You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

279 lines
11 KiB

/*
* Copyright (C) 2018 The Android Open Source Project
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "utils/grammar/parsing/parser.h"
#include <unordered_map>
#include "utils/grammar/parsing/parse-tree.h"
#include "utils/grammar/rules-utils.h"
#include "utils/grammar/types.h"
#include "utils/zlib/zlib.h"
#include "utils/zlib/zlib_regex.h"
namespace libtextclassifier3::grammar {
namespace {
inline bool CheckMemoryUsage(const UnsafeArena* arena) {
// The maximum memory usage for matching.
constexpr int kMaxMemoryUsage = 1 << 20;
return arena->status().bytes_allocated() <= kMaxMemoryUsage;
}
// Maps a codepoint to include the token padding if it aligns with a token
// start. Whitespace is ignored when symbols are fed to the matcher. Preceding
// whitespace is merged to the match start so that tokens and non-terminals
// appear next to each other without whitespace. For text or regex annotations,
// we therefore merge the whitespace padding to the start if the annotation
// starts at a token.
int MapCodepointToTokenPaddingIfPresent(
const std::unordered_map<CodepointIndex, CodepointIndex>& token_alignment,
const int start) {
const auto it = token_alignment.find(start);
if (it != token_alignment.end()) {
return it->second;
}
return start;
}
} // namespace
Parser::Parser(const UniLib* unilib, const RulesSet* rules)
: unilib_(*unilib),
rules_(rules),
lexer_(unilib),
nonterminals_(rules_->nonterminals()),
rules_locales_(ParseRulesLocales(rules_)),
regex_annotators_(BuildRegexAnnotators()) {}
// Uncompresses and build the defined regex annotators.
std::vector<Parser::RegexAnnotator> Parser::BuildRegexAnnotators() const {
std::vector<RegexAnnotator> result;
if (rules_->regex_annotator() != nullptr) {
std::unique_ptr<ZlibDecompressor> decompressor =
ZlibDecompressor::Instance();
result.reserve(rules_->regex_annotator()->size());
for (const RulesSet_::RegexAnnotator* regex_annotator :
*rules_->regex_annotator()) {
result.push_back(
{UncompressMakeRegexPattern(unilib_, regex_annotator->pattern(),
regex_annotator->compressed_pattern(),
rules_->lazy_regex_compilation(),
decompressor.get()),
regex_annotator->nonterminal()});
}
}
return result;
}
std::vector<Symbol> Parser::SortedSymbolsForInput(const TextContext& input,
UnsafeArena* arena) const {
// Whitespace is ignored when symbols are fed to the matcher.
// For regex matches and existing text annotations we therefore have to merge
// preceding whitespace to the match start so that tokens and non-terminals
// appear as next to each other without whitespace. We keep track of real
// token starts and precending whitespace in `token_match_start`, so that we
// can extend a match's start to include the preceding whitespace.
std::unordered_map<CodepointIndex, CodepointIndex> token_match_start;
for (int i = input.context_span.first + 1; i < input.context_span.second;
i++) {
const CodepointIndex token_start = input.tokens[i].start;
const CodepointIndex prev_token_end = input.tokens[i - 1].end;
if (token_start != prev_token_end) {
token_match_start[token_start] = prev_token_end;
}
}
std::vector<Symbol> symbols;
CodepointIndex match_offset = input.tokens[input.context_span.first].start;
// Add start symbol.
if (input.context_span.first == 0 &&
nonterminals_->start_nt() != kUnassignedNonterm) {
match_offset = 0;
symbols.emplace_back(arena->AllocAndInit<ParseTree>(
nonterminals_->start_nt(), CodepointSpan{0, 0},
/*match_offset=*/0, ParseTree::Type::kDefault));
}
if (nonterminals_->wordbreak_nt() != kUnassignedNonterm) {
symbols.emplace_back(arena->AllocAndInit<ParseTree>(
nonterminals_->wordbreak_nt(),
CodepointSpan{match_offset, match_offset},
/*match_offset=*/match_offset, ParseTree::Type::kDefault));
}
// Add symbols from tokens.
for (int i = input.context_span.first; i < input.context_span.second; i++) {
const Token& token = input.tokens[i];
lexer_.AppendTokenSymbols(token.value, /*match_offset=*/match_offset,
CodepointSpan{token.start, token.end}, &symbols);
match_offset = token.end;
// Add word break symbol.
if (nonterminals_->wordbreak_nt() != kUnassignedNonterm) {
symbols.emplace_back(arena->AllocAndInit<ParseTree>(
nonterminals_->wordbreak_nt(),
CodepointSpan{match_offset, match_offset},
/*match_offset=*/match_offset, ParseTree::Type::kDefault));
}
}
// Add end symbol if used by the grammar.
if (input.context_span.second == input.tokens.size() &&
nonterminals_->end_nt() != kUnassignedNonterm) {
symbols.emplace_back(arena->AllocAndInit<ParseTree>(
nonterminals_->end_nt(), CodepointSpan{match_offset, match_offset},
/*match_offset=*/match_offset, ParseTree::Type::kDefault));
}
// Add symbols from the regex annotators.
const CodepointIndex context_start =
input.tokens[input.context_span.first].start;
const CodepointIndex context_end =
input.tokens[input.context_span.second - 1].end;
for (const RegexAnnotator& regex_annotator : regex_annotators_) {
std::unique_ptr<UniLib::RegexMatcher> regex_matcher =
regex_annotator.pattern->Matcher(UnicodeText::Substring(
input.text, context_start, context_end, /*do_copy=*/false));
int status = UniLib::RegexMatcher::kNoError;
while (regex_matcher->Find(&status) &&
status == UniLib::RegexMatcher::kNoError) {
const CodepointSpan span{regex_matcher->Start(0, &status) + context_start,
regex_matcher->End(0, &status) + context_start};
symbols.emplace_back(arena->AllocAndInit<ParseTree>(
regex_annotator.nonterm, span, /*match_offset=*/
MapCodepointToTokenPaddingIfPresent(token_match_start, span.first),
ParseTree::Type::kDefault));
}
}
// Add symbols based on annotations.
if (auto annotation_nonterminals = nonterminals_->annotation_nt()) {
for (const AnnotatedSpan& annotated_span : input.annotations) {
const ClassificationResult& classification =
annotated_span.classification.front();
if (auto entry = annotation_nonterminals->LookupByKey(
classification.collection.c_str())) {
symbols.emplace_back(arena->AllocAndInit<AnnotationNode>(
entry->value(), annotated_span.span, /*match_offset=*/
MapCodepointToTokenPaddingIfPresent(token_match_start,
annotated_span.span.first),
&classification));
}
}
}
std::sort(symbols.begin(), symbols.end(),
[](const Symbol& a, const Symbol& b) {
// Sort by increasing (end, start) position to guarantee the
// matcher requirement that the tokens are fed in non-decreasing
// end position order.
return std::tie(a.codepoint_span.second, a.codepoint_span.first) <
std::tie(b.codepoint_span.second, b.codepoint_span.first);
});
return symbols;
}
void Parser::EmitSymbol(const Symbol& symbol, UnsafeArena* arena,
Matcher* matcher) const {
if (!CheckMemoryUsage(arena)) {
return;
}
switch (symbol.type) {
case Symbol::Type::TYPE_PARSE_TREE: {
// Just emit the parse tree.
matcher->AddParseTree(symbol.parse_tree);
return;
}
case Symbol::Type::TYPE_DIGITS: {
// Emit <digits> if used by the rules.
if (nonterminals_->digits_nt() != kUnassignedNonterm) {
matcher->AddParseTree(arena->AllocAndInit<ParseTree>(
nonterminals_->digits_nt(), symbol.codepoint_span,
symbol.match_offset, ParseTree::Type::kDefault));
}
// Emit <n_digits> if used by the rules.
if (nonterminals_->n_digits_nt() != nullptr) {
const int num_digits =
symbol.codepoint_span.second - symbol.codepoint_span.first;
if (num_digits <= nonterminals_->n_digits_nt()->size()) {
const Nonterm n_digits_nt =
nonterminals_->n_digits_nt()->Get(num_digits - 1);
if (n_digits_nt != kUnassignedNonterm) {
matcher->AddParseTree(arena->AllocAndInit<ParseTree>(
nonterminals_->n_digits_nt()->Get(num_digits - 1),
symbol.codepoint_span, symbol.match_offset,
ParseTree::Type::kDefault));
}
}
}
break;
}
case Symbol::Type::TYPE_TERM: {
// Emit <uppercase_token> if used by the rules.
if (nonterminals_->uppercase_token_nt() != 0 &&
unilib_.IsUpperText(
UTF8ToUnicodeText(symbol.lexeme, /*do_copy=*/false))) {
matcher->AddParseTree(arena->AllocAndInit<ParseTree>(
nonterminals_->uppercase_token_nt(), symbol.codepoint_span,
symbol.match_offset, ParseTree::Type::kDefault));
}
break;
}
default:
break;
}
// Emit the token as terminal.
matcher->AddTerminal(symbol.codepoint_span, symbol.match_offset,
symbol.lexeme);
// Emit <token> if used by rules.
matcher->AddParseTree(arena->AllocAndInit<ParseTree>(
nonterminals_->token_nt(), symbol.codepoint_span, symbol.match_offset,
ParseTree::Type::kDefault));
}
// Parses an input text and returns the root rule derivations.
std::vector<Derivation> Parser::Parse(const TextContext& input,
UnsafeArena* arena) const {
// Check the tokens, input can be non-empty (whitespace) but have no tokens.
if (input.tokens.empty()) {
return {};
}
// Select locale matching rules.
std::vector<const RulesSet_::Rules*> locale_rules =
SelectLocaleMatchingShards(rules_, rules_locales_, input.locales);
if (locale_rules.empty()) {
// Nothing to do.
return {};
}
Matcher matcher(&unilib_, rules_, locale_rules, arena);
for (const Symbol& symbol : SortedSymbolsForInput(input, arena)) {
EmitSymbol(symbol, arena, &matcher);
}
matcher.Finish();
return matcher.chart().derivations();
}
} // namespace libtextclassifier3::grammar