In QPDFParser::parse create dictionaries on the fly

Also, don't search for /Contents name unless the result is used.

5 files changed, 80 insertions, 65 deletions

diff --git a/libqpdf/QPDFParser.cc b/libqpdf/QPDFParser.cc
index fd57c6f3..d2b2af6a 100644
--- a/libqpdf/QPDFParser.cc
+++ b/libqpdf/QPDFParser.cc
@@ -74,7 +74,7 @@ QPDFParser::parse(bool& empty, bool content_stream)
         stack.clear();
         stack.emplace_back(
             input,
-            (tokenizer.getType() == QPDFTokenizer::tt_array_open) ? st_array : st_dictionary);
+            (tokenizer.getType() == QPDFTokenizer::tt_array_open) ? st_array : st_dictionary_key);
         frame = &stack.back();
         return parseRemainder(content_stream);
 
@@ -242,60 +242,44 @@ QPDFParser::parseRemainder(bool content_stream)
             continue;
 
         case QPDFTokenizer::tt_dict_close:
-            if (frame->state == st_dictionary) {
-                // Convert list to map. Alternating elements are keys.  Attempt to recover more or
-                // less gracefully from invalid dictionaries.
-                std::set<std::string> names;
-                for (auto& obj: frame->olist) {
-                    if (obj) {
+            if (frame->state <= st_dictionary_value) {
+                // Attempt to recover more or less gracefully from invalid dictionaries.
+
+                auto& dict = frame->dict;
+                if (frame->state == st_dictionary_value) {
+                    QTC::TC("qpdf", "QPDFParser no val for last key");
+                    warn(
+                        frame->offset,
+                        "dictionary ended prematurely; using null as value for last key");
+                    dict[frame->key] = QPDF_Null::create();
+                }
+
+                if (!frame->olist.empty()) {
+                    std::set<std::string> names;
+                    for (auto& obj: frame->olist) {
                         if (obj->getTypeCode() == ::ot_name) {
                             names.insert(obj->getStringValue());
                         }
                     }
-                }
-
-                std::map<std::string, QPDFObjectHandle> dict;
-                int next_fake_key = 1;
-                for (auto iter = frame->olist.begin(); iter != frame->olist.end();) {
-                    // Calculate key.
-                    std::string key;
-                    if (*iter && (*iter)->getTypeCode() == ::ot_name) {
-                        key = (*iter)->getStringValue();
-                        ++iter;
-                    } else {
-                        for (bool found_fake = false; !found_fake;) {
-                            key = "/QPDFFake" + std::to_string(next_fake_key++);
-                            found_fake = (names.count(key) == 0);
+                    int next_fake_key = 1;
+                    for (auto const& item: frame->olist) {
+                        while (true) {
+                            const std::string key = "/QPDFFake" + std::to_string(next_fake_key++);
+                            const bool found_fake = (dict.count(key) == 0 && names.count(key) == 0);
                             QTC::TC("qpdf", "QPDFParser found fake", (found_fake ? 0 : 1));
+                            if (found_fake) {
+                                warn(
+                                    frame->offset,
+                                    "expected dictionary key but found non-name object; inserting "
+                                    "key " +
+                                        key);
+                                dict[key] = item;
+                                break;
+                            }
                         }
-                        warn(
-                            frame->offset,
-                            "expected dictionary key but found non-name object; inserting key " +
-                                key);
-                    }
-                    if (dict.count(key) > 0) {
-                        QTC::TC("qpdf", "QPDFParser duplicate dict key");
-                        warn(
-                            frame->offset,
-                            "dictionary has duplicated key " + key +
-                                "; last occurrence overrides earlier ones");
-                    }
-
-                    // Calculate value.
-                    ObjectPtr val;
-                    if (iter != frame->olist.end()) {
-                        val = *iter;
-                        ++iter;
-                    } else {
-                        QTC::TC("qpdf", "QPDFParser no val for last key");
-                        warn(
-                            frame->offset,
-                            "dictionary ended prematurely; using null as value for last key");
-                        val = QPDF_Null::create();
                     }
-
-                    dict[std::move(key)] = val;
                 }
+
                 if (!frame->contents_string.empty() && dict.count("/Type") &&
                     dict["/Type"].isNameAndEquals("/Sig") && dict.count("/ByteRange") &&
                     dict.count("/Contents") && dict["/Contents"].isString()) {
@@ -335,7 +319,7 @@ QPDFParser::parseRemainder(bool content_stream)
                 stack.emplace_back(
                     input,
                     (tokenizer.getType() == QPDFTokenizer::tt_array_open) ? st_array
-                                                                          : st_dictionary);
+                                                                          : st_dictionary_key);
                 frame = &stack.back();
                 continue;
             }
@@ -364,15 +348,13 @@ QPDFParser::parseRemainder(bool content_stream)
             continue;
 
         case QPDFTokenizer::tt_name:
-            {
-                auto const& name = tokenizer.getValue();
-                addScalar<QPDF_Name>(name);
-
-                if (name == "/Contents") {
-                    b_contents = true;
-                } else {
-                    b_contents = false;
-                }
+            if (frame->state == st_dictionary_key) {
+                frame->key = tokenizer.getValue();
+                frame->state = st_dictionary_value;
+                b_contents = decrypter && frame->key == "/Contents";
+                continue;
+            } else {
+                addScalar<QPDF_Name>(tokenizer.getValue());
             }
             continue;
 
@@ -415,13 +397,21 @@ QPDFParser::parseRemainder(bool content_stream)
             addNull();
         }
     }
-    return {}; // unreachable
 }
 
 void
 QPDFParser::add(std::shared_ptr<QPDFObject>&& obj)
 {
-    frame->olist.emplace_back(std::move(obj));
+    if (frame->state != st_dictionary_value) {
+        // If state is st_dictionary_key then there is a missing key. Push onto olist for
+        // processing once the tt_dict_close token has been found.
+        frame->olist.emplace_back(std::move(obj));
+    } else {
+        if (auto res = frame->dict.insert_or_assign(frame->key, std::move(obj)); !res.second) {
+            warnDuplicateKey();
+        }
+        frame->state = st_dictionary_key;
+    }
 }
 
 void
@@ -429,7 +419,16 @@ QPDFParser::addNull()
 {
     const static ObjectPtr null_obj = QPDF_Null::create();
 
-    frame->olist.emplace_back(null_obj);
+    if (frame->state != st_dictionary_value) {
+        // If state is st_dictionary_key then there is a missing key. Push onto olist for
+        // processing once the tt_dict_close token has been found.
+        frame->olist.emplace_back(null_obj);
+    } else {
+        if (auto res = frame->dict.insert_or_assign(frame->key, null_obj); !res.second) {
+            warnDuplicateKey();
+        }
+        frame->state = st_dictionary_key;
+    }
     ++frame->null_count;
 }
 
@@ -496,6 +495,15 @@ QPDFParser::warn(QPDFExc const& e) const
 }
 
 void
+QPDFParser::warnDuplicateKey()
+{
+    QTC::TC("qpdf", "QPDFParser duplicate dict key");
+    warn(
+        frame->offset,
+        "dictionary has duplicated key " + frame->key + "; last occurrence overrides earlier ones");
+}
+
+void
 QPDFParser::warn(qpdf_offset_t offset, std::string const& msg) const
 {
     warn(QPDFExc(qpdf_e_damaged_pdf, input->getName(), object_description, offset, msg));
diff --git a/libqpdf/qpdf/QPDFParser.hh b/libqpdf/qpdf/QPDFParser.hh
index ef5be98e..3abe6c92 100644
--- a/libqpdf/qpdf/QPDFParser.hh
+++ b/libqpdf/qpdf/QPDFParser.hh
@@ -31,8 +31,9 @@ class QPDFParser
     QPDFObjectHandle parse(bool& empty, bool content_stream);
 
   private:
-    struct StackFrame;
-    enum parser_state_e { st_dictionary, st_array };
+    // Parser state.  Note:
+    // state < st_dictionary_value == (state = st_dictionary_key || state = st_dictionary_value)
+    enum parser_state_e { st_dictionary_key, st_dictionary_value, st_array };
 
     struct StackFrame
     {
@@ -43,7 +44,9 @@ class QPDFParser
         }
 
         std::vector<std::shared_ptr<QPDFObject>> olist;
+        std::map<std::string, QPDFObjectHandle> dict;
         parser_state_e state;
+        std::string key;
         qpdf_offset_t offset;
         std::string contents_string;
         qpdf_offset_t contents_offset{-1};
@@ -57,6 +60,7 @@ class QPDFParser
     template <typename T, typename... Args>
     void addScalar(Args&&... args);
     bool tooManyBadTokens();
+    void warnDuplicateKey();
     void warn(qpdf_offset_t offset, std::string const& msg) const;
     void warn(std::string const& msg) const;
     void warn(QPDFExc const&) const;
@@ -83,7 +87,6 @@ class QPDFParser
     int int_count = 0;
     long long int_buffer[2]{0, 0};
     qpdf_offset_t last_offset_buffer[2]{0, 0};
-
 };
 
 #endif // QPDFPARSER_HH
diff --git a/qpdf/qtest/qpdf/bad36-recover.out b/qpdf/qtest/qpdf/bad36-recover.out
index ac05acd9..9aacd729 100644
--- a/qpdf/qtest/qpdf/bad36-recover.out
+++ b/qpdf/qtest/qpdf/bad36-recover.out
@@ -1,6 +1,6 @@
 WARNING: bad36.pdf (trailer, offset 764): unknown token while reading object; treating as string
-WARNING: bad36.pdf (trailer, offset 715): expected dictionary key but found non-name object; inserting key /QPDFFake2
 WARNING: bad36.pdf (trailer, offset 715): dictionary ended prematurely; using null as value for last key
+WARNING: bad36.pdf (trailer, offset 715): expected dictionary key but found non-name object; inserting key /QPDFFake2
 /QTest is implicit
 /QTest is direct and has type null (2)
 /QTest is null
diff --git a/qpdf/qtest/qpdf/bad36.out b/qpdf/qtest/qpdf/bad36.out
index cee3c286..e60d8685 100644
--- a/qpdf/qtest/qpdf/bad36.out
+++ b/qpdf/qtest/qpdf/bad36.out
@@ -1,6 +1,6 @@
 WARNING: bad36.pdf (trailer, offset 764): unknown token while reading object; treating as string
-WARNING: bad36.pdf (trailer, offset 715): expected dictionary key but found non-name object; inserting key /QPDFFake2
 WARNING: bad36.pdf (trailer, offset 715): dictionary ended prematurely; using null as value for last key
+WARNING: bad36.pdf (trailer, offset 715): expected dictionary key but found non-name object; inserting key /QPDFFake2
 /QTest is implicit
 /QTest is direct and has type null (2)
 /QTest is null
diff --git a/qpdf/qtest/qpdf/issue-335a.out b/qpdf/qtest/qpdf/issue-335a.out
index 456bc475..c5b64465 100644
--- a/qpdf/qtest/qpdf/issue-335a.out
+++ b/qpdf/qtest/qpdf/issue-335a.out
@@ -51,6 +51,7 @@ WARNING: issue-335a.pdf (trailer, offset 563): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 596): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 597): name with stray # will not work with PDF >= 1.2
 WARNING: issue-335a.pdf (trailer, offset 600): unexpected )
+WARNING: issue-335a.pdf (trailer, offset 134): dictionary has duplicated key /L
 WARNING: issue-335a.pdf (trailer, offset 601): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 648): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 649): name with stray # will not work with PDF >= 1.2
@@ -74,6 +75,7 @@ WARNING: issue-335a.pdf (trailer, offset 563): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 596): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 597): name with stray # will not work with PDF >= 1.2
 WARNING: issue-335a.pdf (trailer, offset 600): unexpected )
+WARNING: issue-335a.pdf (trailer, offset 164): dictionary has duplicated key /L
 WARNING: issue-335a.pdf (trailer, offset 601): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 648): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 649): name with stray # will not work with PDF >= 1.2
@@ -97,6 +99,7 @@ WARNING: issue-335a.pdf (trailer, offset 563): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 596): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 597): name with stray # will not work with PDF >= 1.2
 WARNING: issue-335a.pdf (trailer, offset 600): unexpected )
+WARNING: issue-335a.pdf (trailer, offset 231): dictionary has duplicated key /L
 WARNING: issue-335a.pdf (trailer, offset 601): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 648): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 649): name with stray # will not work with PDF >= 1.2
@@ -448,6 +451,7 @@ WARNING: issue-335a.pdf (trailer, offset 1168): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 1328): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 1329): name with stray # will not work with PDF >= 1.2
 WARNING: issue-335a.pdf (trailer, offset 1332): unexpected )
+WARNING: issue-335a.pdf (trailer, offset 1033): dictionary has duplicated key /L
 WARNING: issue-335a.pdf (trailer, offset 1333): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 1344): unexpected )
 WARNING: issue-335a.pdf (trailer, offset 1428): unexpected )