Move parameters from CPDF_TextPageFind::FindFirst() to ctor.
Then remove redundant members and make more members const.
Also simplify FindFirst() to remove unnecessary work.
Change-Id: I9d4de76c3a05600b007224d8316a1c08999a5b52
Reviewed-on: https://pdfium-review.googlesource.com/c/34855
Reviewed-by: Tom Sepez <tsepez@chromium.org>
Commit-Queue: Lei Zhang <thestig@chromium.org>
diff --git a/core/fpdftext/cpdf_textpagefind.cpp b/core/fpdftext/cpdf_textpagefind.cpp
index 248e61a..093d335 100644
--- a/core/fpdftext/cpdf_textpagefind.cpp
+++ b/core/fpdftext/cpdf_textpagefind.cpp
@@ -32,62 +32,51 @@
return true;
}
-} // namespace
+WideString GetStringCase(const WideString& wsOriginal, bool bMatchCase) {
+ if (bMatchCase)
+ return wsOriginal;
-CPDF_TextPageFind::CPDF_TextPageFind(const CPDF_TextPage* pTextPage)
- : m_pTextPage(pTextPage) {
- ASSERT(m_pTextPage);
-
- m_strText = m_pTextPage->GetAllPageText();
+ WideString wsLower = wsOriginal;
+ wsLower.MakeLower();
+ return wsLower;
}
-CPDF_TextPageFind::~CPDF_TextPageFind() {}
+} // namespace
+
+CPDF_TextPageFind::CPDF_TextPageFind(const CPDF_TextPage* pTextPage,
+ const WideString& findwhat,
+ const Options& options,
+ Optional<size_t> startPos)
+ : m_pTextPage(pTextPage),
+ m_strText(GetStringCase(pTextPage->GetAllPageText(), options.bMatchCase)),
+ m_findWhat(GetStringCase(findwhat, options.bMatchCase)),
+ m_options(options) {
+ if (!m_strText.IsEmpty()) {
+ m_findNextStart = startPos;
+ m_findPreStart = startPos.value_or(m_strText.GetLength() - 1);
+ }
+}
+
+CPDF_TextPageFind::~CPDF_TextPageFind() = default;
int CPDF_TextPageFind::GetCharIndex(int index) const {
return m_pTextPage->CharIndexFromTextIndex(index);
}
-bool CPDF_TextPageFind::FindFirst(const WideString& findwhat,
- const Options& options,
- Optional<size_t> startPos) {
- if (m_strText.IsEmpty() || m_bMatchCase != options.bMatchCase)
- m_strText = m_pTextPage->GetAllPageText();
- WideString findwhatStr = findwhat;
- m_findWhat = findwhatStr;
- m_options = options;
- m_bMatchCase = options.bMatchCase;
+bool CPDF_TextPageFind::FindFirst() {
if (m_strText.IsEmpty())
return true;
- size_t len = findwhatStr.GetLength();
- if (!m_bMatchCase) {
- findwhatStr.MakeLower();
- m_strText.MakeLower();
- }
- m_bMatchWholeWord = options.bMatchWholeWord;
- m_findNextStart = startPos;
- if (!startPos.has_value()) {
- if (!m_strText.IsEmpty())
- m_findPreStart = m_strText.GetLength() - 1;
- } else {
- m_findPreStart = startPos;
- }
-
- m_csFindWhatArray.clear();
+ size_t len = m_findWhat.GetLength();
size_t i = 0;
for (i = 0; i < len; ++i)
- if (findwhatStr[i] != ' ')
+ if (m_findWhat[i] != ' ')
break;
if (i < len)
- ExtractFindWhat(findwhatStr);
+ ExtractFindWhat();
else
- m_csFindWhatArray.push_back(findwhatStr);
- if (m_csFindWhatArray.empty())
- return false;
-
- m_resStart = 0;
- m_resEnd = -1;
- return true;
+ m_csFindWhatArray.push_back(m_findWhat);
+ return !m_csFindWhatArray.empty();
}
bool CPDF_TextPageFind::FindNext() {
@@ -158,7 +147,7 @@
}
}
}
- if (m_bMatchWholeWord && bMatch) {
+ if (m_options.bMatchWholeWord && bMatch) {
bMatch = IsMatchWholeWord(m_strText, nResultPos.value(), endIndex);
}
nStartPos = endIndex + 1;
@@ -185,8 +174,8 @@
if (m_strText.IsEmpty() || !m_findPreStart.has_value())
return false;
- CPDF_TextPageFind find_engine(m_pTextPage.Get());
- if (!find_engine.FindFirst(m_findWhat, m_options, 0))
+ CPDF_TextPageFind find_engine(m_pTextPage.Get(), m_findWhat, m_options, 0);
+ if (!find_engine.FindFirst())
return false;
int order = -1;
@@ -216,13 +205,14 @@
return true;
}
-void CPDF_TextPageFind::ExtractFindWhat(const WideString& findwhat) {
- if (findwhat.IsEmpty())
+void CPDF_TextPageFind::ExtractFindWhat() {
+ if (m_findWhat.IsEmpty())
return;
+
int index = 0;
while (1) {
Optional<WideString> word =
- ExtractSubString(findwhat.c_str(), index, TEXT_SPACE_CHAR);
+ ExtractSubString(m_findWhat.c_str(), index, TEXT_SPACE_CHAR);
if (!word)
break;
diff --git a/core/fpdftext/cpdf_textpagefind.h b/core/fpdftext/cpdf_textpagefind.h
index 1f73918..9f3ac13 100644
--- a/core/fpdftext/cpdf_textpagefind.h
+++ b/core/fpdftext/cpdf_textpagefind.h
@@ -25,19 +25,22 @@
bool bConsecutive = false;
};
- explicit CPDF_TextPageFind(const CPDF_TextPage* pTextPage);
+ CPDF_TextPageFind(const CPDF_TextPage* pTextPage,
+ const WideString& findwhat,
+ const Options& options,
+ Optional<size_t> startPos);
~CPDF_TextPageFind();
- bool FindFirst(const WideString& findwhat,
- const Options& options,
- Optional<size_t> startPos);
+ // Should be called immediately after construction.
+ bool FindFirst();
+
bool FindNext();
bool FindPrev();
int GetCurOrder() const;
int GetMatchedCount() const;
protected:
- void ExtractFindWhat(const WideString& findwhat);
+ void ExtractFindWhat();
bool IsMatchWholeWord(const WideString& csPageText,
size_t startPos,
size_t endPos);
@@ -48,16 +51,14 @@
private:
UnownedPtr<const CPDF_TextPage> const m_pTextPage;
- WideString m_strText;
- WideString m_findWhat;
+ const WideString m_strText;
+ const WideString m_findWhat;
std::vector<WideString> m_csFindWhatArray;
Optional<size_t> m_findNextStart;
Optional<size_t> m_findPreStart;
int m_resStart = 0;
int m_resEnd = -1;
- Options m_options;
- bool m_bMatchCase = false;
- bool m_bMatchWholeWord = false;
+ const Options m_options;
};
#endif // CORE_FPDFTEXT_CPDF_TEXTPAGEFIND_H_
diff --git a/fpdfsdk/fpdf_text.cpp b/fpdfsdk/fpdf_text.cpp
index a6e5c5f..9934c0c 100644
--- a/fpdfsdk/fpdf_text.cpp
+++ b/fpdfsdk/fpdf_text.cpp
@@ -276,10 +276,11 @@
options.bMatchWholeWord = !!(flags & FPDF_MATCHWHOLEWORD);
options.bConsecutive = !!(flags & FPDF_CONSECUTIVE);
auto find = pdfium::MakeUnique<CPDF_TextPageFind>(
- CPDFTextPageFromFPDFTextPage(text_page));
- find->FindFirst(
+ CPDFTextPageFromFPDFTextPage(text_page),
WideStringFromFPDFWideString(findwhat), options,
start_index >= 0 ? Optional<size_t>(start_index) : pdfium::nullopt);
+ find->FindFirst();
+
// Caller takes ownership.
return FPDFSchHandleFromCPDFTextPageFind(find.release());
}
