Implement findFirst and findLast in InputSource

Preparing to refactor some pattern searching code to use these instead
of their own memchr loops. This should simplify the code that replaces
PCRE.

5 files changed, 157 insertions, 0 deletions

diff --git a/libtests/build.mk b/libtests/build.mk
index 22d9299e..2a272799 100644
--- a/libtests/build.mk
+++ b/libtests/build.mk
@@ -6,6 +6,7 @@ BINS_libtests = \
 	concatenate \
 	flate \
 	hex \
+	input_source \
 	lzw \
 	md5 \
 	pcre \
diff --git a/libtests/input_source.cc b/libtests/input_source.cc
new file mode 100644
index 00000000..091a1ea1
--- /dev/null
+++ b/libtests/input_source.cc
@@ -0,0 +1,108 @@
+#include <iostream>
+#include <qpdf/BufferInputSource.hh>
+#include <qpdf/PointerHolder.hh>
+#include <qpdf/Buffer.hh>
+#include <qpdf/QPDFTokenizer.hh>
+
+static PointerHolder<Buffer>
+get_buffer()
+{
+    size_t size = 3172;
+    PointerHolder<Buffer> b(new Buffer(size));
+    unsigned char* p = b->getBuffer();
+    for (size_t i = 0; i < size; ++i)
+    {
+        p[i] = static_cast<unsigned char>(i & 0xff);
+    }
+    return b;
+}
+
+class Finder: public InputSource::Finder
+{
+  public:
+    Finder(PointerHolder<InputSource> is, std::string const& after) :
+        is(is),
+        after(after)
+    {
+    }
+    virtual ~Finder()
+    {
+    }
+    virtual bool check();
+
+  private:
+    PointerHolder<InputSource> is;
+    std::string after;
+};
+
+bool
+Finder::check()
+{
+    QPDFTokenizer tokenizer;
+    QPDFTokenizer::Token t = tokenizer.readToken(is, "finder", true);
+    if (t == QPDFTokenizer::Token(QPDFTokenizer::tt_word, "potato"))
+    {
+        t = tokenizer.readToken(is, "finder", true);
+        return (t == QPDFTokenizer::Token(QPDFTokenizer::tt_word, after));
+    }
+    return false;
+}
+
+void check(char const* description, bool expected, bool actual)
+{
+    std::cout << description << ": "
+              << ((actual == expected) ? "PASS" : "FAIL")
+              << std::endl;
+}
+
+int main()
+{
+    PointerHolder<Buffer> b1 = get_buffer();
+    unsigned char* b = b1->getBuffer();
+    // Straddle block boundaries
+    memcpy(b + 1022, "potato", 6);
+    // Overlap so that the first check() would advance past the start
+    // of the next match
+    memcpy(b + 2037, "potato potato salad ", 20);
+    PointerHolder<InputSource> is =
+        new BufferInputSource("test buffer input source", b1.getPointer());
+    Finder f1(is, "salad");
+    check("find potato salad", true,
+          is->findFirst("potato", 0, 0, f1));
+    check("barely find potato salad", true,
+          is->findFirst("potato", 1100, 945, f1));
+    check("barely find potato salad", true,
+          is->findFirst("potato", 2000, 45, f1));
+    check("potato salad is too late", false,
+          is->findFirst("potato", 1100, 944, f1));
+    check("potato salad is too late", false,
+          is->findFirst("potato", 2000, 44, f1));
+    check("potato salad not found", false,
+          is->findFirst("potato", 2045, 0, f1));
+    check("potato salad not found", false,
+          is->findFirst("potato", 0, 1, f1));
+
+    // Put one more right at EOF
+    memcpy(b + b1->getSize() - 12, "potato salad", 12);
+    check("potato salad at EOF", true,
+          is->findFirst("potato", 3000, 0, f1));
+
+    is->findFirst("potato", 0, 0, f1);
+    check("findFirst found first", true,
+          is->tell() == 2056);
+    check("findLast found potato salad", true,
+          is->findLast("potato", 0, 0, f1));
+    check("findLast found at EOF", true,
+          is->tell() == 3172);
+
+    // Make check() bump into EOF
+    memcpy(b + b1->getSize() - 6, "potato", 6);
+    check("potato but not salad salad at EOF", false,
+          is->findFirst("potato", 3000, 0, f1));
+    check("findLast found potato salad", true,
+          is->findLast("potato", 0, 0, f1));
+    check("findLast found first one", true,
+          is->tell() == 2056);
+
+    return 0;
+}
diff --git a/libtests/libtests.testcov b/libtests/libtests.testcov
index ddbccd24..a5fe625f 100644
--- a/libtests/libtests.testcov
+++ b/libtests/libtests.testcov
@@ -16,3 +16,11 @@ bits write zero bits 0
 Pl_ASCIIHexDecoder ignore space 0
 Pl_ASCIIHexDecoder no-op flush 0
 Pl_ASCIIHexDecoder partial flush 1
+InputSource read next block 1
+InputSource find EOF 1
+InputSource out of range 0
+InputSource first char matched but not string 0
+InputSource start_chars matched but not check 0
+InputSource not enough bytes 0
+InputSource findLast found more than one 0
+InputSource found match at buf[0] 0
diff --git a/libtests/qtest/input_source.test b/libtests/qtest/input_source.test
new file mode 100644
index 00000000..89a1c21a
--- /dev/null
+++ b/libtests/qtest/input_source.test
@@ -0,0 +1,26 @@
+#!/usr/bin/env perl
+require 5.008;
+use warnings;
+use strict;
+
+chdir("input_source") or die "chdir testdir failed: $!\n";
+
+require TestDriver;
+
+my $td = new TestDriver('InputSource');
+
+cleanup();
+
+$td->runtest("input source tests",
+	     {$td->COMMAND => "input_source"},
+	     {$td->FILE => "input_source.out",
+	      $td->EXIT_STATUS => 0},
+             $td->NORMALIZE_NEWLINES);
+
+cleanup();
+
+$td->report(1);
+
+sub cleanup
+{
+}
diff --git a/libtests/qtest/input_source/input_source.out b/libtests/qtest/input_source/input_source.out
new file mode 100644
index 00000000..6af379c6
--- /dev/null
+++ b/libtests/qtest/input_source/input_source.out
@@ -0,0 +1,14 @@
+find potato salad: PASS
+barely find potato salad: PASS
+barely find potato salad: PASS
+potato salad is too late: PASS
+potato salad is too late: PASS
+potato salad not found: PASS
+potato salad not found: PASS
+potato salad at EOF: PASS
+findFirst found first: PASS
+findLast found potato salad: PASS
+findLast found at EOF: PASS
+potato but not salad salad at EOF: PASS
+findLast found potato salad: PASS
+findLast found first one: PASS