Miscellaneous changes to grammar utilities [part I].

PiperOrigin-RevId: 641900294
google-research · Jun 10, 2024 · 9300462 · 9300462
1 parent 033780f
commit 9300462
Show file tree

Hide file tree

Showing 8 changed files with 25 additions and 26 deletions.
diff --git a/nisaba/interim/grm2/paths/BUILD.bazel b/nisaba/interim/grm2/paths/BUILD.bazel
@@ -21,5 +21,6 @@ cc_library(
     hdrs = ["paths.h"],
     deps = [
         "@org_openfst//:fst",
+        "@org_openfst//:util",
     ],
 )
diff --git a/nisaba/interim/grm2/paths/paths.h b/nisaba/interim/grm2/paths/paths.h
@@ -30,8 +30,11 @@
 #include <string>
 #include <vector>
 
+#include "fst/arc.h"
 #include "fst/fst.h"
+#include "fst/properties.h"
 #include "fst/string.h"
+#include "fst/util.h"
 
 namespace fst {
 

diff --git a/nisaba/interim/grm2/rewrite/BUILD.bazel b/nisaba/interim/grm2/rewrite/BUILD.bazel
@@ -21,8 +21,8 @@ cc_library(
     hdrs = ["base_rule_cascade.h"],
     deps = [
         ":rewrite",
-        "@com_google_absl//absl/log",
         "@com_google_absl//absl/strings",
+        "@com_google_absl//absl/types:span",
         "@com_google_protobuf//:protobuf_lite",
         "@org_openfst//:fst",
         "@org_openfst//:lib_lite",
@@ -35,6 +35,7 @@ cc_library(
     hdrs = ["parentheses.h"],
     deps = [
         "@com_google_absl//absl/log",
+        "@com_google_absl//absl/types:span",
         "@org_openfst//:fst",
     ],
 )
@@ -46,6 +47,7 @@ cc_library(
         "//nisaba/interim/grm2/paths",
         "//nisaba/interim/grm2/string:stringprint",
         "@com_google_absl//absl/log",
+        "@com_google_protobuf//:protobuf_lite",
         "@org_openfst//:fst",
         "@org_openfst//:mpdt",
         "@org_openfst//:pdt",
@@ -66,8 +68,6 @@ cc_library(
         "@org_openfst//:far",
         "@org_openfst//:fst",
         "@org_openfst//:fst-decl",
-        "@org_openfst//:mpdt",
-        "@org_openfst//:pdt",
     ],
 )
 
@@ -78,6 +78,7 @@ cc_library(
         ":base_rule_cascade",
         ":rewrite_manager",
         "@com_google_absl//absl/log",
-        "@org_openfst//:fst",
+        "@com_google_absl//absl/strings",
+        "@com_google_protobuf//:protobuf_lite",
     ],
 )
diff --git a/nisaba/interim/grm2/rewrite/base_rule_cascade.h b/nisaba/interim/grm2/rewrite/base_rule_cascade.h
@@ -29,17 +29,16 @@
 #include <string>
 #include <vector>
 
-#ifndef NO_GOOGLE
-#include "google/protobuf/repeated_field.h"
-#endif  // NO_GOOGLE
 #include "fst/arcsort.h"
 #include "fst/compose.h"
 #include "fst/fst.h"
 #include "fst/string.h"
 #include "fst/symbol-table.h"
-#include "absl/log/log.h"
+#include "fst/vector-fst.h"
 #include "absl/strings/string_view.h"
+#include "absl/types/span.h"
 #include "nisaba/interim/grm2/rewrite/rewrite.h"
+#include "google/protobuf/repeated_ptr_field.h"
 
 namespace rewrite {
 
@@ -188,8 +187,8 @@ class BaseRuleCascade {
   bool Matches(absl::string_view input, absl::string_view output) const;
 
  private:
-  bool LabelsToDebugString(const std::string &output,
-                           const std::vector<Label> &labels,
+  bool LabelsToDebugString(absl::string_view output,
+                           absl::Span<const Label> labels,
                            std::string *debug) const;
 
   void PrintDebugSymbol(Label label, std::ostream &ostrm) const;
@@ -500,8 +499,8 @@ bool BaseRuleCascade<Arc>::Matches(absl::string_view input,
 // Private methods.
 
 template <class Arc>
-bool BaseRuleCascade<Arc>::LabelsToDebugString(const std::string &output,
-                                               const std::vector<Label> &labels,
+bool BaseRuleCascade<Arc>::LabelsToDebugString(absl::string_view output,
+                                               absl::Span<const Label> labels,
                                                std::string *debug) const {
   debug->clear();
   // This really only makes sense in BYTE mode.

diff --git a/nisaba/interim/grm2/rewrite/parentheses.h b/nisaba/interim/grm2/rewrite/parentheses.h
@@ -22,6 +22,7 @@
 
 #include "fst/fst.h"
 #include "absl/log/log.h"
+#include "absl/types/span.h"
 
 // This is used by (M)PdtCompose and and associated grammar managers when they
 // express parentheses and/or assignments as transducers.
@@ -76,8 +77,8 @@ void MakeParenthesesVector(
 template <class Arc>
 void MakeAssignmentsVector(
     const ::fst::Fst<Arc> &assignments_transducer,
-    const std::vector<std::pair<typename Arc::Label, typename Arc::Label>>
-        &parens,
+    absl::Span<const std::pair<typename Arc::Label, typename Arc::Label>>
+        parens,
     std::vector<typename Arc::Label> *assignments) {
   using Label = typename Arc::Label;
   std::map<Label, Label> assignment_map;

diff --git a/nisaba/interim/grm2/rewrite/rewrite.h b/nisaba/interim/grm2/rewrite/rewrite.h
@@ -25,8 +25,6 @@
 #include <utility>
 #include <vector>
 
-#ifndef NO_GOOGLE
-#endif  // NO_GOOGLE
 #include "fst/extensions/mpdt/compose.h"
 #include "fst/extensions/pdt/compose.h"
 #include "fst/arcsort.h"
@@ -42,6 +40,7 @@
 #include "absl/log/log.h"
 #include "nisaba/interim/grm2/paths/paths.h"
 #include "nisaba/interim/grm2/string/stringprint.h"
+#include "google/protobuf/repeated_ptr_field.h"
 
 // Generic rewrite utilities for string inputs.
 

diff --git a/nisaba/interim/grm2/rewrite/rewrite_manager.h b/nisaba/interim/grm2/rewrite/rewrite_manager.h
@@ -24,15 +24,7 @@
 #include <utility>
 #include <vector>
 
-#ifndef NO_GOOGLE
-#include "google/protobuf/repeated_field.h"
-#endif  // NO_GOOGLE
 #include "fst/extensions/far/far.h"
-#include "fst/extensions/mpdt/compose.h"
-#include "fst/extensions/mpdt/mpdt.h"
-#include "fst/extensions/pdt/compose.h"
-#include "fst/extensions/pdt/pdt.h"
-#include "fst/extensions/pdt/shortest-path.h"
 #include "fst/arcsort.h"
 #include "fst/fst-decl.h"
 #include "fst/intersect.h"
@@ -43,6 +35,7 @@
 #include "absl/strings/string_view.h"
 #include "nisaba/interim/grm2/rewrite/parentheses.h"
 #include "nisaba/interim/grm2/rewrite/rewrite.h"
+#include "google/protobuf/repeated_ptr_field.h"
 
 namespace rewrite {
 

diff --git a/nisaba/interim/grm2/rewrite/rule_cascade.h b/nisaba/interim/grm2/rewrite/rule_cascade.h
@@ -25,8 +25,10 @@
 #include <vector>
 
 #include "absl/log/log.h"
+#include "absl/strings/string_view.h"
 #include "nisaba/interim/grm2/rewrite/base_rule_cascade.h"
 #include "nisaba/interim/grm2/rewrite/rewrite_manager.h"
+#include "google/protobuf/repeated_ptr_field.h"
 
 namespace rewrite {
 namespace internal {
@@ -35,7 +37,7 @@ namespace internal {
 // optional MPDT assignment rule name).
 class RuleTriple {
  public:
-  explicit RuleTriple(const std::string &rule_def)
+  explicit RuleTriple(absl::string_view rule_def)
       : RuleTriple(RuleTriple::ParseRuleDef(rule_def)) {}
 
   const std::string &Rule() const { return rules_[0]; }
@@ -53,7 +55,7 @@ class RuleTriple {
       : rules_(std::move(rules)) {}
 
   // Actually parses the rules into the pieces we need.
-  static std::array<std::string, 3> ParseRuleDef(const std::string &rule_def) {
+  static std::array<std::string, 3> ParseRuleDef(absl::string_view rule_def) {
     std::array<std::string, 3> result;
     auto main_pos = rule_def.find('$');
     if (main_pos == std::string::npos) main_pos = rule_def.find(':');