Crawtext.yaml

######## example_param.init ######################################
##
############## Description du script lui-même pour l'utilisateur de ctManager, ainsi que les formats d'entrée/sorties ############################
##

script:
  name: crawl_trial
  desc: "Launch a crawl on the www"
  type: data collection
  author: jp cointet
  tags: [crawl, web]
  inputs:
   -
      type: text
      categ: records
      structure: isi	
  outputs:
    -
      type: sqlite
      categ: biblio
      structure: reseaulu

##
############## Paramètres du scripts (construction du formulaire utilisateur et variables renvoyées aux script) ############
##

params:
  # corpus_type:
  #    label: Corpus type
  #    type:
  #      widget: list
  #      multiple: no
  #    choices: ['isi','csv']#, 'pubmed', 'text','scopus']
  #    default: isi
  # year_fiel:
  #   label: 'If your csv file include a year entry, please indicate the column name' 
  #   type: string
  #   default: ""
  # corpus_origin_file:
  #    type: string
  #    default: ""
  # output_type:
  #   label: Output Format
  #   type:
  #     widget: list
  #     multiple: no
  #   choices: ['reseaulu']#, 'classique']
  #   default: reseaulu
  # reinit_db:
  #   label: Build a new database or update a previous one ?
  #   type:
  #     widget: list
  #     multiple: no
  #   choices: ['yes','no']
  #   default: 'yes'
  # project_name:
  #   type: string
  #   default: ""
  inlinks_min:
    label: 'Minimum number of incoming links to consider a webpage'
    type: string
    default: '2'
  depth:
    label: 'Maximum number of corpus extension steps'    
    type: string
    default: '5'
  query:
    label: 'Enter your query (boolean operator like "AND" is allowed)'    
    type: string
    default: 'You really should enter a query, otherwise...'
  project_name:
    type: string
    default: ""
  project_path:
    type: string
    default: ""
  max_pages_number:
    label: 'Enter the max number of pages to visit (caution: the corpus may be far smaller than this number)'
    type: string
    default: '10000'
#  result_path:

    # inlinks_min=parameters.get('inlinks_min',1)
    #  depth=parameters.get('depth',10)
    #  query=parameters.get('query','You really should enter a query, otherwise...')
    #  result_path=parameters.get('result_path','output')