nopaque/app/static/js/CorpusAnalysis/CQiClient.js

class CQiClient {
  constructor(corpusId) {
    this.socket = io(
      '/corpora/corpus/corpus_analysis',
      {
        auth: {corpus_id: corpusId},
        transports: ['websocket'],
        upgrade: false
      }
    );
    this.connected = false;
    this.corpora = new CQiCorpusCollection(this.socket);
  }

  connect() {
    return new Promise((resolve, reject) => {
      this.socket.emit('cqi.connect', response => {
        if (response.code === 200) {
          this.connected = true;
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  disconnect() {
    return new Promise((resolve, reject) => {
      this.socket.emit('cqi.disconnect', response => {
        if (response.code === 200) {
          this.connected = false;
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  ping() {
    return new Promise((resolve, reject) => {
      this.socket.emit('cqi.ping', response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }
}


class CQiCorpusCollection {
  constructor(socket) {
    this.socket = socket;
  }

  get(corpusName) {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: corpusName};

      this.socket.emit('cqi.corpora.get', args, response => {
        if (response.code === 200) {
          resolve(new CQiCorpus(this.socket, response.payload));
        } else {
          reject(response);
        }
      });
    });
  }

  list() {
    return new Promise((resolve, reject) => {
      this.socket.emit('cqi.corpora.list', response => {
        if (response.code === 200) {
          resolve(response.payload.map(x => {return new CQiSubcorpus(this.socket, x);}));
        } else {
          reject(response);
        }
      });
    });
  }
}


class CQiCorpus {
  constructor(socket, attrs) {
    this.socket = socket;
    this.charset = attrs.charset;
    this.name = attrs.name;
    this.properties = attrs.properties;
    this.size = attrs.size;
    this.alignmentAttributes = new CQiAlignmentAttributeCollection(this.socket, this);
    this.positionalAttributes = new CQiPositionalAttributeCollection(this.socket, this);
    this.structuralAttributes = new CQiStructuralAttributeCollection(this.socket, this);
    this.subcorpora = new CQiSubcorpusCollection(this.socket, this);
  }

  getVisualizationData() {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: this.name};

      this.socket.emit('cqi.corpora.corpus.get_visualization_data', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  getCorpusData() {
    return new Promise((resolve, reject) => {
      const dummyData = {
          "num_tokens": 2000,    // number of tokens in the corpus
          "num_unique_words": 500,    // number of unique words in the corpus
          "num_unique_lemmas": 200,    // number of unique lemmas in the corpus
          "num_sentences": 90,    // number of sentences in the corpus
          "average_sentence_length": 11,   // average number of tokens per sentence in the corpus
          "num_ent_types": 30,    // number of entities in the corpus
          "num_unique_ent_types":10,
          "ent_type_freqs": {
            "str": 10,    // number of ent_types with ent_type "str"
            // ...
        },
          "texts": [
              {
                  "num_tokens": 11,    // number of tokens in the text
                  "num_unique_words": 12,    // number of unique words in the text
                  "word_freqs": {    // frequency of unique words in the text (sorted by frequency)
                    "str": "int",    // number of tokens with word "str"
                    // ...
                  },
                  "num_unique_lemmas": 15,    // number of unique lemmas in the text
                  "lemma_freqs": {    // frequency of unique lemmas in the text (sorted by frequency)
                    "str": "int",    // number of tokens with lemma "str"
                    // ...
                  },
                  "num_sentences": 4,    // number of sentences in the text
                  "average_sentence_length": 3,   // average number of tokens per sentence in the text
                  "num_ent_types": 12,    // number of ent_types in the text
                  "num_unique_ent_types": 28,    // number of unique ent_types in the text
                  "num_entities_by_id": {
                      "1": "int",    // number of entities with id 1
                      // ...
                  },            
                  "author": "Author Name",
                  "title": "Titel",
                  "publishing_year": 1950
              },
              {
                "num_tokens": 15,    // number of tokens in the text
                "num_unique_words": 4,    // number of unique words in the text
                "word_freqs": {    // frequency of unique words in the text (sorted by frequency)
                  "str": "int",    // number of tokens with word "str"
                  // ...
                },
                "num_unique_lemmas": 90,    // number of unique lemmas in the text
                "lemma_freqs": {    // frequency of unique lemmas in the text (sorted by frequency)
                  "str": "int",    // number of tokens with lemma "str"
                  // ...
                },
                "num_sentences": 11,    // number of sentences in the text
                "average_sentence_length": 3,   // average number of tokens per sentence in the text
                "num_ent_types": 4,    // number of ent_types in the text
                "num_unique_ent_types": 300,    // number of unique ent_types in the text
                "num_entities_by_id": {
                    "1": "int",    // number of entities with id 1
                    // ...
                },            
                "author": "Author Name",
                "title": "Titel 1",
                  "publishing_year": 1962
              },
              {
                "num_tokens": 11,    // number of tokens in the text
                "num_unique_words": 12,    // number of unique words in the text
                "word_freqs": {    // frequency of unique words in the text (sorted by frequency)
                  "str": "int",    // number of tokens with word "str"
                  // ...
                },
                "num_unique_lemmas": 64,    // number of unique lemmas in the text
                "lemma_freqs": {    // frequency of unique lemmas in the text (sorted by frequency)
                  "str": "int",    // number of tokens with lemma "str"
                  // ...
                },
                "num_sentences": 52,    // number of sentences in the text
                "average_sentence_length": 3,   // average number of tokens per sentence in the text
                "num_ent_types": 45,    // number of ent_types in the text
                "num_unique_ent_types": 68,    // number of unique ent_types in the text
                "num_entities_by_id": {
                    "1": "int",    // number of entities with id 1
                    // ...
                },            
                "author": "Author Name",
                "title": "Titel 2",
                "publishing_year": 1850
              },
              {
                "num_tokens": 56,    // number of tokens in the text
                "num_unique_words": 13,    // number of unique words in the text
                "word_freqs": {    // frequency of unique words in the text (sorted by frequency)
                  "str": "int",    // number of tokens with word "str"
                  // ...
                },
                "num_unique_lemmas": 43,    // number of unique lemmas in the text
                "lemma_freqs": {    // frequency of unique lemmas in the text (sorted by frequency)
                  "str": "int",    // number of tokens with lemma "str"
                  // ...
                },
                "num_sentences": 45,    // number of sentences in the text
                "average_sentence_length": 56,   // average number of tokens per sentence in the text
                "num_ent_types": 8792,    // number of ent_types in the text
                "num_unique_ent_types": 56758,    // number of unique ent_types in the text
                "num_entities_by_id": {
                    "1": "int",    // number of entities with id 1
                    // ...
                },            
                "author": "Author Name",
                "title": "Titel 3",
                "publishing_year": 1504
              },
              {
                "num_tokens": 54345,    // number of tokens in the text
                "num_unique_words": 561,    // number of unique words in the text
                "word_freqs": {    // frequency of unique words in the text (sorted by frequency)
                  "str": "int",    // number of tokens with word "str"
                  // ...
                },
                "num_unique_lemmas": 546,    // number of unique lemmas in the text
                "lemma_freqs": {    // frequency of unique lemmas in the text (sorted by frequency)
                  "str": "int",    // number of tokens with lemma "str"
                  // ...
                },
                "num_sentences": 5427,    // number of sentences in the text
                "average_sentence_length": 657,   // average number of tokens per sentence in the text
                "num_ent_types": 3465,    // number of ent_types in the text
                "num_unique_ent_types": 45,    // number of unique ent_types in the text
                "num_entities_by_id": {
                    "1": "int",    // number of entities with id 1
                    // ...
                },            
                "author": "Author Name",
                "title": "Titel 4",
                "publishing_year": 1712
              },                            
              {
                "num_tokens": 4354,    // number of tokens in the text
                "num_unique_words": 45234,    // number of unique words in the text
                "word_freqs": {    // frequency of unique words in the text (sorted by frequency)
                  "testwort": 50,    // number of tokens with word "str"
                  "testwort2": 1
                },
                "num_unique_lemmas": 15,    // number of unique lemmas in the text
                "lemma_freqs": {    // frequency of unique lemmas in the text (sorted by frequency)
                  "testlemma": 11,    // number of tokens with lemma "str"
                  "testlemma2": 1
                },
                "num_sentences": 90,    // number of sentences in the text
                "average_sentence_length": 7,   // average number of tokens per sentence in the text
                "num_ent_types": 19,
                "num_unique_ent_types": 5,    // number of unique ent_types in the text
                "num_entities_by_id": {
                    "1": "int",    // number of entities with id 1
                    // ...
                },            
                "author": "Author Name 2",
                "title": "Titel 5",
                "publishing_year": 1951
              }
          ]
      };

      resolve(dummyData);
      /*
      const args = {corpus_name: this.name};

      this.socket.emit('cqi.corpora.corpus.getCorpusData', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
      */
    })
  }

  drop() {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: this.name};

      this.socket.emit('cqi.corpora.corpus.drop', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  query(subcorpus_name, queryString) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.name,
        subcorpus_name: subcorpus_name,
        query: queryString
      };

      this.socket.emit('cqi.corpora.corpus.query', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  // nopaque specific CQi extension
  paginate(page=1, perPage=20) {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: this.name, page: page, per_page: perPage};

      this.socket.emit('cqi.corpora.corpus.paginate', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  updateDb() {
    const args = {corpus_name: this.name};

    this.socket.emit('cqi.corpora.corpus.update_db', args);
  }
}


class CQiAlignmentAttributeCollection {
  constructor(socket, corpus) {
    this.corpus = corpus;
    this.socket = socket;
  }

  get(alignmentAttributeName) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        alignment_attribute_name: alignmentAttributeName
      };

      this.socket.emit('cqi.corpora.corpus.alignment_attributes.get', args, response => {
        if (response.code === 200) {
          resolve(new CQiAlignmentAttribute(this.socket, this.corpus, response.payload));
        } else {
          reject(response);
        }
      });
    });
  }

  list() {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: this.corpus.name};

      this.socket.emit('cqi.corpus.alignment_attributes.list', args, response => {
        if (response.code === 200) {
          resolve(response.payload.map(x => {return new CQiAlignmentAttribute(this.socket, this.corpus, x);}));
        } else {
          reject(response);
        }
      });
    });
  }
}


class CQiAlignmentAttribute {
  constructor(socket, corpus, attrs) {
    this.socket = socket;
    this.corpus = corpus;
    this.name = attrs.name;
    this.size = attrs.size;
  }
}


class CQiPositionalAttributeCollection {
  constructor(socket, corpus) {
    this.corpus = corpus;
    this.socket = socket;
  }

  get(positionalAttributeName) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        positional_attribute_name: positionalAttributeName
      };

      this.socket.emit('cqi.corpora.corpus.positional_attributes.get', args, response => {
        if (response.code === 200) {
          resolve(new CQiPositionalAttribute(this.socket, this.corpus, response.payload));
        } else {
          reject(response);
        }
      });
    });
  }

  list() {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: this.corpus.name};

      this.socket.emit('cqi.corpus.positional_attributes.list', args, response => {
        if (response.code === 200) {
          resolve(response.payload.map(x => {return new CQiPositionalAttribute(this.socket, this.corpus, x);}));
        } else {
          reject(response);
        }
      });
    });
  }
}


class CQiPositionalAttribute {
  constructor(socket, corpus, attrs) {
    this.socket = socket;
    this.corpus = corpus;
    this.lexiconSize = attrs.lexicon_size;
    this.name = attrs.name;
    this.size = attrs.size;
  }
}


class CQiStructuralAttributeCollection {
  constructor(socket, corpus) {
    this.corpus = corpus;
    this.socket = socket;
  }

  get(structuralAttributeName) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        structural_attribute_name: structuralAttributeName
      };

      this.socket.emit('cqi.corpora.corpus.structural_attributes.get', args, response => {
        if (response.code === 200) {
          resolve(new CQiStructuralAttribute(this.socket, this.corpus, response.payload));
        } else {
          reject(response);
        }
      });
    });
  }

  list() {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: this.corpus.name};

      this.socket.emit('cqi.corpus.structural_attributes.list', args, response => {
        if (response.code === 200) {
          resolve(response.payload.map(x => {return new CQiStructuralAttribute(this.socket, this.corpus, x);}));
        } else {
          reject(response);
        }
      });
    });
  }
}


class CQiStructuralAttribute {
  constructor(socket, corpus, attrs) {
    this.socket = socket;
    this.corpus = corpus;
    this.hasValues = attrs.has_values;
    this.name = attrs.name;
    this.size = attrs.size;
  }
}


class CQiSubcorpusCollection {
  constructor(socket, corpus) {
    this.corpus = corpus;
    this.socket = socket;
  }

  get(subcorpusName) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        subcorpus_name: subcorpusName
      };
      this.socket.emit('cqi.corpora.corpus.subcorpora.get', args, response => {
        if (response.code === 200) {
          resolve(new CQiSubcorpus(this.socket, this.corpus, response.payload));
        } else {
          reject(response);
        }
      });
    });
  }

  list() {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: this.corpus.name};

      this.socket.emit('cqi.corpora.corpus.subcorpora.list', args, response => {
        if (response.code === 200) {
          resolve(response.payload.map(x => {return new CQiSubcorpus(this.socket, this.corpus, x);}));
        } else {
          reject(response);
        }
      });
    });
  }
}


class CQiSubcorpus {
  constructor(socket, corpus, attrs) {
    this.socket = socket;
    this.corpus = corpus;
    this.fields = attrs.fields;
    this.name = attrs.name;
    this.size = attrs.size;
  }

  drop() {
    return new Promise((resolve, reject) => {
      const args = {corpus_name: this.corpus.name, subcorpus_name: this.name};

      this.socket.emit('cqi.corpora.corpus.subcorpora.subcorpus.drop', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  dump(field, first, last) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        subcorpus_name: this.name,
        field: field,
        first: first,
        last: last
      };

      this.socket.emit('cqi.corpora.corpus.subcorpora.subcorpus.dump', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  export(context=50) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        subcorpus_name: this.name,
        context: context
      };

      this.socket.emit('cqi.corpora.corpus.subcorpora.subcorpus.export', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  partial_export(matchIdList, context=50) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        subcorpus_name: this.name,
        match_id_list: matchIdList,
        context: context
      };

      this.socket.emit('cqi.corpora.corpus.subcorpora.subcorpus.partial_export', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  fdst_1(cutoff, field, attribute) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        subcorpus_name: this.name,
        cutoff: cutoff,
        field: field,
        attribute: attribute
      };

      this.socket.emit('cqi.corpora.corpus.subcorpora.subcorpus.fdist_1', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  fdst_2(cutoff, field1, attribute1, field2, attribute2) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        subcorpus_name: this.name,
        cutoff: cutoff,
        field1: field1,
        attribute1: attribute1,
        field2: field2,
        attribute2: attribute2
      };

      this.socket.emit('cqi.corpora.corpus.subcorpora.subcorpus.fdist_1', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }

  // nopaque specific CQi extension
  paginate(page=1, perPage=20, context=50) {
    return new Promise((resolve, reject) => {
      const args = {
        corpus_name: this.corpus.name,
        subcorpus_name: this.name,
        page: page,
        per_page: perPage,
        context: context
      };

      this.socket.emit('cqi.corpora.corpus.subcorpora.subcorpus.paginate', args, response => {
        if (response.code === 200) {
          resolve(response.payload);
        } else {
          reject(response);
        }
      });
    });
  }
}