should take the form of: 
# 
#     ${Section__Key}

# The configuration file is written in Python. Configuration data goes
# into class attributes. If the module is "A.B.C" then the
# configuration information should be in the class named "C". If there
# is a conflict then use the parameter "akara_name" to set the full
# module name.

# class C1:
#    akara_name = "akara.demo.C"
#    name = "this is for one module ending with C"
#
# class C2:
#    akara_name = "akara.example.C"
#    name = "this is for another C module"

# An extension module should get its configuration class using
#   akara.module_config(name)
# where "name" is the full module name (usually __name__).
# The name can be omitted in which case __name__ is used.

### Section 1: Global Akara Environment
#  The directives in this section affect the overall operation of
#  Akara, such as the number of concurrent requests it can handle and
#  where it should place its PID and log files.
#

class Akara:
    #  Listen: interface name (optional) and port to listen for HTTP requests
    Listen = ${Akara__Port}
    #  To specify the interface name use this format:
    #  Listen = "localhost:8880"

    # ServerRoot = os.path.join(sys.prefix, "share", "akara")

    # To simplify, set this to the location of akara.conf
    ConfigRoot = "."

    #  PidFile: Filename which holds the process id of a running Akara
    #  server. The file is created when Akara starts and removed when it
    #  exists. It contains a single line: the pid followed by a newline.
    #
    PidFile = "logs/akara.pid"

    #  ModuleDir: directory containing the Akara extension modules
    #  Akara loads all of the *.py files in that directory
    #
    ModuleDir = "modules"

    # ModuleCache: directory containing the module cache databases.
    # Akara only creates such databases if the akara.caching
    # feature is being used to cache GET requests
    ModuleCache = "caches"

    ####
    #  Different options controlling the number of pre-forked server
    #  process to run at any one time.
    #
    #  MaxServers: maximum number of servers to run at any one time
    #  (this is therefore the maximum number of simultaneous connections)
    MaxServers           = 40
    #
    #  A 'spare' server is one which is waiting to handle an HTTP request
    #  MinSpareServers: minimum number of spare servers
    MinSpareServers      =  1
    #  MaxSpareServers: maximum number of spare servers
    MaxSpareServers      =  2

    #  MaxRequestsPerServer: restart a server after this many requests
    MaxRequestsPerServer =   5000


    #### Log configuration
    #  ErrorLog: The location of the error log file.
    #
    ErrorLog = "logs/error.log"

    #  AccessLog: The location of the access log file.
    #  Uses the Apache combined log format
    #
    AccessLog = "logs/access.log"

    #  LogLevel: Set the severity level for Akara logging messages.
    #  Messages below the given log level are not written. The levels are,
    #  from highest to lowest:
    #      CRITICAL, ERROR, WARN, INFO, DEBUG
    #  Also, WARNING is an alias for WARN
    #
    LogLevel = "${Akara__LogLevel}"

### Section 2: List of extension modules to install

# These are module names found on the Python path

MODULES = [
    "freemix_akara.contentdm",
    "freemix_akara.oai",
    "dplaingestion.oai",
    "dplaingestion.couch",
    "dplaingestion.create_fetcher",
    "dplaingestion.fetchers.fetcher",
    "dplaingestion.fetchers.oai_verbs_fetcher",
    "dplaingestion.fetchers.absolute_url_fetcher",
    "dplaingestion.fetchers.file_fetcher",
    "dplaingestion.fetchers.ia_fetcher",
    "dplaingestion.fetchers.mwdl_fetcher",
    "dplaingestion.fetchers.nypl_fetcher",
    "dplaingestion.fetchers.uva_fetcher",
    "dplaingestion.fetchers.nara_fetcher",
    "dplaingestion.fetchers.edan_fetcher",
    "dplaingestion.fetchers.hathi_fetcher",
    "dplaingestion.create_mapper",
    "dplaingestion.mappers.mapper",
    "dplaingestion.mappers.dublin_core_mapper",
    "dplaingestion.mappers.mods_mapper",
    "dplaingestion.mappers.harvard_mapper",
    "dplaingestion.mappers.bpl_mapper",
    "dplaingestion.mappers.mwdl_mapper",
    "dplaingestion.mappers.getty_mapper",
    "dplaingestion.mappers.primo_mapper",
    "dplaingestion.mappers.missouri_mapper",
    "dplaingestion.akamod.enrich",
    "dplaingestion.akamod.enrich-subject",
    "dplaingestion.akamod.enrich-type",
    "dplaingestion.akamod.enrich-format",
    "dplaingestion.akamod.enrich_date",
    "dplaingestion.akamod.select-id",
    "dplaingestion.akamod.shred",
    "dplaingestion.akamod.geocode",
    "dplaingestion.akamod.oai-set-name",
    "dplaingestion.akamod.dpla-list-records",
    "dplaingestion.akamod.dpla-list-sets",
    "dplaingestion.akamod.harvard_enrich_location",
    "dplaingestion.akamod.mdl-enrich-location",
    "dplaingestion.akamod.mwdl_enrich_location",
    "dplaingestion.akamod.nara_enrich_location",
    "dplaingestion.akamod.scdl_enrich_location",
    "dplaingestion.akamod.uiuc_enrich_location",
    "dplaingestion.akamod.scdl_geocode_regions",
    "dplaingestion.akamod.filter_empty_values",
    "dplaingestion.akamod.artstor_select_isshownat",
    "dplaingestion.akamod.artstor_identify_object",
    "dplaingestion.akamod.cdl_identify_object",
    "dplaingestion.akamod.contentdm_identify_object",
    "dplaingestion.akamod.move_date_values",
    "dplaingestion.akamod.enrich_location",
    "dplaingestion.akamod.remove_property",
    "dplaingestion.akamod.lookup",
    "dplaingestion.akamod.indiana_identify_object",
    "dplaingestion.akamod.kentucky_identify_object",
    "dplaingestion.akamod.georgia_identify_object",
    "dplaingestion.akamod.bhl_contributor_to_collection",
    "dplaingestion.akamod.copy_prop",
    "dplaingestion.akamod.cleanup_value",
    "dplaingestion.akamod.set_prop",
    "dplaingestion.akamod.enrich_language",
    "dplaingestion.akamod.mwdl_enrich_state_located_in",
    "dplaingestion.akamod.artstor_cleanup",
    "dplaingestion.akamod.nypl_identify_object",
    "dplaingestion.akamod.nypl_coll_title",
    "dplaingestion.akamod.nypl_select_hasview",
    "dplaingestion.akamod.mwdl_cleanup_field",
    "dplaingestion.akamod.ia_identify_object",
    "dplaingestion.akamod.ia_set_rights",
    "dplaingestion.akamod.dc_clean_invalid_dates",
    "dplaingestion.akamod.edan_select_id",
    "dplaingestion.akamod.dc_clean_invalid_dates",
    "dplaingestion.akamod.decode_html",
    "dplaingestion.akamod.artstor_spatial_to_dataprovider",
    "dplaingestion.akamod.david_rumsey_identify_object",
    "dplaingestion.akamod.dedup_value",
    "dplaingestion.akamod.set_type_from_physical_format",
    "dplaingestion.akamod.capitalize_value",
    "dplaingestion.akamod.artstor_cleanup_creator",
    "dplaingestion.akamod.replace_substring",
    "dplaingestion.akamod.uiuc_cleanup_spatial_name",
    "dplaingestion.akamod.remove_list_values",
    "dplaingestion.akamod.usc_enrich_location",
    "dplaingestion.akamod.hathi_identify_object",
    "dplaingestion.akamod.texas_enrich_location",
    "dplaingestion.akamod.set_spec_type",
    "dplaingestion.akamod.usc_set_dataprovider",
    "dplaingestion.akamod.compare_with_schema",
    "dplaingestion.akamod.mdl_state_located_in",
    "dplaingestion.akamod.scdl_format_to_type",
    "dplaingestion.marc_code_to_relator",
    "dplaingestion.akamod.dpla_mapper",
    "dplaingestion.akamod.set_context",
    "dplaingestion.akamod.strip_html"
    ]

### Section 3: Other module configuration goes here

class geocode: 
    twofishes_base_url = "${Twofishes__BaseUrl}"

class lookup:
    # Key is passed in query param.
    # Value is name of the dictionary from lookup module.
    lookup_mapping = {
        'test': 'test_subst',
        'test2': 'test_2_subst',
        'country_alias': 'COUNTRY_ALIAS',
        'iso639_3': 'iso639_3_subst',
        'scdl_fix_format': 'SCDL_FIX_FORMAT'
    }

class identify_object:
    IGNORE = 0
    PENDING = 1

class contentdm_identify_object(identify_object):
    pass

class indiana_identify_object(identify_object):
    pass

class kentucky_identify_object(identify_object):
    pass

class artstor_identify_object(identify_object):
    pass

class georgia_identify_object(identify_object):
    pass

class nypl_identify_object(identify_object):
    pass

class ia_identify_object(identify_object):
    pass

class david_rumsey_identify_object(identify_object):
    pass

class hathi_identify_object(identify_object):
    pass


class type_conversion:
    # Map of "format" or "physical description" substring to
    # sourceResource.type.  This format field is considered first, and these
    # values should be as specific as possible, to avoid false assignments,
    # because this field is usually pretty free-form, unlike the type fields.
    type_for_phys_keyword = [
        ('holiday card', 'image'),
        ('christmas card', 'image'),
        ('mail art', 'image'),
        ('postcard', 'image'),
    ]
    # Map of type-related substring to desired sourceResource.type.
    # For simple "if substr in str" matching.  Place more specific
    # patterns higher up, before more general ones.
    type_for_ot_keyword = [
        ('photograph', 'image'),
        ('still image', 'image'),
        ('sample book', 'image'),
        ('book', 'text'),
        ('specimen', 'image'),
        ('electronic resource', 'interactive resource'),
        # Keep "textile" above "text"
        ('textile', 'image'),
        ('text', 'text'),
        ('frame', 'image'),
        ('costume', 'image'),
        ('object', 'physical object'),
        ('statue', 'image'),
        ('sculpture', 'image'),
        ('container', 'image'),
        ('jewelry', 'image'),
        ('furnishing', 'image'),
        ('furniture', 'image'),
        # Keep "moving image" above "image"
        ('moving image', 'moving image'),
        # And, yes, "MovingImage" is a valid DC type.
        ('movingimage', 'moving image'),
        ('image', 'image'),
        ('drawing', 'image'),
        ('print', 'image'),
        ('painting', 'image'),
        ('illumination', 'image'),
        ('poster', 'image'),
        ('appliance', 'image'),
        ('tool', 'image'),
        ('electronic component', 'image'),
        ('publication', 'text'),
        ('magazine', 'text'),
        ('journal', 'text'),
        ('postcard', 'image'),
        ('correspondence', 'text'),
        ('writing', 'text'),
        ('manuscript', 'text'),
        # keep "equipment" above "audio" ("Audiovisual equipment")
        ('equipment', 'image'),
        ('cartographic', 'image'),
        ('notated music', 'image'),
        ('mixed material', ['image', 'text']),
        ('audio', 'sound'),
        ('sound', 'sound'),
        ('oral history recording', 'sound'),
        ('finding aid', 'collection'),
        ('online collection', 'collection'),
        ('online exhibit', 'interactive resource'),
        ('moving image', 'moving image'),
        ('motion picture', 'moving image'),
        ('film', 'moving image'),
        ('video game', 'interactive resource'),
        ('video', 'moving image')
    ]


class enrich_type(type_conversion):
    pass