TomTBT commited on
Commit
09c2426
1 Parent(s): cc05f6a

Big change, reference idx and text inside text

Browse files
Files changed (1) hide show
  1. pmc_open_access_xml.py +48 -104
pmc_open_access_xml.py CHANGED
@@ -71,15 +71,15 @@ _SUBSETS = {
71
  }
72
  _BASELINE_DATE = "2022-09-03"
73
 
74
- REFS_KEYS = ["pmid_ref", "unknown_pub_ref", "figure_ref", "table_ref", "formula_ref", "box_ref", "code_ref",
75
  "quote_ref", "chem_ref", "supplementary_ref", "footnote_ref", "graphic_ref", "media_ref"]
76
  CONTENT_KEYS = ["introduction", "methods", "results", "discussion", "conclusion",
77
- "front", "body", "back", "figure", "table", "formula", "box",
78
  "code", "quote", "chem", "supplementary", "footnote"]
79
  begin_doc_rgx = re.compile("""<!DOCTYPE.*""")
80
  def clean_raw(xml_text):
81
  """
82
- Fixes the formating of xml of files and returns it.
83
  Some have bad formating but they can be fixed/improved
84
  """
85
  #Some XML can't be parsed because they are not starting with the DOCTYPE declaration
@@ -93,16 +93,16 @@ def clean_raw(xml_text):
93
  return xml_text
94
 
95
  # Tag name to "reference type" linking
96
- TAG_DIC = {"fig":(" ##FIG## ","figure_ref"), "table-wrap":(" ##TAB## ","table_ref"),
97
- "array":(" ##TAB## ","table_ref"), "boxed-text":(" ##BOX## ","box_ref"),
98
- "graphic":(" ##GRAPH## ","graphic_ref"), "inline-graphic":(" ##GRAPH## ","graphic_ref"),
99
- "media":(" ##MEDIA## ","media_ref"), "inline-media":(" ##MEDIA## ","media_ref"),
100
- "disp-formula":(" ##FORMU## ","formula_ref"), "inline-formula":(" ##FORMU## ","formula_ref"),
101
- "table-wrap-foot":(" ##FOOTN## ","footnote_ref"), "fn-group":(" ##FOOTN## ","footnote_ref"),
102
- "code":(" ##CODE## ","code_ref"), "chem-struct-wrap":(" ##CHEM## ","chem_ref"),
103
- "disp-quote":(" ##QUOTE## ","quote_ref"), "speech":(" ##QUOTE## ","quote_ref"),
104
- "supplementary-material":(" ##SUPPL## ","supplementary_ref"),
105
- "inline-supplementary-material":(" ##SUPPL## ","supplementary_ref")}
106
 
107
  def get_ref_indexes(ref_el_l, refs_pmid, refs_nonpmid_keys):
108
  """
@@ -113,9 +113,9 @@ def get_ref_indexes(ref_el_l, refs_pmid, refs_nonpmid_keys):
113
  count_ref_d = defaultdict(lambda:0)
114
  reference_d = {}
115
  for k, v in refs_pmid.items():
116
- reference_d[k] = (v, " ##REF## ", "pmid_ref")
117
  for i, k in enumerate(refs_nonpmid_keys):
118
- reference_d[k] = (i, " ##UREF## ", "unknown_pub_ref")
119
 
120
  refs_key_l = []
121
  for el in ref_el_l:
@@ -133,15 +133,14 @@ def parseout_el_refs(el, rids):
133
  Removes xml namespace from the text for storage savings, such as:
134
  - xmlns:xlink="http://www.w3.org/1999/xlink"
135
  - xmlns:mml="http://www.w3.org/1998/Math/MathML"
136
-
137
  Extract then from the text all the references founds to the rids dictionnary,
138
- and replace them by keywords of the corresponding family (eg " ##FIG## " for a figure,
139
- " ##TAB## " for a table, or " ##MATHS## " for mathematical formulas)
140
 
141
  The range reference (e.g. 1-3 or 15-17) are replaced by their range (1,2,3 or 15,16,17)
142
 
143
- Returns the parsed text, the identifiers for the references and the references text that
144
- were replaced by the keywords. (eg, "Figure 2" was a hypertext reference and got replaced by " ##FIG## ")
145
  """
146
  for xref in el.xpath(".//xref"):
147
  inner_text = "".join(xref.itertext())
@@ -172,22 +171,22 @@ def parseout_el_refs(el, rids):
172
  #### START HANDLING REF RANGE ########
173
  try:
174
  if has_ref_range is None:
175
- if ref_kword in [" ##UREF## ", " ##REF## "]: # Otherwise it's a year
176
  has_ref_range = res_reftext[ref_class][-1].isnumeric() and int(res_reftext[ref_class][-1]) < 500
177
 
178
- if has_ref_range and ref_kword in [" ##UREF## ", " ##REF## "]:
179
  if tail=="-":
180
  ref_rstart = int(res_reftext[ref_class][-1])
181
  tail = ", "
182
  elif ref_rstart is not None:
183
  ref_rstop = int(res_reftext[ref_class][-1])
184
- new_ref_kwords = [ref_kword]
185
  for i in range(ref_rstart+1, ref_rstop):
186
  new_rid = re.sub(str(ref_rstop), str(i), rid, count=1)
187
  ref_idx_, ref_kword_, ref_class_ = rids[new_rid]
188
  res_rid[ref_class_].insert(-1, ref_idx_)
189
  res_reftext[ref_class_].insert(-1, str(i))
190
- new_ref_kwords.insert(-1, ref_kword_)
191
  ref_kword = ", ".join(new_ref_kwords)
192
  ref_rstart = None
193
  except (KeyError, ValueError):
@@ -197,9 +196,9 @@ def parseout_el_refs(el, rids):
197
 
198
  prev_el = xref.getprevious()
199
  if prev_el is None:
200
- parent.text = "".join([(parent.text if parent.text else ""), ref_kword, tail])
201
  else:
202
- prev_el.tail = "".join([(prev_el.tail if prev_el.tail else ""), ref_kword, tail])
203
  parent.remove(xref)
204
 
205
  text = etree.tostring(el, with_tail=False, encoding='unicode', method='xml')
@@ -212,7 +211,7 @@ def parseout_el_refs(el, rids):
212
 
213
  text = "".join([tag_txt, text[tag_start:]])
214
 
215
- return text, res_rid, res_reftext
216
 
217
 
218
  def get_references(article_tree):
@@ -226,7 +225,7 @@ def get_references(article_tree):
226
  references_nonpmid = []
227
  references_nonpmid_keys = []
228
  refs = article_tree.find(".//ref-list")
229
- if refs is None: #Some don't have any references
230
  return {}, [], []
231
  refs = refs.findall("ref")
232
  for i, ref in enumerate(refs):
@@ -236,10 +235,10 @@ def get_references(article_tree):
236
  pmid = int(pubid.text)
237
  break
238
  if pmid is not None and pmid<100000000:
239
- #In an article (oa_comm:PMC2679651), broken PMID were found (>10e9).
240
  #May be several of those. Not sure what to do with them, and what threshold to use
241
  #Keeping them would result in loosing info about the reference (article title, authors, ...)
242
-
243
  #Only the PMID is kept, as it links to the documents in pubmed abstract dataset.
244
  references_pmid[ref.attrib["id"]] = str(pmid)
245
  else:
@@ -272,7 +271,7 @@ def construct_datadict(article_tree):
272
 
273
  Useful information about the tags can be found here: https://jats.nlm.nih.gov/archiving/tag-library/1.3/
274
  """
275
- res_content_d, res_reference_d, res_reference_text_d = {}, defaultdict(dict), defaultdict(dict)
276
 
277
  refs_pmid, refs_nonpmid, refs_nonpmid_keys = get_references(article_tree)
278
  reference_count = len(refs_pmid)+len(refs_nonpmid)
@@ -305,7 +304,7 @@ def construct_datadict(article_tree):
305
  # After testing, no question were found in the dataset, so I commented that part
306
  # question_l = []
307
  # for el in article_tree.xpath(".//question-preamble|.//question|.//answer|.//explanation"):
308
- # text, _, _ = parseout_el_refs(el, {})
309
  # question_l.append(text)
310
  # res_content_d["question"] = "\n".join(question_l)
311
  # for el in article_tree.xpath(".//question-wrap-group|.//question-wrap|.//answer-set|.//explanation"):
@@ -319,16 +318,14 @@ def construct_datadict(article_tree):
319
  |.//table-wrap-foot|.//fn-group|.//chem-struct-wrap\
320
  |.//code|.//disp-quote|.//speech")
321
  rids, key_l = get_ref_indexes(ref_el_l, refs_pmid, refs_nonpmid_keys)
322
- text_l_d, refs_l_d, refs_text_l_d = defaultdict(list),defaultdict(list),defaultdict(list)
323
  for el, key in zip(ref_el_l[::-1], key_l[::-1]):
324
  #The iteration is done backward to always process first the most inner reference,
325
  # it makes the processing is agnostic to structure rules differences between articles
326
- new_text, new_xref_id, new_xref_text = parseout_el_refs(el, rids)
327
 
328
  ref_class = rids[key][2]
329
  text_l_d[ref_class].insert(0, new_text)
330
- refs_l_d[ref_class].insert(0, new_xref_id)
331
- refs_text_l_d[ref_class].insert(0, new_xref_text)
332
 
333
  repl_xref = etree.Element("xref", attrib={"rid":key})
334
  repl_xref.tail = el.tail
@@ -338,52 +335,30 @@ def construct_datadict(article_tree):
338
  # Finally, the discovered references and text are added to the result
339
  for ref_k in REFS_KEYS[2:]: #Slicing from 2, to not add pmid and unknown ref here
340
  res_content_d[ref_k[:-4]] = text_l_d[ref_k]#"\n".join(text_l_d[ref_k])
341
- if ref_k in ["graphic_ref", "media_ref"]: #They don't take references
342
- continue
343
- for ref_k2 in REFS_KEYS:
344
- tmp_l = [refs_d[ref_k2] for refs_d in refs_l_d[ref_k]]
345
- res_reference_d[ref_k[:-4]][ref_k2] = list(chain(*tmp_l)) # [:-4] slicing to remove the "_ref" part
346
- tmp_l = [refs_d[ref_k2] for refs_d in refs_text_l_d[ref_k]]
347
- res_reference_text_d[ref_k[:-4]][ref_k2] = list(chain(*tmp_l))
348
-
349
- def add_part_to_dics(part, text_l, refs_l, ref_texts_l):
350
- """Repeated code to add various parts to the document"""
351
- res_content_d[part] = text_l #"\n".join(text_l)
352
- for ref_k in REFS_KEYS:
353
- tmp_l = [refs_d[ref_k] for refs_d in refs_l]
354
- res_reference_d[part][ref_k] = list(chain(*tmp_l))
355
- tmp_l = [refs_d[ref_k] for refs_d in ref_texts_l]
356
- res_reference_text_d[part][ref_k] = list(chain(*tmp_l))
357
-
358
- path_l, text_l, refs_l, refs_text_l = [], [], [], []
359
  t_paths, t_texts_lowcase = [], []
360
  for part in ["front", "body", "back"]: #Iterate parts and insert first front and back
361
- tmp_path_l, tmp_text_l, tmp_refs_l, tmp_refs_text_l = [], [], [], []
362
  tmp_t_paths, tmp_t_texts_lowcase = [], []
363
  part_el = article_tree.find(".//"+part)
364
  if part_el is None:
365
- res_content_d[part] = []#""
366
- for target_key in REFS_KEYS:
367
- res_reference_d[part][target_key] = []
368
- res_reference_text_d[part][target_key] = []
369
  continue
370
  #Only the outermost p are kept, to prevent duplication.
371
  #Also seen title with p inside. not(ancestor::title) prevents duplication of that p
372
  for el in part_el.xpath(".//p[not(ancestor::p) and not(ancestor::title)]| .//title[not(ancestor::p) and not(ancestor::title)]"):
373
- new_text, new_xref_id, new_xref_text = parseout_el_refs(el, rids)
374
  tmp_path_l.append(article_tree.getelementpath(el))
375
  tmp_text_l.append(new_text)
376
- tmp_refs_l.append(new_xref_id)
377
- tmp_refs_text_l.append(new_xref_text)
378
  if el.tag=="title":
379
  tmp_t_paths.append(tmp_path_l[-1])
380
- tmp_t_texts_lowcase.append(new_text.lower())
381
- if part=="body": #We keep the body for processing right bellow.
382
  path_l, text_l = tmp_path_l, tmp_text_l
383
- refs_l, refs_text_l = tmp_refs_l, tmp_refs_text_l
384
  t_paths, t_texts_lowcase = tmp_t_paths, tmp_t_texts_lowcase
385
  else:
386
- add_part_to_dics(part, tmp_text_l, tmp_refs_l, tmp_refs_text_l)
387
 
388
  # Figuring from the titles which are the different categories
389
  mask_intro = np.array(["introduction" in t_text or "background" in t_text for t_text in t_texts_lowcase]).astype(bool)
@@ -395,10 +370,7 @@ def construct_datadict(article_tree):
395
  for mask, name_section in zip([mask_intro, mask_metho, mask_resul, mask_discu, mask_concl],
396
  ["introduction", "methods", "results", "discussion", "conclusion"]):
397
  if not np.any(mask):
398
- res_content_d[name_section] = []#""
399
- for target_key in REFS_KEYS:
400
- res_reference_d[name_section][target_key] = []
401
- res_reference_text_d[name_section][target_key] = []
402
  continue
403
 
404
  filtered_path_l = list(compress(t_paths, mask))
@@ -407,17 +379,12 @@ def construct_datadict(article_tree):
407
  root_path = root_path[:root_path.rindex("/")]
408
  mask_contents = np.array([path.startswith(root_path) for path in path_l]).astype(bool)
409
  processed_mask |= mask_contents
410
- add_part_to_dics(name_section, list(compress(text_l, mask_contents)),
411
- list(compress(refs_l, mask_contents)), list(compress(refs_text_l, mask_contents)))
412
 
413
  processed_mask = ~processed_mask #Finally, add the body part as everything that don't belong to previous categories
414
- add_part_to_dics("body", list(compress(text_l, processed_mask)),
415
- list(compress(refs_l, processed_mask)), list(compress(refs_text_l, processed_mask)))
416
-
417
- res_reference_d = dict(res_reference_d)
418
- res_reference_text_d = dict(res_reference_text_d)
419
 
420
- return (res_content_d, res_reference_d, res_reference_text_d, reference_count)
421
 
422
  class OpenAccessXMLConfig(datasets.BuilderConfig):
423
  """BuilderConfig for the PMC Open Access Subset."""
@@ -456,7 +423,7 @@ class OpenAccessXML(datasets.GeneratorBasedBuilder):
456
  "results": datasets.features.Sequence(datasets.Value("string")),
457
  "discussion": datasets.features.Sequence(datasets.Value("string")),
458
  "conclusion": datasets.features.Sequence(datasets.Value("string")),
459
-
460
  "front": datasets.features.Sequence(datasets.Value("string")),
461
  "body": datasets.features.Sequence(datasets.Value("string")),
462
  "back": datasets.features.Sequence(datasets.Value("string")),
@@ -478,25 +445,6 @@ class OpenAccessXML(datasets.GeneratorBasedBuilder):
478
  "glossary": datasets.features.Sequence(
479
  {"acronym": datasets.Value("string"), "definition": datasets.Value("string")}
480
  ),
481
-
482
- "references": {k_cont:{k_ref:datasets.features.Sequence(datasets.Value("string" if k_ref=="pmid_ref" else "int32")) for k_ref in REFS_KEYS} for k_cont in CONTENT_KEYS},
483
- "references_text": {k_cont:{k_ref:datasets.features.Sequence(datasets.Value("string")) for k_ref in REFS_KEYS} for k_cont in CONTENT_KEYS},
484
- # -> With the 2 level dict, each item looks like this:
485
- # "introduction":{"pmid_ref": datasets.features.Sequence(datasets.Value("string")),
486
- # "unknown_pub_ref": datasets.features.Sequence(datasets.Value("string")),
487
- # "figure_ref": datasets.features.Sequence(datasets.Value("string")),
488
- # "table_ref": datasets.features.Sequence(datasets.Value("string")),
489
- # "formula_ref": datasets.features.Sequence(datasets.Value("string")),
490
- # "box_ref": datasets.features.Sequence(datasets.Value("string")),
491
- # "code_ref": datasets.features.Sequence(datasets.Value("string")),
492
- # "quote_ref": datasets.features.Sequence(datasets.Value("string")),
493
- # "chem_ref": datasets.features.Sequence(datasets.Value("string")),
494
- # "supplementary_ref": datasets.features.Sequence(datasets.Value("string")),
495
- # "footnote_ref": datasets.features.Sequence(datasets.Value("string")),
496
- # "graphic_ref": datasets.features.Sequence(datasets.Value("string")),
497
- # "media_ref": datasets.features.Sequence(datasets.Value("string")),
498
- # },
499
-
500
  "n_references": datasets.Value("int32"),
501
  "license": datasets.Value("string"),
502
  "retracted": datasets.Value("string"),
@@ -606,7 +554,7 @@ class OpenAccessXML(datasets.GeneratorBasedBuilder):
606
  except etree.XMLSyntaxError: #In some files, xml is broken
607
  continue
608
 
609
- content_d, reference_d, reference_text_d, n_ref = construct_datadict(article_tree)
610
  glossary = np.array([[k,v] for k,v in content_d["glossary"].items()])
611
  data = {
612
  "introduction": content_d["introduction"],
@@ -630,8 +578,6 @@ class OpenAccessXML(datasets.GeneratorBasedBuilder):
630
  "media": content_d["media"],
631
  # "question": content_d["question"],
632
  "unknown_pub": content_d["unknown_pub"],
633
- "references": reference_d,
634
- "references_text": reference_text_d,
635
  "glossary": {"acronym":glossary[:,0], "definition":glossary[:,1]} if len(glossary)>0 else {"acronym":[], "definition":[]},
636
  "n_references": n_ref,
637
  "pmid": data["PMID"],
@@ -670,7 +616,7 @@ class OpenAccessXML(datasets.GeneratorBasedBuilder):
670
  except etree.XMLSyntaxError: #In some files, xml is broken
671
  continue
672
 
673
- content_d, reference_d, reference_text_d, n_ref = construct_datadict(article_tree)
674
  glossary = np.array([[k,v] for k,v in content_d["glossary"].items()])
675
  data = {
676
  "introduction": content_d["introduction"],
@@ -694,8 +640,6 @@ class OpenAccessXML(datasets.GeneratorBasedBuilder):
694
  "media": content_d["media"],
695
  # "question": content_d["question"],
696
  "unknown_pub": content_d["unknown_pub"],
697
- "references": reference_d,
698
- "references_text": reference_text_d,
699
  "glossary": {"acronym":glossary[:,0], "definition":glossary[:,1]} if len(glossary)>0 else {"acronym":[], "definition":[]},
700
  "n_references": n_ref,
701
  "pmid": data["PMID"],
 
71
  }
72
  _BASELINE_DATE = "2022-09-03"
73
 
74
+ REFS_KEYS = ["pmid_ref", "unknown_pub_ref", "figure_ref", "table_ref", "formula_ref", "box_ref", "code_ref",
75
  "quote_ref", "chem_ref", "supplementary_ref", "footnote_ref", "graphic_ref", "media_ref"]
76
  CONTENT_KEYS = ["introduction", "methods", "results", "discussion", "conclusion",
77
+ "front", "body", "back", "figure", "table", "formula", "box",
78
  "code", "quote", "chem", "supplementary", "footnote"]
79
  begin_doc_rgx = re.compile("""<!DOCTYPE.*""")
80
  def clean_raw(xml_text):
81
  """
82
+ Fixes the formating of xml of files and returns it.
83
  Some have bad formating but they can be fixed/improved
84
  """
85
  #Some XML can't be parsed because they are not starting with the DOCTYPE declaration
 
93
  return xml_text
94
 
95
  # Tag name to "reference type" linking
96
+ TAG_DIC = {"fig":("FIG","figure_ref"), "table-wrap":("TAB","table_ref"),
97
+ "array":("TAB","table_ref"), "boxed-text":("BOX","box_ref"),
98
+ "graphic":("GRAPH","graphic_ref"), "inline-graphic":("GRAPH","graphic_ref"),
99
+ "media":("MEDIA","media_ref"), "inline-media":("MEDIA","media_ref"),
100
+ "disp-formula":("FORMU","formula_ref"), "inline-formula":("FORMU","formula_ref"),
101
+ "table-wrap-foot":("FOOTN","footnote_ref"), "fn-group":("FOOTN","footnote_ref"),
102
+ "code":("CODE","code_ref"), "chem-struct-wrap":("CHEM","chem_ref"),
103
+ "disp-quote":("QUOTE","quote_ref"), "speech":("QUOTE","quote_ref"),
104
+ "supplementary-material":("SUPPL","supplementary_ref"),
105
+ "inline-supplementary-material":("SUPPL","supplementary_ref")}
106
 
107
  def get_ref_indexes(ref_el_l, refs_pmid, refs_nonpmid_keys):
108
  """
 
113
  count_ref_d = defaultdict(lambda:0)
114
  reference_d = {}
115
  for k, v in refs_pmid.items():
116
+ reference_d[k] = (v, "REF", "pmid_ref")
117
  for i, k in enumerate(refs_nonpmid_keys):
118
+ reference_d[k] = (i, "UREF", "unknown_pub_ref")
119
 
120
  refs_key_l = []
121
  for el in ref_el_l:
 
133
  Removes xml namespace from the text for storage savings, such as:
134
  - xmlns:xlink="http://www.w3.org/1999/xlink"
135
  - xmlns:mml="http://www.w3.org/1998/Math/MathML"
136
+
137
  Extract then from the text all the references founds to the rids dictionnary,
138
+ and replace them by keywords of the corresponding family (eg "##FIG##4##Doe 2022##" for a figure,
139
+ "##TAB##0##Table 1##" for a table, or "##MATHS##1##(2)##" for mathematical formulas)
140
 
141
  The range reference (e.g. 1-3 or 15-17) are replaced by their range (1,2,3 or 15,16,17)
142
 
143
+ Returns the parsed text
 
144
  """
145
  for xref in el.xpath(".//xref"):
146
  inner_text = "".join(xref.itertext())
 
171
  #### START HANDLING REF RANGE ########
172
  try:
173
  if has_ref_range is None:
174
+ if ref_kword in ["UREF", "REF"]: # Otherwise it's a year
175
  has_ref_range = res_reftext[ref_class][-1].isnumeric() and int(res_reftext[ref_class][-1]) < 500
176
 
177
+ if has_ref_range and ref_kword in ["UREF", "REF"]:
178
  if tail=="-":
179
  ref_rstart = int(res_reftext[ref_class][-1])
180
  tail = ", "
181
  elif ref_rstart is not None:
182
  ref_rstop = int(res_reftext[ref_class][-1])
183
+ new_ref_kwords = [f"##{ref_kword}##{ref_idx}##{inner_text}##"]
184
  for i in range(ref_rstart+1, ref_rstop):
185
  new_rid = re.sub(str(ref_rstop), str(i), rid, count=1)
186
  ref_idx_, ref_kword_, ref_class_ = rids[new_rid]
187
  res_rid[ref_class_].insert(-1, ref_idx_)
188
  res_reftext[ref_class_].insert(-1, str(i))
189
+ new_ref_kwords.insert(-1, f"##{ref_kword_}##{ref_idx_}##{str(i)}##")
190
  ref_kword = ", ".join(new_ref_kwords)
191
  ref_rstart = None
192
  except (KeyError, ValueError):
 
196
 
197
  prev_el = xref.getprevious()
198
  if prev_el is None:
199
+ parent.text = "".join([(parent.text if parent.text else ""), f"##{ref_kword}##{ref_idx}##{inner_text}##", tail])
200
  else:
201
+ prev_el.tail = "".join([(prev_el.tail if prev_el.tail else ""), f"##{ref_kword}##{ref_idx}##{inner_text}##", tail])
202
  parent.remove(xref)
203
 
204
  text = etree.tostring(el, with_tail=False, encoding='unicode', method='xml')
 
211
 
212
  text = "".join([tag_txt, text[tag_start:]])
213
 
214
+ return text
215
 
216
 
217
  def get_references(article_tree):
 
225
  references_nonpmid = []
226
  references_nonpmid_keys = []
227
  refs = article_tree.find(".//ref-list")
228
+ if refs is None: #Some don't have any references
229
  return {}, [], []
230
  refs = refs.findall("ref")
231
  for i, ref in enumerate(refs):
 
235
  pmid = int(pubid.text)
236
  break
237
  if pmid is not None and pmid<100000000:
238
+ #In an article (oa_comm:PMC2679651), broken PMID were found (>10e9).
239
  #May be several of those. Not sure what to do with them, and what threshold to use
240
  #Keeping them would result in loosing info about the reference (article title, authors, ...)
241
+
242
  #Only the PMID is kept, as it links to the documents in pubmed abstract dataset.
243
  references_pmid[ref.attrib["id"]] = str(pmid)
244
  else:
 
271
 
272
  Useful information about the tags can be found here: https://jats.nlm.nih.gov/archiving/tag-library/1.3/
273
  """
274
+ res_content_d = {}
275
 
276
  refs_pmid, refs_nonpmid, refs_nonpmid_keys = get_references(article_tree)
277
  reference_count = len(refs_pmid)+len(refs_nonpmid)
 
304
  # After testing, no question were found in the dataset, so I commented that part
305
  # question_l = []
306
  # for el in article_tree.xpath(".//question-preamble|.//question|.//answer|.//explanation"):
307
+ # text = parseout_el_refs(el, {})
308
  # question_l.append(text)
309
  # res_content_d["question"] = "\n".join(question_l)
310
  # for el in article_tree.xpath(".//question-wrap-group|.//question-wrap|.//answer-set|.//explanation"):
 
318
  |.//table-wrap-foot|.//fn-group|.//chem-struct-wrap\
319
  |.//code|.//disp-quote|.//speech")
320
  rids, key_l = get_ref_indexes(ref_el_l, refs_pmid, refs_nonpmid_keys)
321
+ text_l_d = defaultdict(list)
322
  for el, key in zip(ref_el_l[::-1], key_l[::-1]):
323
  #The iteration is done backward to always process first the most inner reference,
324
  # it makes the processing is agnostic to structure rules differences between articles
325
+ new_text = parseout_el_refs(el, rids)
326
 
327
  ref_class = rids[key][2]
328
  text_l_d[ref_class].insert(0, new_text)
 
 
329
 
330
  repl_xref = etree.Element("xref", attrib={"rid":key})
331
  repl_xref.tail = el.tail
 
335
  # Finally, the discovered references and text are added to the result
336
  for ref_k in REFS_KEYS[2:]: #Slicing from 2, to not add pmid and unknown ref here
337
  res_content_d[ref_k[:-4]] = text_l_d[ref_k]#"\n".join(text_l_d[ref_k])
338
+
339
+ path_l, text_l = [], []
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
340
  t_paths, t_texts_lowcase = [], []
341
  for part in ["front", "body", "back"]: #Iterate parts and insert first front and back
342
+ tmp_path_l, tmp_text_l = [], []
343
  tmp_t_paths, tmp_t_texts_lowcase = [], []
344
  part_el = article_tree.find(".//"+part)
345
  if part_el is None:
346
+ res_content_d[part] = []
 
 
 
347
  continue
348
  #Only the outermost p are kept, to prevent duplication.
349
  #Also seen title with p inside. not(ancestor::title) prevents duplication of that p
350
  for el in part_el.xpath(".//p[not(ancestor::p) and not(ancestor::title)]| .//title[not(ancestor::p) and not(ancestor::title)]"):
351
+ new_text = parseout_el_refs(el, rids)
352
  tmp_path_l.append(article_tree.getelementpath(el))
353
  tmp_text_l.append(new_text)
 
 
354
  if el.tag=="title":
355
  tmp_t_paths.append(tmp_path_l[-1])
356
+ tmp_t_texts_lowcase.append(new_text.lower())
357
+ if part=="body": #We keep the body for processing right bellow.
358
  path_l, text_l = tmp_path_l, tmp_text_l
 
359
  t_paths, t_texts_lowcase = tmp_t_paths, tmp_t_texts_lowcase
360
  else:
361
+ res_content_d[part] = tmp_text_l
362
 
363
  # Figuring from the titles which are the different categories
364
  mask_intro = np.array(["introduction" in t_text or "background" in t_text for t_text in t_texts_lowcase]).astype(bool)
 
370
  for mask, name_section in zip([mask_intro, mask_metho, mask_resul, mask_discu, mask_concl],
371
  ["introduction", "methods", "results", "discussion", "conclusion"]):
372
  if not np.any(mask):
373
+ res_content_d[name_section] = []
 
 
 
374
  continue
375
 
376
  filtered_path_l = list(compress(t_paths, mask))
 
379
  root_path = root_path[:root_path.rindex("/")]
380
  mask_contents = np.array([path.startswith(root_path) for path in path_l]).astype(bool)
381
  processed_mask |= mask_contents
382
+ res_content_d[name_section] = list(compress(text_l, mask_contents))
 
383
 
384
  processed_mask = ~processed_mask #Finally, add the body part as everything that don't belong to previous categories
385
+ res_content_d["body"] = list(compress(text_l, processed_mask))
 
 
 
 
386
 
387
+ return (res_content_d, reference_count)
388
 
389
  class OpenAccessXMLConfig(datasets.BuilderConfig):
390
  """BuilderConfig for the PMC Open Access Subset."""
 
423
  "results": datasets.features.Sequence(datasets.Value("string")),
424
  "discussion": datasets.features.Sequence(datasets.Value("string")),
425
  "conclusion": datasets.features.Sequence(datasets.Value("string")),
426
+
427
  "front": datasets.features.Sequence(datasets.Value("string")),
428
  "body": datasets.features.Sequence(datasets.Value("string")),
429
  "back": datasets.features.Sequence(datasets.Value("string")),
 
445
  "glossary": datasets.features.Sequence(
446
  {"acronym": datasets.Value("string"), "definition": datasets.Value("string")}
447
  ),
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
448
  "n_references": datasets.Value("int32"),
449
  "license": datasets.Value("string"),
450
  "retracted": datasets.Value("string"),
 
554
  except etree.XMLSyntaxError: #In some files, xml is broken
555
  continue
556
 
557
+ content_d, n_ref = construct_datadict(article_tree)
558
  glossary = np.array([[k,v] for k,v in content_d["glossary"].items()])
559
  data = {
560
  "introduction": content_d["introduction"],
 
578
  "media": content_d["media"],
579
  # "question": content_d["question"],
580
  "unknown_pub": content_d["unknown_pub"],
 
 
581
  "glossary": {"acronym":glossary[:,0], "definition":glossary[:,1]} if len(glossary)>0 else {"acronym":[], "definition":[]},
582
  "n_references": n_ref,
583
  "pmid": data["PMID"],
 
616
  except etree.XMLSyntaxError: #In some files, xml is broken
617
  continue
618
 
619
+ content_d, n_ref = construct_datadict(article_tree)
620
  glossary = np.array([[k,v] for k,v in content_d["glossary"].items()])
621
  data = {
622
  "introduction": content_d["introduction"],
 
640
  "media": content_d["media"],
641
  # "question": content_d["question"],
642
  "unknown_pub": content_d["unknown_pub"],
 
 
643
  "glossary": {"acronym":glossary[:,0], "definition":glossary[:,1]} if len(glossary)>0 else {"acronym":[], "definition":[]},
644
  "n_references": n_ref,
645
  "pmid": data["PMID"],