def save(self, item: CitationData, _: None) -> None:
citation_locations = item.citation_locations
key_s2_ids = item.key_s2_ids
entity_infos = []
citation_index = 0
for citation_key, locations in citation_locations.items():
if citation_key not in key_s2_ids:
logging.warning( # pylint: disable=logging-not-lazy
"Not uploading bounding box information for citation with key "
+ "%s because it was not resolved to a paper S2 ID.",
citation_key,
)
continue
for cluster_index, location_set in locations.items():
boxes = cast(List[BoundingBox], list(location_set))
entity_info = EntityInformation(
id_=f"{citation_key}-{cluster_index}",
type_="citation",
bounding_boxes=boxes,
data={
"key": citation_key,
"paper_id": key_s2_ids[citation_key]
},
)
entity_infos.append(entity_info)
citation_index += 1
upload_entities(item.s2_id, item.arxiv_id, entity_infos,
self.args.data_version)
def upload_sentences(processing_summary: PaperProcessingResult,
data_version: Optional[int]) -> None:
entity_infos = []
for entity_and_location in processing_summary.localized_entities:
sentence = cast(SentenceEntity, entity_and_location.entity)
boxes = [cast(BoundingBox, l) for l in entity_and_location.locations]
entity_info = EntityInformation(
id_=f"{sentence.tex_path}-{sentence.id_}",
type_="sentence",
bounding_boxes=boxes,
data={
"text": sentence.text,
"tex": sentence.tex,
"tex_start": sentence.start,
"tex_end": sentence.end,
},
)
entity_infos.append(entity_info)
upload_entities(
processing_summary.s2_id,
processing_summary.arxiv_id,
entity_infos,
data_version,
)
def upload_terms(processing_summary: PaperProcessingResult,
data_version: Optional[int]) -> None:
entity_infos = []
for entity_and_location in processing_summary.localized_entities:
term = cast(TermEntity, entity_and_location.entity)
boxes = [cast(BoundingBox, l) for l in entity_and_location.locations]
entity_info = EntityInformation(
id_=f"{term.tex_path}-{term.id_}",
type_="term",
bounding_boxes=boxes,
data={
"name": term.name,
"definitions": term.definitions,
"sources": term.sources,
"val": term.val,
},
)
entity_infos.append(entity_info)
upload_entities(
processing_summary.s2_id,
processing_summary.arxiv_id,
entity_infos,
data_version,
)
def upload_equations(processing_summary: PaperProcessingResult,
data_version: Optional[int]) -> None:
entity_infos = []
for entity_and_location in processing_summary.localized_entities:
equation = cast(Equation, entity_and_location.entity)
boxes = [cast(BoundingBox, l) for l in entity_and_location.locations]
entity_info = EntityInformation(
id_=f"{equation.tex_path}-{equation.id_}",
type_="equation",
bounding_boxes=boxes,
data={"tex": equation.tex},
)
entity_infos.append(entity_info)
upload_entities(
processing_summary.s2_id,
processing_summary.arxiv_id,
entity_infos,
data_version,
)
def upload_symbols(
processing_summary: PaperProcessingResult, data_version: Optional[int]
) -> None:
arxiv_id = processing_summary.arxiv_id
entities = [es.entity for es in processing_summary.entities]
symbols = cast(List[SerializableSymbol], entities)
symbols_by_id = {sid(s): s for s in symbols}
entity_infos: List[EntityUploadInfo] = []
# Load MathML matches for partially matching of symbols.
matches: Matches = {}
matches_path = os.path.join(
directories.arxiv_subdir("symbol-matches", processing_summary.arxiv_id),
"matches.csv",
)
if os.path.exists(matches_path):
for match in file_utils.load_from_csv(matches_path, Match):
if match.queried_mathml not in matches:
matches[match.queried_mathml] = []
matches[match.queried_mathml].append(match)
else:
logging.warning(
"Could not find symbol matches information for paper %s.", arxiv_id,
)
# Load parent-child relationships for symbols.
children: Dict[SymbolId, List[SymbolId]] = defaultdict(list)
parents: Dict[SymbolId, SymbolId] = {}
children_path = os.path.join(
directories.arxiv_subdir("detected-symbols", arxiv_id), "symbol_children.csv"
)
if os.path.exists(children_path):
for parent in file_utils.load_from_csv(children_path, SerializableChild):
pid = f"{parent.tex_path}-{parent.equation_index}-{parent.symbol_index}"
cid = f"{parent.tex_path}-{parent.equation_index}-{parent.child_index}"
parents[cid] = pid
children[pid].append(cid)
else:
logging.warning(
"Could not find file mapping from symbol to their children for paper %s.",
arxiv_id,
)
# Load contexts that the symbols appear in. Sort them by the symbol MathML.
context_data_missing = False
contexts_path = os.path.join(
directories.arxiv_subdir("contexts-for-symbols", arxiv_id), "contexts.csv",
)
if not os.path.exists(contexts_path):
logging.warning( # pylint: disable=logging-not-lazy
"Contexts have not been found for symbols for arXiv paper %s. "
+ "Symbol data will be uploaded without contexts.",
arxiv_id,
)
context_data_missing = True
symbol_contexts = {}
mathml_contexts = defaultdict(list)
if not context_data_missing:
for context in file_utils.load_from_csv(contexts_path, Context):
tex_path = context.tex_path
symbol_id = f"{tex_path}-{context.entity_id}"
symbol_contexts[symbol_id] = context
symbol = symbols_by_id[symbol_id]
mathml_contexts[symbol.mathml].append(context)
# Prepare collections of formulae that each symbol was found in.
symbol_formulas = {}
mathml_formulas: Dict[str, Set[DefiningFormula]] = defaultdict(set)
for symbol in symbols:
if (
symbol.is_definition
and symbol.equation is not None
and symbol.relative_start is not None
and symbol.relative_end is not None
):
highlighted = wrap_span(
symbol.equation,
symbol.relative_start,
symbol.relative_end,
before=r"\htmlClass{match-highlight}{",
after="}",
braces=True,
)
formula = DefiningFormula(
tex=highlighted,
tex_path=symbol.tex_path,
equation_id=str(symbol.equation_index),
)
symbol_formulas[sid(symbol)] = formula
mathml_formulas[symbol.mathml].add(formula)
entity_infos = []
for localized_entity in processing_summary.entities:
symbol = cast(SerializableSymbol, localized_entity.entity)
boxes = [
BoundingBox(l.left, l.top, l.width, l.height, l.page)
for l in localized_entity.locations
]
# Get context and formula of the symbol, and other matching ones.
symbol_context = symbol_contexts.get(sid(symbol))
matching_contexts = mathml_contexts.get(symbol.mathml, [])
other_context_texs = []
other_context_sentence_ids = []
for c in matching_contexts:
matching_sentence_id = f"{c.tex_path}-{c.sentence_id}"
if matching_sentence_id not in other_context_sentence_ids:
other_context_texs.append(c.snippet)
other_context_sentence_ids.append(matching_sentence_id)
matching_formulas = mathml_formulas.get(symbol.mathml, set())
other_formula_texs = []
other_formula_ids = []
for f in matching_formulas:
equation_id = f"{f.tex_path}-{f.equation_id}"
if equation_id not in other_formula_ids:
other_formula_texs.append(f.tex)
other_formula_ids.append(equation_id)
# Package up data for the symbol.
tags: List[str] = []
MAX_BOX_HEIGHT = 0.1
for b in boxes:
if b.height > MAX_BOX_HEIGHT:
logging.debug( # pylint: disable=logging-not-lazy
"Detected large bounding box for symbol with height %f for entity %s of paper "
+ "%s. Entity will be given a tag indicating it is unexpectedly large.",
b.height,
f"{localized_entity.entity.tex_path}-{localized_entity.entity.id_}",
arxiv_id,
)
tags.append("large")
break
data: EntityData = {
"tex": f"${symbol.tex}$",
"tex_start": symbol.start,
"tex_end": symbol.end,
"type": symbol.type_,
"mathml": symbol.mathml,
"mathml_near_matches": [m.matching_mathml for m in matches[symbol.mathml]],
"snippets": other_context_texs,
"defining_formulas": other_formula_texs,
"is_definition": symbol.is_definition or False,
"tags": tags,
}
# Create links between this symbol, its sentence, and related symbols.
sentence_id = (
f"{symbol_context.tex_path}-{symbol_context.sentence_id}"
if symbol_context is not None
else None
)
parent_id = parents.get(sid(symbol))
child_ids = children.get(sid(symbol), [])
relationships: EntityRelationships = {
"equation": EntityReference(
type_="equation", id_=f"{symbol.tex_path}-{symbol.equation_index}",
),
"parent": EntityReference(type_="symbol", id_=parent_id),
"children": [EntityReference(type_="symbol", id_=id_) for id_ in child_ids],
"sentence": EntityReference(type_="sentence", id_=sentence_id)
if sentence_id is not None
else EntityReference(type_="sentence", id_=None),
"defining_formula_equations": [
EntityReference(type_="equation", id_=id_) for id_ in other_formula_ids
],
"snippet_sentences": [
EntityReference(type_="sentence", id_=id_)
for id_ in other_context_sentence_ids
],
}
# Save all data for this symbol
entity_information = EntityUploadInfo(
id_=sid(symbol),
type_="symbol",
bounding_boxes=boxes,
data=data,
relationships=relationships,
)
entity_infos.append(entity_information)
upload_entities(
processing_summary.s2_id, arxiv_id, entity_infos, data_version,
)
def save(self, item: SymbolData, _: None) -> None:
symbols_with_ids = item.symbols_with_ids
boxes = item.boxes
matches = item.matches
symbol_contexts = item.symbol_contexts
mathml_contexts = item.mathml_contexts
symbol_formulas = item.symbol_formulas
mathml_formulas = item.mathml_formulas
symbol_ids_by_symbol_object_ids = {}
for symbol_with_id in symbols_with_ids:
symbol_ids_by_symbol_object_ids[id(
symbol_with_id.symbol)] = symbol_with_id.symbol_id
entity_infos = []
for symbol_with_id in symbols_with_ids:
symbol = symbol_with_id.symbol
# TODO(andrewhead): move this filtering condition into 'parse_equation'
if symbol.tex in ["$|$", "|"]:
continue
symbol_id = symbol_with_id.symbol_id
# Get context and formula of the symbol, and other matching ones.
context = symbol_contexts.get(symbol_id)
matching_contexts = mathml_contexts.get(symbol.mathml, [])
other_context_texs = []
other_context_sentence_ids = []
for c in matching_contexts:
matching_sentence_id = f"{c.tex_path}-{c.sentence_id}"
if (matching_sentence_id not in other_context_sentence_ids
# and c.sentence_id != context.sentence_id
):
other_context_texs.append(c.snippet)
other_context_sentence_ids.append(matching_sentence_id)
formula = symbol_formulas.get(symbol_id)
matching_formulas = mathml_formulas.get(symbol.mathml, [])
other_formula_texs = []
other_formula_ids = []
for f in matching_formulas:
equation_id = f"{f.tex_path}-{f.equation_id}"
if equation_id not in other_formula_ids:
# and (
# : formula is None or equation_id != formula.equation_id
# )
other_formula_texs.append(f.tex)
other_formula_ids.append(equation_id)
box = boxes.get(symbol_id)
if box is None:
continue
data: EntityData = {
"tex":
f"${symbol.tex}$",
"tex_start":
symbol.start,
"tex_end":
symbol.end,
"mathml":
symbol.mathml,
"mathml_near_matches":
[m.matching_mathml for m in matches[symbol.mathml]],
# "snippet": context.snippet,
"snippets":
other_context_texs,
"defining_formulas":
other_formula_texs,
"is_definition":
symbol.is_definition or False,
}
# if formula is not None:
# data['formula'] = formula.tex
create_symbol_id_string: Callable[[SymbolId], str] = (
lambda sid:
f"{sid.tex_path}-{sid.equation_index}-{sid.symbol_index}")
sentence_id = (f"{context.tex_path}-{context.sentence_id}"
if context is not None else None)
parent_id: Optional[str] = None
for other_symbol_with_id in symbols_with_ids:
other_symbol_id = other_symbol_with_id.symbol_id
other_symbol = other_symbol_with_id.symbol
try:
other_symbol.children.index(symbol)
parent_id = create_symbol_id_string(other_symbol_id)
except ValueError:
continue
child_ids = []
for child_symbol in symbol.children:
child_symbol_id = symbol_ids_by_symbol_object_ids[id(
child_symbol)]
string_id = create_symbol_id_string(child_symbol_id)
child_ids.append(string_id)
relationships: EntityRelationships = {
"equation":
EntityReference(
type_="equation",
id_=f"{symbol_id.tex_path}-{symbol_id.equation_index}",
),
"parent":
EntityReference(type_="symbol", id_=parent_id),
"children": [
EntityReference(type_="symbol", id_=id_)
for id_ in child_ids
],
"sentence":
EntityReference(type_="sentence", id_=sentence_id) if
sentence_id is not None else EntityReference(type_="sentence",
id_=None),
"defining_formula_equations": [
EntityReference(type_="equation", id_=id_)
for id_ in other_formula_ids
],
"snippet_sentences": [
EntityReference(type_="sentence", id_=id_)
for id_ in other_context_sentence_ids
],
# "snippet_sentence": EntityReference(
# type_="sentence", id_=f"{symbol_id.tex_path}-f{context.sentence_id}"
# )
# if context is not None
# else None,
# "formula_equation": EntityReference(
# type_="equation",
# id_=f"{symbol_id.tex_path}-f{formula.equation_id}"
# if formula is not None
# else None,
# ),
}
entity_information = EntityInformation(
id_=
f"{symbol_id.tex_path}-{symbol_id.equation_index}-{symbol_id.symbol_index}",
type_="symbol",
bounding_boxes=[box],
data=data,
relationships=relationships,
)
entity_infos.append(entity_information)
upload_entities(item.s2_id, item.arxiv_id, entity_infos,
self.args.data_version)
def upload_terms(processing_summary: PaperProcessingResult,
data_version: Optional[int]) -> None:
arxiv_id = processing_summary.arxiv_id
contexts = file_utils.load_from_csv(
os.path.join(
directories.arxiv_subdir("contexts-for-glossary-terms", arxiv_id),
"contexts.csv",
),
Context,
)
contexts_by_entity = {(c.tex_path, c.entity_id): c for c in contexts}
# Assemble contexts that should be shown for each term.
contexts_by_term: Dict[str, List[Context]] = defaultdict(list)
for entity_and_location in processing_summary.localized_entities:
term = cast(Term, entity_and_location.entity)
if (term.tex_path, term.id_) in contexts_by_entity:
contexts_by_term[term.text].append(
contexts_by_entity[(term.tex_path, term.id_)])
entity_infos = []
for entity_and_location in processing_summary.localized_entities:
term = cast(Term, entity_and_location.entity)
context = contexts_by_entity.get((term.tex_path, term.id_))
boxes = [cast(BoundingBox, l) for l in entity_and_location.locations]
# Cluster bounding boxes, in case any of these terms are defined as a macro (in which)
# case all appearances of that term on the same page will have been lumped together.
clusters = cluster_boxes(boxes, vertical_split=0.005)
for i, cluster in enumerate(clusters):
entity_info = EntityInformation(
id_=f"{term.tex_path}-{term.id_}-{i}",
type_="term",
bounding_boxes=list(cluster),
data={
"name":
term.text,
"definitions":
term.definitions,
"definition_texs":
term.definitions,
"sources":
term.sources,
"snippets":
[c.snippet for c in contexts_by_term.get(term.text, [])],
},
relationships={
"sentence":
EntityReference(
type_="sentence",
id_=f"{context.tex_path}-{context.sentence_id}"
if context is not None else None,
),
"snippet_sentences": [
EntityReference(type_="sentence",
id_=f"{c.tex_path}-{c.sentence_id}")
for c in contexts_by_term.get(term.text, [])
],
},
)
entity_infos.append(entity_info)
upload_entities(
processing_summary.s2_id,
processing_summary.arxiv_id,
entity_infos,
data_version,
)
def upload_definitions(processing_summary: PaperProcessingResult,
data_version: Optional[int]) -> None:
term_infos = []
definition_infos = []
for entity_and_location in processing_summary.localized_entities:
boxes = [cast(BoundingBox, l) for l in entity_and_location.locations]
entity = entity_and_location.entity
if entity.id_.startswith("definition"):
definition = cast(Definition, entity)
definition_info = EntityInformation(
id_=definition.id_,
type_="definition",
bounding_boxes=boxes,
data={
"definiendum": definition.definiendum,
"definition": definition.text,
"tex": definition.tex,
},
relationships={
"sentence":
EntityReference(
type_="sentence",
id_=f"{definition.tex_path}-{definition.sentence_id}"
if definition.sentence_id is not None else None,
),
},
)
definition_infos.append(definition_info)
if entity.id_.startswith("definiendum") or entity.id_.startswith(
"term-reference"):
term = cast(TermReference, entity)
term_info = EntityInformation(
id_=term.id_,
type_="term",
bounding_boxes=boxes,
data={
"name": term.text,
"definitions": term.definitions,
"definition_texs": term.definition_texs,
"sources": term.sources,
"term_type": term.type_ or "unknown"
},
relationships={
"sentence":
EntityReference(
type_="sentence",
id_=f"{term.tex_path}-{term.sentence_id}"
if term.sentence_id is not None else None,
),
"definitions": [
EntityReference(type_="definition", id_=d)
for d in term.definition_ids
],
},
)
term_infos.append(term_info)
# Upload definitions before terms, because terms hold references to definitions that can
# only be resolved once the definitions have been uploaded.
upload_entities(
processing_summary.s2_id,
processing_summary.arxiv_id,
definition_infos,
data_version,
)
upload_entities(
processing_summary.s2_id,
processing_summary.arxiv_id,
term_infos,
data_version,
)
def save(self, item: SymbolData, _: None) -> None:
symbols_with_ids = item.symbols_with_ids
boxes = item.boxes
matches = item.matches
symbol_sentences = item.symbol_sentences
symbol_ids_by_symbol_object_ids = {}
for symbol_with_id in symbols_with_ids:
symbol_ids_by_symbol_object_ids[id(
symbol_with_id.symbol)] = symbol_with_id.symbol_id
entity_infos = []
for symbol_with_id in symbols_with_ids:
symbol = symbol_with_id.symbol
symbol_id = symbol_with_id.symbol_id
box = boxes.get(symbol_id)
if box is None:
continue
data: EntityData = {
"tex":
f"${symbol.tex}$",
"tex_start":
symbol.start,
"tex_end":
symbol.end,
"mathml":
symbol.mathml,
"mathml_near_matches":
[m.matching_mathml for m in matches[symbol.mathml]],
}
sentence_key = symbol_sentences.get(symbol_id)
sentence_id = (
f"{sentence_key.tex_path}-{sentence_key.sentence_id}"
if sentence_key is not None else None)
child_ids = []
for child_symbol in symbol.children:
child_symbol_id = symbol_ids_by_symbol_object_ids[id(
child_symbol)]
string_id = f"{child_symbol_id.tex_path}-{child_symbol_id.equation_index}-{child_symbol_id.symbol_index}"
child_ids.append(string_id)
relationships: EntityRelationships = {
"children": [
EntityReference(type_="symbol", id_=id_)
for id_ in child_ids
],
"sentence":
EntityReference(type_="sentence", id_=None)
if sentence_id is None else EntityReference(type_="sentence",
id_=sentence_id),
}
entity_information = EntityInformation(
id_=
f"{symbol_id.tex_path}-{symbol_id.equation_index}-{symbol_id.symbol_index}",
type_="symbol",
bounding_boxes=[box],
data=data,
relationships=relationships,
)
entity_infos.append(entity_information)
upload_entities(item.s2_id, item.arxiv_id, entity_infos,
self.args.data_version)
def upload_term_definitions(
processing_summary: PaperProcessingResult, data_version: Optional[int]
) -> None:
" Upload textual terms and their definitions. "
# Group contextual snippets for each term.
term_infos = []
contexts_by_term_name: Dict[TermName, List[Context]] = defaultdict(list)
for entity_summary in processing_summary.entities:
entity = entity_summary.entity
context = entity_summary.context
if is_textual_term(entity) and context is not None:
contexts_by_term_name[entity.text].append(context) # type: ignore
# Construct mapping from definitions to the sentences that contain them.
contexts_by_definition: Dict[EntityId, Context] = {}
for entity_summary in processing_summary.entities:
entity_id = entity_summary.entity.id_
context = entity_summary.context
if (entity_id.startswith("definition")) and context is not None:
contexts_by_definition[entity_id] = context
# Upload information for each term.
for entity_summary in processing_summary.entities:
boxes = [cast(BoundingBox, l) for l in entity_summary.locations]
entity = entity_summary.entity
context = entity_summary.context
if not is_textual_term(entity):
continue
term = cast(TermReference, entity)
# Assemble list of snippets that include this term.
contexts_matching_term = contexts_by_term_name.get(term.text, [])
snippets = [c.snippet for c in contexts_matching_term]
snippet_sentences = [
f"{c.tex_path}-{c.sentence_id}" for c in contexts_matching_term
]
# Create links to the sentences containing definitions for this term.
definition_sentences: List[Optional[str]] = []
for definition_id in term.definition_ids:
if definition_id not in contexts_by_definition:
definition_sentences.append(None)
definition_context = contexts_by_definition[definition_id]
definition_sentences.append(
f"{definition_context.tex_path}-{definition_context.sentence_id}"
)
term_info = EntityUploadInfo(
id_=term.id_,
type_="term",
bounding_boxes=boxes,
data={
"name": term.text,
"term_type": term.type_ or "unknown",
"definitions": term.definitions,
"definition_texs": term.definition_texs,
"sources": term.sources,
# A list of all other sentences the term appearse elsewhere in the paper.
"snippets": snippets,
},
relationships={
# Link the term to the sentence it belongs to. This link is necessary to enable
# visual filtering in the UI where, when a term is clicked, the sentence is
# highlighted and all others are lowlighted.
"sentence": EntityReference(
type_="sentence",
id_=f"{context.tex_path}-{context.sentence_id}"
if context is not None
else None,
),
# IDs of the sentences that contain each of the definitions for a term. These IDs
# can be used to establish links that take a user to the site of a definition.
"definition_sentences": [
EntityReference(type_="sentence", id_=id_)
for id_ in definition_sentences
],
# The IDs of each sentence where the term appears elsewhere in the paper (i.e.,
# for each of the 'snippets' in the entity data above. Used to link from a snippet
# that is shown in a list of snippets to where that snippet appears in the paper.
"snippet_sentences": [
EntityReference(type_="sentence", id_=id_)
for id_ in snippet_sentences
],
},
)
term_infos.append(term_info)
upload_entities(
processing_summary.s2_id, processing_summary.arxiv_id, term_infos, data_version,
)
