Sakurai/pdf.js
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #10066 from timvandermeij/find-controller

Browse Source 
Refactor the find controller
This commit is contained in:
Tim van der Meij 2018-09-15 20:38:08 +02:00 committed by GitHub
commit ed32f6a082
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 141 additions and 140 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

266
web/pdf_find_controller.js
View File

@ -16,7 +16,6 @@
import { createPromiseCapability } from 'pdfjs-lib'; import { createPromiseCapability } from 'pdfjs-lib';
import { getCharacterType } from './pdf_find_utils'; import { getCharacterType } from './pdf_find_utils';
import { getGlobalEventBus } from './dom_events'; import { getGlobalEventBus } from './dom_events';
import { scrollIntoView } from './ui_utils';
const FindState = { const FindState = {
FOUND: 0, FOUND: 0,
@ -25,8 +24,6 @@ const FindState = {
PENDING: 3, PENDING: 3,
}; };
const FIND_SCROLL_OFFSET_TOP = -50;
const FIND_SCROLL_OFFSET_LEFT = -400;
const FIND_TIMEOUT = 250; // ms const FIND_TIMEOUT = 250; // ms
const CHARACTERS_TO_NORMALIZE = { const CHARACTERS_TO_NORMALIZE = {
@ -48,8 +45,8 @@ const CHARACTERS_TO_NORMALIZE = {
*/ */
class PDFFindController { class PDFFindController {
constructor({ pdfViewer, eventBus = getGlobalEventBus(), }) { constructor({ pdfViewer, eventBus = getGlobalEventBus(), }) {
this.pdfViewer = pdfViewer; this._pdfViewer = pdfViewer;
this.eventBus = eventBus; this._eventBus = eventBus;
this.onUpdateResultsCount = null; this.onUpdateResultsCount = null;
this.onUpdateState = null; this.onUpdateState = null;
@ -57,35 +54,50 @@ class PDFFindController {
this.reset(); this.reset();
// Compile the regular expression for text normalization once. // Compile the regular expression for text normalization once.
let replace = Object.keys(CHARACTERS_TO_NORMALIZE).join(''); const replace = Object.keys(CHARACTERS_TO_NORMALIZE).join('');
this.normalizationRegex = new RegExp('[' + replace + ']', 'g'); this._normalizationRegex = new RegExp(`[${replace}]`, 'g');
}
get pageMatches() {
return this._pageMatches;
}
get pageMatchesLength() {
return this._pageMatchesLength;
}
get selected() {
return this._selected;
}
get state() {
return this._state;
} }
reset() { reset() {
this.startedTextExtraction = false;
this.extractTextPromises = [];
this.pendingFindMatches = Object.create(null);
this.active = false; // If active, find results will be highlighted. this.active = false; // If active, find results will be highlighted.
this.pageContents = []; // Stores the text for each page. this._pageMatches = [];
this.pageMatches = []; this._pageMatchesLength = null;
this.pageMatchesLength = null; this._state = null;
this.matchesCountTotal = 0; this._selected = { // Currently selected match.
this.selected = { // Currently selected match.
pageIdx: -1, pageIdx: -1,
matchIdx: -1, matchIdx: -1,
}; };
this.offset = { // Where the find algorithm currently is in the document. this._offset = { // Where the find algorithm currently is in the document.
pageIdx: null, pageIdx: null,
matchIdx: null, matchIdx: null,
}; };
this.pagesToSearch = null; this._extractTextPromises = [];
this.resumePageIdx = null; this._pageContents = []; // Stores the text for each page.
this.state = null; this._matchesCountTotal = 0;
this.dirtyMatch = false; this._pagesToSearch = null;
this.findTimeout = null; this._pendingFindMatches = Object.create(null);
this._resumePageIdx = null;
this._dirtyMatch = false;
this._findTimeout = null;
this._firstPagePromise = new Promise((resolve) => { this._firstPagePromise = new Promise((resolve) => {
const eventBus = this.eventBus; const eventBus = this._eventBus;
eventBus.on('pagesinit', function onPagesInit() { eventBus.on('pagesinit', function onPagesInit() {
eventBus.off('pagesinit', onPagesInit); eventBus.off('pagesinit', onPagesInit);
resolve(); resolve();
@ -94,48 +106,29 @@ class PDFFindController {
} }
executeCommand(cmd, state) { executeCommand(cmd, state) {
if (this.state === null || cmd !== 'findagain') { if (this._state === null || cmd !== 'findagain') {
this.dirtyMatch = true; this._dirtyMatch = true;
} }
this.state = state; this._state = state;
this._updateUIState(FindState.PENDING); this._updateUIState(FindState.PENDING);
this._firstPagePromise.then(() => { this._firstPagePromise.then(() => {
this._extractText(); this._extractText();
clearTimeout(this.findTimeout); clearTimeout(this._findTimeout);
if (cmd === 'find') { if (cmd === 'find') {
// Trigger the find action with a small delay to avoid starting the // Trigger the find action with a small delay to avoid starting the
// search when the user is still typing (saving resources). // search when the user is still typing (saving resources).
this.findTimeout = setTimeout(this._nextMatch.bind(this), FIND_TIMEOUT); this._findTimeout =
setTimeout(this._nextMatch.bind(this), FIND_TIMEOUT);
} else { } else {
this._nextMatch(); this._nextMatch();
} }
}); });
} }
/**
* Called from the text layer when match presentation is updated.
*
* @param {number} pageIndex - The index of the page.
* @param {number} matchIndex - The index of the match.
* @param {Array} elements - Text layer `div` elements.
* @param {number} beginIdx - Start index of the `div` array for the match.
*/
updateMatchPosition(pageIndex, matchIndex, elements, beginIdx) {
if (this.selected.matchIdx === matchIndex &&
this.selected.pageIdx === pageIndex) {
let spot = {
top: FIND_SCROLL_OFFSET_TOP,
left: FIND_SCROLL_OFFSET_LEFT,
};
scrollIntoView(elements[beginIdx], spot,
/* skipOverflowHiddenElements = */ true);
}
}
_normalize(text) { _normalize(text) {
return text.replace(this.normalizationRegex, function (ch) { return text.replace(this._normalizationRegex, function(ch) {
return CHARACTERS_TO_NORMALIZE[ch]; return CHARACTERS_TO_NORMALIZE[ch];
}); });
} }
@ -148,8 +141,8 @@ class PDFFindController {
*/ */
_prepareMatches(matchesWithLength, matches, matchesLength) { _prepareMatches(matchesWithLength, matches, matchesLength) {
function isSubTerm(matchesWithLength, currentIndex) { function isSubTerm(matchesWithLength, currentIndex) {
let currentElem = matchesWithLength[currentIndex]; const currentElem = matchesWithLength[currentIndex];
let nextElem = matchesWithLength[currentIndex + 1]; const nextElem = matchesWithLength[currentIndex + 1];
// Check for cases like "TAMEd TAME". // Check for cases like "TAMEd TAME".
if (currentIndex < matchesWithLength.length - 1 && if (currentIndex < matchesWithLength.length - 1 &&
@ -160,7 +153,7 @@ class PDFFindController {
// Check for cases like "thIS IS". // Check for cases like "thIS IS".
for (let i = currentIndex - 1; i >= 0; i--) { for (let i = currentIndex - 1; i >= 0; i--) {
let prevElem = matchesWithLength[i]; const prevElem = matchesWithLength[i];
if (prevElem.skipped) { if (prevElem.skipped) {
continue; continue;
} }
@ -215,8 +208,9 @@ class PDFFindController {
} }
_calculatePhraseMatch(query, pageIndex, pageContent, entireWord) { _calculatePhraseMatch(query, pageIndex, pageContent, entireWord) {
let matches = []; const matches = [];
let queryLen = query.length; const queryLen = query.length;
let matchIdx = -queryLen; let matchIdx = -queryLen;
while (true) { while (true) {
matchIdx = pageContent.indexOf(query, matchIdx + queryLen); matchIdx = pageContent.indexOf(query, matchIdx + queryLen);
@ -228,16 +222,18 @@ class PDFFindController {
} }
matches.push(matchIdx); matches.push(matchIdx);
} }
this.pageMatches[pageIndex] = matches; this._pageMatches[pageIndex] = matches;
} }
_calculateWordMatch(query, pageIndex, pageContent, entireWord) { _calculateWordMatch(query, pageIndex, pageContent, entireWord) {
let matchesWithLength = []; const matchesWithLength = [];
// Divide the query into pieces and search for text in each piece. // Divide the query into pieces and search for text in each piece.
let queryArray = query.match(/\S+/g); const queryArray = query.match(/\S+/g);
for (let i = 0, len = queryArray.length; i < len; i++) { for (let i = 0, len = queryArray.length; i < len; i++) {
let subquery = queryArray[i]; const subquery = queryArray[i];
let subqueryLen = subquery.length; const subqueryLen = subquery.length;
let matchIdx = -subqueryLen; let matchIdx = -subqueryLen;
while (true) { while (true) {
matchIdx = pageContent.indexOf(subquery, matchIdx + subqueryLen); matchIdx = pageContent.indexOf(subquery, matchIdx + subqueryLen);
@ -258,27 +254,24 @@ class PDFFindController {
} }
// Prepare arrays for storing the matches. // Prepare arrays for storing the matches.
if (!this.pageMatchesLength) { if (!this._pageMatchesLength) {
this.pageMatchesLength = []; this._pageMatchesLength = [];
} }
this.pageMatchesLength[pageIndex] = []; this._pageMatchesLength[pageIndex] = [];
this.pageMatches[pageIndex] = []; this._pageMatches[pageIndex] = [];
// Sort `matchesWithLength`, remove intersecting terms and put the result // Sort `matchesWithLength`, remove intersecting terms and put the result
// into the two arrays. // into the two arrays.
this._prepareMatches(matchesWithLength, this.pageMatches[pageIndex], this._prepareMatches(matchesWithLength, this._pageMatches[pageIndex],
this.pageMatchesLength[pageIndex]); this._pageMatchesLength[pageIndex]);
} }
_calculateMatch(pageIndex) { _calculateMatch(pageIndex) {
let pageContent = this._normalize(this.pageContents[pageIndex]); let pageContent = this._normalize(this._pageContents[pageIndex]);
let query = this._normalize(this.state.query); let query = this._normalize(this._state.query);
let caseSensitive = this.state.caseSensitive; const { caseSensitive, entireWord, phraseSearch, } = this._state;
let phraseSearch = this.state.phraseSearch;
const entireWord = this.state.entireWord;
let queryLen = query.length;
if (queryLen === 0) { if (query.length === 0) {
// Do nothing: the matches should be wiped out already. // Do nothing: the matches should be wiped out already.
return; return;
} }
@ -295,46 +288,46 @@ class PDFFindController {
} }
this._updatePage(pageIndex); this._updatePage(pageIndex);
if (this.resumePageIdx === pageIndex) { if (this._resumePageIdx === pageIndex) {
this.resumePageIdx = null; this._resumePageIdx = null;
this._nextPageMatch(); this._nextPageMatch();
} }
// Update the match count. // Update the match count.
const pageMatchesCount = this.pageMatches[pageIndex].length; const pageMatchesCount = this._pageMatches[pageIndex].length;
if (pageMatchesCount > 0) { if (pageMatchesCount > 0) {
this.matchesCountTotal += pageMatchesCount; this._matchesCountTotal += pageMatchesCount;
this._updateUIResultsCount(); this._updateUIResultsCount();
} }
} }
_extractText() { _extractText() {
if (this.startedTextExtraction) { // Perform text extraction once if this method is called multiple times.
if (this._extractTextPromises.length > 0) {
return; return;
} }
this.startedTextExtraction = true;
this.pageContents.length = 0;
let promise = Promise.resolve(); let promise = Promise.resolve();
for (let i = 0, ii = this.pdfViewer.pagesCount; i < ii; i++) { for (let i = 0, ii = this._pdfViewer.pagesCount; i < ii; i++) {
let extractTextCapability = createPromiseCapability(); const extractTextCapability = createPromiseCapability();
this.extractTextPromises[i] = extractTextCapability.promise; this._extractTextPromises[i] = extractTextCapability.promise;
promise = promise.then(() => { promise = promise.then(() => {
return this.pdfViewer.getPageTextContent(i).then((textContent) => { return this._pdfViewer.getPageTextContent(i).then((textContent) => {
let textItems = textContent.items; const textItems = textContent.items;
let strBuf = []; const strBuf = [];
for (let j = 0, jj = textItems.length; j < jj; j++) { for (let j = 0, jj = textItems.length; j < jj; j++) {
strBuf.push(textItems[j].str); strBuf.push(textItems[j].str);
} }
// Store the pageContent as a string.
this.pageContents[i] = strBuf.join(''); // Store the page content (text items) as one string.
this._pageContents[i] = strBuf.join('');
extractTextCapability.resolve(i); extractTextCapability.resolve(i);
}, (reason) => { }, (reason) => {
console.error(`Unable to get page ${i + 1} text content`, reason); console.error(`Unable to get text content for page ${i + 1}`, reason);
// Page error -- assuming no text content. // Page error -- assuming no text content.
this.pageContents[i] = ''; this._pageContents[i] = '';
extractTextCapability.resolve(i); extractTextCapability.resolve(i);
}); });
}); });
@ -342,47 +335,46 @@ class PDFFindController {
} }
_updatePage(index) { _updatePage(index) {
if (this.selected.pageIdx === index) { if (this._selected.pageIdx === index) {
// If the page is selected, scroll the page into view, which triggers // If the page is selected, scroll the page into view, which triggers
// rendering the page, which adds the textLayer. Once the textLayer is // rendering the page, which adds the textLayer. Once the textLayer is
// build, it will scroll onto the selected match. // build, it will scroll onto the selected match.
this.pdfViewer.currentPageNumber = index + 1; this._pdfViewer.currentPageNumber = index + 1;
} }
let page = this.pdfViewer.getPageView(index); const page = this._pdfViewer.getPageView(index);
if (page.textLayer) { if (page.textLayer) {
page.textLayer.updateMatches(); page.textLayer.updateMatches();
} }
} }
_nextMatch() { _nextMatch() {
let previous = this.state.findPrevious; const previous = this._state.findPrevious;
let currentPageIndex = this.pdfViewer.currentPageNumber - 1; const currentPageIndex = this._pdfViewer.currentPageNumber - 1;
let numPages = this.pdfViewer.pagesCount; const numPages = this._pdfViewer.pagesCount;
this.active = true; this.active = true;
if (this.dirtyMatch) { if (this._dirtyMatch) {
// Need to recalculate the matches, reset everything. // Need to recalculate the matches, reset everything.
this.dirtyMatch = false; this._dirtyMatch = false;
this.selected.pageIdx = this.selected.matchIdx = -1; this._selected.pageIdx = this._selected.matchIdx = -1;
this.offset.pageIdx = currentPageIndex; this._offset.pageIdx = currentPageIndex;
this.offset.matchIdx = null; this._offset.matchIdx = null;
this.hadMatch = false; this._resumePageIdx = null;
this.resumePageIdx = null; this._pageMatches.length = 0;
this.pageMatches = []; this._pageMatchesLength = null;
this.matchesCountTotal = 0; this._matchesCountTotal = 0;
this.pageMatchesLength = null;
for (let i = 0; i < numPages; i++) { for (let i = 0; i < numPages; i++) {
// Wipe out any previously highlighted matches. // Wipe out any previously highlighted matches.
this._updatePage(i); this._updatePage(i);
// Start finding the matches as soon as the text is extracted. // Start finding the matches as soon as the text is extracted.
if (!(i in this.pendingFindMatches)) { if (!(i in this._pendingFindMatches)) {
this.pendingFindMatches[i] = true; this._pendingFindMatches[i] = true;
this.extractTextPromises[i].then((pageIdx) => { this._extractTextPromises[i].then((pageIdx) => {
delete this.pendingFindMatches[pageIdx]; delete this._pendingFindMatches[pageIdx];
this._calculateMatch(pageIdx); this._calculateMatch(pageIdx);
}); });
} }
@ -390,28 +382,27 @@ class PDFFindController {
} }
// If there's no query there's no point in searching. // If there's no query there's no point in searching.
if (this.state.query === '') { if (this._state.query === '') {
this._updateUIState(FindState.FOUND); this._updateUIState(FindState.FOUND);
return; return;
} }
// If we're waiting on a page, we return since we can't do anything else. // If we're waiting on a page, we return since we can't do anything else.
if (this.resumePageIdx) { if (this._resumePageIdx) {
return; return;
} }
let offset = this.offset; const offset = this._offset;
// Keep track of how many pages we should maximally iterate through. // Keep track of how many pages we should maximally iterate through.
this.pagesToSearch = numPages; this._pagesToSearch = numPages;
// If there's already a `matchIdx` that means we are iterating through a // If there's already a `matchIdx` that means we are iterating through a
// page's matches. // page's matches.
if (offset.matchIdx !== null) { if (offset.matchIdx !== null) {
let numPageMatches = this.pageMatches[offset.pageIdx].length; const numPageMatches = this._pageMatches[offset.pageIdx].length;
if ((!previous && offset.matchIdx + 1 < numPageMatches) || if ((!previous && offset.matchIdx + 1 < numPageMatches) ||
(previous && offset.matchIdx > 0)) { (previous && offset.matchIdx > 0)) {
// The simple case; we just have advance the matchIdx to select // The simple case; we just have advance the matchIdx to select
// the next match on the page. // the next match on the page.
this.hadMatch = true;
offset.matchIdx = (previous ? offset.matchIdx - 1 : offset.matchIdx = (previous ? offset.matchIdx - 1 :
offset.matchIdx + 1); offset.matchIdx + 1);
this._updateMatch(/* found = */ true); this._updateMatch(/* found = */ true);
@ -426,13 +417,12 @@ class PDFFindController {
} }
_matchesReady(matches) { _matchesReady(matches) {
let offset = this.offset; const offset = this._offset;
let numMatches = matches.length; const numMatches = matches.length;
let previous = this.state.findPrevious; const previous = this._state.findPrevious;
if (numMatches) { if (numMatches) {
// There were matches for the page, so initialize `matchIdx`. // There were matches for the page, so initialize `matchIdx`.
this.hadMatch = true;
offset.matchIdx = (previous ? numMatches - 1 : 0); offset.matchIdx = (previous ? numMatches - 1 : 0);
this._updateMatch(/* found = */ true); this._updateMatch(/* found = */ true);
return true; return true;
@ -441,7 +431,7 @@ class PDFFindController {
this._advanceOffsetPage(previous); this._advanceOffsetPage(previous);
if (offset.wrapped) { if (offset.wrapped) {
offset.matchIdx = null; offset.matchIdx = null;
if (this.pagesToSearch < 0) { if (this._pagesToSearch < 0) {
// No point in wrapping again, there were no matches. // No point in wrapping again, there were no matches.
this._updateMatch(/* found = */ false); this._updateMatch(/* found = */ false);
// While matches were not found, searching for a page // While matches were not found, searching for a page
@ -454,30 +444,30 @@ class PDFFindController {
} }
_nextPageMatch() { _nextPageMatch() {
if (this.resumePageIdx !== null) { if (this._resumePageIdx !== null) {
console.error('There can only be one pending page.'); console.error('There can only be one pending page.');
} }
let matches = null; let matches = null;
do { do {
let pageIdx = this.offset.pageIdx; const pageIdx = this._offset.pageIdx;
matches = this.pageMatches[pageIdx]; matches = this._pageMatches[pageIdx];
if (!matches) { if (!matches) {
// The matches don't exist yet for processing by `_matchesReady`, // The matches don't exist yet for processing by `_matchesReady`,
// so set a resume point for when they do exist. // so set a resume point for when they do exist.
this.resumePageIdx = pageIdx; this._resumePageIdx = pageIdx;
break; break;
} }
} while (!this._matchesReady(matches)); } while (!this._matchesReady(matches));
} }
_advanceOffsetPage(previous) { _advanceOffsetPage(previous) {
let offset = this.offset; const offset = this._offset;
let numPages = this.extractTextPromises.length; const numPages = this._extractTextPromises.length;
offset.pageIdx = (previous ? offset.pageIdx - 1 : offset.pageIdx + 1); offset.pageIdx = (previous ? offset.pageIdx - 1 : offset.pageIdx + 1);
offset.matchIdx = null; offset.matchIdx = null;
this.pagesToSearch--; this._pagesToSearch--;
if (offset.pageIdx >= numPages || offset.pageIdx < 0) { if (offset.pageIdx >= numPages || offset.pageIdx < 0) {
offset.pageIdx = (previous ? numPages - 1 : 0); offset.pageIdx = (previous ? numPages - 1 : 0);
@ -487,33 +477,33 @@ class PDFFindController {
_updateMatch(found = false) { _updateMatch(found = false) {
let state = FindState.NOT_FOUND; let state = FindState.NOT_FOUND;
let wrapped = this.offset.wrapped; const wrapped = this._offset.wrapped;
this.offset.wrapped = false; this._offset.wrapped = false;
if (found) { if (found) {
let previousPage = this.selected.pageIdx; const previousPage = this._selected.pageIdx;
this.selected.pageIdx = this.offset.pageIdx; this.selected.pageIdx = this._offset.pageIdx;
this.selected.matchIdx = this.offset.matchIdx; this.selected.matchIdx = this._offset.matchIdx;
state = (wrapped ? FindState.WRAPPED : FindState.FOUND); state = (wrapped ? FindState.WRAPPED : FindState.FOUND);
// Update the currently selected page to wipe out any selected matches. // Update the currently selected page to wipe out any selected matches.
if (previousPage !== -1 && previousPage !== this.selected.pageIdx) { if (previousPage !== -1 && previousPage !== this._selected.pageIdx) {
this._updatePage(previousPage); this._updatePage(previousPage);
} }
} }
this._updateUIState(state, this.state.findPrevious); this._updateUIState(state, this._state.findPrevious);
if (this.selected.pageIdx !== -1) { if (this._selected.pageIdx !== -1) {
this._updatePage(this.selected.pageIdx); this._updatePage(this._selected.pageIdx);
} }
} }
_requestMatchesCount() { _requestMatchesCount() {
const { pageIdx, matchIdx, } = this.selected; const { pageIdx, matchIdx, } = this._selected;
let current = 0, total = this.matchesCountTotal; let current = 0, total = this._matchesCountTotal;
if (matchIdx !== -1) { if (matchIdx !== -1) {
for (let i = 0; i < pageIdx; i++) { for (let i = 0; i < pageIdx; i++) {
current += (this.pageMatches[i] && this.pageMatches[i].length) || 0; current += (this._pageMatches[i] && this._pageMatches[i].length) || 0;
} }
current += matchIdx + 1; current += matchIdx + 1;
} }

15
web/text_layer_builder.js
View File

@ -15,8 +15,11 @@
import { getGlobalEventBus } from './dom_events'; import { getGlobalEventBus } from './dom_events';
import { renderTextLayer } from 'pdfjs-lib'; import { renderTextLayer } from 'pdfjs-lib';
import { scrollIntoView } from './ui_utils';
const EXPAND_DIVS_TIMEOUT = 300; // ms const EXPAND_DIVS_TIMEOUT = 300; // ms
const MATCH_SCROLL_OFFSET_TOP = -50;
const MATCH_SCROLL_OFFSET_LEFT = -400;
/** /**
* @typedef {Object} TextLayerBuilderOptions * @typedef {Object} TextLayerBuilderOptions
@ -243,9 +246,17 @@ class TextLayerBuilder {
let isSelected = (isSelectedPage && i === selectedMatchIdx); let isSelected = (isSelectedPage && i === selectedMatchIdx);
let highlightSuffix = (isSelected ? ' selected' : ''); let highlightSuffix = (isSelected ? ' selected' : '');
// Scroll the selected match into view.
if (this.findController) { if (this.findController) {
this.findController.updateMatchPosition(pageIdx, i, textDivs, if (this.findController.selected.matchIdx === i &&
begin.divIdx); this.findController.selected.pageIdx === pageIdx) {
const spot = {
top: MATCH_SCROLL_OFFSET_TOP,
left: MATCH_SCROLL_OFFSET_LEFT,
};
scrollIntoView(textDivs[begin.divIdx], spot,
/* skipOverflowHiddenElements = */ true);
}
} }
// Match inside new div. // Match inside new div.