From 69a5fb70472a2112d2bbb938ee3718250d2364cf Mon Sep 17 00:00:00 2001
From: m-holger <m-holger@kubitscheck.org>
Date: Fri, 26 Aug 2022 23:55:56 +0100
Subject: Add methods InputSource::fastRead, fastUnRead and fastTell

Provide buffered input for QPDFTokenizer.
---
 include/qpdf/InputSource.hh | 68 +++++++++++++++++++++++++++++++++++++++++++++
 libqpdf/QPDFTokenizer.cc    |  8 ++----
 2 files changed, 71 insertions(+), 5 deletions(-)

diff --git a/include/qpdf/InputSource.hh b/include/qpdf/InputSource.hh
index 9feb8ec3..e9d99cdb 100644
--- a/include/qpdf/InputSource.hh
+++ b/include/qpdf/InputSource.hh
@@ -93,6 +93,12 @@ class QPDF_DLL_CLASS InputSource
     // efficient.
     virtual void unreadCh(char ch) = 0;
 
+    // The following methods are for use by QPDFTokenizer
+    inline qpdf_offset_t fastTell();
+    inline bool fastRead(char&);
+    inline void fastUnread(bool);
+    inline void loadBuffer();
+
   protected:
     qpdf_offset_t last_offset;
 
@@ -111,6 +117,68 @@ class QPDF_DLL_CLASS InputSource
     };
 
     std::shared_ptr<Members> m;
+
+    // State for fast... methods
+    static const qpdf_offset_t buf_size = 128;
+    char buffer[buf_size];
+    qpdf_offset_t buf_len = 0;
+    qpdf_offset_t buf_idx = 0;
+    qpdf_offset_t buf_start = 0;
 };
 
+inline void
+InputSource::loadBuffer()
+{
+    this->buf_idx = 0;
+    this->buf_len = qpdf_offset_t(read(this->buffer, this->buf_size));
+    // NB read sets last_offset
+    this->buf_start = this->last_offset;
+}
+
+inline qpdf_offset_t
+InputSource::fastTell()
+{
+    if (this->buf_len == 0) {
+        loadBuffer();
+    } else {
+        auto curr = tell();
+        if (curr < this->buf_start ||
+            curr >= (this->buf_start + this->buf_len)) {
+            loadBuffer();
+        } else {
+            this->last_offset = curr;
+            this->buf_idx = curr - this->buf_start;
+        }
+    }
+    return this->last_offset;
+}
+
+inline bool
+InputSource::fastRead(char& ch)
+{
+    // Before calling fastRead, fastTell must be called to prepare the buffer.
+    // Once reading is complete, fastUnread must be called to set the correct
+    // file position.
+    if (this->buf_idx < this->buf_len) {
+        ch = this->buffer[this->buf_idx];
+        ++(this->buf_idx);
+        ++(this->last_offset);
+        return true;
+
+    } else if (this->buf_len == 0) {
+        return false;
+    } else {
+        seek(this->buf_start + this->buf_len, SEEK_SET);
+        fastTell();
+        return fastRead(ch);
+    }
+}
+
+inline void
+InputSource::fastUnread(bool back)
+{
+    this->last_offset -= back ? 1 : 0;
+    seek(this->last_offset, SEEK_SET);
+}
+
 #endif // QPDF_INPUTSOURCE_HH
diff --git a/libqpdf/QPDFTokenizer.cc b/libqpdf/QPDFTokenizer.cc
index 721ea7ea..56550266 100644
--- a/libqpdf/QPDFTokenizer.cc
+++ b/libqpdf/QPDFTokenizer.cc
@@ -974,11 +974,11 @@ QPDFTokenizer::readToken(
     bool allow_bad,
     size_t max_len)
 {
-    qpdf_offset_t offset = input->tell();
+    qpdf_offset_t offset = input->fastTell();
 
     while (this->state != st_token_ready) {
         char ch;
-        if (input->read(&ch, 1) == 0) {
+        if (!input->fastRead(ch)) {
             presentEOF();
 
             if ((this->type == tt_eof) && (!this->allow_eof)) {
@@ -1013,9 +1013,7 @@ QPDFTokenizer::readToken(
     bool unread_char;
     char char_to_unread;
     getToken(token, unread_char, char_to_unread);
-    if (unread_char) {
-        input->unreadCh(char_to_unread);
-    }
+    input->fastUnread(unread_char);
 
     if (token.getType() != tt_eof) {
         input->setLastOffset(offset);
-- 
cgit v1.2.3-54-g00ecf