Refactor `ResourceFinder`: streamline logic and optimize data structure handling.

m-holger
1 parent eca6d7ab
Showing 4 changed files with 47 additions and 57 deletions
libqpdf/QPDFAcroFormDocumentHelper.cc
libqpdf/QPDFPageObjectHelper.cc
libqpdf/ResourceFinder.cc
libqpdf/qpdf/ResourceFinder.hh
@@ -489,7 +489,7 @@ namespace
       public:
         ResourceReplacer(
             std::map<std::string, std::map<std::string, std::string>> const& dr_map,
-            std::map<std::string, std::map<std::string, std::set<size_t>>> const& rnames);
+            std::map<std::string, std::map<std::string, std::vector<size_t>>> const& rnames);
         ~ResourceReplacer() override = default;
         void handleToken(QPDFTokenizer::Token const&) override;
  
@@ -501,7 +501,7 @@ namespace
  
 ResourceReplacer::ResourceReplacer(
     std::map<std::string, std::map<std::string, std::string>> const& dr_map,
-    std::map<std::string, std::map<std::string, std::set<size_t>>> const& rnames)
+    std::map<std::string, std::map<std::string, std::vector<size_t>>> const& rnames)
 {
     // We have:
     // * dr_map[resource_type][key] == new_key
@@ -510,22 +510,18 @@ ResourceReplacer::ResourceReplacer(
     // We want:
     // * to_replace[key][offset] = new_key
  
-    for (auto const& rn_iter: rnames) {
-        std::string const& rtype = rn_iter.first;
+    for (auto const& [rtype, key_offsets]: rnames) {
         auto dr_map_rtype = dr_map.find(rtype);
         if (dr_map_rtype == dr_map.end()) {
             continue;
         }
-        auto const& key_offsets = rn_iter.second;
-        for (auto const& ko_iter: key_offsets) {
-            std::string const& old_key = ko_iter.first;
+        for (auto const& [old_key, offsets]: key_offsets) {
             auto dr_map_rtype_old = dr_map_rtype->second.find(old_key);
             if (dr_map_rtype_old == dr_map_rtype->second.end()) {
                 continue;
             }
-            auto const& offsets = ko_iter.second;
-            for (auto const& o_iter: offsets) {
-                to_replace[old_key][o_iter] = dr_map_rtype_old->second;
+            for (auto const& offs: offsets) {
+                to_replace[old_key][offs] = dr_map_rtype_old->second;
             }
         }
     }
@@ -534,19 +530,20 @@ ResourceReplacer::ResourceReplacer(
 void
 ResourceReplacer::handleToken(QPDFTokenizer::Token const& token)
 {
-    bool wrote = false;
     if (token.getType() == QPDFTokenizer::tt_name) {
-        std::string name = QPDFObjectHandle::newName(token.getValue()).getName();
-        if (to_replace.contains(name) && to_replace[name].contains(offset)) {
-            QTC::TC("qpdf", "QPDFAcroFormDocumentHelper replaced DA token");
-            write(to_replace[name][offset]);
-            wrote = true;
+        auto it1 = to_replace.find(token.getValue());
+        if (it1 != to_replace.end()) {
+            auto it2 = it1->second.find(offset);
+            if (it2 != it1->second.end()) {
+                QTC::TC("qpdf", "QPDFAcroFormDocumentHelper replaced DA token");
+                write(it2->second);
+                offset += token.getRawValue().length();
+                return;
+            }
         }
     }
     offset += token.getRawValue().length();
-    if (!wrote) {
-        writeToken(token);
-    }
+    writeToken(token);
 }
  
 void
@@ -543,14 +543,14 @@ bool
 QPDFPageObjectHelper::removeUnreferencedResourcesHelper(
     QPDFPageObjectHelper ph, std::set<std::string>& unresolved)
 {
-    bool is_page = (!ph.oh().isFormXObject());
+    const bool is_page = !ph.oh().isFormXObject();
     if (!is_page) {
         QTC::TC("qpdf", "QPDFPageObjectHelper filter form xobject");
     }
  
     ResourceFinder rf;
     try {
-        auto q = ph.oh().getOwningQPDF();
+        auto q = ph.qpdf();
         size_t before_nw = (q ? q->numWarnings() : 0);
         ph.parseContents(&rf);
         size_t after_nw = (q ? q->numWarnings() : 0);
@@ -575,23 +575,25 @@ QPDFPageObjectHelper::removeUnreferencedResourcesHelper(
     QPDFObjectHandle resources = ph.getAttribute("/Resources", true);
     std::vector<QPDFObjectHandle> rdicts;
     std::set<std::string> known_names;
-    std::vector<std::string> to_filter = {"/Font", "/XObject"};
     if (resources.isDictionary()) {
-        for (auto const& iter: to_filter) {
+        for (auto const& iter: {"/Font", "/XObject"}) {
             QPDFObjectHandle dict = resources.getKey(iter);
             if (dict.isDictionary()) {
                 dict = resources.replaceKeyAndGetNew(iter, dict.shallowCopy());
                 rdicts.push_back(dict);
-                auto keys = dict.getKeys();
-                known_names.insert(keys.begin(), keys.end());
+                known_names.merge(dict.getKeys());
             }
         }
     }
  
     std::set<std::string> local_unresolved;
-    auto names_by_rtype = rf.getNamesByResourceType();
-    for (auto const& i1: to_filter) {
-        for (auto const& n_iter: names_by_rtype[i1]) {
+    auto const& names_by_rtype = rf.getNamesByResourceType();
+    for (auto const& i1: {"/Font", "/XObject"}) {
+        auto it = names_by_rtype.find(i1);
+        if (it == names_by_rtype.end()) {
+            continue;
+        }
+        for (auto const& n_iter: it->second) {
             std::string const& name = n_iter.first;
             if (!known_names.contains(name)) {
                 unresolved.insert(name);
@@ -610,7 +612,7 @@ QPDFPageObjectHelper::removeUnreferencedResourcesHelper(
     // unresolved names, and for page objects, we avoid removing any such names found in nested form
     // XObjects.
  
-    if ((!local_unresolved.empty()) && resources.isDictionary()) {
+    if (!local_unresolved.empty() && resources.isDictionary()) {
         // It's not worth issuing a warning for this case. From qpdf 10.3, we are hopefully only
         // looking at names that are referencing fonts and XObjects, but until we're certain that we
         // know the meaning of every name in a content stream, we don't want to give warnings that
@@ -3,8 +3,8 @@
 void
 ResourceFinder::handleObject(QPDFObjectHandle obj, size_t offset, size_t)
 {
-    if (obj.isOperator() && (!this->last_name.empty())) {
-        static std::map<std::string, std::string> op_to_rtype = {
+    if (obj.isOperator() && !last_name.empty()) {
+        static const std::map<std::string, std::string> op_to_rtype{
             {"CS", "/ColorSpace"},
             {"cs", "/ColorSpace"},
             {"gs", "/ExtGState"},
@@ -17,19 +17,15 @@ ResourceFinder::handleObject(QPDFObjectHandle obj, size_t offset, size_t)
             {"Do", "/XObject"},
         };
         std::string op = obj.getOperatorValue();
-        std::string resource_type;
         auto iter = op_to_rtype.find(op);
-        if (iter != op_to_rtype.end()) {
-            resource_type = iter->second;
-        }
-        if (!resource_type.empty()) {
-            this->names.insert(this->last_name);
-            this->names_by_resource_type[resource_type][this->last_name].insert(
-                this->last_name_offset);
+        if (iter == op_to_rtype.end()) {
+            return;
         }
+        names.insert(last_name);
+        names_by_resource_type[iter->second][last_name].push_back(last_name_offset);
     } else if (obj.isName()) {
-        this->last_name = obj.getName();
-        this->last_name_offset = offset;
+        last_name = obj.getName();
+        last_name_offset = offset;
     }
 }
  
@@ -37,15 +33,3 @@ void
 ResourceFinder::handleEOF()
 {
 }
-
-std::set<std::string> const&
-ResourceFinder::getNames() const
-{
-    return this->names;
-}
-
-std::map<std::string, std::map<std::string, std::set<size_t>>> const&
-ResourceFinder::getNamesByResourceType() const
-{
-    return this->names_by_resource_type;
-}
@@ -10,15 +10,22 @@ class ResourceFinder final: public QPDFObjectHandle::ParserCallbacks
     ~ResourceFinder() final = default;
     void handleObject(QPDFObjectHandle, size_t, size_t) final;
     void handleEOF() final;
-    std::set<std::string> const& getNames() const;
-    std::map<std::string, std::map<std::string, std::set<size_t>>> const&
-    getNamesByResourceType() const;
+    std::set<std::string> const&
+    getNames() const
+    {
+        return names;
+    }
+    std::map<std::string, std::map<std::string, std::vector<size_t>>> const&
+    getNamesByResourceType() const
+    {
+        return names_by_resource_type;
+    }
  
   private:
     std::string last_name;
     size_t last_name_offset{0};
     std::set<std::string> names;
-    std::map<std::string, std::map<std::string, std::set<size_t>>> names_by_resource_type;
+    std::map<std::string, std::map<std::string, std::vector<size_t>>> names_by_resource_type;
 };
  
 #endif // RESOURCEFINDER_HH