don't output duplicate filenames in PDBSourceLineWriter. r=mark at http://breakpad.appspot.com/43001

git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@431 4c0a9323-5329-0410-9bdc-e9ce6186880e

src/common/windows/pdb_source_line_writer.cc

@@ -118,11 +118,13 @@ bool PDBSourceLineWriter::PrintLines(IDiaEnumLineNumbers *lines) {
       return false;
     }
 
-    DWORD source_id;
+    DWORD dia_source_id;
-    if (FAILED(line->get_sourceFileId(&source_id))) {
+    if (FAILED(line->get_sourceFileId(&dia_source_id))) {
       fprintf(stderr, "failed to get line source file id\n");
       return false;
     }
+    // duplicate file names are coalesced to share one ID
+    DWORD source_id = GetRealFileID(dia_source_id);
 
     DWORD line_num;
     if (FAILED(line->get_lineNumber(&line_num))) {
@@ -216,7 +218,16 @@ bool PDBSourceLineWriter::PrintSourceFiles() {
         return false;
       }
 
-      fwprintf(output_, L"FILE %d %s\n", file_id, file_name);
+      wstring file_name_string(file_name);
+      if (!FileIDIsCached(file_name_string)) {
+        // this is a new file name, cache it and output a FILE line.
+        CacheFileID(file_name_string, file_id);
+        fwprintf(output_, L"FILE %d %s\n", file_id, file_name);
+      } else {
+        // this file name has already been seen, just save this
+        // ID for later lookup.
+        StoreDuplicateFileID(file_name_string, file_id);
+      }
       file.Release();
     }
     compiland.Release();

src/common/windows/pdb_source_line_writer.h

@@ -35,6 +35,7 @@
 
 #include <atlcomcli.h>
 
+#include <hash_map>
 #include <string>
 
 struct IDiaEnumLineNumbers;
@@ -44,6 +45,7 @@ struct IDiaSymbol;
 namespace google_breakpad {
 
 using std::wstring;
+using stdext::hash_map;
 
 // A structure that carries information that identifies a pdb file.
 struct PDBModuleInfo {
@@ -137,6 +139,37 @@ class PDBSourceLineWriter {
   // its uuid and age.
   bool PrintPDBInfo();
 
+  // Returns true if this filename has already been seen,
+  // and an ID is stored for it, or false if it has not.
+  bool FileIDIsCached(const wstring &file) {
+    return unique_files_.find(file) != unique_files_.end();
+  };
+
+  // Cache this filename and ID for later reuse.
+  void CacheFileID(const wstring &file, DWORD id) {
+    unique_files_[file] = id;
+  };
+
+  // Store this ID in the cache as a duplicate for this filename.
+  void StoreDuplicateFileID(const wstring &file, DWORD id) {
+    hash_map<wstring, DWORD>::iterator iter = unique_files_.find(file);
+    if (iter != unique_files_.end()) {
+      // map this id to the previously seen one
+      file_ids_[id] = iter->second;
+    }
+  };
+
+  // Given a file's unique ID, return the ID that should be used to
+  // reference it. There may be multiple files with identical filenames
+  // but different unique IDs. The cache attempts to coalesce these into
+  // one ID per unique filename.
+  DWORD GetRealFileID(DWORD id) {
+    hash_map<DWORD, DWORD>::iterator iter = file_ids_.find(id);
+    if (iter == file_ids_.end())
+      return id;
+    return iter->second;
+  };
+
   // Returns the function name for a symbol.  If possible, the name is
   // undecorated.  If the symbol's decorated form indicates the size of
   // parameters on the stack, this information is returned in stack_param_size.
@@ -156,6 +189,13 @@ class PDBSourceLineWriter {
   // The current output file for this WriteMap invocation.
   FILE *output_;
 
+  // There may be many duplicate filenames with different IDs.
+  // This maps from the DIA "unique ID" to a single ID per unique
+  // filename.
+  hash_map<DWORD, DWORD> file_ids_;
+  // This maps unique filenames to file IDs.
+  hash_map<wstring, DWORD> unique_files_;
+
   // Disallow copy ctor and operator=
   PDBSourceLineWriter(const PDBSourceLineWriter&);
   void operator=(const PDBSourceLineWriter&);