Split SymbolTable source location by file, line, and column.

PiperOrigin-RevId: 579255556

centipede/BUILD

@@ -109,6 +109,7 @@ cc_library(
         ":util",
         "@com_google_absl//absl/log",
         "@com_google_absl//absl/log:check",
+        "@com_google_absl//absl/status",
         "@com_google_absl//absl/strings",
         "@com_google_absl//absl/types:span",
     ],

centipede/symbol_table.cc

@@ -20,10 +20,14 @@
 #include <ostream>
 #include <string>
 #include <string_view>
+#include <vector>
 
 #include "absl/log/check.h"
 #include "absl/log/log.h"
+#include "absl/strings/match.h"
+#include "absl/strings/numbers.h"
 #include "absl/strings/str_cat.h"
+#include "absl/strings/str_split.h"
 #include "absl/strings/strip.h"
 #include "absl/types/span.h"
 #include "./centipede/command.h"
@@ -34,10 +38,6 @@
 
 namespace centipede {
 
-bool SymbolTable::Entry::operator==(const Entry &other) const {
-  return this->func == other.func && this->file_line_col == other.file_line_col;
-}
-
 bool SymbolTable::operator==(const SymbolTable &other) const {
   return this->entries_ == other.entries_;
 }
@@ -63,8 +63,8 @@ void SymbolTable::ReadFromLLVMSymbolizer(std::istream &in) {
 
 void SymbolTable::WriteToLLVMSymbolizer(std::ostream &out) {
   for (const Entry &entry : entries_) {
-    out << entry.func << std::endl;
-    out << entry.file_line_col << std::endl;
+    out << entry.func << '\n';
+    out << entry.file_line_col() << '\n';
     out << std::endl;
   }
 }
@@ -154,4 +154,33 @@ void SymbolTable::SetAllToUnknown(size_t size) {
   }
 }
 
+void SymbolTable::AddEntry(std::string_view func,
+                           std::string_view file_line_col) {
+  if (absl::StrContains(file_line_col, "?")) {
+    entries_.emplace_back(
+        Entry{std::string(func), std::string(file_line_col), 0, 0});
+    return;
+  }
+  const std::vector<std::string_view> file_line_col_split =
+      absl::StrSplit(file_line_col, ':');
+  CHECK_LE(file_line_col_split.size(), 3);
+  CHECK_GE(file_line_col_split.size(), 1)
+      << "Unexpected symbolizer input format when getting source location: "
+      << file_line_col;
+  int line = -1;
+  int col = -1;
+  if (file_line_col_split.size() >= 2) {
+    CHECK(absl::SimpleAtoi(file_line_col_split[1], &line))
+        << "Unable to convert line number string to an int: "
+        << file_line_col_split[1];
+  }
+  if (file_line_col_split.size() == 3) {
+    CHECK(absl::SimpleAtoi(file_line_col_split[2], &col))
+        << "Unable to convert column number string to an int: "
+        << file_line_col_split[2];
+  }
+  entries_.emplace_back(
+      Entry{std::string(func), std::string(file_line_col_split[0]), line, col});
+}
+
 }  // namespace centipede

centipede/symbol_table.h

@@ -22,6 +22,13 @@
 #include <string_view>
 #include <vector>
 
+#include "absl/log/check.h"
+#include "absl/log/log.h"
+#include "absl/status/status.h"
+#include "absl/strings/match.h"
+#include "absl/strings/numbers.h"
+#include "absl/strings/str_cat.h"
+#include "absl/strings/str_split.h"
 #include "absl/types/span.h"
 #include "./centipede/control_flow.h"
 #include "./centipede/pc_info.h"
@@ -72,28 +79,41 @@ class SymbolTable {
   const std::string &func(size_t idx) const { return entries_[idx].func; }
 
   // Returns source code location for idx-th entry,
-  const std::string &location(size_t idx) const {
-    return entries_[idx].file_line_col;
+  std::string location(size_t idx) const {
+    return entries_[idx].file_line_col();
   }
 
   // Returns a full human-readable description for idx-th entry.
   std::string full_description(size_t idx) const {
     return func(idx) + " " + location(idx);
   }
 
+  // Add function name and file location to symbol table.
+  void AddEntry(std::string_view func, std::string_view file_line_col);
+
+ private:
   // Defines a symbol table entry.
   struct Entry {
     std::string func;
-    std::string file_line_col;
-    bool operator==(const Entry &other) const;
+    std::string file;
+    int line = -1;
+    int col = -1;
+    bool operator==(const Entry &other) const = default;
+    std::string file_line_col() const {
+      if (absl::StrContains(file, "?")) {
+        return file;
+      }
+      std::string ret = file;
+      if (line >= 0) {
+        absl::StrAppend(&ret, ":", line);
+      }
+      if (col >= 0) {
+        absl::StrAppend(&ret, ":", col);
+      }
+      return ret;
+    }
   };
 
-  // Add function name and file location to symbol table.
-  void AddEntry(std::string_view func, std::string_view file_line_col) {
-    entries_.push_back({std::string(func), std::string(file_line_col)});
-  }
-
- private:
   std::vector<Entry> entries_;
 };
 

centipede/symbol_table_test.cc

@@ -41,5 +41,21 @@ TEST(SymbolTableTest, SerializesAndDeserializesCorrectly) {
   EXPECT_EQ(input, output_stream.str());
 }
 
+TEST(SymbolTableTest, SerializesAndDeserializesCorrectlyWithUnknownFile) {
+  std::string input =
+      R"(?
+    ?
+
+)";
+  std::istringstream input_stream(input);
+  SymbolTable symbol_table;
+
+  symbol_table.ReadFromLLVMSymbolizer(input_stream);
+
+  std::ostringstream output_stream;
+  symbol_table.WriteToLLVMSymbolizer(output_stream);
+  EXPECT_EQ(input, output_stream.str());
+}
+
 }  // namespace
 }  // namespace centipede