Merge pull request #1399 from m-holger/parse

Refactor calls to QPDFParser::parse

Merge pull request #1399 from m-holger/parse
Refactor calls to QPDFParser::parse
m-holger · GitHub
2 parents be25fc30 8b0eaaf7
Showing 4 changed files with 136 additions and 83 deletions
libqpdf/QPDFObjectHandle.cc
libqpdf/QPDFParser.cc
libqpdf/QPDF_objects.cc
libqpdf/qpdf/QPDFParser.hh
@@ -1498,10 +1498,7 @@ QPDFObjectHandle::parse(
     // the string.
     Buffer buf(const_cast<std::string&>(object_str));
     auto input = BufferInputSource("parsed object", &buf);
-    qpdf::Tokenizer tokenizer;
-    bool empty = false;
-    auto result = QPDFParser(input, object_description, tokenizer, nullptr, context, false)
-                      .parse(empty, false);
+    auto result = QPDFParser::parse(input, object_description, context);
     size_t offset = QIntC::to_size(input.tell());
     while (offset < object_str.length()) {
         if (!isspace(object_str.at(offset))) {
@@ -1621,7 +1618,6 @@ QPDFObjectHandle::parseContentStream_data(
     Tokenizer tokenizer;
     tokenizer.allowEOF();
     auto sp_description = QPDFParser::make_description(description, "content");
-    bool empty = false;
     while (QIntC::to_size(input.tell()) < stream_length) {
         // Read a token and seek to the beginning. The offset we get from this process is the
         // beginning of the next non-ignorable (space, comment) token. This way, the offset and
@@ -1629,8 +1625,7 @@ QPDFObjectHandle::parseContentStream_data(
         tokenizer.nextToken(input, "content", true);
         qpdf_offset_t offset = input.getLastOffset();
         input.seek(offset, SEEK_SET);
-        auto obj =
-            QPDFParser(input, sp_description, "content", tokenizer, context).parse(empty, true);
+        auto obj = QPDFParser::parse_content(input, sp_description, tokenizer, context);
         if (!obj) {
             // EOF
             break;
@@ -1690,8 +1685,7 @@ QPDFObjectHandle::parse(
     StringDecrypter* decrypter,
     QPDF* context)
 {
-    return QPDFParser(*input, object_description, tokenizer, decrypter, context, false)
-        .parse(empty, false);
+    return QPDFParser::parse(*input, object_description, tokenizer, empty, decrypter, context);
 }
 qpdf_offset_t
 #include <qpdf/QPDFParser.hh>
+#include <qpdf/BufferInputSource.hh>
 #include <qpdf/QPDF.hh>
 #include <qpdf/QPDFObjGen.hh>
 #include <qpdf/QPDFObjectHandle.hh>
@@ -15,6 +16,96 @@ using namespace std::literals;
 using ObjectPtr = std::shared_ptr<QPDFObject>;
 QPDFObjectHandle
+QPDFParser::parse(InputSource& input, std::string const& object_description, QPDF* context)
+{
+    qpdf::Tokenizer tokenizer;
+    bool empty = false;
+    return QPDFParser(
+               input,
+               make_description(input.getName(), object_description),
+               object_description,
+               tokenizer,
+               nullptr,
+               context,
+               false)
+        .parse(empty, false);
+}
+
+QPDFObjectHandle
+QPDFParser::parse_content(
+    InputSource& input,
+    std::shared_ptr<QPDFObject::Description> sp_description,
+    qpdf::Tokenizer& tokenizer,
+    QPDF* context)
+{
+    bool empty = false;
+    return QPDFParser(
+               input, std::move(sp_description), "content", tokenizer, nullptr, context, true)
+        .parse(empty, true);
+}
+
+QPDFObjectHandle
+QPDFParser::parse(
+    InputSource& input,
+    std::string const& object_description,
+    QPDFTokenizer& tokenizer,
+    bool& empty,
+    QPDFObjectHandle::StringDecrypter* decrypter,
+    QPDF* context)
+{
+    return QPDFParser(
+               input,
+               make_description(input.getName(), object_description),
+               object_description,
+               *tokenizer.m,
+               decrypter,
+               context,
+               false)
+        .parse(empty, false);
+}
+
+std::pair<QPDFObjectHandle, bool>
+QPDFParser::parse(
+    InputSource& input,
+    std::string const& object_description,
+    qpdf::Tokenizer& tokenizer,
+    QPDFObjectHandle::StringDecrypter* decrypter,
+    QPDF& context)
+{
+    bool empty{false};
+    auto result = QPDFParser(
+                      input,
+                      make_description(input.getName(), object_description),
+                      object_description,
+                      tokenizer,
+                      decrypter,
+                      &context,
+                      true)
+                      .parse(empty, false);
+    return {result, empty};
+}
+
+std::pair<QPDFObjectHandle, bool>
+QPDFParser::parse(
+    BufferInputSource& input, int stream_id, int obj_id, qpdf::Tokenizer& tokenizer, QPDF& context)
+{
+    bool empty{false};
+    auto result = QPDFParser(
+                      input,
+                      std::make_shared<QPDFObject::Description>(
+                          QPDFObject::ObjStreamDescr(stream_id, obj_id)),
+                      "",
+                      tokenizer,
+                      nullptr,
+                      &context,
+                      true,
+                      stream_id,
+                      obj_id)
+                      .parse(empty, false);
+    return {result, empty};
+}
+
+QPDFObjectHandle
 QPDFParser::parse(bool& empty, bool content_stream)
 {
     // This method must take care not to resolve any objects. Don't check the type of any object
@@ -1154,9 +1154,7 @@ QPDFObjectHandle
 QPDF::readTrailer()
 {
     qpdf_offset_t offset = m->file->tell();
-    bool empty = false;
-    auto object =
-        QPDFParser(*m->file, "trailer", m->tokenizer, nullptr, this, true).parse(empty, false);
+    auto [object, empty] = QPDFParser::parse(*m->file, "trailer", m->tokenizer, nullptr, *this);
     if (empty) {
         // Nothing in the PDF spec appears to allow empty objects, but they have been encountered in
         // actual PDF files and Adobe Reader appears to ignore them.
@@ -1174,13 +1172,11 @@ QPDF::readObject(std::string const&amp; description, QPDFObjGen og)
 {
     setLastObjectDescription(description, og);
     qpdf_offset_t offset = m->file->tell();
-    bool empty = false;
     StringDecrypter decrypter{this, og};
     StringDecrypter* decrypter_ptr = m->encp->encrypted ? &decrypter : nullptr;
-    auto object =
-        QPDFParser(*m->file, m->last_object_description, m->tokenizer, decrypter_ptr, this, true)
-            .parse(empty, false);
+    auto [object, empty] =
+        QPDFParser::parse(*m->file, m->last_object_description, m->tokenizer, decrypter_ptr, *this);
     if (empty) {
         // Nothing in the PDF spec appears to allow empty objects, but they have been encountered in
         // actual PDF files and Adobe Reader appears to ignore them.
@@ -1294,10 +1290,7 @@ QPDF::validateStreamLineEnd(QPDFObjectHandle&amp; object, QPDFObjGen og, qpdf_offset
 QPDFObjectHandle
 QPDF::readObjectInStream(BufferInputSource& input, int stream_id, int obj_id)
 {
-    bool empty = false;
-    auto object =
-        QPDFParser(input, stream_id, obj_id, m->last_object_description, m->tokenizer, this)
-            .parse(empty, false);
+    auto [object, empty] = QPDFParser::parse(input, stream_id, obj_id, m->tokenizer, *this);
     if (empty) {
         // Nothing in the PDF spec appears to allow empty objects, but they have been encountered in
         // actual PDF files and Adobe Reader appears to ignore them.
@@ -11,96 +11,70 @@
 class QPDFParser
 {
   public:
-    QPDFParser() = delete;
+    static QPDFObjectHandle
+    parse(InputSource& input, std::string const& object_description, QPDF* context);
-    // This constructor is only used by QPDFObjectHandle::parse overload taking a QPDFTokenizer.
-    // ABI: remove when removing QPDFObjectHandle::parse overload.
-    QPDFParser(
+    static QPDFObjectHandle parse_content(
+        InputSource& input,
+        std::shared_ptr<QPDFObject::Description> sp_description,
+        qpdf::Tokenizer& tokenizer,
+        QPDF* context);
+
+    // For use by deprecated QPDFObjectHandle::parse.
+    static QPDFObjectHandle parse(
         InputSource& input,
         std::string const& object_description,
         QPDFTokenizer& tokenizer,
+        bool& empty,
         QPDFObjectHandle::StringDecrypter* decrypter,
-        QPDF* context,
-        bool parse_pdf) :
-        input(input),
-        object_description(object_description),
-        tokenizer(*tokenizer.m),
-        decrypter(decrypter),
-        context(context),
-        description(make_description(input.getName(), object_description)),
-        parse_pdf(parse_pdf)
-    {
-    }
+        QPDF* context);
-    QPDFParser(
+    // For use by QPDF. Return parsed object and whether it is empty.
+    static std::pair<QPDFObjectHandle, bool> parse(
         InputSource& input,
         std::string const& object_description,
         qpdf::Tokenizer& tokenizer,
         QPDFObjectHandle::StringDecrypter* decrypter,
-        QPDF* context,
-        bool parse_pdf) :
-        input(input),
-        object_description(object_description),
-        tokenizer(tokenizer),
-        decrypter(decrypter),
-        context(context),
-        description(make_description(input.getName(), object_description)),
-        parse_pdf(parse_pdf)
-    {
-    }
+        QPDF& context);
-    // Used by parseContentStream_data only
-    QPDFParser(
-        InputSource& input,
-        std::shared_ptr<QPDFObject::Description> sp_description,
-        std::string const& object_description,
+    static std::pair<QPDFObjectHandle, bool> parse(
+        BufferInputSource& input,
+        int stream_id,
+        int obj_id,
         qpdf::Tokenizer& tokenizer,
-        QPDF* context) :
-        input(input),
-        object_description(object_description),
-        tokenizer(tokenizer),
-        decrypter(nullptr),
-        context(context),
-        description(std::move(sp_description)),
-        parse_pdf(true)
+        QPDF& context);
+
+    static std::shared_ptr<QPDFObject::Description>
+    make_description(std::string const& input_name, std::string const& object_description)
     {
+        using namespace std::literals;
+        return std::make_shared<QPDFObject::Description>(
+            input_name + ", " + object_description + " at offset $PO");
     }
-    // Used by readObjectInStream only
+  private:
     QPDFParser(
         InputSource& input,
-        int stream_id,
-        int obj_id,
+        std::shared_ptr<QPDFObject::Description> sp_description,
         std::string const& object_description,
         qpdf::Tokenizer& tokenizer,
-        QPDF* context) :
+        QPDFObjectHandle::StringDecrypter* decrypter,
+        QPDF* context,
+        bool parse_pdf,
+        int stream_id = 0,
+        int obj_id = 0) :
         input(input),
         object_description(object_description),
         tokenizer(tokenizer),
-        decrypter(nullptr),
+        decrypter(decrypter),
         context(context),
-        description(
-            std::make_shared<QPDFObject::Description>(
-                QPDFObject::ObjStreamDescr(stream_id, obj_id))),
-        parse_pdf(true),
+        description(std::move(sp_description)),
+        parse_pdf(parse_pdf),
         stream_id(stream_id),
         obj_id(obj_id)
     {
     }
-    ~QPDFParser() = default;
-
-    QPDFObjectHandle parse(bool& empty, bool content_stream);
-
-    static std::shared_ptr<QPDFObject::Description>
-    make_description(std::string const& input_name, std::string const& object_description)
-    {
-        using namespace std::literals;
-        return std::make_shared<QPDFObject::Description>(
-            input_name + ", " + object_description + " at offset $PO");
-    }
-
-  private:
     // 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 };
@@ -123,6 +97,7 @@ class QPDFParser
         int null_count{0};
     };
+    QPDFObjectHandle parse(bool& empty, bool content_stream);
     QPDFObjectHandle parseRemainder(bool content_stream);
     void add(std::shared_ptr<QPDFObject>&& obj);
     void addNull();
@@ -146,7 +121,7 @@ class QPDFParser
     QPDFObjectHandle::StringDecrypter* decrypter;
     QPDF* context;
     std::shared_ptr<QPDFObject::Description> description;
-    bool parse_pdf;
+    bool parse_pdf{false};
     int stream_id{0};
     int obj_id{0};