Merge pull request #19 from wellcometrust/reorganise

Reorganise sub-module structure

deep_reference_parser/__init__.py

@@ -2,9 +2,9 @@
 # distracting on the command line. These lines here (while undesireable)
 # reduce the level of verbosity.
 
+import os
 import sys
 import warnings
-import os
 
 os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2"
 
@@ -19,20 +19,16 @@
 
 from .common import download_model_artefact
 from .deep_reference_parser import DeepReferenceParser
-from .logger import logger
-from .model_utils import get_config
-from .reference_utils import (
-    break_into_chunks,
-    labels_to_prodigy,
+from .io import (
     load_tsv,
-    prodigy_to_conll,
-    prodigy_to_lists,
     read_jsonl,
     read_pickle,
-    write_json,
     write_jsonl,
     write_pickle,
     write_to_csv,
-    write_txt,
+    write_tsv,
 )
+from .logger import logger
+from .model_utils import get_config
+from .reference_utils import break_into_chunks
 from .tokens_to_references import tokens_to_references

deep_reference_parser/__version__.py

@@ -1,5 +1,5 @@
 __name__ = "deep_reference_parser"
-__version__ = "2020.3.0"
+__version__ = "2020.3.1"
 __description__ = "Deep learning model for finding and parsing references"
 __url__ = "https://github.com/wellcometrust/deep_reference_parser"
 __author__ = "Wellcome Trust DataLabs Team"

deep_reference_parser/deep_reference_parser.py

@@ -47,7 +47,7 @@
     save_confusion_matrix,
     word2vec_embeddings,
 )
-from .reference_utils import load_tsv, read_pickle, write_pickle, write_to_csv
+from .io import load_tsv, read_pickle, write_pickle, write_to_csv
 
 
 class DeepReferenceParser:
@@ -456,7 +456,7 @@ def build_model(
 
         self.model = model
 
-        logger.debug(self.model.summary(line_length=150))
+#        logger.debug(self.model.summary(line_length=150))
 
     def train_model(
         self, epochs=25, batch_size=100, early_stopping_patience=5, metric="val_f1"

deep_reference_parser/io/__init__.py

@@ -1 +1,2 @@
-from .io import read_jsonl, write_jsonl
+from .io import (load_tsv, read_jsonl, read_pickle, write_jsonl, write_pickle,
+                 write_to_csv, write_tsv)

deep_reference_parser/io/io.py

@@ -6,9 +6,74 @@
 """
 
 import json
+import pickle
+import csv
+import os
+import pandas as pd
 
 from ..logger import logger
 
+def _split_list_by_linebreaks(tokens):
+    """Cycle through a list of tokens (or labels) and split them into lists
+    based on the presence of Nones or more likely math.nan caused by converting
+    pd.DataFrame columns to lists.
+    """
+    out = []
+    tokens_gen = iter(tokens)
+    while True:
+        try:
+            token = next(tokens_gen)
+            if isinstance(token, str) and token:
+                out.append(token)
+            else:
+                yield out
+                out = []
+        except StopIteration:
+            if out:
+                yield out
+            break
+
+def load_tsv(filepath, split_char="\t"):
+    """
+    Load and return the data stored in the given path.
+
+    Expects data in the following format (tab separations).
+
+      References   o       o
+                   o       o
+               1   o       o
+               .   o       o
+                   o       o
+             WHO   title   b-r
+       treatment   title   i-r
+      guidelines   title   i-r
+             for   title   i-r
+            drug   title   i-r
+               -   title   i-r
+       resistant   title   i-r
+    tuberculosis   title   i-r
+               ,   title   i-r
+            2016   title   i-r
+
+
+
+    Args:
+        filepath (str): Path to the data.
+        split_char(str): Character to be used to split each line of the
+            document.
+
+    Returns:
+        a series of lists depending on the number of label columns provided in 
+        filepath.
+
+    """
+
+    df = pd.read_csv(filepath, delimiter=split_char, header=None, skip_blank_lines=False)
+    out = [list(_split_list_by_linebreaks(column)) for _, column in df.iteritems()]
+
+    logger.info("Loaded %s training examples", len(out[0]))
+
+    return tuple(out)
 
 def write_jsonl(input_data, output_file):
     """
@@ -61,3 +126,71 @@ def read_jsonl(input_file):
     logger.debug("Read %s lines from %s", len(out), input_file)
 
     return out
+
+
+def write_to_csv(filename, columns, rows):
+    """
+    Create a .csv file from data given as columns and rows
+
+    Args:
+        filename(str): Path and name of the .csv file, without csv extension
+        columns(list): Columns of the csv file (First row of the file)
+        rows: Data to write into the csv file, given per row
+    """
+
+    with open(filename, "w") as csvfile:
+        wr = csv.writer(csvfile, quoting=csv.QUOTE_ALL)
+        wr.writerow(columns)
+
+        for i, row in enumerate(rows):
+            wr.writerow(row)
+    logger.info("Wrote results to %s", filename)
+
+
+def write_pickle(input_data, output_file, path=None):
+    """
+    Write an object to pickle
+
+    Args:
+        input_data(dict): A dict to be written to json.
+        output_file(str): A filename or path to which the json will be saved.
+        path(str): A string which will be prepended onto `output_file` with
+            `os.path.join()`. Obviates the need for lengthy `os.path.join`
+            statements each time this function is called.
+    """
+
+    if path:
+
+        output_file = os.path.join(path, output_file)
+
+    with open(output_file, "wb") as fb:
+        pickle.dump(input_data, fb)
+
+
+def read_pickle(input_file, path=None):
+    """Create a list from a jsonl file
+
+    Args:
+        input_file(str): File to be loaded.
+        path(str): A string which will be prepended onto `input_file` with
+            `os.path.join()`. Obviates the need for lengthy `os.path.join`
+            statements each time this function is called.
+    """
+
+    if path:
+        input_file = os.path.join(path, input_file)
+
+    with open(input_file, "rb") as fb:
+        out = pickle.load(fb)
+
+    logger.debug("Read data from %s", input_file)
+
+    return out
+
+def write_tsv(token_label_pairs, output_path):
+    """
+    Write tsv files to disk
+    """
+    with open(output_path, "w") as fb:
+        writer = csv.writer(fb, delimiter="\t")
+        writer.writerows(token_label_pairs)

deep_reference_parser/prodigy/__init__.py

@@ -6,3 +6,5 @@
 from .reach_to_prodigy import ReachToProdigy, reach_to_prodigy
 from .reference_to_token_annotations import TokenTagger, reference_to_token_annotations
 from .spacy_doc_to_prodigy import SpacyDocToProdigy
+from .misc import prodigy_to_conll
+from .labels_to_prodigy import labels_to_prodigy

deep_reference_parser/prodigy/labels_to_prodigy.py

@@ -0,0 +1,57 @@
+def labels_to_prodigy(tokens, labels):
+    """
+    Converts a list of tokens and labels like those used by Rodrigues et al,
+    and converts to prodigy format dicts.
+
+    Args:
+        tokens (list): A list of tokens.
+        labels (list): A list of labels relating to `tokens`.
+
+    Returns:
+        A list of prodigy format dicts containing annotated data.
+    """
+
+    prodigy_data = []
+
+    all_token_index = 0
+
+    for line_index, line in enumerate(tokens):
+        prodigy_example = {}
+
+        tokens = []
+        spans = []
+        token_start_offset = 0
+
+        for token_index, token in enumerate(line):
+
+            token_end_offset = token_start_offset + len(token)
+
+            tokens.append(
+                {
+                    "text": token,
+                    "id": token_index,
+                    "start": token_start_offset,
+                    "end": token_end_offset,
+                }
+            )
+
+            spans.append(
+                {
+                    "label": labels[line_index][token_index : token_index + 1][0],
+                    "start": token_start_offset,
+                    "end": token_end_offset,
+                    "token_start": token_index,
+                    "token_end": token_index,
+                }
+            )
+
+            prodigy_example["text"] = " ".join(line)
+            prodigy_example["tokens"] = tokens
+            prodigy_example["spans"] = spans
+            prodigy_example["meta"] = {"line": line_index}
+
+            token_start_offset = token_end_offset + 1
+
+        prodigy_data.append(prodigy_example)
+
+    return prodigy_data

deep_reference_parser/prodigy/misc.py

@@ -0,0 +1,38 @@
+import spacy
+
+
+def _join_prodigy_tokens(text):
+    """Return all prodigy tokens in a single string
+    """
+
+    return "\n".join([str(i) for i in text])
+
+
+def prodigy_to_conll(docs):
+    """
+    Expect list of jsons loaded from a jsonl
+    """
+
+    nlp = spacy.load("en_core_web_sm")
+    texts = [doc["text"] for doc in docs]
+    docs = list(nlp.tokenizer.pipe(texts))
+
+    out = [_join_prodigy_tokens(i) for i in docs]
+
+    out_str = "DOCSTART\n\n" + "\n\n".join(out)
+
+    return out_str
+
+
+def prodigy_to_lists(docs):
+    """
+    Expect list of jsons loaded from a jsonl
+    """
+
+    nlp = spacy.load("en_core_web_sm")
+    texts = [doc["text"] for doc in docs]
+    docs = list(nlp.tokenizer.pipe(texts))
+
+    out = [[str(token) for token in doc] for doc in docs]
+
+    return out

deep_reference_parser/prodigy/prodigy_to_tsv.py

@@ -20,7 +20,7 @@
 
 msg = Printer()
 
-ROWS_TO_PRINT=15
+ROWS_TO_PRINT = 15
 
 
 class TokenLabelPairs:
@@ -375,8 +375,6 @@ def prodigy_to_tsv(
 
     with open(output_file, "w") as fb:
         writer = csv.writer(fb, delimiter="\t")
-        # Write DOCSTART and a blank line
-        # writer.writerows([("DOCSTART", None), (None, None)])
         writer.writerows(merged_pairs)
 
     # Print out the first ten rows as a sense check