Refactor PDF text extraction to support configurable margins for header, footer, left, and right

This commit is contained in:
Richard Roberson 2025-02-25 15:53:45 -07:00
parent f948601e70
commit 92220de5ea
5 changed files with 200 additions and 57 deletions

View file

@ -1,6 +1,6 @@
'use client'; 'use client';
import { Fragment, useState, useRef, useCallback, useEffect } from 'react'; import { Fragment, useState, useRef, useEffect } from 'react';
import { Dialog, DialogPanel, Transition, TransitionChild, Listbox, ListboxButton, ListboxOptions, ListboxOption, Button } from '@headlessui/react'; import { Dialog, DialogPanel, Transition, TransitionChild, Listbox, ListboxButton, ListboxOptions, ListboxOption, Button } from '@headlessui/react';
import { useConfig, ViewType } from '@/contexts/ConfigContext'; import { useConfig, ViewType } from '@/contexts/ConfigContext';
import { ChevronUpDownIcon, CheckIcon } from '@/components/icons/Icons'; import { ChevronUpDownIcon, CheckIcon } from '@/components/icons/Icons';
@ -21,32 +21,54 @@ const viewTypes = [
]; ];
export function DocumentSettings({ isOpen, setIsOpen, epub }: DocViewSettingsProps) { export function DocumentSettings({ isOpen, setIsOpen, epub }: DocViewSettingsProps) {
const { viewType, skipBlank, epubTheme, textExtractionMargin, updateConfigKey } = useConfig(); const {
viewType,
skipBlank,
epubTheme,
headerMargin,
footerMargin,
leftMargin,
rightMargin,
updateConfigKey
} = useConfig();
const { createFullAudioBook } = useEPUB(); const { createFullAudioBook } = useEPUB();
const [progress, setProgress] = useState(0); const [progress, setProgress] = useState(0);
const [isGenerating, setIsGenerating] = useState(false); const [isGenerating, setIsGenerating] = useState(false);
const [localMargin, setLocalMargin] = useState(textExtractionMargin); const [localMargins, setLocalMargins] = useState({
header: headerMargin,
footer: footerMargin,
left: leftMargin,
right: rightMargin
});
const abortControllerRef = useRef<AbortController | null>(null); const abortControllerRef = useRef<AbortController | null>(null);
const selectedView = viewTypes.find(v => v.id === viewType) || viewTypes[0]; const selectedView = viewTypes.find(v => v.id === viewType) || viewTypes[0];
//console.log(localMargin, textExtractionMargin); // Sync local margins with global state
// Sync local margin with global state
useEffect(() => { useEffect(() => {
setLocalMargin(textExtractionMargin); setLocalMargins({
}, [textExtractionMargin]); header: headerMargin,
footer: footerMargin,
left: leftMargin,
right: rightMargin
});
}, [headerMargin, footerMargin, leftMargin, rightMargin]);
// Handler for slider change (updates local state only) // Handler for slider change (updates local state only)
const handleMarginChange = useCallback((event: React.ChangeEvent<HTMLInputElement>) => { const handleMarginChange = (margin: keyof typeof localMargins) => (event: React.ChangeEvent<HTMLInputElement>) => {
setLocalMargin(Number(event.target.value)); setLocalMargins(prev => ({
}, []); ...prev,
[margin]: Number(event.target.value)
}));
};
// Handler for slider release // Handler for slider release
const handleMarginChangeComplete = useCallback(() => { const handleMarginChangeComplete = (margin: keyof typeof localMargins) => () => {
if (localMargin !== textExtractionMargin) { const value = localMargins[margin];
updateConfigKey('textExtractionMargin', localMargin); const configKey = `${margin}Margin`;
if (value !== (useConfig)[configKey as keyof typeof useConfig]) {
updateConfigKey(configKey as 'headerMargin' | 'footerMargin' | 'leftMargin' | 'rightMargin', value);
} }
}, [localMargin, textExtractionMargin, updateConfigKey]); };
const handleStartGeneration = async () => { const handleStartGeneration = async () => {
setIsGenerating(true); setIsGenerating(true);
@ -154,28 +176,92 @@ export function DocumentSettings({ isOpen, setIsOpen, epub }: DocViewSettingsPro
</div>} </div>}
{!epub && <div className="space-y-6"> {!epub && <div className="space-y-6">
<div className="mt-4 space-y-2"> <div className="mt-4 space-y-2">
<label className="block text-sm font-medium text-foreground"> <label className="block text-sm font-medium text-foreground mb-4">
Text Extraction Margin Adjust extraction margins (experimental)
</label> </label>
<div className="flex justify-between"> <div className="grid grid-cols-1 sm:grid-cols-2 gap-2">
<span className="text-xs">0%</span> {/* Header Margin */}
<span className="text-xs font-bold">{Math.round(localMargin * 100)}%</span> <div className="space-y-1">
<span className="text-xs">20%</span> <div className="flex justify-between">
<span className="text-xs">Header</span>
<span className="text-xs font-bold">{Math.round(localMargins.header * 100)}%</span>
</div>
<input
type="range"
min="0"
max="0.2"
step="0.01"
value={localMargins.header}
onChange={handleMarginChange('header')}
onMouseUp={handleMarginChangeComplete('header')}
onKeyUp={handleMarginChangeComplete('header')}
onTouchEnd={handleMarginChangeComplete('header')}
className="w-full bg-offbase rounded-lg appearance-none cursor-pointer accent-accent [&::-webkit-slider-runnable-track]:bg-offbase [&::-webkit-slider-runnable-track]:rounded-lg [&::-webkit-slider-thumb]:appearance-none [&::-webkit-slider-thumb]:h-4 [&::-webkit-slider-thumb]:w-4 [&::-webkit-slider-thumb]:rounded-full [&::-webkit-slider-thumb]:bg-accent [&::-moz-range-track]:bg-offbase [&::-moz-range-track]:rounded-lg [&::-moz-range-thumb]:appearance-none [&::-moz-range-thumb]:h-4 [&::-moz-range-thumb]:w-4 [&::-moz-range-thumb]:rounded-full [&::-moz-range-thumb]:bg-accent"
/>
</div>
{/* Footer Margin */}
<div className="space-y-1">
<div className="flex justify-between">
<span className="text-xs">Footer</span>
<span className="text-xs font-bold">{Math.round(localMargins.footer * 100)}%</span>
</div>
<input
type="range"
min="0"
max="0.2"
step="0.01"
value={localMargins.footer}
onChange={handleMarginChange('footer')}
onMouseUp={handleMarginChangeComplete('footer')}
onKeyUp={handleMarginChangeComplete('footer')}
onTouchEnd={handleMarginChangeComplete('footer')}
className="w-full bg-offbase rounded-lg appearance-none cursor-pointer accent-accent [&::-webkit-slider-runnable-track]:bg-offbase [&::-webkit-slider-runnable-track]:rounded-lg [&::-webkit-slider-thumb]:appearance-none [&::-webkit-slider-thumb]:h-4 [&::-webkit-slider-thumb]:w-4 [&::-webkit-slider-thumb]:rounded-full [&::-webkit-slider-thumb]:bg-accent [&::-moz-range-track]:bg-offbase [&::-moz-range-track]:rounded-lg [&::-moz-range-thumb]:appearance-none [&::-moz-range-thumb]:h-4 [&::-moz-range-thumb]:w-4 [&::-moz-range-thumb]:rounded-full [&::-moz-range-thumb]:bg-accent"
/>
</div>
{/* Left Margin */}
<div className="space-y-1">
<div className="flex justify-between">
<span className="text-xs">Left</span>
<span className="text-xs font-bold">{Math.round(localMargins.left * 100)}%</span>
</div>
<input
type="range"
min="0"
max="0.2"
step="0.01"
value={localMargins.left}
onChange={handleMarginChange('left')}
onMouseUp={handleMarginChangeComplete('left')}
onKeyUp={handleMarginChangeComplete('left')}
onTouchEnd={handleMarginChangeComplete('left')}
className="w-full bg-offbase rounded-lg appearance-none cursor-pointer accent-accent [&::-webkit-slider-runnable-track]:bg-offbase [&::-webkit-slider-runnable-track]:rounded-lg [&::-webkit-slider-thumb]:appearance-none [&::-webkit-slider-thumb]:h-4 [&::-webkit-slider-thumb]:w-4 [&::-webkit-slider-thumb]:rounded-full [&::-webkit-slider-thumb]:bg-accent [&::-moz-range-track]:bg-offbase [&::-moz-range-track]:rounded-lg [&::-moz-range-thumb]:appearance-none [&::-moz-range-thumb]:h-4 [&::-moz-range-thumb]:w-4 [&::-moz-range-thumb]:rounded-full [&::-moz-range-thumb]:bg-accent"
/>
</div>
{/* Right Margin */}
<div className="space-y-1">
<div className="flex justify-between">
<span className="text-xs">Right</span>
<span className="text-xs font-bold">{Math.round(localMargins.right * 100)}%</span>
</div>
<input
type="range"
min="0"
max="0.2"
step="0.01"
value={localMargins.right}
onChange={handleMarginChange('right')}
onMouseUp={handleMarginChangeComplete('right')}
onKeyUp={handleMarginChangeComplete('right')}
onTouchEnd={handleMarginChangeComplete('right')}
className="w-full bg-offbase rounded-lg appearance-none cursor-pointer accent-accent [&::-webkit-slider-runnable-track]:bg-offbase [&::-webkit-slider-runnable-track]:rounded-lg [&::-webkit-slider-thumb]:appearance-none [&::-webkit-slider-thumb]:h-4 [&::-webkit-slider-thumb]:w-4 [&::-webkit-slider-thumb]:rounded-full [&::-webkit-slider-thumb]:bg-accent [&::-moz-range-track]:bg-offbase [&::-moz-range-track]:rounded-lg [&::-moz-range-thumb]:appearance-none [&::-moz-range-thumb]:h-4 [&::-moz-range-thumb]:w-4 [&::-moz-range-thumb]:rounded-full [&::-moz-range-thumb]:bg-accent"
/>
</div>
</div> </div>
<input <p className="text-xs text-muted mt-2">
type="range" Adjust margins to exclude content from edges of the page during text extraction
min="0"
max="0.2"
step="0.01"
value={localMargin}
onChange={handleMarginChange}
onMouseUp={handleMarginChangeComplete}
onKeyUp={handleMarginChangeComplete}
onTouchEnd={handleMarginChangeComplete}
className="w-full bg-offbase rounded-lg appearance-none cursor-pointer accent-accent [&::-webkit-slider-runnable-track]:bg-offbase [&::-webkit-slider-runnable-track]:rounded-lg [&::-webkit-slider-thumb]:appearance-none [&::-webkit-slider-thumb]:h-4 [&::-webkit-slider-thumb]:w-4 [&::-webkit-slider-thumb]:rounded-full [&::-webkit-slider-thumb]:bg-accent [&::-moz-range-track]:bg-offbase [&::-moz-range-track]:rounded-lg [&::-moz-range-thumb]:appearance-none [&::-moz-range-thumb]:h-4 [&::-moz-range-thumb]:w-4 [&::-moz-range-thumb]:rounded-full [&::-moz-range-thumb]:bg-accent"
/>
<p className="text-xs text-muted">
{"Don't"} include content from outer rim of the page during text extraction (experimental)
</p> </p>
</div> </div>
<Listbox <Listbox

View file

@ -16,6 +16,10 @@ type ConfigValues = {
skipBlank: boolean; skipBlank: boolean;
epubTheme: boolean; epubTheme: boolean;
textExtractionMargin: number; textExtractionMargin: number;
headerMargin: number;
footerMargin: number;
leftMargin: number;
rightMargin: number;
}; };
/** Interface defining the configuration context shape and functionality */ /** Interface defining the configuration context shape and functionality */
@ -28,6 +32,10 @@ interface ConfigContextType {
skipBlank: boolean; skipBlank: boolean;
epubTheme: boolean; epubTheme: boolean;
textExtractionMargin: number; textExtractionMargin: number;
headerMargin: number;
footerMargin: number;
leftMargin: number;
rightMargin: number;
updateConfig: (newConfig: Partial<{ apiKey: string; baseUrl: string; viewType: ViewType }>) => Promise<void>; updateConfig: (newConfig: Partial<{ apiKey: string; baseUrl: string; viewType: ViewType }>) => Promise<void>;
updateConfigKey: <K extends keyof ConfigValues>(key: K, value: ConfigValues[K]) => Promise<void>; updateConfigKey: <K extends keyof ConfigValues>(key: K, value: ConfigValues[K]) => Promise<void>;
isLoading: boolean; isLoading: boolean;
@ -52,6 +60,10 @@ export function ConfigProvider({ children }: { children: ReactNode }) {
const [skipBlank, setSkipBlank] = useState<boolean>(true); const [skipBlank, setSkipBlank] = useState<boolean>(true);
const [epubTheme, setEpubTheme] = useState<boolean>(false); const [epubTheme, setEpubTheme] = useState<boolean>(false);
const [textExtractionMargin, setTextExtractionMargin] = useState<number>(0.07); const [textExtractionMargin, setTextExtractionMargin] = useState<number>(0.07);
const [headerMargin, setHeaderMargin] = useState<number>(0.07);
const [footerMargin, setFooterMargin] = useState<number>(0.07);
const [leftMargin, setLeftMargin] = useState<number>(0.07);
const [rightMargin, setRightMargin] = useState<number>(0.07);
const [isLoading, setIsLoading] = useState(true); const [isLoading, setIsLoading] = useState(true);
const [isDBReady, setIsDBReady] = useState(false); const [isDBReady, setIsDBReady] = useState(false);
@ -72,6 +84,10 @@ export function ConfigProvider({ children }: { children: ReactNode }) {
const cachedSkipBlank = await getItem('skipBlank'); const cachedSkipBlank = await getItem('skipBlank');
const cachedEpubTheme = await getItem('epubTheme'); const cachedEpubTheme = await getItem('epubTheme');
const cachedMargin = await getItem('textExtractionMargin'); const cachedMargin = await getItem('textExtractionMargin');
const cachedHeaderMargin = await getItem('headerMargin');
const cachedFooterMargin = await getItem('footerMargin');
const cachedLeftMargin = await getItem('leftMargin');
const cachedRightMargin = await getItem('rightMargin');
// Only set API key and base URL if they were explicitly saved by the user // Only set API key and base URL if they were explicitly saved by the user
if (cachedApiKey) { if (cachedApiKey) {
@ -90,6 +106,10 @@ export function ConfigProvider({ children }: { children: ReactNode }) {
setSkipBlank(cachedSkipBlank === 'false' ? false : true); setSkipBlank(cachedSkipBlank === 'false' ? false : true);
setEpubTheme(cachedEpubTheme === 'true'); setEpubTheme(cachedEpubTheme === 'true');
setTextExtractionMargin(parseFloat(cachedMargin || '0.07')); setTextExtractionMargin(parseFloat(cachedMargin || '0.07'));
setHeaderMargin(parseFloat(cachedHeaderMargin || '0.07'));
setFooterMargin(parseFloat(cachedFooterMargin || '0.07'));
setLeftMargin(parseFloat(cachedLeftMargin || '0.07'));
setRightMargin(parseFloat(cachedRightMargin || '0.07'));
// Only save non-sensitive settings by default // Only save non-sensitive settings by default
if (!cachedViewType) { if (!cachedViewType) {
@ -104,6 +124,10 @@ export function ConfigProvider({ children }: { children: ReactNode }) {
if (cachedMargin === null) { if (cachedMargin === null) {
await setItem('textExtractionMargin', '0.07'); await setItem('textExtractionMargin', '0.07');
} }
if (cachedHeaderMargin === null) await setItem('headerMargin', '0.07');
if (cachedFooterMargin === null) await setItem('footerMargin', '0.07');
if (cachedLeftMargin === null) await setItem('leftMargin', '0.0');
if (cachedRightMargin === null) await setItem('rightMargin', '0.0');
} catch (error) { } catch (error) {
console.error('Error initializing:', error); console.error('Error initializing:', error);
@ -181,6 +205,18 @@ export function ConfigProvider({ children }: { children: ReactNode }) {
case 'textExtractionMargin': case 'textExtractionMargin':
setTextExtractionMargin(value as number); setTextExtractionMargin(value as number);
break; break;
case 'headerMargin':
setHeaderMargin(value as number);
break;
case 'footerMargin':
setFooterMargin(value as number);
break;
case 'leftMargin':
setLeftMargin(value as number);
break;
case 'rightMargin':
setRightMargin(value as number);
break;
} }
} catch (error) { } catch (error) {
console.error(`Error updating config key ${key}:`, error); console.error(`Error updating config key ${key}:`, error);
@ -198,6 +234,10 @@ export function ConfigProvider({ children }: { children: ReactNode }) {
skipBlank, skipBlank,
epubTheme, epubTheme,
textExtractionMargin, textExtractionMargin,
headerMargin,
footerMargin,
leftMargin,
rightMargin,
updateConfig, updateConfig,
updateConfigKey, updateConfigKey,
isLoading, isLoading,

View file

@ -233,8 +233,8 @@ export function EPUBProvider({ children }: { children: ReactNode }) {
audioChunks.push(audioBuffer); audioChunks.push(audioBuffer);
// Add a small pause between sections (500ms of silence) // Add a small pause between sections (1s of silence)
const silenceBuffer = new ArrayBuffer(24000); const silenceBuffer = new ArrayBuffer(48000);
audioChunks.push(silenceBuffer); audioChunks.push(silenceBuffer);
} catch (error) { } catch (error) {

View file

@ -77,8 +77,19 @@ const PDFContext = createContext<PDFContextType | undefined>(undefined);
* @param {ReactNode} props.children - Child components to be wrapped by the provider * @param {ReactNode} props.children - Child components to be wrapped by the provider
*/ */
export function PDFProvider({ children }: { children: ReactNode }) { export function PDFProvider({ children }: { children: ReactNode }) {
const { setText: setTTSText, stop, currDocPageNumber: currDocPage, currDocPages, setCurrDocPages } = useTTS(); const {
const { textExtractionMargin } = useConfig(); setText: setTTSText,
stop,
currDocPageNumber: currDocPage,
currDocPages,
setCurrDocPages
} = useTTS();
const {
headerMargin,
footerMargin,
leftMargin,
rightMargin
} = useConfig();
// Current document state // Current document state
const [currDocURL, setCurrDocURL] = useState<string>(); const [currDocURL, setCurrDocURL] = useState<string>();
@ -106,7 +117,12 @@ export function PDFProvider({ children }: { children: ReactNode }) {
const loadCurrDocText = useCallback(async () => { const loadCurrDocText = useCallback(async () => {
try { try {
if (!pdfDocument) return; if (!pdfDocument) return;
const text = await extractTextFromPDF(pdfDocument, currDocPage, textExtractionMargin); const text = await extractTextFromPDF(pdfDocument, currDocPage, {
header: headerMargin,
footer: footerMargin,
left: leftMargin,
right: rightMargin
});
// Only update TTS text if the content has actually changed // Only update TTS text if the content has actually changed
// This prevents unnecessary resets of the sentence index // This prevents unnecessary resets of the sentence index
if (text !== currDocText || text === '') { if (text !== currDocText || text === '') {
@ -116,7 +132,7 @@ export function PDFProvider({ children }: { children: ReactNode }) {
} catch (error) { } catch (error) {
console.error('Error loading PDF text:', error); console.error('Error loading PDF text:', error);
} }
}, [pdfDocument, currDocPage, setTTSText, currDocText, textExtractionMargin]); }, [pdfDocument, currDocPage, setTTSText, currDocText, headerMargin, footerMargin, leftMargin, rightMargin]);
/** /**
* Effect hook to update document text when the page changes * Effect hook to update document text when the page changes

View file

@ -26,12 +26,15 @@ export function convertPDFDataToURL(pdfData: Blob): Promise<string> {
} }
// Text Processing functions // Text Processing functions
export async function extractTextFromPDF(pdf: PDFDocumentProxy, pageNumber: number, margin = 0.07): Promise<string> { export async function extractTextFromPDF(
pdf: PDFDocumentProxy,
pageNumber: number,
margins = { header: 0.07, footer: 0.07, left: 0.07, right: 0.07 }
): Promise<string> {
try { try {
const page = await pdf.getPage(pageNumber); const page = await pdf.getPage(pageNumber);
const textContent = await page.getTextContent(); const textContent = await page.getTextContent();
// Get page viewport to help with positioning
const viewport = page.getViewport({ scale: 1.0 }); const viewport = page.getViewport({ scale: 1.0 });
const pageHeight = viewport.height; const pageHeight = viewport.height;
const pageWidth = viewport.width; const pageWidth = viewport.width;
@ -39,34 +42,32 @@ export async function extractTextFromPDF(pdf: PDFDocumentProxy, pageNumber: numb
const textItems = textContent.items.filter((item): item is TextItem => { const textItems = textContent.items.filter((item): item is TextItem => {
if (!('str' in item && 'transform' in item)) return false; if (!('str' in item && 'transform' in item)) return false;
// Get all transform matrix values
const [scaleX, skewX, skewY, scaleY, x, y] = item.transform; const [scaleX, skewX, skewY, scaleY, x, y] = item.transform;
// Check for reasonable scale values (not too small or too large) // Basic text filtering
if (Math.abs(scaleX) < 1 || Math.abs(scaleX) > 20) return false; if (Math.abs(scaleX) < 1 || Math.abs(scaleX) > 20) return false;
if (Math.abs(scaleY) < 1 || Math.abs(scaleY) > 20) return false; if (Math.abs(scaleY) < 1 || Math.abs(scaleY) > 20) return false;
// Check for reasonable skew values (should be close to 0 for normal text)
if (Math.abs(skewX) > 0.5 || Math.abs(skewY) > 0.5) return false; if (Math.abs(skewX) > 0.5 || Math.abs(skewY) > 0.5) return false;
// Filter out positions in header/footer areas using configurable margin // Calculate margins in PDF coordinate space (y=0 is at bottom)
const topMargin = pageHeight * margin; const headerY = pageHeight * (1 - margins.header); // Convert from top margin to bottom-based Y
const bottomMargin = pageHeight * (1 - margin); const footerY = pageHeight * margins.footer; // Footer Y stays as is since it's already bottom-based
if (y < topMargin || y > bottomMargin) { const leftX = pageWidth * margins.left;
const rightX = pageWidth * (1 - margins.right);
// Check margins - remember y=0 is at bottom of page in PDF coordinates
if (y > headerY || y < footerY) { // Y greater than headerY means it's in header area, less than footerY means footer area
return false; return false;
} }
// Filter out positions in left/right margin areas // Check horizontal margins
const leftMargin = pageWidth * margin; if (x < leftX || x > rightX) {
const rightMargin = pageWidth * (1 - margin);
if (x < leftMargin || x > rightMargin) {
return false; return false;
} }
// Check for reasonable x position values // Sanity check for coordinates
if (x < 0 || x > pageWidth) return false; if (x < 0 || x > pageWidth) return false;
// Filter out empty strings or strings with only whitespace
return item.str.trim().length > 0; return item.str.trim().length > 0;
}); });