From 3073d0f4d1d158d46578a1a07ba2a739fbdfd2a8 Mon Sep 17 00:00:00 2001 From: Andrea Canciani Date: Tue, 2 Nov 2010 09:38:51 +0100 Subject: [PATCH 2/3] Remove cursor from TextPool It is supposed to make the searching faster, but it only works when the words are already sorted. A more general approach is needed to get acceptable performance on generic documents. --- poppler/TextOutputDev.cc | 14 ++------------ poppler/TextOutputDev.h | 2 -- 2 files changed, 2 insertions(+), 14 deletions(-) diff --git a/poppler/TextOutputDev.cc b/poppler/TextOutputDev.cc index 576bcc9..b6b92bc 100644 --- a/poppler/TextOutputDev.cc +++ b/poppler/TextOutputDev.cc @@ -512,8 +512,6 @@ TextPool::TextPool() { minBaseIdx = 0; maxBaseIdx = -1; pool = NULL; - cursor = NULL; - cursorBaseIdx = -1; } TextPool::~TextPool() { @@ -580,14 +578,8 @@ void TextPool::addWord(TextWord *word) { } // insert the new word - if (cursor && wordBaseIdx == cursorBaseIdx && - word->primaryCmp(cursor) > 0) { - w0 = cursor; - w1 = cursor->next; - } else { - w0 = NULL; - w1 = pool[wordBaseIdx - minBaseIdx]; - } + w0 = NULL; + w1 = pool[wordBaseIdx - minBaseIdx]; for (; w1 && word->primaryCmp(w1) > 0; w0 = w1, w1 = w1->next) ; word->next = w1; if (w0) { @@ -595,8 +587,6 @@ void TextPool::addWord(TextWord *word) { } else { pool[wordBaseIdx - minBaseIdx] = word; } - cursor = word; - cursorBaseIdx = wordBaseIdx; } //------------------------------------------------------------------------ diff --git a/poppler/TextOutputDev.h b/poppler/TextOutputDev.h index 438aee4..47b134b 100644 --- a/poppler/TextOutputDev.h +++ b/poppler/TextOutputDev.h @@ -230,8 +230,6 @@ private: int maxBaseIdx; // max baseline bucket index TextWord **pool; // array of linked lists, one for each // baseline value (multiple of 4 pts) - TextWord *cursor; // pointer to last-accessed word - int cursorBaseIdx; // baseline bucket index of last-accessed word friend class TextBlock; friend class TextPage; -- 1.7.0.4